The p53 protein is a key mediator of the cellular response to various stress signals. In response to DNA damage, the concentration of p53 can temporally oscillate with fluctuations in both the amplitude and period. The underlying mechanism for p53 variability is not fully understood. Here, we construct a core regulatory network of p53 dynamics comprising the ATM-p53-Wip1 and p53-Mdm2 negative feedback loops. We dissect the contributions of cellular heterogeneity, intrinsic noise, and multiple forms of extrinsic noise to p53 variability in terms of the coefficients of variation of four quantities. Cellular heterogeneity greatly determines the fraction of oscillating cells among a population of isogenic cells. Intrinsic noise—fluctuation in biochemical reactions–has little impact on p53 variability given large amounts of molecules, whereas extrinsic colored noise with proper strength and correlation time contributes much to oscillatory variability in individual cells. With the three sources of noise combined, our results reproduce the experimental observations, suggesting that the long correlation time of colored noise is essential to p53 variability. Compared with previous studies, the current work reveals both the individual and integrated effects of distinct noise sources on p53 variability. This study provides a framework for exploring the variability in oscillations in cellular signaling pathways.

The tumor suppressor p53 mediates the cellular response to various stresses. It was experimentally shown that the concentration of p53 can show oscillations with short or long periods upon DNA damage. The underlying mechanism for this phenomenon is still not fully understood. Here, we construct a network model comprising the ATM-p53-Wip1 and p53-Mdm2 negative feedback loops and ATM autoactivation. We recapitulate the typical features of p53 oscillations including p53 birhythmicity. We show the dependence of p53 birhythmicity on various factors such as the phosphorylation status of ATM. We also perform stochastic simulation and find the noise-induced transitions between two modes of p53 oscillation, which increases the p53 variability in both the amplitude and period. These results suggest that p53 birhythmicity enhances the responsiveness of p53 network, which may facilitate its tumor suppressive function.