Appendix 4

Protein-Protein Interactions

Identifying Protein-Protein Interactions

    It is a frequent occurrence in molecular biology that you will want to decide if two proteins interact.  There are a variety of methods one can use.  In real life each method has advantages and disadvantages.  All these tests depend on the strength and specificity of binding, and each has its weaknesses.  Here, I will treat the methods as if they are more or less equivalent.  The problem with all in vitro protein interaction assays is demonstrating that the molecule you find has a function in vitro.

Yeast Two Hybrid System: The idea is to fuse one of the test proteins to the DNA binding domain of GAL4 in an expression vector, and to fuse the other test protein to a transcriptional activation domain in an expression vector.  Transform both of these vectors into yeast and then monitor the activity of a reporter regulated by an upstream activating sequence containing GAL4 recognition sites.  The reporter will be activated only if the two test proteins interact, thus bringing the activation domain to the upstream activating region.  This has proven to be a very powerful method to monitor protein interactions because the interaction occurs in vivo and expression plasmids can make high concentrations of the fusion proteins.  Its major weakness is that there is a high possibility of false positives.  The two hybrid system can also be used as a screen.  In this case, a cDNA library is fused in frame with the activation domain expression vector and the library is transformed into yeast.  

Serebriiskii et al. (2001).  Biotechniques 20, 634-636.

As two hybrid technology became established, a number of labs began tinkering with the system to determine whether it could be used to identify interactions between proteins and an enlarged class of ligands, including RNA.  

Serebriiskii et al. (2001).  Biotechniques 20, 634-636.

Co-immune precipitation (coIP):  Antibodies are bound by protein A, a cell wall protein made by the bacteria Staphylococcus aureus.  If an antibody recognizes a protein X, and protein A from fixed bacteria, or coupled to insoluble beads is added to the mixture, then the protein and antibody will precipitate.  If protein X recognizes protein Y, then antibodies to protein X will also bring down protein Y, since some of the Y molecules in the solution will by bound by the X.  So the idea is to mix pure X and Y, or use a mixture of proteins in which X is present, add antibody, and see if only protein X precipitates, or if other proteins also precipitate.  It is important to show that the antibody does not cross-react with protein Y (use a western blot), or any other protein in the mixture.  Normally, protein Y is detected because it has been labelled (by in vitro translation), or with an antibody directed against protein Y.  

Once interaction is established, you can determine the domains of interactions or the specific amino-acid residues required for interaction using the same approach.  By using deletion derivatives or point mutations we can narrow down the site of interaction.

GST pull-downs:  The glutathione S-transferase (GST) pull-down technique is a type of co-immune precipitation experiment.  It consists of a GST-tagged bait protein (your protein of interest) that can be used to identify putative binding proteins.  The bait protein, purified from an appropriate expression system, is immobilized on a glutathione affinity gel.  The prey protein could come from previously purified proteins, cell lysates or in vitro transcription/translation reactions.  Unbound proteins are washed off the column.  A very high concentration of glutathione is used to elute the interacting proteins.  Protein-protein interactions can be visualized by SDS-PAGE and either Coomassie or silver staining, western blotting or ³⁵S detection.   Negative controls are absolutely necessary for generating biologically significant results.  The minus bait, plus prey control helps identify and eliminate false positives caused by non-specific binding of proteins to the glutathione.  The plus bait, minus prey helps identify and eliminate false positives caused by non-specific binding of proteins to the GST tag of the bait protein.

Mohd-Sarip et al. (2002).  Mol Cell Biol 22, 7473-7483.

Mohd-Sarip et al. (2002).  Mol Cell Biol 22, 7473-7483.

Cross-linking:  Bi-functional cross-linkers have chemically active moieties at each end that can bind to proteins, often through lysines or tyrosines, connected by a short linker whose length can be adjusted by changing the number of carbon atoms.  The idea is that the cross-linker will join only molecules that are closely associated, because only these molecules will be close enough together, and stably associated for long enough to be cross-linked.  The experiment is to add cross-linker to an extract, separate the proteins by SDS-PAGE, and see if there is a mobility shift of the proteins of interest.  The size of the shift tells you the size of the interacting protein.  The protein of interest is usually detected with antibodies or in vitro translated.

Far westerns and expression-screening:  Another approach is to use a solid-phase assay.  In the far western a protein extract is separated by SDS-PAGE, transferred to a filter, blocked and then the filter is incubated with the labelled protein of interest.  If this protein can bind to any protein in the extract it will be visible as a band after exposure of the filter to film.  Alternatively, an expression library can be screened (usually in bacteria).  The cDNA expression library is plated out, induced to express the protein, the bacteria are  transferred to filters and then the filters are incubated with the labelled protein of interest.  The labelled protein will bind only to those colonies expressing a protein with which the labelled protein interacts.  This is one of many variants of expression screening.