How Proteins Are Made: mRNA, Ribosomes, and the Translation Process

A muscle cell and a neuron carry essentially the same DNA, yet they look and behave nothing alike. The difference lies in which genes are read and which proteins get built. This article follows how proteins are made, from a stretch of DNA to a folded, working molecule. Protein synthesis runs in two stages: transcription copies a gene into messenger RNA (mRNA), and translation uses that message to string amino acids together at a ribosome.

Educational content, not medical advice. Talk to your doctor before changing any medication, treatment, or routine. Information here reflects current research but every person's situation differs.

Why DNA needs a middleman to build anything

DNA is famous, but it does not do much by itself. It is closer to a reference library than a factory: it stores information and rarely leaves the vault. The molecules that carry out the business of life are proteins. They catalyze reactions, give cells their shape, fight pathogens, and relay signals from one cell to the next.

Francis Crick described the central dogma of molecular biology in 1958: sequence information flows from DNA to RNA to protein, and not back out of protein. Retroviruses are the well-known exception to the DNA-to-RNA direction, because they carry an enzyme that copies RNA into DNA. For ordinary cellular genes, though, the road runs one way.

MedlinePlus, the US National Library of Medicine's genetics resource, describes the same two steps: transcription copies the gene's information into RNA in the nucleus, and translation, which "takes place in the cytoplasm," builds the protein one amino acid at a time. Together the two steps are called gene expression.

An analogy helps here. Think of a rare cookbook that can never leave the library. You photocopy one recipe (transcription), carry the copy to the kitchen, and a cook follows it step by step (translation). The original stays safe, and you can make as many copies of the recipe as you need. The photocopy is mRNA, and the cook is the ribosome.

Transcription: copying one gene into mRNA

Transcription begins when an enzyme called RNA polymerase binds a region of DNA known as the promoter, a sequence that marks where a gene starts. The polymerase unwinds the double helix locally and reads one strand, the template strand. It builds a matching strand of RNA, using RNA nucleotides in which uracil (U) stands in for the thymine (T) found in DNA.

In bacteria, which have no nucleus, the new mRNA is immediately available to ribosomes. In eukaryotes (organisms whose cells have a nucleus, including all animals, plants, and fungi), the first copy is a pre-mRNA that must be edited inside the nucleus before it is allowed out. Those edits are not decoration. Each one changes how long the message survives and how well it gets translated.

Three edits before mRNA leaves the nucleus

A eukaryotic pre-mRNA receives a cap, a tail, and a haircut. Each serves a different purpose.

Edit	What happens	Why it matters
5' cap	A modified guanine nucleotide is added to the front end	Protects the message from degradation and gives the translation machinery a handle to grab
Splicing	Non-coding introns are cut out and coding exons are joined	Produces a continuous reading frame and allows one gene to yield several proteins
Poly-A tail	A string of adenine nucleotides is added to the back end	Helps export, boosts translation, and sets the message's lifespan

The cap and the tail

The cap is recognized by a cap-binding protein called eIF4E, one piece of a complex that recruits ribosomes. The tail is added by an enzyme that keeps going until the tail reaches about 250 nucleotides, at which point it can no longer hold on to its partner protein. Poly(A)-binding protein (PABP) coats the tail, promotes export and translation, and inhibits degradation.

Tails do not last forever. In ordinary body cells the tails of cytoplasmic mRNAs gradually shorten, and shorter tails mean less translation and earlier destruction. Once the tail is gone, the cap is removed and the message is degraded, a process that can take many hours. This built-in decay is a feature: it stops a cell from making a protein long after the need has passed.

Splicing and the surprise of alternative splicing

Most eukaryotic genes are interrupted. The stretches that end up in the final protein, called exons, are separated by introns, which are cut out by a molecular machine called the spliceosome. The spliceosome is built from RNA and protein pieces (snRNPs named U1, U2, U4, U5, and U6). U1 recognizes the start of an intron, U2 binds a branch-point adenine, and the remaining pieces rearrange to cut and rejoin the strand.

The cell can choose different combinations of exons from the same gene, a process called alternative splicing. It was first observed in 1977 in adenovirus. It is widely believed that about 95% of multi-exon human genes undergo it, though some scientists argue that many of the observed variants are splicing errors and that the number of biologically meaningful cases is much lower. The extreme record belongs to a fruit fly gene called Dscam, which could in principle produce 38,016 variants.

This is why "one gene, one protein" is a slogan with an expiration date. The idea carried real weight when Beadle and Tatum proposed it, but modern genomics shows that a single gene can be a family of related messages.

The genetic code: 64 codons for 20 amino acids

A finished mRNA is read in groups of three nucleotides called codons. Four bases taken three at a time give 4 × 4 × 4 = 64 possible codons. The standard genetic code uses those 64 to specify the 20 standard amino acids plus start and stop signals. Sixty-one codons are "sense" codons that specify an amino acid, and three are stop codons: UAA (ochre), UAG (amber), and UGA (opal).

Because 61 codons share 20 amino acids, the code is redundant. Several codons can specify the same amino acid, but no codon specifies more than one, so the code has what the genetic-code literature calls redundancy but no ambiguity. The third letter of a codon is often the least important, so a change there frequently leaves the amino acid unchanged. Researchers have long argued that the code is arranged to limit the damage of mistakes, since similar amino acids tend to sit on similar codons.

Codon	Amino acid or signal	Note
AUG	Methionine	The usual start codon
UUU	Phenylalanine	First codon decoded in the lab, in 1961
UGG	Tryptophan	One of only two amino acids with a single codon in the standard code
GAA, GAG	Glutamic acid	Two synonymous codons
GUU, GUC, GUA, GUG	Valine	Four synonymous codons
UAA, UAG, UGA	Stop	Recognized by release factors, not by tRNAs

How the code was cracked

In 1961, Marshall Nirenberg and Heinrich Matthaei fed a synthetic RNA made only of uracil (UUUUU...) into a cell-free protein-making system and got a chain made only of phenylalanine. That single result assigned UUU. Severo Ochoa's laboratory soon assigned AAA to lysine and CCC to proline, and Har Gobind Khorana worked out much of the rest. Nirenberg and Philip Leder determined 54 of the 64 codons. In 1968, Robert Holley, Khorana, and Nirenberg shared the Nobel Prize in Physiology or Medicine for interpreting the code. The full story of the experiment is in our article on the Nirenberg genetic code experiment.

Almost universal, with exceptions

The same codon means the same amino acid in bacteria, yeast, and humans, which is strong evidence that all living things share a common ancestor. Almost is the key word. The first variant was found in human mitochondria in 1979. In the bacterium Mycoplasma, UGA codes for tryptophan instead of acting as a stop, and in a group of yeasts CUG codes for serine.

Two more amino acids come from recoded stop codons. UGA can specify selenocysteine when a hairpin structure called SECIS sits downstream, and UAG can specify pyrrolysine in some organisms. These are sometimes called the 21st and 22nd amino acids.

The cast at the ribosome

Translation needs three kinds of RNA working with a crowd of proteins: the mRNA message, transfer RNA adapters, and the ribosome itself.

tRNA: the adapter that links letters to amino acids

A transfer RNA (tRNA) is a small folded RNA, 74 to 93 nucleotides long. One end carries a three-nucleotide anticodon that pairs with a matching codon on the mRNA. The other end carries the amino acid that the codon specifies. The tRNA is a bilingual translator: it speaks nucleotide at one end and amino acid at the other.

Attaching the right amino acid to the right tRNA is the job of aminoacyl-tRNA synthetases. They come in two classes of ten enzymes each, and the charging reaction spends ATP: amino acid + tRNA + ATP gives an aminoacyl-tRNA + AMP + pyrophosphate. Some synthetases proofread their work. Valine and threonine have similar shapes, so the enzyme that handles valine includes an editing step that hydrolyzes a wrongly attached amino acid. This step matters because the ribosome checks only that the anticodon matches the codon. It cannot see which amino acid the tRNA carries.

Wobble: why 61 tRNAs are not required

If every sense codon needed its own tRNA, cells would need 61 types. In 1966 Crick proposed the wobble hypothesis: the first base of the anticodon, which pairs with the third base of the codon, is less tightly constrained, so one tRNA can recognize several synonymous codons. Most organisms have fewer than 45 tRNA types, and under Crick's scheme the theoretical minimum is 32.

The ribosome: a ribozyme the size of a virus

The ribosome is a molecular machine made of ribosomal RNA (rRNA) and proteins, assembled from a small and a large subunit. The two subunits clamp around the mRNA, and the small one does the decoding while the large one forms the peptide bonds. Bacterial ribosomes are 70S particles, built from a 30S small subunit (16S rRNA plus 21 proteins in E. coli) and a 50S large subunit (23S and 5S rRNA plus 31 proteins). Eukaryotic ribosomes in the cytoplasm are bigger, at 80S, built from 40S and 60S subunits.

The surprise inside is that the catalytic heart is RNA, not protein. The rRNA itself forms the peptide bonds, which makes the ribosome a ribozyme (an RNA enzyme). The proteins mostly stabilize the structure from the surface. That finding is one reason the ribosome features so often in discussions of RNA's early role in life's history.

George Palade first saw ribosomes with an electron microscope in the mid-1950s, and he shared the 1974 Nobel Prize in Physiology or Medicine with Albert Claude and Christian de Duve. The detailed atomic structure came later and earned Venkatraman Ramakrishnan, Thomas Steitz, and Ada Yonath the 2009 Nobel Prize in Chemistry.

Every ribosome has three tRNA docking sites:

• A site (aminoacyl): where a new tRNA with its amino acid enters, or where release factors bind at the end.
• P site (peptidyl): where the tRNA holding the growing chain sits.
• E site (exit): where the empty tRNA leaves.

In eukaryotes, ribosomes float free in the cytoplasm or sit on the rough endoplasmic reticulum (ER), which is rough precisely because ribosomes stud its surface. Mitochondria carry their own 70S ribosomes that resemble bacterial ones, a leftover from their bacterial ancestry covered in how mitochondria became the cell's powerhouses.

Translation step by step

Translation has three phases: initiation, elongation, and termination. Think of the ribosome as a reading head on a tape, except that here the tape is threaded through the machine rather than the machine sliding along it.

Initiation: finding the starting line

In eukaryotes, the cap-binding complex eIF4F does the first grab. It includes eIF4E (which binds the cap), eIF4A (a helicase that unwinds secondary structure), and eIF4G (a scaffold). PABP on the tail also binds eIF4G, which is thought to loop the mRNA into a circle and keep the machinery close to the message.

A small subunit carrying a special initiator tRNA, delivered by a factor called eIF2, binds near the cap and scans along the mRNA until it finds an AUG start codon. A further factor, eIF3, stops the large subunit from joining too early. When the start codon is found, the large subunit clamps on and the ribosome is ready. The initiator tRNA carries methionine (in bacteria, a modified form called formylmethionine), which is why nearly every new protein begins with that amino acid, although it is often trimmed off afterward.

Elongation: adding one amino acid at a time

This three-step cycle is repeated for each amino acid:

1. Decoding. An aminoacyl-tRNA enters the A site, and its anticodon pairs with the codon. A wrong match is rejected most of the time.
2. Peptide bond formation. The rRNA catalyzes a bond between the new amino acid in the A site and the growing chain in the P site, transferring the whole chain onto the tRNA in the A site.
3. Translocation. The ribosome advances exactly one codon. The tRNA holding the chain moves into the P site, the empty tRNA moves to the E site and leaves, and a fresh codon opens in the A site.

The cycle works like a zipper that closes one tooth at a time, always in the same direction. Nothing about it depends on which protein is being built, which is why the same machine can make an enzyme one minute and a structural protein the next.

Termination: the stop signal

There is no tRNA for UAA, UAG, or UGA. When one of them reaches the A site, a release factor binds instead. In eukaryotes a single factor called eRF1 recognizes all three stop codons, and a helper GTPase named eRF3 assists it in freeing the finished chain. Bacteria use two factors, RF1 and RF2, to divide the job. The chain is released, and the ribosome comes apart so that its subunits can be reused.

Many ribosomes can read a single mRNA at once, spaced along it like cars on a single-lane road. Such a cluster is called a polyribosome or polysome, and it lets one message yield many copies of its protein in parallel.

How fast, and how accurately

Translation is quick but not uniform across life. Bacterial ribosomes add up to 17 to 21 amino acids per second, while eukaryotic ribosomes manage up to 6 to 9 per second. A typical protein of 300 amino acids therefore takes about 15 seconds in a fast bacterium and roughly half a minute or more in a human cell. The common rule of thumb that human ribosomes manage 15 to 20 amino acids per second belongs to bacteria.

Feature	Bacteria	Eukaryotes (cytoplasm)
Ribosome	70S (30S + 50S)	80S (40S + 60S)
Elongation speed	Up to about 17 to 21 amino acids per second	Up to about 6 to 9 amino acids per second
First amino acid	Formylmethionine	Methionine
Release factors	RF1 and RF2	eRF1 with eRF3
Location	No nucleus; message is available at once	Ribosomes in cytoplasm and on rough ER

Accuracy is the other half of the story. Estimates of how often a wrong amino acid slips in range from about one in 100,000 to one in 1,000 codons, depending on conditions, and ribosomes abandon a message early about once in 10,000 codons. For a 300-residue protein, that works out to a premature stop in roughly 3% of attempts, which a cell tolerates because it makes so many copies. Defective chains are tagged and recycled.

From chain to machine: protein folding

A newly made protein is a floppy string. To work, it must fold into a precise three-dimensional shape, and the shape is dictated largely by the order of its amino acids. Water-repelling (hydrophobic) amino acids tuck into the interior away from the watery cell, while water-attracting (hydrophilic) ones stay on the surface. Hydrogen bonds, ionic bonds, and disulfide bridges lock the arrangement in place.

Local patterns come first. Alpha helices are coiled ribbons and beta sheets are flat arrangements of strands, and these are combined into the overall tertiary structure. Some proteins also link several chains into a quaternary structure. Folding can start while the chain is still emerging from the ribosome, a process called co-translational folding.

Levinthal's paradox

In 1969 Cyrus Levinthal pointed out that an unfolded chain has an astronomical number of possible shapes. One of his estimates was 3³⁰⁰, or about 10¹⁴³. Sampling them one at a time would take an impractically long time even at nanosecond speeds, yet real proteins fold in milliseconds, and the fastest reactions finish in a few microseconds. Levinthal concluded that folding must follow guided pathways through intermediate states, much as water runs down a funnel-shaped hill rather than sampling every possible route.

Chaperones: folding assistants

In the crowded cell, partly folded chains can stick to each other. Molecular chaperones bind exposed hydrophobic patches, prevent aggregation, and give a chain time to find its shape. They do not supply the answer; they prevent wrong turns. Heat shock proteins are a major chaperone family and have been found in every species examined, from bacteria to humans. Well-studied examples are the Hsp70 family and the barrel-shaped GroEL/GroES complex, both of which use ATP. Energy for the whole business comes from molecules like those described in how ATP powers every living cell.

AlphaFold and the prediction problem

Predicting a protein's shape from its sequence was a grand challenge for half a century. In 2020, DeepMind's AlphaFold placed first in the CASP structure-prediction contest and scored above 90 on the global distance test for around two-thirds of the proteins. The caveat is that predicting the final shape is not the same as explaining how the chain gets there. Our deeper coverage is in AlphaFold and protein folding AI.

Changes after translation

Many proteins are modified after synthesis in ways that change what they do, where they go, or how long they last.

• Phosphorylation, the addition of a phosphate group, can switch enzymes on or off and underlies many signaling pathways.
• Glycosylation, the addition of sugar chains, is typical of proteins bound for the cell surface or for secretion.
• Ubiquitination tags a protein for destruction by the proteasome, the cell's disposal system.
• Cleavage trims a protein into its active form.

Insulin is the classic cleavage example. It is translated as a 110-amino-acid precursor, preproinsulin, directly into the rough ER. A signal peptide is removed to give proinsulin, which folds with three disulfide bonds. Enzymes then excise the middle segment, called the C-peptide, leaving a mature hormone made of a 21-amino-acid A chain and a 30-amino-acid B chain. The clinical side is in insulin and diabetes explained.

When translation or folding goes wrong

Because proteins do the work, a single changed amino acid can have large effects.

In sickle cell disease, a point mutation in the HBB gene on chromosome 11 replaces glutamic acid with valine at position 6 of the beta-globin chain. One amino acid out of roughly 146 is different, and that is enough to change how the hemoglobin behaves.

In cystic fibrosis, the most common mutation, F508del, deletes one phenylalanine at position 508 of the CFTR protein. The chain does not fold normally, is not trafficked to the cell membrane, and is destroyed. F508del accounts for about 70% of cystic fibrosis cases worldwide.

More broadly, failures of folding or clearance are linked to Alzheimer's, Parkinson's, and Huntington's disease. At the extreme are prions, misfolded proteins that induce normal copies to misfold.

Antibiotics that target bacterial ribosomes

Bacterial and eukaryotic ribosomes differ enough that several antibiotics hit the first while mostly sparing the second. The mechanisms below follow the standard summary of protein synthesis inhibitors.

Antibiotic	Target	What it blocks
Tetracycline	30S subunit	Blocks the A site so aminoacyl-tRNAs cannot enter
Streptomycin	30S subunit	Disrupts assembly and proofreading, causing misreading and early stops
Erythromycin	50S subunit	Inhibits peptidyl transfer (peptide bond formation) during elongation
Chloramphenicol	50S subunit	Blocks peptide bond formation, and also acts on mitochondrial ribosomes

Chloramphenicol's effect on mitochondria is a direct consequence of their bacterial-like ribosomes. The bigger public-health problem is that bacteria evolve around these drugs, a topic covered in how antibiotic resistance spreads.

Common confusions about protein synthesis

"The ribosome reads DNA." It does not. Ribosomes read mRNA only. DNA is read by RNA polymerase.

"Every codon codes for an amino acid." Three do not. The stop codons are recognized by proteins (release factors), not by tRNAs.

"A silent mutation does nothing." Changing the third letter of a codon often leaves the amino acid the same, which is why such changes are called synonymous. That does not mean every such change is irrelevant in every context, since mRNA sequences also affect splicing and stability.

"The protein is finished when the ribosome lets go." For many proteins, the released chain is raw material. It still has to fold, may be cut, decorated, or tagged, and has to be delivered to the right place.

"mRNA lives forever." Messages have lifespans set by tail length and other factors, and the vaccines and drugs that use mRNA depend on the same machinery. See how mRNA vaccines work for that application.

What this means

Protein synthesis is a chain of quality-controlled hand-offs: DNA to mRNA, mRNA to ribosome, codon to anticodon, chain to fold. Errors can enter at each hand-off, and the cell has a countermeasure at nearly all of them, from synthetase proofreading to chaperones to the proteasome. The result is a process fast enough to build a 300-amino-acid protein in under a minute and accurate enough that most copies work.

The deeper lesson is that a gene is not a finished product. It is a recipe whose outcome depends on editing, timing, and folding. That is why the same DNA can yield a neuron and a muscle cell, and why a single misplaced amino acid can cause disease.

Key Takeaways

• The central dogma: DNA to mRNA (transcription) to protein (translation).
• A eukaryotic mRNA gets a 5' cap, a poly-A tail, and has its introns spliced out before export. About 95% of multi-exon human genes are estimated to undergo alternative splicing.
• The genetic code uses 64 codons, 61 for amino acids and three for stop, and it is nearly universal.
• tRNAs pair anticodons with codons, aminoacyl-tRNA synthetases load the correct amino acid, and the ribosome's rRNA forms the peptide bonds.
• Bacterial ribosomes add up to 17 to 21 amino acids per second and eukaryotic ones up to 6 to 9.
• Chaperones help new chains fold, and misfolding underlies diseases such as cystic fibrosis.

FAQ

Q: What is the difference between transcription and translation? A: Transcription copies a gene's DNA sequence into mRNA, in the nucleus of eukaryotic cells. Translation reads that mRNA at a ribosome and builds a protein from amino acids. Both are parts of gene expression.

Q: Why do eukaryotic genes have introns? A: Why introns exist is still debated. Their practical effect is that cells can splice exons in different combinations, so one gene can yield several protein variants. Some introns also contain regulatory sequences or give rise to small functional RNAs.

Q: What happens to misfolded proteins? A: Chaperones first try to refold them. Chains that stay misfolded are usually tagged with ubiquitin and destroyed by the proteasome. If that system is overwhelmed, misfolded proteins can aggregate, as in several neurodegenerative diseases.

Q: How do antibiotics target translation? A: Several antibiotics bind bacterial 70S ribosomes, which differ structurally from eukaryotic 80S ribosomes. Tetracycline blocks the A site, streptomycin causes misreading, and erythromycin and chloramphenicol interfere with peptide bond formation. Chloramphenicol also affects mitochondrial ribosomes.

Q: What is alternative splicing and why does it matter? A: Alternative splicing joins different combinations of exons from the same gene, producing protein variants with different functions. About 95% of multi-exon human genes are estimated to be spliced this way, although how many of those variants are functionally meaningful is still being worked out.

Q: How fast is translation? A: Bacterial ribosomes add up to 17 to 21 amino acids per second, and eukaryotic ribosomes up to 6 to 9. A 300-amino-acid protein therefore takes tens of seconds, and multiple ribosomes on one mRNA multiply the output.

Q: Do all proteins start with methionine? A: Almost all new chains start with methionine because AUG is the usual start codon and the initiator tRNA carries it. In bacteria it is a modified form, formylmethionine. The first amino acid is often removed later, so the mature protein may begin with something else.

Sources

• MedlinePlus Genetics: How do cells make proteins from genes?
• NHGRI Genetic Code glossary entry
• Translation (biology), Wikipedia
• Ribosome, Wikipedia
• Genetic code, Wikipedia
• Alternative splicing, Wikipedia
• Polyadenylation, Wikipedia
• Wobble base pair, Wikipedia
• Aminoacyl-tRNA synthetase, Wikipedia
• Protein folding, Wikipedia
• Protein synthesis inhibitor, Wikipedia
• Sickle cell disease, Wikipedia
• Cystic fibrosis, Wikipedia
• Insulin, Wikipedia

---

Read next:

• How Cells Divide: Mitosis, Meiosis, and the Cell Cycle Explained
• How Photosynthesis Works: Light, Dark Reactions, and the Calvin Cycle
• How Does DNA Work? The Molecular Blueprint of Life Explained