Tandem mass spectrometry with ultrahigh mass accuracy clarifies peptide identification by database retrieval

Abstract

A platform was developed to analyze MS/MS spectra from large peptides with low part-per-million mass accuracy, including a commercial-grade software suite. Termed Middle Down Proteomics, this platform identified 7454 peptides from 2-20 kDa (1472 unique) from 555 proteins after 23 LC-MS/MS injections of Lys-C digests of HeLa-S3 nuclear proteins. Along with greatly increased confidence for both peptide identification (expectation values from 10(-89) to 10(-4)) and characterization (up to 18% of peptides were modified in some LC-MS/MS runs), fragmentation data with <2 ppm accuracy enabled error tolerant and routine multiplexed database searching-all clearly demonstrated in this study.

Figure 1

Multiplexed peptide identification using high-resolution MS/MS and tailored software. In a single MS/MS experiment, multiple precursor peptide ions were identified by high-resolution fragment ions. Panel A shows overlapping isotopic distributions from a histone H4 (P62805) peptide (AA 61–78) with 10 b-ion and 9 y-ions identifying it and a lamin A/C (Q5TCI9) peptide (AA 284–308) with 4 b-ions and 5 y-ions matching (B) and (C).

Figure 2

Automated detection and localization of diverse types of protein variation. Panel A shows the fragmentation map of a histone H1.4 (P10412) peptide (AA 1–20) with 5 b-ions and 11 y-ions characterizing the peptide as N-terminally acetylated and threonine 17 phosphorylated. Panel B compares the fragmentation maps of β/γ-actin (P60709, P63261) to a γ-actin variant (AA 69–84). A complementary pair of fragment ions clearly distinguishes the modified form of β/γ-actin from the γ-actin variant. Panel C shows the fragmentation map of a RBM25 (Q2TA72) peptide (AA 1–62) with 5 b-ions partially characterizing a putative dimethylation on arginine 8 with an N-terminal acetylation.

Figure 3

Effect of multiple injections on the number of new unique identifications. Because of the stochastic nature of data-dependent acquisition of MS/MS spectra, the number of unique identifications can be increased by replicate LC-MS/MS injections.

Figure 4

Summary of peptide masses identified and their expectation values from a GeLC-MS/MS analysis of the human nuclear proteome. Panel A shows the number of peptides identified in each 1000 Da mass bin, emphasizing the ability of high-resolution MS/MS ability to identify peptides greater than 2 kDa (26% of the peptides detected). Panel B shows the distribution of the E-values from each of the peptides identified. E-values below 10⁻⁴ are confidently identified and do not require manual validation.