D3.270 - A novel workflow for cell metabolomics: impact of anaphylaxis in a human endothelial system

Cell metabolomics faces challenges in analyzing scarce cell populations, determining optimal cell numbers and distinguishing metabolites derived from cells from those in culture media. To address these issues, we have developed a novel workflow for cell metabolomics analyses.

CD3+ cells were isolated from a healthy donor and distributed into 30 vials (5 replicates for 6 increasing cell numbers ranging from 50000 to 1000000). Samples were analyzed by untargeted lipidomics in positive and negative electrospray ionization modes (ESI+/ESI-). Data processing included quality assurance and imputation of missing values, followed by Spearman correlation analyses between cell numbers and metabolite abundances to determine which features were derived from the CD3+ cells. Only features with a significant (p-value<0.05) and strong (ρ≥0.7) correlation were kept. PCA and clustering analyses were also performed.

Once optimized, the above workflow was applied to an in vitro human anaphylaxis serum-endothelial cell (AX-EC) system. Sera from AX patients were categorized, according to the suspected, clinically determined mechanism, into IgE-mediated (n=8), IgG-mediated (n=8) and MRGPRX2-mediated (n=8) reactions. Serum samples were collected during the reaction (acute) and 2 weeks after (baseline). Then, ECs monolayers plated at 50000 cells/well were incubated with the paired sera for 1h. After washing, ECs were collected for lipidomic analysis. A 4-point calibration line ranging from 25000–100000 ECs was prepared similarly to CD3+ cells. Once selected the metabolites derived from cells statistical differences according to timepoints and/or anaphylactic mechanism were determined by two-way mixed ANOVA or Aligned Rank Transform tests.

For CD3+ cells, 1262 and 249 chemical signals were detected in ESI+ and ESI-, respectively. The 250000–750000 cell range yielded the highest number of significantly correlated features (ESI+ k = 197 and ESI- k = 14). PCA using all correlated features in the 50000-1000000 cell range showed clear clustering across replicates.

For the AX-EC system, 614 and 732 chemical signals were identified in ESI+ and ESI-, respectively, with the optimal range being 25000–100000 cells (ESI+ k = 288 and ESI- k = 347 correlated features). Statistical analysis identified k = 96 ESI+ and k = 97 ESI- metabolites that significantly (p-value < 0.05) differed between acute and baseline timepoints. Out of those, 95 metabolites were annotated, mainly including fatty acids, carnitines, glycerophospholipids and sphingolipids, whose levels were increased in the acute phase. No differences were found in ECs after incubation with sera samples from different mechanisms.

This workflow optimizes cell number requirements for metabolomics, differentiating cell-derived metabolites and supporting specific cellular profiling. Furthermore, we found a distinct lipidomic fingerprint in ECs in response to acute or baseline patient’s sera, highlighting the metabolic alteration of these cells in AX.