Adequate enrichment of extracellular vesicles in laboratory medicine

Extracellular vesicles (EVs) and their clinical applications have been recognized as one of the most fast‐growing and promising fields. Numerous groups including us have expressed their concerns on the idea of isolating “pure” EVs for different purposes, as there are no specific markers to date that can examine the exact EV purity. Typically, major instrument need and EV yielding loss are primary compensations of getting EVs with higher purity. As such, especially in biomarker‐based clinical tests, ceaseless efforts for obtaining more pure EV are worth of discussion. We here propose alternative philosophy to tone down the importance of purity, but to emphasize the significance in clinical laboratory medicine to measure EVs obtained by well‐controlled isolation methods. As a result, EVs are enriched to the adequate degree to meet certain clinical applications such as disease diagnosis and/or prognosis. In this review, we will justify the adequate enrichment rationale with advances of EV definitions, markers and contaminants, EV isolation methods, as well as the conclusions acquired from clinical research using less pure but ease‐to‐operate EV isolation methods.


| INTRODUCTION
As a vesicle compartment with the lipid bilayer structure containing various bioactive molecules, extracellular vesicles (EVs) have become one of the most attractive nano-scaled particles in the clinical fields including diagnostics, prognosis and therapeutics. [1][2][3] EVs have been found in almost all kinds of human biofluids. [4][5][6] It has been widely agreed that to apply EV measurement to clinical utility, fast and reproducible methods for isolating EVs from clinical samples are essential. 7 EVs is a collective term that consists of rapidly growing numbers of subtypes and/or analogues (see Section 2), and there are no consensus criteria to define the purity of EVs (see Section 3).
Clinical application of EVs is an interdisciplinary task. Factors include but should not be limited to the efficacy of biomarkers; the cost of labor and economy to perform the test(s), simplicity of procedure and the feasibility of performing the test(s) in clinical settings should all be taken into consideration. 8 Thus, rather than seeking for the ultimate goal of recovering EVs with the highest purity, researchers should also pay more attention to such essential factors to accelerate the clinical translation of EV-relevant findings.
Here, we propose the idea of pursuing the significance in clinical laboratory medicine rather than the restricted purity if the adequate enrichment of EVs can be justified. Here, the adequate enrichment of EVs refers to the state that EVs obtained using a certain standardized method can be reliably applied to solve certain clinical questions in the field of laboratory medicine. If adequate enrichment is justified, the co-isolation contaminants adherent to such EV isolation method(s) should not be treated as confounders but part of the analysis. To support this opinion, we will provide an overview on the complicated terms of EVs and the research advances of biomarker explorations to distinguish EVs and co-isolation contaminants. Finally, we emphasize the importance of adequate enrichment by reviewing and discussing EV-relevant clinical conclusions in the field of laboratory medicine obtained via EVs isolated with different methods.
Largely, the above-mentioned EV subtypes are of difference in biogenesis, physicochemical properties and biological activities (Table 1). In general, exosomes and MVs are formed in different mechanisms. Exosomes originate restrictively from multivesicular endosomes (MVEs) through the inward budding of endosomal membrane. 4,11,21 In contrast, MVs, also called ectosomes, are formed via outward budding of the plasma membrane. 22,23 The other EV subtypes usually have larger sizes with diameters greater than 500 nm (Table 1). Among them, APBs are formed in the late stage of apoptosis, and primarily removed by phagocytic cells. 24 Oncosomes refer to EVs originated from malignant tumor cells that are shed from amoeboid cancer cells through membrane blebs. 12,16,17 Exophers are found in neurons and muscles of Caenorhabditis elegans 18 as well as murine cardiomyocytes, 19 while they can even secret whole organelles. 18 Cells can leave migrasomes opposite to the migration direction during the breaking of retraction fibers. 20 Thus, EVs is an umbrella term for a variety of particles with high heterogeneity; however, it should be noted that EV subtypes often overlap as they may share similar sizes, densities or cargoes (Table 1.). 25 As diversified nanoparticles with the lipid bilayer structure can be found in human biofluids, EVs is considered as a proper term for the classification of these particles, while extracellular particles (EPs) is deemed a proper term when EV purity is hard to be examined per MISEV2018 recommendations. 9 Indeed, such EV T A B L E 1 Biogenesis, physicochemical properties and biological activities of typical EV subtypes.

AND/OR CONTAMINANTS IN BIOFLUIDS
Numerous literature have described protocols for purifying/isolating/separating/concentrating EVs. [31][32][33][34][35][36] Furthermore, there are multiple approaches that are recommended to evaluate EV purity. Webber et al. proposed a simple method by comparing the ratio of EV particles to protein abundance to assess EV purity. If the ratios are >3 � 10 10 particles/μg, the isolates are deemed with high purity, while ratios of 2 � 10 9 -2 � 10 10 and <1.5 � 10 9 particles/μg correspond to low purity and impure, respectively. 37 However, the widely used nanoparticle analytic approaches for particle size measurement cannot discriminate EVs from nonvesicular materials (e.g., membrane fragments, lipoproteins or nucleic acid -protein complex), 31 which potentially leads to overestimation on the EV concentration.
Most researchers also use protein markers to assess the purity of EV preparation prior to downstream analyses, frequently by measuring the concentration of a specific EV surface-marker antigen (e.g., by ELISA assay or Western blot/FACS/array). [38][39][40] However, as emphasized by ISEV, there are still no universal protein markers to confirm the present of EVs or co-isolating contaminates. Instead, in MISEV2018, a comprehensive list of proteins has been recommended to demonstrate the presence of the lipid-bilayer (Category 1), the cytosolic contents (Category 2), the co-isolated constituents of non-EV structures (Category 3), specific hallmarks for EV subtypes (Category 4) and protein partner of receptors on the EV surface (Category 5), respectively. 9 Among them, proteins in Category 1-3 are strongly recommended by ISEV for bulk EV preparations to illustrate the presence of EV-like vesicles and the removal of contaminants, in other words, the purity of an EV preparation.
In contrast, major contaminants of EV isolations tend to be sample type-specific (Table 2). Extremely high abundant proteins are known as co-isolating contaminants in blood EVs, such as albumin 62 and fibrinogen (plasma-derived samples). 60 In addition, immunoglobulin M (IgM), 35,63 lipoproteins 42,58-60 as well as chylomicrons 42 mimic EVs in particle size, 59 density 64 and even some of the chemical properties. It has been estimated that the concentration of lipoprotein particles is 20-100 folds higher than EVs in plasma. 42 Uromodulin (also called Tamm-Horsfall protein, THP) is the predominant contaminant of urine EVs. 38,39,61,65 THP could entrap EVs by forming polymers and co-precipitate with EVs during low-speed (e.g. 17,000 �g) centrifugation, leading to reduced recovery of EVs. 66 To maintain its high viscosity, saliva contains abundant mucin glycoproteins and salivary enzymes such as α-amylase, that are primary contaminants of saliva EVs. 44,67 Fat globules and casein micelles are known as major contaminants in breast milk EVs. 47,48 Similar to blood EV preparations, albumin is one the primary co-isolating contaminants in EVs acquired from amniotic fluid, 50 CSF 52,54 and synovial fluid. 55,56 Interestingly, some EV "markers" could also be found in non-EV fractions with high concentrations. By isolating EVs from blood plasma using size exclusion chromatography (SEC) columns prepared with different types of resins (Sepharose 2B, CL-4B and Sephacryl S400), Baranyai et al. has found that the majority of CD63 and TSG101 are distributed in the protein-rich fractions. 70 This suggests that these EV markers are actually not primarily exist or encapsulated in EVs, and using their abundance to demonstrate the EV recovery may be interfered by their non-EV-bound portion.
Hence, with current knowledge, it still tends not to be feasible to use certain EV/contaminant markers to quantify or evaluate the EV purity. Especially, if commonly agreed EV markers (such as CD63) are even coisolating contaminants, it will fundamentally challenge a variety of study conclusions.

| EV ISOLATION METHODS
Indeed, numerous EV isolation methods are available, and their advantages and disadvantages have been intensively reviewed by other groups. 71,72 As such, in this review, we will not expand such discussion too much, while providing summarized information in Table 3.
In general, ultracentrifugation (UC) and densitygradient ultracentrifugation (DGUC) can acquire EVs with high purity. As a sacrifice, they require major instruments and long operating duration. In addition, albumin is the primary protein contaminant for UC, 62 while lipoproteins that have similar densities to EVs, such as high-density lipoproteins (HDL), cannot be sufficiently removed in DGUC. 6 Other methods include ZHANG ET AL.
-3 of 11 T A B L E 2 Protein markers reported to demonstrate the presence of EVs and contaminants in human biofluids.  42,74 ; SEC could also be combined with polymer-based precipitation to separate EV subtypes. 75 Hence, each of these approaches has their potentials to meet specific clinical demands, while novel and/or optimized EV isolation techniques are more than welcome to move the field forward. 7

ENRICHMENT OF EVS TOWARD LABORATORY MEDICINE
Technical innovations had greatly facilitated EV isolation; but, few of the above-mentioned methods were specifically designed to enrich EVs from biofluids for laboratory tests under regular conditions of clinical settings . 1 We propose that to choose EV separation methods in laboratory medicine, the importance of EV purity should be toned down, while more attention should be paid to their practical feasibility in clinical settings. If EVs are enriched to the adequate degree to show significantly diagnostic and/or prognostic powers, it should be deemed adequate enrichment of EVs toward laboratory medicine.
Indeed, adequate enrichment of EVs has successful examples that are ready for clinical use. Chen et al. used an NMPA-approved medical device (Supbio Co., Ltd., Guangzhou, China) to enrich EVs from human serum; T A B L E 4 Different EV isolation methods in clinically relevant biomarker researches. with this UF-based device, they found that as compared with direct detection of CA125 in serum, CA125 in EV isolates (or EPs) had significantly higher diagnostic power in ovarian cancer. 89 It is known that serum EVs isolated by UF suffers from severe lipoprotein contamination; even this, the simple and fast UF method makes it feasible to verify a very old cancer biomarker in the context of EVs or EPs. In addition, as an FDA-approved laboratory developed test, ExoDx Prostate (IntelliScore) (EPI) test has been used to assess high-grade prostate cancer (HGPC) to inform whether to proceed with prostate biopsy. 1 EPI uses UF to isolate EVs from urine, followed by PCR analysis to probe prostate cancer biomarkers. 88 It is known that urine THP cannot be efficiently removed by UF; however, the clinical significance of EPI has been widely proven. This suggests that the isolation of EVs with UF is an adequate enrichment for the EPI tests.

Biofluid
In addition, numerous EV-biomarker studies are of significance to justify the importance of EV measurement for disease diagnosis/prognosis (Table 4). In general, UCbased methods were the most used ones in these clinical researches 56,[90][91][92][93][94] ; especially, a number of diagnostic biomarkers of cancers 92 were found with EVs with "relatively high purity". In addition, other "high purity" methods like affinity capture 95 or the combination of SEC and UC 56 were both employed by clinical researchers. Unfortunately, it usually takes at least 3-6 h to isolate EVs, and ultracentrifuges are essential for these operations. In this regard, Torres Crigna et al performed an inter-laboratory comparison on UC-based EV isolations, while they found that particle yields differed significantly between groups even the standardized procedure was used. 96 These studies suggest that the quality control of the EV isolation with UC methods is of concerns in laboratory medicine. As a result, some of the "relatively low purity" methods like ultrafiltration (UF) 1,88 and polymer-based precipitation 95 were employed by biomarker research that showed their significance in laboratory medicine. Although these tests have not been approved for clinical uses, they are potentially enriching EVs adequately regarding certain diseases.

| CONCLUSION
Increasing evidence has supported the opinion that the clinical significance of EV-relevant biomarkers is not solely determined by how pure the EVs are in the discovery stage. Adequate enrichment of EVs should be a priority over ceaseless purity improvement when the beneficial power of disease diagnosis/prognosis has been validated. In addition, as an interdisciplinary field, pathologists should also pay more attention to practical issues in laboratory medicine, for example, the capacity to handle multiple samples parallelly, the cost in time, labor and economy when choosing the EV isolation method. Such properties will propel the feasibility of EV biomarker validations in big cohorts, the evidence level of these biomarkers, and their clinical translations.

CONFLICT OF INTEREST STATEMENT
No potential competing interest was reported by the authors.