Science & Technology

Biochips Ahoy!

|

"This is the year we're going to see improvements in health care as a result of Affymetrix chips," declared Mark Hurt, a senior field applications specialist for the biotech company, at a recent seminar at the University of Virginia Medical Center. The seminar was so popular that it had to be moved to a much larger auditorium -- and even then, some of the overflow audience of doctors, researchers and med students had to sit in the aisles. The excitement was over the research and medical treatment possibilities offered by Affymetrix's biochips.

Founded in the early 1990s, California-based Affymetrix has developed system for quickly analyzing the genetic codes of human beings and many other model organisms, including mice, fruit flies, and yeast. Affymetrix's technique combines the photolithography of semiconductor manufacturers with biochemistry to create gene chips. Each gene chip consists of an array of tiny wells measuring 20 microns square on a 1.3 centimeter silicon chip. In each well are DNA probes that can detect the presence of a specific gene. Because the chips are created using up to 75 different layers, the "complexity of making a gene chip is equal to that used in the semiconductor industry," said Hurt. Affymetrix no longer has the biochip field to itself. Corning, for example, has worked out a deal to develop biochips with the Whitehead Institute in Cambridge, Massachusetts, one of the leading public human genome sequencing centers.

Biochips will be an essential technology for deciphering the human genome, i.e., the entire genetic sequence of all human DNA soon to be published by scientists at Celera Genomics and the international Human Genome Organization. The human genome is a string of some 3 billion As, Cs, Gs and Ts, the chemical building blocks of the genetic code. This sequence is being made available to researchers through both public and private databases. Determining what the sequences mean for human health and development will likely take decades to puzzle out.

What role will biochips play? Right now biochips are being used by researchers to find out which genes are expressed in which tissues. Nearly every human cell contains the entire genome, but to create the various tissues that make up the human body, different genes have to be turned on or off. In any given cell -- a liver cell, say, or a skin cell -- perhaps only 15 percent of its genes are active. Biochips can tell researchers what those genes are. They can find out the differences between healthy and diseased tissues and thus suggest different therapeutic targets. For example, Affymetrix has a chip that can be used to distinguished between cancers based on whether the p53 tumor-suppressor gene has been inactivated or not. This is important because more aggressive treatment regimens are called for if the p53 gene is not functioning. At the UVA seminar, Dr. Meng Chen described how her lab used biochips to determine which genes the new drug lisofylline turns off and on. That knowledge will enable researchers to figure out how to protect insulin-protecting pancreatic cells from damage that could lead to diabetes.

Currently, Affymetrix has a biochip that can detect the presence of 12,500 different human genes. "It is our goal as a company to bring you the entire human genome on a single chip array," Hurt told the audience. "Probably not in 2001, but maybe next year." Affymetrix is also working with the pharmaceutical giant Glaxo Wellcome to create a biochip that will distinguish between 100,000 different single-nucleotide polymorphisms, (SNPs, pronounced "snips"). SNPs are what account for human genetic variation ranging from hair and eye color to propensities to diabetes and heart disease. Ten leading pharmaceutical and information technology companies donated $50 million dollars to the public genome project to create the SNPs Consortium. So far 2.5 million SNPs have been discovered. Francis Collins, Director of the National Human Genome Research Institute estimates that out of some 10 million SNPs, about 200,000 will turn out to be functionally important.

Biochips will enable researchers to compare whole constellations of SNPs between people to discover how this or that genetic variation correlates with this or that disease outcome. For example one constellation of SNPs might dispose those who have it to early-onset dementia, while another SNPs pattern might indicate a higher risk of kidney failure. Every human being probably has 40 to 50 genetic "glitches" predisposing them to various diseases, according to Collins.

Pharmaceutical companies are betting that SNPs will hold the key to explaining why specific drugs cause severe side effects in some people while offering tremendous benefits to others. If researchers can distinguish between those patients a medicine will harm vs. those it will help, many therapies that are currently shelved because their side effects cannot be predicted in specific patients could be brought safely to market. SNPs research also raises socially and ethically problematic issues, such as uncovering possible genetic correlations for alcoholism, violence, and intelligence.

Biochips are not cheap, though the price is falling rapidly. A year ago, human biochips cost $2,000 per unit. Currently human biochips cost $1,000, while chips for mice, yeast, and fruit flies cost around $400 to $500. The price for human biochips will probably drop to $500 this year. Once all the human genes are well characterized and all functional human SNPs are known, manufacture of the chips could conceivably be standardized. Then, prices for biochips, like the prices for computer memory chips, would fall through the floor.

Your doctor might one day routinely put a bit of your DNA on a biochip and pop it into an office scanner to perform a full genetic scan as part of your regular physical. Or she might use biochips designed to detect disease organisms to precisely diagnose which kind of germ is causing your stomach flu. Biochips could thus finally usher in the era of medicine tailored to each individual patient.