A Modular Framework for Multiscale Spatial Modeling of Viral Infection and Immune Response in Epithelial Tissue

Abstract

The COVID-19 crisis has shown that classic sequential models for scientific research are too slow and do not easily encourage multidisciplinary scientific collaboration. The need to rapidly understand the causes of differing infection outcomes and vulnerabilities, to provide mechanistic frameworks for the interpretation of experimental and clinical data and to suggest drug and therapeutic targets and to design optimized personalized interventions all require the development of detailed predictive quantitative models of all aspects of COVID-19. Many of these models will require the use of common submodels describing specific aspects of infection (e.g., viral replication) but combine them in novel configurations. As a contribution to this development and as a proof-of-concept for some components of these models, we present a multi-layered 2D multiscale, multi-cell model and associated computer simulations of the infection of epithelial tissue by a virus, the proliferation and spread of the virus, the cellular immune response and tissue damage. Our initial, proof-of-concept model is built of modular components to allow it to be easily extended and adapted to describe specific viral infections, tissue types and immune responses. Immediately after a cell becomes infected, the virus replicates inside the cell. After an eclipse period, the infected cells start shedding diffusing infectious virus, infecting nearby cells, and secretes a short-diffusing cytokine signal. Neighboring cells can take up the diffusing extracellular virus and become infected. The cytokine signal calls for immune cells from a simple model of the systemic immune response. These immune cells chemotax and activate within the tissue in response to the cytokine profile. Activated immune cells can kill underlying epithelial cells directly or by secreting a short-diffusible toxic chemical. Infected cells can also die by apoptosis due to the stress of viral replication. We do not include direct cytokine mediated protective factors in the tissue or distinguish the complexity of the immune response in this simple model. Despite unrealistically fast viral production and immune response, the current base model allows us to define three parameter regimes, where the immune system rapidly controls the virus, where it controls the virus after extensive tissue damage, and where the virus escapes control and infects and kills all cells. We can simulate a number of drug therapy concepts, like delayed rate of production of viral RNAs, reduced viral entry, and higher and lower levels of immune response, which we demonstrate with simulation results of parameter sweeps of select model parameters. From results of these sweeps, we found that successful containment of infection in simulation directly relates to inhibited viral internalization and rapid immune cell recruitment, while spread of infection occurs in simulations with fast viral internalization and slower immune response. In contrast to other simulations of viral infection, our simulated tissue demonstrates spatial and cellular events of viral infection as resulting from subcellular, cellular, and systemic mechanisms. To support rapid development of current and new submodels, we are developing a shared, publicly available environment to support collaborative development of this framework and its components. We warmly invite interested members of the biological, medical, mathematical and computational communities to contribute to improving and extending the framework.