Joseph Johnson Biochemistry
Alzheimer’s Disease (AD) is a debilitating neurodegenerative disease that begins with subtle memory loss and proceeds relentlessly over 7 to 10 years, eventually leading to complete incapacitation and death. Amyloid plaques are a characteristic feature of AD brains and are predominantly composed of a short peptide fragment termed Abeta, which is derived from the amyloid precursor protein (APP) by proteolytic cleavage. One popular hypothesis proposes that Abeta production is the causal event leading to AD. Two proteases (proteins that cleave other proteins) are responsible for the generation of AB from APP, beta-secretase (BACE1) and gamma-secretase. We propose that preventing BACE cleavage of APP will decrease Abeta production, thereby stopping or reversing the course of AD.
A thorough understanding of protein-protein interactions is essential to the overall understanding of protease function. As mentioned earlier, BACE1 is an attractive target for the design of therapeutics that may delay or prevent the occurrence of AD. In order to more effectively predict the possible side effects of BACE1 therapeutics, it is critical to know how it interacts with other proteins. Identification of proteins that interact with BACE1 in a biological system is time-consuming and expensive. Our proposed approach was to test the InsightII software package (Accelrys, Inc.), recently acquired by the Viz Lab, for its ability to predict and analyze these potential protein-protein interactions. The ability to predict interactions between proteins would be a powerful step forward in narrowing the pool of candidate partner proteins that could then be tested in biological systems. Two other software packages were acquired in aid in this study. NAMD is a program designed to simulate the molecular dynamics of large biomolecules, such as proteins. This software is available from the Theoretical and Computational Biophysics Group at the University of Illinois at Urbana-Champaign. The other software is AutoDock, available from the Scripps Research Institute, which calculates the interactions between two molecules. The NAMD software was used to generate different starting structures and AutoDock was used to bring the candidate molecules together and predict whether there were any favorable interactions.
The initial stage of this project was familiarization with the software packages. Once we gained some proficiency with the software, we moved on to simulate different short peptides in order to generate a variety of conformations. Peptides or proteins are generally flexible molecules that can adopt a plethora of conformations. The purpose of the NAMD molecular dynamics software was to grab “snapshots” of the protein as it is was allowed to move in silico (in computer simulated space). In order to successfully dock two biomolecules, it is typically necessary to allow some freedom for movement in both molecules because they are, in reality, flexible. The difficulty of this approach is that the movement of any atom in a simulated system must be calculated based on fundamental physical principles. These calculations equate to computational time; the greater the number of atoms the longer the time required to complete the calculation. Typically, allowing some flexibility is not detrimental because usually one of the molecules being brought together or “docked” is relatively small and fairly rigid. This is not generally true of many biomolecules, proteins being a prime example, which are rather flexible and capable of adopting a myriad of conformations. The power of the coupling a molecular dynamics simulation to generate different starting structures followed by rigid docking of the two biomolecules allows a compromise between obtaining meaningful data and being able to collect them in a reasonable time frame. We are still at the initial stages of the molecular dynamics and docking simulations as we work through the challenges of trying to get a truly random sampling of the seemingly infinite number of possible peptide sequences. BACE1 is has been shown to recognize between 8 to 11 amino acids. Once recognized and bound, a particular peptide is then cleaved into two peptide products. Since there are 20 standard amino acids, that means that there are 820 to 1120 possible sequences or in excess of 1 x 1018 possibilities. We are in the process of generating what we hope will be a representative subset of these immense possibilities, which we can then use to predict the peptide sequences that will favorably interact with BACE1. Once we have these predictions from the computational studies, we will move on to test their validity in the laboratory.