Background: Proteomics is expected to play a key role in cancer biomarker discovery. Although it has become feasible to rapidly analyze proteins from crude cell extracts using mass spectrometry, complex sample composition hampers this type of measurement. Therefore, for effective proteome analysis, it becomes critical to enrich samples for the analytes of interest. Despite that one-third of the proteins in eukaryotic cells are thought to be phosphorylated at some point in their life cycle, only a low percentage of intracellular proteins is phosphorylated at a given time.
Methodology/principal findings: In this work, we have applied chromatographic phosphopeptide enrichment techniques to reduce the complexity of human clinical samples. A novel method for high-throughput peptide profiling of human tumor samples, using Parallel IMAC and MALDI-TOF MS, is described. We have applied this methodology to analyze human normal and cancer lung samples in the search for new biomarkers. Using a highly reproducible spectral processing algorithm to produce peptide mass profiles with minimal variability across the samples, lineal discriminant-based and decision tree-based classification models were generated. These models can distinguish normal from tumor samples, as well as differentiate the various non-small cell lung cancer histological subtypes.
Conclusions/significance: A novel, optimized sample preparation method and a careful data acquisition strategy is described for high-throughput peptide profiling of small amounts of human normal lung and lung cancer samples. We show that the appropriate combination of peptide expression values is able to discriminate normal lung from non-small cell lung cancer samples and among different histological subtypes. Our study does emphasize the great potential of proteomics in the molecular characterization of cancer.