A panoramic perspective on human phosphosites

Abstract

Protein phosphorylation is the most common post-translational reversible modification of proteins and is key in the regulation of many cellular processes. Due to this importance, phosphorylation is extensively studied, resulting in the availability of a large amount of mass spectrometry based phospho-proteomics data. Here, we leverage the information in these large-scale phospho-proteomics datasets, as contained in Scop3P, to analyze and characterize proteome-wide protein phosphorylation sites (P-sites). First, we set out to differentiate correctly observed P-sites from false positive sites using five complementary site properties. We then describe the context of these P-sites in terms of protein structure, solvent accessibility, structural transitions and disorder, and biophysical properties. We also investigate the relative prevalence of disease-linked mutations on and around P-sites. Moreover, we also assess structural dynamics of P-sites in their phosphorylated and unphosphorylated state. Our study shows that the residues that gets phosphorylated are more flexible than their equivalent non-phosphorylated residues. Our structural and biophysical analyses of P-sites in solvent inaccessible (buried) regions of proteins show that these sites are primarily found in multi-site phospho-proteins, where highly dynamic structural transitions can occur upon binding with another protein. Finally, our analysis of the biophysical properties of P-site mutations shows that P-site mutations that occur in structurally rigid regions are more often involved in disease.

A study on experimental bias in post-translational modification predictors

“…Predicted scores of P-sites located in buried regions tend to be lower than those in interface regions, and much lower than those in accessible regions. Similar conclusions can be drawn from secondary structure analysis, where P-sites in random coil regions generally score higher than those in helices and sheets, which is in line with statistical analysis of phosphorylation data 2 .…”

“…The Ramasamy22 data was obtained from our in-house and publicly accessible Scop3P 2,39 database, in which P-sites were annotated from large-scale re-processing of proteomics experiments 40 while requiring additional evidence from PhosphoSitePlus 41 and dbPTM 42 . All annotations were matched with protein sequences from UniProtKB/SwissProt (version 2021-11), discarding proteins when one of its annotations did not match with the sequence data.…”

“…On average, for the one-hot encoded model, W/C/F/Y/I/V/M/L are considered least favourable when in the proximity of a P-site, while for the PLM, this is the case for F/Y/C/V/I/W/L. As P-sites need to be reached by kinases for phosphorylation to take place, the majority of P-sites are found at the solvent exposed area of the protein (more than 80%, while less than 2% are located in buried regions) 2 . Hydrophobic (A/F/I/L/M/V/Y/W) and C residues 45 are more abundant in buried regions 45 , suggesting that the models implicitly learn that P-sites occur less frequently in buried regions.…”

“…The substrate preferences of kinases depend on the structural and physicochemical properties of the target substrate. Phosphorylation sites (from here onward referred to as P-sites) are most found at accessible and disordered regions of a protein 2 . In addition, the amino acid sequence and structure at the P-site are important, with for example proline-directed kinases predominantly targeting S/T residues with a neighbouring proline downstream.…”