A database of mass spectrometric assays for the yeast proteome

Abstract

To the Editor: The current most widely used mass spectrometry (MS)-based proteomic methods sample the available proteome in a quasi-random manner. In each analysis only a subset of the proteins contained in a sample are measured, and replicate analyses sample partly overlapping proteome segments. In addition, these analyses are biased toward abundant proteins1,2. This limits the feasibility of consistently and comprehensively measuring defined sets of proteins (such as proteins constituting signaling or metabolic networks) across different samples, thus precluding the generation of complete datasets to support mathematical modeling of biological processes. To overcome these limitations we previously proposed a targeted proteomic strategy1,2 based on a MS technique called selected reaction monitoring (SRM, also referred to as multiple reaction monitoring (MRM)). This strategy can be used to generate specific, quantitative MS assays for the proteins of interest, which can be then applied for protein detection and quantification in multiple biological samples. First, for each target protein, proteotypic peptides (peptides that unambiguously represent these proteins and are preferentially detectable by MS (ref. 3)) are selected. Then precursor/fragment ion relationships are established, which identify each proteotypic peptide. These consist of pairs of massto-charge ratio (m/z) values that are selected by the two mass analyzers of a triple quadrupole (QQQ) mass spectrometer to isolate the targeted peptide ion and corresponding, diagnostic fragment ion(s). The detector then counts the analytes matching the defined relationship(s) and returns a signal intensity over the chromatographic time. These relationships, termed SRM (or MRM) transitions, therefore effectively constitute MS assays that identify a peptide and, by inference, the corresponding protein in a complex digest. The assays are quantitatively accurate, in particular if isotopically labeled standards are used. SRM measurements result in higher sensitivity (low-attomolar4) and specificity compared to other MS-based proteomic techniques. Yeast proteins spanning all ranges of abundance, from 1.3 million to <100 copies/cell can be identified and quantified by SRM in unfractionated whole yeast digests (P.P. et al., unpublished data). However, despite these favorable characteristics, SRM has not been broadly applied in proteomics, because of the effort required for establishing SRM assays for every protein. The initial selection of optimal proteotypic peptides is complicated by MS signal responses that vary greatly for different tryptic peptides of the same protein. In addition, for each proteotypic peptide, predominant fragment ions have to be selected that define the SRM transitions. As most fragment ion spectra are generated on instruments other than QQQ mass spectrometers, and operating conditions are often poorly documented, the commonly accessible fragment ion spectra can only be a starting point for establishing optimal Figure 1 | Categorization of the yeast proteins represented in MRMAtlas by function and cellular abundance. (a) Grouping of open reading frames by biological processes in S. cerevisiae is according to the Gene Ontology database nomenclature. The same ORF can map to more than one Gene Ontology biological process. (b) Yeast protein abundances in MRMAtlas are derived from a published dataset5. Organelle organization or genesis