The process of mRNA–tRNA translocation

The Elongation Cycle, and Translocation

In the course of protein synthesis, tRNA occupies successively the universally conserved A (aminoacyl), P (peptidyl), and E (exit) sites of the mRNA-programmed ribosome. The elongation cycle of translation is a repeating, three-step process catalyzed by the ribosome and two GTPases, EF-Tu and EF-G (eEF1A and eEF2 in eukaryotes). First, in decoding, EF-Tu delivers an aminoacylated-tRNA (aa-tRNA) molecule as part of the ternary complex of aa-tRNA·EF-Tu·GTP to the A/T site. The orientation with which tRNA enters the ribosome is not favorable for codon–anticodon interaction, and the anticodon stem-loop of the tRNA needs to be strongly kinked at the A/T site to allow the anticodon to interact with mRNA (24). If the aa-tRNA is cognate—i.e., when its three-base anticodon forms complementary base pairs with the codon in the messenger RNA—GTP is hydrolyzed, and the aa-tRNA, upon dissociation of EF-Tu·GDP, is accommodated in the ribosomal A site. The second step, the peptidyl-transferase reaction, is catalyzed by the rRNA of the large subunit and occurs immediately following the accommodation of the aa-tRNA. The transfer of the nascent peptide chain, from peptidyl-tRNA in the ribosomal P site to the accommodated aa-tRNA in the A site, results in the “pretranslocational” (PRE) state of the translating ribosome, with the deacylated tRNA in the P site and a peptidyl-tRNA in the A site.

The final step of the elongation cycle is translocation, which brings the ribosome into the “posttranslocational” (POST) state, with an empty A site, peptidyl-tRNA in the P site, and deacylated tRNA in the E site. Translocation is catalyzed by the binding of EF-G and subsequent GTP hydrolysis. The movement of the tRNA molecules is concomitant with the movement of the bound mRNA chain by the length of three bases, allowing the next codon of mRNA in the ribosomal A site to be presented for decoding.

Each step of the elongation cycle of translation, namely decoding, peptidyl-transfer, and translocation, presents an enigma on its own. Whereas the mechanism of peptidyl-transfer has received particular attention (see ref. 25), and at least part of the tRNA selection step of decoding has been unraveled in molecular detail (6), the molecular processes of EF-G-mediated translocation have been least understood until recently. Key findings include the discovery of the tRNA hybrid states (26) and the results of kinetic experiments that GTP hydrolysis precedes translocation (27).

Important cryo-EM findings relevant to translocation are the visualization of EF-G binding to the ribosome (11, 12, 14); the visualization of tRNA in the P/E hybrid state (13, 18, 23, 28); the discovery of the ratchet motion, a rotation of the small vs. large subunit upon binding of EF-G (29) or eEF2 (22) to the ribosome; and, most recently, the elucidation of the role of GTP hydrolysis in eEF2-induced translocation of the eukaryotic ribosome (23).

Hybrid-State tRNAs and Intersubunit “Ratchet” Motion

It was originally proposed in the 1960s that translocation of tRNAs might occur independently on the two ribosomal subunits during elongation (30, 31). Later, biochemical data such as the results from direct chemical footprinting led to the hybrid-state tRNA model, which describes an intermediate step in translocation (26). Here, the acceptor ends of the A- and P-site tRNAs move into the P and E sites of the large subunit, while the anticodon stem-loops of the tRNAs remain in the A and P sites of the small subunit. This configuration results in occupation by the tRNAs of the A/P and P/E hybrid states. Cryo-EM reconstructions indeed revealed structural evidence for a position of tRNA in the P/E hybrid state (11, 13, 18, 23); however, the A/P hybrid state has thus far eluded structural analysis using cryo-EM or x-ray crystallography.

The initial report of hybrid-state tRNAs by Moazed and Noller (26) suggested that tRNAs occupy the hybrid state immediately after peptide bond formation, but structures representing the PRE ribosome determined by cryo-EM (13) and x-ray crystallography (32–35) showed that in the absence of elongation factors, the tRNAs occupy the classic A/A, P/P, and E/E sites exclusively. Potential explanations for this discrepancy are that the tRNAs are sampling conformations between classic and hybrid states in the PRE ribosome (36), or that the buffer used in the various studies stabilizes one state of binding over the other (i.e., classic over hybrid) (28).

The P/E hybrid state tRNA in the cryo-EM reconstructions was visualized only when EF-G was bound to the ribosome. In addition to bringing the tRNA into the P/E hybrid state, the binding of the factor also induces a new conformation of the ribosome, where the small subunit is in a different orientation with respect to the large subunit, related to the normal conformation by a counterclockwise rotation (“ratchet motion”) (29, 37) [supporting information (SI) Fig. 6 and SI Movie 1]. Evidence for the existence of an intersubunit motion first came from small-angle neutron scatting experiments, but the precise form of this motion could not be deduced (38). It is likely that the ratchet motion of the ribosome even occurs in the absence of ribosomal factors and that it is coincident with the movement of tRNAs from the classic to the hybrid states, which facilitate translocation (39, 40). Hence, the view is emerging that, as suggested by the cryo-EM data, the binding of EF-G/eEF2 in the GTP form stabilizes the ratcheted conformation, and with it, the hybrid-state tRNAs (40, 41). The ratcheted ribosome with hybrid-state tRNAs, therefore, represents a distinct structural intermediate between the PRE and POST ribosomes. In solution, it is likely that the PRE ribosome oscillates between the ratcheted and nonratcheted states, with EF-G catalyzing translocation to the POST ribosome. The POST ribosome, with peptidyl-tRNA in the P-site, is locked in the nonratcheted state (13).

A convincing proof that the ratchet motion is instrumental for translocation is contained in a series of three recent studies. In one, the motion was inhibited by the placement of a crosslink between proteins of the two subunits facing each other in the normal conformation of the ribosome, and complete suppression of translocation was observed (42). Furthermore, two FRET studies, one bulk (40) and the other single-molecule,^¶ have provided conclusive evidence that the motion inferred from the comparison of cryo-EM maps indeed occurs in a translating ribosome.

It is noteworthy, in this context, that the kinetic parameters of tRNA movement have been investigated in detail with the use of fluorescent labels (43, 44). In addition to supporting the existence of the A/P and P/E hybrid states, these data identify distinct hybrid-state intermediates. The movements of the two tRNAs bound to the A and P sites of the small subunit occur separately, such that A/A tRNA and P/E-hybrid tRNA may exist simultaneously in an intermediate ribosomal state. In the absence of elongation factors, single-molecule FRET measurements suggest that the tRNAs transition between the classic, A/P-P/E hybrid, and A/A-P/E intermediate states but spend the majority of time (≈60%) in the classic states (43).

Factor-Free Translocation

Some of the observations discussed above clearly show that the transition between two dramatically different conformations of the ribosome and, along with it, the transition between canonical and hybrid states of the tRNA occur spontaneously, in the absence of EF-G/eEF2. It is compelling to link the structural and dynamic evidence to the long-known observations of factor-free translocation.

Reports in the very early years of ribosome studies (45–47), when very little was known about its biochemistry and structure, indicated that the ribosome is capable of performing the synthesis of oligo-phenylalanine from a poly(U) template even in the absence of elongation factors and GTP. Subsequent work by Spirin's group (48–50) further indicated that the treatment of the ribosome with pCMB, or the removal of ribosome protein S12 (51), stimulates the rate of factor-free translocation significantly. Recent work from Rachel Green's laboratory (52) identified three target cysteine residues, in protein S12 (C104 and C27) and S13 (C87), whose modification by pCMB results in an elevated factor-free translocation rate, albeit with decreased accuracy (see Fig. 1).

Interestingly, ribosomal protein S12 is the very protein identified in cryo-EM maps as the pivot of the ratchet-like motion (13) and was shown to interact with EF-G both by cross-linking studies (53, 54) and cryo-EM (13). A number of subsequent cryo-EM studies found that other translation factors, including IF2 (10), RF3 (18, 55), and RRF (20), all shown to interact with S12, induce the same type of intersubunit movement as EF-G (Fig. 2 and SI Fig. 7). In fact, the above-mentioned C104 residue of S12 is located exactly at the interface between the 30S subunit and these factors. These cryo-EM observations thus provide a link between factor-free and factor-promoted translocation, where S12 plays an apparent regulatory role in controlling the intersubunit motion of the ribosome (see below for a similar role of S13).

One apparent benefit of introducing protein factors into the elongation cycle, in the course of evolution, is that it helps lower the energetic barriers of the various transition states, because factor-free translation occurs at a much lower rate (45, 48). Translocation itself has been shown to be an exergonic reaction (56, 57), but with a high free energy of activation [ΔG^≠ ≈ 96 kJ/mol for factor-free translocation (58, 59)]. Kinetic studies indicated that the free energy of activation is lowered by ≈18 kJ/mol in the presence of EF-G, and by another 10 kJ/mol with the hydrolysis of GTP, which correspond to an ≈1,000-fold increase in translocation rate and another 50-fold increase, respectively (27, 60).

Architecture of the Ribosome: A Molecule Made to Rock

X-ray studies show that the ribosome possesses a unique architecture with built-in instabilities, among them architectural features conducive to the ratchet motion. As we have seen, these properties are apparently required for a whole range of translational steps, not just for translocation. Among the architectural features giving rise to instability are the following.

1. The ribosome is composed of two relatively loosely coupled subunits, a fact that prompted Spirin (31) quite early on to postulate the existence of intersubunit motion as a necessary part of the translational process.

2. The coupling is such that rotational movement around a central core of connections accompanied by torsional stress of these connections is prevalent among the modes shown by normal mode analysis (61, 62). Specifically, the central bridge B2a is formed by RNA–RNA minor groove interactions (32, 63, 64). According to Valle and coworkers (13), the rotational pivot axis passes through a point between bridges B3 and B5a on h44, just above h27 of 16S rRNA (see also ref. 61).

3. The peripheral bridges (i.e., bridges located far away from the rotational pivot axis) are quite flexible and involve at least one protein [e.g., B4, involving S15 and the 715 stem loop of 23S rRNA (65); for definition of bridge nomenclature, see refs. 9 and 19]. The bridge having to absorb the largest movement, B1b, is formed by two proteins, S13 and L5, which disengage and reengage in different constellations (13). It is interesting to look at this bridge in the two constellations, using the quasi-atomic models obtained by real-space refinement (Fig. 3): in the normal configuration of the ribosome (macrostate I), the electrostatic charges of regions of S13 and L5 facing each other are of opposite polarity, lending stability to the ribosome. In the ratcheted configuration (macrostate II), in contrast, the electrostatic charges of regions facing each other are of equal polarity, conveying instability to the proteins and a propensity for gliding along their interface. By extending the motion further, we would reach a constellation where opposite polarities would again face each other, creating another stabilizing interaction. The distance traveled between the “normal” position (macrostate I) and the postulated second antipolar juxtaposition would be 30 Å, which might act as a yardstick defining the maximum amplitude of “allowed” ratchet motion.

4. There are a large number of tertiary contacts, identified by Noller (66), that are slightly outside of the range of canonical A-minor interactions (67) and hence would allow the gliding of adjacent strands of rRNA without locking them into a defined position.

5. The rRNA possesses a number of kink-turns, a new element discovered in the x-ray structure of the 50S subunit of Haloarcula marismortui (68). Kink-turns are elements with built-in instability, allowing extensive hinge motions of some ribosomal components, among these the L1 stalk, the A-site finger (69), and the GTPase-associated center (GAC). These elements and their biological significance are detailed in the following.

The L1 stalk, which is implicated in the transport of the tRNA, pivots around a hinge region in h76 of 23S rRNA and has been observed in different positions both by cryo-EM (13, 21, 28) and x-ray crystallography (64, 70). The position of this stalk in an “inward” position, found in macrostate II, likely blocks the exit path of tRNA. Conversely, an “outward” position, found in macrostate I, leaves the exit path open. Movement from the “inward” to the “outward” positions of the L1 stalk likely coordinates translocation of tRNAs by allowing ejection of E-site tRNA from the ribosome (see ref. 13).

The A-site finger is a long rRNA helix (h38) reaching from the 23S rRNA into the intersubunit region where it forms a contact (termed B1a) with S13 of the small subunit. In Thermus thermophilus, h38 possesses a kink-turn (Kt-38). In E. coli, it has a similar region of instability. Because the tip of the A-site finger must follow the ratchet motion of the small subunit at a peripheral location, its built-in instability may serve to reduce the activation energy for the ratcheting. Shortening of the helix, presumably leading to a disruption of B1a, results in an acceleration of translocation activity, pointing to a role of the A-site finger as a control element maintaining the reading frame (71).

Finally, the GAC, formed by 23S rRNA h43 and h44, is flexibly linked to the main body of the large subunit by h42, which contains kink-turn 42. According to cryo-EM studies, the mobility of the GAC is essential for factor binding, forming part of an “induced fit” mechanism during the binding of EF-Tu (24, 72), EF-G (11), RF3 (18), and IF2 (10).

Hence, in conclusion, the ribosome's architecture is intrinsically unstable, conducive to making the ribosome alternate between two conformations. In the context of translocation, the first of these conformations (macrostate I) is required for initial factor binding, to initiate the first step of the translocation process, whereas the second (macrostate II) is required for staging the hybrid states, triggering GTP hydrolysis, and disengaging the firm grip that holds the mRNA–tRNA moiety in the decoding center, thereby initiating the second step of the translocation process.

Observed Conformations of EF-G

EF-G-catalyzed translocation requires EF-G to assume at least four conformations: (i) the GTP-bound, ribosome-free form; (ii) the GTP-bound conformation of EF-G on the ribosome immediately before GTP hydrolysis (as binding to the ribosome induces the GTPase of EF-G); (iii) the GDP-bound conformation, in which EF-G remains bound to the ribosome; and (iv) the GDP-bound conformation after dissociation of the factor from the ribosome. All four conformations have been observed structurally—by cryo-EM for the ribosome-bound conformations and by x-ray crystallography for the unbound conformations (SI Fig. 8). Such conformational alterations, specifically a hinge-like motion between the N-terminal domains (I, II, and G′) and C-terminal domains (III–V) of EF-G, undoubtedly play a role in the physiological function of the factor.

In the 1990s, the crystal structures of GDP-bound (73) and nucleotide-free (74) EF-G were solved. Both were shown to have similar conformations, with disordered regions in domain 3 and the switch 1 loop. The ribosome-free structure of EF-G·GTP has been solved in two conformations. First, the structure of a mutant EF-G·GDPNP revealed only slight differences when compared with the GDP-bound form (75). Noteworthy differences between these structures are seen in the P-loop, the visible regions of the switch 1 loop, the switch 2 loop, and a slight shift at the tip of domain IV, where a conserved histidine resides. Whereas the γ-phosphate of GDPNP is ordered in the structure, the switch 1 loop is not. This could be due to an artifact from crystal packing or because the crystals were seeded from mutant EF-G·GDP crystals. Recently, the x-ray structure of an EF-G homolog, EF-G-2, complexed with GTP was reported (76). This structure possesses a fully ordered switch 1 loop and a conformation that is distinct from the GDP-bound EF-G structure; specifically, a reorganization of domains III–V with respect to domains I, II, and G′. This rearrangement results in a 25-Å shift at the tip of domain IV when compared with the GDP-bound EF-G, a conformation of EF-G undoubtedly favorable for binding to the ribosome as evidenced by the increased affinity for EF-G·GTP over its GDP-bound counterpart (see ref. 77 and references therein).

Attempts to find a system in the bewildering variety of observed structures for EF-G (see, for instance, ref. 78) need to account for the fact that the factor is in different environments, facing different conformational constraints when placed in a crystal context versus bound to the ribosome. Conformational differences relevant for the functioning of the factor while performing its work on the ribosome are best inferred from structures of EF-G bound to the ribosome in an authentic state. Because of the lack of x-ray structures for such a complex, the existing quasi-atomic models of EF-G computed from cryo-EM maps are currently the only source of information. A comparison of the cryo-EM reconstructions of EF-G (11–14) bound to the ribosome with the crystal structures of the factors (assuming that these are close to the solution forms) revealed that EF-G undergoes gross conformational changes when it binds to the ribosome. These changes are again described as a hinge-like joint motion of domains III, IV, and V with respect to domains I, II, and G′ (11, 12) and are coincident with the ratchet motion (29). The hinge-like conformational change in EF-G induced by ribosome binding is in the same direction as that of the GDP- to GTP-bound conformations but with even greater magnitude (≈27 Å) (SI Fig. 8).

The conformations of ribosome-bound EF-G before and after GTP hydrolysis are quite likely similar to those recently elucidated for eEF2 by cryo-EM reconstructions of 80S–eEF2 complexes (23). The cryo-EM maps of modified eEF2 bound to the 80S ribosome in the GTP-state (using GDPNP) and GDP-state (GDP plus sordarin) revealed additional, smaller conformational changes in eEF2 upon GTP hydrolysis, which include a 6-Å shift of domain IV of eEF2 toward the decoding center (23). This conformational change is likely responsible for decoupling the decoding center from the mRNA–tRNA duplex in the A site, a prerequisite for translocation.

The GTP Hydrolysis Mechanism in EF-G-Catalyzed Translocation

GTPases, including EF-Tu and EF-G and their eukaryotic counterparts, are a family of proteins characterized by a GTP-binding domain, which is responsible for the binding and hydrolysis of GTP. GTPases function as molecular switches in several processes of cellular regulation by cycling between an active, GTP-bound state and an inactive, GDP-bound state (see ref. 79). The core, or G domain, is composed of three conserved features in the nucleotide-binding pocket: the phosphate-binding loop (P-loop) and the switch 1 and switch 2 motifs. The P-loop binds the nucleotide via the α- and β-phosphates, while the switch 1 and 2 loops coordinate the γ-phosphate of GTP.

Based on the similarities in structure, particularly of the switch 1 and 2 loops and their conserved ribosomal binding site near the GAC, it is likely that the mechanism of GTP hydrolysis is conserved among the elongation factors. In fact, this mechanism may be extended to all ribosomal binding GTPases involved in initiation, elongation, and termination because of the high conservation in factor structures and the universal GAC binding site on the ribosome. However, despite the availability of several crystal structures of these factors bound with various nucleotides and antibiotics, the mechanism of GTP hydrolysis of the factors is not well understood. This is due mainly to a lack of atomic models of the elongation factors bound to the ribosome. Such a model is obviously important for the understanding of the mechanism because GTP hydrolysis by elongation factors is accelerated on the ribosome by more than seven orders of magnitude (80).

GTPase activation of the smaller G proteins is catalyzed by the binding of a GTPase-activating protein (GAP), which induces conformational changes in, and thus alters the activity of, the G proteins (79). Because there is no GAP per se for the elongation factors, the GAC of the ribosome apparently assumes this role. Recent cryo-EM reconstructions of EF-G·70S (76) and eEF2·80S (23) complexes revealed that the switch 1 loop of EF-G/eEF2 is in fact ordered in the GTP-bound state. After GTP hydrolysis, the switch 1 loop of the ribosome-bound eEF2 becomes disordered, similar to that of the EF-Tu·GDP·aurodox crystal structure (81) and the structure of 70S·EF-Tu·Phe-tRNA^Phe·kirromycin complex studied by cryo-EM, which also represents a post-GTPase state (24). This observation indicates a critical role of the switch 1 loop, as well as the existence of a similar mechanism of GTP hydrolysis in EF-Tu and EF-G/eEF2.

The sarcin–ricin loop (SRL) of the ribosome is a conserved tetraloop in the rRNA of the large subunit, also implicated in the activation of the GTPase activity. Cleavage of the SRL by the ribotoxin α-sarcin or ricin inhibits binding of elongation factors (82, 83). When free in solution, the SRL binds EF-G (84). Cryo-EM showed that the SRL is in close proximity of the switch 1 loop of GTP-bound EF-G and eEF2 (23, 76). Chemical protection assays revealed that the SRL is protected by EF-G in either the GTP-bound (GDPNP) or the GDP-bound (stabilized by fusidic acid) conformation (85). A conserved feature in the cryo-EM reconstructions of the ribosome-bound GTPases, IF2 (86) and eEF2 (23), is a shift in the GTP-binding domain of the factors toward the GAC of the ribosome in the transition from the GTP-bound states to GDP-bound states. This transition may alter the interactions between the SRL and the switch regions of the factors, leading to GTP hydrolysis and the disordering of the switch 1 loop.

Head Rotation of the Small Subunit: A Pivotal Step in Translocation

While the movement of the tRNAs on the large subunit (i.e., the formation of the hybrid-state tRNA) is facilitated by the binding of EF-G·GTP and the concomitant ratcheting of the ribosome, the second step of translocation, the movement of the tRNAs along with the mRNA on the small subunit by one codon, is more difficult to explain. Evidence has accumulated that supports the notion that GTP hydrolysis precedes translocation on the small subunit (44, 87). Toe-printing experiments of EF-G·70S complexes in the presence of either GTP or GDPNP demonstrated that mRNA movement relative to the small subunit occurs after GTP hydrolysis (88). Kinetic data suggest that GTP hydrolysis by EF-G leads to some kind of “unlocking” event that induces a conformational change of the ribosome, on such a scale that it limits the rate of translocation (89).

An explanation of this “unlocking” event was recently proposed on the basis of cryo-EM reconstructions of modified eEF2 bound to the 80S ribosome (23). Using the ADP moiety as a density marker, these authors showed that ADP-ribosylated eEF2 (ADPR-eEF2) undergoes conformational changes that are a direct result of GTP hydrolysis. These changes result in a 6-Å shift in the tip of domain IV of ADPR-eEF2, where a diphthamide residue resides in eEF2 (or a conserved histidine, in EF-G). Judged from the location of the tip, this movement likely severs the connection between the ribosomal decoding center in the body of the small subunit and the mRNA–tRNA duplex in the A site, bound to the head of the small subunit. Once this link is severed, the head is free to rotate around the neck of the small subunit, simultaneously translocating the mRNA–tRNA complex.

Such a head rotation of the subunit, which is independent of ratcheting, has been inferred from the difference in head positions observed both in cryo-EM reconstructions of 80S ribosomes (22, 23) and in x-ray structures of the 70S ribosome (64). Even earlier small-angle neutron scattering data predicted a head rotation of the small subunit to provide the mechanical act of translocation (90). Genetic mutations also suggest a necessary interaction between the head of the small subunit and the tRNAs during translocation, because mutations to a conserved adenine in the head-region of 16S rRNA (A1339 in E. coli) confer a >18-fold decrease in translation activity (91). Not surprisingly, A1339 interacts intimately with P-site tRNA in the PRE ribosome (32).

Analysis of the small subunit from the x-ray structures of vacant 70S ribosomes from E. coli revealed that residues G1338 to U1341 (E. coli numbering) in the head and residue A790 in the shoulder of the small subunit form a block between the P and E sites that opens or closes as a function of head movement (64). In the x-ray structure of the 70S ribosome in the PRE state (32) and the cryo-EM structure of the 80S ribosome in that same state (92), the distance between A1340 and A790 α-carbons is ≈18 Å and thus would impede movement of the tRNA stem from the P to E site. Binding of eEF2 opens this distance to >26 Å in the 80S ribosome (22, 23), which would accommodate the passing of the anticodon stem-loop of tRNA on the small subunit from the P to the E site (64). In the cryo-EM maps, however, eEF2 is bound and this block is “open,” but the P/E-site tRNA remains in the hybrid state, interacting strongly with the G1338–U1341 region of the head (Fig. 4). The strong interaction seen between the P/E-site tRNA and the head of the small subunit contrasts with tRNAs in the classic states, where the main interactions are with the body of the small subunit (32–34). This result suggests that in a translating ribosome, the head movement physically drags the anticodon stem-loop of P/E site tRNA toward the E site through a connection with the G1338–U1341 region in the head of the small subunit after ratcheting and GTP hydrolysis. At this point, the P/E hybrid tRNA is likely to be committed to the E site through a strong interaction formed with the L1 stalk of the large subunit after ratcheting of the ribosome (13, 21, 22), an interaction that may assist the transition phase. Disruptions of the interaction between the head region of the small subunit and the tRNAs must be induced by a combination of back-ratcheting of the small subunit and back-rotation of the head, both of which occur upon dissociation of eEF2/EF-G·GDP. We speculate that it is this event that allows P/E tRNA to finally pass through the gate between the head (G1338–U1341) and body (A790) of the small subunit to fully occupy the classic E/E state.

Finally, it is likely that the binding of EF-G indirectly provides the driving force for head rotation in the small subunit: The binding of EF-G·GTP and ribosomal ratcheting are coupled events that are accompanied by a bending of h44 at a point 2/3 along its length, producing a shift of the decoding center by ≈8 Å toward the P site (93). This shift is required to fully accommodate domain IV of EF-G in the ribosomal A site. Directly connected with h44 is h28 in 16S rRNA, which makes up the neck of the small subunit (Fig. 1). It is plausible that EF-G-induced movement at the top of h44 will cause torsional stress in h28. Thus, the head of the small subunit would be primed to rotate as soon as the rigid link between the mRNA–tRNA duplex, still bound to the small subunit's head in the A site, and the decoding center in the body of the small subunit is removed upon GTP hydrolysis (see Fig. 1).

Synthesis of a Model for mRNA–tRNA Translocation

Recent evidence from structural, kinetic, and biochemical data, some of which has been reviewed above, point toward a two-step mechanism of translocation. The binding of EF-G·GTP (or EF-G·GDPNP) induces a first phase of translocation by stabilizing the tRNA molecules in the hybrid state. This occurs because the deacylation of the tRNA at the P site liberates the tRNA CCA, such that it can proceed to the E site on the 50S subunit, and allows the small subunit to ratchet forward while the mRNA–tRNA remains locked with the small subunit at its P site. At the same time, the vacating of the P site on the large subunit by the P/E-tRNA creates the precondition for the formation of the A/P hybrid state (43). Once this state is attained and stabilized by EF-G·GTP, the second step of translocation will start. This step is catalyzed by GTP hydrolysis on EF-G and entails the coupled movement of the mRNA and the anticodon loops of the tRNAs with respect to the small subunit.

The recent crystal structure of EF-G-2·GTP presents a conformation of EF-G that is ideally tailored to interacting with the PRE ribosome (76). Modeling of EF-G in this conformation with the PRE ribosome suggests that the binding of domains I, II, and G′ to the large subunit of the ribosome would position the tip of domain IV precisely at the codon–anticodon binding region of A-site tRNA (SI Fig. 8). Domain IV of EF-G, specifically the conserved histidine residue at its tip, comes into intimate contact with the codon–anticodon moiety formed by the mRNA and A-site tRNA. This interaction could stabilize the base pairing of the codon–anticodon of A-site tRNA during ratcheting of the ribosome, as suggested earlier (22). This reasoning would in fact explain the adverse effects on translocation when the conserved histidine in domain IV of EF-G is mutated (94). (For the eukaryotic case, the same scenario will hold, with EF-G replaced by eEF2 and the critical histidine residue by diphthamide.)

The movement of the decoding center in the direction of mRNA–tRNA translocation also stabilizes the A-site tRNA in the A/P hybrid state. Importantly, placement of tRNAs in the hybrid state is more favorable for translocation to occur as compared with tRNAs in the classic states (39, 43). Therefore, by stabilizing tRNAs in the hybrid states, EF-G binding directs the translocation process to go forward, which explains the catalytic potential of EF-G even in the absence of GTP hydrolysis (27).

Binding of EF-G and ratcheting of the ribosome are followed by GTP hydrolysis, causing a shift of domain IV, which in turn detaches the mRNA–tRNA complex from the decoding center (23). The mRNA–tRNA complex is temporarily freed from the body of the small subunit but likely maintains the strong interactions with the subunit's head. A head rotation in the small subunit is likely responsible for translocation of the mRNA chain by one codon but, because the tRNAs maintain their contacts with the head throughout the rotation, they technically remain in their hybrid states. Disruptions of these interactions must be induced by a combination of reverse-ratcheting of the small subunit and back-rotation of the head that both occur upon release of eEF2/EF-G. The final result is the POST-state ribosome with tRNAs in the canonical P and E sites (Fig. 5).

Conclusion

The ribosome, as we have seen, has intrinsic properties facilitating translocation. The most important one is an architecture that lends a specific kind of instability to the molecule and allows the ribosome to alternate between two distinct conformations separated by a small activation barrier. The change from one conformation to the other is along a pathway predicted as one of the highest-ranking modes by normal mode analysis of the x-ray structure (61, 62). This means, in other words, that the architecture has developed such that a minimum of energy is needed to go from the “normal” conformation to the “ratcheted” conformation, which is productive for the ultimate events of translocation.

This kind of parsimonious design is ubiquitous in all life forms. Human locomotion is an example on the macroscopic scale: during the act of walking, the legs are used as pendula, swinging with close to their innate resonance frequency and thus requiring a minimum of energy for sustenance of the movement. In the same way, energy supplied by thermal bombardment is channeled, by virtue of the ribosome's unique architecture, into a motion that is most productive in terms of translation.

Cryo-EM and the methods of flexible fitting by real-space refinement have allowed us to produce quasi-atomic models of the ribosome in both states (SI Fig. 6). An analysis of these models—particularly by watching an animation in which the two models alternate while the ribosome rotates (SI Movie 1)—makes it clear that the motion involves two interrelated events: (i) a conformational change in the intersubunit bridges and (ii) a change in the entire rRNA framework of each of the subunits, which resembles an elastic deformation, and accommodates the changes in the bridges.

The principal role of EF-G/eEF2, then, is in decreasing the energy barrier between the two ribosomal conformations. Specifically, the role of these factors is twofold: in bringing the ribosome into a conformation that is consistent with the binding of the tRNAs at the hybrid sites, and in unlocking the mRNA–tRNA moiety from the decoding center, thereby allowing the small subunit head movement to occur that is required for completion of translocation.

An “ur-ribosome” with the same basic architecture, and the same properties of instability, in a world in which elongation factors had not yet made their appearance, would already have functioned, albeit with very much decreased efficiency and accuracy. The very small rate of protein synthesis in these initial stages means that the evolution of the complex G-proteins (EF-G/eEF2, EF-Tu/eEF1A, IF2/eIF5B, and RF3/eRF3) tailored precisely to the ribosome's architecture could only proceed at a snail's pace. A similar line of reasoning holds for the impact of the addition of ribosomal proteins such as S12 and S13, which greatly improved the accuracy of factor-free translocation. Once these helper proteins had developed, however, the resultant much higher efficiency and accuracy of protein synthesis would lead to today's overflowing richness of life forms.

Supplementary Material