Protein-detecting arrays may be divided into those that use antibodies and those based on using tags. In the ELISA assay, antibodies to specific proteins are attached to a solid support, such as a microtiter plate or glass slide. The protein sample is then added and if the target protein is present, it binds its complementary antibody. Bound proteins are detected by adding a labeled second antibody.
Another antibody-based protein-detecting array is the antigen capture immunoassay (Fig. 9.25). Much like the ELISA, this method uses antibodies to various proteins bound to a solid surface. The experimental protein sample is isolated and labeled with a fluorescent dye. If two conditions are being compared, proteins from sample 1 can be labeled with Cy3, which fluoresces green, and proteins from sample 2 can be labeled with Cy5, which fluoresces red. The samples are added to the antibody array, and complementary proteins bind to their cognate antibodies. If both sample 1 and 2 have identical proteins that bind the same antibody, the spot will fluoresce yellow. If sample 1 has a protein that is missing in sample 2, then the spot will be green. Conversely, if sample 2 has a protein missing from sample 1, the spot will be red. This method is good for comparing protein expression profiles for two different conditions.
In the third method, the direct immunoassay or reverse-phase array, the proteins of the experimental sample are bound to the solid support (Fig. 9.26). The proteins are then probed with a specific labeled antibody.
Both presence and amount of protein can be monitored. For example, proteins from different patients with prostate cancer can be isolated and spotted onto glass slides. Each sample can be examined for specific protein markers or the presence of different cancer proteins. The levels of certain proteins may be related to the stages of prostate cancer. This immunoassay helps researchers to decipher these correlations.
The main problem with immuno-based arrays is the antibody. Many antibodies cross-react with other cellular proteins, which generates false positives. In addition, binding proteins to solid supports may not be truly representative of intracellular conditions. The proteins are not purified or separated; therefore, samples contain very diverse proteins. Some proteins will bind faster and better than others. Also, proteins of low abundance may not compete for binding sites. Another problem is that many proteins are found in complexes, so other proteins in the complex may mask the antibody binding site.
Rather than using antibodies, protein interaction arrays use a fusion tag to bind the protein to a solid support (Fig. 9.27). The use of protein arrays to determine protein interactions and protein function is a natural extension of yeast two-hybrid assays and co-immunoprecipitation. Protein arrays can assess thousands of proteins at one time, making this a powerful technique for studying the proteome. Protein arrays are often used in yeast because its proteome contains only about 6000 proteins. Libraries have been constructed in which each protein is fused to a His6 or GST tag. The proteins are then attached by the tags to a solid support such as a glass slide coated with nickel or glutathione. To build the array, each protein is isolated individually and spotted onto the glass slide. The tagged proteins bind to the slide and other cellular components are washed away. Each spot has only one unique tagged protein.
Once the array is assembled, the proteins can be assessed for a particular function. In the laboratory of Michael Snyder at Yale University, the yeast proteome has been screened for proteins that bind calmodulin (a small Ca2+ binding protein) or phospholipids (Fig. 9.28). Both calmodulin and phospholipid were tagged with biotin and incubated with a slide coated with each of the yeast proteins bound to the slide via
His6-nickel interactions. The biotin-labeled calmodulin or phospholipid was then visualized by incubating the slide with Cy3-labeled streptavidin. (Streptavidin binds very strongly to biotin.) The results identified 39 different calmodulin binding proteins (only six had been identified previously), and 150 different phospholipid binding proteins.