Mechanistic model-driven exometabolomic characterisation of human dopaminergic neuronal metabolism

German Preciat, Agnieszka B. Wegrzyn, Edinson Lucumi Moreno, Cornelius C.w. Willacey, Jennifer Modamio, Fátima Liliana Monteiro, Diana C. El Assal, Alissa Schurink, Miguel A. P. Oliveira, Zhi Zhang, Ben Cousins, Hulda S. Haraldsdóttir, Siham Hachi, Susanne Zach, Germán Leparc, Yin Tat Lee, Bastian Hengerer, Santosh Vempala, Michael A. Saunders, Amy C. Harms, Enrico Glaab, Jens C. Schwamborn, Ines Thiele, Thomas Hankemeier, Ronan M. T. FlemingbioRxiv 2021 · DOI

TL;DR

This paper presents the development and validation of sophisticated computational models of human dopaminergic neuron metabolism. These models, built using a novel pipeline that integrates literature data with transcriptomic and metabolomic measurements, accurately predict metabolic fluxes and the consequences of metabolic perturbations, offering a powerful new tool for understanding Parkinson's disease.

What Problem Does This Paper Address?

Parkinson's disease (PD) is characterized by the degeneration of dopaminergic neurons, a process heavily influenced by their unique metabolic demands and vulnerabilities. Understanding the intricate metabolic network of these neurons, especially in the context of disease, is crucial for developing effective treatments. However, creating accurate, cell-type-specific metabolic models that can predict how these cells respond to metabolic changes has been a significant challenge. Existing models often lack the specificity or predictive power needed to capture the nuances of neuronal metabolism and its dysregulation in PD.

Methods

The authors developed a novel computational framework, XomicsToModel, to generate context-specific metabolic models. This pipeline integrates:

  • Manual Literature Curation: Extensive review of biochemical literature to identify active/inactive genes and reactions specific to dopaminergic neurons.
  • Multi-Omics Data Integration: Incorporation of quantitative transcriptomic and exometabolomic data from human stem cell-derived dopaminergic neurons.
  • Thermodynamic Constraint-Based Modeling: A key innovation that ensures models are not only flux-consistent but also thermodynamically feasible, using the thermoKernel algorithm.
  • Ensemble Modeling: Generation of numerous candidate models by varying pipeline parameters to identify those with the highest predictive accuracy.
  • Novel Objective Functions: Exploration of various objective functions (e.g., flux entropy maximization) to best predict metabolic behavior in non-growing cells like neurons.
  • Uncertainty Reduction Pipeline: A method to identify the most informative metabolites to measure in future experiments to further refine model constraints.

Datasets

  • Human Stem Cell-Derived Dopaminergic Neurons: Used for generating transcriptomic and targeted exometabolomic data.
  • Literature Data: Comprehensive manual curation of biochemical literature on neuronal metabolism.
  • Existing Metabolic Reconstructions: Recon3D was used as a foundational generic human metabolic model.
  • Exometabolomic Data: Generated from fresh and spent culture media using four complementary mass spectrometry platforms (LC-MS and GC-MS).

Key Insights and Ideas

  • Thermodynamic Constraints are Crucial: Incorporating thermodynamic feasibility significantly improves the accuracy and biological relevance of metabolic models, especially for complex cellular systems.
  • Non-Growing Cell Modeling is Different: Standard metabolic modeling approaches optimized for growing cells (e.g., maximizing biomass) are not suitable for non-replicating neurons. New objective functions and constraints are needed to capture maintenance and electrophysiological activity.
  • Prioritizing Future Experiments: The developed uncertainty reduction pipeline can guide experimental design, ensuring that future metabolomic measurements provide the most valuable information for model refinement.
  • Dopaminergic Neurons are Metabolically Unique: The models highlight specific metabolic pathways and dependencies that make these neurons vulnerable, particularly in energy metabolism.

Key Results

  • Two High-Fidelity Models: The authors generated two primary models: iDopaNeuroC (condition-specific, prioritizing in vitro data) and iDopaNeuroCT (cell-type-specific, prioritizing literature data).
  • Accurate Perturbation Prediction: The iDopaNeuroC model achieved high qualitative accuracy (correct/total = 0.78 for complex I inhibition) and moderate quantitative accuracy (Spearman rho = 0.48) in predicting metabolic responses to perturbations like complex I and V inhibition, and glucose-to-galactose switch.
  • Model Generalizability: The developed modeling approaches and pipeline are generalizable to other cell types and diseases involving metabolic dysfunction.
  • Identification of Key Metabolic Pathways: The models revealed the reliance of dopaminergic neurons on glycolysis and oxidative phosphorylation, and highlighted the impact of perturbations on these pathways.

Limitations and Future Work

  • In Vitro vs. In Vivo Differences: The authors acknowledge potential molecular divergences between in vitro models and in vivo substantia nigra dopaminergic neurons, suggesting future work with improved cell culture protocols.
  • Metabolite Measurement Challenges: Certain metabolites showed significant deviations between manufacturer specifications and measured concentrations, highlighting the difficulty in accurately quantifying all components in complex media.
  • Model Refinement: While the models are highly predictive, further experimental data, particularly for unmeasured metabolites, is needed to fully constrain all exchange reactions.
  • Personalized Models: The authors suggest that similar methodologies could be used to generate context-specific models for individual PD patients to unravel underlying metabolic similarities.

Summary updated

Abstract

Abstract Starting with a comprehensive generic reconstruction of human metabolism, we generated high-quality, constraint-based, genome-scale, cell-type and condition specific models of metabolism in human dopaminergic neurons, the cell type most vulnerable to degeneration in Parkinson ’ s disease. They are a synthesis of extensive manual curation of the biochemical literature on neuronal metabolism, together with novel, quantitative, transcriptomic and targeted exometabolomic data from human stem cell-derived, midbrainspecific, dopaminergic neurons in vitro . Thermodynamic constraint-based modelling enabled qualitatively accurate and moderately quantitatively accurate prediction of dopaminergic neuronal metabolite exchange fluxes, including predicting the consequences of metabolic perturbations in a manner also consistent with literature on monogenic mitochondrial Parkinson ’ s disease. These dopaminergic neurons models provide a foundation for a quantitative systems biochemistry approach to metabolic dysfunction in Parkinson ’ s disease. Moreover, the plethora of novel mathematical and computational approaches required to develop them are generalisable to study any other disease associated with metabolic dysfunction.
BibTeX
@misc{preciat2021mechanistic,
  author = {German Preciat and Agnieszka B. Wegrzyn and Edinson Lucumi Moreno and Cornelius C.W. Willacey and Jennifer Modamio and Fátima Liliana Monteiro and Diana C. El Assal and Alissa Schurink and Miguel A. P. Oliveira and Zhi Zhang and Ben Cousins and Hulda S. Haraldsdóttir and Siham Hachi and Susanne Zach and Germán Leparc and Yin Tat Lee and Bastian Hengerer and Santosh Vempala and Michael A. Saunders and Amy C. Harms and Enrico Glaab and Jens C. Schwamborn and Ines Thiele and Thomas Hankemeier and Ronan M. T. Fleming},
  title = {{Mechanistic model-driven exometabolomic characterisation of human dopaminergic neuronal metabolism}},
  year = {2021},
  doi = {10.1101/2021.06.30.450562},
  howpublished = {bioRxiv},
}

Loading related content...

Figure