I'm trying to insert a 120bp segment into a 4k vector using Gibson Assembly. The purpose is to replace a 27bp portion of the vector with NNKs. The insert is ordered as an ssDNA oligo, and then made double-stranded after annealing to a primer, and completing the second strand with Klenow. It has 36 bp overlap with the vector. The vector backbone is prepared using inverse PCR, because there are no restriction sites around the target area. I then use DpnI to digest the WT vector, and then gel extraction to remove all the DpnI fragments (which can be up to 1kb and tend to ligate back into the vector during Gibson Assembly, so gel extraction is necessary). After cleaning/concentrating the vector product with the Zymo kit, I mix the dsDNA insert with the vector backbone (1:5 vector to insert ratio) with Gibson Assembly, and then chemically transform into homemade MegaX (DH10B) competent cells. The goal is to maximize transformation efficiency, since I'm creating a library.
This has worked for me once, several months ago, using a similar insert that was only 87 bp long (21/18 bp overlaps). Since then, however, it has consistently failed-- I get zero colonies, or at most, a few tiny colonies. I later switched to a larger insert (120bp) since some said that the T5 exonuclease of Gibson Assembly could totally chew up a small fragment. Still, no luck! I can transform the original vector template successfully each time, proving that the vector, antibiotic resistance, transformation protocol, and cells are all good. I've also run a positive control with my Gibson mix, which has worked.
I get colonies if I skip the gel extraction step, but then half of the colonies contain weird combinations of the insert and DpnI fragments that were not removed, so that's not desirable.
Any other thoughts for troubleshooting? I can't understand why this worked over several repeats in the past, and suddenly cannot work anymore. Thank you!
Hey Shivani,
I had a very similar problem with no colonies when I used gel extraction. Turned out that exposing the gel to uv light was the problem. I did control experiments and found even 30s on the lowest setting reduced transformation efficiency ~100x. I would recommend using sybersafe to stain the gel and cutting it on a blue light. Hope that helps.
Alex
Recombinant DNA techniques should always be viewed pragmatically, use the easiest method first and try different approaches when it doesn't work instead of trying to troubleshoot. In my opinion, Gibson assembly is not better than simply PCR assembly when you have small fragments. Whilst PCR assembly works with small quantities of DNA and optimising the PCR conditions, followed by sequencing 2-3 candidates, Gibson assembly only works when you have high concentrations of pur DNA fragments, and often it is the concentration that ends up being too low for getting colonies. Of course, the quality of competent cells are also important, and we always make our own because they are at least 10-20 times better than commercial preps (and much cheaper), but I find PCR assembly much easier to handle. If I had to build a 120bp fragment and insert it into a 4kb vector, I would do this by PCR assembly, or instead get a company to synthesize the fragment with convenient flanking sites, and just subclone it with restriction sites.
Alex-- the UV light issue is a really good point! When this all last worked for me, I was using a collab's blue light box. I'll try that next, thanks!
Jurgen-- a previous student had tried PCR assembly in the past, but it hadn't worked so well for her, which is why we switched to Gibson. But, I may go back to that. A company is already synthesizing the 120bp ssDNA oligo (the target region within it is actually just 27 bp, so the rest of it are flanking sites). I would also use restriction sites to subclone, except that there are no appropriate unique restriction sites for this. If Gibson continues to be a problem, I'll engineering in some convenient restriction sites. Thanks for the tip!
- Badges
- Science topic