The quinoxaline di-N-oxide DCQ blocks breast cancer metastasis in vitro and in vivo by targeting the hypoxia inducible factor-1 pathway

Breast cancer is the most frequently diagnosed malignancy in women. Heterogeneous distribution of hypoxic regions is commonly detected within invasive breast tumors [14, 26]. These resistant regions are known to have larger tumor sizes, and to consist of worse grade tumors that are invasive and metastatic [25]. The use of hypoxia-activated drugs is among the leading approaches to circumvent hypoxia challenges [16]. This strategy provides selective tumor suppression, thus offering minimal toxicity to non-hypoxic tissues. Several hypoxia-activated compounds have reached clinical trials for the treatment of a variety of solid tumors [30‐36]. However, so far these drugs have shown moderate to low activity against breast cancer. Here, we document for the first time that DCQ is a hypoxia-activated cytotoxin effective against breast cancers that harbor different p53 status and have different invasive potential. We show that DCQ induces apoptosis and reduces HIF-1α regulated pathways. In human breast cancer xenograft model, DCQ attenuated metastasis to the lungs and liver and significantly reduced tumor growth as compared to control mice (DMSO-treated). Additionally, DCQ prolonged the survival of 50% of treated mice beyond 60 days, by that time, however, 90% of untreated mice have already died. These data suggest that targeting breast cancer with the hypoxia-activated drug DCQ is a potential therapeutic approach to suppress metastasis through the reduction of HIF-1α and downstream activated pathways.

It has been shown that DCQ exerts its cytotoxic activity through different cellular mechanisms in a cell-type specific manner [18‐22]. In this study, DCQ was found to reduce the viability and colony forming ability of MCF-7 and MDA-MB-231 via the induction of apoptosis, an effect that was significantly enhanced under hypoxia. Interestingly, the p53 mutant and invasive MDA-MB-231 cell line was more sensitive to DCQ than p53 wild type MCF-7 cells. In order to rule out the possible involvement of p53 in drug sensitivity, we silenced functional p53 in MCF-7 cells, and found that DCQ-induced apoptosis was not enhanced, suggesting that the higher drug sensitivity in MDA-MB-231, which provides additional selectivity for tumor versus normal tissue toxicity, may not be attributed to p53 status. This extends previous observations that DCQ-induced apoptosis in HCT116 human colon cancer cells is independent of p53 [19], rendering DCQ a promising molecule, especially when considering that more than 50% of human cancers bear mutated or abnormal p53.

Most hypoxia-activated drugs undergo bio-reductive activation, which generates intra cellular reactive radicals, namely hydroxyl radicals that break cellular DNA ultimately leading to apoptosis [16, 17]. Using alkaline comet assay, we have previously demonstrated that DCQ, when combined with radiotherapy, induces DNA damage [19, 22, 24]. In accordance with these findings, DCQ increased the levels of the DNA damage marker gamma-H2AX in both breast cancer cell lines. Together with previous findings [24], these data provide evidence that DCQ is a potent DNA damaging agent under hypoxic conditions.

ROS are known to mediate the DNA damaging effects of hypoxia-activated prodrugs [17]. In our breast cancer system, DCQ significantly increased the levels of intracellular ROS under hypoxia after 1 hour of exposure. Interestingly, a more pronounced increase of ROS generation was observed in MDA-MB-231 (90%) than MCF-7 cells (60%), which could explain the higher sensitivity of MDA-MB-231 cells to DCQ. In fact, it has been documented that MCF-7 cells are resistant to ROS-dependent apoptosis due to the high basal levels of NRF2, a transcription factor involved in protective responses to oxidants, as well as to elevated levels of intracellular glutathione [36]. In both breast cancer cell lines, DCQ-mediated cell death appears to require ROS production since drug toxicity was reversed by the antioxidants vitamin E and DTT.

Almost 50% of breast cancer patients treated for localized disease develop metastases [25]. Evidence suggests that hypoxia contributes positively to breast tumor invasion and metastasis [25, 26]. Recent studies show that hypoxia aids in the development or maintenance of breast cancer stem cells [37, 38]. HIF-1α is a major molecular player by which hypoxia orchestrates such complex patterns of gene expression in most solid tumors including breast cancer [39‐41]. Among other hypoxia-activated prodrugs, DCQ-induced cell death was accompanied by a reduction in the levels of HIF-1α in many cell lines [18‐22]. In MDA-MB-231 cells, DCQ reduces both basal and hypoxia-induced accumulation of HIF-1α via proteasomal degradation. In MCF-7 cells, it has been reported that the activation of p53, by DNA damaging agents, targets HIF-1α to proteasome-dependent degradation [28]. Here we show that DCQ induces DNA damage in MCF-7 cells, however, unlike other DNA damaging agents, it does not induce a proteasome-dependent degradation of HIF-1α. Moreover, silencing p53 led to a non-significant reversal of DCQ-induced down regulation of HIF-1α. These data provide evidence that DCQ does not target HIF-1α to degradation. Since, treating MCF-7 cells with DCQ did not affect the mRNA transcript levels of HIF-1α (data not shown), we suspected a possible anti-translational activity of DCQ. One of the major oncogenic pathways that shift the balance towards HIF-1α accumulation, via increased translation, is the AKT/mTOR pathway [8, 42‐44]. In fact, Zhan et al. reported that in Hela cells, TPZ suppresses the translation of HIF-1α in an eIF2α dependent manner [29]. This suppression was accompanied by an increase in the levels of p-AKT (Ser 473) and a reduction of p-mTOR (Ser 2448) [29]. Similar to TPZ, DCQ induced an increase in the levels of p-AKT (Ser 473) and a reduction of p-mTOR (Ser 2448), suggesting a possible anti-translational activity in MCF-7 cells. Taken together, we show that DCQ reduces HIF-1α in MDA-MB-231 and MCF-7 cells via distinct mechanisms and possibly independently of its DNA damaging effect (Additional file 1: Figure S5).

HIF-1α is known to activate vital pathways involved in metastasis [8]. In our breast cancer system, DCQ not only reduced HIF-1α but also attenuated hypoxia-induced invasion. The reversal of hypoxia induced VEGF secretion is in accordance with the anti-angiogenic effect of DCQ documented earlier in mouse mammary epithelial cells in C57BL mice [23]. HIF-1α plays a central role in metastasis via direct regulation of several EMT regulators such as Snail, SLUG, and TWIST [11‐14]. Here we report that targeting breast cancer cells with the hypoxia-activated drug, DCQ, reduces hypoxia-induced EMT, as indicated by the reduction of hypoxia-induced cell migration in MCF-7 cells and downregulation of TWIST in MDA-MB-231. In MDA-MB-231, the incomplete abrogation of HIF-1α and TWIST possibly explains the limited effect of DCQ on hypoxia-induced migration in this cell line [12]. In fact, exposure of MDA-MB-231 to hypoxia did not significantly alter their behavior, which may be due to the considerably high basal levels of HIF-1α in this cell line or to the constitutive activation of several oncogenic pathways involved in angiogenesis and EMT in these cells [12, 13, 25, 45].

The present study shows that the hypoxia-activated drug, DCQ, is more effective than the clinically used drug tirapazamine (TPZ) against breast cancer cell lines. DCQ induces ROS-dependent, p53-independent apoptosis in breast cancer cell lines in vitro. Additionally, DCQ reduces primary tumor volume, reduces secondary site invasion, and significantly increases survival in vivo. DCQ exhibits its unique pro-apoptotic and anti-metastatic properties via the reduction of HIF-1α and downstream signaling cascades.

DCQ was synthesized from 5,6-dichlorobenzofurazan oxide and dibenzoylmethane via the Beirut Reaction [46]. RPMI 1640, 10X trypsin-EDTA, 0.25X trypsin-EDTA, Dulbecco’s Phosphate Buffered Saline (PBS), Fetal Bovine Serum (FBS), sodium bicarbonate, penicillin-streptomycin (P/S), DMSO, trypan blue, vitamin E, and Dithiothreitol (DTT) were purchased from Sigma, St. Louis, MO, USA. Propidium iodide (PI) and CM-H₂DCFDA were purchased from Invitrogen Molecular Probes, Eugene, OR, USA. Protease Inhibitor MG-132 was purchased from Roche Applied Science, Penzberg, Germany.

The human breast cell lines MCF-7 (p53 wildtype, noninvasive [27]), MDA-MB-231 (p53 mutant, invasive, [27]) and MCF-10A (p53 wildtype, immortalized normal breast [27]) were originally obtained from ATCC and were constantly tested for mycoplasma contamination. No authentication was performed by the authors. Cells were grown in RPMI 1640 modified medium supplemented with 10% heat-inactivated FBS and 1% Pen-Strep. All cells were maintained in a humidified atmosphere of 5% CO₂ and 95% air. DCQ (7.5 mg) was dissolved in 1 ml of DMSO then diluted in media (DMSO ≤ 0.1%). For hypoxia exposure, cells were placed in a tightly sealed chamber (Bactron III, SHEL LAB, UK) at 37°C. The desired oxygen level was monitored by an Ohmeda Oxymeter and maintained by pumping a gas mixture of 1% O₂, 5% CO₂, and 94% N₂, unless otherwise mentioned.

Cells (10⁴) were treated with drugs 24 hours after plating. Cell viability was tested 6 hours or 24 hours after treatment by the non-radioactive cell proliferation kit (Promega Corporation, Madison, USA), an MTT-based method. Clonogenic survival was performed as described previously [19]. Briefly, cells were treated with DCQ for 3 hours under normoxia (21% O₂) or hypoxia (1%, 10% O₂), trypsinized, then re-plated at a density of 300 living cells/100 mm petri dish without the drug. 12 days later, cells were stained with aqueous 0.5% solution of crystal violet and the number of colonies having more than 50 cells was recorded.

Flow cytometric analysis was performed as previously described [24]. Briefly, cells were treated with DCQ at 50% confluency, and incubated for 6 hours under normoxia or hypoxia and the PI-stained DNA content was measured using Fluorescence Activated Cell Sorter (FACS) flow cytometer (Becton Dickinson, Research Triangle, NC). For Annexin V/PI, after treatment, cell pellets were washed with PBS and re-suspended in 100 μl Annexin-V-Fluos labeling solution (20 μl annexin reagent and 20 μl PI (50 μg/ml) in 1000 μl incubation buffer pH 7.4 (10 mM Hepes/NaOH, 140 mM NaCl, 5 mM CaCl₂). Samples were incubated for 15 mins at room temperature and 0.5 ml incubation buffer was added. Cellular fluorescence was then measured by flow FACS.

Western blots were performed as described earlier [19]. Membranes were probed with the primary antibodies: p21 ((C-19)-G), p53 (DO-1), p-p53, HIF-1α (Novus Biologicals, Littleton, USA), gamma-H2AX, TWIST and GAPDH followed by horseradish peroxidase-conjugated anti-mouse, anti-rabbit, or anti-goat IgG-HRP (Santa-Cruz, California, USA). All blots have been generated from triplicate experiments after 6 hours of exposure to DCQ or hypoxia. The concentration of VEGF in the culture supernatant was measured using a quantitative sandwich enzyme immunoassay (R&D Systems Inc., Minneapolis, USA) according to the manufacturer’s instructions. VEGF concentrations were calculated by comparison with a standard curve generated using a 4-parameter logistic curve-fit and on-board software. VEGF concentrations were normalized to amount (pg) per 1 × 10⁶ living cells/10 ml supernatant.

At 50% confluency, cells were incubated with 10 μM of CM-H2DCFDA in media (2% FBS) for 25 mins at 37ºC. After which, cells were washed with PBS and treated with 5 μM DCQ for 1 hour under hypoxia. Control and treated cells were trypsinized, centrifuged, washed with PBS, and fluorescence was immediately evaluated by FACS. For the antioxidant experiments, cells were pretreated with 1 mM vitamin E or DTT for 2 hours, washed with PBS, incubated with CM-H2DCFDA for 25 mins, and the above protocol was performed to determine the levels of ROS generated.

2 × 10⁴ cells were seeded on top of reduced matrigel, in the trans-well upper chamber, in serum free media and the lower chamber was filled with 10% FBS media. Cells were either treated with 2.5 μM DCQ (5 μM of DCQ is toxic after 24 hours) or vehicle (supplemented in the upper chamber only) under normoxia and hypoxia for 24 hours. Then the lower side of the trans-well membrane was fixed in 4% formaldehyde and cells were stained with Hoechst and counted using fluorescence microscopy.

siRNAs were used to knockdown p53 gene expression. siRNA p53 (p53 siRNA (h): sc-29435), control si RNA (Control siRNA-A: sc-37007) were purchased from Santa Cruz, California, USA. The siRNAs were transfected into MCF-7 cells using lipofectamine 2000 transfection reagent (Invitrogen, New York, USA) following manufacturer’s instructions. After 24 hours, cells were treated with DCQ placed under hypoxia for 6 hours, and proteins were extracted for western blot analysis.

This study was approved by the Institutional Animal Care and Utilization Committee (IACUC) of the American University of Beirut. Eight weeks aged female NOD-SCID mice (average weight 23 g) were injected sub-dermally (s.d.) with 4 × 10⁶ MDA-MB-231 cells into the subcutaneous area of the neck region, in incomplete media. Treatment groups included DCQ (17 mg/Kg in 50 μL DMSO) or DMSO (50 μL) as control, which were administered intraperitoneally (i.p.), twice per week for 4 weeks. Toxicity studies have shown that 17 mg/Kg in 50 μL DMSO is a tolerated dose based on mice behavior and body weight. The weight of animals was determined daily, and the date of death was recorded. Survival curves were plotted using Kaplan-Meier method. Tumor growth and progression were monitored weekly by measurement of tumor size (length, width, height) with a caliper device. These data were plotted as the average tumor size of 22 mice/treatment group. For histological assessment, at the indicated time points (week 5, 7, and 9), 3 mice were anesthetized with isofurane and sacrificed by cervical dislocation. Biopsies from lungs, liver, and primary tumor sites were obtained. Tissues were sectioned at 6 μm thickness using a paraffin microtome and mounted on slides and stained by H&E or probed against HIF-1α or PECAM and were examined by light microscopy. Three tumors (week 5) per treatment group (DMSO versus DCQ) were assayed for lumen containing micro-vessels. Micro-vessels were counted from four random fields at low magnification.