Pooled, High-Quality Reference RNA For Human Microarrays

Universal Human Reference RNA for microarray expression profiling

Pooled, High-Quality Reference RNA for Human Microarrays

Natalia Novoradovskaya • Terry Payette • Alexey Novoradovsky • Jeff Braman

Stratagene

Nicki Chin • Alexander Pergamenschikov • Michael Fero • David Botstein

Department of Genetics, Stanford University

Universal Human Reference RNA for microarray gene expression profiling is a mixture of high-quality total RNA isolated from 10 human cell lines. OD_260/280 ratio of individual cell line RNA composing the reference RNA is minimally 1.8. The RNA is intact and devoid of RNase activity as judged by gel electrophoresis. The reference is efficiently labeled with radiolabeled and fluorescent nucleotides and hybridizes to a majority of microarray probe elements providing a consistent internal control and allowing data comparison between different microarray experiments.

Reliable and precise microarray gene expression profiling relies on comparison of hybridization efficiency between experimental and reference RNA target samples^1,2. Differences in hybridization intensity between these RNA targets reflect relative differences in gene expression levels. The easiest experiment of this type to perform and analyze involves comparing the expression profile of an experimental RNA sample resulting from exposure to some perturbation relative to an unperturbed reference sample. The reference sample in this experimental design is relatively easy to obtain. However, for comparison of multiple experimental RNA samples, a common reference composed of a collection of experimental RNA samples is required. The importance of this common reference was emphasized by Eisen and Brown¹ when they stated “This [referring to the reference] maintains the essential internal-control aspect of two-color hybridization, but allows inferred comparisons to be made among large numbers of samples without requiring that every pairwise comparison be performed.” Accordingly, examining a large collection of experimental RNA samples, within and between gene expression experiments, requires a large quantity of reference RNA composed of an equal mixture of all experimental RNA targets. The quantity and complexity of this reference sample makes it extremely problematic to obtain.

To address the necessity for large quantities of reference RNA containing low and high abundance RNA species, Stratagene has developed Universal Human Reference RNA for microarray gene expression profiling.

Total RNA Isolation

The reference RNA is composed of total RNA isolated from cell lines representing different tissues using a modified StrataPrep^® total RNA miniprep procedure.³ This method, based on RNA binding to glass fiber, produces high-quality RNA free of protein, DNA and organic solvent contamination. Each of these contaminants reduces RNA labeling efficiency (data not shown). Individual cell line RNA included in the reference RNA possesses OD_260/280 ratio no less than 1.8. Formaldehyde-agarose gel electrophoresis demonstrates distinct 18S and 28S ribosomal RNA bands with no evidence of degradation (Figure 1). Equal mass quantities of the highest quality total RNA from each cell line are pooled to make the reference RNA.

Optimized Hybridization Coverage of Microarrays

Fig. 1

Cell lines for reference RNA are chosen based on several important criteria. The foremost is that cDNA produced from the reference RNA hybridizes to as many microarray probe elements as possible. The reference RNA cell lines providing these consistent and reproducible RNA populations include testis, brain, liver, skin, breast, cervix, T and B- cells, macrophages, and lipocytes.

When considering optimized coverage on a microarray, a pool of RNA from too many cell lines will result in dilution of low-abundance genes represented in each individual cell line. Therefore, optimized the reference RNA was prepared from the fewest number of cell lines expressing the largest number of genes hybridizing to microarrays (data not shown). To determine the percentage of array probes hybridizing to individual cell line RNA and the reference RNA, total RNA from these samples were reverse transcribed with [a³³-P] dTTP and radiolabeled cDNA was hybridized to Stratagene’s GeneConnection^™ Discovery microarray.⁴ Microarrays were visualized using a phosphorimager (Packard Bioscience) and analyzed using OptiQuant Image Analysis Software. Signal distribution among more than 4000 spots on the microarray demonstrates that cDNA derived from HURR hybridizes to a higher percentage of array probes than cDNA from any individual cell line (Figure 2).

Fig. 2

Further evidence demonstrating this point was obtained by reverse transcribing a single cell line RNA and the reference with Cy5-dUTP and Cy3-dUTP, respectively, followed by hybridization to a 24,000-spot human microarray from Stanford University. The general procedure for this experiment is shown in Figure 3. The array was scanned using an Axon scanner and analyzed using GenePix Pro 3.0 software. Figure 4 shows a portion of the microarray image and the ratios of red to green signals were used to compare gene expression profiles in different cell lines. Once again, the reference RNA hybridized to a larger number of array probes than the individual cell line RNA.

Fig.3

Fig. 4

Conclusions

Universal Human Reference RNA is a unique blend of high-quality total RNA from 10 carefully chosen human cell lines. Using the same reference RNA in different microarray experiments provides a common denominator for accurate and reproducible comparison of gene expression data. In addition, use of the same reference RNA among different research groups allows inter-laboratory comparisons as well. We recommend using Universal Human Reference RNA as a reference sample in any multicolor hybridization experiments using human cDNA microarrays.

REFERENCES

1. Eisen, M.B. and Brown, O. (1999) Methods in Enzymology 303: 179-2.
2. Ross, D.T., et al. (2000) Nature Genetics (2000) 24:  227-235.
3. Dolter, K. et al. (2000) Strategies 13: 12-14.
4. Garrison, D. et al. (2000) Strategies 13:  4-5.