The identification of membrane-associated and secreted genes that are differentially expressed is a useful step in defining new targets for the diagnosis and treatment of cancer. Extracting information on the subcellular localization of genes represented on DNA microarrays is difficult and is limited by the incomplete sequence and annotation that is available in existing databases. Here we combine a biochemical and bioinformatic approach to identify membrane-associated and secreted genes expressed in the MCF-7 breast cancer cell line. Our approach is based on the analysis of differential hybridization levels of RNAs that have been physically separated by virtue of their association with polysomes on the endoplasmic reticulum. This approach is specifically applicable to oligonucleotide microarrays such as Affymetrix, which use single-color hybridization instead of dual-color competitive hybridizations. Assignment to membrane-associated and secreted class membership is based on both the differential hybridization levels and an expression threshold, which are calculated empirically from data collected on a reference set of known cytoplasmic and membrane proteins. This method enabled the identification of 755 membrane-associated and secreted probe sets expressed in MCF-7 cells for which this annotation did not previously exist. The data were used to filter a previously reported expression dataset to identify membrane-associated and secreted genes which are associated with poor prognosis in breast cancer and represent potential targets for diagnosis and treatment. The approach reported here should provide a useful tool for the analysis of gene expression patterns, identifying membrane-associated or secreted genes with biological relevance that have the potential for clinical applications in diagnosis or treatment.
|Number of pages||6|
|State||Published - Dec 1 2004|