We previously showed that MicroSol-IEF is capable of providing high-resolution fractionation of serum samples, resulting in albumin being confined into a single fraction [10, 11]. This fractionation approach has substantially expanded the number of proteins that can be detected by 2-DE since higher protein loads can be analyzed, in most fractions, without interference from the highly abundant albumin. We have also examined a number of commercially available methodologies for depleting abundant proteins from human plasma/serum and found that the Agilent MARS column is highly efficient in depleting the six most abundant proteins (albumin, transferrin, haptoglobin, a-1 antitrypsin, IgG, and IgA) with minimal nonspecific binding of other proteins . Removal of the major proteins allowed higher amounts of serum or plasma to be loaded onto 2-D gels. However, when the minor protein spots were analyzed by LC-ESI-MS/MS, most of these proteins turned out to be proteolytic products of major proteins . To further enhance detection of lower abundance proteins, the depleted plasma/serum was subjected to MicroSol-IEF fractionation followed by 2-DE analysis. This very time-consuming series of 2-D gels only moderately increased the number of protein spots detected (data not shown). Hence, 2-DE is not an efficient method for detecting lower abundance proteins (<mg/mL) of the human plasma/serum proteome.
To overcome these limitations of 2-DE, we developed the protein array pixelation strategy for comprehensive profiling of the human plasma proteome (Fig. 1). The first step is major protein depletion using the Agilent MARS column. Following
reduction and alkylation ofthe unbound (depleted) proteins, MicroSol-IEF is used as the second fractionation step to further reduce the complexity of the plasma proteome. Each fraction is subsequently electrophoresed on 1-D gels, sliced into pixels, digested individually with trypsin, and analyzed by LC-ESI-MS/MS. In initial analysis ofthe data, proteins were identified from peptides that passed the stringent criteria of XCorr > 1.9 (z =1), 2.3 (z =2), 3.75 (z = 3) and DCn > 0.1, which is based on a commonly used relatively stringent published criteria . Due to the concern that the strict XCorr/DCn used may eliminate some correctly identified low-level proteins, we also incorporated an additional scoring scheme, Sf (final score), which was developed by William Lane at Harvard University and is available in the commercial version of SEQUEST Browser. The Sf score examines the XCorr, DCn, Sp, RSp, and Ions scores of SEQUEST using a neural network and combines them into a single score that reflects the strength of peptide assignment on a scale of 0-1. Peptides with Sf score > 0.7 were considered to have a high probability of being correct (William Lane, personal communication). Therefore, peptide assignments by SEQUESTwere also considered positive if they had an Sf value of >0.7, regardless of the XCorr/ACn scores.
In this study, emphasis is given to proteins identified by multiple peptides (>2 peptides) because the chance multipeptide proteins are false positives decreases exponentially with each additional peptide identified . Since multiple peptides with lower XCorr values can provide the same confidence as a single peptide with a high XCorr value , the inclusion of the Sf score in our analysis should not generate a significant increase in false identifications of multipeptide proteins.
Was this article helpful?