We have developed a technology for the identification of genes differentially regulated between biological samples based around the hybridisation of RNA- derived probes onto nylon membranes containing immobilised cDNAs gridded at high density. We used this technology to examine changes in gene expression patterns in gastric epithelial cells infected with a virulent H. pilory strain at different time points. Radioactively labelled first strand cDNA, synthesised from KATO III gastric cell mRNA, was prepared and hybridised to high density arrays of amplified cDNA derived from a collection of 46,302 non-redundant I.M.A.G.E clones. Hybridisation signals were acquired using the GlaxoWellcome developed DGENT PC-based software package.

The large gene expression matrix generated with this approach has been analysed using cluster analysis to reveal correlated patterns of gene expression. The use of principal component analysis has also been explored. We found that a considerable number of genes map into discrete gene clusters suggesting that they belong to defined pathways of gene expression. We are investigating the possibility of applying this approach to gene expression matrixes based on the frequency of appearance of a gene transcript in representative libraries. Our results reinforce the use of statistical analysis in differential gene expression experiments.