Purchase this article with an account.
S.L. Bernstein, Y. Guo, L.E. H. Smith, B.J. Slater, Z. Mehrabyan; Optimized macroarray analysis of human retinal aging . Invest. Ophthalmol. Vis. Sci. 2004;45(13):1833.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: Array analysis is a powerful tool for identifying gene expression and event changes. A major problem in the utilization of membrane macroarrays is the inability to control intermembrane variation, which results in a high false positive identification rate. One advantage of macroarrays is that they are reusable, unlike glass chip microarrays. We report on a method for direct normalization of different macroarray membranes, to enable optimization of retinal aging data obtained by this method from multiple membranes. Methods: Human donor tissues were obtained using an approved IRB exemption protocol. GDA ver. 1.3 human cDNA arrays (N=6)(Invitrogen) with 18,376 double spotted ESTs were reacted with equivalent amounts of 33P radiolabeled cDNA probe generated from total RNA from a single human brain (visual cortex). Membranes were prehybridized, hybridized and stringently washed according to manufacturers recommended protocols, and exposed to a phosphorimaging screen (Kodak) for three days, imaged (molecular dynamics), and membrane spot information translated using commercial technology (Brinervision). Membranes were also exposed to autoradiographic film (Kodak Biomax). Following radiodecay (10 months), membranes were reacted with RNA from individual human donor samples; 3 young and 3 old. Experimental data for each spot was normalized for each membrane based on the brain cDNA control. Results: Wide variation occurred in intensity between membranes, for every spot. Single probe normalization enabled identification of a minimal subset of genes that are candidates for age–related changes. Eliminating ESTs by cutoff signal (200 CPM/young) reduced total EST number to 5440 (29.6% total). 554 (10%) ESTs were expressed at 2–fold or greater in old vs. young. These ESTs included alcohol dehydrogenase–6 and melanoma–1 antigen. In contrast, 3188 (59% or ∼6 fold more) ESTs were expressed at 2 fold or greater levels in young vs. old. ESTs preferentially expressed in young individuals included calbindin–2, anti–elastase and BCL–2. However, common controls such as GAPDH showed wide variation between samples. Conclusions:Despite array optimization, wide variation exists in array analysis using human donor tissue samples. These variations are likely due to multiple variables. Results obtained by pooling even moderate array sample numbers likely over–estimate the total overall genes changing during aging. Results obtained from human tissues must be confirmed by statistically valid numbers and multiple methodologies.
This PDF is available to Subscribers Only