The term names a class of assembly methods that join fragments without leaving the extra bases a restriction site would add, which matters when a junction sits inside a coding sequence, a promoter or a fusion protein. The methods differ in what makes the join rather than in the result.
What the name actually describes
seamless cloning describes any assembly that leaves no residual scar at the junction, so the product carries only the sequence that was designed. The alternative, digesting and ligating at a restriction site, leaves that site's bases behind, which is harmless in a plasmid backbone and not harmless inside a reading frame or a regulatory element.
The chemistries that produce no scar
Three routes dominate: an exonuclease and polymerase mix that chews back matching overhangs designed into the primers, a type of restriction enzyme that cuts outside its own recognition site so the site leaves with the offcut, and a recombinase that joins defined ends. The first needs only primers, the second needs a site placed outside the insert, and the third needs its own fixed ends.
Designing the overlap, which is where it fails
The overlap length, its melting temperature and whether it is unique across the whole assembly are what decide whether an assembly works, and repeated or low complexity sequence at a junction is the commonest cause of a failed reaction. An assembly of many fragments is designed as a set rather than pair by pair, because a junction that resembles another will join in the wrong order.
When a restriction digest is still the better route
A single insert into a well characterised vector at existing sites is faster and cheaper by digestion and ligation, and the scar is in a backbone where nothing reads it. The scarless methods earn their cost on multi-fragment assemblies, on fusion junctions and where a promoter or a tag has to sit at an exact base.
Verifying the junction rather than the colony
Every assembly is confirmed by sequencing across each junction, not by a colony count or a size check, because the failure modes of these methods are small indels at the join and fragments in the wrong order. Both look correct on a gel, which is why a plasmid is sequenced end to end before it is used.
Questions people ask about seamless cloning
What is the scar?
The extra bases a restriction site leaves at a junction. Harmless in a backbone, not inside a reading frame or a promoter.
Which method needs no special sites?
The exonuclease and polymerase route: the overlaps are designed into the primers.
How is an assembly verified?
By sequencing across every junction. Small indels and wrongly ordered fragments both look correct on a gel.