Biomedical Engineering Reference
In-Depth Information
corresponding to one member of the library. Exactly analogously, a prey library
is generated by fusing the genes for the proteins of interest to the coding region
for the transcriptional activation domain of the yeast transcription factor. Yeast
cells of the opposite mating type Bare transformed to give a population of cells,
each of which expresses one of the prey fusion proteins.
All yeast cells used in this genome-scale assay bear a special kind of re-
porter gene that allows them to survive on a selective medium. For example,
they may be defective in the gene
HIS3
necessary to synthesize the amino acid
histidine, but they contain a reporter gene where the
HIS3
gene is fused to a
regulatory DNA region containing the DNA sequence necessary for binding of
the transcription factor used in the assay. Yeast cells in which this reporter gene
is expressed can grow and divide on a medium lacking histidine.
In the final step of the assay, the previously transformed yeast cells of op-
posite mating type are allowed to mate and fuse. The resulting diploid cells are
exposed to a medium lacking histidine. Only cells where prey and bait protein
interact physically will express
HIS3
and will thus survive and form colonies on
this medium. The proteins whose interaction allowed them to survive can be
identified through DNA sequencing of their encoding genes.
The first genome-wide protein interaction screens to which the two-hybrid
assay was applied were carried out in the yeast proteome itself. They yielded
maps of protein interactions involving some 1000 proteins (18,30). Variations of
the approach have been applied successfully to analyze protein interactions in
other microbes, such as the bacterium
Helicobacter pylori
(28), and protein in-
teractions between viral and cellular proteins (4, 11).
The yeast two-hybrid approach has several commonly recognized short-
comings. One of them is the use of fusion proteins, which can lead to bait or
prey misfolding. Another problem is that the assay forces coexpression of pro-
teins in the same compartment of a cell or an organism, although the proteins
may not co-localize in vivo. These shortcomings lead to potentially high false
positive and false negative error rates, i.e., to detection of spurious interactions,
and to a failure to detect actual interactions. These error rates may well exceed
50% (7,31).
3.2. Large-Scale Identification of Protein Complexes
Another class of techniques, which I will illustrate with one prototypical
example, is designed to identify the proteins that are part of a multiprotein com-
plex (13,16). The departure point of a typical experiment (Figure 3) is some pro-
tein A of interest, which is attached to a solid support via a chemical tag that
forms part of the protein. A frequently used tag is glutathione S-transferase
(GST), an enzyme that binds the tripeptide glutathione with high affinity. To
attach this tag to a protein A, a fusion gene containing the coding region of GST
