As microarray researchers know, relativity is a much less sexy concept in gene expression experiments than in theoretical physics. The ultimate aim would be to quantitate absolute levels of expressed transcripts. Until now, however, determination of absolute expression levels has been left to expensive bead-based technologies such as Lynx’s Megasort system. But Aimee Dudley, a postdoctoral fellow, and John Aach, a senior scientist from George Church’s Harvard Medical School lab, have developed a method for absolute quantitation of expression in yeast microarrays.
The traditional design for a microarray experiment involves hybridizing two-color labeled experimental and control samples to arrays, then examining relative expression levels between the two and adjusting for dye bias and other factors. But in this typical experiment, the absolute expression levels of transcripts in the control sample are unknown, so the final measurements of expression are just relative.
To get at this absolute quantitation, Dudley, Aach, and colleagues decided to replace the control sample with oligonucleotides that would be complementary to the yeast probe oligos on the array, ordering them from Qiagen Operon in small quantities. They divided the oligos into two samples, one of which was a five-fold dilution compared to the other; and labeled one sample with Cy3 and the other with Cy5, then hybridized them to the yeast array. In theory, their results would show a five-fold difference in expression levels, and given the known baseline concentration, they would be able to discern the expression level of an unknown sample. This turned out to be the case.
“This [technique] is directly applicable to comparing abundances between genes,” said Dudley. In other words, by comparing the spot intensity for experimental sample (unknown) to that of the pre-measured oligos (known), they could derive the absolute number of transcripts of a particular gene expressed in the experimental sample.
The problem with this experiment was that the group was unable to order the 5,000-plus unique complementary oligos they would need to make up an entire reference sample for a microarray: Operon had not yet designed such a set.
So, to see if a similar concept would work on an entire array, they used defined quantities of complementary “universal” oligos that matched the universal primer tails of PCR products that had been spotted down on an array. They labeled the primer oligos with Cy3 and the RNA with Cy5, then were able to compare the expression level of the RNA to that of the primers.
This comparison of a known quantity of primers to an unknown quantity of transcripts not only allowed the researchers to quantify absolute expression levels, it also made it “easier to compare between experiments than if you use an unknown [quantity] of cDNA as your control sample,” Dudley said.
This method additionally allows researchers to easily detect cross-hybridizations, or situations where oligos did not hybridize very well. These phenomena would be signalled by expression ratios that deviated from the five-fold dilution ratio in the first experiment.
The group has paired this new array system with “masliner,” a program designed by Aach that extends the range of intensities that can be detected on an array, allowing an array user to use smaller amounts of oligos in each hybridization.
The group plans to publish a paper on this system in an upcoming issue of the Proceedings of the National Academy of Sciences, and meanwhile has found “a lot of enthusiasm” for it among other researchers, Dudley said.
“We think that companies like Operon will start producing mixtures of oligos complementary to their spotting oligo sets, once they become more familiar with this work and the interest of the microarray community in such a product,” Dudley said.