Golden Gate and type IIS concepts

Golden Gate cloning is a use of type IIS restriction enzymes, not a separate law of ligation. Enzymes such as BsaI or BbsI cut outside their recognition site. The overhangs, a few bases long and usually four in current designs, are chosen so the parts anneal in one order. Because the recognition site can be placed on the side that is cut away, a finished junction loses the site and is not recut. Engler, Kandzia and Marillonnet described that one-pot logic in PLOS ONE in 2008 (type IIS cloning). This page compares the idea with classical restriction ligation and with Gibson assembly. It does not copy a cycle chart. How any of those joins becomes a colony is in plasmid cloning from insert to colony, and the two older joins are set out in restriction ligation and Gibson assembly.

Enzyme classes can be raised from the molecular biology catalogue. The mix card still decides times and temperatures.

The cut sits outside the site you recognise

A type IIS recognition site is asymmetric. The enzyme binds it and hydrolyses the backbone a short, defined distance downstream. BsaI and BbsI are the examples people mean when they say Golden Gate in everyday cloning. Each leaves a short sticky overhang whose sequence is not the recognition site itself. Change the bases in that offset window and you change the overhang without changing the fact that the enzyme will cut.

That is the design freedom. In classical cloning the overhang is whatever the chosen site already encodes, and two fragments join only if those sites are compatible. In a type IIS plan you assign a different overhang to each junction. Fragment A sticks to fragment B, B to C, and the backbone accepts the ends you reserved for it. A repeated overhang is a branch point: parts can loop, swap or close early. Unique overhangs are the rule that makes order.

The site is lost when it lies in the flanking DNA that the cut discards. The product junction is then ordinary duplex of your overhang sequence, with no recognition site left for the enzyme. Parent molecules and wrong joins that still carry the site remain substrates. Digestion and ligation in one pot therefore push toward the product that can no longer be cut. That push is the concept. It is not a set of microlitres.

One pot, described without a protocol

Ligase needs ATP and a phosphate, as in any sticky-end ligation. The type IIS enzyme needs its own conditions. A shared buffer is a compromise the mix manufacturer has already made. Some workflows hold one temperature where both activities persist. Others cycle between a warmer step that favours the restriction enzyme and a cooler step that favours ligase. Copying a cycle count from a paper or from a different vendor's insert is how a one-pot reaction becomes two half-reactions.

What you should decide, before you touch a cycler, is smaller. Every part must be free of the internal recognition site. Every overhang must be unique and a few bases long, matching the offset that enzyme actually produces. The backbone's sites must point outward so the finished circle has lost them. Dead-end products that still contain a site are supposed to be recut. If your "product" was designed so the site remains, the enzyme will keep destroying it. That is a design error, not a weak ligase.

The 2008 paper is the citation for the concept of precision type IIS cloning in one pot. Laboratory notes in the Addgene protocol collection show the shape of a modern setup. Follow the mix you opened when the two disagree. Addgene's molecular biology reference is background vocabulary. It is not a substitute card.

Side by side with ligation and Gibson

Classical restriction ligation cuts at a site you already have, inside or at the edge of that site, and seals compatible ends. The site often remains, which is useful when you want to recut the clone later and awkward when the site sits in a coding sequence you cannot scar. Self-ligation is the characteristic failure, especially with one enzyme. Methylation can block the site on plasmid DNA from ordinary E. coli. Direction is real only when the two ends differ.

Gibson assembly does not use a restriction site. Overlaps, commonly 15 to 40 base pairs, are chewed back and filled. The junction can be seamless. Repeats in the overlap, a melting-temperature mismatch, and exonuclease loss of a very small fragment are the characteristic failures. There is no recognition site to domesticate, and there is also no enzyme pushing uncut parent back into the reaction.

Golden Gate sits between them. It is still a restriction enzyme and a ligase, so phosphates, ATP and star conditions still matter. The overhangs are designed, like Gibson overlaps, but they are a few bases rather than a few dozen, and they exist because the cut is offset from the site. The site can be lost, which Gibson achieves by never having one, and which classical ligation often cannot. Multipart order is the practical win: many pieces, one tube, order encoded by overhangs. The practical cost is domestication. A gene that contains BsaI cannot be a BsaI part until that site is removed or the enzyme choice changes to one the gene lacks, such as moving a BbsI plan onto a gene that is free of BbsI.

Where the comparison should change your plan

Choose classical ligation when the vector already has two unique, unblocked sites and you are joining one insert. Designing type IIS flanks for that job adds domestication work you may not need. Choose Gibson when the sequences at the junctions are unique enough for 15 to 40 base overlaps and you want a seamless join without a restriction enzyme at all. Choose Golden Gate when you have several parts, you can clear internal sites, and you want the ligation to keep cutting the failures.

Colony screening does not repair a bad overhang table. A missing part can still give a colony if the backbone closed on a repeated overhang. PCR across the whole insert shows size. It misses a swapped middle fragment of similar length. Digest with an enzyme that cuts once per part, or sequence across every junction, before you call the order correct.

An overhang that is palindromic can ligate to itself. Avoid it when you assign the few bases. An overhang that differs by one base from its partner will sometimes ligate and leave a mismatch the cell may repair at random. That is a scar you will not see on a gel. Sequence the junction.

Safety

Type IIS enzymes and ligase are standard research reagents. The product can encode something your institution has not approved for the bench. Containment is their decision. One-pot assembly is not a diagnostic method and does not make a synthetic construct clinically valid. Cite the 2008 paper as the method concept, not as permission.

A cycler that must actually step

One-pot reactions that cycle between two temperatures assume the block did both steps. After a voltage sag, a cycler can finish the clock on a single temperature and still display a completed programme. The tube then had plenty of ligation or plenty of cutting, and not the alternation you planned. Before you commit a rare part set, rerun a small assembly you have already seen succeed, or at least confirm the block still changes temperature. Write the enzyme name and the overhang table into the specification. A verbal "BsaI Golden Gate" without the overhangs is not enough for a second person to order the right fragments.

What an enquiry should contain

Name the enzyme (BsaI, BbsI, or whichever type IIS enzyme the design actually uses), the number of parts, whether internal sites have been removed, and the overhangs at each junction. Ask for the enzyme and ligase classes that match that design. A part set someone else will synthesise can be discussed through the custom gene synthesis enquiry reference. Junction reads can be framed with the Sanger sequencing enquiry reference. Those pages are enquiry references. They are not a statement that EVRINTH runs synthesis or sequencing. Put the map in the quote request and ask whether a quotation is possible.