Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

A comparison of RNA extraction and sequencing protocols for detection of small RNAs in plasma.

Author(s): Wong RKY, MacMahon M, Woodside JV, Simpson DA

Publication: BMC Genomics, 2019, Vol. 20, Page 446

PubMed ID: 31159762 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   The purpose of this study was to investigate differences between RNA extraction and library preparation methods on NGS profiles of miRNA in plasma. The effect of a plant-based diet on the levels of plant miRNA was also discussed. Fasting EDTA blood was obtained from three individuals before and after a diet change from <2 servings to 8 servings of fruit and vegetables per day that occurred over 4 weeks. Within 2 h of blood draw, plasma was obtained by centrifugation (details not specified) and frozen at -80˚C. RNA was extracted from 600 µL of plasma using the MagnaZol cfRNA Isolation Reagent or from two 200 µL aliquots of plasma using the miRNeasy Serum/Plasma Extraction Kit. miRNA was quantified using the Quubit microRNA Assay kit. Sequencing libraries were prepared from 2 µL of miRNeasy-isolated RNA using CleanTag or from 5 µL of RNA isolated from each kit using NEXTflex and QIAseq. CleanTag and QIAseq libraries were further cleaned using Agencourt AMPure magnetic beads (XP beads for CleantTag and QMN beads for QIASeq) and NEXTflex libraries were purified by PAGE size selection. Libraries were quantified using the Qubit dsDNA HS Assay Kit and sequenced on an Illumina NextSeq 500.

   Summary of Findings:

   For QIAseq libraries, there were very strong correlations between the mean number of reads before and after UMI correction (only included in QIAseq), regardless of extraction method (R²=0.9999 and R²=0.9995), and consequently, all further analysis was done without UMI correction. UMI correction resulted in a smaller decrease in reads in miRNeasy than MagnaZol-isolated specimens which the authors state indicates potentially more miRNA in the miRNeasy extraction.

   Libraries constructed from RNA extracted with MagnaZol had higher mean percentage of reads mapping to miRNA than those extracted using miRNeasy (P<0.05 for QIAseq and P<0.001 for NEXTflex). For MagnaZol-extracted RNA, the percentage of reads mapping to miRNA was higher and the percentage of reads discarded (too long, too short, and adapter dimers) were lower when library construction was with NEXTflex (62.8% and 10.3%, respectively) rather than QIAseq (50.3% and 41.8%, respectively).  Using miRNeasy extracted RNA, the highest percentage of reads mapping to miRNA and lowest percentage of reads discarded were achieved using NEXTflex (18.9% and 11.4%, respectively), followed by CleanTag (17.2% and 31.4%, respectively) and QIAseq (9.5% and 50.6%, respectively). Further investigation of the sequences not mapped to human miRNA mapped to rRNA and Y RNAs and to plant miRNAs. Interestingly, the plant miRNAs did not change in abundance following the dietary switch so the authors attribute their presence to contamination.

   After down-sampling the reads to 550,000, the highest number of miRNAs (most diversity) was found in QIAseq libraries from MagnaZol RNA (471 miRNAs) followed by QIAseq libraries of miRNeasy RNA (451 miRNAs). Fewer miRNAs were detected in NEXTflex libraries (385 miRNAs for miRNeasy RNA and 327 miRNAs for MagnaZol RNA) or the CleanTag miRNeasy library (260 miRNAs). The relative diversity of the libraries was further compared by looking at the average abundance of the top ten miRNAs in the 550,000 reads. Using this metric, the least diverse library was the CleanTag miRNeasy library (524,747 reads) and the most was QIAseq libraries of MagnaZol RNA (391,961 reads, P<0.001). Further down-sampling of the data to 500 reads confirmed the higher diversity of QIAseq libraries of MagnaZol RNA with 428 miRNAs detected versus 328 miRNAs in NEXTflex libraries of MagnaZol RNA, 298 miRNAs in NEXTflex libraries of miRNeasy RNA, 254 miRNAs in QIAseq libraries of miRNeasy RNA and 168 miRNAs in CleanTag miRNeasy libraries. Based on the down-sampling data, the authors determined a minimum of 1 million reads as the minimum sequencing depth since more than half of miRNAs had >5 million reads at this depth and the increase in miRNAs detected with increased depth plateaued at 2 million reads.

   When the same RNA specimen was used for library preparation, 18 miRNAs displayed over-expression (≥2-fold change, P<0.05) and 25 displayed under-expression when library preparation was with NEXTflex rather than QIAseq. Similarly, compared to libraries constructed with miRNeasy RNA, libraries constructed with MagnaZol RNA had over-expression of 10 miRNAs and under-expression of two miRNAs. Hierarchical clustering based on the 100 most differentially detected miRNAs and the 500 most differentially detected isomiRs first clustered based on library preparation and then by RNA extraction method with QIAseq libraries of miRNeasy RNA most consistently clustering specimens from the same individual together. When quantified by real-time PCR, the 10 miRNAs differentially expressed between library methods were significantly strongly correlated with those obtained using QIAseq (R2=0.73 for MagnaZol and R2=0.72 for miRNeasy), modestly correlated with those obtained using NEXTflex and MagnaZol RNA (R2=0.66) and not correlated using NEXTflex and miRNeay RNA.

   Biospecimens

   • Bodily Fluid - Plasma

   Preservative Types

   • Frozen

   Diagnoses:

   • Normal

   Platform:

   Analyte	Technology Platform
   RNA	Next generation sequencing

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Preaquisition	Patient diet	<2 servings of fruit and vegetables per day 8 servings of fruit and vegetables per day
   Analyte Extraction and Purification	Analyte isolation method	MagnaZol cfRNA Isolation Reagent miRNeasy Serum/Plasma extraction kit
   Next generation sequencing Specific	Template modification	CleanTag library NEXTflex library QIAseq library

   View more