Spatio-temporal modeling of signaling protein recruitment to EGFR

Abstract

Background: A stochastic simulator was implemented to study EGFR signal initiation in 3D with single molecule detail. The model considers previously unexplored contributions to receptor-adaptor coupling, such as receptor clustering and diffusive properties of both receptors and binding partners. The agent-based and rule-based approach permits consideration of combinatorial complexity, a problem associated with multiple phosphorylation sites and the potential for simultaneous binding of adaptors.

Results: The model was used to simulate recruitment of four different signaling molecules (Grb2, PLCgamma1, Stat5, Shc) to the phosphorylated EGFR tail, with rules based on coarse-grained prediction of spatial constraints. Parameters were derived in part from quantitative immunoblotting, immunoprecipitation and electron microscopy data. Results demonstrate that receptor clustering increases the efficiency of individual adaptor retainment on activated EGFR, an effect that is overridden if crowding is imposed by receptor overexpression. Simultaneous docking of multiple proteins is highly dependent on receptor-adaptor stability and independent of clustering.

Conclusions: Overall, we propose that receptor density, reaction kinetics and membrane spatial organization all contribute to signaling efficiency and influence the carcinogenesis process.

Figure 1

Hierarchy of EGFR tail binding partners predicted by coarse-grained molecular modeling method. (A) Coarse-grained molecular modeling of EGFR C-terminal tail attached to the asymmetric model of EGFR kinase domain obtained from Zhang et al [45]. (B) Full-length coarse-grained molecular models of Shc (purple), Grb2 (yellow), PLCγ1 (green), and Stat5 (cyan). (C) Simplified systematic interaction profiling of the EGFR tyrosine family used in our study. In this model, EGFR has four cytoplasmic interaction partners, one binding site for each of Stat5 and Grb2, and two binding sites for each of Shc and PLCγ1. Coarse-grained docking method was used to develop the hierarchy of EGFR tail binding partners. Some possible docking methods between the four adaptors and EGFR include (D) PLCγ1 docked to pY992, (E) PLCγ1 docked to pY1173, (F) PLCγ1, Shc, and Grb2 simultaneously docked to EGFR at pY992, pY1148, and pY1068, respectively, and (G) Stat5, Shc, and Grb2 simultaneously docked to EGFR at pY992, pY1148, and pY1068, respectively.

Figure 2

Analysis and simulation of EGFR tyrosine residues phosphorylation kinetics. A431 cells were serum-starved and treated with batimastat for "resting" condition shown in (A), or treated thereafter for 2 min with 20 nM EGF (B). "Rip-Flips" were prepared and membranes immunogold-labeled with anti-EGFR antibodies. Inset in both (A) and (B) confirms EGFR clustering by Hospkins test. Bars, 0.1 μm. Western blotting method was used to analyze phosphorylation kinetics of EGFR tyrosine residues (C) Y992, (D) Y1068, (E) Y1173, and (G) Y1148. Bands were quantified by densitometry and plotted as density of the bands. (F, H) Results of simulations (dashed lines) agree well with the "fast" kinetics and "slow" kinetics data (solid lines), using parameter values estimated by fitting to the data.

Figure 3

Analysis and simulation of the reaction kinetics between the four adaptors and EGFR. (A-B) Membrane sheets were prepared from serum-starved, batimastat-treated A431 cells without (A) or with EGF stimulation (B). Sheets were labeled with 5 nm gold reagents recognizing Shc. Circles in (A, B) highlight Shc label on these membranes. Bars, 0.1 μm. (C-F) Quantitative values of Shc, Stat5, PLCγ1, and Grb2 immunogold labeling on 3 μm² area of membrane, reported as an average of at least 10 membranes. Blots in C-F show results of fractionation experiments, where crude cytosol and membrane fractions were prepared, proteins separated by SDS-PAGE and membranes blotted for Shc, Stat5, PLCγ1 and Grb2. In (G-I), blots report co-precipitation of Shc, Stat5 and PLCγ1 with EGFR over a time course of EGF stimulation. Bands were quantified by densitometry and plotted as density of the bands. In (J-M), simulations of reaction kinetics between the four adaptors and EGFR using experiment-fitted values produce results (black solid line) similar to experimental data (grey dashed line).

Figure 5

Comparison of the sharing and competing docking models. Simulations in (A-B) used the parameter values fitted to A431 cells data, while simulations in (C) used Kholodenko's parameter values [36]. (A) Results of simulations show similar number of adaptors docked to EGFR at steady state, when receptors are overexpressed and clustered. (B) Results of simulations for sharing and competing models at steady state, when receptors are at normal expression levels (50,000 receptors/cell) and either clustered or random. Receptor clustering increases the efficiency of adaptor retainment to EGFR, but sharing does not contribute further efficiency. (C) Results show that use of slow dissociation rates produces a dramatic increase in the shared docking of adaptors on EGFR tails, simulated for 50,000 clustered receptors in the spatial-stochastic model.

Figure 4

Effect of receptor clustering and density on efficiency of adaptor containment near the membrane. (A) Illustration of particles diffusing in and out of preferred domains or islands. Colored traces show 10 second trajectories of 3 diffusing EGFR particles. Receptors have greater probability to enter preferred domains, where they diffuse 3x slower. When outside of domains, particle diffusion is unconstrained. (B) Particles are clustered at every time step when diffusion is governed in silico by the domain approach. Inset shows results of Hopkins test, confirming clustering is significant. Plots in (C) show that receptor clustering increases the efficiency of Grb2 retainment to EGFR at normal levels of expression (50,000/cell). The frequency of Grb2 docking to another receptor after dissociating from a previous binding event is low when receptors are in the randomly distributed (black line). Increasing receptor cluster size increases the efficiency of adaptor containment near the membrane (grey solid line, 6.57 receptors/cluster; grey dashed line, 100 receptors/cluster). In contrast, if EGFR is overexpressed the plots (D) show no difference in Grb2 rebinding.