FAQs Technical questions - Quick Links
- What is the most important success factor for yeast two-hybrid screening?
- Which proteins work well in yeast two-hybrid screens?
- Which libraries are available for screening?
- Can transmembrane proteins work in yeast two-hybrid screens?
- Should I use full-length proteins as baits, or rather protein fragments?
The most important factor for a successful yeast two-hybrid screen is the choice of the bait protein. Some bait proteins produce very reliable interaction data, whereas other baits can give results that are dominated by known false positives and single hits. If a protein is suitable for Y2H screening, is not safely predictable. One hint can be that related proteins (e.g. same protein family), or protein fragments have been screened successfully so far.
The second most important factor is the choice of libraries.
Lastly, it is important to screen under conditions of appropriate stringency. Too high stringency will increase the rate of false negatives, too low stringency will increase the rate of false positives.
At the DKFZ Y2H facility we do a test screen for each bait, with one library of your choice, varying the screening conditions. Based on the prescreening results we choose conditions, that allow a clear separation of positives from the background, and which give a "reasonable" number of hits (in the range of 1 to 50 hits per million library clones). If a prescreen is not successfull, we will cancel the screening and not charge any cost.
- Proteins should be from nuclear or cytoplasmic compartments (but extracellular proteins can work if they do not have disulfide bonds, and the interactions do not depend on glycosylation)
- Proteins should not have transmembrane segements (but transmembrane proteins can work, even though it is not clear how!)
- Proteins should be folded
- Proteins should not activate transcriptional reporters in yeast (auto activation)
We have good experience with pre-transformed libraries from Clontech. Currently, we have libraries available from human brain, fetal brain, testis, liver, heart, skeletal muscle, aorta, lymph node, bone marrow, kidney, liver, ovary and HeLa cells. We use also a human universal normalized library derived form various tissues. In addition, we have libraries available from mouse testis, brain and whole embryo, as well as from A. thaliana, C. elegans and S. cerevisiae.
In addition, we have constructed a library from full-length open reading frames of > 10,000 individually cloned cDNAs. This library has the advantage that it is normalised, and it does not express non-coding parts of the cDNA, or wrong reading frames, which can be advantageous for screens using "sticky" baits.
We also generated a c-termianal version of our ORF libray (AD fusion at c-terminus) covering >10,000 different genes . The library has been used successfully and is an alternative with preference on the correct folding of the n-termial bait protein domain.
Surprisingly, they can. For example, we did a very successful screen with a protein that had four transmembrane domains. Possibly, these proteins insert into the nuclear membrane post-translationally. However, in many cases screens with transmembrane domains are dominated by false positives. Therefore, we recommend to avoid them in yeast two-hybrid bait constructs. An alternative approach is cloning protein fragments (intracellular domains/subdomains).
When using fragments of a bait, there is a danger that those fragments do not fold properly. This can lead non-specific interactions with unfolded preys, which are abundant in classical cDNA libraries. Therefore, fragments should be generated according to domain boundaries, if known. If the organisation of the protein is not known, the full length version may be the best choice, especially if the typical troublesome features (transmembrane domains, coiled coils) are absent.
