Poster M10 - Modeling Synergistic Effects of Integrin and TGF-β Signaling in Epithelial Mesenchymal Transition

Introduction: Within the extracellular matrix (ECM), TGF-β1 is found in a latent, inactive form that is shielded from binding to receptors. When synthesized, TGF-β1 is contained within two prodomain strands, known as Latency Associated Peptide (LAP), which encircle TGF-β1 to form what is known as the Small Latent Complex (SLC). The SLC is then further bound to another polypeptide known as Latent TGF-β Binding Complex (LTBP), forming the Large Latent Complex (LLC). TGF-β1 is capable of inducing a variety of cellular signaling processes: immunomodulation, apoptosis, and of particular interest, epithelial-mesenchymal transition (EMT). For TGF-β1 to contribute to EMT, it has to be released into a soluble form which can bind to receptors on the cell membrane; research has shown that this is mediated by integrins: the binding of integrins ɑv and ɑ5 to LAP and the soluble ECM protein fibronectin (FN) respectively triggers the activation of motor proteins that contribute to the assembly of ECM and contract to release soluble, active TGF-β1 from the latent complexes.
While there is extensive research on the roles of TGF-β1 and integrin signaling in inducing EMT, the synergistic effects of the two remain unexplored. To investigate the combinatorial effects of the two, we developed a deterministic model simulating the integrin-mediated unfolding of the complexes that release active TGF-β1 that can then bind to receptors to induce EMT signaling. Our model builds on previous TGF-β1 signaling models, improving the representation of extracellular activation.

Materials and

Methods: We developed a system of ordinary differential equations (ODEs) describing the interactions between the various species involved in the release of soluble TGF-β1. Each species that is known to exist in cells prior to EMT induction was assigned a basal creation term, a destruction term, and a series of terms corresponding to the influence of other species in the model. We separate each TGF-β1 complex as individual species, resulting in 13 equations in total.
Initial values were determined using a conservation of mass equation and in vitro qPCR data. At steady state, there would be no net change in the concentration of each species; therefore, running the system at steady state should result in outputs of zero. Unknown reaction rates were estimated by optimizing the model to expected steady state values. The system of ODEs was solved over a duration of 24 hours using the ode23 function in MATLAB, which implements the Runge-Kutta method, allowing us to see changes in concentration over time.

Results, Conclusions, and Discussions: The use of computational models is invaluable in uncovering information that can be difficult to find using experimental methods. Additionally, computational models can serve as a starting point for discovery that can then be validated through experimentation. In the case of our model, we were able to simulate integrin mediated TGF-β1 latency complex unfolding and subsequent onset of EMT over a period of 24 hours. Once parameterized, the model yielded steady state concentration values similar to those obtained from experimental methods. Notably, optimization resulted in reversal rates of 0 for integrin-complex formation. This would imply that the formation of complexes between integrin ɑv and LLC, and integrin ɑ5 and FN is highly favored and suggests that the complexes do not separate during the progression of EMT.
The model that we have created can be used for further analysis of and experimentation with the integrin-mediated signaling that contributes to EMT induction. The specific role of each state variable in the system can be investigated, as can variations in concentrations. This can serve as the basis for new directions in further experimental work and as the groundwork potential therapeutic applications.

Acknowledgements (Optional):