Abstract

The identification of protein interactions is a central task in biology today, which has been greatly simplified by the invention of yeast based screening assays such as the yeast two-hybrid system. However, yeast two-hybrid systems exclude several important protein classes, such as integral membrane proteins or transcription factors. Here, we demonstrate the ability of the DUALhunter system to detect interactions between trancriptionally active proteins and to identify novel protein interactions by library screening.

While traditionally, biochemical methods such as co-immunoprecipitation or affinity purification have been used to characterize protein complexes, the invention of yeast based screening assays such as the yeast two-hybrid system ¹ has resulted in a dramatic increase in the number of novel protein interactions described in the literature since they enable the quick identification of novel protein interactions by virtue of cDNA library screening. A protein of interest is assayed against a complex mixture of full-length proteins, protein domains and protein fragments expressed from a cDNA library and those proteins or fragments interacting specifically with it are isolated. The fact that no prior knowledge of the interaction partner is needed makes these systems extremely powerful tools for protein interaction discovery. However, the yeast two-hybrid system is limited to proteins which are able to translocate to the nucleus and which do not display any autonomous capacity for activating transcription ². This excludes several important protein classes, such as integral membrane proteins, transcription factors or strongly acidic proteins. To circumvent these limitations we set out to  develop a flexible screening system which would allow the use of potentially any cytoplasmic protein as a bait for screening, called the “DUALhunter system”.

Principle of the DUALhunter system

The DUALhunter system is based on the well-known split-ubiquitin complementation assay ^3-5. To screen a protein of interest for novel interactors, the protein of interest (the bait) is fused at its N-terminus to the small membrane protein Ost4p and at its C-terminus to a reporter module encompassing the C-terminal half of ubiquitin (Cub) followed by a transcription factor (Figure 1A). Ost4p serves to anchor the bait at the membrane, whereas Cub and the attached transcription factor are needed to detect the protein interaction. Potential interactors (preys) are expressed from a cDNA library as fusions to a modified N-terminal half of ubiquitin (NubG, Figure 1B). Interaction between the bait and a prey results in formation of split-ubiquitin from Cub and NubG. Ubiquitin specific proteases (UBPs) present in the yeast cell recognize the split-ubiquitin and release the attached transcription factor by cleavage of the polypeptide chain after Cub. The transcription factor then translocates to the yeast nucleus and activates a set of cognate reporter genes, thereby converting the protein interaction into a transcriptional readout which is easily measured, for example as growth on a selective minimal medium (Figure 1C).

Figure 1. (A) Principle of the DUALhunter system. A protein of interest (the bait) is inserted between the membrane protein Ost4p and the C-terminal half of ubiquitin (Cub) followed by the artificial transcription factor LexA-VP16. (B) A second protein (the prey) is fused to the mutated N-terminal half of ubiquitin (NubG). (C) If bait and prey interact, Cub and NubG complement to form split-ubiquitin, followed by cleavage and translocation of the transcription factor to the nucleus and transcriptional activation of endogenous reporter genes.

Pairwise interaction assays between known proteins

We chose Uri1p ⁶, a protein with proposed roles in protein translation and folding, as a model protein since our previous experiments have shown that Uri1p cannot be used in a classical yeast two-hybrid assay due to the presence of a long acidic region which autonomously activate transcription. Using a classcial two-hybrid system, the only remaining choice is to divide the protein into subdomains or fragments and to screen those in isolation. However, this may result in a loss of potential interactors. In contrast, the DUALhunter system allows the use of full-length Uri1p. Figure 2 demonstrates pairwise interactions between Uri1p and two of its known interaction partners, Pfd6p and Rbp5p. Yeast was transformed with bait and prey plasmids expressing Uri1p and Pfd6p or Rbp5p respectively, and the interactions were assayed by growth on selective minimal medium. Several unrelated control preys were used to demonstrate the specficity of the interaction.

Figure 2. Pairwise interactions in the DUALhunter system. Yeast co-expressing the indicated baits and preys was plated either on medium selecting for the presence of both bait and prey (left column) or on medium selecting for a protein interaction (right column). NubG: empty vector expressing only NubG. Gal80p, Pex4p and Pfd2p: non-cognate proteins used as negative controls.

Screening cDNA library to identify novel interactors

We then screened the Uri1p bait against a S. cerevisiae cDNA library to identify novel interaction partners. Seven million independent clones were screened, resulting in 63 putative interactors. These clones were analyzed in detail, and clones encoding the same protein were subgrouped. In total, we identified 21 different putative interactors of Uri1p in this screen. Of those, 15 clones interacted reproducibly with Uri1p in a bait-dependency test and did not interact with several unrelated control baits. The 15 novel Uri1p interactors are shown in Table 1. Several interactors were found to be involved in protein translation, supporting a previously suggested role for Uri1p in this process. Co-immunoprecipitation assays confirmed the interactions between Uri1p and several of the newly identified proteins (data not shown). In summary, the results of the cDNA library screen show that Uri1p interacts with several members of the translation and protein folding machinery, confirming the proposed role of Uri1p in protein translation and suggesting a novel role for Uri1p in protein folding. We have also used several other transcriptionally active proteins in the DUALhunter assay, including p53 and its binding partners, and several members of the NF-KkB complex (data not shown). The results show that the DUALhunter system is very flexible and can be used to quickly identify novel interaction partners of those proteins which are unsuitable for use in classical yeast two-hybrid systems.

Table 1. Novel interactors of Uri1p identified by DUALhunter screening. Confirmed: confirmation of screening results by co-immunoprecipitation. ND: not done.

References

1. Fields, S. & Sternglanz, R. Trends Genet. 10, 286-292 (1994).

2. Fashena, S.J. et al. Gene 250, 1-14 (2000).

3. Johnsson, N. & Varshavsky, A. Proc.Natl.Acad.Sci.U.S.A. 91, 10340-10344 (1994).

4. Stagljar, I. et al. Proc. Natl. Acad. Sci. U. S. A. 95, 5187-92 (1998).

5. Thaminy, S. et al. Genome Res 13, 1744-53 (2003).

6. Gstaiger, M. et al. Science 302, 1208-12 (2003).