A new form of microarray analysis, using beads and fiber optics, reveals new subtleties in gene expression, potentially aiding in the diagnosis of cancer and other diseases. Kevin Davies reports.
By Kevin Davies
May 7, 2002 | If the Guinness Book of World Records included a category for the most proteins encoded by a single gene, there would have been a new entry last year. In an extraordinary report published in Cell, Larry Zipursky and colleagues at the University of California at Los Angeles told the bizarre story of a gene in the fruit fly Drosophila melanogaster called Dscam (the name comes from a related gene on human chromosome 21 called Down syndrome cell adhesion molecule), thought to be important in controlling and targeting the growth of neurons during development.
Zipursky's group found that the single Dscam gene has the capacity to produce a staggering 38,000 discrete messenger RNA (mRNA) transcripts (the chemical message read out to produce protein) — more than twice the total number of genes in the entire Drosophila genome. Four coding sections (or exons) within the gene, comprised of more than 90 related sequences, can be edited or spliced together in myriad combinations to produce distinct mRNA molecules. While the human Dscam gene exhibits much less variability, the three human neurexin genes are known to produce thousands of isoforms, resulting from alternative promoters and alternative splicing.
There are about 32,000 genes in the human genome, although uncertainty over the precise tally will likely continue long after the "final" draft is completed next year. Early estimates suggest about 60 percent produce two or more isoforms, and that the transcriptome — the collection of all transcripts produced by the human genome — may be around 70,000. Although only a preliminary estimate, it shows how straightforward cataloguing the human genome was in comparison to surveying the complete repertoire of proteins encoded therein.
Many of these splice variants have important functional significance, either in controlling normal development or in the pathogenesis of disease states, including cancer, making the development of new technologies to monitor the expression profiles of gene isoforms a priority. Writing in the April issue of Nature Biotechnology, researchers from the University of California at San Diego, led by Xiang-Dong Fu, in collaboration with Mark Chee and coworkers from San Diego-based Illumina Inc., describe a method to study alternative splicing on DNA microarrays using a novel bead system and fiber optics. Chee, a cofounder of Illumina, was formerly a senior scientist with Affymetrix, helping to pioneer methods for resequencing DNA.
In the Illumina BeadArray system, oligonucleotides are attached to microspherical beads (3 or 5 µm in diameter) rather than physically attached to a glass slide. These oligonucleotide sequences are specially designed to provide distinct and easily recognizable "addresses" for each bead, and do not correspond to naturally occurring sequences. The beads aggregate on a bed of shallow pits etched into a bundle of about 50,000 optical fibers, with about 100 beads for each "address" to counteract fluctuations in expression level. Streptavidin fluorescence is detected by means of a CCD camera. (The technology is already in use in Illumina's SNP genotyping service, and the platform should be released later this year.)
The new method for monitoring alternatively spliced genes is termed RASL — for RNA-mediated annealing, selection, and ligation. Oligonucleotides of 20 bases in length are designed to match the exon boundaries. These are fused to
|J. M. Yeakley, J-B. Farn, D. Doucet, L. Luo, E. Wickham, Z. Ye, M. S. Chee & X-D. Fu. "Profiling alternative splicing on fiber-optic arrays." Nature Biotechnology 20: 353-358.
the "address" oligonucleotides and other sequences to aid subsequent amplification. After hybridizing to the RNA sample, only sequences that align contiguously across a splice junction will be amplified. These sequences are then hybridized to the addresses on the beads.
Yeakley and coworkers studied the relative levels of expression of around 100 alternatively spliced transcripts from 23 known genes in five cancer cell lines. RASL revealed major differences in the levels of various gene isoforms with known roles in the cell cycle and signaling. For example, the p16 gene (mutations in which cause hereditary melanoma) shows clear variations in two isoforms in different cell lines. Another interesting example concerns the fibroblast growth factor receptor-2 gene, which is expressed in two forms, K or B, depending on the selection of exon 10 or 11, respectively. RASL revealed significant expression changes between the two isoforms, leading to the suggestion that abnormalities in the signaling loops between these two variants could play a role in some cancers.
The current RASL technology has some limitations. Rare transcripts are poorly amplified in the presence of more abundant messages, and the procedure is limited to surveys of known genes and splice variants. But with exciting advances in the molecular profiling of cancer flooding the scientific literature, RASL offers an important new tool for these studies. And as Paula Grabowski, a Howard Hughes Medical Institute investigator at the University of Pittsburgh, points out, RASL should help address "the daunting challenge of how our elaborate proteomes ... are generated from 'relatively modest' genomes."