Multi-scale and multi-compartment approaches to understand host-pathogen dynamics: TB as a case study

Denise Kirschner

Short Abstract
Tuberculosis (TB) is the main cause of death due to infectious disease in the world today. An estimated 2 billion people are carriers of M. tuberculosis, the bacteria that causes TB. The different outcomes from infection with Mtb (primary or active disease, latent infection, disease reactivation) are in large part determined by the formation and function of lung granulomas, structures that dynamically contain the infection in an immune microenvironment but unfortunately also serve as a niche for bacterial survival. Events that play a role in granuloma formation and function occur over a broad range of biological scales - molecular (e.g. cytokines), cellular (e.g. T cells, macrophages, and dendritic cells), tissue (e.g. lung, lymph node) and larger scales (e.g. lymphatics, blood and other organs) - and also time scales (seconds to the lifetime of host). A systems biology approach, combining computational models with data from relevant animal models, offers a path to understanding how these different events influence infection control. A particular challenge for study of Mtb infection is that observing formation of granulomas is not readily accessible within lung tissue.

In this work, multi-scale and multi-compartment computational modeling is coupled with novel experimental data to determine how mechanisms at one biological scale affect dynamics at other scales in order to predict which mechanisms control granuloma formation and function. We develop and apply a methodology to address issues of parameter uncertainty and sensitivity to analyze our computational models. We develop both 2-dimensional (2-D) and 3-D agent-based models to explore these questions, allowing us also to assess the influence of dimensionality and determine what information the simpler model can and cannot provide. Finally, we develop a tool known as tuneable resolution that allows us to control model detail in a mechanistic way. This aids in future studies where performance issues related to efficiency and speed of simulations are important. Finally, our goal of linking immune dynamics occurring in a lymph node to those occurring at the site of infection in the lung has been realized; we present the first hybrid, multi-compartment model exploring the role of dynamics occurring between lymph nodes and lung during infection and determine how mechanisms related to trafficking between these compartments are key to successful infection control.