Introduction
Extracellular vesicles (EVs) are fast becoming a preferential platform for liquid biopsy-based biomarker discovery owing to their rich molecular cargo contents. EVs are released from cells of all tissue types during normal biological processes and have high potential as a source of low abundance biomarkers for a wide array of pathophysiologies [
1,
2]. Given that EVs carry internal cargo, analytical platforms like mass spectrometry are crucial for elucidating their biochemical content. However, studies leveraging techniques such as metabolomics and lipidomics to analyze EV content have so far remained relatively limited.
We have previously examined the role of plasma derived EVs as biomarkers for radiation injury [
3,
4]. Whole-body exposure to acute, large doses (> 2 Gy) of ionizing radiation (IR) can potentially be lethal if not diagnosed and treated expeditiously [
5,
6]. Damage to the vascular endothelium may occur as early as 24 h post-exposure, affecting most commonly the gut and bone marrow [
7,
8]. Cumulative IR exposure for the treatment of various cancers also puts patients at risk of normal tissue toxicity [
9‐
11]. Late effects of IR injury may appear several months delayed, potentially causing life-threatening injuries to the brain, heart, lung, and kidneys. When considering a non-invasive approach for the identification of IR injury biomarkers, we and others have studied plasma, serum, urine, and saliva [
12‐
14]. However, these matrices may contain significant background, obscuring biomarkers of biological importance.
Urine, as a noninvasive biological matrix, contains a plethora of molecular information indicating rapid changes that occur in various physiological conditions. Urine has been studied as a source of biomarkers related to IR exposure [
15]. Urinary EVs have also shown promise as biomarkers for kidney disease, urological cancers, and even neurologic disease [
16‐
18]. However, many reported protocols for the isolation of EVs from urine necessitate the use of large volumes of sample [
19,
20]. To our knowledge, no groups have studied the utility of urinary EVs as a source for biomarkers of radiation damage. The goal of this study was to compare, optimize, and validate a method of EV isolation from small volumes of urine, with an emphasis on downstream LC–MS/MS-based metabolomic and lipidomic analyses. We compared three methods of EV isolation: ultracentrifugation (currently one of the most widely used methods with relatively low throughput, generally requiring large volumes (> 10 mL) of urine); a magnetic bead-based method (MBB); and a size exclusion chromatography (SEC)-based method. Our benchmarks of success of EV isolation included yield and purity for a given volume, compatibility with LC–MS analyses, and scalability to support large batches of samples. Following extensive characterization to confirm EV enrichment and small molecule profiling, our results showed that SEC-based urinary EV isolation gave optimal yields with 0.5 mL of rat urine and resulted in high quality LC–MS data. The method was validated using a pilot study and later tested using a larger cohort of samples utilizing the WAG/RijCmcr rat model of leg-out partial body irradiation (leg-out PBI). This sophisticated model is the only one available to display multiple relevant organ sequelae (gastrointestinal, bone marrow, pulmonary, heart, brain, and kidney radiation injuries) observed after a large single radiation exposure. Finally, we demonstrated the clinical utility of our EV analytical approach in a pilot study of human urine samples from patients receiving thoracic radiation therapy (RT).
To our knowledge, this study is the first comparison of EV isolation methods from small volumes of urine, the first report of a comprehensive method for mass spectrometric analysis of the small molecule and lipid content from EVs isolated from small volumes of urine, and the first report evaluating the efficacy of urinary EVs as biomarkers for radiation injury. The availability of reproducible methods of EV isolation from low volumes of urine that are compatible with mass spectrometry methodologies is an unmet need for enabling biomarker discovery, and for studying disease onset, disease progression and patient response to RT. The reported SEC-based small volume urinary EV isolation method can be extended to other types of biomolecular analyses; to this end, we have developed and provided a bulleted optimized standard operating procedure (SOP) to enable easy implementation in other laboratories.
Materials and methods
Animal care and irradiation protocols
All animal protocols were approved by the Institutional Animal Care and Use Committees (IACUC) at the Medical College of Wisconsin, Milwaukee. WAG/RijCmcr female rats were irradiated at 11–12 weeks of age (~ 155 g). Two groups of rats were randomized for this study: I) No irradiation (n = 5); II) 13 Gy leg-out partial body irradiation (leg-out PBI) (n = 5). A subset of 4 urine samples from these rats were used for optimizing EV isolation. To test the effect of radiation on EV cargo composition, a separate cohort of rats was randomized into (1) No irradiation, vehicle (n = 8), or (2) 13 Gy leg–out PBI (n = 10). The protocol used for leg-out PBI was the same as reported in previous studies [
21,
22] and details can be found in the Additional file
2. Urine was collected 24 h, 14, 30, or 90 days after irradiation. Urine was centrifuged at 13,000 rpm at 4 °C for 10 min and the supernatant stored frozen at − 80 °C until analyses.
Human urine collection
The biospecimen collection study was approved by the Georgetown University-MedStar Health Institutional Review Board (IRB 2013-0049) and eligible patients underwent signed informed consent. Eligibility included patients receiving thoracic radiation therapy and who were able to provide consent for participation. Exclusion criteria included less than 18 years of age, pregnancy, and acute illness. Enrolled patients included 12 patients with non-small cell lung cancer and 1 patient with thymoma who were treated with 35–66 Gy. A clean catch urine sample was collected prior to radiation therapy, following the last fraction of treatment, and approximately 6 weeks after treatment. For the current report, we used samples from 5 patients pre- and immediately post-RT.
Urinary EV isolation
We have submitted all relevant data of our experiments to the EV-TRACK knowledgebase (EV-TRACK ID: EV210076) [
23]. EVs were isolated from urine using 3 independent methods (I) ultracentrifugation (UC) with filtration, (II) size exclusion chromatography (SEC) proceeded by filtration/concentration steps, and (III) a proprietary magnetic bead-based isolation method [
24]. We compared two different initial volumes of urine (0.5 mL or 1 mL). Detailed protocols and checklists can be found in Additional files
1 and
2.
Nanoparticle tracking analysis (NTA)
NTA was performed using a NanoSight NS300 (Malvern Panalytical) equipped with a high sensitivity sCMOS camera, 532 nm laser, and automatic syringe pump. Detailed methods can be found in Additional file
2.
EV immunoblot array
To examine expression of accepted EV specific protein markers in isolated urinary EV samples, we performed the Exo-check Antibody Array (System Biosciences, Palo Alto CA, #EXORAY210A). 30 µg of EV protein was aliquoted and immunoblots were processed and developed according to the manufacturer’s protocol.
Cryogenic electron microscopy
EV samples resuspended in 1 × PBS were sent to the Molecular Electron Microscopy Core at the University of Virginia for analysis. Detailed methods can be found in the Additional file
2.
Mass spectrometry solvents and reagents
All solvents were LC–MS grade. Catalog numbers for all solvents, reagents and standards can be found in the included SOPs in Additional files.
Detailed sample preparation methods and MS data acquisition and processing can be found in Additional file
2.
Methods were developed for quantitation using a QTRAP 5500 LC–MS/MS System (Sciex). Detailed methods, protocols and checklists can be found in Additional files and Additional file
2.
Statistical analyses and visualization
Prior to statistical analyses, MS data were normalized as described above for each dataset. Binary comparisons were done using paired or unpaired t-tests as appropriate, and fold changes were calculated to identify differences in metabolite levels after radiation. All analyses were performed in R (v 4.0.3). Figures were created using R, GraphPad Prism (v. 9.0) and BioRender (
www.BioRender.com).
Discussion
Whole body exposure to high doses of IR can potentially be lethal if radiation injury is not diagnosed and treated expeditiously. Radiation-induced organ damage can manifest within days to years after irradiation. When considering a non-invasive approach for the identification of biomarkers of exposure to IR and of radiation-induced organ damage, we and others have studied molecules in plasma, serum, saliva, and urine. However, these matrices are complex with high abundance molecules like albumins and globulins that can obscure the detection of potential biomarkers of biological importance that have relatively lower abundance.
Extracellular vesicles are fast becoming a platform for biomarker discovery in radiation research as well as in other pathologies. Few studies have investigated the use of a metabolomics approach to analyze EVs derived from urine in the context of IR exposure. Furthermore, the dominant protocols for EV isolation from urine require a large (up to 30 mL) amount of starting volume, which may not be feasible for many studies and clinical translation. The aim of this study was to optimize EV isolation from rat urine and assess radiation-induced alterations in urinary EV metabolic content. Given that EVs have shown tremendous potential as a source for biomarkers for an array of diseases and disorders, our overall goal was to develop a robust and reproducible method of EV isolation from small volumes of urine that is compatible with downstream molecular characterization of EV cargo. Importantly, each of our isolation methods yielded a different number of total detected features. Part of this variance may be explained by the inherent differences in non-EV contaminants co-isolated with each method [
27‐
29]. This is an important consideration particularly when isolating EVs from other biological matrices which may have high protein concentrations, such as plasma or tissue. The ability to successfully isolate EVs from small volumes of urine can be applied beyond metabolomics to miRNA isolation, proteomics, and functional in vitro assays. The method presented is consistent and reliable for isolating, quantifying, and characterizing urinary EVs for broad research and clinical purposes.
We also demonstrate the utility of urinary EVs as an effective source of biomarkers for detecting IR exposure. Previous reports have studied whole urine, whole plasma, and plasma EVs as sources for biomarkers of IR exposure and IR damage [
3,
11,
30‐
33]. Dyslipidemia has previously been observed after radiation exposure [
3,
12,
34‐
37]. Given the importance of lipid composition to EV function, this remains an active area of research. We have previously found that post-IR exposure, plasma EVs are enriched in lipid species indicative of an inflammatory response, namely TAGs [
4]. In that previous study, we found that upregulation of TAGs dissipated by 14 days post-irradiation, a finding similar to our results reported here. Xu et. al previously demonstrated that TAGs are upregulated in the plasma at late time points (15–20 weeks) post irradiation [
13]. TAGs were the bulk of lipids dysregulated 90 days post-irradiation, all of which were upregulated. The functional consequences of these changes remain an active area of research for our group.
These methods are also useful for studying human clinical samples. Though we have only performed a pilot study on 5 patients receiving RT as a part of their thoracic cancer treatment regimens, we were able to detect significant metabolite profiles of urinary EVs from extremely small volumes of urine. Importantly, these differences may also be biologically relevant in the context of IR exposure. The same data can also be extrapolated for correlation of EV-metabolites with tumor response to radiation therapy, once follow-up data become available. As reported here, we found significantly altered levels of key metabolites driving folate and nucleotide metabolism. Previous studies have demonstrated functional consequences of folate processing enzymes and altered folate pools imparted by radiation stress [
38]. Activation of DNA damage and repair mechanisms, as well as altered nucleotide metabolism are canonical biological events post-IR exposure, events that may be linked to radio responsiveness depending on tissue type and disease setting [
39‐
41]. Considering previous reports, with our findings, urinary EVs present themselves as a potentially important biological matrix for monitoring metabolite changes, and ultimately patient response, to RT.
Though these results are of interest, our study contains potential limitations and considerations which should be discussed. First, the WAG/RijCmcr rat model is an extreme measure of radiation exposure. Though it may not directly mirror clinical irradiation protocols, as experienced by patients undergoing radiation therapy, we have included pilot data to demonstrate that our findings are applicable to human urine samples from patients receiving clinically relevant radiation doses. That said, the analysis of human urinary EVs is limited to a small number of patients in this report. This means that the biological relevance of our findings still requires in depth validation, and that the biological interpretations of these findings may be limited. The small sample size means we do not have analysis of changes over time, nor could we identify commonly dysregulated metabolites between rat and human urinary EVs. However, these limitations will be addressed in future studies as we continue to develop methods for monitoring how patients respond to RT. One final consideration of this study is that all human urine samples came from patients with either non-small cell lung cancer or thymoma. Given that we did not analyze urine from human healthy volunteers, it is possible that the differences we observe could be confounded by pre-existing disease.
In conclusion, the urinary EV metabolome and lipidome may be a sensitive and specific early indicator of radiation injury and a platform for monitoring patient response to RT. Furthermore, the approaches developed and validated in this study can be easily applied in diverse areas of biomedical research that seek to leverage molecular analyses of urinary EV profiles for gaining insights into disease onset and progression. Finally, of the methods tested herein, SEC was found to be the preferred method for isolating EVs from small volumes of urine for broad-based mass spectrometric analysis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.