Proteomics

Today the common application of genomic analysis cannot reveal changes in proteins present in cells as they age, but it is a useful screening approach to focus on certain candidates that may change and can be specifically studied (as described elsewhere in this book). Alternatively, direct proteomic analyses using traditional 2-D-gels with or without antibody (Pawelec et al., 1988) or more modern techniques can be applied. Here, we describe an example of a more recently introduced technique using SELDI ProteinChip® technology from Ciphergen (Freemont, CA). This approach combines the retention of proteins to an array surface by chemical or physical interaction followed by analysis by Laser Desorption/Ionisation Time of Flight mass spectrometry. A reproducible ''protein profile'' is achieved according to mass and has been applied successfully to the study of biological samples for the definition of candidate biomarkers of disease. Increasingly, the technology has been combined with powerful software analogues to distinguish ions that hold predictive significance of disease state, either individually or in combination with other ions. This marriage of analytical chemistry and bioinformatics has proved to be a very sensitive and novel proteomic technology, defining a paradigm in the way proteins can be analyzed in complex mixtures.

Proteomics protocol TCC are lysed in 200 ^l lysis buffer on ice (50 mM Tris HCl, 5 mM EDTA [pH 6.0], 2 mM PMSF, 1% Triton X-100), followed by five short bursts with a sonic finger before centrifugation at 150 x g for 20 minutes to pellet cell debris. Lysates are diluted 1:4 in pH 9.0 binding buffer (100 mM Tris Base) and applied to SAX2 (anion exchange) arrays (Ciphergen) in duplicate. Arrays are read in a ProteinChip Reader (Ciphergen) after the application of 2 x 0.6 ^l sinapinic acid matrix. Generated spectra of masses in the range of 2000-16000 Daltons (Da) are calibrated externally by a mixture of known calibrants. Finally, profiles are aligned and normalized to a common mass occurring in all spectra at a consistent intensity. Using a multilayer perceptron Artificial Neural Network (ANN) (Neuroshell 2) with a back propagation algorithm, models can be generated to sort significant ions consistently predictive of known-age clones by stage, using raw data (mass to charge [m/z] value versus intensity) from mass ranges 2-5, 5-10, 10-16 (see Figure 66.4).

The top 100 ions of predictive value are selected from each and put forward to further analysis to rank a top 50. Ions with high predictive value that do not correlate with protein peak/part of peak signals on the original SELDI data are filtered out. In the example using 77 TCC samples shown here, the resulting ion list contained three ion-clusters at 9417-23 Da (cluster A), 9598-601 Da (B), and 9954-60 Da (C). Each cluster was tested individually and in conjunction with the other two for predictive ability. Though each cluster held little predictive value individually, in combination A+C and A+B+C were able to correctly predict the age of the test clones with greater than 70% accuracy. The model was then tested on seven unseen, validation clones. Again, both the top 50 and top 20 ions yielded > 80% accuracy, though each cluster proved individually poor. The success of combining A+C could not be repeated on the unseen data, but of note is the ability of the model to predict with greater than 70% accuracy the combination of A+B+C, reflecting the result of the test data.

The selection of ion clusters around a certain mass and then cross-referencing with the SELDI data for the presence of a peak is an important step to eliminate artifacts from the data processing. For example, the ion with by far the most predictive power is a single ion found at 6463.08 Da but is very much within the noise of the SELDI data. This cannot be a protein or peptide. However, clustered ions in the top 20 could be correlated to peaks on the SELDI, and their predictive power established on seen and unseen data with some degree of success. Potentially, these proteins can be identified by biochemical methods and characterized to establish their true value within the aging model system studied here. This example illustrates the potency but also the potential pitfalls of this powerful combined proteomic/ bioinformatic approach and the degree to which caution must be exercised in interpreting protein pattern data.

Blood Pressure Health

Blood Pressure Health

Your heart pumps blood throughout your body using a network of tubing called arteries and capillaries which return the blood back to your heart via your veins. Blood pressure is the force of the blood pushing against the walls of your arteries as your heart beats.Learn more...

Get My Free Ebook


Post a comment