Science Wiki

A codon is a contiguous sequence of three nucleotides (a trinucleotide triplet) in a messenger RNA (mRNA) molecule that specifies either a single proteinogenic amino acid or the termination of translation. The corresponding 5′→3′ triplets on the non-template (sense/coding) strand of DNA are also called codons by convention, written with deoxythymidine (T) in place of uridine (U). Functioning as the fundamental informational words of the genetic code, codons enable the ribosome and transfer RNA (tRNA) anticodons to translate linear nucleotide sequences into the precise primary amino acid sequences of polypeptide chains that fold into functional proteins across the biosphere.

Combinatorial Information Capacity (4³ = 64)

The genetic code maps an alphabet of four canonical nucleotide bases (A, U, C, and G in RNA) onto the 20 standard proteinogenic amino acids plus a translation termination signal. The total number of distinct codons (*N*) that can be constructed from a sequence of *k* nucleotides is given by the combinatoric relation:

*N* = 4^*k*

A singlet code (*k* = 1) would specify only 4¹ = 4 meanings, and a doublet code (*k* = 2) would specify only 4² = 16 meanings—fewer than the minimum 21 discrete assignments required (20 amino acids and at least one stop signal). A triplet code (*k* = 3) yields 4³ = 64 distinct codons, the smallest integer value of *k* with sufficient informational capacity. During translation elongation, the ribosome translocates along the mRNA in the 5′ → 3′ direction in discrete 3-nucleotide steps, reading one non-overlapping, comma-free triplet at a time.

Experimental Discovery by Crick, Brenner, Nirenberg, and Khorana

The triplet, non-overlapping nature of the genetic code was established in 1961 through bacteriophage genetics by Francis Crick, Sydney Brenner, Leslie Barnett, and R.J. Watts-Tobin, who analyzed acridine-induced frameshift insertions (+) and deletions (−) in the *rIIB* cistron of bacteriophage T4. Whereas a single (+1 or −1) or double (+2 or −2) nucleotide indel abolished gene function, combining three insertions (+3) or three deletions (−3) within the same gene restored the correct downstream reading frame and yielded functional phage plaques. In the same year (1961), Marshall Nirenberg and Heinrich Matthaei demonstrated in a cell-free *Escherichia coli* translation extract that synthetic polyuridylic acid [poly(U)] directed the synthesis of polyphenylalanine, assigning the first codon (UUU = Phe). Subsequent biochemical work by Nirenberg and Philip Leder (using trinucleotide-directed ribosome filter-binding assays in 1964) and Har Gobind Khorana (using chemically synthesized repeating di-, tri-, and tetranucleotide copolymers) completed the assignment of all 64 codons by 1966.

Start Codon Selection (AUG, Shine-Dalgarno, and Kozak Context)

Translation begins and ends at specific boundary codons that establish and terminate the coding sequence:

• Start codon (AUG): The triplet 5′-AUG-3′ is the canonical translation initiation codon across all domains of life and also encodes internal methionine residues. Initiation utilizes a specialized initiator tRNA (tRNA_i^Met in eukaryotes and archaea; tRNA^fMet carrying *N*-formylmethionine in bacteria and eukaryotic organelles) that binds directly to the ribosomal P-site, whereas internal AUG codons are decoded in the A-site by elongator tRNA^Met. In bacteria, the 30S ribosomal subunit locates the start codon via base pairing between the 3′ end of 16S rRNA and an upstream purine-rich Shine-Dalgarno sequence (consensus 5′-AGGAGG-3′, located 5–9 nucleotides 5′ of AUG). In eukaryotes, the 43S preinitiation complex binds the 5′ m⁷G cap and scans 5′→3′ along the 5′-UTR until it encounters the first AUG situated within a favorable Kozak consensus sequence (5′-gccRcc<u>AUG</u>G-3′, where R at −3 is a purine and +4 is G). Alternative near-cognate initiation codons—notably GUG (~14% of *E. coli* genes) and UUG (~3%) in bacteria, and CUG in eukaryotes—are still decoded as fMet or Met by the initiator tRNA.
• Stop codons (UAA, UAG, UGA): These three triplets (termination or nonsense codons) do not specify any canonical amino acid, and no standard tRNA carries a cognate anticodon. Historically derived from suppressor genetics, UAG is termed *amber*, UAA is *ochre*, and UGA is *opal* (or *umber*). When a stop codon enters the ribosomal A-site, it is recognized directly by protein class I release factors via conserved peptide motifs (rather than RNA anticodons). In bacteria, RF1 (via a PXT motif) recognizes UAA and UAG, whereas RF2 (via an SPF motif) recognizes UAA and UGA. In eukaryotes and archaea, a single omnipotent factor (eRF1 / aRF1, via a NIKS motif) recognizes all three stop codons. Upon binding, the universally conserved GGQ motif of the release factor inserts into the peptidyl transferase center and coordinates a nucleophilic water molecule to hydrolyze the ester bond linking the nascent polypeptide to the P-site tRNA, releasing the finished protein.

The first initiator AUG establishes the reading frame and defines the start of the open reading frame (ORF), which extends continuously until the first in-frame termination codon. Because 3 of the 64 triplets are stop codons, an unbiased random sequence contains an in-frame stop codon on average once every 64 / 3 ≈ 21.3 codons; consequently, long continuous ORFs (>100 codons) serve as primary statistical signatures in computational gene prediction.

Co-Translational Recoding of Stop Codons (Selenocysteine and Pyrrolysine)

Stop codons can be contextually reassigned during translation elongation to incorporate the 21st and 22nd genetically encoded proteinogenic amino acids. In the presence of a stem-loop SECIS element (Selenocysteine Insertion Sequence) in the mRNA and a dedicated elongation factor (SelB in bacteria; eEFSec in eukaryotes), UGA is decoded by tRNA^Sec as selenocysteine (Sec, U) in essential redox enzymes such as glutathione peroxidases and thioredoxin reductases. Analogously, in methanogenic archaea (such as *Methanosarcina*) and certain anaerobic bacteria, in-frame UAG codons within methylamine methyltransferase transcripts are decoded by pyrrolysyl-tRNA^Pyl as pyrrolysine (Pyl, O).

Synonymous Codon Family Boxes

Of the 64 triplets in the standard genetic code, 61 are sense codons that specify the 20 standard amino acids, making the genetic code degenerate (redundant, though strictly unambiguous): 18 of the 20 amino acids are encoded by two to six synonymous codons. The degenerate distribution across the 64-codon table is structured as follows:

Synonymous codons are predominantly organized into "family boxes" that share identical nucleotides at the first (5′) and second (middle) positions and vary only at the third (3′) position (e.g., all four GCN codons specify alanine). The three hexacodonic amino acids—leucine, arginine, and serine—occupy one four-codon box plus a two-codon set that differs at the first or second position.

Degeneracy acts as a physiochemical buffer against point mutations and translational misreading. A point mutation at the third codon position (such as CUC → CUG) frequently leaves the encoded amino acid unchanged (leucine) and is termed a silent mutation (synonymous substitution). Furthermore, the chemical topology of the code clusters amino acids with similar hydropathy and polarity by their second-position base: U at the second position (NYN) universally specifies hydrophobic residues (Phe, Leu, Ile, Met, Val), whereas A at the second position (NAN) specifies hydrophilic charged or polar residues (Tyr, His, Gln, Asn, Lys, Asp, Glu). Monte Carlo simulations by Stephen Freeland and Laurence Hurst demonstrate that only ~1 in 10⁶ randomly permuted genetic codes buffer mutational hydropathy changes more effectively than the natural standard code.

The Wobble Hypothesis and Anticodon Position 34

Degeneracy is concentrated at the 3′ third codon position because of the wobble base pair geometry proposed by Francis Crick in 1966. While the ribosomal decoding center strictly enforces canonical Watson-Crick base pairs at the first and second codon positions via minor-groove A-minor interactions (16S rRNA A1492, A1493, and G530), the third codon position (paired with the 5′ nucleotide at position 34 of the tRNA anticodon) permits greater rotational freedom and non-Watson-Crick hydrogen bonding. Post-transcriptional enzymatic modifications of anticodon position 34—such as the deamination of adenosine to inosine (I34)—expand decoding versatility:

Wobble pairing allows a single tRNA isoacceptor species to decode multiple synonymous codons, enabling cells to translate all 61 sense codons using only ~30 to 45 distinct tRNAs (or as few as 22 tRNAs in mammalian mitochondria via unmodified U34 "superwobble").

Codon Usage Bias, RSCU, and Codon Optimization

Synonymous codons are not utilized at equal frequencies within or across genomes. Every species exhibits a characteristic codon usage bias shaped by directional genomic GC-content mutational biases (e.g., >70% GC in *Streptomyces coelicolor* vs. ~19% GC in *Plasmodium falciparum*) and translational selection matching codon frequencies to the intracellular pool of cognate tRNAs. A standard quantitative metric is the Relative Synonymous Codon Usage (RSCU):

*RSCU*_*ij* = *X*_*ij* / (*S*_*i* / *n*_*i*)

where *X*_*ij* is the observed count of codon *j* encoding amino acid *i*, *S*_*i* is the total count of all synonymous codons for amino acid *i*, and *n*_*i* is the degeneracy (number of synonymous codons, 1 ≤ *n*_*i* ≤ 6) for that amino acid. An RSCU value of 1.0 indicates unbiased usage, >1.0 indicates preferred codons, and <1.0 indicates rare codons. Related gene-level metrics include the Codon Adaptation Index (CAI) (ranging from 0 to 1.0, introduced by Paul Sharp and Wen-Hsiung Li in 1987).

Consequently, synonymous mutations are not always physiologically neutral: rare codons cause transient ribosomal pausing that regulates co-translational folding of nascent protein domains, alters mRNA secondary structure, and modulates exonic splicing enhancers during pre-mRNA splicing and mRNA decay via Dhh1/CNOT3 sensing of slow ribosomes. In industrial biotechnology and mRNA vaccine design, codon optimization replaces rare codons in heterologous transgenes with host-preferred high-RSCU codons to increase protein yield by 10- to 100-fold.

Six Reading Frames and Non-Overlapping Triplets

Because the ribosome reads codons in continuous, non-overlapping triplets without internal punctuation, the identity of the first nucleotide at the 5′ initiation site dictates the grouping of every downstream triplet. This phase register is the reading frame. Any double-stranded DNA locus possesses six potential reading frames—three phase registers (+1, +2, and +3) on the 5′→3′ sense strand and three on the complementary 5′→3′ antisense strand—with the functional reading frame specified by the promoter and start codon. While most cellular genes are non-overlapping, compact viral genomes such as bacteriophage phiX174, hepatitis B virus, and SARS-CoV-2 encode overlapping genes read in alternative frames.

As a mechanistic illustration, the mRNA sequence 5′-AUG-CUA-GGC-UAA-3′ is decoded in the +1 frame as: Met – Leu – Gly – [Stop].

Frameshift Mutagenesis vs. Programmed Ribosomal Frameshifting

An insertion or deletion (indel) of a number of nucleotides not divisible by three (*n* ≠ 3*k*) within a protein-coding region shifts the translational register of all downstream triplets and is termed a frameshift mutation. Such mutations arise from replication slippage across homopolymeric tracts during DNA replication or error-prone non-homologous end joining during DNA repair. If the first C in the sequence above is deleted (−1 frameshift), the mRNA shifts to 5′-AUG-UAG-GCU-AA-3′, which translates as Met – [Stop], immediately terminating synthesis. Frameshift mutations completely scramble the downstream amino acid sequence and almost invariably encounter an out-of-frame premature termination codon (PTC), triggering degradation of the aberrant transcript via nonsense-mediated decay (NMD) in eukaryotes. In contrast, in-frame indels of 3*k* nucleotides delete or insert whole amino acids while preserving the downstream reading frame (as in the 3-bp ΔF508 deletion in the *CFTR* gene causing cystic fibrosis).

Conversely, many viruses and some cellular genes exploit programmed ribosomal frameshifting (PRF) as a stoichiometric regulatory mechanism. In retroviruses such as HIV-1 and coronaviruses, a "slippery heptanucleotide sequence" (consensus X-XXY-YYZ) followed by a stable downstream RNA pseudoknot stalls the translating ribosome and induces a controlled −1 nucleotide backward slip at a fixed frequency (~5–30%), enabling the synthesis of the fused Gag-Pol polyprotein in a defined ratio relative to structural Gag proteins.

Conservation Since LUCA and Alternative Translation Tables

The standard 64-codon genetic code (NCBI Translation Table 1) is nearly universal across Bacteria, Archaea, Eukarya, and viruses: the triplet UUU encodes phenylalanine whether translated by a human hepatocyte, an *Arabidopsis* chloroplast, or a deep-sea archaeon. This conservation constitutes primary molecular proof that all extant cellular life evolved from a single Last Universal Common Ancestor (LUCA), preserved by Francis Crick's "frozen accident" constraint (any global reassignment of a sense codon in a complex proteome would introduce lethal missense substitutions across thousands of proteins simultaneously).

Nevertheless, over 30 natural deviations from the standard code have evolved in reduced organelle genomes and specific microbial lineages (commonly via the "codon capture" or "ambiguous intermediate" mechanisms):

• Vertebrate Mitochondria (NCBI Table 2): UGA encodes tryptophan (rather than stop), AUA encodes methionine (rather than isoleucine), and AGA and AGG function as termination signals. In budding yeast (*Saccharomyces*) mitochondria (Table 3), the entire CUN four-codon box is reassigned from leucine to threonine.
• *Mycoplasma*, *Spiroplasma*, and Candidatus *Hodgkinia* Bacteria (Table 4): UGA is reassigned from a stop codon to tryptophan due to loss of release factor RF2 under extreme AT-mutational bias.
• Ciliated Protozoa (Table 6 & Table 10): In *Tetrahymena* and *Paramecium*, UAA and UAG encode glutamine (leaving UGA as the sole stop codon), whereas in *Euplotes* UGA encodes cysteine and in *Condylostoma magnum* all three stop codons specify amino acids except when positioned adjacent to the 3′ poly(A) tail.
• CTG-Clade Fungi (*Candida* *albicans*, Table 12): The codon CUG is decoded as serine (~97%) rather than leucine by a chimeric tRNA_CAG^Ser.

Synthetic Genome Recoding and Expanded Genetic Alphabets

In synthetic biology, researchers systematically engineer the genetic code using orthogonal translation systems—consisting of evolved aminoacyl-tRNA synthetase / suppressor tRNA_CUA pairs (such as *Methanocaldococcus jannaschii* TyrRS or archaeal PylRS) that do not cross-react with endogenous host tRNAs or synthetases—to reassign the amber stop codon UAG to site-specifically incorporate over 200 non-canonical amino acids (ncAAs) carrying bioorthogonal click handles, fluorophores, or post-translational modifications. At the whole-genome scale, Jason Chin and colleagues synthesized the 4-megabase *Escherichia coli* Syn61 genome (2019), replacing all 18,214 occurrences of two serine codons (UCG and UCA) and the UAG stop codon with synonymous triplets, and subsequently deleted their cognate tRNAs and RF1 to create a 61-codon organism completely resistant to viral bacteriophage infection and capable of programmable non-canonical polymer synthesis.

Why are there 64 codons in the genetic code if there are only 20 standard amino acids?

Because RNA uses an alphabet of 4 nucleotide bases (A, U, C, G), a 1-base code yields only 4¹ = 4 combinations and a 2-base code yields only 4² = 16 combinations—both mathematically insufficient to encode 20 amino acids plus a stop signal. A 3-base (triplet) code is the minimum integer length capable of encoding all 20 amino acids, yielding 4³ = 64 combinations, which life exploits through codon degeneracy to buffer against point mutations.

How can a synonymous (silent) mutation cause human genetic disease if it does not change the amino acid sequence?

Although a synonymous mutation encodes the same amino acid, it can disrupt protein production through three primary mechanisms: (1) destroying or creating an exonic splicing enhancer/silencer motif, causing exon skipping during pre-mRNA splicing; (2) altering mRNA secondary structure and stability; or (3) replacing a rare codon with a fast-translating abundant codon (or vice versa), which alters ribosomal elongation pauses required for proper co-translational protein folding (as documented in synonymous polymorphisms of the *MDR1/ABCB1* multidrug transporter).

Why is AUG almost universally used as the translation start codon instead of each gene having a dedicated non-amino-acid start triplet?

In both bacteria and eukaryotes, the initiator codon must base-pair with a charged tRNA in the ribosomal P-site to provide the initial α-carboxyl ester bond onto which the first peptide bond is synthesized. By coupling AUG recognition to upstream structural signals (the Shine-Dalgarno sequence in bacteria or the 5′ cap and Kozak consensus in eukaryotes) and a dedicated initiator tRNA (tRNA^fMet or tRNA_i^Met), the ribosome cleanly distinguishes an initiating AUG from internal methionine AUG codons without sacrificing an extra sense codon.