Here is some informal background information that will hopefully provide support and context if you want to use ET recombination. Perhaps needless to mention, this information has not been refereed, is, in part, anecdotal and should not be cited. Alternatively, you may be looking for

• descriptions and sequence files of the plasmids used for DNA engineering in E.coli by ET cloning and site specific recombination, (additions to this page are ongoing),
• fuller experimental protocols for ET cloning
• the list of oligonucleotides used in Nature Genetics, 20, 123-8 (1998).
• the list of   oligonucleotides2 used in Nature Biotechnology (?)

ET recombination is straightforward and we are not aware of any peculiarities of the protocols we use. The following points may be helpful. So far we have received encouraging stories of success from many labs and, inevitably, several stories of frustration. Three common problems are emerging. First, as included when this advice page was first posted, the most common source of background is carry over of the template plasmid used for the PCR reaction. Our protocol includes a Dpn1 digestion to eliminate the template plasmid leaving the PCR fragment. If this digestion step is not working then you will find that most colonies simply contain the template plasmid. One way in which the Dpn1 digestion can go wrong is if the template plasmid is grown in a dam- E.coli host. It will then not be methylated at GATC sites and consequently, like the PCR product, not be a substrate for Dpn1. Second, the sbcA strains, JC9604 and JC8679 have not been modified in any way to knock-out the endogenous restriction systems. This limits what DNA can be introduced into these strains. In particular, we and others have never been able to transform these strains with BACs. This limitation should also block the introduction of certain PCR fragments. To modify BACs, please use the endogenous BAC host and our plasmid based expression plasmids - either pBADETg or pBADabg . Third, when selecting for integration of antibiotic resistance genes, it can be worthwhile to titrate the selection pressure. Please consider that the commonly used concentrations for most antibiotics are based on the presence of a corresponding resistance gene on a high copy plasmid. Sometimes when these genes are placed in a low copy context, e.g. a BAC, insufficient expression occurs to convey resistance to the usual (high) levels of antibiotics used. Other points - apologies to people who received the PCR template plasmids, pPGK -FRT and pPGK-loxP. These plasmids are more convenient templates than the plasmids pKaZ and pKaX, however require different PCR primers. You can use upper primer 5'' TCTGCAAACCCTATGCTACTCCGTCG 3'' lower primer 5'' TCCCGGCGGATTTGTCCTACTCAGGAGAGCG 3'' We strongly recommend that you try a simple, model experiment first. For example, try the experiment described in Fig. 5 of our Nature Genetics paper. You can use either the E.coli strain, NS3145 as described in Fig. 5, or the more readily available E.coli strain, HB101. So far all selectable marker genes we have tried (kanamycin, chloramphenicol, ampicillin, zeocin, tetracycline, blasticidin, hygromycin) have worked. Of these chloramphenicol is preferable since it gives a lower background of pseudo-resistant colonies (that is, occasional colonies which appear on plates but do not grow when picked for liquid culture). Any target DNA appears to be suitable - we have no experience of unsuitable targets. However, there is a difference in total number of colonies obtained with different targets. As one might expect, high copy targets give more total colonies than medium, than low. Choice of E.coli strain for the pBAD-ETg pBADabg based approach. Both recBC+ and recBC- strains can be used. The commonly available strains, HB101 or DK1, the common BAC host, DH10B, and the common P1 host, NS3145, give good results. IMPORTANT! Make sure that your electrocompetent cells give 10.8, or greater, colonies per microgram after tranformation of a standard high copy plasmid. If you cannot make good electrocompetent cells, then you will have a hard time. So far, other methods of making competent cells have failed to work. In this regard, our protocol for making electrocompetent cells includes the second ice-cold wash omitted in some protocols for making electrocompetent E.coli. An unexpected anecdote has emerged - old electroporation machines (as often relegated for use to electroporate E.coli) can be very unreliable. PCR to generate the linear fragments.

• There is nothing special about the PCR conditions we use. Initially we purified our oligos by HPLC but then found that this is unnecessary as long as the oligo is extracted once with phenol:chloroform. We do not kinase the oligos.
• The most significant source of background is residual template carried over from the PCR reaction. To eliminate the PCR template, a Dpn1 digestion is included in the protocol. If you get complex plasmid products which contain the selectable marker (which is of course in both the template and the PCR product), then a likely reason is that the Dpn1 digestion did not work.
• An instructive case for us of a PCR fragment not working - in this case we were trying to generate a relatively long PCR fragment (~4kb) and needed to force the PCR conditions beyond 35 cycles to get a good yield. (In other cases we have had positive results with long PCR fragments - longest so far is 6.7 kb). We think it is possible that under these sub-optimal conditions the ends of the PCR fragment were frayed or damaged. Obviously the homologous recombination reaction relies on the quality of the homology arms, that is the ends of the linear fragment. For the same reason, another likely source of problems is poor quality of the synthesised oligonucleotides. Now that the EMBL oligonucleotide service has been dropped and we are forced to rely on commercial sources, we know that the quality of oligonucleotides is critical and can vary wildly, even from the same supplier. If a positive control works in parallel, we send back new oligonucleotides and insist on resynthesis without cost.
• Length of the homology arms - if you have read the Nature Genetics paper closely, you will have noticed that we started our experiments with 42 nt homology arms (this was an arbitrary choice based on nothing other than the answer to Life, the Universe and Everything - Hitchhikers Guide to the Galaxy). Although 42 nt clearly works, Fig.1c of the paper shows that longer homology arms work better and we now routinely use 50 nt homology arms. Use longer arms if you want but in our hands 50 is the chosen balance between efficiency and cost. We commonly order oligos that are 68 mers (50 for the homology arm, 18 for the PCR primer) or 75 mers (50 for the homology arms, 25 for the PCR primer). The difference between the 68 mers and the 75 mers is based on efficiency in the PCR reaction. In our hands, 18 nt PCR primers can require optimisation based on, for example, use of the program "Oligo 4", whereas 25 nt PCR primers generally work on any chosen sequence.
• Sequence content of homology arms - We do not apply any criterion to the sequence of the homology arms other than simply to direct the linear fragment to the place we want. In other words, we do not think that the homology arm should, for example, start with an T or C, or be approximately 50% G/C s or anything like this. On the other hand, we have only a finite set of experiences and have not tried to target repetitive sequences. We have had one instructive example of an unexpected result. In this case, in addition to the 2 flanking homology arms, the PCR fragment also carried an unsuspected internal 90 nt patch of homology shared by the target plasmid (it was a piece of lacZ, there are bits and pieces of lacZ in many unexpected places). From this reaction we got 2 products (and only 2), the first being the intended and the second being a rearrangement through the 90 nt homology region. The take home message here is make sure that the linear fragment does not carry other patches of homology to the target (or at least be aware of them - the correct product may still be obtainable).