The story actually started when I was making gene targeting constructs for 3’UTR knock-in of two other genes not described in this paper. For some reason, it seemed impossible for me to target Neo cassette into the BACs containing the two homologue genes. Every colony I picked from my Kan/Clm double selection plates (the colony numbers were suspiciously low) harbored the Neo plasmid (must come from contamination of incompletely digested plasmid) and the untargeted BAC, they never got recombined together (this is what I later painfully realized). Before I realized the nature of the problem, I proceeded to zap the double resistant bugs again with linearized Amp retrieval construct, this time, I got suspiciously full plates of colonies on the Kan/Amp double plates, but none of the colonies contained the correctly recombined targeting construct. This did not make sense in the naïve assumption of Kan/Amp double positive clones can only happen when Neo is retrieved by Amp plasmid. After I figured out the real problem by doing more minipreps at each steps, all these results reconciled, because the Neo targeting cassette was cloned in a cassette that also carries AmpR, the colonies I picked on Kan/Clm plates are Kan/Clm/Amp triple resistant in nature.
I actually solved these problems by going through the alternative: retrieving the genomic region with Amp plasmid first and then target Neo into the retrieved fragment. During the trouble shooting process, another idea that came to my mind was introducing a fourth selection marker (X) other than Amp to be used in the retrieval construct so the final double selection with Kan/X would clean up all the junk colonies that did not go through the correct recombineering steps (this might well be another naïve assumption if I did get to apply it on the original problems I faced).
Anyways, in the search of marker X, I stepped upon the paper as Reference 17 in my paper. The cloning seemed simple enough, so I just went ahead and cloned the wild-type FabI myself, saving the trouble of MTA paperwork. When it was ready, I happened to encounter the problem with a nasty gigantic insert I created, as described in my paper. So, I think, this is a great chance to test this cool FabI thing I just got in hand. If it just worked at the first time, this paper would never happen, however, it faltered at first try. I was trying to move the insert into the pF vector and picked only two colonies for miniprep. One of them, contained only the original Amp plasmid, but still somehow survived the triclosan selection! So I thought, this is crap, the triclosan selection is not stringent enough to weed out the wild type bugs (however, this never happened again in my repeated tests, as shown in paper, triclosan is good even in low concentration). I wanted to raise the stringency of selection but at the same time worried about the transformation efficiency drop, so I went back to the literature again and found the paper as Reference 15 in paper. I eventually made mFabI and solved the problem as described in paper.
After solving the problem, I think this is a good thing to share and decided to make a paper out of it, so I spent more effort to characterize it and found all the fascinating features of mFabI as reported in paper. I actually did the head-to-head comparison between FabI and mFabI in the recombineering experiment mentioned in paper for a reason. After successfully obtained the targeting construct with mFabI, I still got aberrantly sequence rearranged by-products in large volume cultures, for the insert was so nasty and the whole thing was so big and unstable, even with mFabI, E coli can’t hold it in one piece well. I thought FabI might worth a try and tried, but it was no better. I eventually retransformed the construct into Stbl2 cells from Invitrogen and got my midiprep worked! So Stbl2 rocks! Take home message: If you can get your constructs in SW106 in minipreps, but not getting it in large cultures, you may want to give Stbl2 a try.
I want to thank the editor and reviewer of my paper for one more reason. They forced me to show more solid evidence to support my claim (that mFabI is really better than FabI) than anecdotal circumstantial tales, or they wouldn’t let me publish. So I did the experiments in the last part of Result section. The data was really positive and made this paper a strong one.