DNA Chips and Microarrays Part 1
Scientists who wish to use DNA chips or microarrays to pursue their investigations have a growing variety of choices. Vendors offer off-the-shelf and customized versions as well as support for do-it-yourself chips.
by Peter Gwynne and Gary Heebner



ADVERTISERS

Affymetrix
www.affymetrix.com

American Type Culture Collection
www.atcc.org

Axon Instruments
www.axon.com

Biosearch Technologies, Inc.
www.biosearch.com

Commonwealth Biotechnologies, Inc.
www.cbibiotech.com

GeneMachines, Inc.
www.genemachines.com

Genemed Biotechnologies, Inc.
www.genemed.com

Genescan Europe
www.genescan-europe.com

Genetix, Ltd.
www.genetix.co.uk

Hitachi Genetic Systems, Ltd.
www.miraibio.com

IBC/USA Conferences
www.ibcusa.com

Incyte Genomics
www.incyte.com

Infectio Diagnostic
www.infectio.com

MWG Biotech AG
www.mwg-biotech.com

Scanalytics, Inc.
www.scanalytics.com

Schleicher & Schuell, Inc.
www.s-and-s.com

Sequenom, Inc.
www.sequenom.com

Silicon Genetics
www.sigenetics.com

Stratagene
www.stratagene.com

TeleChem International, Inc.
www.arrayit.com

Virtek Vision
www.virtekvision.com


CONTENTS

A a few life scientists may mourn the passing of the days when the concept of "one gene, one experiment" controlled their professional lives. But most of their colleagues have welcomed the arrival of the DNA chips and microarrays that offer researchers the opportunity to run thousands of samples simultaneously in a single experiment under virtually identical conditions. The pharmaceutical industry in particular values the use of microarray technology to screen increasing numbers of molecules in smaller volumes as drug candidates. "There's tremendous excitement about the technology," says Alex Szabo, vice president of Stratagene. "Everyone realizes that it's one of the key technologies in the genomic era."

Microarray technology seems tailor-made for the type of exploration necessary to follow up the initial work on sequencing the genes of humans and other organisms. "Genome projects give you, in a sense, a list of the words in the genomes' vocabulary," says Patrick Brown, professor of biochemistry at Stanford University and a pioneer in developing microarrays for research use. "If you want to learn what words mean in a foreign language you look at how they are used. It's the same for genes. Microarrays as a way of seeing how genes express themselves will be the most widely used application of arrays." Jeff Mooney, business technology manager of Corning Microarray Technologies, extends that thought. "The more we look at the human genome, the more questions people have. Microarray platforms help to answer general and specific questions."

Beyond that, researchers see use of microarrays in such areas as genotyping, studying disease pathways, analyzing single nucleotide polymorphisms (SNPs), and examining proteins. "Expression arrays offer researchers the promise of finding the fundamental causes of disease and identifying new, more precise strategies to diagnose, treat, prevent, and ultimately cure disease," says Stephen Fodor, chairman and CEO of Affymetrix, Inc., the first major manufacturer of arrays.

Plenty of vendors have joined Affymetrix in the microarray marketplace. "There are tens, if not hundreds, of companies out there trying to find the next technology," says Andrew Farquharson, executive vice president of Operon Technologies, Inc. Some newcomers, such as NimbleGen Systems, Inc., aim to follow the model pioneered by Affymetrix and Incyte Genomics, producing microarrays for core facilities in large industrial and academic departments. Others, such as Corning, plan to enter the market with "theme arrays" targeted at specific diseases. Yet more, including Agilent Technologies and German company Graffinity Pharmaceutical Design, GmbH, provide specific services such as printing and fingerprinting arrays designed and used by individual researchers. And several vendors, among them CLONTECH Laboratories, a subsidiary of BD Bioscience, and British firm BioRobotics, Ltd., provide the basic tools necessary for individual researchers to carry out the entire process of producing microarrays, including printing.

Types of Microarray

The concept emerged 10 years ago. "Light-directed, spatially addressable parallel chemical synthesis," a paper by Fodor and colleagues in Science (251:767 – 773, 1991), opened the way for the entire microarray industry. DNA microarrays are microscopic groups of thousands of DNA molecules of known sequences attached to a solid surface such as a nylon membrane or a simple glass microscope slide. Each array consists of an orderly organization of samples that provides a medium for matching known and unknown samples based on base-pairing rules and automating the process of identifying the unknowns. Microarrays come in several varieties, each of which has specific advantages for research and screening.

Brown's team at Stanford, in collaboration with Mark Schena and Ron Davis, working as consultants to Affymetrix, developed the basic technology for what scientists now regard as the traditional type of microarray. It uses lengths of complementary DNA (or cDNA) produced from cellular messenger RNA using the reverse transcriptase polymerase chain reaction (RT-PCR). Stretches of cDNA about 500 to 5,000 bases long are immobilized onto a substrate and exposed to a set of targets either separately or in a mixture. "We thought that the real application of array technology was being able to look at gene expression programs on a very large scale," recalls Brown. "The technology needed to be user-friendly and cheap."

The other form of microarray consists of oligonucleotides or peptide nucleic acids synthesized either insitu on the chip or by conventional synthesis followed by immobilization on the chip. The array is exposed to labeled sample DNA and hybridized in order to determine the identity and abundance of its complementary sequences. Scientists at Affymetrix developed this type of microarray, which is often called a DNA chip. The company notes that its short oligos have strong specificity because they can distinguish single base mismatches. Affymetrix also optimizes hybridization conditions to maximize the sensitivity of oligos of different lengths.

Operon Technologies, Inc. has designed oligonucleotide probes longer than usual. "We've done a number of experiments with oligos of different lengths," says Farquharson. "We've determined that the 70- to 80-base range is optimal for specificity and sensitivity. One of the powers of making oligo arrays is that you can look at your genome and design the arrays to match it."

Arrays can accommodate biomolecules besides DNA. CLONTECH has a unique type of RNA microarray. "We offer over 100 human tissues on the chip," says Paul Siebert, vice president of research. "In cases where we have limited amounts of diseased tissue — in tumor samples, for example — we convert the RNA into cDNA. We also expect to have our first antibody array by the end of this year."

Arrays can also use proteins. Ciphergen Biosystems, for example, has produced a commercial protein chip for use in proteomics research. Some pharmaceutical companies are also using microarray formats to screen the activity of various chemical compounds against targets in their drug discovery programs. In addition some researchers have begun to work with cell and tissue arrays. Several companies now offer these as part of their product lines.

These fundamental platforms for microscale experimentation have resulted from the marriage of several technologies. Robotic engineering, pin technology, molecular biology, DNA sequencing, optical and laser technology, and informatics have all contributed to the development of microarrays. Since few companies can develop expertise in all these areas, individual firms must partner with others to create compatible systems. Alternatively researchers intent on preparing their own arrays must assemble components from several vendors.

Off-the-Shelf vs. Do-It-Yourself

Through much of the past 10 years life scientists intending to use microarrays have faced the decision whether to use commercially available versions of the technology or to embark on their own efforts to design, adapt, and apply their own microarrays to a greater or lesser extent. In fact do-it-yourself microarraying predated the commercial variety. It has continued as a key option for life scientists since then.

As more life scientists have begun to demand microarray technology, the do-it-yourself approach has expanded. So has the need for relevant products and services. "We have a tremendous amount of interest from scientists who want to build their own microarrays specific to their own research using our instrumentation," says Michael Kane, vice president, genomics for Genomic Solutions, Inc. Vendors have responded to the growing needs. "There seem to be more and more entrants each year," adds Stuart Elmes, technical director of BioRobotics, Ltd. "But the market is beginning to pick out a few companies, including ours, that are achieving a critical mass."

One reason for the popularity of do-it-yourself microarraying is the expense of commercial versions. "Cost is still the barrier for this technology to become more accessible to the bench scientist," says Leming Shi, a senior research scientist at BASF Corporation who has created an information website (www.gene-chips.com) about DNA microarrays. "The tool needs to be as accessible and inexpensive and familiar as, say, a microscope," adds Brown, "because the kind of information you can get from it is useful whatever aspect of biology you're studying. Until the commercial companies can provide the technology at a price that makes it a no-brainer for scientists to use, the do-it-yourself approach will have a role."

Critics also charge that commercial chips are somewhat inflexible in terms of applications. While denying any real lack of choice, commercial microarray-makers have started to expand their offerings. "Over the past year Affymetrix has made major technological improvements to provide customers with more information on a single chip," says Fodor. "These improvements have enabled us to form Perlegen Sciences, a genomic subsidiary founded with the objective of scanning 50 human genomes using our GeneChip® technology — a feat impractical through any other methods. In the future, Affymetrix will develop next-generation genotyping arrays developed using the haplotype patterns found by Perlegen." Fodor adds that Affymetrix offers its own do-it-yourself systems via Genetic MicroSystems, which it acquired last year.

Other firms have begun to emerge in the microarray manufacturing market. For example, several scientists at the University of Wisconsin, Madison recently founded NimbleGen to commercialize technology that they had developed in the laboratory. That firm recently announced that it is developing a lower cost DNA microarray system with more flexibility for users. "We look at our role as complementary to Affymetrix's," says Michael Treble, the company's president and CEO. "Our microarrays are cheaper and faster to produce. But our main competition comes from home-grown microarrays."

The Role of Robotics

Laboratory automation has played a major role in allowing scientists to process samples at lower costs. It has done so by using platforms that handle smaller and smaller sample volumes, from microcentrifuge tubes to microwell plates to microarrays. Over a fairly short period of time researchers have shifted from milliliter to microliter to nanoliter sample volumes. Automated workstations and robotics help to reduce the intensive manual labor involved in handling many small samples. Such companies as Packard Instrument and Zymark have developed expertise in laboratory automation.

Robotics not only handle and process small volume samples in high throughput environments. They are also used to place (or spot) extremely small samples onto the solid supports, such as nylon membranes or glass slides, that make up the substrates for microarrays. "Robotics are of the highest importance for microarrays," says Willem Pleusters, head of surface modifications for German company Eppendorf AG. "Otherwise it's impossible to get the quality you need. Regardless of the technology you use, the reproducibility of the array is critical." Sarah Stevens, senior scientist at British firm Genetix, Ltd., agrees. "The robotics for making high-density arrays is actually key," she says. "We have liquid handling capacities on our robots to do tasks such as handling plasma preparations."

This approach makes a stark contrast to experiments based on wet chemistry. That traditional method involves mixing reagents together in a solution and permitting diffusion to play a key role in determining how molecules interact. Solid phase chemistry enables microarray makers to take advantage of chemical bonding to attach a reagent to the solid support. This allows the reagent to be located in a specific position on a surface, where it becomes easily identified and differentiated from other samples on the same surface.

Arrays of regularly organized samples on a flat surface are ideal for experiments in which a large number of small samples has to be subjected to the same conditions. Each spot in such an array can represent a different chemical experiment and can generate unique data. Several identical array patterns printed on a solid surface provide multiple replicates of each sample. Arrays provide both assay miniaturization and duplication of samples.

Suitable Substrates

"Microarrays have five different components once you have your samples and your libraries," explains Steve Lee, director of R&D for MiraiBio Inc. formerly the genetics division of Hitachi Software Engineering America, Ltd. "They include the chip with its surface, the spotter or the ability to do insitu synthesis, a fluidic system to hybridize the chip, a scanner to read it, and — very important — sophisticated programs that allow you to quantify and interpret your results." In the following sections we shall outline basic considerations relevant to these components.

Producing a reliable DNA chip or microarray involves careful control of several critical operations. Significant factors include the density of the array, the method of spotting a sample onto a membrane or slide, and the application that determines the type of material that the spotter places on the substrate.

The density of an array is determined by the size of the spots, the distance between spots (otherwise known as the pitch), and the support material. The method of printing impacts both the spot size and distance between spots. Macroarrays, offered by CLONTECH, Geno Technology, Inc., and Sigma-Aldrich, among other vendors, are usually printed on nylon membranes about 8 cm by 12 cm. in area. They have spots with diameters greater than 300 um. These arrays can accommodate up to 5,000 spots. This format is ideal for researchers in laboratories that may be operating on a limited budget who do not need high throughput applications. "We were the first company to offer a macroarray in nylon three years ago," says CLONTECH's Paul Siebert. "Our approach was to get something out that people could afford without having to buy confocal microscopes and other expensive equipment. We produce microarrays on plastic as well as glass."

Microarrays are usually printed on glass or silicon. In a microarray, the diameter of the spots is typically less than 200 um. Some microarrays offer nearly 10,000 spots, while a high-density oligonucleotide microarray may have 40,000 or more spots (or features) per chip. Incyte Genomics, NEN Life Sciences and Research Genetics, among other companies, offer these chips. Affymetrix notes that its standard and custom arrays have more than 400,000 different features.

Hitting the Spot

Spotters can use pin, ink-jet, and other technologies to deposit samples onto the support material. Several of the more common methods utilize metal pins, which can be either solid or split. When the pins are dipped into wells that contain the compounds of interest, each picks up a small amount of the material. The pin is then brought into contact with the solid support and a nanoliter volume is dispensed at the desired location. In split pins (otherwise known as quills) a slot cut into the head of the pin functions as a reservoir for the compound being spotted. Quill pins are most often used with glass slides, while solid pins are typically used for spotting membranes.

Amersham Pharmacia Biotech, GeneMachines, and other companies offer spotting robots. "We supply equipment for scientists to print their own arrays," says Elmes of BioRobotics. "Our technology can make large numbers of arrays, in the range of hundreds of thousands or millions. But it is more suitable for medium numbers, from hundreds to thousands or tens of thousands of arrays. It's also more flexible than many on the market. You can make arrays with oligonucleotides, PCR products, proteins, or whatever you want."

MiraiBio, meanwhile, has developed a spotter based on a patent pending proprietary pin design originally intended for the semiconductor industry. "It holds a perfect globe of fluid at the end of the four-pronged tip, resulting in consistent spot shape," explains Lee. "It is spring-loaded, rather like a ballpoint pen. It gives scientists the ability to control the spotting speed and where it hits. Our arrayer allows you several adjustments, x, y, and z, so that the surfaces it spots can be very flexible. It can work on very fragile membrane surfaces, for example."

Genetic MicroSystems, meanwhile, has developed a unique pin and ring technology. Because it is less subject to clogging of quills or pins, this method is more adaptable to different types of array that researchers might wish to make.

Ink-jet technology provides another method of spotting microarrays. Adapted from the printer industry and redesigned for use in biotechnological applications, this uses piezoelectric crystal oscillators and an electrode guidance system to deposit the compound in a precise location on the slide or membrane. Companies such as Cartesian Technologies and ProtoGene Laboratories use this technology.

ProtoGene's spotter, for example, can deposit individual nucleotides to form oligonucleotides insitu (that is, on the chip). Using separate print heads for each base eliminates cross-contamination of nucleotides. In addition this technology produces spots very consistent in size and works at very high speed. Agilent, taking advantage of its Hewlett-Packard heritage, also uses ink-jets both to deposit presynthesized oligos and PCR products and, using a different writer system, to deposit individual nucleotides to create sequences of interest at predefined locations.

Affymetrix, the market leader in DNA microarrays, uses a proprietary method of photolithography. The company bases its approach on photolithographic masks similar to those used to produce computer chips. The masks control the light-sensitive removal of protective chemical groups from hydroxyls in regions of the slide that are not masked. This allows the altered nucleotides to react with bases in the reaction solution to grow the DNA sequence. The process is repeated until it produces DNA sequences about 20 to 25 bases in length. This system produces very high-density microarrays. It is fairly expensive in absolute terms although competitive in terms of price per bit of information. Changing or modifying the photolithographic mask also takes time and comes at some significant cost.

Affymetrix and Others

Affymetrix led the way in large-scale production of DNA microarrays. "Because we are able to take advantage of semiconductor manufacturing techniques, we're able to make anywhere from 40 to 400 individual GeneChip arrays at one time," explains Fodor. "To meet the rising demand for our products, we substantially increased our manufacturing capacity with the completion and validation of a second manufacturing facility. The manufacturing capacity of this new facility is designed to keep pace with future demand."

Affymetrix offers a broad range of products and services from its standard GeneChip system to custom services for scientists who have budgets that allow them to purchase custom DNA microarrays from a commercial supplier. The company has also helped academic researchers gain access to these chips by creating an academic pricing structure, which allows university and nonprofit labs to gain access to these products.

A scientific team at NimbleGen has developed a new insitu arraying method that resembles Affymetrix's photolithographic approach. It differs, however, in that it does not require the use of photolithographic masks. "The primary difference is the way we pattern the light," explains Roland Green, the company's vice president and chief technology officer. This new "maskless" technology uses a system of very tiny mirrors from Texas Instruments. It directs ultraviolet light to each desired pixel simply by changing the position of individual mirrors in the system. These mirrors control the path of the light and ultimately determine which pixels will be activated for DNA synthesis. "The process allows us to redesign a chip in a matter of hours," says Green. "Chips are cheaper and faster to produce."

This maskless DNA microarray system has particular interest for life scientists because it can provide them with a great deal of flexibility and customization in producing their microarrays. Rather than make a new photolithographic mask each time the DNA microarray is changed or modified, this system requires the user simply to reprogram the mirrors. That action redirects the light to different pixels on the chip.

San Francisco-based Mergen Ltd. introduced what it calls the ExpressChip (TM) microarray system in October 1999. The company makes the chips in human, rat, and mouse multiple formats. The system contains two identical glass microarray slides, key reagents for hybridization and detection and a detailed instruction manual. Each slide is prespotted with DNA sequences functionally important for cell metabolism and/or disease development. "We are the only company that uses spotted oligos," says Qian Jin Hu, Mergen's president and CEO. "We think that this has much more advantage for quality control. We prepurify the oligos, which gives much better quality control than for insitu preparation. And the cost and feasibility are better."

Corning has entered the field with what it calls index arrays of 10,000 or more genes. The idea is to create a platform less rigid than those commercially available at present. "Our initial topic is looking at expression profiling for people who want to do target identification," says Jeff Mooney. "At the moment we have a yeast array on the market. Our plans right now are to introduce a human array product during the second quarter of this year."

Spot the Spots

However high its purity and spotting quality, a microarray will lose its value without an effective method of reading its results. "The ability to image what you've got is very important," says MiraiBio's Lee. "The factors to include are the sensitivity of the instrument, its resolution, and the ability to get a proper focal depth. The speed of scanning is also a critical factor for core facilities."

The method of signal detection used with DNA chips depends on the type of label used in an experiment. Common tagging methods include fluorescent, radioactive, and enzymatic approaches. Companies that offer systems of these types include Axon Instruments, Genetic MicroSystems, Genomic Solutions, and MiraiBio.

Fluorescent labels are detected with confocal laser scanners specifically designed for use with DNA microarrays. These scanners can eliminate unwanted background fluorescence by limiting the field in which the system picks up signals to those regions above the plane of the array where the substrate is located. This minimizes detection of stray fluorescent signals from the substrate, dust particles, or the slide itself. These scanners often include software for analysis and interpretation of the data.

Radioactive labels can be imaged with a phosphorimager or the much less glamorous but still effective autoradiography film. The most common radiolabels for this procedure are the phosphorus isotopes 32P and 33P. Both have pluses and minuses. 32P produces a stronger signal and costs less. The more expensive 33P produces a weaker signal but can be used over a greater range. Radiolabels are most often used with nylon membrane macroarrays, such as those offered by CLONTECH.

A third detection method is less commonly used than the other two. Research groups can apply enzymatic detection to nylon membrane macroarrays. Enzymatic systems generally rely on a spectrophotometer to automate the detection process. Alternatively a scientist can inspect the result visually. Both Display Systems Biotech and Genzyme carry systems for enzymatic detection.

A Matter of Interpretation

Once they complete an experiment with DNA microarrays, scientists need to interpret its readings. Microarrays with thousands of samples or spots can produce huge volumes of data. Storing and analyzing these data can cause a serious bottleneck in laboratory research. In fact, some researchers hope to perform array experiments first with the large comprehensive chips, such as Affymetrix's gene on a chip products, and then down-size their research efforts by focusing on a specific family of genes.

This way of operating stems in part from the complexity of working with large volumes of data. "Most researchers getting into the field are not accustomed to dealing with large data sets, often across large numbers of sample. They may not be in a position to readily derive pertinent results from these large data sets," says Michael Kane of Genomic Solutions. "They may find out only later that their study design and data analysis strategy do not meet their overall research objectives."

Some researchers have partnered with computer programmers to develop in-house software for their individual research efforts. This is not an easy task, however. A programmer who wants to understand the subtlety of interpreting biological data must have a hard-to-acquire familiarity with biology and sometimes with clinical diagnostics. Scientists who have a strong background in computer programming increasingly find themselves in high demand as recruits to the new field of bioinformatics.

BASF's Leming Shi exemplifies the qualities needed to interpret microarray experiments. Trained as a molecular modeler, he gained his first real exposure to life science during his postdoctoral research. "I was amazed at how much I was sold on the biology," he recalls. "My interest now is in getting the knowledge out of microarray data for gene discovery, drug discovery, or research. My goal is to integrate bioinformatics and cheminformatics."

Those scientists less than enthusiastic about writing programming code have an alternative. Several suppliers have developed commercial software for analyzing and interpreting data from DNA microarrays. Affymetrix, Lion Bioscience, Spotfire, and Silicon Genetics, among others, offer software packages for this purpose. However, Shi warns against the danger of too much mixing and matching. "The field is still not in a very mature stage even in representing molecular data," he says. "Manufacturers have different standards and a lot of companies have different platforms. What is available from the market cannot always do what the individual department or group needs." The implication: "Each group will have to develop its own software for some years to come," Shi says.

Stress on Standards

The informatics area is not the only segment of microarrying that stands to benefit from standardization. "What's clearly missing are technical standards for comparing different systems," says Lee of MiraiBio. "There are several different technologies in the market," adds Elke Zimmermann, Eppendorf's product manager. "Our feeling is that producing arrays is not the only critical step. It is also important to look in detail at all the different aspects of the arrays, including preparation, labeling, and data analysis, to know what you are doing exactly. It is very important to look at quality control and at standardization."

Dirk Vetter, CEO of Graffinity Pharmaceutical Design, echoes that thought. "Scientists are worrying about the data quality of microarrays," he says. "There's no really good way to validate whether you have produced good data or weak data. The field faces the danger of making decisions that involve spending a lot of time and money based on possibly poor data. You need a security filter because you have to have rock-solid data."

The issue has reached beyond the laboratory. "Standardization is definitely a problem right now. At a recent conference in Germany the discussion concerned the fact that researchers don't always believe in the results of other groups because they are using their own methods and standards," Zimmermann remembers. "There's some hesitation to join the club and be frank about results," concurs Vetter.

Scientists involved in the discipline have done more than complain about the lack of standardization. "I can hear people saying very loudly that they would like to form a committee or consortium to set standards on the quality of data from arrays," says Vetter. "Committees are already forming, but it will require quite a long-term effort and will represent a technical challenge."

Some companies have already started to meet the challenge. Affymetrix, for example, instituted internal controls from the start of its manufacturing. "They debugged early," says Vetter. "That may give them the opportunity to come out as a standard." Fodor adds that the company's manufacturing process, leveraged from semiconductor production technology, permits it to make arrays in wafers and to exert quality control over each lot.

Help with Home Brews

The issues of standards and quality control have the most obvious impact on life scientists who choose to go their own way in preparing microarrays. In fact universities and pharmaceutical and biotechnology companies now make life simpler for such scientists than it used to be half a dozen years ago. Research groups and departments in academic laboratories are joining forces to create core facilities for DNA microarray design, production, and even analysis. These facilities can provide a great deal of the expertise required for microarray work. In particular they eliminate the extensive training that each individual researcher requires before embarking on fully fledged microarray work. The core facility concept also allows individual researchers to work together and share the costs of this large financial investment required for production and other necessary microarray instruments.

Nevertheless, core facilities have their limitations. "There will always be people who like the control of having their own operation," says Chris Bailey, gene array product manager of Sigma-Aldrich. So several academic researchers continue to create and use do-it-yourself systems to produce their own DNA chips in the laboratory. Indeed, says Stratagene's Szabo, "The vast majority of experiments are being done with do-it-yourself arrays."

These rely largely on the robotic spotting system originally developed by Patrick Brown for his lab at Stanford. Brown has since made the protocols of his home-brewed technology available to other laboratories to permit them to produce their own DNA microarrays.

Soon after the emergence of commercial microarray technology, several companies started to offer spotting systems for home-brew use. That represented a major advance for the many academic researchers who lacked the engineering expertise required to follow the practices of the Brown lab. Cartesian Technologies, Genetic MicroSystems and Packard Instrument, among others, continue to offer spotting instruments that are more affordable for the academic researcher.

Agilent has taken a new and different approach. "Our technology lets scientists define and then design their own arrays and bring those designs to us to print them," says Mel Kronick, the R&D manager for Agilent's bioresearch solutions group. "Scientists in our technology access program can choose the oligos they want to create for their designs and then have the arrays ready in a matter of days at a very reasonable cost. It's rather like extending the process of ordering oligos to arrays." Mergen, meanwhile, offers full service for its range of arrays.

Do-it-yourself users of arrays stand to gain more than faster production at lower cost. "Everyone is so centered around human microarrays," says Simon Sims, manager of arrays for Sigma-Genosys. "But a significant proportion of scientists out there needs to have custom options to make their own chips for other organisms. We have developed applications for small arrays focused on, for example, apoptosis and the cytokine market. We offer very open systems." Certainly some mouse and rat arrays are available on the market. But the numbers of genes on those chips are underrepresented.

Several vendors provide ancillary products to home-brewers. "The vast majority of experiments are performed by researchers who have a need for simple biological tools to make their arrays and run them to collect useful data," says Stratagene's Szabo. "Our mission is to provide the complementary tools to make the process reliable."

Applications of Arrays

What of the applications of microarrays? Structural genomics has emerged as the prime customer. This involves the determination of the genetic sequences in the DNA of living organisms, such as humans, mice, and Drosophila. It is a huge undertaking that will provide a strong foundation for studies of the function of these genes and of the discovery of drugs that can be used to alter aberrant metabolic functions.

During the week of 15 February 2001, both the government-funded Human Genome Project consortium and the privately funded Celera Genomics published their drafts of the human genome. Analysis of these genome sequences surprisingly revealed that the 3 billion base pairs that make up the 23 pairs of chromosomes of the human genome seem to code for no more than 30,000 to 40,000 genes. Scientists had previously believed that the figure would exceed 80,000. In addition to discovering why the new estimate falls so far short of expectation, the sequence data will provide extraordinary opportunities for life scientists to discover new targets for developing drugs and otherwise intervening in diseases.

The vehicle is functional genomics, which includes applications in such areas as expression profiling, SNP detection/diagnostics, and personalized medicine. Several pharmaceutical companies have drug discovery programs based on functional genomics. These include companies like GENSET, Millennium Pharmaceuticals, Inc., and Lion Biosciences. DNA array technology provides a major tool in most functional genomics programs.

To understand the biochemical processes involved in the normal and diseased states of living cells, for example, researchers can use microarrays to detect differences in gene expression between different cells. The microarray permits researchers to examine thousands of different genes in the same experiment and thus to obtain a good understanding of the relative levels of expression between different genes in an organism.

Single nucleotide polymorphisms are likely to play a significant part in molecular diagnostics, because scientists know that several DNA base changes help to determine an individual's risk of developing a specific disease. As investigators associate specific SNPs with various diseases, they will develop diagnostic tests to screen populations of individuals for those at increased risk of disease. These tests will also prove valuable in screening asymptomatic individuals who are actually in the early stages of specific diseases at the time of screening. Identifying which SNPs are predictors for disease requires screening of many individuals to establish a pattern of SNPs versus disease states. Affymetrix has already developed a SNP microarray for this purpose. Called the GeneChip® HuSNP Mapping Assay, it has approximately 1,500 human SNPs per array.

Any ultimate goal of SNP analysis is personalized medicine. One aspect of the concept envisions doctors running drug metabolism profiles on their patients' DNA to determine how the individual patients will respond to certain commonly used drugs. The vehicle for the test: a SNP DNA microarray. Clinical researchers are on the verge of identifying a number of clinically relevant SNPs associated with cytochrome P450 and drug metabolism. As more information emerges in this area, biomedical researchers will gain a better understanding of the ways in which different individuals respond to the same drug treatment.

The Next Steps

Recent advances in DNA chip technologies have occurred primarily in the area of refining the process of creating DNA chips. Several newly developed spotters allow academic researchers to gain access to this technology. Affymetrix and other suppliers continue to invest heavily in creating new products to meet researchers' changing needs more effectively. Growing numbers of firms offer products and services to do-it-yourself microarrayers. MiraiBio offers a liquid bead instrument that permits scientists to perform 100 automated assays in a single tube.

Furthermore, there always seems to be a new company on the horizon promising ways to develop arrays and gain data from them more accurately, faster, and at a lower cost. "The whole business of microarray systems is going to change a lot," says Sarah Stevens of Genetix. "More and more companies are starting to manufacture premade arrays. A lot of groups working on common organisms will buy premade arrays to get the economies of scale in human, mouse, and Drosophila systems. But we will still see a market for do-it-yourself scientists working on unusual organisms and special organisms."

DNA chips will continue to create opportunities for researchers to discover drugs at faster and faster rates. It will also help them to discover important disease-related SNPs. This relatively young research tool has been quickly embraced by researchers and will be extraordinarily important in the process of scientific research and discovery, both now and in the future. "It's clear that the performance and sensitivity of the arrays haven't reached any limits set by any physical laws," says Stanford University's Brown. "The limits are simply set by the state of the technology. The underlying logic whereby array technology allows you to get the big picture of any biological process is so compelling that it will be played out for many years to come."

Peter Gwynne is a freelance science writer based on Cape Cod, Massachusetts, U.S.A. Gary Heebner is president of Cell Associates, a scientific marketing firm in Chesterfield, Missouri, U.S.A.

WEBLINKS
ADVERTISERS

Affymetrix
www.affymetrix.com

American Type Culture Collection
www.atcc.org

Axon Instruments
www.axon.com

Biosearch Technologies, Inc.
www.biosearch.com

Commonwealth Biotechnologies, Inc.
www.cbibiotech.com

GeneMachines, Inc.
www.genemachines.com

Genemed Biotechnologies, Inc.
www.genemed.com

Genescan Europe
www.genescan-europe.com

Genetix, Ltd.
www.genetix.co.uk

Hitachi Genetic Systems, Ltd.
www.miraibio.com

IBC/USA Conferences
www.ibcusa.com

Incyte Genomics
www.incyte.com

Infectio Diagnostic
www.infectio.com

MWG Biotech AG
www.mwg-biotech.com

Scanalytics, Inc.
www.scanalytics.com

Schleicher & Schuell, Inc.
www.s-and-s.com

Sequenom, Inc.
www.sequenom.com

Silicon Genetics
www.sigenetics.com

Stratagene
www.stratagene.com

TeleChem International, Inc.
www.arrayit.com

Virtek Vision
www.virtekvision.com
FEATURED COMPANIES
and ORGANIZATIONS


Affymetrix, Inc.
www.affymetrix.com

Agilent Technologies
www.agilent.com

Amersham Pharmacia Biotech
www.apbiotech.com

Axon Instruments
www.axonet.com

BASF Corporation
www.basf.com

BioRobotics, Ltd.
www.biorobotics.co.uk

Cartesian Technologies, Inc.
www.cartesiantech.com

Celera Genomics
www.celera.com

Ciphergen Biosystems, Inc.
www.ciphergen.com

CLONTECH Laboratories
www.clontech.com

Corning Microarray Technologies
www.corning.com

Display Systems Biotech, Inc.
www.displaysystems.com

Eppendorf, AG
www.eppendorf.com

GeneMachines
www.genemachines.com

Genetic MicroSystems, Inc.
www.affymetrix.com

Genetix Ltd.(UK)
www.genetix.co.uk

Geno Technology, Inc.
www.genotech.com

Genomic Solutions
www.genomicsolutions.com

GENSET
www.genxy.com

Genzyme Corporation
www.genzyme.com

Graffinity Pharmaceutical Design, GmbH
www.graffinity.com

Human Genome Project
www.nhgri.nih.gov/hgp

Incyte Genomics, Inc.
www.incyte.com

ISIS Pharmaceuticals
www.isispharmaceuticals.com

Lion Bioscience, GmbH
www.lionbioscience.com

Mergen Ltd.
www.mergen.com

Millennium Pharmaceuticals, Inc.
www.mlnm.com

MiraiBio, Inc.
www.miraibio.com

NEN Life Science Products, Inc.
www.nen.com

NimbleGen Systems, Inc.
www.nimblegen.com

Operon Technologies
www.operon.com

Packard Instrument Company
www.packardbioscience.com

Perlegen Sciences
www.perlegen.com

ProtoGene Laboratories, Inc.
www.protogene.com

Research Genetics, Inc.
www.resgen.com

Sigma-Aldrich Corporation
www.sigma-aldrich.com

Sigma-Genosys
www.sigma-aldrich.com

Silicon Genetics
www.sigenetics.com

Spotfire, Inc.
www.spotfire.com

Stanford University
www.stanford.edu

Stratagene
www.stratagene.com

Texas Instruments
www.texasinstruments.com

Zymark Corporation
www.zymark.com
Note: Readers can find out more about the companies and organizations listed by accessing their sites on the World Wide Web (WWW). If the listed organization does not have a site on the WWW or if it is under construction, we have substituted its main telephone number. Every effort has been made to ensure the accuracy of this information. The companies and organizations in this article were selected at random. Their inclusion in this article does not indicate endorsement by either AAAS or Science nor is it meant to imply that their products or services are superior to those of other companies.
Part 2 of this report will appear in the 19 October 2001 issue of Science. It will further explore advances in DNA chips and microarray technology.

This article was published as a special advertising supplement in the 4 May issue of Science