Differential gene expression profiles are dependent upon method of peripheral blood collection and RNA isolation

Abstract

Background: RNA isolation and purification steps greatly influence the results of gene expression profiling. There are two commercially available products for whole blood RNA collection, PAXgene and Tempus blood collection tubes, and each comes with their own RNA purification method. In both systems the blood is immediately lysed when collected into the tube and RNA stabilized using proprietary reagents. Both systems enable minimal blood handling procedures thus minimizing the risk of inducing changes in gene expression through blood handling or processing. Because the RNA purification steps could influence the total RNA pool, we examined the impact of RNA isolation, using the PAXgene or Tempus method, on gene expression profiles.

Results: Using microarrays as readout of RNA from stimulated whole blood we found a common set of expressed transcripts in RNA samples from either PAXgene or Tempus. However, we also found several to be uniquely expressed depending on the type of collection tube, suggesting that RNA purification methods impact results of differential gene expression profiling. Specifically, transcripts for several known PHA-inducible genes, including IFNgamma, IL13, IL2, IL3, and IL4 were found to be upregulated in stimulated vs. control samples when RNA was isolated using the ABI Tempus method, but not using the PAXgene method (p < 0.01, FDR corrected). Sequenom Quantiative Gene Expression (QGE) (SanDiego, CA) measures confirmed IL2, IL4 and IFNgamma up-regulation in Tempus purified RNA from PHA stimulated cells while only IL2 was up-regulated using PAXgene purified (p < 0.05).

Conclusion: Here, we demonstrate that peripheral blood RNA isolation methods can critically impact differential expression results, particularly in the clinical setting where fold-change differences are typically small and there is inherent variability within biological cohorts. A modified method based upon the Tempus system was found to provide high yield, good post-hybridization array quality, low variability in expression measures and was shown to produce differential expression results consistent with the predicted immunologic effects of PHA stimulation.

Figure 1

Differences in RNA quality and yield between PAXgene™ and Tempus™. The tempus system has higher mean yields, improved RNA purity based on OD 260/230 ratios, less degradation based on GAPDH 3'/5 ratios, and a higher number of expressed transcripts based on Percent Present Calls.

Figure 2

A Hierarchical clustering of individual samples for the 1,338 transcripts generated from a union set of transcripts statistically significant as either up- or down-regulated using Tempus™ or PAXgene™ tubes. PHA stimulation is associated with the primary segregation of samples (Blue and Gold Bar), with RNA collection tube as a secondary grouping (Pink and Teal). B. Hierarchical clustering of fold change values within a participant/tube condition for PHA stimulation vs. no PHA stimulation shows consistent grouping by participant despite difference in collection tubes (Note color bars by participant). C. Venn diagram for comparisons of 3 hr PHA stimulation vs. 3 hrs with no stimulation. Substantial up-regulated transcripts using Tempus™ compared to PAXgene™, and a greater number of down-regulated transcripts using PAXgene™ compared to Tempus™.

Figure 3

Differential expression of individual subjects for four transcripts known to be differentially expressed upon addition of PHA: IL2, IL4, IFNG, and IL3. Using the Tempus ABI system, all four transcripts are identified as up-regulated by PHA stimulation, with an FDR-adjusted p value < 0.01 and a fold change > 2. These four transcripts are not identified using the same criteria from data generated using PAX system.

Figure 4

Sequenom QGE assay: Tempus™ appears to have slightly higher differential expression estimates than PAXgene™ for IL2, IL4 and IFNG. IL4, IFNγ, were statistically significant for Tempus ™ samples (P < 0.05). Both Tempus™ and PAXgene™ systems show IL2 as up-regulated (p < 0.05).