The aggregate-forming amyloid beta protein isn’t the only factor in Alzheimer’s, but by now, researchers have little doubt that it plays a major role in the disease. Genetic mutations that increase its production or make it more likely to form clumps account for virtually all the early-onset forms of Alzheimer’s. Factors that directly affect amyloid beta’s ability to accumulate in the brain also contribute to more common late-onset cases. In one of the most striking examples, scientists in Iceland reported last summer [see “Gene Mutation Strongly Protects Against Alzheimer’s”] that carriers of a gene mutation that roughly halves amyloid beta production live longer, experience a markedly slower cognitive decline in old age, and have a very low risk of Alzheimer’s, compared with non-carriers.
Some Alzheimer’s researchers have suggested mimicking this genetic effect with drugs that dial down amyloid beta production. In principle, if such drugs were safe enough, even healthy middle-aged people could take them daily to ward off dementia—just as tens of millions around the world now take “statin” drugs to reduce cholesterol production and thereby prevent heart attacks and strokes.
Yet despite an abundance of scientific discoveries about Alzheimer’s over the past two decades, scientists have had great difficulty in developing drugs that safely lower the production of amyloid beta in the brain. Only recently have they begun to hit this elusive target in a way that could plausibly be used for long-term dosing.
“The challenge has been to develop a drug that will exploit the opportunity without bringing any toxicity along with it,” says Sam Gandy, director of Alzheimer’s research at Mount Sinai School of Medicine, and author of a recent essay on the subject in the New England Journal of Medicine.
The rise and fall of gamma secretase inhibitors
The first major strategy for lowering amyloid beta production aimed at gamma secretase, an enzyme complex that helps cleave amyloid beta away from its mother protein, APP. In test cells and in animal models, chemical compounds that inhibited gamma secretase’s activity strongly reduced amyloid beta production and accumulation. But even in these early, preclinical tests, it was clear that gamma secretase inhibitors (GSIs) were likely to have unwanted side effects.
Like most enzymes, gamma secretase performs multiple functions in the body. One of these involves the so-called Notch signaling pathway, which is important for the normal working of neurons—so that its inhibition could end up doing more harm than good to the brain. (GSIs are now being investigated for use as cancer chemotherapies, since many cancers reply heavily on Notch signaling.) There also have been hints from preclinical studies that GSIs can cause a buildup of APP fragments that are themselves toxic to neurons.
A more definitive picture of unwanted GSI effects came from clinical trials, beginning with trials of Eli Lilly & Co.’s semagacestat. These trials were terminated early after an interim analysis showed that semagacestat-treated Alzheimer’s patients were declining faster cognitively, with worse side-effects—including gastrointestinal effects and skin cancers—than patients who got only placebos. Bristol Myers Squibb’s avagacestat did not fare much better, even though it was designed specifically to inhibit gamma secretase’s cleavage of APP while having a much weaker impact on the Notch pathway.
“Several companies have given up” on GSIs, says Bart de Strooper, a researcher at the University of Leuven in Belgium who helped to discover the gamma secretase complex in the late 1990s. He notes, however, that the GSI concept hasn’t yet been tested as thoroughly as it could be. For example, researchers now know that there are several distinct gamma secretase enzyme complexes, each with its own set of biological functions; the one for cleaving APP may be different from the one for activating the Notch pathway. “There is need for further work to clarify how [these] different enzymes bind their different substrates, and also whether it is possible to target a subclass of the enzymes, which probably would provide more safe roads towards drugs,” he says.
Beta secretase inhibitors – still in play
The other enzyme that cleaves amyloid beta into being is called beta secretase, or BACE-1 (for Beta-site APP-Cleaving Enzyme). It makes the first cut on APP, while gamma secretase makes the second. An early strain of “knockout” mice, bred without the BACE-1 gene, seemed quite healthy, which suggested that a BACE-1 inhibitor drug would not have major adverse side effects.
However, developing an effective inhibitor turned out to be a severe challenge, because of the conflicting characteristics needed by the inhibitor molecule. For example, it had to be big enough to block BACE-1’s relatively large active site, but also biochemically nimble enough to get through the blood-brain barrier as well as neuronal membranes to reach BACE-1.
In recent years, pharmaceutical companies have finally succeeded in making BACE-1 inhibitors that can satisfy these criteria and get to neurons. In addition to performing well in mouse models, these candidate drugs have reduced amyloid beta levels in cerebrospinal fluid in human safety trials, and have been well tolerated at amyloid-beta-lowering doses. Several of these drugs—including compounds from pharma giants Eli Lilly, Eisai, and Merck—are now in clinical trials. Genentech also has made a monoclonal antibody that binds to BACE-1 with high specificity, but as a biologic product delivered by infusion—not a simple pill—it probably would be too expensive for long-term use as a preventive.
There are still some doubts that BACE-1 is the best target for an amyloid beta-lowering drug. Like gamma secretase, BACE-1 works on many other molecules in the brain and body, whose activity might also be affected significantly by a BACE-1 inhibitor. These BACE-1-cleaved molecules include neuregulin-1, an important developmental signaling protein, adhesion molecules that guide neuronal connections, vascular proteins in the retina, and the myelin protein that sheaths nerve fibers. Moreover, a strain of BACE-1 knockout mice reported in 2008 had defects that were reminiscent of human schizophrenia.
“The rationale for [targeting] BACE1 is strong,” notes De Strooper, but “only further clinical work will teach us the therapeutic window in which we can maneuver.”
Modulators and regulators
In principle, a more selective strategy would be to block the specific sites on APP where beta or gamma secretase make their cuts to release amyloid beta, thus impeding or “modulating” these enzymes’ amyloid-making activities. The protective APP mutation in Icelanders appears to do just this, for its effect is to change a single amino acid at a location adjacent to the BACE-1 cleavage site.
Many academic and commercial laboratories are now trying to develop such APP-cleavage modulators, as safer alternatives to direct enzyme inhibitors. One example is EnVivo Pharmaceuticals’ EVP-0015962, a small-molecule drug that apparently binds to APP near the gamma secretase cleavage region. With the drug molecule in place, gamma secretase becomes more likely to cut APP at a position that yields a shorter, less aggregation-prone form of amyloid beta.
Researchers also are thinking about targeting separate proteins that somehow regulate gamma secretase and BACE-1. One of these, glycogen synthase kinase 3 beta (GSK3β), helps to control the expression of BACE-1’s gene; a recent report indicates that inhibiting GSK3β reduces BACE-1 gene expression as well as brain pathology and memory deficits in a standard Alzheimer’s mouse model.
Meanwhile De Strooper’s laboratory has been studying beta arrestin 2, a protein that is recruited to the vicinity of certain active receptors on brain cells. Beta arrestin 2’s primary function seems to be to desensitize these active receptors—but for reasons that are unclear, beta arrestin 2 also has a tendency to grab gamma secretase and hold it in an area of the cell membrane where its amyloid-beta-producing activity is maximized. “It puts it where it needs to be,” says Amantha Thathiah, a postdoctoral researcher who was lead author of a recent study of beta arrestin 2, published in Nature Medicine. Thathiah found that genetically knocking down beta arrestin 2 levels in Alzheimer’s mice sharply reduced amyloid beta levels. She and her colleagues also looked at autopsied tissue from two separate groups of Alzheimer’s patients, and confirmed that the expression levels of beta arrestin 2 were elevated, specifically in the brain regions that are most affected by the disease. “This suggests to us that beta arrestins are really a viable target,” Thathia says.
In principle, says De Strooper, beta arrestin 2 or its associated neuronal receptors could be targeted in a way that spares two of the most worrisome off-target pathways hit by ordinary GSIs: Notch signaling, and the accumulation of apparently neurotoxic APP fragments.
He adds that all these candidate amyloid-beta-lowering drugs require appropriate clinical testing—which is probably not as a dementia-slowing therapy in frail Alzheimer’s patients. “The companies had better think how they want to profile their drugs,” he says. “If it is in preventive settings, the right clinical studies should be performed in the age class that is going to benefit from such treatments.”
Gandy echoes the sentiment. “We need a way to intervene that has clinical impact,” he says.