By David N. Leff

Ever since the O.J. Simpson trial, forensic DNA testing has stayed in the news as biotech's sure-fire tool for nailing the guilty and freeing the innocent.

This molecular approach works not just to identify murderers, rapists and deadbeat dads. It also catches perpetrators that are themselves molecules rather that people. Most recently, it has captured a war criminal in the war against Alzheimer's disease (AD).

The culprit's name is BACE - short for "beta-site APP-cleaving enzyme" - and its rap sheet appears in today's issue of Science, dated Oct. 22, 1999, under the title "b-secretase cleavage of Alzheimer's amyloid precursor protein by the transmembrane aspartic protease BACE."

The paper's senior author is molecular biologist Martin Citron, who leads the Alzheimer's disease research program at Amgen Inc. in Thousand Oaks, Calif. His name heads a list of 24 co-authors, all but one of them Amgen scientists.

Almost everyone knows that AD's mark of Cain is a slew of senile neuritic plaques wrapped around slowly dying neurons in the brain. They congregate especially in the hippocampal region, which plays a crucial hand in memory and cognition. Those plaques consist largely of amyloid-beta peptide - made mainly of starch-like proteinaceous fibrils. Before those peptides can get in their neurotoxic licks in an AD patient's brain, they have to cut loose from their parent sequence, amyloid precursor peptide (APP).

The enzymes that do that cutting are prime therapeutic targets for inhibitors that could slow, or perhaps stop, the inexorable, downhill course of the disease. "So companies and academics worldwide," Citron told BioWorld Today, "have for the last couple of years been looking for these enzymes. But they have been very difficult to isolate - and now we've got one of the two.

"There are actually three enzymes in the AD game," he explained, "alpha, beta and gamma secretases. Alpha-secretase cleaves in the middle of the Ab-peptide region, and therefore blocks production of the A-beta peptide. "Beta and gamma secretases," Citron went on, "are the enzymes that cut the Ab peptide free from its APP precursor, which lets it produce that plaque-forming peptide. So to sum up," he added, "alpha secretase is the good guy, and beta and gamma secretases are the bad guys."

A Walk Through Expression Cloning

Citron described how the Amgen team succeeded in isolating the elusive bad-guy beta-secretase, as a first step in finding a drug to block its brain-trashing mayhem. "Our principal idea," he recounted, "was not to go the usual biochemical-purification route, in which you grind up cells or tissue and then search for the enzyme at the protein level. Rather we chose to go for the secretase gene and expression-cloning strategy. So we started with human cells that express the amyloid precursor protein, and have all the mechanisms in them to make beta peptide.

"We took those expressed genes from human kidney cells," Citron went on. "That might seem somewhat weird, because AD is a brain disease. But it was shown years ago that in principle all cells in the body can make beta-peptide, so all cells in the body should have both beta and gamma secretase. Therefore we felt that we could use the peripheral kidney cells, which are just technically easier to handle than brain cells.

"Then we put all kinds of unknown genes into those cells, and assembled them in 8,600 groups of 100 genes each. We looked at which groups caused increased beta-peptide production from the cells, and subdivided them into smaller groups, until we had narrowed the effect down to a single gene."

From that unique sequence the co-authors cloned BACE, their APP-cleaving enzyme. To prove that it worked, they could thank a now-celebrated Swedish family, in which many members acquire a severe form of early-onset AD, due to a mutation in their APP genes - discovered in 1992.

"What those people have," Citron pointed out, "is an amino-acid exchange right near the APP cleavage site of beta-secretase. This makes their APP a better substrate, so they make a lot more amyloid-beta. That's why the Swedish mutation is widely used in academic labs and in industry, just because it yields more Ab per APP molecule, and it increases the sensitivity of the method.

"So what we requested of our new enzyme is that it cleave three mutant APP forms with the same specificity as the native beta-secretase does. The Swedish mutant did it better than wild type, and wild type proved better than another rare mutant that has been previously shown not to be cleaved well by beta-secretase."

Next, Therapeutic Payoff: Enzyme Inhibitors

Amgen is now embarked on the next logical step - going after an orally available, small-molecule inhibitor of their bad-guy BACE enzyme. "Our strategy," Citron said, "is similar to inhibiting approaches that have been used to block other proteases in the past - such as the anti-HIV protease blocker. Essentially one would take the purified protein, incubate it with a purified substrate, and then screen thousands of compounds for their ability to inhibit this in vitro cleavage reaction.

"Once we've identified such compounds," he continued, "we would test their efficiency in tissue-culture cell models. If they are effective there, then we would go into transgenic animal models of amyloid formation and test the inhibitors preclinically."

One of the company's next steps, Citron allowed, will be to undertake such in vivo preclinical testing. At some future date, he foresees human clinical trials. "A trial would involve many AD patients, probably at an early stage of the disease," he observed. "And the read-out would call for cognitive markers. So you take two groups. The control cohort, untreated with inhibitor, would deteriorate in the course of the study; the other would be treated. Then if things work, they won't deteriorate, or do so at a slower rate. These trials," Citron concluded, "would take a long time. I can't say how many years."