Skip to main content
. 2014 Sep 4;10(9):e1003795. doi: 10.1371/journal.pcbi.1003795

Figure 7. Analysis of error evolution with upgraded topology suggests A375-specific signaling rearrangement.

Figure 7

Here, we implement the modeling concept illustrated in Figure 2F. (A) Error evolution displayed as errors for models trained with 0–96 h dataset (abscissas) against errors for models trained with 0–48 h dataset (ordinates). Each point represents a model trained to experimental data upon DMSO (triangle), U0126 (square), AZD6244 (circle) and Sorafenib (diamond). All models with errors at 96 h and 48 h below 0.2 RMSE capture the trend in the data (see Figure S2 for simulations and error calculations). Hence, models above 0.2 RMSE suggest a topological disagreement. The diagonal shows the models, whose errors exhibited no change upon retraining and simulation. CREB and ERK1/2 upon treatment with specific MEK1 inhibitor AZD6244 show a high error both at 48 h and 96 h, indicating that a disagreement in MEK1-ERK1/2 regulation is present from the first acquirement onwards. This disagreement is not present with specific MEK1 inhibitor U0126. The region of confidence constrains the models that have simultaneously a low error at 48 h and a high error at 96 h. This is the case of ERK1/2 both in control conditions and Sorafenib, which suggests a potential rearrangement in the canonical MAPK pathway. (B) Plotting ERK1/2 trajectory for retrained simulations and experimental data supports the suggestion by the model that MEK1 regulation follows canonical pathway implemented in the prior knowledge network upon treatment with specific inhibitor U0126, but (C) not upon specific MEK1 inhibitor AZD6244, indicating differential mechanisms of action of both inhibitors. (D) Trajectory plot for ERK1/2 confirms lack of regulation by MEK1 throughout early and late time points upon treatment with both Sorafenib inhibitor (upper panel) and control conditions (lower panel). The observation that canonical MEK1-ERK1/2 regulation is present upon U0126 suggests that he rearrangement is upstream of ERK1/2, and as supported by literature it could be specific in A375 melanoma cell line.