Poster L30 - Agent Based Modeling of Invasion Scenarios in the Gut Microbiome

Introduction: Agent-based modeling (ABM) offers a computational approach to recreating bacterial populations within biofilms, particularly relevant to understanding the gut microbiome. We introduce an ABM framework where bacteria are represented as spherical particles subjected to pushing forces, growing and replicating according to chemical feeding patterns. Nutrients and antimicrobials are modeled as chemical gradients, solved using the Finite Volume Method within a two-dimensional system.

The gut microbiome, crucial to human health, is highly dynamic and contains trillions of bacteria. Many of these bacteria are thought to form biofilms on the intestinal walls, comprising both beneficial host-commensal bacteria and harmful pathogens. This study applies ABM to explore the conditions under which invasive bacteria can displace an existing microbial population in the gut. Additionally, we investigate the regulatory role of host-generated antimicrobials, secreted by endothelial cells, on microbial dynamics.

Our model provides insights into the interactions within the gut microbiome and the impact of dynamic antimicrobial production by the host towards regulation or inhibition of invasion. Understanding these mechanisms is vital for developing therapeutic strategies to maintain a healthy microbiome and combat pathogenic invasions.

Materials and

Methods: Simulations were performed using an in-house agent-based model (ABM) built using Python and the numpy and FiPy packages. The ABM simulates bacteria as spherical particles in a 2D box using Numpy for data storage and processing, and includes various modules for physics, replication, and growth within the simulation. Bacteria are categorized by species with parameters like growth rate and density, and may be motile (planktonic) or passive within a biofilm.

The simulation uses scalar fields to represent bacterial density, nutrient concentrations, and other factors, with values stored in 50x50 scalar matrices. The concentration solver module models chemical solutes using a steady-state solution to a PDE, solved with the FiPy solver.

Initialization involves placing bacteria randomly within the fluid environment, with default numbers for each species. Planktonic bacteria are assigned random angular directions. The timestep module calls physics and agent modules to simulate biofilm growth in 15-minute increments. This includes planktonic movement, physical constraints to prevent overlaps and wall intersections, concentration solving, and bacterial growth and division. Growth is calculated using the Monod equation, and bacteria exceeding a certain radius are split into new bacteria.

During simulations, we allow for continuous seeding of the simulation environment with invasive bacteria, as well as for erosion effects including the continuous sloughing of upper level bacteria from the biofilm and periodic removal of biofilm chunks, as would be the case due to bulk erosive effects.

Results, Conclusions, and Discussions: Overall, we find that the outcome of an invasion scenario depends on the relative growth rates of invading or host bacteria, but may also be influenced by the feeding behaviors and pre-existing conditions of the environment. Over the course of several simulations, we allow invader growth rates to vary across several orders of magnitude to investigate the relative advantages of several possible scenarios.

We find that in the simplest case of a two species system, several factors determine the eventual outcome of an invasion scenario within the gut. In general, there is a strong advantage towards the pre-existing species within a biofilm, with new invaders constantly fighting against erosion effects. However, bacteria capable of feeding on the inner surface of a biofilm have a significant advantage over bacteria feeding on nutrients from the lumen. Bottom level growth within the biofilm allows bacteria to push competitors up to be eroded.

Within the biofilm, antimicrobials secreted from the bottom surface can strongly influence bacteria growth. We find that selective secretion of antimicrobial peptides, especially in response to molecules signaling the presence of invasive bacteria, have a stronger effect than equivalent total levels of inhibition emitted statically. As invasive bacterial growth rates increase, host effect inhibition maintains biofilm health longer when applied dynamically in response to signaling.

Acknowledgements (Optional):