Molecular Biology
Translation (Protein Synthesis)
From mRNA codons to a folded polypeptide, one amino acid every 50 milliseconds
Translation is the process that reads an mRNA into a chain of amino acids. It runs in three stages: initiation assembles a ribosome on the start codon; elongation adds one amino acid per codon at roughly 20 per second; termination ends at a stop codon and releases the finished protein. Twenty aminoacyl-tRNA synthetases — one per amino acid — quietly enforce the genetic code by charging each tRNA with the correct amino acid before it ever reaches the ribosome.
- Three stagesInitiation → Elongation → Termination
- Start codonAUG (Met / fMet in bacteria)
- Stop codonsUAA, UAG, UGA
- Rate~20 amino acids / sec
- Error rate~1 in 10,000 per codon
- Energy cost~4 high-energy phosphates per residue
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
The three-stage program
Translation begins with an mRNA, charged tRNAs, two free ribosomal subunits, and factors. By the end, the polypeptide has started folding even before the last residue is added.
5' ━━━━━━━━━━━━━━━━━━━━━━━━━ 3'
AUG CCU AAG GCU GAA UAA
M P K A E STOP
INITIATION → ELONGATION → TERMINATION
(cap or SD) (decode → (release factor
(small + peptide at stop codon
AUG + bond → → free peptide,
large) translocate) recycle subunits)Initiation — finding the start
The hardest part of translation is finding the right AUG. The cell has to start at the right one — typically the first one in the right context — every single time.
Bacteria use the Shine-Dalgarno sequence: a purine-rich stretch (consensus AGGAGG) ~8 nt upstream of the start codon that base-pairs with the 3' end of 16S rRNA. The 30S parks on the SD, the AUG falls into the P site, and fMet-tRNA (formylmethionine) is delivered by IF2-GTP. IF1 blocks the A site, IF3 keeps the 50S off, and once the 70S assembles GTP hydrolyzes and elongation begins.
Eukaryotes use the cap-and-scan model. The 40S, preloaded with Met-tRNA via eIF2-GTP, recognizes the m7G cap through the eIF4F complex (eIF4E, eIF4G, eIF4A) and scans 3' along the mRNA, melting structure with eIF4A helicase, until it lands on an AUG in good Kozak context (consensus gccRccAUGG). Recognition triggers eIF2-GTP hydrolysis and eIF5B-driven 60S joining.
Some viral and stress-response mRNAs bypass cap scanning entirely using internal ribosome entry sites (IRESs). Hepatitis C, polio, and apoptotic translation all use this trick.
Elongation — one codon, one amino acid
Each elongation cycle adds one amino acid, shifts the frame by three nucleotides, and repeats ~20 times per second.
- Decoding. An aminoacyl-tRNA escorted by EF-Tu-GTP (eEF1A-GTP) enters the A site. Correct codon-anticodon pairing triggers GTP hydrolysis and tRNA accommodation; wrong pairs dissociate first. This kinetic proofreading step is where most of the code's accuracy comes from.
- Peptide bond formation. The A-site amino group attacks the P-site peptidyl-tRNA carbonyl, transferring the growing chain to the A-site tRNA. 23S rRNA catalyzes — the ribozyme. No input energy needed; it was paid when the synthetase charged the tRNA.
- Translocation. EF-G-GTP (eEF2-GTP) ratchets the mRNA by three nucleotides — empty tRNA moves P → E → out, peptidyl-tRNA moves A → P. The A site is empty for the next codon.
Aminoacyl-tRNA synthetases — silent enforcers of the code
The ribosome doesn't actually know the genetic code. It only knows how to base-pair anticodons to codons and how to make peptide bonds. The mapping from codon to amino acid is enforced one step earlier, by the aminoacyl-tRNA synthetases — 20 enzymes, one per amino acid.
Each synthetase recognizes both its amino acid and its cognate tRNAs, activates the amino acid with ATP, and covalently attaches it to the tRNA's 3' end. Synthetases handling chemically similar pairs (Ile vs Val, Thr vs Ser) have a separate editing site that hydrolyzes mischarged tRNAs. Missense mutations in synthetases are devastating because they corrupt the code globally — a leaky valyl-tRNA synthetase that occasionally accepts threonine sprinkles wrong residues across every Val-containing protein in the cell.
Termination and recycling
Stop codons are read by release factors, not tRNAs. Bacteria use RF1 (UAA/UAG) and RF2 (UAA/UGA); RF3-GTP recycles them. Eukaryotes use a single eRF1 for all three plus eRF3-GTP. The release factor's GGQ motif redirects the peptidyl transferase center from bond formation to ester hydrolysis, freeing the polypeptide. RRF + EF-G (bacteria) or ABCE1 (eukaryotes) then split the ribosome back into subunits for reuse.
Translation across organisms
| E. coli | S. cerevisiae | Human cytosol | Human mitochondria | Plant chloroplast | Archaea | |
|---|---|---|---|---|---|---|
| Ribosome | 70S | 80S | 80S | 55S (bacterial-like) | 70S | 70S |
| Initiator tRNA | fMet-tRNA | Met-tRNAᵢ | Met-tRNAᵢ | fMet-tRNA | fMet-tRNA | Met-tRNA (no formyl) |
| Start-codon recognition | Shine-Dalgarno sequence | Cap + scanning | Cap + scanning (Kozak) | Direct positioning, no SD/cap | SD-like | SD or scanning (variable) |
| Initiation factors | IF1, IF2, IF3 | ~12 eIFs | ~12 eIFs | mtIF2, mtIF3 | IF1, IF2, IF3 | aIF1, aIF1A, aIF2, aIF5B |
| Stop codons | UAA, UAG, UGA (RF1, RF2) | 3 stops (eRF1) | 3 stops (eRF1) | UAA, UAG (UGA = Trp!) | 3 stops | 3 stops (aRF1) |
| Coupling | Co-transcriptional | Decoupled (nuclear/cytoplasmic) | Decoupled | Co-transcriptional in matrix | Co-transcriptional | Co-transcriptional |
| Rate | ~20 aa/sec | ~10 aa/sec | ~6–10 aa/sec on ER, ~20 cytosol | ~5 aa/sec | ~10 aa/sec | ~10–20 aa/sec |
Real numbers
- Elongation rate: ~20 aa/sec in E. coli and mammalian cytosol; ~5–10 aa/sec on the ER and in mitochondria.
- Energy cost: ~4 high-energy phosphates per amino acid — 2 ATP for tRNA charging, 2 GTP for elongation (EF-Tu, EF-G).
- A typical mammalian cell makes ~10⁹ proteins per minute; protein synthesis consumes 30–50% of cellular energy.
- Misincorporation: ~1 in 10⁴. Frameshift: ~1 in 10⁵.
- Polysome density: one ribosome per ~80–100 nt on a heavily-translated mRNA.
- Human genome: ~20,000 protein-coding genes, producing >100,000 distinct proteins by splicing and PTMs.
- Each E. coli ribosome produces ~10 proteins/min; with ~70,000 ribosomes, that's ~700,000 proteins/min per cell.
Variants and drugs
- Selenocysteine (Sec, 21st) — encoded by UGA + SECIS element. Glutathione peroxidase, thioredoxin reductase, deiodinases.
- Pyrrolysine (Pyl, 22nd) — UAG-encoded in some methanogenic archaea.
- IRES translation — viral and stress-response mRNAs bypass cap-dependent initiation.
- Programmed -1 frameshifting — HIV gag-pol generates two proteins from one ORF.
- uORFs — short upstream ORFs gate downstream translation (GCN4/ATF4 stress response).
- Nonsense-mediated decay (NMD) — destroys mRNAs with premature stops; loss causes neurodevelopmental disorders.
- Inhibitors: cycloheximide (eEF2), puromycin (chain-terminator mimic), diphtheria toxin (ADP-ribosylates eEF2), ricin (depurinates 28S rRNA), fusidic acid (bacterial EF-G), mupirocin (bacterial isoleucyl-tRNA synthetase). Ataluren promotes therapeutic readthrough of premature stops in some Duchenne and CF cases.
Common misconceptions
- "The ribosome reads the genetic code." No — aminoacyl-tRNA synthetases do, by charging tRNAs correctly. The ribosome only checks codon-anticodon base pairs.
- "Met is always the first amino acid." The N-terminal Met is often cleaved post-translationally; mature proteins frequently start with the second residue.
- "All AUGs are start codons." Kozak context (eukaryotes) or Shine-Dalgarno (bacteria) selects the start; internal AUGs encode internal methionines.
- "The genetic code is universal." Mitochondria, ciliates, and mycoplasmas all reassign codons. Human mitochondria read UGA as tryptophan, not stop.
- "Faster ribosome = more protein." Often the opposite — local pauses at slow codons promote correct co-translational folding; speeding through risks misfolding.
- "Translation antibiotics are safe." They prefer bacterial 70S, but mitochondria use 70S-derived 55S — which is why aminoglycosides cause hearing loss and chloramphenicol causes aplastic anemia.
Frequently asked questions
How does translation start?
By assembling a ribosome on the start codon (AUG, coding for methionine). In bacteria, the small 30S subunit binds the Shine-Dalgarno sequence — a purine-rich stretch ~8 nt upstream of AUG that base-pairs with 16S rRNA. fMet-tRNA enters with initiation factors IF1, IF2-GTP, IF3; the 50S joins, GTP hydrolyzes, and elongation begins. In eukaryotes, the 40S binds the 5' cap with eIF4F, scans 5' → 3' until it finds a Kozak-context AUG, and then recruits the 60S — the cap-and-scan model. eIF2-GTP delivers initiator Met-tRNA. Internal ribosome entry sites (IRES) can bypass cap-dependent initiation, used by viruses and stress-response mRNAs.
How does the cell pick the right amino acid for each codon?
Aminoacyl-tRNA synthetases. There are 20 of them — one per amino acid — and each one recognizes both its amino acid and the correct tRNAs for that amino acid. The synthetase "charges" the tRNA by attaching the amino acid to its 3' end via an ATP-driven aminoacyl bond. Many synthetases also have proofreading (editing) sites that hydrolyze mischarged tRNAs. After charging, the tRNA itself does the codon recognition: its anticodon loop pairs with the mRNA codon at the ribosome A site. So the synthetases enforce the genetic code; the ribosome only reads what's already correct. Get one synthetase wrong and an entire amino acid gets systematically misincorporated.
How fast is translation?
About 20 amino acids per second in both bacteria and eukaryotes. A 300-residue protein takes ~15 seconds; titin (~30,000 residues) takes ~25 minutes. The rate-limiting step is GTP hydrolysis on EF-Tu (eEF1A in eukaryotes), not chemistry of bond formation. Elongation is also where the cell does most of its translational regulation — codon usage, tRNA availability, and mRNA secondary structure all modulate local speed, which affects co-translational folding.
How accurate is translation?
Misincorporation rate is about 1 in 10,000 per codon — roughly 1 wrong amino acid per protein for an average-length polypeptide. Two stages contribute: aminoacyl-tRNA synthetases proofread their charging reactions, and EF-Tu performs kinetic proofreading at the ribosome A site (a wrong tRNA dissociates faster than it accommodates). Without these mechanisms the error rate would be 100-fold worse. Translation is less accurate than DNA replication (~10⁻⁹) because protein errors don't propagate — a misfolded protein can be replaced; a mutated DNA cannot.
What's a polysome?
Multiple ribosomes simultaneously translating the same mRNA. The first ribosome leaves a clear path on the mRNA after ~80 nucleotides; the second initiates and follows. Polysomes are a sign of high translation demand. In bacteria, polysomes are often coupled to RNA polymerase that's still transcribing the mRNA — translation chases transcription in real time. In eukaryotes, mRNAs typically circularize via eIF4F-poly(A) interaction, so terminating ribosomes are recycled back to the 5' end. Polyribosome profiling (sucrose gradient centrifugation) is the standard technique for measuring translational activity.
What happens at a stop codon?
Release factors recognize UAA, UAG, or UGA in the A site. Bacteria use RF1 (for UAA, UAG) and RF2 (for UAA, UGA); RF3-GTP recycles them. Eukaryotes use a single eRF1 (recognizes all three) plus eRF3-GTP. The release factor stimulates the ribosome's peptidyl transferase center to hydrolyze the bond between the peptide and the P-site tRNA, releasing the finished protein. Then the recycling factors (RRF + EF-G in bacteria; ABCE1 in eukaryotes) split the ribosome back into subunits, which are reused. Premature stop codons trigger nonsense-mediated decay; readthrough mutations can be therapeutic — ataluren promotes readthrough of premature stops in some Duchenne muscular dystrophy and cystic fibrosis cases.