Question 11 of 20 Fundamentals Practical
How does Gibson assembly join DNA fragments, and when would you choose it?
Short answer
Three enzymes work together in one tube at 50 C. An exonuclease chews back 5' ends to expose single-stranded overhangs, matching fragments anneal through 20 to 40 base pairs of shared sequence, a polymerase fills gaps and a ligase seals the nicks. Choose it when you want a seamless join, several fragments at once, or no usable restriction sites.
Why
Gibson and colleagues described the method in 2009 as an isothermal single-reaction way to assemble overlapping DNA, from genes up to whole genomes. The overlaps are usually created by PCR primers with a 5' tail matching the neighboring fragment. A common design is a 60 base primer with 30 bases annealing to the target and 30 matching the adjacent piece.
Because the join needs only shared sequence, no restriction site is left behind and no scar appears between parts. That matters for fusion proteins where an extra few bases would shift the reading frame. Addgene's protocol warns against strong secondary structure in the homology region, because hairpins reduce annealing.
The method has limits. Repeated sequences can assemble in the wrong order. Very short fragments and high GC overlaps work poorly.
Background from uncut or re-circularized vector is common, and one trick is to split an antibiotic resistance gene across two parts, so only a correct assembly can grow. Whatever the method, the seams between parts are exactly what must be sequenced afterwards.
A strong answer covers
- Exonuclease, polymerase and ligase in one isothermal reaction
- 20 to 40 base pair overlaps made by PCR primers
- Seamless joins with no restriction scar
- Several fragments in a single step
- Sequence the seams afterwards
Common mistakes
- Designing overlaps with strong hairpins or repeated sequence
- Assuming colonies on the plate mean the assembly is correct
- Skipping sequencing of the junctions because the size looked right
The interviewer might ask next
How would you reduce background colonies from uncut vector in an assembly reaction?