
Some projects need one gene in many places: the same coding sequence has to go into a mammalian expression vector, a lentiviral backbone, a fluorescent fusion construct and a bacterial expression vector, and you want to compare the results.
With restriction cloning, this means designing new primers, buying new enzymes and running a separate subcloning workflow for every vector. Gateway removes most of this. It uses site-specific recombination from lambda phage biology to move a sequence between vectors that have been built to accept it, with no restriction digest and no ligation at any point.
Originally developed by Invitrogen and now sold by Thermo Fisher, the system has two reactions, BP and LR, and they run in opposite directions. This guide explains why the direction is fixed, what the system is good for, and the scars it leaves in your construct.
At a glance
- Mechanism: Lambda integrase recombines defined att sites. No restriction enzymes, no ligase, and no added energy cofactor.
- Design: Add four guanines to the 5' end of your attB primers. Check the reading frame across both att junctions before you commit.
- Watch for: The ccdB gene in empty destination vectors, which will kill a standard cloning strain, and the att scars in the expressed protein.
How BP and LR reactions work
Most cloning methods need a ligase to seal a break in the DNA backbone, and the ligase needs ATP or NAD+ to do it. Gateway needs neither.
The enzyme is lambda integrase, a tyrosine recombinase. It cuts the DNA backbone by attacking it with a tyrosine in its own active site, which leaves the enzyme covalently attached to the cut end. The energy of the original bond is held in this new enzyme-to-DNA bond rather than released, so when the partner strand resolves the junction, the bond can be remade without any outside energy. This is why the reaction is efficient in a tube with nothing but the two plasmids and the enzyme mix.
A recombination reaction like this is reversible in principle. Gateway makes it one-way by controlling which accessory proteins are present in each mix:
- The BP reaction (attB × attP, giving attL and attR) makes your entry clone. You start with a PCR product flanked by attB sites, or an existing attB plasmid, and recombine it with a donor vector carrying attP sites. The BP Clonase mix contains integrase and integration host factor (IHF). It does not contain excisionase, and without excisionase the reverse reaction cannot proceed.
- The LR reaction (attL × attR, giving attB and attP) moves the sequence from the entry clone into a destination vector. The LR Clonase mix contains integrase, IHF and excisionase. Excisionase works as a switch: it stabilizes the complex that takes the insert out of the entry clone and destabilizes the one that would put it back.
The practical point is that the two mixes are not interchangeable. If a BP reaction fails, using LR Clonase will not rescue it.
What the system is good for
The value is in splitting your work in two. The BP reaction is the only step where you design anything. Once you have an entry clone and have sequenced it, everything after that is a recombination reaction with a vector someone else designed.
This means you can run several LR reactions in parallel, one per destination vector, from the same verified entry clone: no new primers, no new digests, no new sequencing of the insert itself. For a gene you expect to study in five contexts, you do the careful work once.
MultiSite Gateway extends this to combinations. It uses orthogonal att site variants (attL1, attL2, and so on) so that up to four entry clones recombine into one destination vector in a defined order. This is how researchers assemble a promoter, a gene and a terminal tag in one reaction, and then swap one element without rebuilding the rest.
Cost and the att scars
Gateway is expensive to run. The Clonase mixes are proprietary, and the donor and destination vectors are engineered products you have to buy. The system earns its cost when the same gene has to go into many vectors. For one gene into one vector, almost any other method is cheaper.
The other cost is in the sequence. Gateway is not seamless. Site-specific recombination leaves att sequences of about 25 bp flanking your gene in the final construct, and they cannot be removed.
If the cassette is expressed, those scars are translated. They become extra amino acids at one or both ends of your protein. Twenty-five base pairs is eight or nine residues. In many experiments this makes no difference, and in others it changes how the protein folds or whether a terminal tag still works. Before you commit to Gateway for an expression construct, look at where the scars sit relative to your start codon, your tag and any domain near the terminus, and decide whether you can tolerate them.
Plan it first
Two things need checking before you order anything, and both are sequence problems rather than bench problems:
- The reading frame: Your gene has to stay in frame across both att junctions, in every destination vector you intend to use. Because the att sites are fixed and the destination vectors are built by someone else, this is arithmetic you do once per vector, and getting it wrong invalidates every clone downstream of the entry clone.
- What the scars do to the protein: A construct can be perfectly in frame and still express something you cannot use, if the extra residues sit inside a domain or block a terminal tag.
In Photo51, you can build the entry clone and the final construct, inspect both att junctions, and translate across them to see exactly what the expressed protein will begin and end with before you buy a Clonase reaction. You can also watch our step-by-step videos: Gateway BP Cloning in Photo51 and Gateway LR Cloning in Photo51.
Troubleshooting
Your plates are bare after transforming an empty destination vector
Symptom: You ran low on a destination vector and tried to grow up more by transforming it into your usual cloning strain. Nothing grew.
Cause: Empty destination vectors carry the ccdB gene between the att sites, as a negative selection marker. CcdB is a bacterial toxin that attacks DNA gyrase and traps it on the DNA, which stops replication and kills the cell. In a normal cloning strain, an empty destination vector is lethal. This is the point of it: after an LR reaction, only cells carrying a successful insert survive, which is what keeps the background so low.
Fix: Propagate empty destination vectors in a ccdB-resistant strain such as DB3.1. These carry a gyrA mutation that stops CcdB binding while leaving the gyrase working. Use a normal strain for the product of an LR reaction, where the ccdB cassette has been replaced by your gene.
The BP reaction gives no entry clone colonies
Symptom: Your primers look right, the PCR gave a clean band, and the BP transformation gives nothing.
Cause: The ends of your PCR product. Integrase needs a little flanking DNA beyond the att sequence itself to bind and position the complex, and a primer that starts immediately at the att site does not give it any.
Fix: Add four guanines to the absolute 5' end of each attB primer, upstream of the attachment sequence. This is in the vendor protocol, and leaving them out is a common reason a first BP reaction fails. If you already ordered primers without them, you will need new primers.
A MultiSite LR reaction gives no correct clones
Symptom: Three entry clones and one destination vector go into one LR reaction, and nothing comes out correct, although each entry clone works on its own.
Cause: Stoichiometry, meaning the relative molar amounts of the partners. A four-way recombination needs all the partners present in comparable molar amounts, and plasmids of different sizes at the same mass concentration are not in comparable molar amounts.
Fix: Calculate equimolar amounts from each plasmid's length rather than pipetting equal volumes or equal masses. If one entry clone is much larger than the others, this correction is substantial. Verify each entry clone separately by a single LR into a simple destination vector first, so that a failure tells you something.
The construct is correct but the protein does not behave
Symptom: Sequencing confirms your insert and the frame, but the protein is the wrong size on a gel, or a terminal tag does not work.
Cause: The att scars. Roughly eight or nine extra amino acids at the junction are part of your protein now.
Fix: You cannot remove them after the fact, so this is a design decision rather than a repair. Check what the scars add by translating the final construct, and if they sit somewhere you cannot tolerate, use a seamless method for that particular construct. Gateway is worth its scars when you need many vectors; it is not worth them for one.