Deciphering the Genetic Code
In May 1961, a young NIH biochemist named Marshall Nirenberg mixed a synthetic RNA into a tube of broken-open bacterial cells. The RNA was made of nothing but uracil. The tube made protein, and that protein was pure polyphenylalanine. This was a chain of one single amino acid. That result cracked open one of biology's deepest ciphers within five years. It answered a question that had haunted molecular biology since the structure of DNA was solved. How does a sequence of four nucleotide letters specify a sequence of twenty amino acid letters? The problem was not trivial. DNA had already been shown to be the hereditary material. The Central Dogma had established that genetic information flows from DNA to RNA to protein. But knowing the direction of information flow says nothing about the actual dictionary used for translation. Theorists including Francis Crick and the physicist George Gamow had proposed elaborate, mathematically elegant schemes. These described how nucleotide triplets might encode amino acids, but none had experimental support. The code could in principle have been overlapping or non-overlapping, comma-free or comma-containing. There was no guarantee it was even a triplet code. This held true until genetic experiments by Crick, Barnett, Brenner, and Watts-Tobin in 1961. Those experiments established that three bases specify one amino acid. What Nirenberg and his colleague Heinrich Matthaei did was turn a purely theoretical puzzle into a solvable biochemical problem. They built a cell-free protein-synthesizing system. They fed it synthetic messenger RNAs of known and controlled sequence. Then they read out which amino acid got incorporated into protein. This was the genetic code equivalent of the Meselson-Stahl experiment in its decisiveness. It offered a simple, physically direct readout that could not be argued away by competing theories. Over the next five years, this approach was refined further. Nirenberg's group and Har Gobind Khorana's chemical synthesis of defined RNA sequences both contributed. Together they assigned all 64 codons to their amino acids or stop signals. This article walks through exactly how the poly-U experiment worked. It explains why it was such a clean test. It shows how the follow-up triplet-binding assay let the whole code be filled in systematically. It also covers what pitfalls the experimenters had to route around. It sets up the next layers of the story. These include the code's remarkable structural logic (redundancy, near-universality). They also include the wobble rules that explain why tRNA numbers do not match codon numbers.
- The genetic code maps 64 possible mRNA triplets (codons) onto 20 amino acids plus stop signals; it needed to be determined experimentally, not just theorized.
- Nirenberg and Matthaei (1961) added synthetic poly-U RNA to a cell-free E. coli extract and got a polypeptide made entirely of phenylalanine, proving UUU = phenylalanine.
- The cell-free system used real ribosomes, tRNAs, amino acids, and energy sources from lysed bacteria, with translation driven by whatever mRNA was added.
- Nirenberg and Philip Leder's 1964 triplet-binding assay used short, defined trinucleotides to trigger ribosome-tRNA binding, letting them assign codons without needing full-length synthesis of every RNA.
- Har Gobind Khorana's organic-chemical synthesis of repeating RNA sequences (e.g., poly-UC, poly-UUC) filled in remaining codons and confirmed the triplet, non-overlapping, comma-free reading frame.
- By 1966 all 64 codons were assigned, revealing a redundant (degenerate) code that later analysis showed is nearly universal across life, setting up the study of Wobble Hypothesis and codon usage.
Feeding chemically defined synthetic mRNAs into a cell-free translation system and reading out the resulting protein sequence let Nirenberg, Matthaei, Leder, and Khorana directly assign each of the 64 mRNA codons to its amino acid or stop signal.
The core idea
The genetic code is the fixed correspondence between nucleotide triplets in messenger RNA, called codons, and the amino acids incorporated into protein during translation. Because there are four bases (A, U, G, C) and codons are three bases long, there are 4³ = 64 possible codons. There are only 20 standard amino acids used in proteins, plus the need for start and stop signals. This immediately implies redundancy: most amino acids are specified by more than one codon, a property called degeneracy. Before 1961, this arithmetic was understood in principle, but nobody knew which triplets meant which amino acids. The code itself, the dictionary, was an empirical unknown. Deciphering the code required a system in which one could control the input (the mRNA sequence) and directly observe the output (the protein made), stripped of the complexity of a living, dividing cell. The Nirenberg-Matthaei breakthrough was to build exactly such a system: a cell-free extract of E. coli that retained all the translation machinery, ribosomes, transfer RNAs, aminoacyl-tRNA synthetases, initiation and elongation factors, and ATP/GTP-generating enzymes, but had no intact genome directing its own protein synthesis. Left alone, this extract's endogenous mRNA would eventually degrade and residual translation would stop. Add a synthetic RNA of known, simple composition, and whatever protein appears must have been directed by that synthetic message. This is the essence of an in vitro translation system, a tool still central to modern molecular biology and biotechnology. The second conceptual pillar, established slightly earlier by genetic means, was that the code is read in triplets, non-overlapping, and without punctuation between codons (a "comma-free," fixed-frame reading). Crick, Brenner, and colleagues showed this using frameshift mutations in the bacteriophage T4 rII gene: single insertions or deletions of one or two bases destroyed the gene's function, but insertions or deletions of three bases (or combinations that summed to a multiple of three) often restored a functional, near-normal protein. This told them the ribosome reads a message in fixed groups of three nucleotides from a defined start point, without overlap, so a single-base insertion shifts every downstream codon and garbles the whole message. That triplet, non-overlapping frame is the grammatical rule; Nirenberg's experiments then filled in the actual vocabulary, codon by codon.
Mechanism
Step 1: Building the cell-free translation system. Nirenberg and Matthaei grew E. coli to log phase, then broke the cells open (using a French press or grinding with alumina) to release their cytoplasmic contents without denaturing the machinery inside. Centrifugation removed intact cells and debris, leaving an "S-30" extract, the supernatant after spinning at roughly 30,000 × g, that contains ribosomes, tRNAs, aminoacyl-tRNA synthetases, translation factors, and residual endogenous mRNA. The extract was pre-incubated to let endogenous mRNA finish being translated and degrade, run down the background "noise" of the system. To this exhausted extract, the experimenters added the essential raw materials for protein synthesis: a mix of the 20 amino acids, ATP and GTP (energy for aminoacyl-tRNA charging and ribosome translocation), and Mg²⁺, which is critical for ribosome and codon-anticodon stability.
Step 2: The radioactive readout. To measure protein synthesis quantitatively, they ran twenty parallel reaction tubes. In each tube, one of the 20 amino acids was radioactively labeled (with ¹⁴C), while the other 19 were present unlabeled. If a synthetic mRNA directed incorporation of a particular amino acid into protein, that tube's protein, precipitated with hot trichloroacetic acid (TCA) and collected on a filter, would carry the radioactive label; the label would show up as protein-bound, TCA-precipitable radioactivity. If that amino acid was not being used, the tube would show only background counts.
Step 3: Adding poly-U as the test message. The key experiment used synthetic polyuridylic acid, poly-U, an RNA made entirely of uracil bases, synthesized enzymatically using the enzyme polynucleotide phosphorylase (discovered by Severo Ochoa's group), which can polymerize nucleotide diphosphates without needing a template. Poly-U has an unambiguous, maximally simple sequence: every codon read from it, in any frame, is UUU. When poly-U was added to the 20 parallel extracts on 27 May 1961, only the tube containing radioactive phenylalanine showed a massive spike in TCA-precipitable counts, more than 15-fold above background in the original experiment. All other 19 tubes stayed near baseline. The inference was immediate and unambiguous: UUU codes for phenylalanine.
Step 4: Extending to other homopolymers. The same logic was repeated with poly-A (polyadenylic acid) and poly-C (polycytidylic acid), synthesized the same enzymatic way. Poly-A directed incorporation of lysine, showing AAA = lysine. Poly-C directed incorporation of proline, showing CCC = proline. Poly-G was harder to work with because it forms unusual secondary structure (tetraplexes) that impaired its use as a template, an early hint that RNA structure, not just sequence, matters for translation, a theme that recurs throughout molecular biology.
Step 5: Mixed copolymers to break the degeneracy. Homopolymers gave only four codon assignments (for the four "pure" triplets), but the code has 64 codons to fill. Nirenberg's group, and independently Ochoa's group, next used polynucleotide phosphorylase to make random copolymers with defined, controlled ratios of two bases, for instance poly-UC made with 3 parts U to 1 part C. Because the enzyme incorporates bases roughly according to their relative concentration in the reaction, this generates an RNA with a statistically predictable mix of codons: mostly UUU, but also UUC, UCU, CUU, and rarer CCU, CUC, UCC, CCC, each present at a calculable frequency based on the input ratio. By comparing the amount of each amino acid incorporated with the calculated frequency of each possible codon composition, the researchers could infer probable assignments, though this method gave the base composition of a codon (which three bases) without pinning down their exact order. This ambiguity, knowing that a codon contained "two U and one C" without knowing whether it was UUC, UCU, or CUU, was the central limitation of the random copolymer approach.
Step 6: The triplet-binding assay resolves the ordering problem. In 1964, Nirenberg, working with Philip Leder, devised a second, more surgical technique that sidestepped the need to synthesize full-length, ordered mRNAs and did not require actual polypeptide synthesis at all. They discovered that a short synthetic RNA trinucleotide, just three bases long, could bind to a ribosome and, if it matched a tRNA's anticodon, cause that specific aminoacyl-tRNA to bind the ribosome-mRNA complex. This ribosome-tRNA-trinucleotide ternary complex was large enough to be retained on a nitrocellulose filter, whereas free aminoacyl-tRNA passed through. So the assay worked like this: mix ribosomes, a defined trinucleotide (say, GUU), and a single radioactively labeled aminoacyl-tRNA species (say, ¹⁴C-valyl-tRNA), then filter. If the trinucleotide's sequence matched the anticodon of that tRNA, the radioactive complex was retained on the filter and counted by scintillation. By testing each of the possible trinucleotides, made accessible through organic synthesis, against each labeled aminoacyl-tRNA, they could assign specific codons unambiguously, resolving the order problem that copolymers left open. This "trinucleotide binding assay" assigned roughly 50 of the 64 codons.
Step 7: Khorana's chemical synthesis fills in the rest. Har Gobind Khorana's laboratory took a complementary, more laborious route: organic synthesis of RNA molecules with precisely known, repeating sequences, such as poly-UG (alternating UGUGUG...) or repeating trinucleotide units like poly-UUC. Because the exact sequence was chemically defined rather than statistical, translating these RNAs in the cell-free system gave unambiguous readouts. A repeating dinucleotide like poly-UG, read in three possible frames, gives repeating codons UGU-GUG-UGU... or GUG-UGU-GUG..., producing an alternating copolypeptide, and comparing observed protein sequences across different repeat lengths (di-, tri-, and tetranucleotide repeats) let Khorana's group cross-validate and complete the code. By 1966, the codon assignments were essentially complete, including identification of the three stop codons (UAA, UAG, UGA) as sequences that caused ribosomes to halt and release the polypeptide rather than incorporate any amino acid.
The evidence
The founding experiment: Nirenberg and Matthaei, May 1961. Working in Nirenberg's small NIH laboratory, Heinrich Matthaei carried out the poly-U experiment described above. The result was stark: the tube with radioactive phenylalanine and poly-U showed roughly 15- to 20-fold more TCA-precipitable radioactivity than any control, while poly-U added to a tube without other amino acids, or amino acids added without poly-U, produced only background counts. Crucially, the control tubes, no added synthetic RNA, or poly-U with a different labeled amino acid, all stayed flat. This ruled out the trivial alternative explanations: that the extract was simply degrading or that phenylalanine was being incorporated non-specifically. The experiment was presented at the International Congress of Biochemistry in Moscow in August 1961, where Nirenberg's short, initially under-attended talk caused a sensation once Crick, in the audience, recognized its importance and arranged for it to be given a second, larger hearing.
Why the poly-U result was so persuasive. The logic of the experiment mirrors the logic of the best classic molecular biology experiments, such as the Hershey-Chase experiment or Meselson-Stahl: a single manipulable variable (which RNA is added, which amino acid is labeled) produces an unambiguous, quantitative, binary readout (radioactivity precipitated or not). There was no room for a "maybe": either a tube lit up or it did not. Because poly-U's sequence is unambiguous in any reading frame (it is always UUU regardless of where you start reading, unlike a mixed sequence where frame matters), the experiment did not even require prior knowledge of the reading frame to give a clean answer for this one codon.
Confirmatory and extending experiments. The result was rapidly and independently reproduced. Severo Ochoa's laboratory, using the same polynucleotide phosphorylase chemistry (an enzyme Ochoa's group had itself extensively characterized), obtained consistent results and extended the copolymer approach, sometimes racing Nirenberg's group to particular assignments. The subsequent triplet-binding assay (Nirenberg and Leder, 1964) and Khorana's repeating-sequence synthesis (throughout 1964-1966) were independent methods using different chemistry and different logic, yet they converged on the same codon table. This convergence across three technically distinct approaches, statistical copolymer incorporation, direct trinucleotide-ribosome-tRNA binding, and defined repeating chemical synthesis, is what gave the scientific community full confidence that the code table was correct rather than an artifact of any one method's quirks.
What was proved, precisely. These experiments proved three things simultaneously: (1) that specific nucleotide sequences in mRNA determine specific amino acids in protein, empirically validating the whole hypothesis of a nucleic-acid-to-protein code; (2) the actual dictionary, which triplets mean which amino acids, for all 64 codons; and (3) that a cell-free extract, lacking any intact chromosome, could faithfully translate an artificial message according to the same rules used by living cells, showing the code is a property of the translation machinery (ribosomes, tRNAs, synthetases) itself, not of the particular gene or organism being read. Later sequencing of natural genes and comparison to their protein products (beginning with things like the coat protein of MS2 phage) confirmed that the codon table derived from these synthetic-RNA experiments correctly predicted real, naturally occurring mRNA-to-protein translations, closing the loop between artificial reconstruction and natural biology.
Worked examples
Example 1: Predicting amino acid ratios from a mixed copolymer. Suppose polynucleotide phosphorylase is used to make a random RNA copolymer from a reaction mix containing UTP and GTP in a ratio of 4:1 (i.e., 80% U, 20% G), with bases incorporated independently and randomly at that ratio. What fraction of codons should be UUU, and what fraction should contain exactly two U and one G (in any order)? For a random copolymer, the probability of any specific triplet sequence is the product of the individual base probabilities, since incorporation is random and independent. P(UUU) = 0.8 × 0.8 × 0.8 = 0.512, so 51.2% of codons are UUU. For codons with exactly two U's and one G, there are three possible orderings: UUG, UGU, GUU. Each has probability 0.8 × 0.8 × 0.2 = 0.128. Summed across the three orderings: 3 × 0.128 = 0.384, so 38.4% of codons have this composition (though the copolymer experiment alone cannot tell you which ordering corresponds to which amino acid, only the triplet-binding assay or Khorana's defined sequences can). If this poly-UG (4:1) RNA is translated and the resulting protein is found to be roughly 51% phenylalanine, about 38% of a mix of amino acids consistent with a 2U:1G codon, and small remaining fractions matching rarer compositions, this is exactly consistent with UUU = Phe (matching the calculated 51.2%) and tells you that whichever amino acids appear at about 12.8% each individually (recall each of UUG, UGU, GUU occurs at 12.8%) are candidates for those specific orderings, to be resolved definitively by the trinucleotide-binding assay. This calculation is precisely the kind of reasoning Nirenberg, Ochoa, and their collaborators did in 1962-1963 to narrow down candidate codons before Leder's binding assay nailed down exact sequences.
Example 2: Reading a frameshift result to test the triplet, non-overlapping code. Consider a hypothetical (simplified, in the spirit of the actual Crick-Brenner rII experiments) short coding sequence and a mutant that has a single extra base (a "+1" insertion) inserted early in the gene: ``` Wild type mRNA: AUG GCU UUC AAG UGA Met Ala Phe Lys STOP (normal 4-residue protein + stop) +1 insertion mutant: AUG GGC UUU CAA GUG A... Met Gly Phe Gln Val (frame shifted; reads into what was formerly "downstream") ``` Because the ribosome reads in fixed, non-overlapping groups of three starting from the AUG start codon, inserting one extra base after the second codon shifts every subsequent triplet boundary by one position. The result is a completely different, garbled amino acid sequence downstream of the insertion, and, critically, the original in-frame UGA stop codon is destroyed (it's now split across two shifted triplets, GUG-A...), so translation runs on into sequence that used to be untranslated, likely hitting a new stop codon much later or not at all. This is exactly the logic Crick, Barnett, Brenner, and Watts-Tobin used with rII mutants in T4 phage: they showed that single '+1' or '+1' combined with a second '-1' insertion nearby could restore a functional (or partially functional) protein, because the frame is only disrupted between the two mutations and correctly restored downstream of the second, compensating mutation. Three insertions together (or three deletions), summing to a net change of zero mod 3, also restored function, confirming triplets, not doublets or quadruplets, are the unit read by the ribosome. This genetic proof of a triplet, non-overlapping, fixed-frame code is what made the subsequent codon-by-codon assignments by Nirenberg's group interpretable as a definite table rather than an ambiguous pattern.
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "The genetic code was deduced purely by theoretical reasoning about codon combinatorics." | Theorists (Gamow and others) proposed many elegant but ultimately incorrect schemes; the actual dictionary required direct biochemical experiments (poly-U, triplet binding) to determine. |
| "Poly-U experiments proved the code is a triplet code." | That was established separately, by genetic frameshift experiments (Crick, Brenner, Barnett, Watts-Tobin, 1961). Poly-U assumed a triplet frame and used it to assign UUU specifically. |
| "Each amino acid has exactly one codon." | The code is degenerate: most amino acids have 2-6 synonymous codons (e.g., leucine has six), which is why 64 codons map onto only 20 amino acids plus stop signals. |
| "The cell-free system just showed RNA can bind ribosomes, not that it directs specific protein sequence." | The radioactive-amino-acid readout specifically showed which amino acid was incorporated into TCA-precipitable protein, directly linking RNA sequence to protein composition, not mere binding. |
| "The genetic code is completely universal with zero exceptions." | It is nearly universal, but real exceptions exist, e.g., some codon reassignments in mitochondrial genomes and a few ciliate species, discovered decades after the original code table. |
The scientist behind it
Marshall W. Nirenberg trained as a biochemist, earning his PhD at the University of Michigan studying hexokinase, before joining the National Institutes of Health, where he pivoted toward the then-wide-open problem of the genetic code. Working with almost no institutional fanfare in a modest NIH lab, he and postdoctoral fellow Heinrich Matthaei spent long hours through 1960-1961 building and troubleshooting the cell-free E. coli translation system, an unglamorous, technically exacting effort that many larger, more famous labs (including Ochoa's) were also racing to complete. Nirenberg's insistence on rigorous, quantitative controls, twenty parallel tubes, one labeled amino acid each, is what gave the poly-U result its unambiguous, immediately convincing character rather than a merely suggestive one.
Origins · 1 minMarshall Warren Nirenberg and how this was discovered
Nirenberg shared the 1968 Nobel Prize in Physiology or Medicine with Har Gobind Khorana and Robert Holley for cracking the genetic code and elucidating its role in protein synthesis. He later shifted his research toward neuroscience and developmental biology at NIH, where he remained for his entire career.
Where this matters today
The codon table deciphered by this work underlies essentially every technology that reads or writes genetic information for human benefit. In medicine, understanding codon-to-amino-acid mapping is fundamental to interpreting the effects of point mutations, whether a DNA change is silent (synonymous, no amino acid change, thanks to code degeneracy), missense (changes the amino acid, as in sickle-cell disease where a single codon change converts glutamate's codon to valine's), or nonsense (creates a premature stop codon), each with different clinical consequences and different therapeutic strategies (nonsense-suppression drugs, for instance, are designed specifically to make ribosomes read through premature stop codons). In biotechnology, recombinant protein production, from insulin to monoclonal antibodies to mRNA vaccines, depends on codon optimization: choosing synonymous codons that match the tRNA abundance of the host expression system (bacterial, yeast, or mammalian cells) to maximize translation efficiency, a direct engineering application of code degeneracy. The mRNA vaccine platforms developed against SARS-CoV-2 are a particularly direct descendant of this work: designing a synthetic mRNA that reliably directs ribosomes, in human cells, to synthesize a specific viral spike protein sequence requires exact knowledge of the codon table and codon-optimization principles first established through Nirenberg's cell-free system. In agriculture and synthetic biology, engineers routinely recode genes with synonymous, host-preferred codons to boost expression of transgenes in crop plants, or use expanded/alternative genetic codes (incorporating non-standard amino acids via reassigned stop codons) to build proteins with novel chemical properties. In evolutionary biology and forensics, the near-universality of the code (with known, catalogued exceptions in mitochondria and some microbes) is a foundational piece of evidence for common descent, and codon-level sequence comparison underlies molecular phylogenetics, DNA barcoding for species identification, and paternity/forensic DNA matching.
Ask the tutor: How did researchers use codon-usage frequency data from the copolymer experiments to make quantitative predictions before the triplet-binding assay confirmed exact codon order?
Why couldn't scientists just theoretically predict the genetic code before doing these experiments?
Knowing that codons must be triplets (4³ = 64 possibilities for 20 amino acids) tells you nothing about which specific triplet corresponds to which specific amino acid; that correspondence is a historical, biochemical fact fixed early in the evolution of life, not something derivable from pure logic or mathematics. Several theorists, notably George Gamow, proposed mathematically elegant "comma-free" or overlapping coding schemes in the 1950s that turned out to be entirely wrong once tested. The actual dictionary had to be read out empirically, by feeding defined synthetic messages into a translation system and observing what protein resulted, exactly as Nirenberg, Matthaei, Leder, and Khorana did between 1961 and 1966.
What exactly is in a "cell-free" or "in vitro" translation system, and why does it work without a living cell?
An in vitro translation system is a lysate (broken-open cell extract) containing all the molecular machinery needed for protein synthesis, ribosomes, transfer RNAs, aminoacyl-tRNA synthetases, initiation/elongation/release factors, ATP and GTP for energy, and Mg²⁺ and other ions, but no intact chromosome directing its own gene expression. Translation does not require an intact, living cell membrane or nucleus; it only requires these biochemical components working together. Once residual endogenous mRNA is exhausted or degraded, adding any exogenous mRNA (natural or synthetic) will be translated using this borrowed machinery, exactly as it would inside a living cell. Modern versions of this same principle (e.g., the PURE system, wheat germ extract, rabbit reticulocyte lysate) remain workhorse tools in labs today.
Why did poly-G give unreliable results compared to poly-U, poly-A, and poly-C?
Polyguanylic acid (poly-G) tends to fold into unusual, highly stable secondary structures called G-quartets or tetraplexes, where four guanine bases hydrogen-bond together in a planar arrangement, stacking into a stable helical structure quite unlike the extended single-stranded conformation ribosomes expect. This aberrant structure impaired poly-G's ability to be bound and read by ribosomes in the cell-free system, making results with poly-G unreliable and inconsistent compared to the other three homopolymers. This was an early, somewhat serendipitous demonstration that RNA secondary structure, not just primary sequence, can affect translation, a principle that later became central to understanding riboswitches, ribosomal RNA structure, and mRNA secondary structure effects on translation efficiency.
How did the discovery relate to the earlier proof, from Crick and colleagues, that the code is triplet and non-overlapping?
The Crick-Brenner-Barnett-Watts-Tobin frameshift experiments (1961) used purely genetic methods, insertion and deletion mutations in bacteriophage T4's rII gene, to show that the reading frame is fixed, triplet, and non-overlapping, without ever identifying which triplet meant which amino acid. That was a grammatical proof: it established the rules of reading, not the vocabulary. Nirenberg's biochemical experiments then assumed this triplet-frame rule and used it to determine the actual vocabulary, which specific triplet corresponds to which specific amino acid. The two lines of evidence are complementary: one genetic, proving the reading rule; one biochemical, filling in the dictionary under that rule.
Did every codon assignment come from Nirenberg's lab, or did others contribute significantly?
Several groups contributed essential, independent pieces. Severo Ochoa's laboratory (which had earlier characterized polynucleotide phosphorylase, the very enzyme used to make synthetic RNAs) ran parallel copolymer experiments and made significant codon assignments, sometimes in direct competition with Nirenberg's group. Har Gobind Khorana's laboratory contributed a technically distinct and crucial method: organic chemical synthesis of RNAs with exactly known, repeating sequences, which resolved ordering ambiguities that copolymer experiments left unsettled and helped complete and cross-validate the whole table. The 1968 Nobel Prize recognized this shared, converging effort, awarded jointly to Nirenberg, Khorana, and Robert Holley (who separately determined the first complete tRNA sequence, revealing the anticodon loop).
Is the genetic code exactly universal across all living things?
It is nearly universal but not absolutely so, a nuance discovered well after the original 1960s work. The "standard" code deciphered by Nirenberg, Ochoa, and Khorana applies to the vast majority of nuclear genomes across bacteria, archaea, and eukaryotes. However, mitochondrial genomes in many species use a slightly different code (for example, human mitochondria read UGA as tryptophan rather than as a stop codon), and a handful of organisms, including some ciliate protozoa, reassign certain codons in their nuclear genomes as well. These exceptions are thought to reflect code evolution after the code's core structure was already established very early in life's history, and they do not undermine the overall near-universality that supports common ancestry.
What comes next after knowing the codon table itself?
Knowing the table raises further questions this article sets up but does not fully resolve. Why is the code redundant in the particular pattern it shows (synonymous codons for one amino acid often differing only in the third position)? That regularity is explored under Properties of the Genetic Code, including its apparent robustness to point mutations and its near-universality across life. A second, related puzzle is arithmetic: cells have far fewer than 61 distinct tRNA species for 61 sense codons, so how does a limited set of tRNAs read all the codons accurately? That is resolved by Francis Crick's Wobble Hypothesis, which describes relaxed pairing rules at the third codon position.