Computational modelling identifies primary mediators of crosstalk between DNA damage and oxidative stress responses

Abstract

Cells exposed to toxicants, such as drugs, activate a wide variety of stress pathways, often simultaneously. Two important pathways that can influence cell fate and consequently adverse reactions are the oxidative stress response (OSR) and the DNA damage response (DDR). Previous studies have presented evidence of crosstalk between the OSR and DDR. We aimed to develop computational models to describe experimentally observed dynamics of both OSR and DDR proteins in liver (HepG2) cells in vitro upon exposure to various concentrations of either diethyl maleate (DEM; an agent primarily invoking oxidative stress) or etoposide (an agent primarily causing DNA damage). With these models, we aimed to identify the key interactions that cause crosstalk and their importance in describing protein dynamics. We developed a new model for the OSR pathway, coupled it to a previously developed model for the DDR pathway, and extended the resulting combined model based on multiple potential modes of crosstalk described in the literature. The different models were applied to previously published data of HepG2 GFP-reporter cells with time-dynamic information on the relative amount of proteins important for the OSR (NRF2, SRXN1) or DDR (p53, p21, BTG2 and MDM2). The developed models properly described key OSR and DDR protein dynamics, and in silico knockdowns of key model components in most cases led to a moderate effect on the connected pathway. The largest effect occurred after knockdown of p21, which resulted in a substantial decrease in NRF2 and SRXN1. We expect these models could play a role in adversity predictions by coupling our models with other models that predict cell fate or adversity based on the expression of specific proteins.

Fig 1. DDR model describes data from HepG2 cells exposed to etoposide.

A) Simulations (red) of the DDR model for BTG2, MDM2, p21 and p53 combined with data (black line represents the mean, grey points the measurements per replicate) for these proteins after exposure to 2.5 and 25 µM etoposide. B) Schematic overview of the model for the DDR. C) Data for NRF2 and SRXN1 HepG2 reporter cell lines after exposure to six concentrations of etoposide. The colour represents the dose and the shaded area indicates the standard deviation across replicates.

Fig 2. OSR model describes data from HepG2 cells exposed to DEM when NRF2 stimulates KEAP1 production.

A) Schematic overview of the model for the OSR. B) Experimental data along with simulations of the OSR model with (red) and without (blue) NRF2-dependent KEAP1 production. The black line represents the mean, and the grey points the measurements per replicate after exposure to 20 and 200 µM DEM. C) Data for MDM2 and p53 after exposure to eight concentrations of DEM. The colour indicates the dose and the shaded area indicates the standard deviation.

Fig 3. Crosstalk models to describe etoposide-induced OSR activity.

A) Schematic overview of the model for the DDR and OSR upon DNA damage induction. Pathways are connected by crosstalk via p21 and p53p. B) Simulations of four different model versions (coloured lines) for NRF2 and SRXN1 combined with data (black line represents the mean, grey points the measurements per replicate) for these proteins after exposure to 2.5 and 25 µM etoposide.

Fig 4. Crosstalk models to describe DEM-induced DDR activity.

A) Schematic overview of the model for the DDR and OSR upon oxidative stress. Pathways are connected by crosstalk via oxidative stress, NRF2 and SRXN1. B) & C) Different model versions without (B) and with (C) adapted MDM2-dependent p53 degradation. Simulations (coloured lines) for MDM2 and p53 combined with data (black line represents the mean, grey points the measurements per replicate) for these proteins after exposure to 20 and 200 µM DEM.

Fig 5. In silico knockdown predictions on the basis of crosstalk models.

A-B) Simulation of best crosstalk model versions (blue or pink) and combined crosstalk models (red) for exposure to 25 µM etoposide (A) or 200 µM DEM (B). Simulations (coloured lines) and data (black line represents the mean, grey points the measurements per replicate) are shown for MDM2, NRF2, p53 and SRXN1. C-D) Simulations with (coloured lines) or without knockdowns (black line) for MDM2, NRF2, p53 and SRXN1, for exposure to 25 µM etoposide (C) or 200 µM DEM (D).