Metabolomic Profiling of Extracellular Vesicles and Alternative Normalization Methods Reveal Enriched Metabolites and Strategies to Study Prostate Cancer-Related Changes

Abstract

Body fluids are a rich source of extracellular vesicles (EVs), which carry cargo derived from the secreting cells. So far, biomarkers for pathological conditions have been mainly searched from their protein, (mi)RNA, DNA and lipid cargo. Here, we explored the small molecule metabolites from urinary and platelet EVs relative to their matched source samples. As a proof-of-concept study of intra-EV metabolites, we compared alternative normalization methods to profile urinary EVs from prostate cancer patients before and after prostatectomy and from healthy controls.

Methods: We employed targeted ultra-performance liquid chromatography-tandem mass spectrometry to profile over 100 metabolites in the isolated EVs, original urine samples and platelets. We determined the enrichment of the metabolites in the EVs and analyzed their subcellular origin, pathways and relevant enzymes or transporters through data base searches. EV- and urine-derived factors and ratios between metabolites were tested for normalization of the metabolomics data.

Results: Approximately 1 x 10¹⁰ EVs were sufficient for detection of metabolite profiles from EVs. The profiles of the urinary and platelet EVs overlapped with each other and with those of the source materials, but they also contained unique metabolites. The EVs enriched a selection of cytosolic metabolites including members from the nucleotide and spermidine pathways, which linked to a number of EV-resident enzymes or transporters. Analysis of the urinary EVs from the patients indicated that the levels of glucuronate, D-ribose 5-phosphate and isobutyryl-L-carnitine were 2-26-fold lower in all pre-prostatectomy samples compared to the healthy control and post-prostatectomy samples (p < 0.05). These changes were only detected from EVs by normalization to EV-derived factors or with metabolite ratios, and not from the original urine samples.

Conclusions: Our results suggest that metabolite analysis of EVs from different samples is feasible using a high-throughput platform and relatively small amount of sample material. With the knowledge about the specific enrichment of metabolites and normalization methods, EV metabolomics could be used to gain novel biomarker data not revealed by the analysis of the original EV source materials.

Keywords: exosomes; extracellular vesicles; metabolomics; platelets; prostate cancer; urine.

Figure 1

Validation of EV sample quality. A. Electron microscopy (EM) shows typical round morphology and size range of urinary EVs. B. Urinary EVs with multiple membrane layers (arrows) were seen occasionally. C.-D. Immuno-labelling of EV-markers CD59 and CD63 showed their presence on the urinary EVs by EM. E. The size distribution of the urinary EVs in the EM images indicated that most of the EVs were small, < 100 nm in diameter (N = 3, total of 425 EVs). F. Platelet EVs showed typical morphology and somewhat larger size than urinary EVs by EM. G. Western blotting of the samples from each step of the urinary EV isolation confirmed the enrichment of CD9, TSG101, CD59 and CD63 as well as the absence of cellular organelle proteins calnexin, TOMM20 and GM130 in the EVs. Equal quantities of protein were loaded from all urine-derived samples. Supernatant (sn).

Figure 2

Characterization of the urinary EV samples applied to metabolomics by western blotting and Nanoparticle tracking analysis. A. Western blotting of the urinary EV samples in the metabolomics study with EV-markers demonstrated significant variation in EV quantity from different donors. Two out of three samples from prostate cancer patients of Helsinki Urological Biobank project (HUB.1-3) obtained before prostatectomy (pre) contained more EVs than the samples from the same patients after prostatectomy (post) or from healthy controls. B. The urinary EV concentrations in these samples measured by nanoparticle tracking analysis and by quantification of CD9 band optical density (OD) from the western blot correlated well. C. Size distribution of the urinary EVs applied to metabolomics was obtained by nanoparticle tracking analysis showing that the EV sizes did not vary much between samples (N = 8).

Figure 3

Comparison of the metabolite content and pathways between urinary EVs and platelet EVs. A. Venn-diagram of the metabolites above the quantification limit in urinary and platelet EVs showed an > 50% overlap between the two EV types (derived from healthy individuals). The two EV types contained both EV-type specific (11 and 5 in urinary and platelet EVs, respectively) and common metabolites (21), as well as metabolites that were above the quantification limit in all samples of one EV type, but below this limit in one or more samples of the other (9 in each). Both EV types contained metabolites belonging to five different categories (highlighted with a color code in the image). B. Metabolite set enrichment analysis depicted the metabolic pathways that showed hits in both EV types (number of hits shown) and the total number of metabolites included in the metabolite panel from these pathways. platelet EVs (pEVs), urinary EVs (uEVs). “Intracellular signaling via adenosine…” continues with “receptor A2A/B and adenosine”.

Figure 4

Comparison of the metabolite content in the EVs and their source materials. A. Venn-diagram of the metabolites above the detection limit in all urinary EVs (uEVs) from controls and the matched original urine samples showed overlap, but also sample type specific metabolites. Urine contained more unique compounds (17) than uEVs (1) indicating efficient purification of these from the uEV samples. B. Although platelet EVs (pEVs) and platelets had a highly overlapping metabolite content, the pEVs contained more unique metabolites (11) than platelets (1).

Figure 5

Subcellular localization of the EV metabolites. Database searches using Human Metabolome Database and Small Molecule Pathway Database indicated that the subcellular localization of the metabolites found above quantification limit in the urinary and platelet EVs was mainly cytosolic. Many of the EV metabolites were assigned, in addition to cytosol, to other subcellular locations (the charts marked “+ other”) including mitochondria, ER, peroxisomes and lysosomes. Endoplasmic reticulum (ER), platelet EVs (pEVs), urinary EVs (uEVs).

Figure 6

Most significant changes in the urinary EV metabolites in prostate cancer. A. Metabolite concentrations in the individual urinary EV samples were normalized to the CD9 optical density (OD) determined by western blotting, an EV-derived parameter. The analysis indicated lower levels of four metabolites in the pre-prostatectomy samples (pre) in comparison to post-prostatectomy (post) and healthy control samples. Pre- and post-prostatectomy samples from the same patients (HUB. 1-3) are connected with lines. B. Ratios between two metabolites indicated lower levels of glucuronate, isobutyryl-L-carnitine and D-ribose 5-phosphate in the pre-prostatectomy samples as in A, but also changes in other metabolites. Statistical significance is indicated for the comparisons of pre-prostatectomy group to control and post-prostatectomy groups separately (small brackets) or to the combined control and post-prostatectomy group (large brackets). p < 0.05 (*), p < 0.01 (**).