Molecular Cell

Article

Identification of Two DistinctHybrid State Intermediates on the RibosomeJames B. Munro,1 Roger B. Altman,1 Nathan O’Connor,1 and Scott C. Blanchard1,*1Department of Physiology and Biophysics, Weill Medical College of Cornell University, 1300 York Avenue,

New York, NY 10021, USA

*Correspondence: scb2005@med.cornell.eduDOI 10.1016/j.molcel.2007.01.022

SUMMARY

High spatial and time resolution single-moleculefluorescence resonance energy transfer mea-surements have been used to probe the struc-tural and kinetic parameters of transfer RNA(tRNA) movements within the aminoacyl (A) andpeptidyl (P) sites of the ribosome. Our investiga-tion of tRNA motions, quantified on wild-type,mutant, and L1-depleted ribosome complexes,reveals a dynamic exchange between threemetastable tRNA configurations, one of whichis a previously unidentified hybrid state in whichonly deacylated-tRNA adopts its hybrid (P/E)configuration. This new dynamic informationsuggests a framework in which the formationof intermediate states in the translocation pro-cess is achieved through global conformationalrearrangements of the ribosome particle.

INTRODUCTION

Enzymes are dynamic machines, undergoing multiscale

conformational changes during catalysis (Frauenfelder

et al., 1991; Wolf-Watz et al., 2004; Schnell et al., 2004;

Ridge et al., 2006). Atomic resolution structures of the

ribosome (Schuwirth et al., 2005; Korostelev et al., 2006;

Selmer et al., 2006) provide an essential foundation for un-

derstanding the mechanism of protein synthesis and shift

the focus of ribosome research toward reconciling these

static structural snapshots with the fundamentally dy-

namic nature of translation. Remodeling of the ribosome

structure takes place upon subunit association (Blaha

et al., 2002), transfer RNA (tRNA) and translation factor

binding (Ogle et al., 2001; Valle et al., 2003b), peptide

bond formation (Schmeing et al., 2005), and translocation

(Agrawal et al., 1999b; Frank and Agrawal, 2000; Valle

et al., 2003a). Ribosome conformation is also sensitive

to mutations (Gabashvili et al., 1999; Vila-Sanjurjo et al.,

2003), buffer conditions (Agrawal et al., 1999a; Muth

et al., 2001), and the binding of small-molecule inhibitors

(Yonath, 2005; Ogle and Ramakrishnan, 2005; Moore

and Steitz, 2003). A complete understanding of translation

therefore depends critically on defining the structural and

Molecul

kinetic landscape of ribosome conformations and their

relation to the mechanism of protein synthesis.

During the elongation phase of translation, tRNAs rap-

idly and directionally translocate in �30 A steps through

structurally distinct aminoacyl (A), peptidyl (P), and exit (E)

sites at the interface of the two ribosomal subunits (30S

and 50S in bacteria). The speed and accuracy of these

translocation processes are fueled by elongation factor-

dependent GTP hydrolysis (Rodnina et al., 1997; Wilden

et al., 2006).

The dynamic remodeling of tRNA position on the ribo-

some where specific interactions must be broken and

reformed is of fundamental importance to the mechanism

of translocation. These include base-pairing interactions

between the universally conserved CCA termini of tRNA

and the A and P loops within the large subunit peptidyl-

transferase center (PTC), recognition elements within the

small subunit decoding site that bind the anticodon-codon

complexes, and bridge elements spanning the subunit

interface that contact the central regions of tRNA (Koros-

telev et al., 2006; Selmer et al., 2006). These conserved in-

teractions help maintain the proper reading frame of trans-

lation (Namy et al., 2006) and prevent untimely tRNA

dissociation.

The ribosome’s capacity to maintain a firm grasp on

peptidyl-tRNA during its movement from the A to the P

site has been the focus of ribosome research for several

decades. Spirin first proposed that the ribosome must ‘‘un-

lock’’ its grip on peptidyl-tRNA in the A site prior to translo-

cation to the P site (Spirin, 1968). Bretcher hypothesized

that unlocking places tRNAs in hybrid positions, distinct

from their classical binding sites (Crick, 1958; Bretscher,

1968). Elegant structural interrogations of tRNA-ribosome

interactions have since ascertained that the ribosome’s

intrinsic capacity to direct tRNAs toward ‘‘hybrid’’ configu-

rations plays a key role in the translocation process

(Sharma et al., 2004; Dorner et al., 2006). Hybrid tRNA con-

figurations are formed in the absence of elongation factor-

G (EF-G) and arise from the movements of A and P site

tRNA acceptor stems within the large (50S) subunit, inde-

pendent of the anticodon-codon complexes that remain

stably bound on the small (30S) subunit (Moazed and Nol-

ler, 1989). A tRNA configuration in which both A and P site

tRNAs adopt hybrid configurations (A/P-P/E) is an authen-

tic intermediate in the translocation process (Dorner et al.,

2006), and its formation is affected by the aminoacylation

ar Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 505

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

state of tRNA (Semenkov et al., 2000; Sharma et al., 2004;

Blanchard et al., 2004a).

Here, by using high spatial and time resolution single-

molecule fluorescence resonance energy transfer (FRET)

measurements, we show that, in addition to the well-

established ‘‘classical’’ (A/A-P/P) and ‘‘hybrid’’ (A/P-P/E)

states, a previously uncharacterized configuration of

tRNA exists in which only deacylated-tRNA adopts a hy-

brid state (A/A-P/E). In an effort to establish the rates of

classical and hybrid states interconversion, we have im-

plemented the use of the QuB software package (http://

www.qub.buffalo.edu) to aid the analysis of more than

3000 single-molecule FRET trajectories of ribosome parti-

cles carrying site-specifically dye-labeled A and P site

tRNA. QuB is a tool previously established for the study

of single-ion channel function that has more recently

been adapted for the quantification of single motor protein

movements (Milescu et al., 2006). Our preliminary kinetic

analysis of wild-type and specifically mutated ribosome

complexes supports a model in which two distinct hybrid

states are formed by global rearrangements in ribosome

conformation whose activation energies are of the same

magnitude as those required for translocation catalyzed

by EF-G-dependent GTP hydrolysis (�70 kJ/mol) (Katunin

et al., 2002; Studer et al., 2003).

These data provide an initial structural and kinetic

framework for understanding the physical characteristics

of translocation intermediates and how an interplay of ri-

bosome-tRNA interactions determines the conformational

state of the translating particle. In this view, elements of

initiator tRNA, translation factors, and the growing peptide

chain may function to regulate specific elongation pro-

cesses through the modification of ribosome normal

modes (Tama et al., 2004; Wang et al., 2004).

By relating our data to global structural changes ob-

served in the ribosome associated with movement of de-

acylated-tRNA into a P/E hybrid state (Valle et al., 2003a),

we provide an estimate of apparent ‘‘unlocking’’ and

‘‘locking’’ rates of the ribosome that suggest that unlock-

ing is �2-fold faster than the rate of translocation (Studer

et al., 2003). We also observe that the growing peptide

chain accelerates the rate of unlocking, suggesting that,

analogous to the initiation to elongation transition in tran-

scription (Yin and Steitz, 2002; Boeger et al., 2005), the

growing peptide mediates structural transitions within

the ribosome that may regulate early translation events.

RESULTS

Deacylated- and Peptidyl-tRNAs Fluctuate

between Classical and Two Distinct Hybrid State

tRNA Configurations on the Ribosome

To explore the nature of tRNA hybrid configurations on

the ribosome, initiation complexes were prepared by

using E. coli components, including Cy3-labeled fMet-

tRNAfMet(Cy3-s4U8) in the P site and a phenylalanine

(UUC) codon in the A site. Initiation complexes were im-

mobilized on passivated quartz surfaces as previously de-

506 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Else

scribed (Blanchard et al., 2004a, 2004b), and the A site

was filled enzymatically by incubation with the ternary

complex of EF-Tu(GTP)-Phe-tRNAPhe(Cy5-acp3U47).

Dye-labeled tRNAs were purified to homogeneity prior to

use. Fluorophores within surface-immobilized complexes

were excited by the evanescent wave generated by total

internal reflection (TIR) of a 532 nm wavelength laser. Pho-

tons emitted by Cy3/Cy5 fluorophores were collected at

a rate of 25 frames per second by using a high-numerical

aperture objective and imaged separately onto a cooled

back-thinned CCD. As previously described, dye-labeled

tRNAs are competent in tRNA selection, peptide bond

formation, and translocation. On the millisecond time-

scale, fluorescence anisotropy contributes insignificantly

to FRET measurements (Blanchard et al., 2004a, 2004b).

Surface-immobilized ribosomes were efficiently converted

to pretranslocation complexes (�90%) as measured by

the fraction of ribosomes per image showing FRET. In

the absence of EF-G, <5% of ribosome complexes trans-

locate during the observation period as demonstrated by

toeprinting and puromycin reactivity (Blanchard et al.,

2004a).

On the ribosome, Cy3 and Cy5 fluorophores linked to

the elbow regions of A and P site tRNAs are separated

by a distance near their R0 value (�60 A), making the sys-

tem ideal for detecting small changes (�5–20 A) in their

relative distance. By recording Cy3 and Cy5 fluorescence

intensities from single ribosome complexes, the FRET ef-

ficiency relationship (FRET = ICy5/[ICy5 + ICy3]) can there-

fore be used to estimate time-dependent changes in the

intermolecular distance between tRNAs. FRET trajecto-

ries from hundreds of single pretranslocation complexes

were recorded simultaneously at a higher time resolution

(2.5-fold) than was possible previously (Blanchard et al.,

2004a) without appreciable loss of signal to noise

(R7:1). Mutagenesis was used to probe the nature of

each FRET state observed. Histograms, each composed

of the FRET trajectories of R250 individual molecules,

greatly aided in the assessment of kinetic behaviors of

the various ribosome complexes investigated.

Predominantly, tRNAs occupy (�60% of the time) a

high-FRET classical configuration (A/A-P/P) on wild-type

pretranslocation ribosome complexes carrying an fMet-

Phe dipeptide on the A site tRNA (Figure 2; see Table S1

in the Supplemental Data available with this article online)

(Blanchard et al., 2004a). Inspection of the single-mole-

cule FRET trajectories shows that occupancy of the high-

FRET state is punctuated by rapid transitions to at least

two metastable (R100 ms) intermediate-FRET tRNA con-

figurations (Figure 1). For simplicity, we refer to the three

FRET states observed as the classical (C) state (�0.55

FRET), hybrid state-1 (H1) (�0.39 FRET), and hybrid

state-2 (H2) (�0.24 FRET) rather than by their absolute

values (Figure 1; Table S2). Short-lived transitions to 0

FRET, which arise from photophysical blinking of the

Cy3 or Cy5 dye molecules, have been minimized through

the use of an optimized oxygen-scavenging system

containing a cocktail of triplet state quenchers (our

vier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Figure 1. Single-Molecule FRET Trajectories Show Three tRNA Configurations on the Ribosome

(A) Fluorescence and FRET trajectories reveal the existence of three distinct FRET states as indicated by C, H1, and H2 (see text). Cy3 and Cy5 fluo-

rescence are shown in green and red, respectively. FRET efficiency (FRET = ICy5/[ICy3 + ICy5]) is shown in blue. The idealization of the data is overlaid in

red on the FRET trace.

(B) The boxed region in (A) is expanded to highlight specific FRET transitions.

unpublished data). Changes in FRET between classical

and hybrid states lie within the linear region of the Forster

relationship, and it can therefore be estimated that transi-

tions from the classical state to hybrid state-1 (DFRET z0.16) and hybrid state-2 (DFRET z 0.31) correspond to

increases in distance between tRNAs of �7 A and �15 A,

respectively.

The apparent transition rates between tRNA configura-

tions were estimated by using the freely available QuB

software package, a Markov model-based kinetic analysis

tool. In an attempt to understand the kinetic parameters of

the transitions between FRET states, two Markov chain

models consisting of a fixed number of states were con-

sidered: (1) a four-state model including a high-FRET clas-

Molecular

sical state, a single low-FRET hybrid state, and two zero-

FRET states—a blinking state and a photobleached state;

(2) a five-state model that contained an additional interme-

diate-FRET hybrid state. FRET traces were idealized to

each of these schemes by using a hidden Markov model-

based segmental k-means algorithm (Qin, 2004). The re-

sulting sequence of dwell times was used to optimize ki-

netic parameters of the model using a maximum-likelihood

estimation procedure (Qin et al., 1996, 1997). This analysis

yielded a unique set of rate constants for each single-mol-

ecule trajectory that was reasonably insensitive to varia-

tions in initial parameter values. Mean rates and standard

errors were calculated for each model parameter, and

single-molecule FRET traces were simulated with the

Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 507

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Figure 2. Increasing Peptide Length Increases Hybrid State Occupancy

(A) Single-molecule FRET trajectories were summed into histograms to reveal the population behavior of complexes bearing deacylated-tRNAfMet

in the P site and (left) Phe-tRNAPhe, (center) Met-Phe-tRNAPhe, or (right) fMet-Phe-tRNAPhe in the A site. Scales shown at the top of each panel of

histograms indicate the relative populations. Zero-FRET states arise from both blinking and photobleaching.

508 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

four- and five-state models. These data were then summed

into population histograms for direct comparison with ex-

perimental data. The maximized log-likelihood per transi-

tion calculated for the four- and five-state models and

visual inspection of experimental and simulated histo-

grams convincingly show that the five-state model better

approximates the experimental data (Figure S1). More-

over, the FRET values observed before and after transi-

tions identified in the five-state model idealization clearly

show a clustering of transitions between three distinct non-

zero FRET states (Figures 2 and 3). These so-called transi-

tion density plots (McKinney et al., 2006) further support

the existence of at least two distinct hybrid states. Since

it is logical that deacylated-tRNA must exit the large sub-

unit P site prior to the arrival of peptidyl-tRNA there, we

posit that formation of hybrid state-2 (lowest FRET) results

from the independent movement of deacylated-tRNA

to the P/E hybrid state (A/A-P/E). In this model, hybrid

state-1 therefore results from the simultaneous move-

ments of both tRNAs to hybrid configurations (A/P-P/E).

The Peptide Modulates tRNAs’ Equilibria

on the Ribosome

The peptide’s influence on tRNA fluctuations to hybrid

states was examined by preparing initiation complexes

with different acylated species of Cy3-labeled tRNAfMet

in the P site: deacylated-tRNAfMet, Met-tRNAfMet, or

fMet-tRNAfMet. Incubation of these complexes with Cy5-

labeled Phe-tRNAPhe ternary complex efficiently gener-

ated pretranslocation complexes bearing A site tRNA of

varied peptide lengths: Phe-tRNAPhe, Met-Phe-tRNAPhe,

and fMet-Phe-tRNAPhe. Analysis of these complexes

showed that increased peptide length correlates with

increased occupation of hybrid tRNA configurations (Fig-

ure 2; Table S1), which can be attributed largely to an

increased rate of deacylated-tRNA exiting the classical

state (Table S3). Absence of the formyl group on the

dipeptide (Met-Phe) moderately stabilized the classical

state. Truncation of the peptide to a single amino acid

(Phe) further exacerbated this trend.

Deletion of Ribosomal Protein L1 Stabilizes

the Classical State by Disrupting tRNA

Binding in the E Site

The gene encoding ribosomal protein L1, rplA, was

knocked out of the E. coli genome by a ‘‘gene gorging’’

protocol (Herring et al., 2003). Ribosomes isolated from

this strain were specifically lacking L1, as shown by

SDS-PAGE analysis (Figure S2A). L1-depleted ribosome

complexes were initiated with dye-labeled fMet-tRNAfMet,

surface immobilized, and converted to pretranslocation

complexes as described.

Molecula

As predicted by the model, significant stabilization

of the classical state was observed (Figure 3; Table S1).

Kinetic analysis also showed that both hybrid tRNA con-

figurations were significantly destabilized (Tables S1 and

S3). These observations strongly support the hypothesis

that deacylated-tRNA adopts a P/E state in both hybrid

state-1 and hybrid state-2. This preliminary assignment

also suggests that transitions among hybrid states corre-

spond to independent motions of peptidyl-tRNA between

the large subunit A and P sites.

Specific tRNA-rRNA Interactions Affect the

Equilibrium Positions of tRNA on the Ribosome

tRNA binding in the P site is characterized by base-pairing

interactions between tRNA residues C74 and C75 and

G2252 and G2251 of the P loop of 23S ribosomal RNA

(rRNA) (Samaha et al., 1995). Similarly, binding at the A

site includes a base-pairing interaction between C75 of

tRNA and G2553 of the A loop (Kim and Green, 1999).

To test the hypothesis that hybrid states-1 and -2 correlate

with specific interactions of the 30 ends of tRNAs with

rRNA within the PTC, experiments were performed with

ribosomes that contain single point mutations in either

the A loop (G2553C) or P loop (G2252C) known to alter the

translocation mechanism (Dorner et al., 2006).

According to the model, disrupting the C75-G2553 base

pair should destabilize A site binding of peptidyl-tRNA fa-

voring the previously described hybrid state-1 (A/P-P/E)

(Sharma et al., 2004; Dorner et al., 2006); disruption of

the C74-G2252 base pair should destabilize deacylated-

tRNA binding in the P site, favoring the previously unchar-

acterized hybrid state-2 (A/A-P/E). As shown in Figure 3,

the single-molecule data support these predictions. The

occupancy of the classical state is dramatically reduced

in both mutant complexes in favor of FRET states as-

signed to hybrid state-1 (A/P-P/E) and hybrid state-2 (A/

A-P/E) (Table S1). Although both hybrid configurations

are observed in each system, A loop mutant ribosomes

predominantly occupy hybrid state-1, and P loop mutant

ribosomes strongly favor hybrid state-2. These data pro-

vide striking evidence that hybrid state-1 results from va-

cancy of the large subunit A site and hybrid state-2 from

vacancy of the large subunit P site.

The Assignment of Hybrid tRNA Configurations Is

Confirmed by Puromycin Reactivity

Further validation of the assignment of FRET states was

made by using a single-molecule puromycin assay, which

tests for peptide occupancy at the P site (Blanchard et al.,

2004a). This assay is based on the reaction of puromycin

with a fluorophore-labeled peptide linked to the CCA

terminus of tRNA followed by rapid dissociation of the

(B) One-dimensional histograms of the population data were fit to the sum (black line) of four single Gaussian distributions (red) to estimate the mean

values and relative occupancies of each FRET state. Histograms overlaid in blue represent simulated data.

(C) Initial and final FRET values for each transition are summed into two-dimensional histograms and show that transitions occur among distinct FRET

states.

(D) Histograms generated by the simulation of single-molecule FRET data.

r Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 509

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Figure 3. Mutation of A, P, and E Sites Reveals the Physical Nature of tRNA Binding Sites

Data are displayed as in Figure 2. (Far left) L1-depleted complexes, (center left) wild-type complexes, (center right) G2553C-mutated complexes, (far

right) G2252C-mutated complexes.

510 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

puromycin dye product from the ribosome. The model of

the physical nature of hybrid states-1 and -2 predicts

that the occupancy of hybrid state-1 should correlate

with increased puromycin reactivity due to the placement

of the peptide in the P site. Complexes that favor either the

classical or the hybrid state-2 configuration should have

greatly diminished puromycin reactivity due to the pep-

tide’s location within the A site.

Control experiments show that, in the absence of an A

site tRNA, stopped-flow delivery of 2 mM puromycin to

wild-type ribosome complexes results in a rapid decrease

of fluorescence in the image area. Complexes bearing the

A loop mutation, G2553C, react at a similar rate, indicating

that under these conditions the binding of puromycin to

the A site is not significantly altered. Complexes prepared

with the P loop mutation, G2252C, react considerably

slower than both the wild-type and A loop mutant com-

plexes, most likely reflecting poor substrate positioning

for the nucleophilic addition reaction (Figure S3).

The puromycin reactivities of wild-type and mutant pre-

translocation complexes are shown in Figure 4. As ex-

pected, wild-type complexes react slowly with puromycin,

consistent with the predominant residence of peptidyl-

tRNA in an unreactive classical state (A/A) and short-lived

transitions to a puromycin-reactive hybrid state configura-

Figure 4. Puromycin Reactivity Confirms the Assignment of

FRET States to Distinct tRNA Configurations on the Ribosome

A single-molecule puromycin assay was used to report on the position

of the peptide within wild-type, G2252C-mutated, and G2553C-mu-

tated ribosome complexes bearing Cy3-Met-tRNAfMet in the A site. Re-

action with puromycin correlates with the occupancy of an A/P hybrid

state. Puromycin (2 mM) was stopped-flow delivered to surface-immo-

bilized complexes, and reaction progress was followed as the loss of

fluorescence over time. After consideration of the rates of Cy3 photo-

bleaching, and peptidyl-tRNA drop off from the A site (see discussion

in Figure S3), the reactivities of the three complexes were the following:

wild-type (red squares), k1 = 0.081/s; G2252C (blue circles), k1 = 0.036/

s; and G2553C (green triangles), k1 = 0.53/s. In agreement with the pre-

dicted positions of the peptide moiety, the rates of puromycin reaction

are faster for A loop mutant ribosomes and slower for P loop mutant

ribosomes than observed for wild-type pretranslocation complexes.

Molecula

tion (A/P) (Semenkov et al., 1992; Sharma et al., 2004;

Youngman et al., 2004; Blanchard et al., 2004a). Corre-

spondingly, ribosome complexes bearing the A loop mu-

tation show an increase in the rate of puromycin reactivity

compared with the wild-type complex that is consistent

with increased occupancy in hybrid state-1 wherein pep-

tidyl-tRNA occupies an A/P hybrid state. P loop mutant

complexes show a further decrease in puromycin reactiv-

ity consistent with the assignment of hybrid state-2

wherein peptidyl-tRNA predominantly occupies the un-

reactive classical state (A/A). These data strongly support

our initial assignment of hybrid states-1 and -2 as A/P-P/E

and A/A-P/E tRNA configurations, respectively.

DISCUSSION

Here we demonstrate the power of single-molecule FRET

methods to provide insights into dynamic structural

processes related to the remodeling of ribosome-tRNA in-

teractions and the nature of hybrid tRNA configurations.

Kinetic and structural analysis of the motions of fluores-

cently labeled tRNA on the ribosome establishes the exis-

tence of two structurally and kinetically distinct hybrid

tRNA configurations. The existence of hybrid states was

first evidenced in chemical probing experiments (Moazed

and Noller, 1989) and later characterized kinetically

through single-molecule FRET experiments (Blanchard

et al., 2004a). However, the assignment of two distinct hy-

brid states as reported here was not previously possible

due to their transient nature and the time scales with which

they had been probed.

Figure 5. Equilibrium Model of tRNA Motions on the Ribo-

some

The two-subunit ribosome (yellow) is shown bound to tRNA (blue) and

mRNA (green). The three states observed, classified as classical,

hybrid state-1, and hybrid state-2, are defined by the motions of

tRNA acceptor stems with respect to the large subunit. The apparent

rates of transition between states are identified as tabulated in Table

1 and Table S3. Conformational changes of the ribosome that may

be important to tRNA motions, as described in the text, are shown

as changes in A, P, and E site color.

r Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 511

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Table 1. Kinetics Analysis of Single-Molecule FRET Trajectories Reveals the Effects of Peptide Length andModification of tRNA Binding Sites

Apparent Transition Rates

kC/H1 kH1/C kH1/H2 kH2/H1 kC/H2 kH2/C

Wild-Typea

fMet-Phe 1.58 ± 0.19 5.41 ± 0.31 3.61 ± 0.26 2.84 ± 0.21 1.28 ± 0.14 4.56 ± 0.32

Met-Phe 1.40 ± 0.14 6.65 ± 0.43 5.10 ± 0.33 3.89 ± 0.30 1.09 ± 0.17 4.42 ± 0.37

Phe 1.10 ± 0.10 6.29 ± 0.33 3.71 ± 0.30 2.69 ± 0.24 0.66 ± 0.08 4.49 ± 0.30

Laboratory Strains

fMet-Pheb 1.82 ± 0.13 5.22 ± 0.31 3.78 ± 0.27 3.00 ± 0.24 1.39 ± 0.15 3.17 ± 0.25

DL1c,d 1.26 ± 0.12 7.07 ± 0.40 4.16 ± 0.38 2.57 ± 0.24 0.78 ± 0.08 4.83 ± 0.40

G2252Cb,d 4.27 ± 0.61 3.16 ± 0.39 3.51 ± 0.43 3.98 ± 0.35 3.33 ± 0.69 1.85 ± 0.40

G2553Cb,d 6.20 ± 0.68 3.64 ± 0.49 3.07 ± 0.48 4.13 ± 0.52 2.32 ± 0.49 2.15 ± 0.40

Rate constants for transitions between each state shown in Figure 5 were calculated by averaging those derived from maximum-likelihood optimization of each trace in QuB. Data from wild-type complexes are labeled according to the peptide or amino acid on

the A site tRNA. Rate constants are given in units of s�1 and are represented as an average ± standard error.a Ribosomes isolated from E. coli MRE600.b Ribosomes isolated from E. coli DH10 as described (Dorner et al., 2006).c Ribosomes isolated from E. coli BL21(DE3).d Contains fMet-Phe-tRNAPhe in the A site.

Given the correlation time of the 25 kDa tRNA molecule

(<<1 ms at 25�C; Komoroski and Allerhand, 1972), simple

thermal fluctuations in tRNA position on a static ribosome

cannot explain the relatively slow time scales of classical

and hybrid states-1 and -2 interconversion (�1–10/s).

The activation energies for classical-hybrid transitions

are similar in magnitude to complete translocation cata-

lyzed by EF-G and GTP hydrolysis (�67 kJ/mol) yet signif-

icantly lower than that required for spontaneous translo-

cation (�96 kJ/mol) (Semenkov et al., 2000; Katunin

et al., 2002; Studer et al., 2003). This suggests that hy-

brid states formation is accompanied by large-scale

rearrangements in ribosome conformation that are inde-

pendent of EF-G binding. A key finding of the present

work is that A and P site tRNA motions on the ribosome

are often uncoupled, which gives rise to a hybrid state in

which only the deacylated P site tRNA enters a hybrid

state. Interestingly, this new hybrid state is �20% more

stable than the established A/P-P/E hybrid state.

Dispersed Kinetics Are Observed in tRNA

Motions on the Ribosome

Visual inspection of the single-molecule FRET trajectories

reporting on tRNA position within the ribosome reveals

a strikingly broad range of kinetic behaviors. This hetero-

geneity is inherent to the stochastic nature of tRNA mo-

tions but may also relate to a physical heterogeneity in ri-

bosome composition and/or structure. Direct observation

of such dispersed kinetics in biological systems is a pro-

found and unique benefit of single-molecule experimenta-

tion (Xie, 2002) that may be used in the classification of

specific behaviors of subpopulations in an ensemble. A

complete discussion of the heterogeneous behavior we

512 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Else

observe is beyond the scope of this manuscript and will

be discussed elsewhere. Here, we provide a framework

for describing tRNA and ribosome dynamics based on

average rate constants obtained for the specific systems

investigated. The validity of this approach is demonstrated

by the ability to recapitulate the population behavior of

each ribosome complex investigated through simulation

of single-molecule trajectories (Figures 2 and 3).

These average rate constants represent a lower bound

on the actual kinetics of the system (Table S3). To arrive at

a more accurate kinetic representation of each system in-

vestigated, the subset of molecules that photobleached

prior to visiting all states in the model was eliminated

from consideration. Even after disregarding these mole-

cules (�25%–60%), the distribution of kinetic behaviors

remains broad. Nonetheless, simulated data generated

from average rate constants recapitulates the experimen-

tal data remarkably well (Figures S4 and S5). The mean ki-

netic values derived in this analysis (Table 1) represent an

upper bound on the kinetic parameters of each system.

Hybrid States Arise from an Induced-Fit Mechanism

If tRNAs moved on a static ribosome, modifications to one

tRNA binding site would have no effect on the stability of

other sites. We find that the specific modifications studied

here affect the rates of multiple kinetic parameters. These

observations can be explained by two mechanisms: (1)

mutation of the ribosome allosterically alters the ground

state energies of all ribosome conformations, and (2) tran-

sitions between classical and hybrid states are mediated

by sampling events faster than the time resolution of our

measurements. Based on previously established transla-

tion mechanisms (Pape et al., 1999; Blanchard et al.,

vier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

2004b; Schmeing et al., 2005), we posit that fast sampling

of states by tRNA, consistent with an induced-fit mecha-

nism, parsimoniously explains much of the kinetic behav-

ior of the ribosome complexes investigated. Fast fluctua-

tions in tRNA motions that report on the induced-fit

mechanism will require faster optical detection methods

using avalanche photodiodes.

Depletion of L1 from the Ribosome Stabilizes

the Classical State

The occupation of deacylated-tRNA in the P/E state is ac-

companied by interactions with components of the large

subunit E site, in particular with ribosomal protein L1 (Valle

et al., 2003a). The rate of translocation is also slowed in ri-

bosomes lacking the L1 protein (Subramanian and Dabbs,

1980). We observe that both hybrid tRNA configurations,

wherein deacylated-tRNA occupies the large subunit E

site, are destabilized in ribosomes lacking L1. This desta-

bilization manifests as an �25% increase in the rates

returning to the classical state from both hybrid state-1

(kH1/C) and hybrid state-2 (kH2/C) with respect to the

wild-type complex (Table 1, Figure 3). However, L1-de-

pleted ribosomes also show an�45% decrease in the ap-

parent rates of exit from the classical state to both hybrid

states. These data are explained if deacylated-tRNA rap-

idly samples the E site of the large subunit much faster

than the time resolution of our data to induce structural

transitions within the ribosome that lead to its capture at

the E site. While it remains formally possible that the L1

protein takes hold of deacylated-tRNA in the P site and

pulls it into the E site (Valle et al., 2003a), we believe that

it is more plausible that the L1 protein indirectly facilitates

the structural transitions of the ribosome and the observed

hybrid states by increasing the affinity of deacylated-tRNA

for the E site.

Point Mutations in the A Loop or P Loop

Destabilize the Classical State

Ribosome complexes bearing the P loop mutation,

G2252C, transition out of the classical state to both hybrid

state-1 (kC/H1) and hybrid state-2 (kC/H2) more than 2-

fold faster than wild-type ribosome complexes (Table 1).

The P loop mutation also increases by �50% the rate at

which peptidyl-tRNA leaves its hybrid configuration re-

turning to the classical state (kH1/H2) (Table S3). These

observations are consistent with the G2252C mutation

disrupting base-pairing interactions with the CCA end of

tRNA in both the P/P and A/P states (Moazed and Noller,

1989; Samaha et al., 1995). Similarly, the A loop mutation,

G2553C, increases the rates of transition from classical to

hybrid state-1 (kC/H1), and from hybrid state-2 to hybrid

state-1 (kH2/H1). These observations are congruous with

the G2553-C75 base pair playing a key role in maintaining

peptidyl-tRNA in the A site (Kim and Green, 1999). The ob-

servation that transitions back to both mutated binding

sites are also decreased by �2.5-fold (kH1/C and kH2/C)

can be adequately explained by an induced-fit regime.

Molecu

Ribosome Unlocking and Locking Rates

Can Be Estimated from tRNA Motions

The observation of independent A and P site tRNA mo-

tions allows us to determine that the rate at which deacy-

lated-tRNA adopts a P/E hybrid state is rate limiting in the

classical-hybrid equilibrium. This result is consistent with

the steric constraints of the system. By analogy to struc-

tural differences observed in ribosomes carrying only

deacylated-tRNA in either P/P or P/E configurations

(Agrawal et al., 1999b; Frank and Agrawal, 2000; Valle

et al., 2003a), we posit that the observed rate of this key

structural transition, defined by the sum of rates forming

hybrid states-1 and -2 (kC/H1+kC/H2), reports directly

on the unlocking mechanism. This aggregate rate

(�2.86/s) suggests that the unlocking mechanism has a

minimum activation energy (DGz = �RT*ln[h*(kC/H1 +

kC/H2)/kBT]) of �70.4 kJ/mol (Table 2). This rate is ap-

proximately two times faster than the observed rate of

translocation under similar buffer conditions (Studer

et al., 2003).

An estimation of the rate of locking is derived in the Ex-

perimental Procedures. The locking mechanism has lower

activation energy (�68.8 kJ/mol) and is therefore relatively

fast (�5.4/s; Table 2) by comparison, giving rise to the ob-

servation that tRNAs spend a majority of time (�60%) in

the classical (locked) configuration (Figure 2, Table S1).

Consistent with previous reports, the growing peptide

increases the rate of fluctuations to hybrid states (Table

1, Figure 2) by lowering the activation energies for hybrid

Table 2. Peptide Length and Modification of the tRNABinding Sites Influence the Rates of Ribosomal Lockingand Unlocking

Rates of Putative Ribosome Conformational Changes

kunlocking klocking kA/A kA/P

Wild-Typea

fMet-Phe 2.86 5.37 9.02 2.31

Met-Phe 2.49 5.82 11.8 2.36

Phe 1.76 5.76 10.0 1.79

Laboratory Strains

fMet-Pheb 3.21 4.50 9.00 2.61

DL1c,d 2.04 6.25 11.2 1.91

G2252Cb,d 7.60 2.93 6.67 4.51

G2553Cb,d 8.52 3.44 6.71 5.50

The rate of unlocking was calculated from kC/H1 + kC/H2.

klocking is as described in the Experimental Procedures. The

rates kA/A and kA/P define the motions of peptidyl-tRNA in

the A site. kA/A is defined by kH1/H2 + kH1/C, and kA/P is de-rived in Table S1. Rates are in units of s�1.a Ribosomes isolated from E. coli MRE600.b Ribosomes isolated from E. coli DH10 as described (Dorner

et al., 2006).c Ribosomes isolated from E. coli BL21(DE3).d Contains fMet-Phe-tRNAPhe in the A site.

lar Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 513

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

states formation (Blanchard et al., 2004a). However, we

now observe that the dominant effect of the growing pep-

tide is to increase by �30% the rate at which deacylated-

tRNA transitions to its hybrid configuration (kC/H2). The

rate at which peptidyl-tRNA exits the classical state is

not significantly affected. Thus, interactions between the

A site peptide and the large subunit appear to promote un-

locking through either a steric or an allosteric mechanism.

The increase in the rate of peptidyl-tRNA transitions both

into and out of the A/P state (kH2/H1 and kH1/H2), ob-

served specifically in the case of Met-Phe-tRNAPhe, sug-

gests that formylation of the peptide screens charge-

charge interactions between the growing peptide and

the exit tunnel that affect peptidyl-tRNA motions.

Deletion of L1 slows the rate of unlocking and acceler-

ates the rate of locking consistent with L1 stabilizing hybrid

state intermediates important for translocation. Both A and

P loop mutant ribosomes show an increase in the rate of

unlocking and a decrease in the rate of locking (Table 2).

Previous studies have shown that the A loop mutation ac-

celerates the rate of translocation while the P loop muta-

tion slows translocation (Dorner et al., 2006). These data

suggest that unlocking is necessary, but not sufficient,

for rapid translocation. Instead, the translocation process

appears most sensitive to peptidyl-tRNA adopting its hy-

brid configuration (Noller et al., 2002; Dorner et al., 2006).

Hybrid State-1, the Presumed Translocation

Intermediate, Is Formed by Coupled Conformational

Changes of the Ribosome

Approximately 45% of the time (kC/H2/[kC/H1 + kC/H2]),

the motion of deacylated-tRNA to its hybrid P/E configura-

tion is uncoupled from movement of A site tRNA, as is re-

flected in the transition density plots in Figures 2 and 3.

The isolated transition of deacylated-tRNA out of the large

subunit P site is coupled to a large activation energy of

�72.4 kJ/mol (determined by kC/H2). The energetic bar-

rier for this transition is lowered by the P loop mutation

(DDGz z �2.4 kJ/mol) consistent with the disruption of

C74-G2252 base pair (DGz of a single G-C pair is �2–10

kJ/mol). The activation energy of the P/P to P/E transition

may be directly correlated with ribosome unlocking that

entails rotation of the small subunit head domain toward

the E site (Agrawal et al., 1999b; Frank and Agrawal,

2000; Valle et al., 2003a). This conformational change is

expected to reconfigure interactions with P site tRNA at

the decoding site to allow its transition to the hybrid state.

The independent movement of peptidyl-tRNA into its A/

P configuration is also associated with a large activation

energy of �70.4 kJ/mol (determined by kH2/H1) and is

only slightly decreased by mutation of the A loop (DDGzz�0.93 kJ/mol). This suggests that the formation of the pep-

tidyl-tRNA hybrid state is also coupled to conformational

change of the ribosome that is distinct from unlocking.

We speculate that remodeling of the A site finger intersu-

bunit bridge, B1a, located between A and P site tRNA

(Valle et al., 2003a; Schmeing et al., 2005), may be critical

to this conformational transition. The significance of such

514 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Elsev

a structural transition can be tested with mutant ribosomes

bearing A site finger truncations that increase the rate of

translocation and frameshifting (Komoda et al., 2006).

While our data demonstrate that A and P site tRNA mo-

tions can be uncoupled, �55% of the time tRNAs appear

to adopt their hybrid configurations simultaneously. The

energetic barrier for moving both tRNAs to their hybrid

configurations, determined by the rate kC/H1 (DGz z71.8 kJ/mol), is significantly lower than the sum of ener-

gies calculated for their independent movements. Simul-

taneous motions of tRNA and the coupling of ribosome

conformational changes would be favorable for rapid

translation. Models proposing the independent motion

of deacylated-tRNA in translocation have been recently

reported (Pan et al., 2006), and it will be important to inves-

tigate whether the ribosome’s capacity to control the

motions of deacylated- and peptidyl-tRNA independently

has mechanistic and/or regulatory significance.

An Initiation to Elongation Transition May Be

Established by Initiator tRNA

The observed rates of ribosome unlocking in vitro are ap-

proximately two times faster than the rate of translocation

measured under similar conditions (Studer et al., 2003);

however, they are approximately 1.5 to five times slower

than estimated translation elongation rates in vivo (ap-

proximately five to 15 amino acids per second; Parker

and Friesen, 1980; Neidhardt, 1987; Pavlov and Ehren-

berg, 1996). While this discrepancy may relate to the lower

temperature and higher magnesium ion concentration

used in our present analysis, it may also be attributed to

the unique nature of the initiation complex as initiator

tRNAfMet translocates out of the P site �3.5 times slower

than elongator tRNAMet (Studer et al., 2003).

Transition of tRNAfMet to the P/E state may be slowed

by the interactions of tRNAfMet’s G-C-rich identity element

that enhance its affinity for the P site (Seong and RajBhan-

dary, 1987). A functional role of the A site peptide at this

stage of translation is supported by the fact that E. coli

open reading frames show a bias toward specific amino

acids in the +1 position (Sato et al., 2001). Moreover,

in vitro polysome studies show an increased spacing of

ribosomes on mRNA consistent with early elongation pro-

cesses being slow relative to later events (Underwood

et al., 2005). Further experiments will be necessary to ex-

plore the structural and functional implications of the

determinants of hybrid states formation and their role in

promoting translocation. It will also be important to inves-

tigate if the rates of hybrid states formation are signifi-

cantly different for elongator tRNAs.

The present study shows that early elongation events in

translation bear striking similarity with the well-character-

ized initiation to elongation transition in transcription (von

Hippel, 1998; Yin and Steitz, 2002; reviewed by Boeger

et al. [2005]). We therefore posit that an interplay between

the initiator tRNA, the growing peptide, and the ribosome

may play an important role in early elongation that pre-

vents peptidyl-tRNA dissociation (Semenkov et al.,

ier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

2000). Future single-molecule investigations of ribosome

dynamics as they relate to tRNA motions will benefit

greatly from direct labeling of the ribosome and translation

factors that will allow simultaneous measurements of ribo-

some conformational change and tRNA motions.

EXPERIMENTAL PROCEDURES

Preparation of Ribosome Complexes, tRNA,

and Translation Factors

Wild-type and mutant tight-coupled 70S ribosomes were purified from

E. coli MRE600, BL21(DE3), and DH10 and used to prepare initiation

complexes following procedures described in the Supplemental Ex-

perimental Procedures or as previously described (Blanchard et al.,

2004a, 2004b; Dorner et al., 2006). The formation of pretranslocation

complexes is described in the Supplemental Data.

Single-Molecule Fluorescence Experiments

Data were acquired using a lab-built prism-based TIR fluorescence

microscope. All single-molecule experiments were performed in Tris-

polymix buffer (pH 7.5), 15 mM MgOAc, in an oxygen-scavenging en-

vironment optimized for the extension of dye lifetime and limited blink-

ing (1 unit/mL glucose oxidase, 10 units/mL catalase, 0.1% v/v glucose).

The Cy3 fluorophore was directly excited by using a Ventus 532 nm

laser (Laser Quanta), and data were collected using a 1.2 NA 603 wa-

ter-immersion objective (Nikon) at 25 frames per second with a Photo-

metrics Cascade 512B CCD camera (Roper Scientific). Data were ac-

quired by using MetaMorph software (Universal Imaging Corporation).

Kinetic Analysis of Single-Molecule FRET Data

Single-molecule fluorescence trajectories were identified and ex-

tracted from multiple wide-field movies by custom-made MATLAB

programs according to four criteria: (1) the mean total fluorescence in-

tensity (ICy5 + ICy3) exceeded a predetermined threshold, (2) Cy3 fluo-

rescence lifetime exceeded 1 s, (3) FRET efficiency exceeded 0.1, and

(4) fluorescence correlation coefficient fell outside the interval [�0.2,

0.2]. The histograms in all figures are composed of the first 2 s of

each FRET trace. The first 400 frames (16 s) of each FRET trace

were idealized to Markov chain models by using a segmental k-means

algorithm in QuB (Qin, 2004). The FRET amplitudes, standard devia-

tions, and initial probabilities of the model states were obtained by

analysis of the one-dimensional histograms in Figures 2B and 3B. A

small population of traces (�5%), poorly idealized due to drift and/or

spurious noise, was eliminated from further analysis.

Markov chains were fit to the sequence of dwell times by a maxi-

mum-likelihood estimation procedure (Qin et al., 1996, 1997). Traces

from wild-type MRE600 ribosome complexes were fit to two different

Markov models: one containing a single hybrid state and the second

containing both hybrid state-1 and hybrid state-2. The sum of the max-

imized log-likelihood per transition for all traces was corrected for the

different number of parameters by the Akaike information criterion

(AIC) (Akaike, 1974). The relative AIC values, 869.2 for the model

with two hybrid states and 1580.2 for the model with one hybrid state,

confirmed that the data are better fit by the model with two hybrid

states. The optimized kinetic parameters from each trace were aver-

aged to form a single model, and single-molecule FRET traces were

simulated in QuB.

To verify that hybrid states-1 and -2 were distinct FRET states and

not a continuum of FRET values, transition density plots were con-

structed from the idealized data. These plots were constructed by

compiling a two-dimensional histogram of FRET values before and

after each transition (McKinney et al., 2006). The resulting peaks

shown in Figures 2 and 3 and Figures S4 and S5 indicate the relative

frequency of each transition and verify that hybrid states-1 and -2

are distinct FRET states.

Molecul

Derivation of klocking and kA/P

The complete five-state model of tRNA dynamics on the ribosome is

shown in Figure S7, which includes the three states shown in Figure 5

as well as the photobleached (PB) and blinking (B) states. For both

klocking and kA/P, we seek to calculate the rate of transition from two

states in dynamic exchange into a single state. The process for deter-

mining such a rate is outlined below for klocking and is identical for kA/P.

Both are special cases of the more general procedure demonstrated

by Colquhoun and Hawkes (1981). Due to the minimal blinking rate,

we have neglected to exchange between H1 and H2 and B.

We seek to determine the probability density function (PDF) describ-

ing transitions out of a set of two connected states (H1 and H2) into

a single state (C),

fH/CðtÞ= w1expð�k1tÞ+ w2expð�k2tÞ;

where w1 and w2 are normalization constants and k1 and k2 are time

constants determined below. The master equation expressing the

probability Pi/C(t) of the system being found in the classical state at

time t, given that it began in H1 or H2 at t = 0 in terms of the transition

rates between states is

dPi/CðtÞdt

= Pi/H1ðtÞkH1/C + Pi/H2ðtÞkH2/C

where i is either H1 or H2. The quantity dPi/C(t)/dt is the probability

density of transitions from i to C. We therefore need to obtain the tran-

sition probabilities Pi/H1(t) and Pi/H2(t). This is most easily done by

combining these probabilities into the 2 3 2 matrix PHH and noting

that they satisfy analogous master equations, which are expressed

in matrix notation as

dPHHðtÞdt

= PHHðtÞKHH (1)

where KHH is the matrixof transition rates betweenhybridstates-1and -2,

KHH =

�kH1/H1 kH1/H2

kH2/H1 kH2/H2

�;

which is a submatrix of the complete 5 3 5 matrix of transition rates, K.

The diagonal elements of K (and of KHH) are defined such that the rows

of K sum to zero. Using standard matrix methods, PHH(t) may be ob-

tained from the spectral decomposition of KHH. For this, we calculate

the eigenvalues l1,2 and eigenvectors v1,2 of KHH, which are readily

obtained analytically, or numerically in MATLAB (for example). We

can then express the solutions to (Equation 1) as

PHHðtÞ= A1expðl1tÞ+ A2expðl2tÞ

where A1,2 are obtained from the eigenvectors of KHH. These equations

are then inserted into dPi/C(t)/dt. As there are two hybrid states in our

model (i.e., i = H1,H2),

fH/CðtÞ= pH1

dPH1/CðtÞdt

+ pH2

dPH2/CðtÞdt

;

where pH1 and pH2 are the relative probabilities of the system begin-

ning in H1 and H2 at t = 0, respectively. Grouping terms with like expo-

nentials then defines the weighting constants w1 and w2 and makes it

clear that l1 = k1 and l2 = k2.

We can then define klocking as the average rate of transition from the

hybrid states to the classical state

1

klocking

hhti=ðN

0

tfH/CðtÞdt

so that

klocking =l2

1l22

w1l22 + w2l2

1

:

An analogous expresssion is defined for kA/P.

ar Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 515

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Supplemental Data

Supplemental Data include four tables, seven figures, Supplemental

Experimental Procedures, and Supplemental References and can be

found with this article online at http://www.molecule.org/cgi/content/

full/25/4/505/DC1/.

ACKNOWLEDGMENTS

We gratefully thank Dr. Rachel Green and her laboratory (Howard

Hughes Medical Institute, Johns Hopkins University) for providing

ribosomes bearing A and P loop mutations and for comments on the

manuscript. We also thank Kevin Sanbonmatsu (Los Alamos National

Laboratory) for his helpful comments and discussions. We would also

like to acknowledge Anton Rosenbaum for his preparation of Cy3-

Met-tRNAfMet. This work was supported by the National Institute of

General Medical Sciences (1R01GM079238-01) and the Alice Bohm-

falk Charitable Trust.

Received: October 27, 2006

Revised: December 14, 2006

Accepted: January 22, 2007

Published: February 22, 2007

REFERENCES

Agrawal, R.K., Penczek, P., Grassucci, R.A., Burkhardt, N., Nierhaus,

K.H., and Frank, J. (1999a). Effect of buffer conditions on the position

of tRNA on the 70S ribosome as visualized by cryoelectron micros-

copy. J. Biol. Chem. 274, 8723–8729.

Agrawal, R.K., Heagle, A.B., Penczek, P., Grassucci, R.A., and Frank,

J. (1999b). EF-G-dependent GTP hydrolysis induces translocation

accompanied by large conformational changes in the 70S ribosome.

Nat. Struct. Biol. 6, 643–647.

Akaike, H. (1974). A new look at statistical model identification. IEEE

Trans. Auto. Cont. 19, 716–723.

Blaha, G., Burkhardt, N., and Nierhaus, K. (2002). Formation of 70S

ribosomes: large activation energy is required for the adaptation of

exclusively the small ribosomal subunit. Biophys. Chem. 96, 153–161.

Blanchard, S.C., Kim, H.D., Gonzalez, R.L., Puglisi, J.D., and Chu, S.

(2004a). tRNA dynamics on the ribosome during translation. Proc.

Natl. Acad. Sci. USA 101, 12893–12898.

Blanchard, S.C., Gonzalez, R.L., Kim, H.D., Chu, S., and Puglisi, J.D.

(2004b). tRNA selection and kinetic proofreading in translation. Nat.

Struct. Mol. Biol. 11, 1008–1014.

Boeger, H., Bushnell, D.A., Davis, R., Griesenbeck, J., Lorch, Y., Strat-

tan, J.S., Westover, K.D., and Kornberg, R.D. (2005). Structural basis

of eukaryotic gene transcription. FEBS Lett. 579, 899–903.

Bretscher, M.S. (1968). Translocation in protein synthesis: a hybrid

structure model. Nature 218, 675–677.

Colquhoun, D., and Hawkes, A.G. (1981). On the stochastic properties

of single ion channels. Proc. R. Soc. Lond. B. Biol. Sci. 211, 205–235.

Crick, F.H.C. (1958). On protein synthesis. Symp. Soc. Exp. Biol. 12,

138–163.

Dorner, S., Brunelle, J.L., Sharma, D., and Green, R. (2006). The hybrid

state of tRNA binding is an authentic translation elongation intermedi-

ate. Nat. Struct. Mol. Biol. 13, 234–241.

Frank, J., and Agrawal, R.K. (2000). A ratchet-like inter-subunit reorga-

nization of the ribosome during translocation. Nature 406, 318–322.

Frauenfelder, H., Sligar, S.G., and Wolynes, P.G. (1991). The energy

landscapes and motions of proteins. Science 254, 1598–1603.

Gabashvili, I.S., Agrawal, R.K., Grassucci, R., Squires, C.L., Dahlberg,

A.E., and Frank, J. (1999). Major rearrangements in the 70S ribosomal

3D structure caused by a conformational switch in 16S ribosomal RNA.

EMBO J. 18, 6501–6507.

516 Molecular Cell 25, 505–517, February 23, 2007 ª2007 Elsev

Herring, C.D., Glasner, J.D., and Blattner, F.R. (2003). Gene replace-

ment without selection: regulated suppression of amber mutations in

Escherichia coli. Gene 331, 153–163.

Katunin, V.I., Savelsbergh, A., Rodnina, M.V., and Wintermeyer, W.

(2002). Coupling of GTP hydrolysis by elongation factor G to transloca-

tion and factor recycling on the Ribosome. Biochemistry 41, 12806–

12812.

Kim, D.F., and Green, R. (1999). Base-pairing between 23S rRNA and

tRNA in the ribosomal A site. Mol. Cell 4, 859–864.

Komoda, T., Sato, N.S., Phelps, S.S., Namba, N., Joseph, S., and

Suzuki, T. (2006). The A site finger in 23S rRNA acts as a functional

attenuator for translocation. J. Biol. Chem. 281, 32303–32309.

Komoroski, R.A., and Allerhand, A. (1972). Natural-abundance carbon-

13 Fourier-transform nuclear magnetic resonance spectra and spin

lattice relaxation times of unfractionated yeast transfer-RNA. Proc.

Natl. Acad. Sci. USA 69, 1804–1808.

Korostelev, A., Trakhanov, S., Laurberg, M., and Noller, H.F. (2006).

Crystal structure of a 70S ribosome-tRNA complex reveals functional

interactions and rearrangements. Cell 126, 1065–1077.

McKinney, S.A., Joo, C., and Ha, T. (2006). Analysis of single-molecule

FRET trajectories using hidden Markov modeling. Biophys. J. 91,

1941–1951.

Milescu, L.S., Yildiz, A., Selvin, P.R., and Sachs, F. (2006). Maximum

likelihood estimation of molecular motor kinetics from staircase

dwell-time sequences. Biophys. J. 91, 1156–1168.

Moazed, D., and Noller, H.F. (1989). Intermediate states in the move-

ment of transfer RNA in the ribosome. Nature 342, 142–148.

Moore, P.B., and Steitz, T.A. (2003). The structural basis of large ribo-

somal subunit function. Annu. Rev. Biochem. 72, 813–850.

Muth, G.W., Chen, L., Kosek, A.B., and Strobel, S.A. (2001). PH-de-

pendent conformational flexibility within the ribosomal peptidyl trans-

ferase center. RNA 7, 1403–1415.

Namy, O., Moran, S.J., Stuart, D.I., Gilbert, R.J.C., and Brierley, I.

(2006). A mechanical explanation of RNA pseudoknot function in pro-

grammed ribosomal frameshifting. Nature 441, 244–247.

Neidhardt, F.C. (1987). Escherichia coli and Salmonella typhimurium:

Cellular and Molecular Biology (Washington, D.C.: American Society

for Microbiology).

Noller, H.F., Yusupov, M.M., Yusupova, G.Z., Baucom, A., and Cate,

J.H.D. (2002). Translocation of tRNA during protein synthesis. FEBS

Lett. 514, 11–16.

Ogle, J.M., and Ramakrishnan, V. (2005). Structural insights into trans-

lational fidelity. Annu. Rev. Biochem. 74, 129–177.

Ogle, J.M., Brodersen, D.E., Clemons, W.M., Tarry, M.J., Carter, A.P.,

and Ramakrishnan, V. (2001). Recognition of cognate transfer RNA by

the 30S ribosomal subunit. Science 292, 897–902.

Pan, D., Kirillov, S., Zhang, C., Hou, Y., and Cooperman, B.S. (2006).

Rapid ribosomal translocation depends on the conserved 18-55

base pair in P-site transfer RNA. Nat. Struct. Mol. Biol. 13, 354–359.

Pape, T., Wintermeyer, W., and Rodnina, M. (1999). Induced fit in initial

selection and proofreading of aminoacyl-tRNA on the ribosome.

EMBO J. 18, 3800–3807.

Parker, J., and Friesen, J.D. (1980). ‘‘Two out of three’’ codon reading

leading to mistranslation in vivo. Mol. Gen. Genet. 177, 439–445.

Pavlov, M.Y., and Ehrenberg, M. (1996). Rate of translation of natural

mRNAs in an optimized in vitro system. Arch. Biochem. Biophys.

328, 9–16.

Qin, F. (2004). Restoration of single-channel currents using the seg-

mental k-means method based on hidden Markov modeling. Biophys.

J. 86, 1488–1501.

ier Inc.

Molecular Cell

Two Hybrid State Intermediates on the Ribosome

Qin, F., Auerbach, A., and Sachs, F. (1996). Estimating single-channel

kinetic parameters from idealized patch-clamp data containing missed

events. Biophys. J. 70, 264–280.

Qin, F., Auerbach, A., and Sachs, F. (1997). Maximum likelihood

estmation of aggregated Markov processes. Proc. R. Soc. Lond. B.

Biol. Sci. 264, 375–383.

Ridge, K.D., Abdulaev, N.G., Zhang, C., Ngo, T., Brabazon, D.M., and

Marino, J.P. (2006). Conformational changes associated with recep-

tor-stimulated guanine nucleotide exchange in a heterotrimeric G-pro-

tein a-subunit. J. Biol. Chem. 281, 7635–7648.

Rodnina, M.V., Savelsbergh, A., Katunin, V.I., and Wintermeyer, W.

(1997). Hydrolysis of GTP by elongation factor G drives tRNA move-

ment on the ribosome. Nature 385, 37–41.

Samaha, R.R., Green, R., and Noller, H.F. (1995). A base pair between

tRNA and 23S rRNA in the peptidyl transferase center of the ribosome.

Nature 377, 309–314.

Sato, T., Terabe, M., Watanabe, H., Gojobori, T., Hori-Takemoto, C.,

and Miura, K. (2001). Codon and base biases after the initiation codon

of the open reading frames in the Escherichia coli genome and their

influence on the translation efficiency. J. Biochem. (Tokyo) 129, 851–

860.

Schmeing, T.M., Huang, K.S., Strobel, S.A., and Steitz, T.A. (2005). An

induced-fit mechanism to promote peptide bond formation and ex-

clude hydrolysis of peptidyl-tRNA. Nature 438, 520–524.

Schnell, J.R., Dyson, H.J., and Wright, P.E. (2004). Structure, dynam-

ics, and catalytic function of dihydrofolate reductase. Annu. Rev. Bio-

phys. Biomol. Struct. 33, 119–140.

Schuwirth, B.S., Borovinskaya, M.A., Hau, C.W., Zhang, W., Vila-San-

jurjo, A., Holton, J.M., and Cate, J.H.D. (2005). Structures of the bac-

terial ribosome at 3.5A resolution. Science 310, 827–834.

Selmer, M., Dunham, C.M., Murphy, F.V., Weixlbaumer, A., Petry, S.,

Kelley, A.C., Weir, J.R., and Ramakrishnan, V. (2006). Structure of

the 70S ribosome complexed with mRNA and tRNA. Science 313,

1935–1942.

Semenkov, Y.P., Rodnina, M.V., and Wintermeyer, W. (2000). Ener-

getic contribution of tRNA hybrid state formation to translocation

catalysis on the ribosome. Nat. Struct. Biol. 7, 1027–1031.

Semenkov, Y., Shapkina, T., Makhno, V., and Kirillov, S. (1992). Puro-

mycin reaction for the A site-bound peptidyl-tRNA. FEBS Lett. 296,

207–210.

Seong, B.L., and RajBhandary, U.L. (1987). Escherichia coli formylme-

thionine tRNA: mutations in GGG-CCC sequence conserved in antico-

don stem of initiator tRNAs affect initiation of protein synthesis and

conformation of anticodon loop. Proc. Natl. Acad. Sci. USA 84, 334–

338.

Sharma, D., Southworth, D.R., and Green, R. (2004). EF-G-indepen-

dent reactivity of a pre-translocation-state ribosome complex with

the aminoacyl tRNA substrate puromycin supports an intermediate

(hybrid) state of tRNA binding. RNA 10, 102–113.

Spirin, A.S. (1968). On the mechanism of ribosome function. The hy-

pothesis of locking-unlocking of subparticles. Dokl. Akad. Nauk

SSSR 179, 1467–1470.

Molecula

Studer, S.M., Feinberg, J.S., and Joseph, S. (2003). Rapid kinetic anal-

ysis of EF-G-dependent mRNA translocation in the ribosome. J. Mol.

Biol. 327, 369–381.

Subramanian, A.R., and Dabbs, E.R. (1980). Functional studies on ri-

bosomes lacking protein L1 from mutant Escherichia coli. Eur. J. Bio-

chem. 112, 425–430.

Tama, F., Miyashita, O., and Brooks, C.L. (2004). Normal mode based

flexible fitting of high-resolution structure into low-resolution experi-

mental data from cryo-EM. J. Struct. Biol. 147, 315–326.

Underwood, K.A., Swartz, J.R., and Puglisi, J.D. (2005). Quantitative

polysome analysis identifies limitations in bacterial cell-free protein

synthesis. Biotechnol. Bioeng. 91, 425–435.

Valle, M., Zavialov, A., Sengupta, J., Rawat, U., Ehrenberg, M., and

Frank, J. (2003a). Locking and unlocking of ribosomal motions. Cell

114, 123–134.

Valle, M., Zavialov, A., Li, W., Stagg, S.M., Sengupta, J., Nielsen, R.C.,

Nissen, P., Harvey, S.C., Ehrenberg, M., and Frank, J. (2003b). Incor-

poration of aminoacyl-tRNA into the ribosome as seen by cryo-elec-

tron microscopy. Nat. Struct. Biol. 10, 899–906.

Vila-Sanjurjo, A., Ridgeway, W.K., Seymaner, V., Zhang, W., Santoso,

S., Yu, K., and Cate, J.H.D. (2003). X-ray crystal structures of the WT

and a hyper-accurate ribosome from Escherichia coli. Proc. Natl.

Acad. Sci. USA 100, 8682–8687.

von Hippel, P. (1998). An integrated model of the transcription complex

in elongation, termination, and editing. Science 281, 660–665.

Wang, Y., Rader, A.J., Bahar, I., and Jernigan, R.L. (2004). Global ribo-

some motions revealed with elastic network model. J. Struct. Biol. 147,

302–314.

Wilden, B., Savelsbergh, A., Rodnina, M.V., and Wintermeyer, W.

(2006). Role and timing of GTP binding and hydrolysis during EF-G-de-

pendent tRNA translocation on the ribosome. Proc. Natl. Acad. Sci.

USA 103, 13670–13675.

Wolf-Watz, M., Thai, V., Henzler-Wildman, K., Hadjipavlou, G., Eisen-

messer, E.Z., and Kern, D. (2004). Linkage between dynamics and

catalysis in a thermophilic-mesophilic enzyme pair. Nat. Struct. Mol.

Biol. 11, 945–949.

Xie, X.S. (2002). Single-molecule approach to dispersed kinetics and

dynamic disorder: probing conformational fluctuation and enzymatic

dynamics. J. Chem. Phys. 117, 11024–11032.

Yin, Y.W., and Steitz, T.A. (2002). Structural basis for the transition

from initiation to elongation transcription in T7 RNA polymerase. Sci-

ence 298, 1387–1395.

Yonath, A. (2005). Antibiotics targeting ribosomes: resistance, selec-

tivity, synergism, and cellular regulation. Annu. Rev. Biochem. 74,

649–679.

Youngman, E.M., Brunelle, J.L., Kochaniak, A.B., and Green, R. (2004).

The active site of the ribosome is composed of two layers of conserved

nucleotides with distinct roles in peptide bond formation and peptide

release. Cell 117, 589–599.

r Cell 25, 505–517, February 23, 2007 ª2007 Elsevier Inc. 517