Bottom-up modeling approach for the quantitative estimation of parameters in pathogen-host interactions

Abstract

Opportunistic fungal pathogens can cause bloodstream infection and severe sepsis upon entering the blood stream of the host. The early immune response in human blood comprises the elimination of pathogens by antimicrobial peptides and innate immune cells, such as neutrophils or monocytes. Mathematical modeling is a predictive method to examine these complex processes and to quantify the dynamics of pathogen-host interactions. Since model parameters are often not directly accessible from experiment, their estimation is required by calibrating model predictions with experimental data. Depending on the complexity of the mathematical model, parameter estimation can be associated with excessively high computational costs in terms of run time and memory. We apply a strategy for reliable parameter estimation where different modeling approaches with increasing complexity are used that build on one another. This bottom-up modeling approach is applied to an experimental human whole-blood infection assay for Candida albicans. Aiming for the quantification of the relative impact of different routes of the immune response against this human-pathogenic fungus, we start from a non-spatial state-based model (SBM), because this level of model complexity allows estimating a priori unknown transition rates between various system states by the global optimization method simulated annealing. Building on the non-spatial SBM, an agent-based model (ABM) is implemented that incorporates the migration of interacting cells in three-dimensional space. The ABM takes advantage of estimated parameters from the non-spatial SBM, leading to a decreased dimensionality of the parameter space. This space can be scanned using a local optimization approach, i.e., least-squares error estimation based on an adaptive regular grid search, to predict cell migration parameters that are not accessible in experiment. In the future, spatio-temporal simulations of whole-blood samples may enable timely stratification of sepsis patients by distinguishing hyper-inflammatory from paralytic phases in immune dysregulation.

Dynamic Interplay of Host and Pathogens in an Avian Whole-Blood Model

“…We adapted our human whole-blood model (17,(28)(29)(30) and generated state-based models (SBMs) that simulate the immune reactions during infection in avian whole-blood samples. In order to cope with known differences between fungal and bacterial infection scenarios in avian blood, we implemented slightly different models for bacteria and fungi.…”

“…In line with our previous studies on whole-blood infections in humans (17,(28)(29)(30), the experimental whole-blood infection assay was complemented by virtual infection modeling. By calibrating the virtual infection model to experimental data, the functional characteristics of the immune response in avian whole blood were quantified.…”

Dissecting Candida albicans Infection from the Perspective of C. albicans Virulence and Omics Approaches on Host–Pathogen Interaction: A Review

“…This analysis provides new insights into systemic host responses and may ultimately lead to the development of new anti-microbial regimens. Lehnert et al [130] constructed a bottom-up modeling approach, from a state-based model to an agent-based model, to simulate C. albicans infection in a human whole-blood model. The authors reported that polymorphonuclear neutrophils (PMN) cells are important effector cells in killing C. albicans in human blood.…”

“…The authors reported that polymorphonuclear neutrophils (PMN) cells are important effector cells in killing C. albicans in human blood. The authors also surmised that a systemic medicine approach that utilizes the predictive power of virtual infection models will play a crucial role in infectious disease diagnosis in the future [130]. …”

Identification of human proteins vulnerable to multiple organisms

“…virus infects host), or on the population level (pathogen infections affecting a human population). Interactions among host and pathogen protein networks are called pathogen-host interactomes and investigating them may allow us to apprehend and be aware of the functioning of the host immune system and these important interacting proteins (IIPs) present in the host cell could be utilized as potential drug targets [2]. We have analyzed 3,905 interactions (edges) from 182 pathogens consisting of 1,188 proteins (nodes) that interact with 668 human proteins (nodes) from The Pathogen-Host Interaction Interaction Search Tool (PHISTO).…”