Microarray Readers: Pushing the Envelope

To truly reap the benefits of the flood of information coming out of sequencing factories worldwide, investigators must move beyond the traditional notion of "one-gene, one-experiment," in favor of highly parallel, automation-friendly, and miniaturized assays. One such tool is the microarray--a matrix of biomaterials attached to a support such as glass or plastic.1-3 Using microarrays, scientists can perform hundreds or thousands of experiments in parallel, all thanks to a chip usually no bigge

Jorge Cortese
Dec 9, 2001
To truly reap the benefits of the flood of information coming out of sequencing factories worldwide, investigators must move beyond the traditional notion of "one-gene, one-experiment," in favor of highly parallel, automation-friendly, and miniaturized assays. One such tool is the microarray--a matrix of biomaterials attached to a support such as glass or plastic.1-3 Using microarrays, scientists can perform hundreds or thousands of experiments in parallel, all thanks to a chip usually no bigger than a microscope slide.

DNA microarrays, as the name suggests, are arrays of oligonucleotides, cDNAs, or other DNA targets, which promise to advance several biomedical goals. First, microarrays allow researchers to track global gene expression patterns characteristic of normal and disease states, and to identify genes that are up- or down-regulated when a drug or signaling factor is added.1 Such information can advance drug discovery efforts. Second, since gene expression patterns vary between physiological states or in different regions of an organ, such as the brain, there is considerable interest in the generation of comparative three-dimensional maps of gene expression in normal and diseased tissues.4 These maps can be produced by combining laser capture microdissection (LCM),5 single-cell DNA amplification, and microarray analysis. Finally, arrays can aid single-nucleotide polymorphism (SNP)3 and mutation pattern mapping, thus potentially individualizing medical treatments.

Scientists can label samples for microarray analysis with either radionuclides or fluorophores. Direct or indirect fluorescence labeling is generally preferred, because it enables direct sample-to-sample comparison on a single chip. Direct labeling involves the introduction of a covalently linked fluorophore in the nucleic acid sequence. For differential, two-color labeling, probe families with similar chemical structure but different spectroscopic properties are desirable. Examples include the cyanine (Cy) dyes (e.g., Cy3 and Cy5), as well as Eugene, Ore.-based Molecular Probes Inc.'s Alexa™ probes.6 Indirect labeling techniques create labeling sites on the biopolymer, followed in some systems by amplification of these linking sites and labeling with Cy3/Cy5.

Once the array is labeled, however, researchers need an imager or scanner to collect all of the data. Continuing The Scientist's ongoing series on microarray tools,7-9 LabConsumer now reviews the instrumentation used to read nucleic acid-derived microarray assays.

Scanners vs. Imagers

Microarray readers can be either scanners or imagers. Typically, scanners use one or more lasers and scan dual-fluorophore samples either sequentially or simultaneously; however, simultaneous scanning produces fewer registration errors. Union City, Calif.-based Axon Instruments Inc. offers the GenePix™ 4000 B Array Scanner that simultaneously reads at two excitation wavelengths, 532 and 635 nm.

The alternative to scanning arrays is imaging them with a charge-coupled device (CCD) camera.10 Unlike scanners, CCD-based imagers can link images covering a very large surface area, so that many slides can be imaged sequentially. Scanners can focus more energy to excite fluorophores and thus collect more light in less time. The resulting shorter integration times should be a big advantage over CCD imaging, however, this higher power output is not necessarily advantageous, since fluorophores can reach a saturation point at which further excitation leads to photobleaching or a non-linear emission response.

Mark Rand, senior product manager and applications scientist for Applied Precision Inc. of Issaquah, Wash., has carried out extensive studies on the process of imaging arrays, and explains that "a key difference with scanners is that CCD cameras can collect far more photons per pixel without reaching saturation." This makes the imagers' longer integration times and lower excitation power a benefit, rather than a drawback.

San Leandro, Calif.-based Alpha Innotech Corp.'s Alpha Array™ Reader is a high-throughput imager. Its optical system is enhanced by the NovaRay™ Light Management System, which reduces overall acquisition time: The instrument can image a 16 x 22-mm array in under one second. The instrument also overcomes some of the inherent problems with laser scanning, especially lateral displacement noise and photobleaching effects. With eight excitation wavelengths, the Alpha Array Reader can detect numerous fluorophores.

Some companies derive microarray imagers from general imaging devices. For example, Amersham Biosciences of Piscataway, N.J., recently entered this market with the Typhoon 9410, a high-end imager with extensive capabilities to tackle one- and two-dimensional gels, blots, tissue sections, macroarrays, and microarrays. The instrument can also be "flavored" to undertake specific needs in storage-phosphor, fluorescence, or chemiluminescence detections.

UVP Inc. of Upland, Calif., also bridges the gap between general lab imaging and microarray work. Three of the company's eight imaging workstations can gather microarray data. The BioMicro™ System is a microscope-compatible imager that can carry out high-resolution gel- or plate-imaging analysis, fluorescence microscopy, and microarray measurements. The BioChemi™ System has a cooled CCD camera and faster optics capable of visible light or colorimetric measurements, fluorescence detection, and chemiluminescence or chemifluorescence imaging. UVP's OptiChem™ System has a 1.4 mega-pixel, super-cooled CCD device (75°C below ambient) with five orders of magnitude linear dynamic range. The OptiChem is also capable of imaging and analyzing fluorescent microarrays, chemiluminescent membrane arrays, and microplate arrays.

Unlike laser scanners, imagers are not limited to single-wavelength excitation. For example, Applied Precision's arrayWorx™ instruments use white light.11 A white light beam--one that contains all visible wavelengths--is directed through an excitation filter in a filter wheel to give monochromatic illumination to the sample. The filter wheel allows researchers to acquire up to four wavelengths while the slide is stationary, eliminating spatial registration artifacts. The cooled scientific-grade CCD provides low noise and high quantum efficiency. The camera collects partial images (panels) from the designated area and reconstructs them using Stitch-by-Position™ Image Registration Technology.

Several arrayWoRx instruments are available. The arrayWoRxe Biochip Reader-Basic handles up to four-color simultaneous analysis, and includes two standard channels, Cy3 and Cy5; the arrayWoRxe Biochip Reader-Standard runs full area scanning even at 3.25-µm resolution and handles image files up to 500 MB in size. The Automated arrayWoRx Microarray Scanners include the same capabilities of the other arrayWoRx scanners as well as automatic slide loading and analysis for up to 25 slides. Vysis Inc. of Downers Grove, Ill., offers another CCD imaging microarray reader. The instrument, which uses a Xenon-lamp source, is part of the company's GenoSensor™ System, which includes the GenoSensor Array 300 consisting of 287 targets for postnatal and cancer research applications.

Companies can construct general scanning/imaging systems to detect both fluorophores and radionuclides. Stamford, Conn.-based Fujifilm Medical Systems USA Inc.'s FLA-5000 Science Imaging System can scan both fluorescent and radioisotopic labels in an area of up to 40 x 46 cm. With three standard lasers (437/532/635 nm) and ports for another two, researchers can use the device to read most fluorophores. An optional second photomultiplier allows simultaneous detection of two dyes in a single scan. The new Fujifilm FLA-8000 Fluorescence Image Analyzer can efficiently image macroarrays and a variety of other applications, such as labeled cells and radiolabeled samples; it has two standard lasers and one optional laser. Resolution can be as low as 5 µm for fluorescence images and 10 µm for radioisotope images. Although most microarrays are read using fluorescence, this instrument incorporates a newly designed Phosphor Imaging Plate (IP™ BAS-SR0813) that is scanned by the FLA-8000's red laser and emits blue fluorescence by photo-stimulated luminescence.

Overcoming Technical Hurdles

Courtesy of Tecan

Tecan's LS 200 scanner

By their very nature, microarrays push imaging equipment to their detection limits. These arrays can contain tens of thousands of 50-150-µm diameter spots per square centimeter, but they tend to be biased toward known and abundant gene products, even though many interesting genes are often expressed at relatively low levels. In practice, microarray readers can usually detect transcripts with a relative abundance of between one copy in 100,000 and one copy in 500,000, or 3 to 10 copies per cell.3,4 The sensitivity required for this low-level fluorescence detection is approximately two-to-five molecules per square micrometer, with a linear dynamic range of five orders of magnitude.

Image optimization increases demands on equipment. For instance, researchers often try to minimize fluorophore photobleaching (photochemical damage) and "bleed-through" (contamination of detected emission with that of additional fluorophores). Pixel size and resolution are also critical parameters for microarray scanners and imagers. In general, resolutions of 5-10 µm are needed to resolve common features. Santa Clara, Calif.-based Affymetrix's GeneArray™ Scanner can focus its argon-ion laser (488 nm) down to 3 µm.

Scanners use standard microscope objectives, producing images akin to moderately magnified fluorescent micrographs. Unfortunately, the large lateral focal plane is not homogenous, so the objective must be moved across the microarray or the array must be moved across the objective's field. This is a technical problem that increases non-uniformity across an array, and each instrument solves it in a slightly different manner.

Alameda, Calif.-based MiraiBio's FMBIO® IIe tackles this problem by moving the slide under a fixed objective. It uses a high-power, solid-state laser, and the optical unit moves under the sample as it is scanned using a high-speed rotating polygonal mirror; two optical fiber arrays and two photomultipliers accelerate the reading process. The device can read up to 48 medium-density, multicolor arrays simultaneously.

Another problem is the signal-to-noise ratio of array quantification: How does the instrument differentiate data from junk? One approach is to use confocal microscopy. In principle, confocal optics reject fluorescence from non-scanned focal planes, leading to lower backgrounds.12 But this requires the use of precisely aligned pinholes, heavy-duty moving components, and alignment that is maintained in every scan. Furthermore, scientists may find working with 2-µm confocal slices impractical when a depth-of-field of 20-30 µm suffices for most current quantitative purposes.

GeneFocus®, a division of Biomedical Photometrics Inc., based in Waterloo, Ontario, produces several microarray confocal-laser scanners in its DNAscope™ line: DNAscope IV, V, and LM. GeneFocus also offers an open- systems version of its confocal-based scanners, the Open Frame Research DNAscope, suitable for the development of new solid or "wet" biochips for a wide selection of fluorescent probes; high-resolution versions can image Affymetrix's GeneChips®.

Background contributions can also be minimized by non-confocal techniques. PerkinElmer Life Science of Boston offers two instruments with dark-field illumination to enhance background rejection. The GeneTAC™ 2000 Microarray Analyzer is part of the GeneTAC Biochip Solution from Genomic Solutions of Ann Arbor, Mich.7 The GeneTAC 2000 is compatible with various substrates including glass, plastic, ceramic, and metal. It can also detect luminescence. The alternative GeneTAC LSIV Microarray Analyzer has four excitation lasers and two additional photomultipliers. Both instruments use a 24-slide carousel for high-throughput applications.

Courtesy of Virtek Vision Corp.

Virtek's ChipReader Microarray Scanner

Another way to eliminate background noise and photobleaching is to optimize the microarray reader to operate at low laser power. Waterloo, Ontario-based Virtek Vision Corp.'s ChipReader™ Microarray Scanner is a very small, single-slide reader that uses this approach. In addition, researchers can control each channel's output continuously and independently during scanning. The ChipReader is a two-color scanner available at various resolutions, down to 3 µm. The ChipReader also has high sensitivity as it can detect less than 2.6 femtograms of labeled DNA on a 100-µm diameter spot with a reproducible signal-to-noise ratio greater than nine.

The final technical issue is the quality of the microarray itself. If the slide is not perfectly flat, the images will not be uniform or consistent, regardless of resolving power. Agilent Technologies Inc. of Wilmington, Del., offers a solution to this problem with its DNA Microarray Scanner, which uses dynamic auto focus. By adjusting the focus continually during scanning, the instrument minimizes the effect of glass slide aberrations, improving uniformity and enhancing sensitivity.

One typical test of array quality is the "concordance" correlation: Labeling an array with the same source RNA divided in two pools and labeled with two probes (e.g., Cy3 and Cy5) should lead to an array with a theoretical differential gene expression ratio of 1.0. From these experiments, users can estimate the minimal differential gene expression value that the array can detect.

Final Notes

A number of issues still exist regarding array-reading technologies. The first is data quality. Comparing data obtained in different labs can be problematic, and part of the blame rests on the array readers. However, analysis of a single array with instruments based on different reading principles indicates that inter-laboratory comparisons of experimental data are warranted.13 Such comparisons should help researchers take advantage of both their collaborators' unpublished work as well as research appearing in the literature.

Another issue is the convergence between imaging and scanning. Scanners will likely dominate in the short-term, but imaging is bound to become the principal form of analysis. Scanners may have an edge for very high-density arrays (if required for massive clinical testing). Alternatively, researchers may adopt techniques such as two-photon or near-infrared confocal microscopy,14 endowed with more restricted focus depth and low background.

As microarray usage becomes commonplace, microarray images will join the standard record-keeping photographs of Western blots, agarose gels, or tissue sections. Thus, a general-purpose imager will be ideal, especially from a financial point of view. Some already exist, such as MiraiBio's FMBIO II, which can detect DNA in both agarose gels and sequencing gels, in addition to scanning microarrays. Array reading can be incorporated as a module of operation in many general imagers, but it seems unlikely that the most sophisticated microarray readers will take this all-in-one approach.

Courtesy of Packard BioScience

Packard BioScience's ScanArray Express

Since the needs of microarray laboratories vary greatly between individual labs and university-level facilities, several companies offer lines of microarray readers so that labs can purchase equipment to meet their needs. For example, Meriden, Conn.-based Packard BioScience Co.'s ScanArray® line of upgradeable instruments ranges from the ScanArray LITE Microarray Analysis System, a two-color scanner, to high-end instruments such as the five-laser ScanArray 5000 XL. These instruments support applications ranging from standard microarray reading to sophisticated four-color SNP genotyping. The instrument features sequential scanning, the ability to decrease scan laser power to diminish photobleaching, and a focus depth optimized to eliminate reading artifacts caused by dust particles or fingerprints.

For all the hype surrounding them, microarrays are like a large collection of test tubes, miniaturized on a microscope slide. They answer a single question: What genes are expressed in a given sample? Yet they are part of a new era of multitasking that can and will change the way researchers do science.

Jorge D. Cortese (jorge_cortese@mindspring.com) is a freelance technical writer in Durham, N.C.
1. D.J. Lockhart, E.A. Winzeler, "Genomics, gene expression and DNA arrays," Nature, 405:827-36, 2000.

2. M. Schena et al., "Microarrays: Biotechnology's discovery platform for functional genomics," Trends in Biotechnology, 16:301-6, 1998.

3. M. Schena, ed., Microarray Biochip Technology, Natick, Mass.: Eaton Publishing, 2000.

4. Z. Luo, D.H. Geschwind, "Microarray applications in neuroscience," Neurobiology of Disease, 8:183-93, April 2001.

5. M.R. Emmert-Buck et al., "Laser capture microdissection," Science, 274:998-1001, 1996.

6. J.D. Cortese, "Let it shine," The Scientist, 14[6]:24, March 20, 2000.

7. J.D. Cortese, "Array of options," The Scientist, 14[11]:26, May 29, 2000.

8. J.D. Cortese, "The array of today," The Scientist, 14[17]:25, Sept. 4, 2000.

9. M. Brush, "Making sense of microchip array data," The Scientist, 15[9]:25, April 30, 2001.

10. J.D. Cortese, "Microscopy paraphernalia," The Scientist, 14:26, Dec. 11, 2000.

11. J.D. Cortese, "Light Idea," The Scientist, 14[17]:24, Sept. 4, 2000.

12. J. Pawley, "Fundamental limits in confocal microscopy," In: Handbook of Biological Confocal Microscopy, J.B. Pawley, ed., New York: Plenum Press, 1995, pp. 19-37.

13. L. Ramdas et al., "Comparative evaluation of laser-based microarray scanners," BioTechniques, 31:546-52, Sept. 2001.

14. E. Waddell et al., "High-resolution near-infrared imaging of DNA microarrays with time-resolved acquisition of fluorescence lifetimes," Analytical Chemistry, 72:5907-17, Dec. 15, 2000.

Supplemental Materials