SEC + UC successfully isolated sEVs from plasma
The sEVs fractions were isolated directly by UC, SEC and SEC + UC methods (Fig. 1), and characterized according to MISEV2018 guideline 16. Transmission electron microscope (TEM) images showed that the sEVs isolated from human plasma by three different isolation pipelines had intact membrane structures and similar morphology (Fig. 2A-C). We applied nanoparticle tracking analysis (NTA) to measure the mean diameters of isolated fractions, and most particles were at 60–100 nm (Fig. 2D-F). Coomassie blue staining showed the total protein level of SEC + UC, UC and SEC fractions visually (Fig. 2G), and verified SEC + UC had the minimum protein content. Western blotting confirmed several sEVs positive markers, including CD9 (60232-1, 1:1000, Proteintech), CD81 (66866-1, 1:1000, Proteintech), and HSP90 (60318-1, 1:1000, Proteintech) (Fig. 2H). Under the premise of the same protein quantity, UC preserved the highest level of CD81, as compared to SEC and SEC + UC. Meanwhile, the CD9 and HSP90 levels of the SEC + UC sEVs fraction were similar to that of UC and much higher than that of SEC. Generally, our SEC + UC isolation pipeline kept all three sEV-associated biomarkers, revealing the successful sEVs recovery of SEC + UC isolation assay.
SEC + UC showed comparable recovery of sEVs with higher purity in contrast to a single-step UC or SEC isolation
According to the NTA quantificationresults (Fig. 3A), the SEC + UC procedure isolated 3.03 × 1010 particles from 1 mL plasma, slightly lower than UC (5.87 × 1010/mL plasma), while SEC assay isolated the largest number of particles at 5.97 × 1011/mL plasma. BCA protein quantification was also performed, the
results showed that the sEVs fraction isolated by SEC + UC (6.15 µg/mL plasma) had a significantly lower protein amount, as compared to UC (209.8 µg/mL plasma) and SEC (65.6 µg/mL plasma, Fig. 3B), which suggested a successful removal of lipoproteins and other contaminant particles. In addition, protein abundance per particle was also calculated in Fig. 3C. Considering lipoprotein particles have a higher protein proportion than sEVs, our results showed that SEC (0.11 fg protein/particles) and SEC + UC (0.20 fg protein/particles) isolated particles had better purity of sEVs than UC (3.58 fg protein/particles).
Common putative contaminants, such as Albumin (16475-1, 1:5000, Proteintech) and apolipoprotein A1 (Apo-A1) (66206-1, 1:1000, Proteintech), and one sEVs negative biomarker, Calnexin (10427-2, 1:500, Proteintech) were all detected in the sEVs fractions isolated by UC, SEC, and SEC + UC, along with the plasma input. Albumin is the most abundant protein in human blood plasma, and considered as the most commonly polluted indicator. Apo-A1 is a major component of HDL and CM. Western blotting showed lipoprotein particles were co-isolated with sEVs (Fig. 3D), and one-step SEC and SEC + UC pipelines had fewer contaminants of lipoproteins, as compared to UC. In addition, all three pipelines showed the negative level of Calnexin. In summary, SEC + UC combination assay could acquire a higher yield of sEVs with fewer contaminants, as compared to a single-step UC or SEC.
SEC + UC kept more protein species detectable by MS than a single-step UC or SEC isolation
We continued by analyzing proteins with MS. In proteomic analysis, there are 992 protein species identified in the sEVs fractions isolated by SEC + UC, much higher than the sEVs fractions isolated by UC (453) or SEC (682) alone (Fig. 4A). The heatmap shows all protein species of three isolation groups had a variation expression level (Fig. 4B). All protein species were displayed according to SEC + UC protein level from high to low. Generally, 345 protein species were identified in SEC + UC but not in SEC, and 591 protein species were identified in SEC + UC but not in UC. Those results suggested that SEC + UC kept more protein species detectable by MS than single-step UC or SEC isolation, especially for the proteins of low-abundance.
Then, we investigated what roles these newly identified low-abundance proteins mainly played. Gene Ontology (GO) / Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment of the 345 identified sEVs proteins in SEC + UC but not in SEC was performed to explore their functions (Fig. 4C). The enriched GO terms were analysed into three categories: biological process (BP), cellular component (CC), molecular function (MF). For BP, GO terms include mRNA splicing, via spliceosome (GO:0000398), RNA export from nucleus (GO:0006405), osteoblast differentiation (GO:0001649). For CC, GO terms include extracellular exosome (GO:0070062), membrane (GO:0016020), nucleoplasm (GO:0005654). For MF, GO terms include poly(A) RNA binding (GO:0044822), protein binding (GO:0005515), RNA binding (GO:0003723). The top 10 enriched GO terms were shown in Fig. 4C, left. Additionally, the KEGG pathway analysis indicated 21 significantly enriched pathways, including spliceosome (hsa03040), dopaminergic synapse (hsa04728), platelet activation (hsa04611) (Fig. 4C, right).
We also performed the GO/KEGG enrichment of 591 identified proteins in SEC + UC but not in UC as well. The enriched GO terms were slightly different. For BP, GO terms include cell-cell adhesion (GO:0098609), viral transcription (GO:0019083), nuclear-transcribed mRNA catabolic process, nonsense-mediated decay (GO:0000184). For CC, GO terms include extracellular exosome (GO:0070062), membrane (GO:0016020), focal adhesion (GO:0005925). For MF, GO terms include poly(A) RNA binding (GO:0044822), protein binding (GO:0005515), GTPase activity (GO:0003924). The top 10 enriched GO terms were shown in Fig. 4D, left. Additionally, the KEGG pathway analysis indicated 46 significantly enriched pathways, including platelet activation (hsa04611), regulation of actin cytoskeleton (hsa04810), spliceosome (hsa03040) (Fig. 4D, right).
We also compared our results with the current public EV datasets, Vesiclepedia and Exocarta (Fig. 4E-G). There were 584 species of previously identified sEV-associated proteins (the intersecting area in Venn diagrams) identified by SEC + UC, much more than those of UC (180) and SEC (366) alone. Additionally, there were also more proteins which not reported before in the SEC + UC group (360). Thus, we demonstrated that the SEC + UC group maintained more protein species to be detectable in MS, including the most identified proteins and some unidentified low-abundant proteins, as compared to UC and SEC.
SEC + UC kept EV-associated proteins more detectable by MS and discarded more contaminants than a single-step UC or SEC isolation
Then we ranked all detected proteins by their abundance and marked some common EV biomarkers in all three groups. Detailed analysis of MS results suggested that sEV-associated proteins showed higher rank in SEC + UC group than both UC group and SEC group evidently (Fig. 5A, C). Generally, CD9 and CD81 ranked highest in all three groups, suggested them as robust sEVs biomarkers. Flotillim-1 (Flot1) is considered as a highly specific exosome biomarker, which does not appear in other EVs. Here our results showed that Flot1 was only detected by MS in SEC + UC group, suggested a better detection ability for exosome components of our SEC + UC pipeline.
Lipoproteins (Apo-E, Apo-L1, Apo-A1, Apo-A2) and IgG antibody fragments (IgG L chain and IgG H chain) are the most common contaminations in plasma sEVs proteomic analysis. MS results analyzed that all contaminants in UC showed a higher abundance rank than SEC and SEC + UC, except Apo-E, which suggests UC can remove Apo-E efficiently. The most contaminations in SEC were Apo-E and Apo-L1, while SEC showed a lower abundance rank in Apo-A1, Apo-A2 and IgG H chain. In terms of SEC + UC, they showed a similiar protein rank among Apo-E, Apo-L1, and Apo-A1. As compared to SEC, the combination method showed a better removal effect on Apo-E and two IgG fractions (Fig. 5B, D). Additionally, isolated sEVs from all three pipelines did not contain Argonaute1, Argonaute2, GM130, PMP70, or Tamm-Horsfall protein, suggested that they were not sEVs associated, as previous reported 17. Moreover, MS analysis also revealed that SEC + UC kept more histones, such as HMGN2, HIST1H1E, H1FX, H1F0, RBBP4, which might be derived from EVs released from cell death (eg. apoptosis, necrosis, NETosis).