| RNArticles Validation of Array Analysis Results The advent of array analysis has allowed researchers to rapidly analyze expression data for several thousand genes at a time. Typically, RNA is isolated from both a control and an experimental sample, and labeled during reverse transcription or subsequent aRNA amplification. These labeled samples are then hybridized to separate, identical arrays; comparing the signal from each element of the arrays allows the researcher to determine if the gene represented by the element has different levels of expression in the control and test samples. While seemingly straightforward, the technique involves a cascade of steps that are highly susceptible to variability, from manufacturing of the array, to reverse transcription and labeling of the samples, to differences in hybridization. Array data might reflect this introduced variability rather than actual changes in mRNA expression profiles, leading to representation levels of rare transcripts that are sometimes hard to replicate (poor precision). Array users also have different parameters that define "up" or "down" regulation of a gene. Some researchers set a 2-fold difference in expression level as significant while others require a 4-fold difference in expression for significant increased representation. It is therefore important to validate array data using a separate method of expression analysis. The first paper below describes "multiple-array" analysis and subsequent verification of differentially expressed mRNA transcripts using relative quantitative RT-PCR. In the second paper, Rajeevan and colleagues use real-time RT-PCR to validate clones that showed a 2-4 fold difference in expression across 2 different RNA samples. Welsh, J., Zarrinkar, P., Sapinso, L., Kern, S., Behling, C., Monk, B., X Lockhart, D., Burger, R. and Hampton, G. (2001) Analysis of gene expression profiles in normal and neoplastic ovarian tissue samples identifies candidate molecular markers or epithelial ovarian cancer. Proc. Natl. Acad. Sci. 98(3):1176-1181. Recently, Welsh and coworkers have taken the array concept a bit further, by creating an "array of arrays," consisting of 49 individual arrays on a single glass wafer, using gaskets to separate the individual array sections during the hybridization step. This allowed simultaneous analysis of 49 RNA samples, creating over 250,000 data points. The authors used this technique to better understand the cascade of gene expression involved in ovarian cancer; to this end they used RNA samples from tumors, normal ovaries, B cells (to help gauge infiltrating immune cells) and ovarian cell lines. The results allowed them to categorize the tumors into three subgroups, and to propose several candidate diagnostic markers. The authors used an RT-PCR assay to evaluate the expression of the candidate genes. Amplification of an internal standard within the RT-PCR procedure controls for quantity and quality differences in the RNA sample. The authors used the 18S rRNA primers and competimers provided in Ambion's QuantumRNA™ 18S Internal Standards Kit. The QuantumRNA Internal Standards contain patented "competimer technology" to allow quantitative multiplex PCR. Rajeevan M.S., Vernon S.D., Taysavang N., and Unger E.R. (2001) Validation of Array-Based Gene Expression Profiles by Real-time (Kinetic) RT-PCR. J. Mol. Diagnostics 3(1):26-30. In this second paper, Rajeevan et al. used real-time RT-PCR to confirm differential expression of genes identified by high-density filter arrays (HDFA). Overall, 71% of the genes with >2-fold difference in expression level in HDFA were confirmed to be differentially expressed by real-time RT-PCR. Although 88% of the HDFA results were validated when HDFA showed a greater than 4-fold difference in expression or when the HDFA hybridization intensity was strong, only 1 of 4 genes that were uniquely expressed in one of the samples was confirmed to be differentially expressed by real-time RT-PCR. In addition, among the 14 differentially expressed genes confirmed by real-time RT-PCR, 10 genes showed expression differences by RT-PCR that were greater in magnitude than that determined by HDFA. This group concludes that genes identified by array analysis, with a two- to fourfold difference in expression level between samples, cannot be rejected or accepted as "real" without further validation. Confirmation of Array Data is Highly Recommended Publications with conclusions based on gene array data typically include supporting data generated from classical, highly controlled experiments measuring differential gene expression such as RT-PCR, Northern blots, and RNase protection assays. Manufacturers of gene arrays also recommend corroboration of results by alternative methods due to sequence-dependent hybridization characteristics and variations inherent to any hybridization reaction. Ambion provides many tools to validate array data: - QuantumRNA™ 18S Internal Standards - One of the best internal standards for use in relative quantitative RT-PCR. Can be RT-PCR amplified to the level of any mRNA.
- Competitive Quantitative RT-PCR Kits - A gel-based method for quantitating specific gene expression by RT-PCR.
- FirstChoice™ Human Tumor/Normal Adjacent Tissue RNA - Pure, intact total RNA for gene expression analysis within individual tumor samples and adjacent normal tissues.
- FirstChoice™ Northern Blots - Pre-made Northern blots, containing RNA from specific mouse or human tissues; saves time; optimized for maximum sensitivity.
- NorthernMax™ Complete Northern Blotting Kit - Provides RNase-free reagents for Northern analysis and an extensive Instruction Manual for optimization and troubleshooting of the procedure.
- RPA III™ Ribonuclease Protection Assay Kit - A sensitive, single-tube method for detection, quantitation, and mapping of mRNA transcripts; completely compatible with multi-target analysis.
|