By Aaron J. Sender
We’ve got universal remotes and universal keys. So why not a universal microarray?
That’s what Caltech physicist Stephen Quake and his PhD student Michael van Dam wondered: Would it be possible to make a single array that would work for the study of any organism — from yeast to human? It turns out, in theory at least, that the answer is yes.
The quest for the universal array began about two years ago when Quake’s lab, having developed what looked like a very promising microfluidics technology for combinatorial chemical synthesis, was looking for a practical way to test it. They chose DNA molecules, “in part because we thought it would be a nice test because the chemistry is very well worked out. And also in part because we thought it might be useful for gene expression analysis,” says Quake.
They began looking into what sequences to put on the array, and it hit them. “We could put everything on the array,” says Quake. “We realized that the microfluidics would allow us in a very easy way to make all possible combinations of n-mers, as opposed to designing arbitrary sequences.” The idea of putting every oligo representing any gene ever found, or to be found, was intriguing. But was it possible?
“We were kind of skeptical whether it would actually work or not at first,” says van Dam. To be specific enough to distinguish between any possible transcript, they thought, the oligos would have to be longer than was practical to put on a chip. The only way to know for sure, though, was to do the math.
They set out to calculate the number of nucleotides in an oligo necessary for every gene to be uniquely represented and compare the results with known genes in various databases. The answer turned out to be somewhere between 10 and 16, depending on the size of the organism. “That’s very doable,” says van Dam.
But that means a chip with the minimum 10-base-long oligos would require about a million spots to be truly universal. At 13 bases the number jumps to 64 million spots.
Although today these numbers are unthinkable, van Dam insists that as the micofluidic technologies the Quake lab is developing get better, squeezing millions of spots on a chip won’t be a stretch. “Currently we have 100-micron channels,” he says. “We hope to get that down to about one micron.”
The Caltech physicists say that their universal key to unlock the gene expression patterns of any organism offers several benefits over other arrays. First, current arrays are spotted with specially designed probes meant to represent specific genes. “Mistakes are inevitable,” says Quake. “If you make a mistake in setting up your array — a computer error, you mislabel your cDNA library — it becomes the property of the array and you can’t really do anything without the very expensive process of making more arrays and redoing the experiments.”
With the universal array, however, because all possible sequences are already on the array, they never have to be redesigned and remanufactured when errors or new genes are discovered. “You just rerun the data analysis,” says Quake.
Their process of synthesizing the oligos is also more efficient than Affymetrix’s photolithography method, they say. Quake and van Dam use standard DNA synthesis chemistry, which has more than 99 percent efficiency in stringing bases together. “On an Affymetrix chip, you end up having all these sequences which are shorter than what they should be,” says van Dam.
But it’ll be some time yet before you can use the one-chip-for-all to explore your favorite organism. The Quake lab is still working on the prototype. The researchers are trying to find the right chemistry for the surface of the arrays as well as the perfect combination of solvents to go with the microfluidic channels. But they are optimistic that these hurdles will soon be worked out. “We hope that it will have some commercial value in the relatively near future,” says van Dam.