Comparison of blood RNA isolation methods from samples stabilized in Tempus tubes and stored at a large human biobank

Abstract

Background: More than 50,000 adult and cord blood samples were collected in Tempus tubes and stored at the Norwegian Institute of Public Health Biobank for future use. In this study, we systematically evaluated and compared five blood-RNA isolation protocols: three blood-RNA isolation protocols optimized for simultaneous isolation of all blood-RNA species (MagMAX RNA Isolation Kit, both manual and semi-automated protocols; and Norgen Preserved Blood RNA kit I); and two protocols optimized for large RNAs only (Tempus Spin RNA, and Tempus 6-port isolation kit). We estimated the following parameters: RNA quality, RNA yield, processing time, cost per sample, and RNA transcript stability of six selected mRNAs and 13 miRNAs using real-time qPCR.

Findings: Whole blood samples from adults (n = 59 tubes) and umbilical cord blood (n = 18 tubes) samples collected in Tempus tubes were analyzed. High-quality blood-RNAs with average RIN-values above seven were extracted using all five RNA isolation protocols. The transcript levels of the six selected genes showed minimal variation between the five protocols. Unexplained differences within the transcript levels of the 13 miRNA were observed; however, the 13 miRNAs had similar expression direction and they were within the same order of magnitude. Some differences in the RNA processing time and cost were noted.

Conclusions: Sufficient amounts of high-quality RNA were obtained using all five protocols, and the Tempus blood RNA system therefore seems not to be dependent on one specific RNA isolation method.

Keywords: Biobank; Cord blood; Epigenetics; Gene expression; MiRNA; MoBa; Noncoding RNA; RNA isolation; Tempus tubes; The Norwegian Mother and Child Cohort Study.

Fig. 1

Comparison of RNA yields for the adult and cord blood samples collected in Tempus tubes. The RNA yield from adult blood samples (n = 12 Tempus tubes per protocol, n = 11 Tempus tubes for the Tempus 6-port protocol) and the RNA yield from cord blood samples (n = 3 Tempus tubes per protocol, except the ABI system, where n = 6 Tempus tubes). *The RNA yields from cord blood samples were significantly higher than for adult blood samples, and **the RNA yields obtained from cord blood samples using the two MagMAX protocols were significantly higher than the RNA yields obtained using the other three protocols (p < 0.05). Each bar represents the average RNA yield and the error bars indicate ± SE

Fig. 2

RNA transcript levels of six target genes from adult and cord blood samples. RNA isolated from adult and cord blood samples using five different protocols and analyzed by qPCR. a Relative transcript levels of six genes from adult blood samples collected in Tempus tubes (n = 11–12 Tempus tubes for each protocol). b Relative transcript levels of six genes from cord blood samples collected in Tempus tubes (n = 3–6 Tempus tubes for each protocol). The Tempus 6-port samples were used as reference samples (calibrators) and all other samples were compared against the reference samples. Each bar represents the average log2-transformed fold change values; fold change = 2^−∆∆Cq. The error bars indicate ± SE and the stippled lines indicate ± twofold

Fig. 3

Non-normalized raw Cq-value and coefficients of variation of 13 miRNAs. RNA isolated from adult and cord blood using three different RNA isolation protocols optimized for simultaneous isolation of all RNA species and analyzed by qPCR. The non-normalized raw Cq-values for adult blood (n = 12 Tempus tubes) and cord blood (n = 3 Tempus tubes) samples collected in the Tempus tubes. a Average non-normalized raw Cq-values for adult blood samples; b the average non-normalized raw Cq-values for cord blood samples. The error bars indicate ± SE. There are no significant differences in the raw Cq-values between the three RNA isolation protocols. The coefficients of variation (CV) of the average raw Cq-values were calculated for adult and cord blood samples for each RNA isolation protocol for the 13 miRNAs. c CVs for adult blood samples (ranging from 4.5 to 18.8 %); d CVs for cord blood samples (ranging from 1.7 to 10.9 %). CV of 10 and 20 % is indicated by stippled lines. Each point represents the average CV of samples for one miRNA from one Tempus tubes from one protocol

Fig. 4

Log2-NRQ values for 13 miRNAs for adult blood samples. The normalized relative quantity values [NRQ = 2^−ΔCq(sample); where the ΔCq (sample) = Cq (miRNA) − Cq (internal control)]) of 13 miRNAs from adult samples collected in Tempus tubes (n = 12 Tempus tubes for adult blood). a Log2-transformed NRQ-value of 13 miRNAs for the three RNA isolation protocols. Each bar represents the log2-transformed NRQ values and the error bar indicates ± SE; b correlation between MagMax semi-automated and manual protocols; c correlation between MagMax semi-automated and Norgen protocols; d correlation between MagMax manual and Norgen protocols. There are significant (p < 0.001) correlations between the NRQ-values of the three protocols. Each point represents the average of log2-NRQ values of three technical replicates from one Tempus tube, i.e., 13 miRNAs × 12 blood samples from 12 Tempus tubes

Fig. 5

Log2-NRQ values for 13 miRNAs for cord blood samples. The normalized relative quantity values [NRQ = 2^−ΔCq(sample); where the ΔCq (sample) = Cq (miRNA) − Cq (internal control)]) of 13 miRNAs from cord samples collected in the Tempus tubes (n = 3 Tempus tubes for adult blood). a Log2-transformed NRQ-value of 13 miRNAs for the three RNA isolation protocols. Each bar represents the log2-transformed NRQ values and the error bar indicates ± SE; b correlation between MagMax semi-automated and manual protocols; c correlation between MagMax semi-automated and Norgen protocols; d correlation between MagMax manual and Norgen protocols. There are significant (p < 0.001) correlations between the NRQ-values of the three protocols. Each point represents the average of log2-NRQ values of three technical replicates from one Tempus tube, i.e., 13 miRNAs × 3 blood samples from 3 Tempus tubes

Fig. 6

Non-normalized raw Cq-value for Tempus spin and 6-port. The non-normalized raw Cq-values for adult blood and cord blood samples collected in the Tempus tubes. a Average non-normalized raw Cq-values for adult blood samples (three replicates from pooled RNA samples per protocol); b average non-normalized raw Cq-values for cord blood samples (three replicates from pooled RNA samples per protocol). The error bars indicate ± SE