Overlap assembly design rules

Overlap assembly fails in the design more often than in the pipetting. Gibson-style joins use a shared overlap, commonly planned around 15 to 40 base pairs, and a mix that chews back ends, fills gaps and ligates them in one isothermal reaction. When the colonies are wrong or absent, the useful question is which overlap rule broke. The method is the one Gibson and colleagues described in Nature Methods in 2009 (enzymatic DNA assembly). The broader choice between this join and a restriction ligation is in restriction ligation and Gibson assembly. Archiving the colony that survives is in plasmid cloning from insert to colony.

Follow the assembly mix you actually have for temperature and time. A classical planning point is near 50 Celsius. The tube in front of you may differ. Enzyme classes are listed from the molecular biology catalogue as a sourcing start, not as a copied insert.

What the mix is doing while the design either holds or folds

A 5-prime exonuclease resects each end, exposing the overlap as a single strand. Complementary overlaps anneal. A polymerase fills the gaps. A ligase seals the nicks. Every fragment in the tube sees the same temperature and the same exonuclease. Design rules exist because that single condition cannot pamper a weak overlap and a tiny fragment at the same time.

The overlap has to be unique. If the same 20 bases occur at two junctions, or if they occur inside a fragment, the anneal can land in the wrong register. The product is often a deletion between the repeated stretches, and it is circular enough to transform. Colony screening that only asks "is there an insert-sized PCR product?" can bless that deletion when the primers sit outside it.

Melting temperature has to be similar across junctions. Calculate it as a starting comparison, in the same salt assumption for every overlap, and treat the number as a rank rather than a promise. The mix's temperature is fixed. Your overlaps have to live there together.

A reasoning path from symptom to redesign

Start with the plate only long enough to see whether any circular DNA formed. Then go back to the sequences. Search each overlap against every fragment with NCBI BLAST or a local aligner, and ask whether it appears once. Write down the melting temperature of each overlap side by side. Measure the insert length that remains after you subtract both overlaps. If that remainder is a small fraction of the fragment, put the exonuclease problem at the top of the list.

Only after those three checks should you blame the mix, the competent cells or the plates. A known plasmid transformation still belongs beside the assembly, because an empty plate has more than one cause. A vector-only assembly, without the insert fragments, tells you whether the backbone can circularise on a repeat of its own ends.

The Addgene Gibson assembly protocol is a public note of the reaction's shape. Use it to see which controls people bother to run. The incubation on your mix card wins if they differ. protocols.io records variants. None of those pages replace the overlap sequences in your file.

Repeats in the overlap

Direct repeats and shared primer tails are the common version of this fault. You designed junction A with the same 25 bases you used at junction C because both fragments meet a repeated promoter or a repeated linker. The exonuclease does not know which copy you meant. Annealing picks a register that deletes the DNA between the copies.

Fix the sequence, not the colony count. Slide the overlap into neighbouring unique bases, or split a repeated element so each junction sees a different tail. If the gene itself is repetitive, overlap assembly may be the wrong join, and a restriction ligation at sites you have already shown to be unique is the calmer plan. Say that early. Screening fifty colonies will not invent uniqueness that the sequence lacks.

Melting temperature mismatch

Rank the overlaps. If one is a 15-base AT stretch and another is a 40-base GC stretch, they do not belong in one isothermal reaction without a redesign. Lengthen the weak side toward the upper part of the 15 to 40 base habit, or shorten and mild the strong side, until the calculated melting temperatures sit together. Hairpins inside an overlap behave like a melting problem: the overlap prefers itself to its partner. Move off the hairpin even if the length looked fine.

Primers that encode the overlap can also disagree with the template by a one-base error from ordering. That mismatch lowers the effective melting temperature and can scar the junction. Read the primer file against the template before you blame the exonuclease.

Too little overlap, and a fragment chewed to nothing

Too little overlap on a short insert is a double constraint. You need enough bases to anneal, and you need enough fragment left after chew-back that the piece still exists. If the insert is barely longer than its two overlaps, those two needs collide. The exonuclease, which is happy to resect a long backbone by a few dozen bases, can erase the small fragment.

Raise the molar amount of the short piece relative to the vector so a few molecules anneal before they vanish. Keep the incubation inside the window the mix states, rather than leaving a small assembly for the time you use on a hundred-kilobase plan. If the insert is a synthetic cassette, order it already inside a small carrier and cut it out, or switch that one junction to a restriction ligation. The 2009 paper is the conceptual source. It is not a reason to incubate a 100-base pair piece for as long as a large assembly just because both are called Gibson.

What not to keep doing

Do not add more cycles of hope in the form of a second assembly on top of an unsequenced first one. Do not lengthen every overlap to 80 bases because one junction failed; you will create new repeats and more chew. Do not trust a single pretty colony PCR band across a repetitive junction. Sequence the overlaps of two independent colonies. If they disagree, the design is still unstable.

Safety

The mix is a research reagent. An assembly that deletes a toxin gene's regulatory region, or that joins the wrong fragments, is still DNA you may transform. Your institution decides containment for the host and the insert. This troubleshooting note is not a clinical method and does not validate a construct for any diagnostic use.

Put the overlap bases in the file, not only in the email

When a design moves between people, the sentence "use 25 base overlaps" is not a design. In a busy week, or when a file is rebuilt from memory after a delayed reply, the repeats and the melting temperatures are the first things to be dropped. Write each overlap sequence into the specification, with its calculated melting temperature beside it, before you ask anyone to synthesise the fragments. Heat and humidity do not rewrite the bases. They do encourage a verbal shortcut. The file is the shortcut you can check.

What to ask for

Send the fragment map, the overlap sequences, the melting temperatures and which junction failed. If you want the fragments made rather than amplified, frame that against the custom gene synthesis enquiry reference. If a colony needs a junction read, frame that against the Sanger sequencing enquiry reference. Both are enquiry references. They do not mean EVRINTH synthesises DNA or runs sequencing. Use the quote request for the mixes or the fragments you actually need, and ask whether a quotation is possible.