A little bit of automation could take some of the mess and guessing out of microarraying, according to some participants at Lab Automation 2003 in Palm Springs, Calif., last week.
Michael McNeely, the CTO of Salt Lake City-based BioMicro, began his talk with the proclamation that inconsistent results resulting from manual microarraying have “plagued the microarray industry.”
McNeely went on — naturally — to promote his company’s solution to this problem: an automated hybridization station. The system, the microarray user interface, or MAUI, consists of four “bays” for microarray slides on a block. The slides are covered with the company’s patented air bladders, plastic chambers that contain the sample, which are attached to tubes that use air to pump the sample across the surface of the arrays.
The idea behind this pumping of sample is to enable the user to have “exquisite control of fluid,” according to McNeely, so that the sample will be evenly spread out on the array, and the results of the experiment will be more consistent. Researchers at UCSF who compared MAUI to a “control” manual hybridization found that its use resulted in a threefold increase in sensitivity of the arrays, said McNeely.
Currently, he said, the MAUI is optimized to hybridization at a temperature of 65 degrees Fahrenheit, although the company is planning to optimize it for a temperature of 72 degrees within a month.
BioMicro, founded in 1997, has placed 10 systems, which sell for $15,000 each, and has five paying customers. Additionally, it sells the disposable plastic bladders for $29 each.
A second-generation instrument that the company soon plans to launch includes multiple bladders per chip, for second-stage microarray slides that have a low number of probes repeated multiple times on the chip. “The trend we’re seeing,” said McNeely, is that users “are moving away from whole genome analysis to target gene analysis.”
For the third generation, BioMicro is ambitiously planning to combine microfluidics with sample processing, allowing users to convert mRNA to cDNA, aliquot the product, add the buffers, and dilute the sample along a concentration gradient.
BioMicro is still looking for what McNeely called “qualified validation partners” for the MAUI system.
The 10-person company recently hired Jim Kuo as its new CEO.
Testing The Differences
A couple of other presentations at the tail end of the conference focused on clinical applications of microarrays. Although these talks were only tangentially related to automation, they provided some useful insights.
Dominic Sinicropi, of Genomic Health, discussed the Redwood City, Calif., clinical genomics company’s comparison of TaqMan and Affymetrix arrays. Company scientists wanted to see whether they could use microarrays to perform quantitative gene expression analysis — rather than relative expression analysis. So they designed a study to compare Affymetrix chips to Applied Biosystems’ TaqMan assay, which has been traditionally used for quantitative gene expression analysis.
In the first experiment, the scientists used for the TaqMan assay a gene-specific panel of 80 oligo primers that were synthesized to be complementary to target sequences on the Affymetrix Hu 95A chip. They performed this assay on Jurkat T cell RNA, then also hybridized a portion of this same sample to the Affymetrix chips. Using this single sample, they found that the “expression profiles determined by these two technologies actually didn’t correlate very well,” said Sinicropi. But, in another experiment comparing a sample of Jurkat cells to a cancer sample of pooled patient breast tumors, they found that the TaqMan and Affymetrix results did track each other well, while TaqMan showed a dynamic range that was twice that of Affymetrix. In this second study, they used a single round of sample amplification. But when the group repeated this experiment with two rounds of amplification, they found that the results correlated less well, Sinicropi reported.
Despite some of the discrepancies using multiple amplifications and single samples, however, Sinicropi said the scientists concluded that “microarrays can be used for quantitative correlation of expression profiles.”
In another comparative microarray study, Mark Erlander of Arcturus, a Mountain View, Calif., microgenomics company, discussed a comparison of two different types of tissue samples: formalin fixed vs. frozen biopsies.
The comparison came up during a collaboration between Arcturus and Dennis Sgroi, a pathologist at Massachusetts. General Hospital and assistant professor at Harvard Medical School. Sgroi has been using the company’s laser capture microdissection instruments to grab epithelial cells from different types of cancerous breast tissue and obtain gene expression profiles for them. Along the way, he and the Arcturus scientists tried to resolve the issue of whether fresh frozen breast biopsy samples were required for future diagnostic profiling efforts, or whether samples of tissues that were fixed in formalin and embedded in paraffin (FFPE) would work well enough.
While the group expected that the RNA in the FFPE samples would degrade too soon for robust expression profiling, it found that RNA remains intact in 10 percent formalin.
The group compared samples of 50 ng, one frozen and one FFPE, from the same basic tissue, by linear amplification and hybridization to the same 12,000-probe, 60-mer oligo chips, and the results correlated well. Even after the FFPE sample was eight days old, there was still an 80 percent correlation.
The Arcturus-MGH group hopes that this work will enable pathologists to combine standard pathological immunohistochemistry-based analysis of cancer cells with RNA analysis, using the same samples and, of course, the laser capture microdissection instrument.