Comparison of Alternative MS/MS and Bioinformatics Approaches for Confident Phosphorylation Site Localization

Over the past years, phosphoproteomics
has advanced to a prime
tool in signaling research. Since then, an enormous amount of information
about in vivo protein phosphorylation events has
been collected providing a treasure trove for gaining a better understanding
of the molecular processes involved in cell signaling. Yet, we still
face the problem of how to achieve correct modification site localization.
Here we use alternative fragmentation and different bioinformatics
approaches for the identification and confident localization of phosphorylation
sites. Phosphopeptide-enriched fractions were analyzed by multistage
activation, collision-induced dissociation and electron transfer dissociation
(ETD), yielding complementary phosphopeptide identifications. We further
found that MASCOT, OMSSA and Andromeda each identified a distinct
set of phosphopeptides allowing the number of site assignments to
be increased. The postsearch engine SLoMo provided confident phosphorylation
site localization, whereas different versions of PTM-Score integrated
in MaxQuant differed in performance. Based on high-resolution ETD
and higher collisional dissociation (HCD) data sets from a large synthetic
peptide and phosphopeptide reference library reported by Marx et al.
[Nat. Biotechnol. 2013, 31 (6), 557–564], we show that an Andromeda/PTM-Score probability
of 1 is required to provide an false localization rate (FLR) of 1%
for HCD data, while 0.55 is sufficient for high-resolution ETD spectra.
Additional analyses of HCD data demonstrated that for phosphotyrosine
peptides and phosphopeptides containing two potential phosphorylation
sites, PTM-Score probability cutoff values of <1 can be applied
to ensure an FLR of 1%. Proper adjustment of localization probability
cutoffs allowed us to significantly increase the number of confident
sites with an FLR of <1%.Our findings underscore the need for the
systematic assessment of FLRs for different score values to report
confident modification site localization.