Comprehensive profiling of extracellular RNA in HPV-induced cancers using an improved pipeline for small RNA-seq analysis

Abstract

Extracellular RNAs (exRNAs) have attracted great attention due to their essential role in cell-to-cell communication as well as their potential as non-invasive disease biomarkers. However, at present, there is no consensus on the best method to profile exRNA expression, which leads to significant variability across studies. To address this issue, we established an experimental pipeline for comprehensive profiling of small exRNAs isolated from cell culture. By evaluating six RNA extraction protocols, we developed an improved method for robust recovery of vesicle-bound exRNAs. With this method, we performed small RNA sequencing of exosomes (EXOs), microvesicles (MVs) and source cells from 14 cancer cell lines. Compared to cells, EXOs and MVs were similarly enriched in tRNAs and rRNAs, but depleted in snoRNAs. By miRNA profiling analysis, we identified a subset of miRNAs, most noticeably miR-122-5p, that were significantly over-represented in EXOs and MVs across all 14 cell lines. In addition, we also identified a subset of EXO miRNAs associated with cancer type or human papillomavirus (HPV) status, suggesting their potential roles in HPV-induced cancers. In summary, our work has laid a solid foundation for further standardization on exRNA analysis across various cellular systems.

Seqpac: A New Framework for small RNA analysis in R using Sequence-Based Counts

“…In our HeLa cell experiment, the main fragment was 61 nt, which indicates a unique fragment given that we had a maximum read length of 75 nt. Even though the methods used in Tong et al (25) and Xu et al (26, 27) were restricted to a maximum read length of 50 nt, we found traces of this fragment in the pile of fragments with unverifiable length of ≥ 50 nt. It must be emphasized, however, that we tried to validate the 5’ ETS rRF in yet another dataset, Snoek et al (62) (SRA accession: PRJNA413777), but here we failed to detect anything in the 5’ EST region.…”

“…To further illustrate the strengths of seqpac, we reanalyzed a recently published dataset by Tong et al 2020 (25) (SRA access: SRP285629). This dataset contains 42 fastq files originating from 14 human cancer cell lines, where RNA was extracted from cells, as well as exosomes and microvesicles of these cells.…”

“…Tong et al (25) used a feature-based counting strategy. This strategy first aligns sequences to a reference genome, often allowing for multiple mismatches, and discards sequences that fails to align.…”

“…Some of these issues with feature-based counting can be illustrated with the Tong et al dataset (25) using seqpac’s workflow. It must be emphasized, however, that our critic is not specifically aimed against Tong et al, whose work we admire, but rather against the feature-based counting strategies that hundreds of studies have been using.…”

“…Tong et al from 2020 (25) (SRA accession: PRJNA666144; ENA download: https://www.ebi.ac.uk/ena/browser/view/PRJNA666144) were used for exemplifying the strengths of sequence-based counting in detecting severe bias in cancer cell line experiements. PAC generation and reannotation was performed similarly to the Kang et al dataset (MATERIALS AND METHODS 2.1) with a few exceptions.…”