Written and maintained by CASRAI Editorial Board
Last updated
Gibson assembly joins DNA fragments seamlessly, in a defined order, in a single tube at a single temperature. There are no restriction sites, no scars at the junctions, and no separate ligation step. The method was described by Daniel Gibson at the J. Craig Venter Institute, and its appeal is that three enzymes with apparently incompatible jobs are made to cooperate in one isothermal reaction.
Understanding what each of those enzymes is doing is what separates a reliable assembly from one that produces empty vector and unexplained colonies.
The three enzymes and what each one does
The reaction contains three activities working on the same DNA at the same time.
- T5 exonuclease chews back from the 5′ ends of each fragment. Because it removes the 5′ strand, what it leaves behind is a single-stranded 3′ overhang. This is the step that creates the sticky ends — but unlike restriction cloning, you designed those ends yourself, in the primers.
- Phusion DNA polymerase extends from the annealed 3′ ends and fills the single-stranded gaps left behind once complementary overhangs have found each other.
- Taq DNA ligase seals the remaining nicks, covalently joining the backbone so the product is a continuous molecule rather than an annealed complex.
The order matters conceptually even though everything is in one tube: exonuclease acts on ends, annealing happens between complementary overhangs, and polymerase and ligase act on the annealed product. A fragment with no complementary partner simply gets chewed and never proceeds.
One tube, one temperature
The reaction is isothermal at 50 °C, typically for 15 to 60 minutes, with 60 minutes optimal for harder assemblies. That single incubation replaces the digest, gel purification, dephosphorylation and overnight ligation of restriction cloning.
The practical consequence of isothermal one-pot chemistry is that you cannot troubleshoot by staging the reaction. There is no intermediate you can run on a gel to see how far it got. Diagnosis is therefore almost entirely about what you put in — overlap design and fragment quality — rather than about the incubation itself.
Overlap design is the whole method
Adjacent fragments must share identical terminal sequence. You add that homology in the PCR primers, so the overlap is a design decision made before any assembly reagent is opened.
Published guidance from the reagent manufacturer scales the overlap with the complexity of the assembly:
- 2–3 fragments: 15–25 nt overlaps, with the 15-minute reaction.
- Up to 6 fragments: 20–80 nt overlaps, with the 1-hour reaction.
- Overall recommended range 15–40 bp, with an overlap melting temperature at or above 48 °C.
That Tm threshold is the part most often ignored, and it is not arbitrary: the overlaps have to stay annealed at the 50 °C reaction temperature. A 20 bp overlap that happens to be AT-rich can fall below it, and the assembly fails for a reason that looks nothing like a design error on paper.
How many fragments, and how much of each
Up to six fragments can be assembled in one step, and efficiency falls as either the number of fragments or their length increases. Recommended input is 0.02–0.5 pmol total when assembling one or two fragments into a vector, rising to 0.2–1.0 pmol for four to six fragments.
Note that those are molar amounts, not masses. This is the most common quantitation mistake in the method: equal masses of a 6 kb vector and a 500 bp insert are wildly different molar amounts, and a ratio set by nanograms rather than picomoles will silently starve the reaction of one component.
Why assemblies fail
- Overlap Tm below 48 °C — the junction cannot hold at 50 °C. Recalculate rather than lengthening blindly.
- Ratios set by mass, not moles — see above; convert first.
- Residual template plasmid — the commonest cause of a background lawn of colonies that all contain the original construct. If the vector was PCR-amplified from a plasmid, the template survives unless you remove it; a DpnI digest of the methylated template before assembly is the standard countermeasure.
- Impure or partially degraded fragments — the exonuclease acts on every free end it finds, including ends you did not intend to expose.
- Repeats or inverted repeats at junctions — homology-based assembly cannot distinguish two identical overlaps, so a repeated sequence can join in the wrong orientation or drop a fragment entirely.
- Too many fragments at once — efficiency is known to fall with fragment count. A six-way assembly that refuses to work often succeeds as two sequential three-way assemblies.
Where it fits against the alternatives
Gibson assembly is the right tool when you want a seamless junction, a defined fragment order, or more fragments than restriction cloning can practically handle. It is the wrong tool when a single well-placed restriction site would do the job in an afternoon, or when the junction sequence is constrained in a way that makes designing homology arms awkward.
Classical restriction-enzyme digestion remains simpler for routine subcloning, and every Gibson workflow still depends on clean PCR to generate fragments with the designed overlaps. Verify products on an agarose gel before assembly, and confirm junctions afterwards by Sanger sequencing — a colony that grows is evidence of a circular plasmid, not evidence of the construct you designed.
Frequently asked questions
What temperature and how long?
50 °C, isothermally, for 15 to 60 minutes. Fifteen minutes suits two- or three-fragment assemblies; an hour is optimal and is what more complex assemblies need.
How long should the overlaps be?
15–25 nt for two or three fragments, 20–80 nt for up to six, within an overall recommended range of 15–40 bp — and in every case with a melting temperature of at least 48 °C.
How many fragments can I assemble at once?
Up to six. Efficiency decreases as the number of fragments or their length rises, so splitting a large assembly into sequential stages is often faster in practice than forcing one reaction.
Why do all my colonies contain the empty original plasmid?
Almost always surviving template from the PCR that produced the vector fragment. Digest the methylated parental plasmid with DpnI before assembling.
Why does the polymerase not just chew everything up?
The exonuclease and the polymerase act on different substrates: T5 exonuclease works on free 5′ ends, while the polymerase extends annealed 3′ ends and fills gaps. Fragments that have annealed through their designed overlaps present the polymerase substrate; unannealed ends do not.
Do I need to phosphorylate my fragments?
No. Unlike a conventional ligation, the nick sealing is done by Taq DNA ligase on a filled-in, annealed junction, so there is no separate 5′-phosphate preparation step for the user to get wrong.
References
- Gibson Assembly — method overview, New England Biolabs
- Gibson Assembly Master Mix instruction manual (E2611) — NEB manual E2611
- T5 exonuclease-dependent assembly offers a low-cost method for efficient cloning and site-directed mutagenesis — PMC6379645
- Seamless insert-plasmid assembly at high efficiency and low cost — PMC4830597
- Plasmids 101: Gibson assembly and other long-homology based cloning methods — Addgene
