Kinetic Model of Amyloid Beta Distribution and Allometric Scaling from Mouse to Monkey and Human
Tatiana Karelina(1), Eugenia Kazimirova(1), Kirill Zhudenkov(1), Oleg Demin Jr(1), Yasong Lu(2), Timothy Nicholas(2), Sridhar Duvvuri(2), Hugh A. Barton(2)
(1)Institute for Systems Biology SPb, Moscow, Russia; (2)Pfizer Worldwide Research and Development, Groton, CT, USA
Objectives: Abnormal accumulation and deposition of amyloid beta (Aβ) in the brain is considered one of the causes of Alzheimer's disease. Decreasing Aβ levels in brain could be a potential therapy, it is thus important to understand how Aβ distributes among different compartments. Here we present an updated model, describing Aβ kinetics in mouse and its allometric scaling to monkey and human.
Methods: The model describes three types of Aβ (Aβ40, Aβ42 and Aβr, all other forms), which distribute in compartments for brain cells (BC), brain interstitial fluid (BIF), cerebrospinal fluid (CSF), plasma, and peripheral tissues (PT). Aβ is generated in BIF and PT. Aβ concentrations in each compartment change due to transport or degradation. Bulk flow from BIF to CSF and to lymph is also taken into account. All the calculations and fitting were done in DBSolve Optimum software. Most of the kinetic data for model calibration have been obtained in wild-type mouse, so aggregation of Aβ was not considered. Allometric scaling of mouse model was used to describe CSF 13C-Aβ in the monkey and human in the stable isotope labeling kinetic (SILK) studies. The mouse model was also tested against reported CSF and brain Aβ time courses after a dose of a γ-secretase inhibitor (GSI).
Results: The model satisfactorily describes most of the literature Aβ kinetics data. The set of parameters obtained for mouse allowed adequate allometric scaling to monkey and human, with Aβ production fitted for each species. The mouse model qualitatively reproduced the time shift between the effects of the GSI on brain and CSF Aβ, but with additional fitting of some parameters controlling Aβ distribution between BIF and BC. Local sensitivity analysis also has shown that the shape of time course of the brain Aβ concentration is sensitive to these parameters.
Conclusions: Parameter values in the updated model provided better descriptions of existing data. Difficulty in achieving better fitting of the GSI data likely stems from poor understanding of Aβ distribution in brain, partly due to lack of quality data on BC and BIF concentrations. Such information would be important for further improvement of the model.
