Size exclusion chromatography is the preferred method for isolating EVs from low volumes of urine for mass spectrometry-based metabolomics analysis
To optimize a reliable method for EV isolation from small volumes of urine, we first compared the yield and purity of EVs isolated from two initial volumes of rat urine (0.5 mL or 1 mL) using three independent isolation methods (Fig. 1): ultracentrifugation with filtration (UC), size exclusion chromatography (SEC), and a proprietary magnetic bead-based commercial method (MBB). After isolation, we measured the size distribution and concentration of EVs from samples using nanoparticle tracking analysis (NTA). NTA showed all three methods resulted in EVs of similar size (Additional file 2: Figure S1), though there were differences in the average size of detected particles (Additional file 3: Table S1). We found that the MBB method provided the highest concentration (particles/mL) of EVs at both initial volumes, but SEC resulted in the highest number of total EVs (yield) based upon the entire volume of working solution obtained by each method (Additional file 2: Figure S1). UC was least efficient (Additional file 2: Figure S1). Typical yields from 0.5 mL of rat urine using the SEC method ranged from 1.42E9 to 2.82E9 particles/mL, while 1.0 mL of urine yielded between 4.66E9 to 1.13E10 particles/mL (Additional file 3: Table S1). We next used cryogenic electron microscopy (Cryo EM) to confirm EV enrichment and analyze the biophysical properties of isolated vesicles. We found that UC and SEC yielded EVs with of varied sizes, though most within the expected range of < 200 nm in diameter (Additional file 2: Figure S2). The MBB method, however, seemed to yield smaller vesicles of a more uniform size, an observation other groups have made using polymer-based EV isolation methods (Additional file 2: Figure S2).
To compare resultant mass spectrometry data quality, we next investigated small molecule profiles of EVs isolated via each method separately using quadrupole time of flight mass spectrometry coupled with ultra-performance liquid chromatography (QToF-LCMS)-based untargeted metabolomics analysis. The total number of features (m/z and retention time pairs) identified for UC (0.5 mL = 4837 and 1 mL = 5141) and SEC (0.5 mL = 5031 and 1 mL = 5096) were comparable. EVs isolated using the MBB method had a significantly higher number of detected features (0.5 mL = 7450 and 1 mL = 7135) (Fig. 2A). Though increasing the starting volume of urine to 1 mL increased EV yield and concentration, the number of detected features only marginally increased (Fig. 2A). Interestingly, examination of total ion current chromatograms (TICs) and Manhattan plots revealed a cluster of features which were uniquely detected within the MBB samples (outlined in red, Fig. 2B–C). Further investigation of the unique chromatographic peaks in MBB samples revealed m/z patterns which were consistent with polymer contaminants. EV preparation using the UC and SEC methods generated MS spectra free of contaminants with profiles similar to each other. We did not observe significant differences in the total mass spectrometry signal between EVs derived from either 0.5 mL or 1 mL of urine (Fig. 2B–C). There was significant overlap in the number of unique features identified by each isolation method, with both 0.5 mL and 1 mL of urine (Fig. 2D–E). Considering the high EV yield, lack of contaminating material, quality of mass spectrometry data and high throughput capability we determined SEC as the optimal method for isolating EVs from 0.5 mL of urine.
The observation of “classical” EV-markers in antibody array immunoblots was used to ascertain isolation and enrichment of bona fide EVs. In accordance with MISEV guidelines , we performed a detailed characterization of the EV preparations. Firstly, we evaluated the expression of transmembrane proteins (such as CD63 and CD81) as well as cytosolic proteins (such as TSG101, ALIX) in the urinary EV preparations. We found that the urinary EV preparations were positive for known EV markers including cluster of differentiation 63 (CD63), cluster of differentiation 81 (CD81), tumor susceptibility gene 101 (TSG101), ALG-2-interacting Protein X (ALIX), intracellular adhesion molecule (ICAM), Annexin5, epithelial cell adhesion molecule (EpCAM), and flotilin1 (Flot1) (Additional file 2: Figure S2). Importantly, pooled samples of fractions not expected to contain EVs showed no enrichment in these targets (Additional file 2: Figure S2).
Pilot study validates EV isolation method and potential utility of urinary EVs as a source of small molecule radiation biomarkers
We next performed a pilot study to investigate the utility of urinary EVs as biomarkers of radiation injury. We isolated EVs using our SEC method from 0.5 mL of urine from a small cohort of rats (n = 5 per group) either sham irradiated or exposed to 13 Gy leg-out PBI and performed QToF-LCMS to characterize their small molecule profiles (Fig. 3A). Principal component analysis (PCA) demonstrated distinct separation in the small molecule profiles of EVs from irradiated rats (Fig. 3B). Visualization using a volcano plot identified many features which were significantly dysregulated (Fig. 3C). A heatmap also showed distinct differential expression patterns between irradiated and sham irradiated EV samples (Fig. 3D). Overall, we identified a total of 72 features which were significantly altered (FDR adjusted p-value < 0.05) in the radiation group (Additional file 3: Table S2). Ultimately, we putatively annotated 21 of these features covering a broad range of endogenous metabolites such as lipids, prostaglandins, peptides or amino acid derivatives, and small molecules such as adrenaline (Additional file 3:Table S3).
Urinary EVs allow for the identification of radiation biomarkers in a large rat cohort
This pilot study confirmed the potential of urinary EVs as a source of biomarkers for radiation exposure. To build on these findings, we applied these methods to a larger cohort of 18 rats and a total of 72 rat urine samples to study biochemical profiles of urine derived EVs obtained from rats exposed to 13 Gy leg-out PBI (Fig. 4A). The WAG/RijCmcr rat model is the most advanced partial body irradiation model, used to assess mitigators of radiation injury to normal tissue . This model is routinely used in our studies related to understanding long-term effects of radiation exposure and the identification, and development, of potential radiomitigators.
In accordance with our observations in the pilot study, LC–MS/MS-based targeted metabolomics and lipidomics revealed robust changes in urinary EV profiles following irradiation. Visualization using PCA showed distinct separation between sham and irradiated groups, starkest 24 h, and 90 days post-irradiation (Fig. 4B). No metabolites or lipids were significantly dysregulated (FDR adjusted p-value < 0.05) at either 14 or 30-days post-irradiation. Interestingly, the majority of the significantly altered molecules (FDR adjusted p-value < 0.05) post-irradiation were lipids (Additional file 3: Table 4). 24 h post irradiation, several lipid classes were downregulated including triglycerides (TAG), phosphatidylcholines (PC), sphingomyelins (SM), hexosyl ceramides (HCER), free fatty acids (FFAs), lysophosphatidylcholines (LPCs) and phosphatidylethanolamines (PE) (Fig. 4C, Additional file 3: Table S4). By 90 days post-irradiation, many of these lipid species had reversed in their relative abundance with several lipids showing a significant upregulation compared to sham irradiated rats (Fig. 4C). The majority of upregulated lipids 90 days post-irradiation were TAGs.
Urinary EVs demonstrate clinical utility for identifying RT biomarkers in human urine samples
Since the goal of our method comparison is to ultimately develop urinary EV-based biochemical analyses to understand how patients respond to RT in the clinic, we next sought to validate our methods in human urine samples. In a pilot study of 5 thoracic cancer patients receiving RT as a part of their treatment regimen, we collected urine pre- and immediately post-RT, and isolated EVs using the above SEC-based method. First, we validated enrichment of urine EVs using immunoblot and NTA, as described previously (Additional file 2: Figure S3). We next leveraged our UPLC-QToF-MS platform to analyze the small molecule profiles within our human urine EV samples. Similar to our findings in rat urine EVs, we detected a substantial number of features (4362 in ESI + and 3111 in ESI-) with low noise TICs (Fig. 5A). To identify biologically relevant altered metabolites, we used LC–MS/MS targeted metabolomics to quantitate the biochemical profile of human urinary EVs. Using our targeted panel consisting of 360 multiple reaction monitoring (MRM) transitions, covering 270 polar metabolites, we were able to reliably quantify 175 MRMs, corresponding to 152 metabolites, with coefficients of variation less than 0.2, indicating stable and reliable quantification of those metabolites (Additional file 3: Table S5).
We next sought to identify whether these metabolite profiles significantly changed post-RT. Interestingly, we found 11 metabolites which were stably quantified and significantly altered post-RT in this pilot study (Table 1, Fig. 5B). Some of these metabolites are involved in nucleotide (Xanthylic acid, imidazole and dTTP) and folate metabolism (5-methyltetrahydrofolic acid and flavin mononucleotide). These changes may implicate signs of DNA damage and/or impaired DNA synthesis upon irradiation.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.