High-Throughput and Automated Acoustic Trapping of Extracellular Vesicles to Identify microRNAs With Diagnostic Potential for Prostate Cancer.

Anson Ku,Thomas Laurell,Karina D Sørensen,Mikael Evander,Hans Lilja,Michael Borre,Yvonne Ceder,Jacob Fredsøe

doi:10.3389/fonc.2021.631021

Abstract

Molecular profiling of extracellular vesicles (EVs) offers novel opportunities for diagnostic applications, but the current major obstacle for clinical translation is the lack of efficient, robust, and reproducible isolation methods. To bridge that gap, we developed a microfluidic, non-contact, and low-input volume compatible acoustic trapping technology for EV isolation that enabled downstream small RNA sequencing. In the current study, we have further automated the acoustic microfluidics-based EV enrichment technique that enables us to serially process 32 clinical samples per run. We utilized the system to enrich EVs from urine collected as the first morning void from 207 men referred to 10-core prostate biopsy performed the same day. Using automated acoustic trapping, we successfully enriched EVs from 199/207 samples (96%). After RNA extraction, size selection, and library preparation, a total of 173/199 samples (87%) provided sufficient materials for next-generation sequencing that generated an average of 2 × 106 reads per sample mapping to the human reference genome. The predominant RNA species identified were fragments of long RNAs such as protein coding and retained introns, whereas small RNAs such as microRNAs (miRNA) accounted for less than 1% of the reads suggesting that partially degraded long RNAs out-competed miRNAs during sequencing. We found that the expression of six miRNAs was significantly different (Padj < 0.05) in EVs isolated from patients found to have high grade prostate cancer [ISUP 2005 Grade Group (GG) 4 or higher] compared to those with GG3 or lower, including those with no evidence of prostate cancer at biopsy. These included miR-23b-3p, miR-27a-3p, and miR-27b-3p showing higher expression in patients with GG4 or high grade prostate cancer, whereas miR-1-3p, miR-10a-5p, and miR-423-3p had lower expression in the GG4 PCa cases. Cross referencing our differentially expressed miRNAs to two large prostate cancer datasets revealed that the putative tumor suppressors miR-1, miR-23b, and miR-27a are consistently deregulated in prostate cancer. Taken together, this is the first time that our automated microfluidic EV enrichment technique has been found to be capable of enriching EVs on a large scale from 900 μl of urine for small RNA sequencing in a robust and disease discriminatory manner.

Highlights

Extracellular vesicles (EVs) are 50–1,000 nm membraneencapsulated particles that are secreted by outward budding or fusion of the multi-vesicular endosome with the plasma membrane
Of the mapped RNA species revealed that the average length of the reads across samples was 21 nt in length (Figure 4A), which was expected as the libraries were size selected for inserts ~10–30 nt long
We found that a total of 80% of the mapped reads were derived from long RNAs such as protein coding, retained intron, processed transcript, lincRNA, antisense, and nonsense-mediated decay

Summary

Introduction

Extracellular vesicles (EVs) are 50–1,000 nm membraneencapsulated particles that are secreted by outward budding or fusion of the multi-vesicular endosome with the plasma membrane. In pathological conditions like cancer, EVs have been shown to facilitate disease progression Due to their reported diseasespecific content and accessibility, EVs have been proposed as a potential new, non-invasive source of biomarkers during routine liquid biopsy [2]. There are a number of commercial and published methods for EV enrichment such as size-exclusion separation (qEV), polymer precipitation (Exoquick), membrane filtration, affinity-based purification, and microfluidics based separation, but many are not suitable for clinical translation due to long incubation time (Exoquick) or cost (immuno-affinity based purification) [3, 4]. The number of automated EV isolation methods remains few (qEV, membrane filtration, iDEP), has yet to be tested on large scale, or is not low sample-volume compatible [3, 5,6,7,8]. We have utilized an optimized pipeline to interrogate miRNA expression from urinary EVs by next-generation sequencing [12]

Methods

Results

Conclusion