Comparing cloning methods for a short insert

A short insert is easy to lose. An exonuclease can chew it away, and a gel can hide it as a faint band or let it run off the end. You are choosing among restriction ligation, Gibson assembly, Golden Gate, and annealed oligonucleotides, using three facts: how long the fragment is, which scar you can tolerate, and how you will prove orientation. This is an application note for that choice. It is not a volume table.

The chemistry of restriction ligation and Gibson assembly is in restriction ligation and Gibson assembly. The colony you keep still follows plasmid cloning from insert to colony.

What each class does with a small piece of DNA

Restriction ligation cuts at known sites and seals compatible ends with ligase. It suits a short insert when two different sites give you direction and you accept that the fragment may be invisible on the gel. Judge the ligation by colonies you then sequence, not by a band you might have imagined. If the ends are compatible or blunt, phosphatase belongs on the vector only, so the backbone cannot close as easily, and the insert must still carry phosphates. A leftover restriction site is the usual scar. Keep it when you will cut there again. Remove it from the plan when the protein or the later map cannot stand it.

Gibson assembly uses an overlap, commonly designed around 15 to 40 base pairs, and a mix of exonuclease, polymerase, and ligase activities. The 2009 Nature Methods paper is the method citation. The exonuclease is the short-insert problem. It exposes single strands so overlaps can anneal. A fragment that is only as long as that chew, or barely longer, can be eaten before a stable duplex remains. Stop and do not force Gibson on a linker-sized piece. Orientation comes from unique overlaps. The same short tail on both ends lets the fragment flip and still anneal. The scar is the overlap sequence you wrote into the primers. Design it on purpose.

Golden Gate, in the type IIS sense published in 2008, uses enzymes that cut outside their recognition site. The overhang can be any short sequence you choose, and the recognition site can be left out of the final product if it sat in the flanking DNA that is cut away. That is why a short insert made as an oligonucleotide with type IIS tails is a natural fit: you never needed a visible restriction fragment, and the scar can be the four bases you wanted or no added scar at all. Orientation is the overhang code. Two fragments that share an overhang will ligate in the wrong neighbourhood. Follow the enzyme card. Do not copy a vendor's microlitre table into the plan.

Annealed oligonucleotides are the class for a very short linker, tag, or site. Two single strands are annealed to a duplex with the overhangs the vector expects. If the vector was dephosphorylated, the oligos have to bring the 5-prime phosphates. The gel will often show nothing you can cut out. Do not wait for a band. Ligate, transform, and confirm by sequence. Asymmetric overhangs set direction. Identical blunt ends or a palindromic overhang do not. The "scar" is the sequence you ordered. There is no hidden site unless you included one.

Choose, then stop if the check is missing

Write the three facts before you order oligos or enzymes. Length: a handful of bases, a short peptide, or a fragment you can already see as a real band. Scar: site retained, overlap seam, designed overhang, or the exact oligo. Orientation: directional ends, a diagnostic digest, or a sequencing primer that reads across the junction and would look wrong if the fragment were backwards.

Eliminate. If the insert is short enough that an exonuclease would remove it, cross out Gibson. If you cannot tolerate a remaining restriction site and you have no second site strategy, cross out a single-site ligation or redesign it. If the only confirmation you imagined was "I will see it on the gel," cross out every class until a sequence or a directional test is on the page. A faint band is not an orientation assay.

If two classes remain, pick the one whose confirmation is shorter. A linker is often annealed oligos plus one read. A scarless short peptide often fits Golden Gate better than a site you must later remove. A short piece you can already amplify from a trusted plasmid, with sites you will keep, fits restriction ligation. The Sanger sequencing enquiry reference is how you specify the read that makes any of these a clone rather than a candidate.

Stop when the plate exists and you still cannot tell direction. Do not screen thirty colonies with a PCR that amplifies either way. Add the directional primer or the digest, or sequence. Stop when a Gibson reaction was built on an insert shorter than the overlap logic. Redesign in another class. Stop when the faint band you excised might have been primer-dimer. A dimer ligates. Sequence will show it, and another extraction of the same gel will show it again.

Failures that look like a short clone

A colony with no insert is common when the short piece was never in excess and the vector could close. The gel of the purified insert was a stain blob. Phosphatase on the vector, or Golden Gate overhangs that cannot self-match, is the redesign. Repeating the same purification will repeat the empty colony.

A Gibson colony with a deletion has often lost the short middle fragment. The overlaps found each other and skipped the insert. That is the chew and the length, not bad luck. Move the join to oligos or type IIS.

A Golden Gate reaction with repeated overhangs ligates in more than one order. A short insert then sits backwards or between the wrong neighbours. Recode the overhangs. The 2008 paper is the reason the overhang is yours to design. The enzyme does not rescue a duplicated code.

An annealed duplex left on a warm block can fall apart before it meets the vector. Use a short double strand while it is still a duplex, and sequence the clone. A control plasmid tells you whether the cells were fine.

Safety

Oligos, restriction enzymes, and assembly mixes are ordinary molecular-biology reagents. Follow the sheet and the waste rules for recombinant plates. A short insert can still encode a peptide your institution restricts. The method class does not change that decision. This page is not an approval to clone it.

A short band and a long run

People lose short inserts by borrowing the gel recipe from a genomic digest: a low percentage and a long run, and the piece leaves the gel before anyone photographs it. For a fragment you still hope to see, use a percentage and a time that keep small DNA on the gel, and load a marker that has a small band. If the marker's small band is gone, your insert is not "faint." It is absent from the slice. Then switch to a method that does not need the slice, and confirm by sequence.

What to ask for

Name the class, the insert length, the scar you can tolerate, and the orientation test. Enzymes and competent cells are lines in the molecular biology catalogue. If the short sequence should be made as a specified fragment rather than annealed in the lab, the custom gene synthesis enquiry reference is a specification prompt only. Put the length and the confirmation plan on the quote request. Ask whether a quotation is possible. A request that says only "clone this linker" has not chosen a method.