Affymex: a Bioinformatics Online Tool for Optimizing Abundant Protein Depletion and Mass Spectrometry Analysis

Abstract

Specific removal of high-abundant proteins from a proteomic sample can improve the ability of mass spectrometry (MS) to detect low-abundant proteins and enhance biomarker discovery (1,2). Immunoaffinity separation of proteins is one of the best methods for this purpose (3–5). Peptide mass fingerprints have been extensively used for MS to identify proteins and develop proteomic profiles (6–8). However, there is no method to date to predict how a particular high-abundant protein interferes with the MS analysis of specific low-abundant proteins or biomarkers. In order to rationally guide depletion of the high-abundant proteins for MS analysis, we applied peptide mass fingerprints and developed an online program (Affymex) for analyzing molecular interference between high-abundant and low-abundant proteins. Affymex generates isobaric charts by quantifying the degree of overlap of peptide mass fingerprints derived from an interfering high-abundant protein with that from a target low-abundant protein. This degree of overlap is extrapolated by a single integer called mass match value (MMV), which measures how much a particular high-abundant protein would interfere with specific low-abundant proteins during MS analysis. As illustrated in Figure 1, the closer two peptide masses, the higher the MMV generated, and the greater the predicted interference between them. MMV is a central parameter that can be correlated to a variety of other parameters such as molecular weight, isoelectric point, concentration, posttranslational modifications of intact proteins, and/or their peptide mass fingerprint. The data can be plotted in 3-D for visualization of the proteomic landscape. The algorithm was written in C-sharp computer language, with a graphical module written in VB.NET hosted on a Microsoft® .NET web server. The database architecture on the web server was designed to include SwissProt protein sequence database (www.expasy.org) as a primary source of every target protein and an internal database as a secondary source for the predigested high-abundant proteins or other potentially interfering proteins. To predict the degree of high-abundant protein interference with low-abundant protein detection, the following algorithm was implemented. Affymex calculates molecular mass proximity of peptides derived from interfering high-abundant protein (iMM) to the molecular mass of the peptides derived from that of a target low-abundant protein (tMM). The degree of overlap of an iMM to a tMM is visualized by isobar (MMV = f[iMM ~ tMM]) (Figure 1). The isobars are presented graphically as a function of the isotopic peptide mass of the target protein (Figure 1, x-axis). The closer the iMM to tMM, the higher the MMV (MMVmax [iMM = tMM]). Every isobar is indicated according to its source of matching highabundant protein and presents the real value of tMM. For example, what would be the degree of overlap of the iMM derived after trypsin digest of a high-abundant protein [α-2-macroglobulin (A2MG)] with the tMM derived from a protein of interest, interleukin-6 (IL-6)? Figures 1 and 2 provide examples of how Affymex was applied to answer this question. When the A2MG-derived peptide with an