Microarrays have been around since the early 1990s, but the necessary equipment to use the technology could set back a laboratory nearly $200,000. Then Patrick O. Brown of the Howard Hughes Medical Institute, and his colleagues at Stanford University, published their first paper in Science in 1995 (M. Schena, et al., see table, paper #1), using their home-grown, low-cost automated microarrays, and the revolution was launched. Brown's extraordinary impact on the field is aptly demonstrated by his citations: nearly two dozen papers with over 100 citations each, including the 1995 Science paper and its remarkable 1,000+ citations, and another Science paper, from 1997, with more than 800 (see table, paper #2).

Last year, Brown made this publication’s annual list of "hot" scientists by virtue of six highly cited papers published over the preceding two years (see Science Watch, 12(2):1-2, March-April 2001). Two papers by Brown and colleagues can currently be found in the Biology Top Ten on page 8 of this issue (paper #5 and #8).

Another of Brown’s 2001 accolades brought even wider recognition: his selection, in the field of genomics, as one of Time magazine’s handful of "America’s best in science and medicine."

Brown, 47, received his bachelor's in chemistry in 1976 from the University of Chicago, where he also earned both his M.D. and a Ph.D. degree in biochemistry, studying with Nicholas Cozzarelli. After a three-year residency in pediatrics at Children's Memorial Hospital in Chicago, Brown moved on to do a post-doc in microbiology and immunology at University of California, San Francisco, with Michael Bishop. In 1988, he became a Howard Hughes investigator and moved to Stanford University, where he is now a professor of biochemistry.

When you set out to construct your first microarrays nearly 10 years ago, what were your goals?

Brown: Our immediate goal at the time was to be able to genotype a complete genome in a single hybridization step. We had successfully developed a biochemical method to fractionate an entire genome based on genotype, but we needed a way to read out, at high resolution, what segments of the genetic map were represented in a DNA sample comparable in complexity to an entire genome. We needed a high-resolution, ordered physical representation of a genome that would perform well in such a hybridization, be robust and quantitative, and allow us to look at a very large number of sequences in this single hybridization. We also wanted something that would be really cheap and easy to use. That was fundamentally important, and it’s something that has yet to become enough of a guiding principle for the companies that are selling microarrays. If you can get the cost down low enough, you can use a microarray just like a microscope to get useful molecular information about what a genome is doing in practically any process or problem that you're studying in biology, and you can get very valuable, rich information. The simple technology, small scale, and the use of comparative fluorescence to make the performance quantitative and robust, were natural consequences of our design goals.

A few years ago, you and your students posted the MGuide on the web, a guide that anyone could use to make inexpensive microarrays. What was your motivation?

Brown: It's a funny thing. To me it seems that it is more the norm than the exception that scientists are glad to share protocols that are more detailed than what a journal would ever publish in a materials-and-methods section. In other words, any scientist in our situation would have willingly published the MGuide, but no journal would publish it. So it was practical and simple to post it on the web, and that made it more easily accessible. And everyone in the lab was all for it and eager to help make this easy for other people to do. In fact, Joe DeRisi, who was a student in my lab, now runs this fantastic course at Cold Spring Harbor to teach people everything they need to know to be self-sufficient in making and using microarrays. More than anyone else, he was the person who actually assembled the protocols for the MGuide and put this all together. So it was a natural thing to do.

How have microarrays changed the way you do science?

Brown: That sounds like a simple question, but it’s not. I'd say maybe two-thirds of the experiments done in my lab use microarrays in one way or another. In that sense, microarrays have changed everything about the way I do science. But that's kind of a trivial answer. So you have to contrast it with what we were doing before, which involved taking specific problems, developing appropriate ideas and tools, and incrementally working our way through the problems. What we're doing now is closer to fully embracing the potential of genomic approaches and, wherever possible, trying to do experiments in a systematic way with the deliberate aim of taking what we learn about the specific system or process we’re studying and embedding that knowledge in a more comprehensive framework.

Let me give you an example. We have a project now where the very specific focus is trying to dissect why it is that when you deliver an antigen-specific signal to a T cell and at the same time co-stimulate it via another receptor, you’ll get a fundamentally different physiological response than when you engage the T-cell receptor by itself. This is a specific interesting example of a huge class of questions about the language that cells use to communicate with one another and to orchestrate their physiological responses between each other. Well, the point is that we do this experiment, which involves looking at literally hundreds of samples of cells and a whole bunch of different, very carefully orchestrated in vitro physiological time courses. So we're systematically approaching the problem. It’s just one of what will ultimately be thousands or millions of experiments of that kind. But even though it’s a body of work in its own right, we're conceiving of it as part of a systematic picture that aims to understand the bigger picture of the genome-expression program and how genome and environment interact to send out specific sets of instructions to specific cells. And as we build up this systematic picture, our ability to make sense both of the regulatory logic and the physiology of this specific process gets better, and conversely, the information from this single study provides part of the framework that helps us to understand what we see in other gene-expression programs.
continued