Towards DNA-damage induced autophagy: A Boolean model of p53-induced cell fate mechanisms

Highlights

• First synthetic gene network of multiple cell fates induced by DNA damage based on the U87 glioblastoma cell line.

• The model describes cell fate determination for three phenotypes: apoptosis, senescence and autophagy.

• The model was generalized to describe other three different cell lines: MCF-7, A549 and U2OS.

• For p53-deficient cell lines the model contemplates the alternative AMPK pathway which compensates this deficiency.

• The approach is useful for designing experiments of loss and gain of function perturbations to induce a specific phenotype.

Abstract

How a cell determines a given phenotype upon damaged DNA is an open problem. Cell fate decisions happen at cell cycle checkpoints and it is becoming clearer that the p53 pathway is a major regulator of cell fate decisions involving apoptosis or senescence upon DNA damage, especially at G1/S. However, recent results suggest that this pathway is also involved in autophagy induction upon DNA damage. To our knowledge, in this work we propose the first model of the DNA damage-induced G1/S checkpoint contemplating the decision between three phenotypes: apoptosis, senescence, and autophagy. The Boolean model is proposed based on experiments with U87 glioblastoma cells using the transfection of miR-16 that can induce a DNA damage response. The wild-type case of the model shows that DNA damage induces the checkpoint and the coexistence of the three phenotypes (tristable dynamics), each with a different probability. We also predict that the positive feedback involving ATM, miR-16, and Wip1 has an influence on the tristable state. The model predictions were compared to experiments of gain and loss of function in other three different cell lines (MCF-7, A549, and U2OS) presenting agreement. For p53-deficient cell lines such as HeLa, H1299, and PC-3, our model contemplates the experimental observation that the alternative AMPK pathway can compensate this deficiency. We conclude that at the G1/S checkpoint the p53 pathway (or, in its absence, the AMPK pathway) can regulate the induction of different phenotypes in a stochastic manner in the U87 cell line and others.

Introduction

It is well known that Tumor suppressor p53 protein (p53) is the master regulator of cell fate in response to DNA damage (see latest review by Hafner [1]). Low levels of p53 induce cell cycle arrest and/or senescence, whereas high levels induce cell death pathways [2]. However, inactivation of p53-mediated-cell death is a major step in tumor development [3]. While a role for the p53 pathway in apoptosis is better established [4], its involvement in the induction of autophagy is not fully understood. Our main interest here is to explain how DNA damage induces a given cell phenotype.

Several studies show that the DNA damage-regulated autophagy modulator 1 (DRAM1), which is regulated by p53, modulates autophagy in response to DNA damage [[5], [6], [7], [8], [9]]. This molecule seems to represent the link between the p53 pathway and autophagy induction (see the review by Mathiassen [10]). The study of Mauthe et al. [11] revealed that resveratrol-mediated autophagy depends on the downregulation of Wip1 (protein phosphatase, Mg2 + /Mn2 + dependent 1D), a p53 regulated gene in cancer cell lines such as U2OS cells and MCF-7 [11]. Zou and colleagues [12] showed that ATM (ATM serine/threonine kinase) is involved in autophagy of U87 cells [12]. Wip1 is a negative regulator of the ATM/p53 pathway, downregulation of Wip1 thereby increases this pathway activity and leads to the induction of the G1/S (and G2/M) cell cycle checkpoint, whereas its overexpression inhibits checkpoints induction [[13], [14], [15]].

In addition, the studies of Brichkina et al. [16] and Le Guezennec et al. [17] reported that the deletion or knockdown of Wip1 can activate autophagy via the ATM/AMPK/TSC2/mTORC (AMP-activated protein kinase, Tuberin, Mammalian target of rapamycin complex 1) signaling pathway. In more detail, ATM activates TSC2 via AMPK, and TSC2 inhibits mTOR inducing autophagy. Moreover, Oxidized low-density lipoprotein (OxLDL) promotes intracellular ROS production and induces oxidative DNA damage. OxLDL is a well-known regulator of cholesterol efflux in macrophages. Deletion or knockdown of Wip1 via OxLDL activated ATM-mediates atherosclerosis and is essential for the formation of foam cells and atherosclerotic plaques. Thus, these two studies suggested that the autophagy phenotype can be achieved via targeting of Wip1 in foam cells and atherosclerosis.

The MicroRNA-16 (miR-16) is a master regulator of tumor suppression and plays an essential role in response to DNA damage [18]. Lately, various studies show that the targeting of Wip1 by miR-16 can affect cell fate decision in cancer cells in response to DNA damage [[19], [20], [21], [22]]. Interestingly, Huang et al. found evidence that miR-16 is involved in autophagy induction [23]. Furthermore, miR-16 activation occurs through ATM-mediated transcription inhibiting Wip1 expression [18,20]. An already identified double-negative feedback loop involving ATM, miR-16, and Wip1 plays a role in the induction of autophagy in U2OS cells (and possibly in U87 cells) [18,20,24]. The information above suggests that these molecules play a role in autophagy regulation.

With the immense complexity of biological systems in terms of its interactive nature, the development of network-based models of these systems has been recognized as a valuable approach to study many biological processes [[25], [26], [27], [28]]. Particularly, Boolean models can provide an informative and coherent qualitative description of the network dynamics [29]. Experimentally testable predictions can be obtained using this method, which are associated to less-understood aspects of the integration of network analysis and dynamic modeling [30]. One of the main features of qualitative models is to study the influence of regulatory circuits (also known as feedback loops), which are closed paths between two or more nodes in the network. Determining the regulatory function of a circuit in the network can help to elucidate cell-fate decision processes.

Motivated by these facts, we developed a Boolean model of the G1/S checkpoint regulatory network (see Fig. 1) to investigate cell-fate induction contemplating three possible phenotypes: autophagy, apoptosis and senescence. To our knowledge this is the first model in the literature dealing with more than two phenotypes.