Vision-based analyzer fingerprints disease

Andrew Wilson, Editor, [email protected]

Applied Biosystems (ABI; Foster City, CA, USA; www.appliedbiosystems.com) was recently asked by the US Food and Drug Administration (Jefferson, AR, USA; www.fda.gov/nctr) to determine why a specific diabetes drug can cause liver problems. ABI decided to use its Model 1700 analyzer, which uses gene expression to provide information about how cells react to such drugs.

The researchers cover a microarray surface with messenger RNA (mRNA)-the form of RNA that carries information from DNA in the nucleus to the ribosome sites of protein synthesis in the cell-containing samples extracted from blood or tissue samples of rats. These mRNA molecules are then spotted with oligonucleotides, which are short sequences of synthesized DNA or RNA that bind to their counterparts and are used as probes. Thus, they can be used to match a region where a mutation is known to occur.

The resulting hybrid molecules are chemoluminescent at 458 nm. So every spot corresponding to how a particular a gene is formed or “expressed” in the sample glows to varying degrees. How brightly it glows depends on how much of its particular mRNA appeared in the sample. Those spots corresponding to genes not expressed remain dark.

Microarrays are covered with glowing spots whose intensity signals the level of expression for each gene. Strongly expressed genes are represented by brightly glowing spots. Unexpressed genes are represented by spots that remain dark.

“One single measurement can result in more than 30,000 normalized value points or data points,” says Roland Wicki, ABI director of marketing. “With more than 50 or 100 of these arrays, millions of data points need to be compared.”

Researchers load a microarray onto a motorized stage under the system’s image sensor. While a much simpler optical system suffices to image microarrays in their chemoluminescent light, not all the spots light up, so there is some difficulty registering the array within the image field. To solve this problem, ABI adds a fluorescent dye to every spot along with the oligonucleides.

When excited by a flash of light from an array of LEDs, the fluorescent dye responds by emitting light at 658 nm. “Most fluorescent dyes have something on the order of 30-40 nm difference between the excitation and emission,” King says, “so the excitation light must be blocked by a narrow-pass optical-interference filter to prevent it saturating the detectors.

Narrow-bandwidth filters, however, only work properly with light rays moving parallel to the optical axis. This requires a two-element optical system. The first lens is a 35-mm lens with a 50-mm focal length mounted with the microarray at its conjugate focus. These photographic lenses are corrected to have flat image planes when focused at infinity, so putting the flat microarray at the conjugate focus provides a collimated beam for the interference filter. A second lens, identical to the first, focuses the beam to form an image on a 1777 × 2361-pixel CCD array.

The system exposes six images for each microarray. Two images (one for each half of the microarray) are taken with an exposure time of 25 s to locate every spot on the microarray and two images with an exposure time of 5 s are required to obtain measurements of brightly luminescent spots. Two images with an exposure time of 24 s are used to image spots that are not fluorescing as brightly. This combination of long and short exposures substantially increases the imaging system’s dynamic range. Image information is then processed on a host PC using proprietary ABI application software to acquire and process the images.

The software performs a primary analysis to convert spot brightness to normalized measures of gene expression, and the results are recorded in an Oracle 9i database.