Quantitative systems pharmacology model of key intraneuronal pathways regulating toxic protein accumulation in Alzheimer’s disease
Mark Meerson (1), Tatiana Karelina (1), Diana Clausznitzer (2)
(1) InSysBio, Moscow, Russia, (2) Abbvie Deutschland GmbH & Co KG, Ludwigshafen, Germany
Objectives: Alzheimer’s disease (AD) is a complex neurodegenerative disorder, its pathogenesis mechanism is still debated and there is no effective treatment or prevention therapy. Apart from the accumulation of protein aggregates there are many intracellular processes altered in AD brain, which are related to Aβ and tau pathology, and are believed to be the cause or contributors to toxicity. They include, for example, protein degradation and secretion machinery and lipid metabolism, as demonstrated by changes in specific biomarkers in AD pathology. Although some of these cellular pathways [1], as well as, e.g. amyloid or tau pathology progression have been modelled previously, there is a lack of insights in the quantitative aspects of their interactions in pathology and health. Quantitative systems pharmacology (QSP) provides a quantitative framework to study the interaction of complex disease mechanisms.
The objective of this study was to develop a QSP model describing the key neuronal processes observed to be affected in AD and potentially influencing, or significantly influenced by, tau and Aβ pathology. The model is aimed at describing both, baselines of target variables in healthy human brain and their dynamics during disease progression.
Methods: The proposed model describes several neuronal processes in human brain disturbed in AD by means of ordinary differential equations (ODEs). Based on a literature review, several pathways interacting with Aβ and tau accumulation were selected: protein degradation (via autophagic-lysosomal system, ALS, and cytoplasmic proteolytic systems), protein secretion (via exosome secretion), as well as sphingolipid and cholesterol metabolism.
Protein degradation and secretion were described using a mechanistic approach, which allows for calibration to quantitative data (e.g. [2]) and comparison with in vitro data. Most of the important regulatory hubs (e. g. mTOR in the ALS) were described by algebraic functions, which were derived using a quasi-steady state (QSS) assumption and are dependent on various cellular components. The model includes activating and inhibiting interactions between variables and functions, allowing for the reflection of complexity of cellular metabolic regulation. Here, we purposefully focused on the dynamics of dysregulation of neuronal homeostasis during AD pathology describing several drivers of the dysregulation, including (but not limited to) Abeta and Tau protein pathology, in terms of explicit functions derived from published data.
Calibration of the model was carried out using public domain data on baseline concentrations of different metabolites and vesicle content in human brain tissues. The model was validated on in vivo clinical data sets and in vitro data for interventions into different cellular pathways.
Results: The model correctly describes the baseline concentrations of cellular components in healthy human brain. Steady-state levels of 87% of variables in brain compartment were compared to published quantitative data. 77% of them match data within 2-fold range and 80% of these variables in turn match the data within 20% error. Comparing model simulations to data for pharmacological treatments we show that cellular responses to various compounds (e.g., rapamycin, vinblastine, ACAT inhibitors) are described correctly by the model, demonstrating that our model captures complex interactions between cellular pathways.
Finally, the model is able to reproduce the dysregulation of different cellular pathways in AD compared to healthy subjects. In particular, the model matches observed biomarkers indicating impairment of protein degradation and secretion, demonstrating that the model correctly captures the age-dependent breakdown of protein degradation and secretion machinery and its exacerbation by amyloid and tau pathology in AD.
Conclusions: The model describes key AD-related cellular processes and we showed that it matches levels of different cellular components in healthy subjects, as well as key pathway interactions and dysregulation in AD. The model can be used as a mechanistic description of processes driving initial accumulation of toxic proteins in AD and of the feedback mechanisms leading to further development of AD progression. In future work, it should be integrated with dynamic models of amyloid and tau pathology to explore their interactions during disease progression. The model provides a quantitative framework for hypothesis generation and testing in biomedical research and drug development for AD.
References:
[1] Lloret-Villas, A., Varusai, T. M., Juty, N., Laibe, C., Le Novere, N., Hermjakob, H., & Chelliah, V. (2017). The impact of mathematical modeling in understanding the mechanisms underlying neurodegeneration: Evolving dimensions and future directions. CPT: Pharmacometrics and Systems Pharmacology, 6(2), 73–86. https://doi.org/10.1002/psp4.12155
[2] Bordi, M., Berg, M. J., Mohan, P. S., Peterhoff, C. M., Alldred, M. J., Che, S., … Nixon, R. A. (2016). Autophagy flux in CA1 neurons of Alzheimer hippocampus: Increased induction overburdens failing lysosomes to propel neuritic dystrophy. Autophagy, 12(12), 2467–2483. https://doi.org/10.1080/15548627.2016.1239003
