Computational modeling reveals molecular details of epidermal growth factor binding

Abstract

Background: The ErbB family of receptors are dysregulated in a number of cancers, and the signaling pathway of this receptor family is a critical target for several anti-cancer drugs. Therefore a detailed understanding of the mechanisms of receptor activation is critical. However, despite a plethora of biochemical studies and recent single particle tracking experiments, the early molecular mechanisms involving epidermal growth factor (EGF) binding and EGF receptor (EGFR) dimerization are not as well understood. Herein, we describe a spatially distributed Monte Carlo based simulation framework to enable the simulation of in vivo receptor diffusion and dimerization.

Results: Our simulation results are in agreement with the data from single particle tracking and biochemical experiments on EGFR. Furthermore, the simulations reveal that the sequence of receptor-receptor and ligand-receptor reaction events depends on the ligand concentration, receptor density and receptor mobility.

Conclusion: Our computer simulations reveal the mechanism of EGF binding on EGFR. Overall, we show that spatial simulation of receptor dynamics can be used to gain a mechanistic understanding of receptor activation which may in turn enable improved cancer treatments in the future.

Figure 1

Schematic of simulated microscopic events. Each receptor can diffuse to an empty neighboring site, react with a neighboring receptor to form a dimer, and bind ligand. All events are reversible.

Figure 2

Reactions events considered in our model as given in [14].

Figure 3

Comparison of hybrid null-event MC and ODE models in terms of (a) dimerized EGFR in the absence of ligand at a high receptor density and diffusivity (11,000 per μm², D = 2 × 10^-14 m²sec^-1) and assuming no initial dimers, and (b) EGF bound EGFR in the presence of ligand (160 nM) at low receptor density (125 per per μm²) and D = 2 × 10^-15 m²sec^-1. The reactions on the figure indicate the dominant processes responsible for the concentration trajectories. Error bars indicate 2 standard deviations obtained from 10 independent MC simulations.

Figure 4

(a) Evolution of intensity of dimerized receptors with two ligands (high intensity spots) and of monomer plus dimerized receptors with a single ligand bound (low intensity spots) along with the data of single particle tracking experiments by Sako et al. over time intervals of 20 sec. The simulations were performed for a receptor number density of 5500 per μm², a diffusivity of D = 2 × 10^-14 m²sec^-1, and 18% dimers initially. The simulation intensity has been normalized with the experimental data. (b) Comparison of predicted density of Cy3-EGF spots with experimental data of Sako et al. The densities are normalized with the value at 60 sec. Good agreement of simulations with experimental data is found. In both panels, error bars indicate 2 standard deviations obtained from 10 independent MC simulations.

Figure 5

Sequence of reactions resulting in dimerized receptors with both receptors bound to ligand for simulations of Fig. 4. All reactions are reversible.

Figure 6

Contributions of the different reaction mechanisms at 60 sec for different concentrations of EGF with (a) a receptor number density of 5500 receptors per μm² and D = 2 × 10^-14 m²sec^-1, (b) a receptor number density of 125 receptors per μm² and D = 2 × 10^-14 m²sec^-1, and (c) a receptor number density of 125 receptors per μm² and D = 2 × 10^-15 m²sec^-1.

Figure 7

Normalized sensitivity coefficients at 3 different times (20, 40 and 60 sec) calculated by introducing a 20% increase in the kinetic parameter indicated on the x-axis.