Genome-wide comparison of two RNA-stabilizing reagents for transcriptional profiling of peripheral blood.

Peripheral whole blood is relatively easily obtained for monitoring gene expression for biomarker discovery using transcriptomic platforms such as genome-wide microarrays. However, whole blood provides challenges caused by sensitivity for ex vivo incubation and overrepresentation of globin mRNAs. We compared the performance of 2 commercial whole blood preservation methods, TEMPUS (Applied Biosystems, Foster City, CA) and PAXgene (PreAnalytiX, Qiagen BD, Valencia, CA), using 2 RNA amplification protocols and high-density microarrays. Performance of commercial globin mRNA reduction protocol also was studied. Human peripheral blood samples collected with TEMPUS and PAXgene Blood RNA tubes were amplified with the RiboAmp OA 1 Round RNA Amplification Kit (Arcturus; Applied Biosystems) and the Affymetrix (Santa Clara, CA) small sample protocol. Affymetrix globin reduction protocol was applied for total RNA samples. Samples amplified with RiboAmp were hybridized on Illumina Sentrix HumanRef-8 Expression BeadChips (Illumina Inc, San Diego, CA) and subjected to statistical analyses. RiboAmp mRNA amplification did not notably amplify globin mRNA that is overrepresented in RNA isolated by both TEMPUS and PAXgene preservation. Enzymatic depletion of globin transcript reduced the quality of total RNA and is thus not recommendable. Microarray analysis showed acceptable correlation within and between the RNA preservation methods, but altogether 443 transcripts were differentially expressed between RNA samples preserved in TEMPUS and PAXgene tubes. We demonstrated that the 2 tested blood RNA-preservation methods combined with RiboAmp mRNA amplification may be used for microarray experiments without the need for a prior globin RNA reduction. However, because genes involved in immune cell functions and gene regulatory pathways were differentially expressed as a result of the technical bias between the preservation methods, they should not be used in the same analytic setting.