Comparison of Extracellular Vesicle Isolation Methods for miRNA Sequencing

Abstract

MicroRNAs (miRNAs) encapsulated in extracellular vesicles (EVs) are potential diagnostic and prognostic biomarkers. However, discrepancies in miRNA patterns and their validation are still frequent due to differences in sample origin, EV isolation, and miRNA sequencing methods. The aim of the present study is to find a reliable EV isolation method for miRNA sequencing, adequate for clinical application. To this aim, two comparative studies were performed in parallel with the same human plasma sample: (i) isolation and characterization of EVs obtained using three procedures: size exclusion chromatography (SEC), iodixanol gradient (GRAD), and its combination (SEC+GRAD) and (ii) evaluation of the yield of miRNA sequences obtained using NextSeq 500 (Illumina) and three miRNA library preparation protocols: NEBNext, NEXTFlex, and SMARTer smRNA-seq. The conclusion of comparison (i) is that recovery of the largest amount of EVs and reproducibility were attained with SEC, but GRAD and SEC+GRAD yielded purer EV preparations. The conclusion of (ii) is that the NEBNext library showed the highest reproducibility in the number of miRNAs recovered and the highest diversity of miRNAs. These results render the combination of GRAD EV isolation and NEBNext library preparation for miRNA retrieval as adequate for clinical applications using plasma samples.

Keywords: diagnostic biomarker; extracellular vesicle isolation; miRNA sequencing.

Figure 1

Diagram of EV isolation methods used for the comparison. The box on the right explains the characterization of the products obtained at different steps of the three EV separation protocols. Details are given in Materials and Methods.

Figure 2

Composition and structure of EVs isolated with three different methods. (A) Assessment of CD9, total cholesterol, and total protein in SEC, GRAD, and SEC+GRAD fractions. Codes for the quantifications are depicted in the upper-left box. CD9 geometrical mean fluorescence intensity was measured as geometrical mean was plotted on the left axis. Total cholesterol and total proteins were measured as mg/dL and plotted on the right axis. Total protein was represented as the mean of the three replicas. (i) SEC means size exclusion chromatography, (ii) GRAD means iodixanol gradients, and (iii) SEC+GRAD means size exclusion chromatography followed by iodixanol gradients (see Section 4). (B) Cryo-TEM images of EV extractions using SEC, GRAD, and SEC+GRAD. Lower panels (Zoomed Cryo-TEM) are images zoomed in from the corresponding ones shown above.

Figure 3

MiRNA profiles vary depending on the miRNA library preparation protocol. NEB refers to NEB Multiplex Small RNA Library Prep Set for Illumina, NEXT refers to NEXTFlex Small RNA-Seq Kit v3 and SMARTer refers to SMARTer smRNA-Seq Kit (see Section 4). (A) Principal component analysis (PCA) showing that miRNA profiles are separated based on the library preparation protocol. Each protocol is represented by a different color and symbol. The two replicas are depicted in the plot (Replica 1 and Replica 2). The third point is the centroid of the two replicas of each library protocol. (B) Number of different miRNAs identified in each replica of each protocol. (C) Venn diagram showing the relationship between the miRNA profiles detected in two replicas of each library preparation protocol. Number and percentage of miRNAs detected in one or two replicas are indicated. (D) Venn diagram showing the relationship between miRNA profiles detected with different library preparation protocols. Number and percentage of miRNAs detected by one, two, or three miRNA library preparation methods are indicated. Table showing the number of different and unique miRNAs obtained by the three protocols. A test of proportions comparing the three protocols was performed. p values were represented as n.s when no significance was observed and *** when p values were <0.001. MiRNA counts with the corresponding accession numbers are listed in Table S2.

Figure 4

miRNA profiles of circulating EVs vary depending on the EV isolation method. (A) Principal component analysis (PCA) showing that miRNA profiles are separated according to the EV isolation method. Each method was assessed in three independent experiments (R1, R2, and R3). The fourth point is the centroid of the three replicas of each EV isolation method. (B) Number of different miRNAs detected in replicate experiments (R1, R2, and R3) of each EV isolation method. (C) Venn diagram showing the relationship between miRNA profiles of circulating EVs detected in replicate experiments of each EV isolation method. Number and percentage of miRNAs detected in one, two, or three replicas are indicated. (D) Venn diagram showing the relationship between miRNA profiles of circulating EVs obtained with different EV isolation methods. Number and percentage of miRNAs detected by one, two, or three EV isolation methods are indicated. Table showing the number of different and unique miRNAs obtained by the three protocols. A test of proportions comparing the three protocols was performed. p values were represented as n.s when no significance was observed, * for p values ≤ 0.05, and *** for p values < 0.001. Counts of miRNAs with the corresponding accession numbers are listed in Table S4. SEC, size exclusion chromatography; GRAD, iodixanol gradient; SEC+GRAD, combination of size exclusion chromatography and iodixanol gradient.