Classification filtering strategy to improve the coverage and sensitivity of phosphoproteome analysis

Data dependent neutral loss triggered MS3 methodology (NLMS3) is often applied to acquire MS data for the analysis of phosphopeptides. Some phosphopeptides tend to seriously lose the phosphate and result in MS2 spectra with poor fragments and fragment-rich MS3 spectra, while some phosphopeptides do not lose phosphate and result in nice MS2 spectra. Since different phosphopeptides have fragment spectra with different characteristics, filtering all of the phosphopeptide identifications by setting a global filter criteria may be inappropriate and result in low sensitivity. In this study, we developed a classification filtering strategy to improve the phosphopeptide identification and phosphorylation site localization. Phosphopeptide identifications were classified into four classes according to their different characteristics, and then, the identifications from each class of mass spectra were processed and filtered separately using different filtering strategies. It was found that the overlap of phosphopeptide identifications from different classes was low and the classification strategy significantly improved the coverage of the phosphoproteome analysis. Compared with MS2 strategy and multiple stage activation (MSA) strategy, NLMS3 with the classification filtering strategy was demonstrated to have higher sensitivity and higher performance in localizing the phosphorylation to specific sites.