A mix of math and experiments links a main symptom of Alzheimer’s disease to subtle changes in protein dynamics
The current study was carried out by postdoctoral fellow Vanessa Schmidt and PhD student Katharina Baum, with Angelyn Lao from Olaf Wolkenhauer’s lab at the University of Rostock. It builds on previous work from Thomas’ group. In 2009 he showed that a protein called SORLA is involved in the development of the disease. This molecule participates in the movement of APP through the cell and the production of amyloid-beta peptide. Its effects are usually beneficial: increasing the amount of SORLA leads to less A-beta, both in the test tube and animal models. Mice that have been genetically engineered to lack SORLA, on the other hand, produce higher levels of the dangerous amyloids. But the reasons have been unclear.
The processing of APP involves an interplay of so many proteins that a “systems biology” approach, using mathematical modeling, was necessary to describe their roles. “Most mouse models and other studies have used ‘all-or-nothing’ methods, either completely eliminating particular molecules, or raising their amounts to unnaturally high levels,” Thomas says. “Patients experience much subtler changes in protein levels. We needed a way to make small changes in protein expression and watch their effects over long periods of time.” Levels of SORLA drop in many Alzheimer’s patients, but it isn’t completely lacking.
The scientists developed a unique cell-based system in which they could incrementally raise or lower concentrations of APP and SORLA. Then they carried out quantitative studies to study the effects of the changes on the production of amyloid-beta peptides. The next step was for Katherina, Jana and their colleagues to replicate these effects in mathematical models.
A breakthrough came when the scientists applied “Hill kinetics” to the problem. This approach detects cases when the elements of a system produce an effect by cooperating, rather than acting independently. It showed that the production of A-beta depended on some sort of cooperative event, which further experiments exposed as the pair-wise binding of APP proteins.
“This pairing creates an optimal ‘platform’ for enzymes to bind to APP, the first step in producing dangerous fragments,” Thomas says. “That discovery gave us a hint about the role of SORLA. It doesn’t directly stop enzymes from binding to APP, which was one of our early hypotheses. Instead, it interferes with the pairing of APP molecules. It locks up single copies so they don’t bind to each other. This means fewer ideal ‘docking sites’ for the enzymes, and a lower production of A-beta.”
The cell culture method allowed the scientists to observe the effects of a gradual raising or lowering of levels of SORLA. Even small reductions led to significant jumps in the amount of A-beta. “This helps explain how a drop in SORLA of just 25 percent in some Alzheimer’s patients leads to dramatically more fragments,” Thomas says. “It’s due to an increase in the cleavage of APP. Other groups have shown that APP normally forms pairs about 30 to 50 percent of the time. If levels of SORLA drop, that proportion rises. Cells produce more amyloid-beta peptide, leading to accumulations and the dangerous plaques seen in Alzheimer’s disease.”
The study has implications for the development of new therapies, he says. Rather than trying to inhibit the activity of APP-cutting enzymes, which healthy cells might need for other reasons, scientists can look for drugs or other substances that imitate the action of SORLA and block the pairing of APP molecules.
“This is the first mathematical explanation of the anti-Alzheimer effects of SORLA,” Thomas says, “and it helps show how relatively small changes in the ‘dosage’ of this molecule can have big effects on the course of the disease.”
Schmidt V, Baum K, Lao A, Rateitschak K, Schmitz Y, Teichmann A, Wiesner B, Petersen CM, Nykjaer A, Wolf J, Wolkenhauer O, Willnow TE. Quantitative modelling of amyloidogenic processing and its influence by SORLA in Alzheimer’s disease. EMBO J. 2011 Oct 11;31(1):187-200