June 1, 2007 (Vol. 27, No. 11)
Microarray Market Is Growing as Applications for Diagnostics and Research Are Realized
Although DNA microarrays are just beginning to be investigated for diagnostics, they are becoming key tools for preclinical and clinical research. Gene-expression analysis continues to provide information on what is occurring on a molecular level and is crucial to discovering new drug candidates. According to a Frost and Sullivan report, “U.S. DNA Microarray Markets,” the total U.S. revenue was $446 million in 2005, expected to increase to $532 million by 2012.
A number of researchers provided up-to-date reports on DNA microarrays at Select Biosciences’ “Advances in Microarray Technology” conference held last month in Edinburgh.
One of the biggest issues with microarrays has been high cost and difficulties in making them. Febit Biotech’s (www.febit.de) Geniom® platform designs, synthesizes, applies, and analyzes microarrays. Customers can use company-provided designs or design their own. “We want to overcome the hurdle of so few microarrays and so many genes. Our instrument allows people to make and test as many microarrays as there are genes and organisms,” said Peer Staehler, CSO.
The platform utilizes microfluidic biochips that enable several assays: dynamic detection of hybridization and amplification, PCR-on-Chip, longmer synthesis, and signal-enhancing protocols. Most microarrays are analyzed via a confocal scanner with one image. However, dynamic detection offers the ability to take images at any time during the experiment and enables a melting-curve profile. The microfluidic system prevents the liquid from evaporating.
“This gives you a binding profile on the whole chip, on thousands of features, which corresponds to a melting-curve plot. Instead of creating one layer of information, you create 20 or 30 data sets,” added Staehler. This provides benefits such as in SNP-typing, distinguishing a mutant from a wild-type, and for optimization of oligos, quickly identify their binding behavior.
Another feature, Enzyme-on-Chip, enables reverse synthesis of DNA oligos in the biochip. This provides new applications like PCR-on-Chip or primer extension reactions. “We try to fuse the advantage of the microarray being a multiplex analysis with the high data quality of the PCR reaction,” explained Staehler.
More researchers are investigating miRNAs, which appear to be responsible for regulating gene expression. Kreatech Biotechnology (www/kreatech.com) has developed Universal Linkage System (ULS™) to label small RNAs. ULS contains a platinum complex that forms a bond with the N7 position of guanine nucleotide base. There are currently two kits available, the ULS Small RNA labeling kit and the miRACULS miRNA kit.
“When we developed the first two kits, we found out that customers wanted kits for smaller RNA amounts,” said Erik Jan Klok, Ph.D., product development manager, microarrays. “These can handle about one microgram of small RNA, but the new kit can handle about 300 nanograms. It is critical to use less material.”
The ULS method directly labels nucleotides independent of their size. This differs from enzymatic methods, which use linkers or adapters to label the small amount of molecules. As it uses a chemical to label, it is a 15–30 minute reaction, followed by a spin column (KREApure™) to remove dye molecules that haven’t reacted. “Since there are only two components in the reaction, there is easy control over the degree of labeling,” added Dr. Klok. In addition, there isn’t any modification of the original miRNA used for hybridization, which avoids any bias. .
There have been many factors preventing the use of diagnostic microarrays in the clinic. Bertrand Jordan, Ph.D., a consultant for the Marseille-Genopole project, discussed why it is not catching on like many thought it would. Originally there was a lot of discussion about expression profiles predictive or prognostic of various types of cancer, Dr. Jordan said. However, the use of microarrays to identify mutations and copy-number variation is an easier application.
“With microarrays, you can look at many mutations simultaneously and see whether they are present. Usually the clinical importance of the mutation is already established,” Dr. Jordan added. Many FDA-approved microarray diagnostics are of this type, like the Affymetrix(www.affymetrix.com) AmpliChip® CYP450 test, which indicates whether a patient has a version of a given gene.
Arrays that look at expression profiles are more technically difficult because one is looking at hundreds of genes and trying to measure the expression of each one in a tumor sample over a wide dynamic range.
“There is the issue of what the profile actually means and how well you have proven that a profile correlates with a good prognosis or a bad prognosis,” said Dr. Jordan. He explained that there have been many findings not substantiated in further studies because of many possible pit falls. “If the correlation between a profile and prognostic information is good and scientifically valid, it is not necessarily clinically useful. If it is not used to help make clinical decisions, there is no point in paying for expensive tests.”
Streamlined Genomic Labeling
Invitrogen (www.invitrogen.com) recently launched the BioPrime® Total Genomic DNA Labeling System for use in array comparative genomic hybridization applications. This new kit adds dye-labeled nucleotides directly to the synthesis steps, so fluorescence is directly added during the labeling process. “It is a quicker and easier protocol—you go directly out of your enzymatic step and right into your array, or whatever downstream application you have,” explained Steven Roman, Ph.D., senior scientist, R&D. “Streamlining the protocol means you get better yields.”
The kit is optimized for smaller sample size and can accommodate 50 nanograms of genomic DNA. “That is a trend in all assays and systems in biology, going to smaller amounts of input,” added Dr. Roman. “Our kit does some level of amplification that allows a significant output of DNA. The main message is being able to go down in input without having your results suffer.” A master mix of each dye mix is provided for labeling control and test DNA, eliminating extra pipetting steps and stabilizing the components. In addition, using the dyes that are supplied by molecular probes lowers the cost per reaction between 30 and 40%.
Dr. Roman presented data using the kit on two different comparative genomic hybridization platforms: BAC arrays and oligo-based arrays. The challenge with oligo arrays is getting enough signal because the probes on the array are short and less material is hybridized to them in the labeled product. “You have to have good, even labeling and faithful replication of DNA during the labeling process to get good signal and results. Our results on both platforms showed we could go well below the manufacturer’s minimum recommended input amount.”
Data analysis continues to improve with the launch of new bioinformatics tools. Rosetta Biosoftware (www.rosettabio.com) now offers the Rosetta Resolver® gene-expression data analysis system 7.0 with additional features for data storage, exchange, and collaboration. It also provides flexibility to store gene expression data from Agilent, GE Healthcare, Affymetrix, Illumina, as well as TaqMan® or quantitative real-time PCR data. It can also bring in accessory data like drug levels, phenotype, and clinical chemistry data to support toxicology and safety studies.
Technology-specific error models enable one to identify differences in expression even if the number of replicates is small. Other features include cross-species and/or cross-platform analysis and data exchange between internal/external collaborators. An optional component is the qPCR analysis module that imports, stores, and analyzes quantitative real-time qPCR data.
“Often I think the problem is a disconnect between a biologist’s view of the world and the statistical requirements of analysis,” said Tim Bonnert, Ph.D., principal scientist at Rosetta. “I think Resolver brings its flexible nature to analyze, store, and interpret/visualize data.” Dr. Bonnert said the future of microarrays will require the ability to integrate data from outside the original microarray study itself, along with greater understanding of the biology and biological interpretation.
Microarrays are a powerful technique, but the way they are being implemented widely has removed their usefulness, said Nigel Saunders, Ph.D., departmental reader in microbiology, University of Oxford, U.K. He added that the problem is that many people accept them as being technically noisy, challenging tools, and that one needs to go back and sort through the methods so they are reliable. “You can actually get reasonably quantitative data out of your microarrays—most people think you can’t,” Dr. Saunders explained.
It is key to have the right probes and to print them sufficiently and accurately. Dr. Saunders’ group uses dendrimer labeling provided by Genisphere (www.genisphere.com), which enables them to measure more genes and provides more quantitative data. He discussed how to pull single-channel data off dual-channel comparative microarrays and how to analyze it as single-channel data. This is only achieved if the data is sufficiently reproducible and quantitative.
“Microarray analysis is a problem in many settings because people don’t actually distinguish whether they have actually measured anything for a gene or not. The key is to know when you have collected data and only to analyze data that you have collected,” added Dr. Saunders. Another big issue is validation. “There is no such thing as a perfect microarray probe.” What is required is the actual information on the cross-hybridization potentials of each probe. “Validation should be targeted for the genes where you have genuine reasons to believe the data may be coming from somewhere else.”
He concluded that one needs good quality chips, high-sensitivity, reasonably quantitative labeling, correct experimental design, and to only analyze data that has actually been collected.