Each elongation cycle admits one aminoacyl-tRNA into the ribosomal A site, checked against the mRNA codon in two independent kinetic steps (Hopfield/Ninio kinetic proofreading, 1974/1975) — this is why translation is far more accurate than a single binding equilibrium could ever be:
Step 1 — initial selection (before GTP hydrolysis by EF-Tu):
correct codon-anticodon pairing -> fast forward rate k1_f
near-cognate mismatch -> high rejection rate k1_r
Step 2 — proofreading (after GTP hydrolysis, before peptide bond):
correct tRNA -> fast accommodation, rare rejection
near-cognate -> re-checked, mostly rejected & released
Overall discrimination: D = D1 x D2
Error frequency: f_err ~ 1 / D (real ribosomes: f_err ~ 10^-4 to 10^-5 per codon)
Because the two checks are independent, their discrimination factors multiply rather than add — a modest ~10-100x preference at each step compounds into the ~103-105-fold fidelity real ribosomes achieve, at an elongation rate of roughly 15-20 amino acids/second in bacteria.
- Cognate : near-cognate ratio — relative cytoplasmic concentration of matching vs. mismatching tRNAs competing for the A site (this is what makes rare codons slow and error-prone: fewer cognate copies mean more near-cognate encounters before a correct match arrives).
- Proofreading stringency — scales how aggressively step 2 rejects a near-cognate tRNA that slipped past step 1; turning it down reproduces antibiotic/mutant conditions (e.g. streptomycin, ram mutants) that raise the real error rate.
- Common / Rare codon — switches the current codon's cognate tRNA abundance, the single biggest determinant of local elongation speed along a real mRNA.