Identifying expressed genes

Publisher Website
Google Scholar
Abstract
The study of expressed genes has had a great impact on biological research (1). Expressed genes are the basic functional units of genomic DNA. Because these regions cannot be identified from genomic sequence information per se, the gene's products, messenger RNAs or proteins, must be isolated from cells, directly sequenced, and identified. As we steadily build up sizable expression databases that currently include more than 92,000 of the roughly 100,000 total human genes, the process of identifying the remaining undiscovered genes is becoming progressively more difficult. Existing databases of expressed genes now include virtually all of the abundantly expressed human genes—the easier to reach “low hanging fruit on the tree,” as well as many middle and rarely expressed genes. The genes that are still undiscovered are expressed at low levels or are specifically expressed only in certain cell types, developmental stages, or growth conditions. Such genes hold the promise of including key regulatory factors responsible for differentiated phenotypes, developmental progression, or cell growth regulation. As we move forward to identify these genes, highly efficient methods of removing, i.e., subtracting, the bulk of identified, abundant genes from cDNA libraries are required. In this issue of PNAS, Wang and colleagues (2) discover a flaw in current subtraction methods, which are now widely used to identify novel expressed genes. They show that the long poly(A) regions present in most expressed mRNAs generate a serious problem in subtraction reactions. Long poly(dT) regions of tester cDNA, which is generated from the RNA of interest, randomly hybridize with long poly(dA) regions of driver cDNA generated from the comparison cell type, resulting in template loss. This loss particularly affects low abundance mRNAs. Wang et al. predict that this flaw will limit the usefulness of current subtraction methods and result in a fall-off in gene identification …