In order to investigate how proteins interact with each other to control complex biological processes, it is necessary to measure the “state” of many different proteins in a series of related samples. The “state” of a protein includes several parameters: abundance; post-translational modification; activity (if it is an enzyme); subcellular localization; and interaction with other proteins. While recording all of these parameters for every protein involved in a given process would provide the most complete picture of a biological system, this is not currently possible. We are developing high throughput methods that report on two of these parameters: abundance and post-translational modification. Significantly, this technology enables us to focus on specific proteins of interest.
There are three main types of quantitative immunoassay that are compatible with microarray technology and can therefore be readily multiplexed (Fig. 1). In a protein capture experiment (Fig. 1a), every protein in the sample (cellular lysate) is chemically labeled with a fluorophore and the mixture of labeled proteins subsequently applied to a microarray of antibodies. Each antibody captures its cognate protein and the amount of each protein is subsequently quantified by scanning the array for fluorescence. While this strategy is attractive in that it is scalable, it is limited in several ways. First, since proteins are very different from each other in their physical and chemical properties, it is difficult to label every protein in a complex mixture to the same extent. Some proteins become over-labeled and are no longer recognized by their antibodies, while other proteins are under-labeled and hence not detected. Second, since most antibodies crossreact with off-target proteins, they tend to capture other cellular proteins, in addition to their cognate antigens. This leads to inaccuracies in quantification. Finally, the strategy described above does not permit signal amplification beyond that afforded by the use of fluorophores and so is not particularly sensitive.
Figure 1. Types of immunoassays compatible with microarray technology.
To circumvent these issues, a multiplexed sandwich immunoassay can be used instead (Fig. 1b). This entails capturing unlabeled proteins from the sample with immobilized antibodies (capture antibodies) and subsequently quantifying the amount of captured proteins using labeled antibodies that recognize a different epitope on the protein (detection antibodies). Unlike the first strategy, this assay does not require chemical labeling of the proteins. In addition, each protein must be recognized by two different antibodies, thereby reducing inaccuracies arising from antibody cross-reactivity. In collaboration with Peter Sorger’s lab at MIT, we have shown that this strategy can be used to quantify the abundance and phosphorylation states of cellular proteins (PNAS 2003, 100, 9330). While this strategy is well suited to experiments in systems biology, it suffers from one very significant hurdle: most currently available antibodies do not perform well in this assay. In our experience, only about 5% of commercial antibodies function effectively as capture antibodies on microarrays. This correlates with the observation that, while many antibodies can recognize denatured, immobilized proteins in the context of an immunoblot (western blot), most are unable to immunoprecipitate proteins out of solution. Based on this analogy, we have turned our attention to the third strategy for multiplexed protein quantification: the direct immunoassay (Fig. 1c).
In a direct immunoassay, a complex solution (cellular lysate) is itself immobilized on a solid support, typically a nitrocellulose membrane. In a macro format, this is commonly referred to as a dot blot. Liotta and coworkers adapted this assay to a microarray format (abstract). They spotted lysates from patient matched human tissues onto a series of nitrocellulose-coated glass slides and probed each slide with a different antibody. In this way, they were able to follow several molecular markers and pro-survival checkpoint proteins at the microscopic transition stage from histologically normal prostate epithelium to prostate intraepithelial neoplasia to invasive prostate cancer. They referred to this assay as a “reverse phase protein microarray”. While many more antibodies are likely to function in this assay relative to the sandwich-style assay, one drawback of this approach is that, as indicated above, most antibodies cross-react with off-target proteins. Thus the signal arising from a given spot is the sum of the signal arising from the antibody binding to its intended target and from the antibody binding to off-target proteins. In a traditional experiment, one circumvents this problem by performing a western blot: cellular proteins are separated from each other by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to the solid support (membrane) before probing with an antibody. In this way, off-target proteins can be ignored and the assay yields more accurate information. We are developing high throughput technology that recapitulates the desirable features of a western blot by replacing SDS-PAGE with automated nonporous reverse phase column chromatography and by replacing electrophoretic blotting with protein microarray technology (Fig. 2). We are using this high throughput, miniaturized assay to study, on a systems level, the ErbB signaling network in cultured cells. We are also adapting this assay for profile-oriented screening (see small molecule discovery and Mark Sevecka).
Figure 2. Strategy for protein profiling using automated nonporous reverse phase column chromatography coupled with lysate microarray technology.