Team:Pitt/cathelicidinModels/Intro

Cathelicidin Simulation

Before pre-clinical testing is started for a new medical product, computer simulations can be used as a low-investment strategy to investigate product safety. Our team is using a computational modeling approach to investigate the potential of topically applied cathelicidin (an anti-microbial peptide) to reduce acne-associated skin inflammation. Once applied to the skin, the anti-microbial peptides (AMPs) will initially disrupt and kill inflammation-causing bacteria on the skin. However, as pointed out by Harder, et al. (2013), AMPs not only kill bacteria, but recruit immune cells as well. With activation of the adaptive immune system comes additional inflammation, which may counteract the benefits of lowering bacterial sources of inflammation.

To closely examine this process, we are specifically focusing on the balance between inflammation due to bacterial injury vs inflammation due to immune cell recruitment. We have built a simple Boolean network model, where every major system in the skin is modeled as a binary node (may only be 0 or 1). Interactions between systems can now be modeled using digital logic, instead of ordinary differential equations (ODEs). The results of our model suggest that the topical application of cathelicidin could indeed be a viable treatment for acne, showing that although cathelicidin will increase inflammation from the immune system, the inflammation caused by P. acnes injuring cells will decrease, balancing out the overall effect.

Boolean Network Modeling

The most common method of mathematical modeling employed by iGEM teams to date is through deterministic models based on sets of ordinal differential equations (ODEs) and using concentrations of important molecules in the cell as parameters. While these models work very well for small, simple systems, there are three main issues with an ODE model, because 1) models are limited by the number of known rate constants, 2) actually digging and finding rate constants for every interaction is exponentially more difficult in a large scale models and 3) a complex model containing too many constraints will create an unsolvable set of ODEs. Consequently, for larger scale networks many iGEM teams have recently turned to linear piecewise differential equations or stochastic models to use simplification or random iterations to drive solutions. However, if matrix algebra isn’t quite your cup of tea and you can’t stand the resemblance of Monte Carlo simulations to a sandwich (where’s the au jus?), rest assured – there are other options out there.

Boolean network-based computer simulations provide a powerful alternative approach without dependence on rate constants and are ideally suited to modeling large scale networks. Because biological circuits typically involve rapid signaling cascades, the interaction can be treated like electrical impulses. Therefore, each connection of a Boolean network is an instantaneous, rate-independent reaction just like an electrical element. As shown in Figure 2, cellular signals can be modeled as digital bytes rapidly alternating between two states – either 0 or 1. Instead of confronting a highly complex network, where every node is connected with a different rate constant, nodes in a Boolean network are connected through digital logic, which is relatively simpler. Once all the nodes have been established, they can be connected with AND, OR, and NOT logical gates, just like a digital circuit. Once the circuit diagram has been built from AND, OR, and NOT gates, the model is ready to begin churning out data. And the best part? Not a single number was needed to initialize the model!

For more information, see “Boolean modeling of biological regulatory networks: A methodology tutorial.” Assieh Saadatpour, Réka Albert. Methods 62 (2013) 3–12.