miRNA Expression in FirstChoice® Human Brain Reference RNA
Our R&D scientists have received inquiries about the microRNA (miRNA) content of the FirstChoice Human Brain Reference RNA. We felt that this data would provide useful information as it is the most thoroughly characterized, commercially available human tissue RNA. Here, the expression patterns of 297 miRNAs using Applied Biosystems TaqMan MicroRNA Assays for quantitative detection of mature miRNAs was characterized in the FirstChoice Human Brain Reference RNA used in the MAQC Project.
FirstChoice Total RNAs from human lung and placental tissues were compared to the Human Brain Reference RNA to evaluate expression levels of known miRNAs in these tissues. Data were compared using histograms and hierarchical clustering.
Triplicate reactions were averaged for each miRNA, and the raw threshold cycle (CT) values were plotted in histograms generated using JMP® Statistical Analysis Software (Figure 1). CT values were normalized to ten control assays for small nucleolar RNA (snoRNAs), and hierarchical clustering was performed on normalized ΔCT values using the Partek® Genomics Suite v6.2 software (Figure 2).
Using an expression level cutoff of 36.0 CT, 74% of all human miRNAs investigated were expressed in human lung, 78% were expressed in the human brain reference, and 89% were expressed in human placenta.
Figure 1. Distributions of Raw CT for miRNAs in 3 Human Tissues. The distributions of mean CT for 297 human miRNAs and 14 control snoRNAs are shown for placenta, lung, and brain (Human Brain Reference RNA) RNA. Plotted CT values are the average of 3 replicate PCRs. CT values that were considered undetermined by the SDS v2.2 software were set to a threshold of 40 cycles. Green regions within the histograms represent the position of the 14 snoRNA control assays. The boxplots shown above each histogram indicate the mean and median values, the inter-quartile range (sides of boxes), the upper 10% and lower 0.5%, 2.5%, and 10% quantiles. As one can see from the boxplots, placenta RNA had a substantially higher expression level on average than either lung or brain, which were similar. JMP® Statistical Analysis Software was used to build the plots.
Hierarchical clustering (Figure 2) revealed relationships of miRNAs unique to specific tissues.
Figure 2. Hierarchical Clustering of miRNAs in Three Human Tissues. Delta CT values (DCT) for 297 miRNAs were used to generate a heatmap and dendrograms relating miRNAs (tree on left) and placenta, lung, and Human Brain Reference RNA (tree on top). DCT values represent the mean CT of 14 snoRNAs (endogenous controls) minus the average CT (mean of triplicate PCR measurements) of each target miRNA within a given tissue (DCT = Endogenous Control minus Target). Using this formulation, DCT values ranged from ~ +7 to -14 (as seen in the scale bar at the bottom of the heatmap). As DCT increased, or became more positive, expression level increased for a miRNA in a given tissue. As in Figure 1, undetermined CT values were set to 40. As the branching relationship for the tissues indicate, brain and lung were more closely related to one another than either was to placenta. The Average Linkage method with Euclidean distance was used to generate the clustering relationships in the Partek® Genomics Suite v6.2.
Appropriate Control RNAs for miRNA and mRNA Expression Studies
We have profiled many additional tissue types (data not shown), and from this work we propose the use of human brain reference and placenta tissue RNAs as reasonable control RNAs for miRNA and mRNA expression studies. We will continue to characterize the human brain reference RNA as additional miRNAs are identified and new assays for them are released.
Andrew Lemire, Penn Whitley, Bob Setterquist • Applied Biosystems, Austin, TX