Quantitative Proteomics of Human Urine-Isolated Extracellular Vesicles Reveals Significant and Reproducible Enrichment of EV Markers and Renal Drug-Interacting Proteins

Abstract ID 101685 Poster Board 437 Liquid biopsies are becoming a popular method of obtaining a snapshot of an individual's health state for diagnostic purposes, including drug response and disposition. One component of liquid biopsies of note are extracellular vesicles (EVs) that are released by all cell types and contain proteins, metabolites and RNA. These EVs have been detected in all types of biofluids (e.g., blood, urine, cerebral spinal fluid, etc.), however, a common hurdle to utilizing biofluids for exosomal/EV analysis is the reproducible extraction of those vesicles from complex samples in sufficient quantity and purity to provide analytical rigor. While there are several reports on exosome isolation, most focus on isolation of exosomes/EVs from cell culture media or plasma samples. Thus, our aim in this method optimization study was to compare the reproducibility of marker proteins and drug interacting proteins in EVs isolated from human urine using two common enrichment methods–ultracentrifugation and precipitation, utilizing quantitative proteomics. Three independent donor samples were pre-processed via low-speed centrifugation followed by 0.22 μm filtration to remove large particulate matter. Each sample was then split into three subgroups: one control group and two enrichment method groups. EVs from the first enrichment subgroup for each sample were isolated using ultracentrifugation (UC; 100,000 xg, 1 hr), while EVs from the second enrichment subgroup were isolated using the ExoQuick-TC precipitation (PPTBP1 HC2) method (System Biosciences, Palo Alto, CA); the individual enrichment methods were performed in triplicate, on the same day (intraday) and across three separate days (interday). Trypsin digestion of the unprocessed and EV-enriched samples was performed using an optimized method and 1 μg of digested protein each was injected into a nanoLC coupled Thermo Fisher Q Exactive HF mass spectrometer. Proteomic analysis was performed using DIA-NN software. Of the 2142 proteins detected in total, 878 proteins were detected in the PPT-enriched samples, whereas 550 proteins were detected in the UC-enriched samplesHC3 (Figure 1A). The PPT and UC preparations showed 67 and 45 of the top 100 EV proteins,1 respectively, >10 of which had a fold enrichment (FE) ≥1.5 as compared to the unprocessed samples. There were 30 enriched proteins (e.g., ANXA2, ALDOA, LADHA, ENO1, CFL1, CLIC1, and CD63) that were only detected in the EV preparations, whereas, unconcentrated urine samples only showed detectable levels of 16 of these proteins. Additionally, while the UC preparations did not display any unique EV markers, the PPT preparations showed 23 unique EV marker reads (e.g., ACTN4, FLOT1, CD81, and HSP90AA1) (Figure 1B). Five drug disposition related proteins (P-gp, OAT1, MATE1, GUSB, and CNT1) and two drug targets, SGLT2 and URAT1 were enriched between the UC and PPT preparations. Overall, the PPT method showed better intraday and interday reproducibility as compared to the UC method, which was able to distinguish interindividual variability in EV makers and drug interacting proteins. These data suggest potential utility of urine EVs for predicting dispostion and response of drugs in individual patients.