by
Abstract
An NIH-sponsored symposium on filter arrays gave companies a chance to discuss their products and researchers an opportunity to learn more about the strengths and limitations of this exciting new technology.
You've probably seen advertisements for microarrays in scientific magazines, or in a paper from Pat Brown's lab at Stanford University. The technology is being hyped as one that will not only change the way we do experiments, but fundamentally alter the types of questions we can ask. But what are microarrays, and how can a scientist who is not a molecular biology or genomics expert get started with this rapidly evolving technology? To help National Institutes of Health scientists enter the array arena, the 150-member Microarray User Group (an NIH Special Interest Group, formed in September 1998 to allow users or potential users of microarray technologies to share information and reagents) sponsored a mini-symposium in December that focused on filter arrays, one common type of array.
What Is a Microarray?
Arrays go by a variety of different names, such as high-density, micro-, or macroarrays, but all are
conceptually very much the same. Thousands of nucleic acid fragments, usually synthetic oligonucleotides or cDNAs, are affixed to a
solid support - a glass slide, chip, or nylon membrane - in an ordered layout or array. They can then be probed with a
labeled sample, usually cDNA generated from an RNA sample of interest, by methods similar to those used in Southern
analysis. Although the specific labeling and hybridization techniques vary depending on what type of solid support is used,
the means of comparing gene expression is essentially the same. For example, in the case of filter arrays (the focus of
the mini-symposium), the level of hybridization to the same spots on separate but identical filters, or in separate
hybridizations of the same filter to different cDNA probes, are compared. The relative spot intensities correspond to the
relative expression levels of the gene in the two RNA samples used to make the cDNAs.
Using microarrays, one can assay the expression of thousands of genes in a single experiment or series of experiments. There are essentially two types of experiments for which the arrays can be used: analysis of the expression of known genes and gene discovery. The two applications are not necessarily mutually exclusive, but the arrays need to be chosen appropriately depending on the desired goals.
The filter array format is currently the easiest and most economical for an individual scientist to use.
Filter arrays are made and marketed by three companies, Clontech, Research Genetics, and Genome Systems; all three gave presentations at the mini-symposium, providing a useful opportunity for side-by-side comparison.
Evaluating the Expression of Known Genes
The first speaker of the symposium, Alex Chenchik of Clontech, discussed a set of arrays the company has developed for use in studying the expression of known genes. These range from the Human Atlas array, which contains 588 cDNA fragments of known genes, to more limited, application-targeted arrays containing smaller, more specific sets of genes. These limited arrays are useful to scientists interested in studying the expression of a specific group of genes, such as those involved in the cell cycle or neurobiology. While all of the companies produce arrays of human genes, not all have them for other species, and the selection is slimmer. For example, Clontech also produces broad-coverage arrays of 588 known cDNAs for mouse and rat, but currently offers only one application-targeted array.
Because Clontech is working with known genes, the company has been able to take advantage of the cDNA sequences to try to maximize sensitivity and diminish background signal. Many of the cDNA fragments in the Atlas array are not just sequences chosen from the ends of the genes, such as those in other arrays, but have been carefully selected from regions thought to be unique to the individual genes. Additionally, the method used to generate probes is unusual. For most arrays, the probe is generated by a reverse transcription reaction using an oligo-dT primer. Since the oligo-dT will prime off of any mRNA, this creates a probe mixture that, theoretically, contains cDNA representatives of all the mRNA species. The Clontech probe generation is primed using a mixture of specific primers, resulting in a probe that contains only cDNAs representative of the genes in the array.
Discovering New Genes
The next two presenters described arrays designed for gene discovery (an area into which Clontech is also planning to move). Both Genome Systems' GDA and Research Genetics' GeneFilter arrays contain some known genes, but they consist primarily of expressed sequence tags (ESTs), which are cDNA fragments representative of as-yet-unidentified genes. ESTs have been sequenced from a number of cDNA libraries, many by the international genomics research collaboration known as the IMAGE (Integrated Molecular Analysis of Genome Expression) Consortium.
In comparing gene expression between mRNA samples, one might find a number of ESTs that are
differentially expressed. Having this list of ESTs is only the beginning, however. As emphasized by many of the
mini-symposium speakers and later during the panel discussion, it is important to remember that the arrays are just a
screening method and should not be taken as providing totally conclusive evidence. Any suspected changes in gene
expression need to be confirmed by another method, such as Northern analysis or a ribonuclease protection assay. To
facilitate such confirmation, the companies sell clones containing the ESTs that can be used to generate probes for the
analysis. After confirming expression, one must decide how to further characterize the interesting ESTs. For example,
after comparing the ESTs to standard databases such as the Unigene or TIGR (The Institute for Genomic Research) databases,
one possible approach is to try to clone the full-length cDNA identified by the EST.
Research Genetics' GeneFilters are specifically designed for gene discovery. As described by the company's Abdel Elkahlown, the 5 x 7-cm filters contain approximately 5,000 ESTs per filter. The spots on the GeneFilters are actually the PCR amplification products of EST clones. Since the number of identified ESTs is increasing at a rapid pace, Research Genetics plans to release a new filter every 90 days until at least 40,000 genes are represented. As of this writing, there have been three human GeneFilter releases, in addition to a "Known Genes" GeneFilter; rat and mouse arrays are being developed for future release. The EST clones spotted on the filters are verified clones, meaning that the EST clones Research Genetics received from the IMAGE Consortium have been resequenced. Knowing exactly which clone has been spotted makes subsequent analysis easier.
The Genome Systems' GDA filters differ from GeneFilters in a number of different ways. As described by the company's Sam LaBrie, the GDA filters are larger (22 x 22 cm) and contain more ESTs - 18,000 for the human set. GDA filters are also available with mouse ESTs (16,000/filter). To make it easier to detect false positives or negatives, each clone is represented by two spots on the filter. Currently, each spot is the lysate of a bacterial clone containing the appropriate EST. In an effort to decrease the background signal, however, Genome Systems plans to switch from lysates to PCR products of the ESTs. LaBrie went on to discuss the company's new focus on GEM arrays, chip-based arrays that are an outgrowth of Genome Systems' acquisition by Incyte.
Array Analysis
The second speaker from Research Genetics, Elaine Poplin, highlighted its Pathways analysis software. In filter arrays, each sample is radiolabeled with 33P or 32P and hybridized to an individual filter, and the signal is then detected using a phosphorimager. All three companies have software packages for analyzing the phosphorimager data. The software's basic approach is to take the image, set up a grid overlaying the array, and measure the intensity of the spots. The software can then report the data in a variety of ways. One useful summary format is a listing of the genes ranked by their expression levels in one sample relative to a second sample.
Panel Discussion
The mini-symposium ended with a panel discussion with all of the company representatives. Two overriding
themes emerged from the discussion: validation and sensitivity. As before, the speakers emphasized the need to validate any
observed differential expression by other methods. The issue of sensitivity is more complex. All three companies have
similar estimates of how many molecules of RNA their array can detect. A more interesting point, however, was made by
Clontech's Alex Chenchik. He drew a line diagram resembling a topographical map of the ocean floor, and then drew what
would be the water line. Much of the drawing was below this line - representing the rare transcripts, expressed at only a few copies per cell. If one is searching for a specific new gene, a pearl at the bottom of the ocean, the arrays will not be very useful. But what lie above the water line are the genes that can be detected by the current generation of arrays.
Since there are still large numbers of ESTs corresponding to genes with no known functions, using the arrays to harvest
genes at the water's edge can be a very fruitful strategy.
Emily Klotz is a postdoctoral fellow at the National Cancer Institute.
Caleb Brown is an illustrator and biologist living in Montana. By day he drives a delivery van, and by night he draws pictures with his computer.
Large-Scale Gene Expression and Microarray Links and Resources - an extensive collection of Web resources and pointers.
DNA Microarrays - some interesting Web sites ("introductory, detailed, and industrial") and papers on DNA microarrays.
Division of Intramural Research: Microarray Project - information about database design, development, and management, plus downloadable data sets. From the National Human Genome Research Institute.
Related HMS Beagle articles:
One-Step Genome Analysis - an HMS Beagle Profile on Synteni, one of several companies using microarrays to analyze the expression of genes.
Know Thy Cell - an HMS Beagle Meeting Brief outlining new techniques, including microarrays, highlighted at the 1998 annual meeting of the American Society for Cell Biology.