Random Parameter Sampling of a Generic Three-Tier MAPK Cascade Model Reveals Major Factors Affecting Its Versatile Dynamics

The mitogen-activated protein kinase (MAPK) pathway is considered to be a central block in many biological signaling networks. Despite the common core cascade structure, the activation of MAPK in different biological systems can exhibit different types of dynamic behaviors. Computer modeling may help to reveal the mechanisms underlying such variations. However, so far most computational models of the MAPK cascade have been system-specific, or to reflect a particular type among the wide spectrum of possible dynamics. To obtain a general and integrated view of the relationship between the dynamics of MAPK activation and the structures and parameters of the MAPK cascade, we constructed a generic model by comparing previous models covering different specific biological systems. We assumed that reliable qualitative results could be predicted through a qualitative model with pseudo parameters. We used randomly sampled parameters instead of a specific set of “best-fit” parameters to avoid biases towards any particular systems. A range of dynamics behaviors for MAPK activation, including ultrasensitivity, bistability, transient activation and oscillation, were successfully predicted by the generic model. The results indicated that the steady state dynamics (ultrasensitivity and bistability) was jointly determined by the three-tiered structure of the MAPK cascade and the competitive substrate binding in the dual-phosphorylation processes of the central components, while the temporal dynamics (transient activation and oscillation) was mainly affected by the upstream signaling pathway and feedbacks. Moreover, MAPK kinase (MAPKK) played more important roles than the other two components in determining the dynamics of MAPK activation. We hypothesize that this is an important and advantageous property for the regulation and for the functional diversity of MAPK pathways in real cells. Finally, to assist developing generic models for signaling motifs through model comparisons, we proposed a reaction-based database to make the model data more flexible and interoperable.