Limits...
Reduction of systematic bias in transcriptome data from human peripheral blood mononuclear cells for transportation and biobanking.

Ohmomo H, Hachiya T, Shiwa Y, Furukawa R, Ono K, Ito S, Ishida Y, Satoh M, Hitomi J, Sobue K, Shimizu A - PLoS ONE (2014)

Bottom Line: RNA-Seq for 25,223 transcripts also suggested that about 40% of transcripts were systematically biased.On the basis of the results of this study, we established a protocol to reduce systematic bias in the expression levels of RNA transcripts isolated from PBMCs.We believe that these data provide a novel methodology for collection of high-quality RNA from PBMCs for biobank researchers.

View Article: PubMed Central - PubMed

Affiliation: Division of Biobank and Data Management, Iwate Tohoku Medical Megabank Organization, Iwate Medical University, Shiwa-gun, Iwate, Japan.

ABSTRACT
Transportation of samples is essential for large-scale biobank projects. However, RNA degradation during pre-analytical operations prior to transportation can cause systematic bias in transcriptome data, which may prevent subsequent biomarker identification. Therefore, to collect high-quality biobank samples for expression analysis, specimens must be transported under stable conditions. In this study, we examined the effectiveness of RNA-stabilizing reagents to prevent RNA degradation during pre-analytical operations with an emphasis on RNA from peripheral blood mononuclear cells (PBMCs) to establish a protocol for reducing systematic bias. To this end, we obtained PBMCs from 11 healthy volunteers and analyzed the purity, yield, and integrity of extracted RNA after performing pre-analytical operations for freezing PBMCs at -80°C. We randomly chose 7 samples from 11 samples individually, and systematic bias in expression levels was examined by real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR), RNA sequencing (RNA-Seq) experiments and data analysis. Our data demonstrated that omission of stabilizing reagents significantly lowered RNA integrity, suggesting substantial degradation of RNA molecules due to pre-analytical freezing. qRT-PCR experiments for 19 selected transcripts revealed systematic bias in the expression levels of five transcripts. RNA-Seq for 25,223 transcripts also suggested that about 40% of transcripts were systematically biased. These results indicated that appropriate reduction in systematic bias is essential in protocols for collection of RNA from PBMCs for large-scale biobank projects. Among the seven commercially available stabilizing reagents examined in this study, qRT-PCR and RNA-Seq experiments consistently suggested that RNALock, RNA/DNA Stabilization Reagent for Blood and Bone Marrow, and 1-Thioglycerol/Homogenization solution could reduce systematic bias. On the basis of the results of this study, we established a protocol to reduce systematic bias in the expression levels of RNA transcripts isolated from PBMCs. We believe that these data provide a novel methodology for collection of high-quality RNA from PBMCs for biobank researchers.

Show MeSH

Related in: MedlinePlus

Data bias of transcriptome analysis in each condition.A. Correlation analysis of the average of FPKM under eight conditions for each sample. B. Cluster analysis of 56 transcriptomes: eight conditions for each of seven volunteers. C. Pair-wise comparisons of significant differences in gene expression for each sample. The number in each box shows the number of differentially expressed genes (p<0.05, Wilcoxon signed rank test).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4125218&req=5

pone-0104283-g004: Data bias of transcriptome analysis in each condition.A. Correlation analysis of the average of FPKM under eight conditions for each sample. B. Cluster analysis of 56 transcriptomes: eight conditions for each of seven volunteers. C. Pair-wise comparisons of significant differences in gene expression for each sample. The number in each box shows the number of differentially expressed genes (p<0.05, Wilcoxon signed rank test).

Mentions: As an indicator of normalized gene expression levels, FPKM values were calculated using TopHat and Cufflinks (File S1) [22], [23], [24]. Spearman correlation coefficient between ΔCt values measured by qRT-PCR experiments (GAPDH was used as a reference transcript) and the FPKM value calculated from RNA-Seq data was −0.762 (P<2.2×10−16), suggesting that qRT-PCR and RNA-Seq showed consistent quantitative results (Figure S2). Figure 4A shows the average and standard deviation of Pearson correlation coefficients between the Ctrl1 conditions and the seven other conditions. Importantly, the Pearson correlation coefficient between the Ctrl1 and Ctrl2 conditions was 0.9955±0.00057, indicating the markedly high reproducibility of our experimental results between replicates. Additionally, the Pearson correlation coefficient between the Without stab and Ctrl1 conditions was 0.9825±0.010, which was significantly lower than that between the Ctrl1 and Ctrl2 conditions (P = 0.016; Wilcoxon signed rank test). This result suggested that pre-analytical freezing induced a systematic bias in the expression level of the whole transcriptome. As expected, under the 1-Thio, Stab, Lock, and Protect conditions, Pearson correlation coefficients with the Ctrl1 condition were not significantly different from that between the Ctrl1 and Ctrl2 conditions (Figure 4A). However, the Pearson correlation coefficient between the SDS and Ctrl1 conditions was significantly lower than that between the Ctrl1 and Ctrl2 conditions (P = 0.016; Wilcoxon signed rank test). These results showed that the 1-Thio, Stab, Lock, and Protect conditions could reduce systematic bias due to pre-analytical operations. Pair-wise scatter plots between the eight conditions are shown in .


Reduction of systematic bias in transcriptome data from human peripheral blood mononuclear cells for transportation and biobanking.

Ohmomo H, Hachiya T, Shiwa Y, Furukawa R, Ono K, Ito S, Ishida Y, Satoh M, Hitomi J, Sobue K, Shimizu A - PLoS ONE (2014)

Data bias of transcriptome analysis in each condition.A. Correlation analysis of the average of FPKM under eight conditions for each sample. B. Cluster analysis of 56 transcriptomes: eight conditions for each of seven volunteers. C. Pair-wise comparisons of significant differences in gene expression for each sample. The number in each box shows the number of differentially expressed genes (p<0.05, Wilcoxon signed rank test).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4125218&req=5

pone-0104283-g004: Data bias of transcriptome analysis in each condition.A. Correlation analysis of the average of FPKM under eight conditions for each sample. B. Cluster analysis of 56 transcriptomes: eight conditions for each of seven volunteers. C. Pair-wise comparisons of significant differences in gene expression for each sample. The number in each box shows the number of differentially expressed genes (p<0.05, Wilcoxon signed rank test).
Mentions: As an indicator of normalized gene expression levels, FPKM values were calculated using TopHat and Cufflinks (File S1) [22], [23], [24]. Spearman correlation coefficient between ΔCt values measured by qRT-PCR experiments (GAPDH was used as a reference transcript) and the FPKM value calculated from RNA-Seq data was −0.762 (P<2.2×10−16), suggesting that qRT-PCR and RNA-Seq showed consistent quantitative results (Figure S2). Figure 4A shows the average and standard deviation of Pearson correlation coefficients between the Ctrl1 conditions and the seven other conditions. Importantly, the Pearson correlation coefficient between the Ctrl1 and Ctrl2 conditions was 0.9955±0.00057, indicating the markedly high reproducibility of our experimental results between replicates. Additionally, the Pearson correlation coefficient between the Without stab and Ctrl1 conditions was 0.9825±0.010, which was significantly lower than that between the Ctrl1 and Ctrl2 conditions (P = 0.016; Wilcoxon signed rank test). This result suggested that pre-analytical freezing induced a systematic bias in the expression level of the whole transcriptome. As expected, under the 1-Thio, Stab, Lock, and Protect conditions, Pearson correlation coefficients with the Ctrl1 condition were not significantly different from that between the Ctrl1 and Ctrl2 conditions (Figure 4A). However, the Pearson correlation coefficient between the SDS and Ctrl1 conditions was significantly lower than that between the Ctrl1 and Ctrl2 conditions (P = 0.016; Wilcoxon signed rank test). These results showed that the 1-Thio, Stab, Lock, and Protect conditions could reduce systematic bias due to pre-analytical operations. Pair-wise scatter plots between the eight conditions are shown in .

Bottom Line: RNA-Seq for 25,223 transcripts also suggested that about 40% of transcripts were systematically biased.On the basis of the results of this study, we established a protocol to reduce systematic bias in the expression levels of RNA transcripts isolated from PBMCs.We believe that these data provide a novel methodology for collection of high-quality RNA from PBMCs for biobank researchers.

View Article: PubMed Central - PubMed

Affiliation: Division of Biobank and Data Management, Iwate Tohoku Medical Megabank Organization, Iwate Medical University, Shiwa-gun, Iwate, Japan.

ABSTRACT
Transportation of samples is essential for large-scale biobank projects. However, RNA degradation during pre-analytical operations prior to transportation can cause systematic bias in transcriptome data, which may prevent subsequent biomarker identification. Therefore, to collect high-quality biobank samples for expression analysis, specimens must be transported under stable conditions. In this study, we examined the effectiveness of RNA-stabilizing reagents to prevent RNA degradation during pre-analytical operations with an emphasis on RNA from peripheral blood mononuclear cells (PBMCs) to establish a protocol for reducing systematic bias. To this end, we obtained PBMCs from 11 healthy volunteers and analyzed the purity, yield, and integrity of extracted RNA after performing pre-analytical operations for freezing PBMCs at -80°C. We randomly chose 7 samples from 11 samples individually, and systematic bias in expression levels was examined by real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR), RNA sequencing (RNA-Seq) experiments and data analysis. Our data demonstrated that omission of stabilizing reagents significantly lowered RNA integrity, suggesting substantial degradation of RNA molecules due to pre-analytical freezing. qRT-PCR experiments for 19 selected transcripts revealed systematic bias in the expression levels of five transcripts. RNA-Seq for 25,223 transcripts also suggested that about 40% of transcripts were systematically biased. These results indicated that appropriate reduction in systematic bias is essential in protocols for collection of RNA from PBMCs for large-scale biobank projects. Among the seven commercially available stabilizing reagents examined in this study, qRT-PCR and RNA-Seq experiments consistently suggested that RNALock, RNA/DNA Stabilization Reagent for Blood and Bone Marrow, and 1-Thioglycerol/Homogenization solution could reduce systematic bias. On the basis of the results of this study, we established a protocol to reduce systematic bias in the expression levels of RNA transcripts isolated from PBMCs. We believe that these data provide a novel methodology for collection of high-quality RNA from PBMCs for biobank researchers.

Show MeSH
Related in: MedlinePlus