2.12.4. Proteomics Data Processing

Proteomics data analysis involved measurement and assignment of MS intensity signal to each identified protein and was performed using MatchRx software as described previously [34]. Briefly, peak intensities of all the ions in each MS run were extracted from the mzXML files and assigned to Mascot-identified proteins using the MatchRx software using their m/z, retention times and neighbouring peak coordinates. Each MS intensity was adjusted using total median normalization as described previously [34]. For each sample, total MS intensity signal was also calculated by summing intensities of all the MS intensity signals in the run and was used to estimate fraction of MS intensity (FMSI) of each protein as follows:

FMSI were used to examine the enrichment or depletion of each protein in EV fractions compared to Clone 92 cells or supernatants. Proteins showing more than two natural log difference (approximately 7-fold) were considered either enriched or depleted. Since FMSI values were calculated using MS intensities, they may not correspond to true protein abundance and hence were not used to compare levels amongst proteins.

The top 50 EV proteins were selected based on the following criteria for high confidence protein identification:

The protein’s Mascot score had to be ≥40 (<1% FDR) with ≥2 peptides and an FMSI fold change ≥7 compared to cells and supernatant.

Keratins were not included in the top 50 list as their presence can be the result of sample processing.

The FMSI value in SEC isolated EVs had to be >0.

Venn diagrams were generated using the BioVenn website [35]. The common proteins identified in both the ExoCarta [36] and Vesiclepedia [37] databases were used for comparison.