Research ArticleCancer therapy

Modeling chemotherapy-induced stress to identify rational combination therapies in the DNA damage response pathway

See allHide authors and affiliations

Science Signaling  24 Jul 2018:
Vol. 11, Issue 540, eaat0229
DOI: 10.1126/scisignal.aat0229
  • Fig. 1 Illustration of the computational model describing the DNA damage response signaling, gene expression changes, cell cycle stages, and cellular responses.

    The model includes a representation of the cell cycle (stages G1, S, and G2) and the DNA damage repair signaling pathways (gray boxes). Chemotherapy dose and the cell cycle composition before treatment are the model inputs. Doxorubicin leads to DSBs in G2 phase, whereas SN38 and gemcitabine induce SSBs in S phase. DSBs and SSBs trigger distinct individual branches of the signaling model. Stage-dependent cell cycle checkpoints (stop signs) and different DNA repair pathways (NHEJ, HR, and NER/BER) link the signaling model back to the cell cycle model. The model includes p53-dependent apoptosis and drug effects on proliferation and survival signaling. Node-labeled “m_” indicates mRNA. The reactions are labeled with ν1 to ν102; the complete mathematical description of the computational model is available in data file S1. DDR, DNA damage response. qRT-PCR, quantitative reverse transcription polymerase chain reaction.

  • Fig. 2 Experimental data sets used to train the computational model.

    (A) Proliferation and cell death: exemplary microscopy images of untreated U2OS cells at the 24-hour time point and of cells treated with 1 μM doxorubicin at the 48-hour time point (additional images shown in fig. S1). (B) Distribution of cell cycle stages at baseline quantified by flow cytometry measuring DNA content [DAPI (4′,6-diamidino-2-phenylindole)] and DNA synthesis rate [EdU (5-ethynyl-2′-deoxyuridine)]. (C) DNA damage response signaling in U2OS cells was exposed to 0, 0.01, 0.05, and 0.1 μM gemcitabine, SN38, and doxorubicin; the abundance of phosphoproteins was assessed over the subsequent 24 hours. pATM, pCHK1, p53, and p21 were measured by Western blot; representative examples are shown. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

  • Fig. 3 Trained computational model describes signaling and proliferation data as a function of time and dose.

    (A and B) Quantification of DNA damage response protein phosphorylation by Western blot data (represented in Fig. 2C), analyzed as response to 0.01, 0.05, and 0.1 μM chemotherapy over 24 hours (A), and dose-response curves at the 6-hour time point (B). (C) Imaging-based quantification of cell proliferation dynamics with and without chemotherapy exposure by counting nuclear reporter–positive cells. Data are normalized to time point zero. Dots represent data of separate experiments; each replicate was individually used to calibrate the computational model. Lines indicate the corresponding simulation trajectories. Further experimental details are described in Methods.

  • Fig. 4 Sensitivity analysis of cellular responses to perturbation of the DNA damage response signaling in the presence of chemotherapies.

    (A and B) For the chemotherapies, a low dose of 0.001 μM (A) and a high dose of 1 μM (B) were simulated. The inhibition of eight molecular targets within the DNA damage response pathway (ATM, ATR, DNA-PK, CHK1, CHK2, p53, p21, and CHK1/2) was simulated. The effect of chemotherapies is simulated as monotherapy or in combination (bars) and shown on the x axis. The outcome is quantified as log10 fold change of cell number after 3 days of the drug exposure. The vertical dashed line is aligned to the effect of the chemotherapy alone (first bar in each and dashed lines), which is the comparator for the effect of the combinations.

  • Fig. 5 Drug response profiles and experimental validation.

    (A and B) Drug response profiles for gemcitabine and doxorubicin predicted by the computational model (A) and experimental validation using small-molecule inhibitors of ATR, CHK1, DNA-PK, and ATM (B). For the analysis, the drug and inhibitor concentrations were varied over a wide range, and outcome is quantified after 3 days. The black line indicates the stasis line as predicted by the computational model. Blue stars represent the low-dose chemotherapy, and the red stars indicate the high-dose chemotherapy in combination with a targeted inhibitor as chosen for the single-dose combinations (Fig. 4). The dashed white boxes indicate areas where the model prediction and experimental data are discordant.

  • Fig. 6 Dynamics of live and apoptotic cells treated with potentiating drug combinations.

    (A and B) Proliferation and apoptosis assessed in U2OS cells cultured with 0.04 μM gemcitabine and 1 μM ATR inhibitor VE-822 (A), or 0.0016 μM doxorubicin and 0.5 μM DNA-PK inhibitor KU-0060648 (B). The experimental data (dots, representative of separate experiments) are compared to the response simulated by the computational model (solid lines) together with their associated uncertainties (shades).

  • Fig. 7 Response to SN38 and VE-822 (ATR inhibitor) in the U2OS and MS-751 cervical cancer lines in vitro and MS-751 cervical cancer model in vivo.

    (A and B) Time course data of cell number measurements in U2OS cells (A) and MS-751 cells (B) after treatment with 0.01 μM SN38, 1 μM VE-822, or in combination (n = 4), overlaid with the corresponding model simulation. Drugs were added at the 24-hour time point. (C) Growth of MS-751 xenografts in mice treated as indicated on days 17, 24, 31, and 38 (dashed lines) after tumor implantation. nal-IRI was given at a dose of 5 mg/kg and liposomal ATR inhibitor Ls-VE-822 at 20 mg/kg intravenously. For the combination treatment, Ls-VE-822 was injected 24 hours after the nal-IRI injection (n = 10 mice per group). PBS, phosphate-buffered saline.

Supplementary Materials

  • www.sciencesignaling.org/cgi/content/full/11/540/eaat0229/DC1

    Fig. S1. Additional images from live-cell imaging of U2OS cell line.

    Fig. S2. Results of 500 independent model fits from randomized starting parameters.

    Fig. S3. Identifiability analysis of model parameter by analysis of likelihood profiles.

    Fig. S4. Drug response profiles for all chemotherapy and targeted inhibitors predicted by the computational model.

    Fig. S5. Additional, experimentally measured drug response profiles for chemotherapy and targeted inhibitors.

    Fig. S6. Western blot data for U2OS and MS-751 cell lines treated with chemotherapies, ATR inhibitor, or combinations.

    Fig. S7. Additional drug response profiles comparing the potentiating effect of gemcitabine and ATR inhibition between the U2OS cell line and a panel of lung cancer cell lines.

    Table S1. Quantification of model misfit for individual experimental conditions.

    Table S2. Chemotherapy drugs and targeted inhibitors used.

    Table S3. Antibodies used for Western blotting.

    Movie S1. Time-lapse video of U2OS cells treated with a combination of gemcitabine and ATR inhibitor.

    Data file S1. Model parameters and computational analyses.

  • The PDF file includes:

    • Fig. S1. Additional images from live-cell imaging of U2OS cell line.
    • Fig. S2. Results of 500 independent model fits from randomized starting parameters.
    • Fig. S3. Identifiability analysis of model parameter by analysis of likelihood profiles.
    • Fig. S4. Drug response profiles for all chemotherapy and targeted inhibitors predicted by the computational model.
    • Fig. S5. Additional, experimentally measured drug response profiles for chemotherapy and targeted inhibitors.
    • Fig. S6. Western blot data for U2OS and MS-751 cell lines treated with chemotherapies, ATR inhibitor, or combinations.
    • Fig. S7. Additional drug response profiles comparing the potentiating effect of gemcitabine and ATR inhibition between the U2OS cell line and a panel of lung cancer cell lines.
    • Table S1. Quantification of model misfit for individual experimental conditions.
    • Table S2. Chemotherapy drugs and targeted inhibitors used.
    • Table S3. Antibodies used for Western blotting.
    • Legend for movie S1
    • Legend for data file S1

    [Download PDF]

    Other Supplementary Material for this manuscript includes the following:

    • Movie S1 (.mp4 format). Time-lapse video of U2OS cells treated with a combination of gemcitabine and ATR inhibitor.
    • Data file S1 (.pdf format). Model parameters and computational analyses.

Stay Connected to Science Signaling

Navigate This Article