Cloning problem: only frameshift mutants survive - (Aug/13/2007 )
Seconded.
My PhD training was in E coli transcription. Plasmids are full of transcription and DNA replication start sites (both primed by RNA pol), they are circular, and no trasncription terminator is 100% efficient. (Good terminators are ~80% efficient.) Many plasmids have copy numbers of 50 or more per cell. You can now imagine why you are getting expression of your protein.
Going with a low copy numbver system (even a phage system which would get you down to 1 copy per cell) will save you much heartache in the end. And your lab mates will be impressed that you spearheaded this system for your lab, and that you didn't look for a quick fix but instead got to the root of the problem.
If you want to prove to yourself that you are getting expression (I would bet $100 that you are) you could do an RNA prep and reverse-transcription from your frameshift clones, and PCR from the resulting rtDNA.
I think it is unlikely that non-specific or run-through transcription could cause this toxicity. Although I have never researched transcription in detail, I have done a lot of work on gene expression in E.coli. Using expression vectors where the exogenous gene is under a very tightly controlled promoter (e.g. the pBAD arabinose promoter or a T7 promoter in Novagen's pLysS strain) even highly toxic proteins such as restriction enzymes, mRNA interferases and lambda protein E, to name a few, can be cloned. If non-specific transcription generally gave rise to significant levels of protein expression it would be impossible to ever control gene expression to the level that these genes could be cloned in E.coli. I think that expression is occuring, but it is due to residual expression from the yeast ADH1 promoter on pGBKT7.
In this case, E.coli is only being used to generate enough plasmid to transform the yeast strain that the experiment will be performed in. pGBKT7 is being used because it is a plasmid specifically designed for yeast 2-hybrid work, so cloning into a non-specific, low copy number E.coli expression vector, or a phage system, does not get you any further forward. Unless a low copy number alternative to pGBTK7 is available, the long way round would be to make a pGBKT7 derivative with a low copy number origin, but this is a significant amount of work. BEfore going down this road, I would try to a quick fix to quickly get around this problem so that you can get to the experiments that really matter. I suggest repeating the transformation of your ligation, with the following changes:
1. Add of glucose and/or ethanol to the recovery medium and selection plate, as suggested by dpo.
2. Keep the recovery step and selection plates at room temperature, as suggested by perneseblue.
3. Try using an increased kanamycin concentration in the plates. The antibiotic resistance on pGBKT7 is kanamycin, which is an inhibitor of translation. If you are lucky, increasing the kanamycin concentration in your plates may repress translation enough to take the level of expression of your gene below the toxicity threshold - try out a range of kanamycin concentrations using control cells to fund the maximum level the cells can tolerate.
4. Try using another E.coli host - sometimes it helps.
Finally, what is known about the gene you are trying to clone? If you can tell, or guess, from its function why it is toxic to E.coli, it may be possible to overcome the toxicity by adding supplement of some sort to the medium.
You could always try to "clone" your insert into pGBKT7 using the yeast themselves using a technique known as gap repair:
Muhlrad et al., Yeast. 1992 Feb;8(2):79-82.Links
A rapid method for localized mutagenesis of yeast genes.
Basically, you cotransform linear vector and linear insert (which must have some degree of homology at the 5' and 3' ends) into yeast and select for the auxotrophic requirement of the (in your case bait) plasmid, which i think is Trp for pGBKT7. Linear vector cannot replicate but homologous recombination inserts your insert into the vector and those cells which grow should have formed the plasmid you want. Then just recover the plasmid from the yeast and sequence to check you have the correct thing.
Thanks everybody for a lot of great suggestions; I will give everything a try and hope it will work!