ATR Inhibition Broadly Sensitizes Soft-Tissue Sarcoma Cells to Chemotherapy Independent of Alternative Lengthening Telomere (ALT) Status

Only few drugs have shown activity in patients with advanced soft-tissue and the median overall survival is only 18 months. Alterations of genes involved in the DNA damage repair pathway have been associated with sarcoma risk and prognosis. ATR plays a crucial role in maintaining genomic integrity by responding to a large spectrum of DNA damage, including double strand breaks (DSBs) that interfere with replication. The objective of this study is to evaluate the pre-clinical activity of ATR inhibition in soft tissue sarcomas (STS). We explored the ability of the ATR inhibitor, VE-822, to prevent chemotherapy-induced intra-S-phase checkpoint activation and evaluated the antitumor potential of this combination in vitro and in vivo in STS cell lines and in a patient-derived xenograft model. The combination of VE-822 and gemcitabine in vitro was synergistic, inhibited cell proliferation, induced apoptosis, and accumulated in the S phase of the cell cycle with higher efficacy than either single agent alone. The combination also resulted in enhanced γH2AX intranuclear accumulation as a result of DNA damage induction. These effects were unrelated to the alternative lengthening of telomeres pathway. In vivo, the combination of VE-822 and gemcitabine significantly enhanced tumor growth inhibition and progression-free survival in an aggressive model of undifferentiated pleomorphic sarcoma. The combination of ATR inhibitor and chemotherapy is beneficial in pre-clinical models of soft-tissue sarcoma and deserves further exploration in the clinical setting.


VE-822 has antiproliferative activity and induces apoptosis and S-phase cell cycle delay in STS cells.
We studied the sensitivity of 8 STS cell lines to VE-822 and gemcitabine. The IC50 values for VE-822 ranged between 1.3 and 12 µM (Table 1). All the cell lines had impaired TP53 pathway has a result of TP53 mutation/deletion or MDM2 amplification except for two cell lines (IB114 and IB128).
We next assessed the impact of ATR inhibition on cell cycle progression. In response to VE-822, cells accumulated in G0/G1 consistent with G1/S cell cycle arrest. By using the Annexin V-based detection assay, apoptosis was also detected under these conditions (Fig. 1C,D).

Sensitivity of STS cells to VE-822 is not related to alternative lengthening of telomeres.
Given that the cell death induced by ATR inhibitors was suggested to be highly selective for cancer cells that rely on alternative lengthening telomere (ALT) 6 , we assessed the ALT status of our panel of cell lines by using the C-Circle assay as previously described 7 . We then explored the correlation between VE-822 sensitivity and ALT status. The VE-822 IC 50 value ranged from 1.5 to 12 μM for ALT-positive cell lines and from 1.3 to 8 μM for the ALT-negative cell lines with no significant difference between the two groups (Table 1). VE-822 is synergistic with gemcitabine in STS cell lines, increase DNA damage and prevents the checkpoint activation elicited by gemcitabine. The current clinical development of ATR inhibitors is based on combination regimens with additional therapies. We studied the effects of the combination of gemcitabine and VE-822 on the viability of STS cells. Eight STS cell lines were exposed during 72 hours to different combinations of both agents at a constant ratio of 1: 1 VE-822 and gemcitabine were mixed and diluted serially (usually 2 fold-serial dilutions with several concentrations above and below the IC50 for the two drugs). Combination Indices (CIs) were determined according to Chou et al. 8 . The results are described in Table 1. Interestingly, we observed an additive or synergistic effect in all the STS cell lines. We then assessed the level of DNA damage induced by the combination. By analyzing γ-H2AX expression, we found that the combination of VE-822 and gemcitabine induced significantly higher levels of DNA damage than either drug used as single agent alone (Fig. 2). We next investigated whether VE-822 could counteract the checkpoint activation elicited by gemcitabine. Importantly, in all the cell lines tested, VE-822 reduced the level of gemcitabine-induced CHK1 phosphorylation, the downstream ATR target (Fig. 3).

VE-822 and gemcitabine combination induces apoptosis and cell cycle arrest in STS cell lines.
We studied the effects of VE-822 and gemcitabine combination on apoptosis induction after 72 h of drug exposure as well as cell cycle effects after 48 h of treatment. We observed that the drug combination (picomolar amounts of Gemcitabine and micromolar amounts of VE-822) increased the rate of apoptosis in comparison with the drugs used as single agent in all the cell lines but one (IB128, Fig. 1). Furthermore, accumulation in S phase was also observed after treatment with the drug combination ( Fig. 1) in all the cell lines except in IB136.

VE-822 and gemcitabine combination reduces tumor growth in a patient-derived xenograft model of undifferentiated pleomorphic sarcoma.
To further validate our in vitro results, we performed in vivo studies to test the antitumor effects of the VE-822 and gemcitabine combination. Xenografts were generated by subcutaneous implantation in ragγ2C−/− mice of one patient derived undifferentiated pleomorphic sarcoma. Animals were randomized in 4 groups and treated for 3 weeks. These groups included control (NaCl 0.9%), VE-822 (VE-822 alone; 60 mg/kg every day during 3 weeks), gemcitabine (gemcitabine alone; 30 mg/Kg IP, one time per week), and combination. After three weeks of treatment we observed a significant effect on progression free survival (evaluated as the time span from the treatment start and the doubling of the initial tumor volume), median time to doubling was 14.5 days for combination, 9.9 days for VE-822 (p = 0.0014) 10.3 days for gemcitabine, and 8.4 days for the vehicle (Fig. 4). No signs of toxicity were observed with the combination treatment.

Discussion
Genome instability is a crucial hallmark of cancer. Physiologically, DNA damage response pathways maintain genome integrity by repairing DNA damage. Cancer cells are characterized by defects in DDR which results in increased mutational load, replication stress and genome instability. Chibon et al. have shown previously that high level of genome instability is associated with adverse outcome in patients with sarcomas 4 . ATR is an essential DDR kinase activated in cancer cells with high level of replication stress. Therefore, several drugs targeting ATR are currently under clinical development. The first-in-class ATR inhibitor, VX-970 (Vertex Pharmaceutical; now M6620, Merck), has shown target modulation and promising preliminary signs of clinical activity in early-phase studies assessing its safety and efficacy as a single agent and in combination with chemotherapy 9 . Two other oral ATR inhibitors, AZD6738 (Astrazeneca), and BAY-1895344 (Bayer) are currently being tested in phase I or II trials as monotherapy or in combination with DDR agents, radiotherapy, chemotherapy or immunotherapeutics 10,11 . We report here the first study investigating the preclinical activity of ATR inhibition in a panel of pre-clinical models representing the most frequent histological subtypes of STS. Interestingly, we found that all the cell lines tested here displayed some degree of basal ATR activation as illustrated by the constitutive phosphorylation of CHK1 in sarcoma cells in the absence of treatment/damage induction. A recent study has shown that during unperturbed replication, some ATR and CHK1 molecules are activated to limit the activation of the replicative helicase in regions of ongoing DNA replication 12 . Further studies are needed to understand more in depth how is ATR and CHK1 signaling activated in unperturbed sarcoma cells. We found that all the cell lines tested displayed various degree of sensitivity to VE-822 with a median IC50 of 4.9 (range 1. [3][4][5][6][7][8][9][10][11][12]. A recent study has suggested that ATR inhibitors are selectively active in cancer cells that employ the alternative lengthening telomere (ALT) pathway 7 . ALT is particularly common in bone and soft-tissue sarcomas 13 . Contrary to what was observed in osteosarcoma models, our results do not indicate a general hypersensitivity of ALT-positive STS cells when ATR is inhibited. Interestingly, there are in agreement with a recent study which did not identify any correlation between the ALT-or telomerase status of various cell lines and response to the ATR inhibitor VE-821 14 . In that study, the authors compared also the ATR inhibitor sensitivity in isogenic cell lines, in which ALT was active or suppressed and found absolutely no differences. Altogether, these findings do not support a synthetic lethality of ATR targeting in alternative lengthening of telomeres-dependent tumors and suggest that other pathways are involved in sensitivity to ATR inhibition. When the activity of ATR is inhibited, maintenance of genome integrity becomes dependent on functional TP53, with TP53 being essential for arresting cell cycle progression to permit repair 15 . ATR, therefore, represents a promising synthetically lethal target for TP53 deficiency 16,17 . Several studies have confirmed such synthetically lethal interaction by deleting ATR in TP53-deficient mice 17 and by inhibiting its activity in tumor cell lines, which resulted in death induction of cells harboring TP53 defects [18][19][20] . TP53 pathway abrogation appears as a pivotal event in soft tissue sarcoma oncogenesis 21 . However, our results indicate that defective TP53 as result of deletion or mutation or MDM2 gene amplification do not confer greater sensitivity of STS cells to VE-822. This is in line with a recent study investigating the role of TP53 in sensitivity to four different ATR inhibitors in several models of osteosarcomas, breast, and colorectal cancers 22 . The authors were not able to find a correlation between TP53 status and ATR inhibitor sensitivity even if gemcitabine sensitization was more pronounced in TP53-defective models. Altogether, these data suggest that TP53 is probably not a key determinant of the effect of ATR inhibition in tumor cells but only one contributor among other factors depending on the tumor type and the cellular context.
As even for the most sensitive STS lines, IC50 values were above 1 μM, we reasoned that achieving anti-tumor efficacy in vivo using VE-822, would be unlikely. Therefore, we sought to investigate the synergistic activity of VE-822 and gemcitabine when used in combination in STS models. In the present study we observed a synergistic or additive effect in all the cell lines tested. VE-822 strongly potentiated sub-IC 50 levels of gemcitabine to induce S-phase arrest in the majority of the cell lines tested. Moreover, VE-822 synergized with gemcitabine to induce apoptosis in STS cells and does not only inhibit gemcitabine induced checkpoint activation, but also pre-existing CHK1 phosphorylation and/or CHK1 protein levels in general, while enhancing gemcitabine-induced DNA damage. We validated these in vitro results in the in vivo setting by using a patient-derived xenograft model of UPS, the most aggressive STS subtype 23 . As observed in vitro, the combination of VE-822 with gemcitabine significantly inhibited tumor growth in comparison with both drugs used as single agent.
We report here pre-clinical evidence that the combination of an ATRinh and chemotherapy is synergistic in soft-tissue sarcomas. Our results provide a rational for a clinical trial investigating the combination of gemcitabine and an ATR inhibitor in patients with advanced STS.

Methods
All methods were performed in accordance with the relevant French and European Union guidelines and regulations.
Cells and cell culture. All of the STS cell lines used in this study were derived from surgical specimens of patients with STS who underwent surgery at Institut Bergonié, Bordeaux, France and who gave written informed consent (Table 1) as previously described 24 . Each cell line was characterized by array comparative genomic hybridization for every 10 replicates to verify that its genomic profile was still representative of the originating tumor sample. Cells were grown in RPMI medium 1640 (Sigma Life Technologies, Saint Louis, MO) in the presence of 10% fetal calf serum (Dutscher, France) in flasks. Cells were maintained at 37 °C in a humidified atmosphere containing 5% CO 2 .