The Ribosome and the Elongation Cycle

Kinetic proofreading by the ribosome is not a museum piece — it is a live target in modern pharmacology and biotechnology. Aminoglycoside antibiotics (streptomycin, gentamicin, paromomycin) bind the 16S rRNA decoding center and distort the induced-fit geometry used during initial selection, increasing misreading and causing bactericidal mistranslation; understanding the proofreading mechanism explains both their efficacy and mechanisms of bacterial resistance mutations that restore fidelity at the cost of drug binding. Macrolides (erythromycin, azithromycin) block the nascent peptide exit tunnel rather than the PTC directly, stalling elongation; oxazolidinones (linezolid) bind near the PTC A site itself. Fusidic acid and related compounds trap EF-G post-hydrolysis, directly exploiting the mechanical, not chemical, role of GTP hydrolysis in translocation described here. In biotechnology, engineered orthogonal ribosomes and modified EF-Tu variants are used for unnatural amino acid incorporation (genetic code expansion), deliberately exploiting or bypassing natural proofreading checkpoints to permit non-canonical residues to reach the PTC. In human disease, ribosomopathies (Diamond-Blackfan anemia, Treacher Collins syndrome) arise from mutations in ribosomal proteins that subtly perturb assembly or fidelity rather than abolishing translation outright, illustrating that even small kinetic perturbations to this proofreading cycle have organism-level developmental consequences. Cryo-EM has now captured every intermediate described mechanistically here — ternary complex sampling, GTPase activation, accommodation, hybrid states, and EF-G-driven translocation — as discrete structural snapshots, turning what was once purely kinetic inference into a visually complete movie of the elongation cycle. This structural and kinetic foundation sets up what happens immediately after: how the finished polypeptide threads through the exit tunnel toward its ultimate cellular destination and chemical maturation.

Ask the tutor: How would the fidelity math change if a third, independent proofreading checkpoint were added to the elongation cycle?

The bacterial ribosome (70S, ~2.5 MDa) is built from a small subunit (30S: 16S rRNA + ~21 proteins) that reads the mRNA code, and a large subunit (50S: 23S rRNA, 5S rRNA, ~31 proteins) that houses the peptidyl transferase center and the nascent polypeptide exit tunnel. Eukaryotic ribosomes (80S: 40S + 60S) are larger and more protein-rich but organized identically. Three tRNA-binding sites span both subunits: the A site (aminoacyl) accepts incoming charged tRNA, the P site (peptidyl) holds the tRNA carrying the growing chain, and the E site (exit) transiently holds deacylated tRNA before it dissociates. Each site is really a composite: a codon-anticipating "decoding" region on the 30S and an acceptor-stem-binding region on the 50S, connected because a single tRNA spans both subunits. One elongation cycle proceeds as follows. A ternary complex — aminoacyl-tRNA bound to EF-Tu·GTP — diffuses into the empty A site. The 16S rRNA decoding center inspects the codon–anticodon minihelix; if the fit is good, EF-Tu is triggered to hydrolyze GTP, releases the tRNA, and dissociates. The tRNA's 3′-CCA-aminoacyl end then swings ~70 Å from the decoding center into the peptidyl transferase center — this is "accommodation." Once accommodated, the PTC catalyzes nucleophilic attack of the A-site amino group on the P-site ester carbonyl, forming a new peptide bond and leaving a deacylated tRNA in the P site and a peptidyl-tRNA (one residue longer) in the A site. EF-G·GTP then binds and drives translocation: the mRNA and both tRNAs move by exactly one codon, so the peptidyl-tRNA shifts A→P and the deacylated tRNA shifts P→E, regenerating an empty A site for the next ternary complex.

The peptidyl transferase center is worth dwelling on because it overturned a century of assumption that catalysis requires protein. Crystal structures of the 50S subunit (Steitz, Moore, Ramakrishnan, Yonath — 2000 Nobel Prize) showed no protein side chain within 18 Å of the catalytic core; the reaction is carried out entirely by 23S rRNA nucleotides, primarily through substrate positioning and proton shuttling rather than direct acid-base chemistry from a ribozyme functional group. The active site accelerates peptide bond formation roughly 10⁷-fold over the uncatalyzed solution reaction, mostly by entropic effects: locking the nucleophilic amine and the ester carbonyl into the precise geometry and orientation needed for in-line attack, and by desolvating the reacting groups.

The two central quantitative relationships in this article are (1) the free-energy/rate framework for peptide bond formation and translocation and (2) the kinetic proofreading equation that explains fidelity. For any step in the cycle, the observed rate constant follows transition-state theory:

where:

• $k_B T/h$ is the universal frequency factor (~6 × 10¹² s⁻¹ at 37 °C)
• $\Delta G^{\ddagger}$ is the free energy of activation for that step
• $R$, $T$ are the gas constant and absolute temperature

Reading: lowering $\Delta G^{\ddagger}$ by ~4 kcal/mol multiplies the rate by roughly 1000-fold — this is the currency the PTC spends by pre-organizing substrates, and the currency EF-Tu spends by clamping the tRNA in a strained, hydrolysis-competent conformation only when codon–anticodon pairing is correct. The discrimination factor between cognate and near-cognate tRNA at a single binding/selection step is set by the free-energy difference in Watson-Crick versus mismatched pairing:

where:

• $D$ is the specificity ratio (catalytic efficiency ratio) for one selection event
• $\Delta\Delta G$ is the free-energy difference between correct and incorrect codon-anticodon pairing (typically 2-3 kcal/mol for a single mismatch)
• a single $\Delta\Delta G$ of ~2.8 kcal/mol at 37 °C gives $D \approx 100$

Because two W-C mismatches or wobble-type errors give smaller $\Delta\Delta G$, a single binding step alone cannot explain observed fidelities of 10³–10⁴; this motivates kinetic proofreading, in which the same modest $D$ is applied twice at two thermodynamically decoupled checkpoints, multiplying to $D^2 \approx 10^4$.

Goal: derive how GTP hydrolysis converts a modest one-step discrimination ratio into the ribosome's overall high fidelity, and show why an irreversible step in the middle of the pathway is essential.

Step 1 — Initial selection (before GTP hydrolysis). The ternary complex aa-tRNA·EF-Tu·GTP binds the A site. The codon–anticodon minihelix is monitored by universally conserved 16S rRNA bases (A1492, A1493, G530) that flip out and hydrogen-bond into the minor groove of the first two codon–anticodon base pairs, sensing Watson-Crick geometry rather than sequence identity directly. A correct match triggers a conformational signal transmitted through the tRNA body to EF-Tu's GTPase domain, activating GTP hydrolysis. Because the decoding center only samples shape, both cognate and many near-cognate (mismatched) ternary complexes can bind and even trigger hydrolysis, but at different rates:

This is a genuine kinetic (not equilibrium) selection: at physiological ternary-complex concentrations, rejection occurs mostly through dissociation before GTP hydrolysis is triggered, governed by

For cognate pairs, GTPase activation outcompetes dissociation; for near-cognate pairs, the reverse is usually true, so most wrong ternary complexes simply fall off here — this is selection step 1, worth roughly $D_1 \approx 10^2$–$10^3$.

Step 2 — GTP hydrolysis and the irreversible commitment. If activation wins, GTP is hydrolyzed on EF-Tu in ~1 ms. This is the crucial irreversible branch point: hydrolysis to GDP + Pi is (for practical kinetic purposes) not reversed, so the system cannot "go back" and re-examine the same binding event. This irreversibility is what makes proofreading possible — reversible steps cannot proofread, because detailed balance would erase any discrimination gained after the fact.

Step 3 — Proofreading (after hydrolysis, before peptide bond formation). EF-Tu·GDP has low affinity for aa-tRNA and dissociates from the ribosome, releasing the aa-tRNA to "accommodate" — diffuse into the PTC. But this diffusional step takes real time (~10 ms), during which the aa-tRNA can still dissociate from the A site if the codon–anticodon interaction is weak. This is the second, kinetically independent discrimination event:

Near-cognate tRNAs, having survived step 1 by chance, are still weakly bound and preferentially dissociate ("drop off") during accommodation rather than reach the PTC — worth another $D_2 \approx 10$–$100$.

Step 4 — Multiplication of independent factors. Because the two checkpoints are separated by an irreversible step (GTP hydrolysis), the overall discrimination is the product, not the sum or an average, of the two factors:

Numerically, with $D_1 = 200$ and $D_2 = 40$ (typical experimental orders of magnitude from Ehrenberg/Kurland-type kinetic measurements):

which matches observed in vivo missense error frequencies of roughly 1 per 10³–10⁴ codons. This is the entire logic of Hopfield/Ninio kinetic proofreading: spend one GTP per aa-tRNA delivered, in order to insert an irreversible step between two otherwise-correlated selection events, so that the errors made at the two steps are statistically independent and their small individual discriminations multiply.

Limiting case check. If GTP hydrolysis were reversible (imagine an enzyme that could easily re-form GTP from GDP + Pi and "undo" its commitment), then by microscopic reversibility the system could not net-discriminate any better than a single equilibrium step; formally $D_{overall} \to D_1$ only, because step 2 could no longer act as an independent filter. This is exactly why the ribosome couples fidelity to an energy-dissipating, irreversible hydrolysis — proofreading is fundamentally a non-equilibrium phenomenon, and it costs one GTP (~7.3 kcal/mol standard free energy of hydrolysis, more like 11-13 kcal/mol at cellular concentrations) per correctly incorporated residue as the price of that independence.

Goal: show mechanistically how the PTC forms the peptide bond and how EF-G couples a second GTP hydrolysis to precise, unidirectional one-codon movement.

Step 1 — Substrate juxtaposition. After accommodation, the A-site aminoacyl-tRNA's 3′ CCA end sits in the PTC adjacent to the P-site peptidyl-tRNA's 3′ CCA end. Both are held by an extensive network of rRNA hydrogen bonds (notably to A2451 and the "A-loop"/"P-loop" of 23S rRNA) that positions the P-site ester carbonyl carbon roughly in-line with the lone pair of the A-site α-amino nitrogen — an attack trajectory geometrically pre-organized rather than searched for in solution.

Step 2 — Nucleophilic attack. The A-site α-amino group (largely deprotonated/nucleophilic form favored locally by the microenvironment) attacks the carbonyl carbon of the P-site ester, forming a tetrahedral oxyanion intermediate:

Step 3 — Proton shuttle and collapse. The developing negative charge on the former carbonyl oxygen is stabilized transiently (proposed proton-wire mechanisms invoke the 2′-OH of A76 on the P-site tRNA acting as a general acid/base relay, though the precise chemical assistance remains an active area refined by later crystallography). Collapse of the tetrahedral intermediate expels the 3′-O of the P-site tRNA as the leaving group, generating a new peptide bond:

This reaction is entropically catalyzed: the uncatalyzed rate in solution for two similarly juxtaposed esters/amines is negligible over biological timescales; the ribosome's ~10⁷-fold rate enhancement comes almost entirely from removing translational and rotational entropy costs and excluding water, not from covalent catalysis by an active-site residue — consistent with the absence of protein near the PTC.

Step 4 — EF-G binding and the ratchet motion. After peptide bond formation, the ribosome exists in two interconverting global conformations: a non-rotated ("classical") state and a rotated ("hybrid," ~9° intersubunit rotation) state in which the acceptor ends of the tRNAs have already partially shifted (A/P and P/E hybrid states) while their anticodon ends remain in the A and P sites of the 30S. EF-G·GTP binds preferentially to the rotated, hybrid-state ribosome, inserts its domain IV into the A site (mimicking, sterically, part of a tRNA), and its GTP hydrolysis, together with Pi release, drives a large conformational change (partly a "power stroke," partly biased Brownian ratcheting) that completes translocation of both tRNAs and the mRNA by exactly one codon (3 nucleotides):

Step 5 — Fidelity of translocation distance. The single-codon step size is enforced by the geometry of the mRNA channel and the tRNA anticodon loops still base-paired to successive codons; slippage (±1 nucleotide) is rare (frameshifting frequency typically <10⁻⁴ per codon) because breaking and re-forming three correct base pairs simultaneously, in register, is strongly disfavored relative to sliding exactly one triplet under the geometric constraint of the mRNA channel and the swiveling head domain of the 30S subunit.

Limiting case check. If EF-G's GTPase activity is abolished (as with certain mutants or the antibiotic-stabilized states used experimentally), the ribosome can still translocate slowly via spontaneous ratcheting/hybrid-state fluctuations, but at a rate reduced by several orders of magnitude and with loss of directional bias — showing that GTP hydrolysis here, as with EF-Tu, is not chemically required for the physical motion itself but is required to make it fast and unidirectional, i.e., it lowers $\Delta G^{\ddagger}$ and breaks detailed balance to enforce a preferred direction.

Example 1 (foundational) — Overall discrimination from two proofreading steps. Suppose initial selection favors cognate over a specific near-cognate tRNA with $D_1 = 180$, and the accommodation/proofreading step gives $D_2 = 55$ for the same pair (these are realistic, non-round, asymmetric numbers consistent with published stopped-flow kinetic partitioning experiments). Find the overall error frequency expected for this codon.

This lands squarely in the experimentally measured in vivo range (roughly 10⁻³ to 10⁻⁴ per codon), confirming the two-step multiplicative model rather than, say, an additive one (which would predict $D \approx 235$, far too low, and thus is ruled out by matching real error rates).

Example 2 (applied) — Energetic cost of fidelity per protein made. A ribosome synthesizing a 437-residue protein hydrolyzes one GTP per elongation cycle for EF-Tu delivery and one GTP per cycle for EF-G translocation (2 GTP per residue added, ignoring initiation/termination GTPs for this estimate). Using a cellular free energy of GTP hydrolysis of about $\Delta G'_{GTP} \approx -12.5$ kcal/mol (physiological, not standard-state, value):

per mole of this protein synthesized — roughly 25 kcal/mol of GTP energy spent per peptide bond formed (translation is energetically far more expensive than the ~0.5–1 kcal/mol needed thermodynamically just to form one peptide bond; nearly all of this "extra" energy purchases speed and the fidelity machinery, not the bond itself). This illustrates the central trade-off: the cell pays roughly an order of magnitude more free energy than chemically necessary, in exchange for the kinetic proofreading discrimination derived above.

Example 3 (exam/competition level) — Partitioning kinetics at a branch point. At the initial-selection branch point, a near-cognate ternary complex bound in the A site can either dissociate (rate constant $k_{off} = 80\ \text{s}^{-1}$) or trigger GTPase activation and proceed forward (rate constant $k_{cat} = 0.7\ \text{s}^{-1}$). For the cognate ternary complex under the same geometry, $k_{off} = 0.4\ \text{s}^{-1}$ and $k_{cat} = 250\ \text{s}^{-1}$ (asymmetric, realistic orders-of-magnitude values reflecting real single-molecule/stopped-flow ribosome kinetics literature). Compute the partitioning fraction (probability of proceeding forward rather than dissociating) for each case:

For cognate:

For near-cognate:

The step-1 discrimination ratio is the ratio of these partitioning probabilities:

Limiting-case sanity check: if $k_{off}\to 0$ for both species (an idealized ribosome that never releases bound tRNA), $f_{proceed}\to 1$ for both cognate and near-cognate, and $D_1 \to 1$ — all discrimination vanishes, because a step can only discriminate through the competition between forward commitment and rejection; a purely "sticky" A site would accept everything indiscriminately. This confirms that fidelity is a kinetic partitioning phenomenon, not simply a matter of tight binding.

Structural elucidation of the ribosome earned the 2009 Nobel Prize in Chemistry for Venkatraman Ramakrishnan, Thomas A. Steitz, and Ada Yonath, who independently solved atomic-resolution crystal structures of the 30S and 50S subunits around 2000, following decades of painstaking crystallization work pioneered largely by Yonath starting in the late 1970s using ribosomes from extremophile bacteria to obtain well-ordered crystals. Their structures directly confirmed, at atomic resolution, that no protein side chain lies near the peptidyl transferase active site — settling a debate that had simmered since the 1960s about whether the ribosome was fundamentally an RNA enzyme or a protein enzyme scaffolded by RNA. The kinetic proofreading framework itself predates these structures by decades: John Hopfield (1974) and, independently, Jacques Ninio (1975), proposed that biological systems achieve fidelity beyond simple equilibrium binding by coupling recognition to an irreversible, energy-dissipating step — a purely kinetic argument later validated mechanistically for the ribosome through detailed pre-steady-state kinetic work by Marina Rodnina and Wolfgang Wintermeyer, whose stopped-flow and single-molecule FRET experiments in the 1990s-2000s measured the actual rate constants ($k_{off}$, $k_{cat}$, accommodation rates) that make the proofreading scheme quantitative rather than merely conceptual.