The vector's cloning-site region really is scanned for each enzyme's recognition sequence — it's the classic pUC19 polylinker
GAATTC…CCCGGGGATCC…AAGCTT, so it happens to carry one authentic site for all four enzymes. The donor fragment only
carries EcoRI sites at both flanks of the gene (plus one internal BamHI site) — that asymmetry is what drives the outcome:
- EcoRI — cuts the vector once and both insert flanks: the gene fragment is excised with two
AATT sticky overhangs that base-pair with the vector's own AATT ends (checked by reverse-complementing the overhang in code, not just labelled). Real recombinant clones form.
- BamHI — cuts the vector once but the insert only once (internally): no gene fragment is released, so ligation can only re-circularize the empty vector. Every survivor lacks the insert.
- HindIII — cuts the vector once; the insert has no HindIII site at all, so it stays uncut and its ends can't pair with the vector's
AGCT overhangs. Only background survives.
- SmaI — a blunt cutter: cuts the vector once, leaving flush ends. Blunt ends can only ligate to other blunt ends, and the uncut insert has none exposed, so again only the empty vector recircularizes.
Ampicillin selection only tests for the resistance gene on the vector backbone — it can't tell a true recombinant from an empty
vector that simply closed back up on itself. That's why even a "successful" plate can carry background colonies, and why the
insert:vector ratio slider (which biases which molecule's ends find each other first during ligation) only changes the recombinant
fraction when a compatible insert fragment actually exists.
EcoRI G^AATTC → AATT BamHI G^GATCC → GATC
HindIII A^AGCTT → AGCT SmaI CCC^GGG → blunt