"We're very excited to provide more insight into how this bizarre process takes place," says principal investigator Raphael Kopan, Ph.D., professor of medicine and of molecular biology and pharmacology. "The more we understand the way this enzyme works, the easier it will be to design better and more intelligent approaches to tweaking the enzyme to do what we want."
The results are published online in the early edition of the Proceedings of the National Academy of Sciences and will appear in the Oct. 28 print edition. The study was an international collaboration between researchers at the School of Medicine, Merck and Co. Inc., University of Tokyo, Harvard Medical School, University of Tennessee at Memphis, and the K.U. Leuven and Flanders Interuniversity Institute for Biotechnology in Belgium.
The results focus on gamma-secretase, an enzyme that clips a long protein called amyloid precursor protein (APP), which results in fragments that accumulate as brain plaques. The plaques are a hallmark of Alzheimer's disease, making inhibition of gamma-secretase activity a main objective for new Alzheimer's drugs.
Kopan and colleagues previously found the enzyme also is required for another protein called Notch to function. Notch helps produce many cell types and, using a thymus organ culture model system, Kopan's team found gamma-secretase inhibitors had the potential to interfere with production of key immune cells.
"Ideally, the next generation of drugs will be able to prevent gamma-secretase from triggering production of plaques without interfering with the enzyme's role in Notch signaling," Kopan says. "That goal is made easier with every additional glimpse into how the enzyme works."
The team's latest findings, which suggest that gamma-secretase may contain multiples of one subunit, are a step in that direction.
To confirm the enzyme cleaves both APP and Notch, Kopan's team first examined whether the two compete with each other for the enzyme's attention in culture cells. They did indeed find evidence of competition: Notch cleavage was significantly stunted after the addition of C99, the piece of APP upon which gamma-secretase acts. The opposite also was true: In the presence of Notch fragments, there was significantly less production of ABeta40, one product of APP cleavage.
The team also ranked each of seven different gamma-secretase inhibitors in order of its ability to interfere with cleavage of Notch or APP. The rankings were the same for both proteins. Six of the inhibitors also had an identical effect on Notch and APP.
Together, these findings suggest gamma-secretase cleaves both APP and Notch and treats them interchangeably -- rather than distinguishing between the two, it simply clips whichever it runs into first. The presence of either protein can therefore influence gamma-secretase's effect on the other.
Next, the team examined how the same enzyme is responsible for clipping each molecule at two separate sites. A breakthrough came from observing the activity of a particular mutation in Notch that is protected from gamma secretase cleavage. As expected, Kopan's team found that this fragment does not compete with C99. Surprisingly, though, it did bind to the enzyme.
According to Kopan, enzymes can either have one site where they interact with molecules or have separate cutting and binding sites. This study suggests that gamma-secretase belongs to a class of enzymes where, in addition to the active site (which is in limited supply and therefore leads to competition between APP and Notch), there also are "binding" sites, where molecules can latch onto the enzyme without competing with each other and without becoming subject to cleavage.
"The active site is like a mouth -- it chews whatever it touches but can only chew one thing at a time," Kopan explains. "The other site is like a hand -- it's used for holding, and doesn't interfere with the ability of the mouth to chew another object. Maybe one molecule acts as the "hand" serving a meal to the "mouth," which is located on another molecule."
The researchers tested this theory in several ways. Gamma-secretase is a large, complex enzyme composed of four proteins. At its core is a molecule called presenilin. Kopan's team found that antibodies designed to find a tag on one presenilin molecule also could latch onto a different presenilin molecule with a different tag. This implies the two molecules are located close to each other.
Kopan's team confirmed the molecules' close proximity to each other by creating an irreversible chemical bond between the two molecules using a small inhibitor molecule designed by Merck and Co. in England.
"The data generated by our colleagues at Merck shows conclusively that there are two presenilin molecules in very tight proximity to each other," Kopan says. "But we still can't differentiate how the catalytic core of gamma-secretase, the "mouth" of the enzyme, is organized and whether it functions as a single entity or at the interface between two molecules."
To further investigate the complex organization and function of the enzyme, the researchers examined the effects of presenilin mutations found in people who develop the early, genetically linked form of Alzheimer's disease. They reintroduced mutated presenilin proteins from Alzheimer's disease patients to cultured cells missing both presenilin molecules. The mutant proteins failed to completely restore gamma-secretase activity, but the cells still produced ABeta42, the product of APP cleavage that accumulates to form brain plaques. In fact, some even resulted in production of more ABeta42 than when only normal presenilin molecules were present.
Kopan's team hypothesized that perhaps the mutated presenilin molecules influence production of ABeta42 by gamma-secretase activity by interacting with each other differently than do normal molecules, even though they themselves cannot efficiently clip APP or Notch. If a mutated presenilin molecule could be developed that is completely incapable of performing the active functions of gamma-secretase on its own and yet still is capable of increasing production of ABeta42, it would confirm that the enzyme has a functional unit at the interface between two presenilin proteins and suggest that familial forms of Alzheimer's disease are caused by inter-molecular interactions between mutant and normal proteins.
The team was able to observe that exact phenomenon; however, the finding was fleeting and, thus far, has not been reproducible.
"There are many reasons why this experiment shouldn't work, and yet for a short while it did," Kopan says. "Perhaps some component in the experimental conditions that allowed this to happen has changed; however, we don't fully understand what those key variables are and therefore have lost the ability to replicate the result. Our hope is that by publishing this study and proposing this experimental approach we will inspire other scientists to try different pairs of mutations or to develop better experiments while we continue to work on ours."
Schroeter EH, Ilagan MXG, Brunkan AL, Hecimovic S, Li Y, Xu M, Lewis HD, Saxena MT, Strooper BD, Coonrod A, Tomita T, Iwatsubo T, Moore CL, Goate A, Wolfe MS, Shearman M, Kopan R. A presenilin dimer at the core of the gamma-secretase enzyme? Insights from parallel analysis of Notch 1 and APP proteolysis. Proceedings of the National Academy of Sciences, October 2003.
Funding from the National Institutes of Health, the Alzheimer's Association, the Zenith award and the French Foundation supported this research.
The full-time and volunteer faculty of Washington University School of Medicine are the physicians and surgeons of Barnes-Jewish and St. Louis Children's hospitals. The School of Medicine is one of the leading medical research, teaching and patient-care institutions in the nation. Through its affiliations with Barnes-Jewish and St. Louis Children's hospitals, the School of Medicine is linked to BJC HealthCare.