Biomarkers can be discovered by finding plus/minus or quantitative differences between samples
Biomarker discovery was probably the first application envisaged for proteomics. As early as 1982, it was suggested that two-dimensional gels could be used to detect quantitative differences in protein profiles between healthy individuals and those suffering from particular diseases, although at the time there was no easy way to identify the differentially expressed proteins that were discovered. This all changed in the early 1990s with the advent of mass spectrometry techniques that allowed proteins to be identified by correlative database searching. The combination of 2DGE and mass spectrometry soon became the standard way to find potential new protein biomarkers. An initial strategy was to compare silver-stained gels by eye or using visual analysis software. Spots that were present on one gel and absent on another, or spots that showed obvious quantitative differences between gels, were picked and analyzed by mass spectrometry. The proteins contained within the spots were thus identified, and their relative abundance in different samples was confirmed using other methods. This led to the discovery of numerous potential disease biomarkers, many of which offered the prospect of diagnosis for different forms of cancer, but also for cardiovascular disease, neurological disease, autoimmune and inflammatory diseases, and infectious diseases such as hepatitis.
Cancer has been the primary target for proteomic analysis because it is relatively easy to obtain matched samples of disease and healthy tissue from the same patient in sufficient amounts to carry out 2DGE. Good examples of this approach include the pioneering studies of Sam Hanash and colleagues that identified various biomarkers suitable for the diagnosis and classification of different forms of leukemia. One such study identified the protein stathmin (otherwise known as oncoprotein 18), which functions as an intracellular signal relay in the transduction of growth factor signals, as a reliable biomarker for childhood leukemia. The interesting feature of this particular protein is that only the phosphorylated form is implicated in the disease. Pioneering work was also carried out by Julio Celis and colleagues, who initially used 2DGE to study the changes in protein expression that occurred as cultured cells underwent growth transformation. The knowledge gained from this series of investigations was later applied to the analysis of bladder cancer resulting in the discovery of several markers, including different forms of keratin, which can be used to follow the progression of the disease from normal epithelium through the early transitional epithelium stage to the late squamous cell carcinoma. Another protein, called psoriasin, is shed into the urine of squamous cell carcinoma patients and thus has the potential to be developed as a validated biomarker for disease diagnosis. Breast cancer has also received much attention, particularly since proteins can be isolated from nipple aspiration fluid allowing noninvasive diagnosis. Several potential biomarkers have been identified through the comparative 2DGE analysis of bilateral matched samples of fluid taken from women with unilateral breast cancer.
Despite the many successes that have been reported, 2DGE has a number of disadvantages for biomarker discovery, including its low sensitivity and the requirement for relatively large samples. The information content of 2DGE can be improved through multiplex analysis (difference gel electrophoresis) and the sensitivity can be increased through the use of novel protein stains, or by pre-fractionation of the sample prior to separation. Various strategies for pre-fractionation have also been tested in biomarker discovery projects, including approaches that select a particular component of the proteome for analysis or eliminate a certain fraction of the proteome during analysis. The selection of cell surface proteins on cancer cells by labeling the extracellular portion of cell surface proteins on intact cells with a hydrophilic biotin reagent is an example of the first approach. An example of the second approach is the use of narrow pH range gels or simple chromatographic procedures that select proteins with particular physicochemical properties. In these cases, however, it is beneficial to use even larger amounts of the starting material to provide enough of the protein sample to facilitate the identification of low-abundance proteins. Unfortunately, most clinical samples are small and heterogeneous, and are surrounded by contaminating normal tissue, which makes the detection of useful biomarkers much more difficult. One way to address the problem of contamination is to use laser capture microdissection (LCM), a technique in which particular cell populations can be isolated under direct microscopic visualization.