Written and maintained by CASRAI Editorial Board
Last updated
Short answer: The Kozak sequence is the nucleotide context around a eukaryotic start codon that helps ribosomes pick that AUG to begin translation. It is usually written with the AUG included, so yes, the start codon sits inside the conventional consensus. The positions that matter most are a purine (usually A) at -3 and a G at +4, counting the A of AUG as +1.
What the Kozak sequence is
Marilyn Kozak identified the pattern by comparing the sequences surrounding initiation codons in eukaryotic messenger RNAs. In her 1981 survey, “out of 153 messages examined, 151 have either a purine in position -3, or a G in position +4, or both,” and she proposed a favored context of [A/G]XXAUGG (Kozak, Nucleic Acids Research 9(20):5233-5252, 1981). Nonfunctional AUG codons in 5′ noncoding regions usually carried a pyrimidine at -3, which is part of why the context is read as a signal that separates a real start codon from an accidental one.
Promega’s technical note on the topic gives the lineage of the written consensus: it was originally defined as ACCAUGG from a mutation study of preproinsulin, and a later survey of 699 vertebrate mRNAs extended it to GCCGCCACCAUGG. In the same note, the A of the AUG is coordinate 1, a purine at -3 is described as crucial for efficient initiation, and when that purine is absent a G at +4 is described as essential.
Does the Kozak sequence contain the start codon?
In the standard written form, yes. GCCGCCACCAUGG places the AUG at +1 to +3, the -3 purine three bases upstream of the A, and the G at +4 immediately after the codon. Many people also use “Kozak sequence” for only the upstream flank (for example GCCACC) that is placed in front of an ATG when cloning, so check which meaning a protocol or paper intends. The AUG is the anchor; the Kozak context is what surrounds it.
Key positions: which are strongly supported
Numbering starts at the A of AUG as +1; the base before it is -1 (there is no position zero).
| Position | Consensus base | What the sources say |
|---|---|---|
| -3 | Purine, usually A (G also occurs) | Kozak 1981: 151 of 153 messages had a purine at -3, a G at +4, or both. Promega: purine at -3 is crucial for efficient initiation. |
| -6 to -4 | G C C in the extended consensus | Appears in the extended GCCGCCACCAUGG consensus from the 699-mRNA vertebrate survey as summarized by Promega; the notes there single out -3 and +4 as the key positions. |
| +1 to +3 | AUG | The start codon itself. |
| +4 | G | Kozak 1981 and Promega: important, and described as essential when the -3 purine is missing. Xia (2007) disputes the explanation: in 34,169 human protein-coding genes, the prevalence of G at +4 was concluded to be unrelated to translation initiation and instead to reflect constraints on the second amino acid (GCN, alanine, clearly over-represented). |
This is the gap most short explainers leave: the -3 purine has the longest-standing support, while the +4 G has both functional support and a published argument that part of its prevalence comes from amino-acid composition rather than initiation. Both views are in the primary literature, so a construct design should not treat +4 as settled in the same way as -3.
Example: what a Kozak-ready start looks like
As an illustration of the consensus (not a real gene), the RNA context GCCGCCACCAUGG corresponds to the DNA coding-strand sequence GCCGCCACCATGG. Reading from the first A of ATG: -6 to -4 are GCC, -3 is A (a purine), +1 to +3 is ATG, and +4 is G. The base at +4 is also the first base of the second codon, so a G there restricts which amino acid can follow the initiator methionine; this is the link to Xia’s second-codon finding.
What is the best Kozak sequence?
The sources do not name a single universally best sequence. The commonly cited vertebrate consensus is GCCGCCACCAUGG, with the extended survey consensus built from 699 vertebrate mRNAs, and Promega reports that templates carrying CCACCAUGG translated well in rabbit reticulocyte lysate. That result is for that system; it does not establish an optimum for every organism or cell type. Treat “best” as the closest match to the consensus for your host, and confirm with a reporter or expression readout when expression level matters.
What is the consensus sequence?
A consensus sequence is the most frequent base at each position when many sites are aligned. For translation initiation in vertebrates, the written consensus is GCCGCCACCAUGG (RNA) or GCCGCCACCATGG (DNA), and the minimal rule of thumb from Kozak’s original survey is [A/G]XXAUGG.
Kozak sequence vs Shine-Dalgarno sequence
Both help a ribosome find where translation starts, but they work differently and in different cells.
| Feature | Kozak sequence | Shine-Dalgarno sequence |
|---|---|---|
| Organisms | Eukaryotes (vertebrate consensus described above) | Bacteria and archaea |
| Location | Surrounds the AUG, from about -6 through +4 | A purine-rich site upstream of the start codon |
| Mechanism | A sequence context that makes an AUG more or less likely to be recognized as the start | Base-pairs with the 3′ end of the small-subunit (16S) rRNA to position the ribosome |
| Includes the start codon | Conventionally written to include AUG | No; it lies upstream |
Because of this, a bacterial ribosome binding site cannot simply be swapped for a Kozak sequence, and the reverse. Match the initiation signal to the host in which the construct is expressed.
Construct design checklist
- Confirm the intended start AUG is in frame with the coding sequence that follows, for example by finding the reading frame with an ORF finder or by translating the sequence in Expasy Translate.
- Read the sequence around the start in the assembled plasmid, not only in the design file; the bases at -3 and +4 must survive cloning, restriction-site additions and fusion tags.
- If you add an upstream flank such as GCCACC, check that it does not introduce an additional upstream ATG, since nonfunctional upstream AUGs are a known confounder in 5′ regions (Kozak 1981).
- If a G at +4 would change the second amino acid, weigh that against the protein’s needs, remembering Xia’s finding that second-codon identity and +4 are linked.
- For a vendor or repository vector, read the start region in the vector map. Plasmid collections such as Addgene publish full sequences, and tools such as Geneious can annotate them.
- Primer-introduced changes should be verified by sequencing; see the function of primers in PCR for how primer sequence becomes template sequence.
Common misunderstandings
- “Kozak sequence” always means GCCACC. That is a common shorthand for the upstream flank; the consensus as published includes the AUG and the +4 base.
- It is a ribosome binding site like Shine-Dalgarno. It is a context that tunes start-codon recognition, not a base-pairing partner of rRNA.
- Every base is equally important. The -3 and +4 positions are the strongest signals in the primary sources summarized above.
- +4 G is settled. A large human gene analysis argues its frequency reflects second-codon constraints rather than initiation.
FAQ
Does the Kozak sequence contain the start codon?
In the conventional written form, yes: the AUG is at +1 to +3 within GCCGCCACCAUGG. Some protocols use “Kozak” for the flank alone.
Can you give me an example of a coding sequence with a Kozak context?
The DNA coding-strand string GCCGCCACCATGG shows the consensus: A at -3, ATG at +1 to +3, G at +4. It is an illustration of the consensus, not an annotated gene.
What is the consensus sequence?
GCCGCCACCAUGG in vertebrates (RNA form), from a survey of 699 vertebrate mRNAs summarized by Promega; the minimal rule is [A/G]XXAUGG (Kozak 1981).
What is the best Kozak sequence?
No single best sequence is established across organisms. The consensus above is the usual vertebrate reference, and optimality depends on the host and system.
Sources
- Kozak M. Possible role of flanking nucleotides in recognition of the AUG initiator codon by eukaryotic ribosomes. Nucleic Acids Research 9(20):5233-5252, 1981. pmc.ncbi.nlm.nih.gov/articles/PMC327517 (read 2026-11-07).
- Promega. What is a Kozak consensus sequence? promega.com eNote (read 2026-11-07).
- Xia X. The +4G site in Kozak consensus is not related to the efficiency of translation initiation. PLoS ONE 2(2):e188, 2007. pmc.ncbi.nlm.nih.gov/articles/PMC1781341 (read 2026-11-07).
