Connecting in-vivo optical imaging with dynamic modelling of host-pathogen interaction during bacterial infection

Bacterial infections are experienced by almost everyone at least once during their lifetime. Although antibiotic treatments will cure the infection in most cases, bacteria are constantly developing and can become resistant to antibiotics. This way, new strains of bacteria continue to emerge, including strains that have considerable costs in terms of morbidity and mortality. Multiple outbreaks of aggressive forms of E.coli in recent years have demonstrated that these types of bacteria are not completely under control. For these bacteria we need to develop appropriate treatments, which can only be done if we understand how bacteria establish infections.

To find out how a bacterial infection evolves in the body over time, we should investigate this process inside living animals, such as mice, that can represent the course of human disease. Recently, new imaging systems have been developed that can track bacteria inside a mouse without any invasive intervention. This is achieved by using bacteria that emit light at a wavelength larger than the visible range, which can travel through several centimetres of tissue; enough for the light of bacteria in the gut to reach the surface of the animal. It is also possible to use a fluorescent probe inside the mouse to visualize other cells that are important for the immune response of the mouse. With this technique we can see where the bacteria and immune cells are in the body, how many there are, and by imaging at multiple time points, we can also follow their movement. The precision with which we can locate cells in the body is of the order of millimetres.

In order to be able to interpret the large-scale in-vivo images in terms of the small-scale infection mechanisms, we need to link what we can observe with what we know about the infection based on other investigations. These other investigations include detailed post-mortem investigations of separate organs that show exactly where the bacteria are located; for example, whether bacteria are attached to the wall of the gut, or whether they have moved beyond it.

In this project, we will develop techniques and tools that allow us to combine in-vivo image analysis with the available biological knowledge at the cellular and molecular scale. The project includes theoretical modelling of the progress of bacterial infection with deterministic and stochastic methods, as well as experimental in-vivo imaging of bioluminescent bacteria inside a mouse. The modelling work will be performed within the Theoretical Systems Biology group, while the experimental work will be performed within the Molecular Pathogenesis group, both at Imperial College London. By interweaving theoretical modelling with experimental investigation of the disease model, we will be able to feedback theory into experimental design, and experimental results into the definition of the model; both are equally important in order to arrive at good mechanistic models for the infection processes. The model we will thus obtain of the progress of bacterial infection can deliver new insights across spatial and temporal scales and potentially help identify new targets for treatment.

The progression of a bacterial infection takes place over a range of spatial and temporal scales. Biomedical imaging is ideally suited to investigate disease progression in-vivo over time in small animals. To facilitate interpretation of in-vivo imaging results in terms of biological mechanisms, we propose to develop a modelling and statistical analysis framework that links these two. We will apply this to investigate host-pathogen interactions taking place when a mouse is infected with Citrobacter Rodentium, a model for Enterohaemorrhagic Escheria coli (EHEC). Mice infected with bioluminescent bacteria will be imaged using the IVIS system for bioluminescence/fluorescence tomography and X-ray CT. This will show the spatial and temporal distribution of the bacteria and fluorescently targeted immune cells. Our framework will consist of a spatio-temporal model that divides the mouse in several compartments in which biological processes take place that we will model using ordinary and stochastic differential equations. The model will be used to estimate key infection parameters for each compartment at a number of time points in light of the in-vivo measurements. We will base an initial model on available in-vivo imaging data and cellular and molecular data acquired post-mortem. The model will be further informed by performing experiments targeted at filling gaps in our mechanistic understanding, including in-vivo experiments of a wild-type mouse, histological analysis and FACS. We will use the (calibrated) model to simulate the outcome of knocking out parts of the immune system, based on which we will design and perform an experiment with a mutant mouse model. The experimental results will be used to validate and refine our model. This close integration of experimental and modelling efforts will provide immediate feedback between them. The resulting multi-compartment model can increase our understanding of disease mechanisms, and identify new targets for treatment.

Beneficiaries in the short term:
- Commercial developers of in-vivo imaging systems and fluorescent probes, the increased value of the insights that can be obtained by combining in-vivo imaging with modelling could increase the demand for the technology (3 years)

Beneficiaries in the long term:
- Pharmaceutical companies: the increased insight in the mechanisms of disease progression can identify new targets for treatment (5 years)
- The community: containing the spread of bacterial infection by treatment will benefit public health, especially in closely populated areas (10 years)