Transcriptome sequencing of RNA isolated from small volumes of blood stabilized in Tempus solution: a technical assessment of different extraction methods and DNase treatment

Abstract

Transcriptome profiling of human whole blood is used to discover biomarkers of diseases and to assess phenotypic traits. RNA sequencing technologies offer many advantages for transcriptomic profiling over other technologies, including the ability to detect novel transcripts, a wide dynamic range of transcript detection, high specificity and sensitivity and the ability to detect low-abundance transcripts. Recently, finger-stick blood collection systems have allowed a less invasive and quicker collection of peripheral blood that does not necessarily require medical infrastructures. Such non-invasive sampling of small volumes of blood offers practical advantages, allowing large-scale projects. The quality of gene expression data is strictly dependent on the steps used for the sample collection, extraction, preparation and sequencing. Here we have: i. compared the manual and automated RNA extraction of small volumes of blood using the Tempus Spin RNA isolation kit (ThermoFisher Scientific, USA) and the MagMAX™ for Stabilized Blood RNA Isolation Kit (ThermoFisher Scientific, USA), respectively; and ii. assessed the effect of TURBO DNA Free treatment on the transcriptomic data of RNA isolated from small volumes of blood. RNA Libraries were prepped using the QuantSeq 3' FWD mRNA-Seq Library Prep Kit (Lexogen, Austria). Library QC was performed on the LabChip GXII. Libraries were quantified using KAPA Library quantification kit by qPCR on the LightCycler 480 II (Roche Diagnostics, Basel, Switzerland). Libraries were pooled on the Hamilton MicroLab Star (Hamilton, Reno, NV, USA ) and sequenced on the Illumina NextSeq 500 system. The QC of the sequencing data was performed as recommended by Illumina. Reads were mapped to the human genome GRCh38.p13 (Genome Reference Consortium Human Build 38), INSDC Assembly GCA_000001405.28, Dec 2013) using STAR_2.6.1d aligner, and featureCounts v2.0.0 was used to generate the raw counts. We used DESeq2 (v1.32.0) to normalize read counts with standard settings. Normalized data was transformed using variance-stabilizing transform (VST) and removed of the batch effect using limma::removeBatchEffect from Lima package (v3.48.2). Heatmaps, correlation matrices and PCA plots were generated as relevant. Transcriptomic profiles were overall consistent, however the samples isolated manually displayed a higher variability in the transcriptomic data as compared to the other samples. The TURBO DNA Free treatment affected the RNA samples negatively, decreasing the RNA yield and reducing the quality and reproducibility of the transcriptomic data. We conclude that automated extraction systems should be preferred over manual extraction systems for data consistency, and that the TURBO DNA Free treatment should be avoided when working on RNA samples isolated manually from small volumes of blood.