In the late 1970s, the advent of DNA recombination techniques set the stage for a revolution in pharmaceuticals: the birth of the biotechnology industry.

Unknown startups like Amgen, Genentech, Genzyme and Biogen scrambled to harness the natural machinery for weaving proteins within each living cell. With the support of "big pharma," the giant pharmaceutical companies of the past, these startups rolled out scores of blockbuster drugs that addressed diseases like hepatitis and cancer in entirely novel ways.

Today, a new generation of startups is taking advantage of a new revolution--the application of cheap computing power and rapid automation to the drug discovery process.

And like the discoveries that paved the way for the first wave of biotechnology entrepreneurs, today's new techniques of drug discovery represent a profound--and unavoidable--shift in the growth of the global pharmaceuticals industry.

Structurally Sound

Two Cambridge firms, Millennium Pharmaceuticals and Vertex Pharmaceuticals, have been at the forefront of the second technological revolution.

Vertex, founded by a Harvard graduate in 1989, is a pioneer in "structure-based drug design." In theory, the idea is plausible enough: if one knows the precise three-dimensional structure of an enzyme target, one can use a computer to design a perfectly complementary "small molecule drug" to jam into the enzyme and disable it, just as a locksmith can build a key from scratch to open a lock.

Vertex's first major research effort--designing a safe and reliable inhibitor for the HIV protease that assists with the replication of the HIV virus--started with generating a molecular map of the protease enzyme.

The techniques of choice for this sort of cartography are x-ray crystallography and protein nuclear magnetic resonance spectroscopy (NMR), which take a snapshot of a protein at a resolution of less than a billionth of a meter.

"X-ray crystallography allows us to map out the atom-by-atom three-dimensional structure of a drug target, [especially] the active site," says Brum. "The more precise information we can have about the active site helps us to develop a safer, more potent drug."

Vertex scientists solved the structure in 1992 (see picture below) and began to explore several potential sites for artificial blockage, using supercomputers to suggest possible "keys" for the HIV "lock." In practice, the computer's guess is often far from perfect, but computers can help bench-top scientists refine their search for the molecule with the perfect fit.

"We do a lot in the area of computational chemistry," explains Lynne H. Brum, Vertex's vice president of corporate communications. "It helps us understand which chemical compounds [will be successful] on the bench."

Vertex scientists attacked the HIV protease with different drug candidates, using combinatorial chemistry to produce thousands of subtle variations of the computer's best guess. Combinatorial chemistry, explains Harvard Professor of Chemistry Gregory L. Verdine, "is a technology that allows one to make thousands to millions of organic compounds of defined structures in a very rapid period of time."

without combinatorial chemistry, a chemist might spend years synthesizing a few thousand variants of a drug candidate. Using the technology, a single researcher can brew the same number of molecules in minutes.

When a molecule showed promise, it was plugged back into the computer for another round of computer guessing and filtering.