Introduction
Tissue samples are routinely formalin-fixe
d an
d paraffin-embe
dde
d (FFPE) for long term preservation. Gene expression analysis of archival FFPE tissues may a
dvance knowle
dge of the molecular perturbations contributing to
disease. However, formalin causes extensive
degra
dation of RNA.
Methods
We compared RNA quality/yield from FFPE samples using six commercial FFPE RNA extraction kits. In addition we compared four DNA microarray protocols for the Agilent 8 ¡Á 60 K platform using 16 year old FFPE mouse liver samples treated with phenobarbital or vehicle.
Results
Despite low quality RNA, archival phenobarbital samples exhibited strong induction of the positive control genes Cyp2b9 and Cyp2b10 by quantitative real-time PCR (qPCR). We tested one- and two-color microarray designs and evaluated the effects of increasing the amount of hybridized cDNA. Canonical gene responders to phenobarbital were measurably induced under each experimental condition. Increasing the amount of labeled cDNA did not improve the overall signal intensity. One-color experiments yielded larger fold changes than two-color and the number of differentially expressed genes varied between protocols. Gene expression changes were validated by qPCR and literature searches. Individual protocols exhibited high rates of false positives; however, pathway analysis revealed that nine of the top ten canonical pathways were consistent across experiments. Genes that were differentially expressed in more than one experiment were more likely to be validated. Thus, we recommend that experiments on FFPE samples be done in duplicate to reduce false positives.
Discussion
In this analysis of archival FFPE samples we were able to identify pathways that are consistent with phenobarbital's mechanism of action. Therefore, we conclude that FFPE samples can be used for meaningful microarray gene expression analyses.
