Modeling Overview

Our modeling strategy has been divided, for clarity reasons, in 4 pages:

  1. - Overview, where we present our general modeling strategy.

  2. - In silico protein design to evaluate and optimize the anchoring and sensing of cancer cells by our liposomes.

  3. - Global kinetic model, to predict and boost the production of anticancer drug.

  4. - Liposome 2.0, where we summarize the model-driven modifications towards an improved version of our liposomes.

Model-driven understanding and optimization of our therapeutic liposomes

The local production of our anticancer drug within the liposome is the integrated output of cascading reactions. Once the liposome gets close enough to the cancerous cell, two systems are activated to ensure production: First the functional complementation of the split T7 RNA polymerase once HER2 interacts simultaneously with the antibodies Pertuzumab and Trastuzumab and, second, the derepression of the expression of the drug-producing enzyme through sensing of 2-HG by DhdR.

The liposome as a whole hosts a complex network of biochemical reactions. Therefore, modeling is necessary to understand and optimize its functioning. It is a major step to address our main questions:

  • Can the split T7 be successfully functionalized after HER2 recognition?
  • Can 5-FU be produced at a sufficient concentration once the liposome is anchored on a cancer cell?
  • How can we optimize anticancer drug production?

Our strategy

We addressed both a structural challenge regarding the design of a new-to-nature protein complex with embedded cancer cell recognition and transcriptional activation domains, and a functional challenge regarding drug production. Both challenges were tackled with dedicated models involving statistical molecular mechanics and coarse-grained kinetic modeling of biochemical networks, as detailed below.

Figure 1: Overview of our modeling strategy.Figure 1: Overview of our modeling strategy.

We first built a molecular model focusing on the structural optimization of the antibody-linked split T7 RNAP (See our Modeling / In silico protein design page). The goal was to design the linker sequences between the polymerase subunit and the transmembrane domain, and between the transmembrane domain and the anti-HER2 antibody. Indeed, the linker must allow enough movement for each subunit of the polymerase to bind to one another, while limiting the chance of unspecific binding. We thus simulated various protein structures containing different linkers and selected the best one.

In parallel, we constructed a kinetic model describing the biomolecular network of inside a liposome (See our Modeling / Global kinetic model page). The model captures all main processes from biosensing to cell-free gene expression and drug release. The goal was to simulate the dynamics of drug production, identify which factors determine its production, and how they can tuned to boost drug release. We then carried out model-driven experiments to verify the predictions, and could improve the model’s predictive capabilities by feeding experimental data. We performed a sensitivity analysis on our refined model to identify the parameters that impact the most drug production, which helped us rationally design more efficient therapeutic liposomes.

Finally, we combined the results obtained from both the structural and kinetic models to design an improved version of our synthetic liposomes (See our Modeling / Liposome 2.0 page).