Bardoxolone-Methyl (CDDO-Me) impairs tumor growth and radioresistance of oral squamous cell carcinoma (OSCC) cells via accumulation of reactive oxygen species


 Radiotherapy (RT) represents a common treatment strategy for patients suffering from oral squamous cell carcinoma (OSCC). However, application of RT is immanently limited by radio-sensitivity of normal tissue surrounding the tumor sites. In this study, we used normal human epithelial keratinocytes (NHEK) and OSCC cells (Cal-27) as models to investigate radio-modulating and anti-tumor effects of the synthetic triterpenoid 2-cyano-3,12-dioxooleana-1,9,-dien-28-oic acid methyl ester (CDDO-Me). Nanomolar CDDO-Me significantly enhanced antioxidative heme oxygenase-1 (HO-1) levels in NHEK only and not in the OSCC cell line, as shown by immunoblotting. In the presence of CDDO-Me reactive oxygen species (ROS) were found to be reduced in NHEK when applying radiation doses of 8 Gy, whereas ROS levels in OSCC cells rose significantly even without radiation. In parallel, CDDO-Me was shown to enhance metabolic activity in malignant cells only as indicated by significant accumulation of reducing equivalents NADPH/NADH. Clonogenic survival was left unchanged by CDDO-Me treatment in NHEK but revealed to be abolished almost completely in OSCC cells. The latter effect was confirmed by a CDDO-Me induced significant reduction of OSCC tumor xenograft-growth in-ovo applying the chick chorioallantoic membrane (CAM) assay. Our results strongly indicate anti-cancer and radio-sensitizing effects of CDDO-Me treatment in OSCC cells, whereas nanomolar CDDO-Me failed to provoke clear detrimental consequences in non-malignant keratinocytes. We conclude, that the observed differential aftermath of CDDO-Me treatment in malignant OSCC and non-malignant skin cells may be capitalized to broaden the therapeutic range of clinical RT.


Introduction
Malignancies of the oral cavity are among the most common cancers within the European Union.
According to estimates of the European Cancer Information System (ECIS) over 45,000 cancer cases of the lip and oral cavity were diagnosed in 2018, representing a crude incidence rate of 8.9 per 100,000 (1). Despite advances in modern multidisciplinary treatment modalities comprising surgery, radiochemotherapy and targeted pharmacological therapy, the overall outcome of oral squamous cell carcinoma (OSCC) patients still remains dissatisfying (2). Therefore, scienti c efforts have previously been made to overcome clinical limitations and side-effects of OSCC treatment regimes. The therapeutic window for radiotherapy is mainly narrowed by local side effects mainly due to damage of surrounding normal tissue when targeting cancer sites. Aside from the recent implementation and constant advancement of intensity-modulated radiotherapy (IMRT), a further strategy to restrict radiation doses for neighboring normal cells lies within the identi cation of small-molecule drugs allowing for the radiosensitization of cancer cells and ideally with a radio-protective effect on healthy tissue (3)(4)(5).
The synthetic oleanane triterpenoid 2-cyano-3,12-dioxooleana-1,9,-dien-28-oic acid (CDDO) and its C-28 methyl ester (CDDO-Me, Bardoxolone-methyl; Fig. 1A) has been shown to exert bene cial therapeutic activities by suppressing in ammation and oxidative stress in vitro and in vivo at low nanomolar concentrations (6). The BEACON-study (ClinicalTrials.gov Identi er: NCT01351675), a randomized, placebo-controlled phase 3 clinical trial, evaluated CDDO-Me induced effects on the kidney function in 2,185 patients suffering chronic kidney disease and type 2 diabetes. Although the study ultimately had to be terminated due to increased rates of heart failure events, CDDO-Me revealed to increase eGFR and to signi cantly reduce the hazard for the loss of kidney function (7). Besides the inhibition of the nuclear factor κB (NFκB) signaling cascade, activation of the Kelch-like ECH-associated protein 1 (Keap1)/ nuclear factor erythroid 2-related factor (Nrf2) pathway is widely regarded as a major mechanism of action for CDDO-Me related cytoprotective effects (6,8). Stimulation of the Nrf2 pathway mediates the downstream activation of various promoter genes encoding for detoxifying and antioxidative proteins like heme oxygenase 1 (HO-1). The heat-shock protein (HSP)-32 family member HO-1, which has been found in microsomes, mitochondria and nuclei, was demonstrated to catalyze the rate-limiting step of heme catabolism, leading to the formation of biliverdin. The following biliverdin/bilirubin redox cycle system effectively scavenges reactive oxygen species (ROS) and represents a highly conserved cellular control mechanism against oxidative stressors like radiation (9)(10)(11).
Numerous experimental studies highlighted the e cacy of CDDO-Me for both, prevention and treatment of cancer, albeit predominantly at high nanomolar to micromolar concentrations (6,12).
However, differential reactions to radiation exposure of normal and cancer cells at equivalent and physiological achievable CDDO-Me concentrations are preferably required when giving consideration to a future usage in radiotherapy. Previously, CDDO-Me has been demonstrated to mitigate radiation-induced damage in normal epithelial cells but not cancer cells of the lung, breast and colon (13,14).
In this study, we analyzed the implications of low nanomolar CDDO-Me in the radiation response and in ovo tumor growth of the OSCC cell line Cal-27 and matched the results with ndings in normal human epithelial keratinocytes (NHEK) as a model for surrounding healthy skin.

Material And Methods
Cell culture and treatment Cal-27 cells were originally derived from a 56-year old male patient suffering SCC of the tongue and were purchased from Leibniz-Institut DSMZ (Braunschweig, Germany). Cells were cultivated at 37 °C in a 5% CO2 atmosphere using DMEM GlutaMAX medium (Gibco, Eggenstein, Germany), which was supplemented with 10% FCS (Boehringer, Mannheim, Germany).
Primary normal human epidermal keratinocytes (NHEK) originate from the epidermal stratum basale of an adult single donor and were cultivated at 37 °C and 5% CO2 in Keratinocyte Growth Medium 2 (both from PromoCell, Heidelberg, Germany).
Unless stated differently, seeded cells were allowed to attach for 24 h, then culture medium was supplemented with 10 nM CDDO-Me or DMSO as solvent control at 0.1 vol% (both from Selleckchem, Houston, USA) and cells were incubated for further 6 h. Subsequently, cells were treated according to the respective protocol.
Radiation exposure Cells were exposed to 240 kV X-rays using the YXLON Maxishot (Hamburg, Germany) including a 3 mm beryllium lter at a plateau dose rate of 1 Gy/min at 13 mA. Monitoring of the applied doses was performed by a PTW Unidose dosimeter (PTW Freiburg GmbH, Freiburg, Germany).

Cytotoxicity assessment
In order to assess the IC 50 values of CDDO-Me for Cal-27 and NHEK, we performed the ATP dependent CellTiter-Glo® Luminescent Cell Viability Assay (Promega, Madison, USA) according to the manufacturers protocol using a concentration range from 2.5 nM up to 2.5 µM in 11 steps with DMSO (solvent) only and cell free medium serving as negative controls.
For image acquisition we used a Zeiss Axioimager 2i uorescence microscope in combination with the ISIS uorescence imaging system (MetaSystems, Altlussheim, Germany).

Immunoblotting
The XCell Sure Lock™ Mini-Cell Electrophoresis System served as a platform for western blot experiments according to standard protocols. For equalization of protein concentrations, we used the BCA Protein Assay Kit (both from Thermo Scienti c, Westham, USA). Amounts of HO-1 were detected using primary rabbit monoclonal anti-HO-1 (dilution 1:1000, Cell Signaling, Danvers, USA) and secondary HRPconjugated polyclonal goat anti-rabbit (dilution 1:10,000, Thermo Scienti c, Westham, USA). For digital image acquisition we used the myECL™ Imager system (Thermo Scienti c, Westham, USA). For calculation of HO-1/GAPDH-ratios greyscale intensity values were determined by ImageJ software, v. 1.51 (NIH, Bethesda, USA).

Assessment of cellular reactive oxygen species
In order to show whether CDDO-Me decreases the amount of free reactive oxygen species (ROS) after irradiation within NHEK and Cal-27 cells we used the DCFDA Cellular ROS Detection Assay Kit according to the manufacturer's instructions (Abcam, Cambridge, UK). In brief, 2', 7'dichlorodihydro uorescein diacetate (DCFDA) served as a marker which accumulates in living cells and becomes uorescent upon oxidation. Therefore, 0.5 × 10 6 cells (Cal-27) or 0.75 × 10 6 cells (NHEK) were incubated for 24 hours in 6 cm diameter petri dishes. Subsequently, 1 µL/mL medium CDDO-Me stock solution (10 µM in DMSO) was added to the treatment group resulting in 10 nM CDDO-Me, whereas 1 µL/mL Medium DMSO was added to the control group and incubated for another 6 hours before undergoing X-ray irradiation; 45 min before irradiation, cells were stained with 25 µM DCFDA. 55 mM Tert-Butyl Hydrogen Peroxide (TBHP) served as positive control. Immediately after irradiation the cells were detached by trypsinization and measured by ow cytometry using the FACS-Calibur System (BD Biosciences, Franklin Lakes, NJ, USA). Experiments were performed in quadruplicate (10.000 cells/experiment).
Absorbance at 490 nm indicated the amount of formazan using the Multiskan™ FC microplate photometer (Thermo Scienti c, Westham, USA).
Clonogenic survival assay NHEK and Cal-27 cells were cultivated in 6-well plates for 24 hours and treated according to the standard protocol. Experiments were stopped after 9 days by xing cells with 70% ethanol followed by staining with gentian violet. We counted colonies (> 50 cells) manually by using a Zeiss STEMI SV8 stereomicroscope. Experiments were performed in quadruplicate.
DNA double strand break analysis using imaging ow cytometry DNA double strand breaks were assessed using phosphohistone γH2AX as a marker. Cells were detached, xed for 20 minutes in cold 4% PFA pH 7.0 (Roti-Histo x®, Carl Roth, Karlsruhe, Germany), washed twice, permeabilized for 10 minutes using 0.1% Triton X (Sigma Aldrich, Darmstadt, Germany) and washed twice again. Staining was performed for 2 hours at room temperature using mouse γH2AX antibodies primarily coupled with AlexaFluor® 488 (BioLegend, San Diego, CF, USA). After one additional washing step, cells were resuspended in 100 µL PBS, containing 20 µM DRAQ5 for DNA staining, yielding at least 10 6 cells/mL and measured using the ImageStream® X mkII (Luminex, Austin, TX, USA) imaging ow cytometer.
Excitation lasers with 488 nm and 642 nm wavelength were used at laser powers adjusted to the sample with the highest expected signal for each data set.

Chick egg chorioallantoic membrane as tumor xenograft model
The chick egg chorioallantoic membrane (CAM) tumor model was used as previously described (16)(17)(18). Brie y, fertilized chicken eggs were incubated at 37 °C and 60% relative air moisture for 7 days before fenestration and placement of a silicone ring (diameter 5 mm) on the vascularized CAM. A 1:1 solution of matrigel (BD, Heidelberg, Germany) and medium containing Cal-27 cells (1.5 × 10 6 cells/egg) was grafted within the ring. The following day, topical treatment with CDDO-Me (10 nM) or vehicle (0.2% DMSO in NaCl 0.9%) was started and continued for 2 more days. After an incubation period of 4 days at 37 °C, tumors were collected, imaged, xed in phosphate-buffered 4% formaldehyde solution and embedded in para n for immunohistochemical analysis. Slices (5 µm) were stained for H&E and proliferation marker Ki-67 (Dako, Glostrup, Denmark). Mean tumor volume of Cal-27 cancer xenografts was assessed immediately after extraction. Tumor volume was calculated according to the formula: π/6 x length x width² (19).
P-values < 0.05 were regarded as statistically signi cant. Bars indicate mean values ± standard deviation.

CDDO-Me increased heme oxygenase-1 levels in NHEK
Analysis of CDDO-Me-cytotoxicity exhibited IC 50 values of 820 nM in NHEK and 280 nM in Cal-27 (data not shown). We found HO-1 ubiquitously expressed in both, NHEK and Cal-27 as shown by uorescence microscopy (Fig. 1B). HO-1 was shown to be localized throughout the whole cells, albeit exhibiting an accentuation within nuclei compared to cytoplasm. When incubating cells with 10 nM CDDO-Me over 6 hours the subcellular localization of HO-1 was left unchanged (data not shown). However, HO-1 levels increased signi cantly in the whole cell lysate of NHEK in the presence of CDDO-Me as shown by immunoblotting, whereas elevated HO-1 concentrations after CDDO-Me treatment in Cal-27 failed to reach statistical signi cance (Fig. 1C + D).
Treatment with CDDO-Me increased reactive oxygen species, metabolic activity and γH2AX-foci formation after irradiation in OSCC cells Investigation of cellular reactive oxygen species (ROS) activity was based on ow cytometric measurement of ROS-sensitive DCF levels. DCF uorescence intensity rose concomitantly to increasing doses of ionizing radiation in both cell lines ( Fig. 2A + B). While baseline ROS activity was found to be unaltered in the presence of CDDO-Me in non-malignant NHEK, we showed signi cantly increased ROS in Cal-27 during CDDO-Me treatment at low nanomolar concentrations (Fig. 2C + D). Radiation-induced increase of cellular ROS turned out to be unaffected by CDDO-Me in Cal-27, whereas ROS generation by 8 Gy was signi cantly reduced in NHEK. The amount of cellular reducing equivalents NADPH and NADH was demonstrated to be signi cantly enhanced in Cal-27 upon CDDO-Me incubation for 6 hours when applying a concentration range from 1 nM to 10 µM (Fig. 2F). In contrast, no signi cant effects on metabolic activity were detected in NHEK (Fig. 2E).
Finally, CDDO-Me mediated consequences for cellular γH2AX-foci induction were analyzed 6 hours after irradiation using imaging ow cytometry (Fig. 3A + B). Our results demonstrate a signi cant increase in DNA double strand break frequency in both cell lines as indicated by γH2AX-foci ( Fig. 3C + D).

CDDO-Me impaired clonogenic survival and tumor forming capability of OSCC cells
Analyzing clonogenic survival highlighted no signi cant changes when treating NHEK with 10 nM CDDO-Me (Fig. 3E + F). In contrast, CDDO-Me abolished clonogenicity of Cal-27 with or without irradiation almost completely. To verify these ndings, we seeded Cal-27 on top of a vascularized chick egg chorioallantoic membrane and monitored the amount of subsequent tumor growth and Ki67-proliferation index (Fig. 4A). CDDO-Me was shown to reduce tumor volume in a signi cant manner (Fig. 4B). The fraction of Ki67-proliferative cells within the tumor tissue revealed to be reduced by trend even though without reaching statistical signi cance levels ( Fig. 4A + B)

Discussion
Over the last decades radiotherapy (RT) has traditionally played an essential role in the clinical management of oral squamous cell carcinoma (OSCC). Despite the development of various modi cations of conventional RT regimens there is still a narrow ridge between effective cancer treatment resulting in improved patient's outcome and the prevention of adverse effects by damaging neighboring healthy tissue (20).
In this study, the triterpenoid CDDO-Me was demonstrated to exert anti-cancer activity in an OSCC cell line, while not consistently impairing cell homeostasis of primary keratinocytes even in the presence of ionizing radiation. Our ndings are in line with previous studies, which highlighted radio-protective effects of CDDO-Me in normal epithelial cells of the lung, breast and colon but not in cancer cells (13,14,21).
Activation of the Nrf2 pathway followed by downstream up-regulation of antioxidant enzymes like HO-1 is widely regarded as a major mechanism of action of CDDO-Me (6). In physiological settings the Nrf2/HO-1 cascade represents a key mechanism for normal cells to adapt to oxidative stress conditions, mediating enhanced survival, preserved cellular homeostasis and prevention of carcinogenesis (22). Several studies using rodent models and a case report on a HO-1 de cient patient revealed the radioprotective activity of antioxidant HO-1 in normal tissue, including skin (23)(24)(25).
In our study we found signi cantly enhanced HO-1 levels when treating NHEK with CDDO-Me. This observation was accompanied by reduced ROS activity after high dose irradiation, indicating radioprotective effects of CDDO-Me possibly mediated via enhanced expression of cellular HO-1. Sole CDDO-Me administration left ROS activity in NHEK unchanged, which argues for direct pharmacological effects nally leading to augmented HO-1 instead of elevated HO-1 levels due to increased oxidative stress.
Surprisingly, the diminished ROS levels found after high dose (8 Gy) irradiation were not potent enough to result in measurable reduction of γH2AX-foci of NHEK. The missing veri cation of reduced DNA damage subsequent to reduced ROS may in part be explained by the saturation of γH2AX-foci induction commonly observed by administration of high radiation doses (26). Furthermore, even radiation with low linear energy transfer, such as X-rays generates DNA double strand breaks to some extent via ROSindependent direct ionization of target DNA atoms, representing a way of action that is not preventable by elevated antioxidative cellular defense mechanisms.
On the other hand, signi cantly reduced ROS activity in NHEK after 8 Gy points to cytoprotective effects by prevention of oxidative damage to various molecular structures crucial for the maintenance of cell homeostasis, e.g. within cellular membranes or organelles. By all means, low nanomolar CDDO-Me did neither provoke acute toxic effects (IC 50 = 820 nM) nor impairment of viability, clonogenicity or radioresistance in NHEK. A recent in-vivo study even highlighted radioprotective effects for healthy skin when radiation-induced dermatitis was shown to be mitigated when treating mice externally with CDDO-Me (27).
OSCC cells showed an increased sensitivity to CDDO-Me when compared to NHEK. IC50 values of CDDO-Me were roughly 3-fold lower in Cal-27. Nanomolar CDDO-Me abolished clonogenicity of OSCC cells almost completely and signi cantly impaired tumor forming capability of OSCC xenografts grown on the CAM of fertilized chicken eggs. Contrary to the results found in NHEK, we showed signi cantly increased ROS activity subsequent to CDDO-Me treatment. Since redox equivalents NADPH and NADH proved to be enhanced concomitantly upon CDDO-Me administration we hypothesize that CDDO-Me may selectively trigger ROS accumulation in cancer cells via activation of the mitochondrial metabolism as the major source of intracellular ROS. No evidence was found for down-regulated antioxidative defensive mechanisms by CDDO-Me as a potential contributing factor for elevated intracellular ROS levels, since HO-1 concentrations revealed to be even increased by trend in OSCC cells. Contrary to our results found in NHEK this observation failed to reach statistical signi cance, which may in part be referred to elevated baseline HO-1 expression levels in Cal-27. Constitutive up-regulation of antioxidative adaptive mechanisms is widely regarded to be inherent in malignant cells and was shown to be associated with cancer progression and resistance to therapy (22,28).
Furthermore, we postulate that the activation of the antioxidative Keap1/Nrf2 pathway regularly attributed to CDDO-Me, may partially result from elevated ROS levels. Previous studies identi ed the direct interaction of synthetic triterpenoids with Keap1 to be responsible for the Nrf2 pathway initiation (29). However, activated Nrf2 signaling was shown to be only partially involved in the up-regulation of HO-1 mediated by CDDO-derivatives (30). CDDO-Me is well known as a multifunctional drug, which initiates various energy-demanding processes like transcriptional pathways or apoptosis (6). The required activation of mitochondrial metabolism for cellular energy supply accompanied by inevitable ROS generation may nally stimulate the up-regulation of antioxidative enzymes like HO-1. We further postulate that CDDO-Me induced ROS-accumulation may vary among the analyzed cell lines, as no signi cantly changed ROS activity was observed in NHEK and previously shown in U937 cells isolated from histiocytic lymphoma (30).
However, further research is needed to illuminate the precise mechanisms and clinical relevance of CDDO-Me induced ROS generation in OSCC cells.
Recent scienti c research highlighted the ambivalent role of ROS in cancer (28,31). While elevated baseline ROS levels commonly found in highly metabolic active cancer cells are capable to mediate prooncogenic characteristics, excessive amounts of intracellular ROS were shown to induce senescence and/or apoptosis. The latter effect is widely exploited by conventional treatment regimens like chemotherapy or radiotherapy (32,33). CDDO-Me induced ROS accumulation in malignant cells could explain the signi cantly increased frequency of γH2AX-foci following high dose irradiation. Consistently, OSCC colony formation was signi cantly reduced by combined treatment with CDDO-Me pointing to radiosensitizing effects in OSCC cells.
In summary, CDDO-Me selectively impaired proliferation, clonogenicity and radioresistance in OSCC cells, but not in NHEK. The use of CDDO-Me as a radiomodulating drug during radiotherapy of OSCC patients may optimize treatment effectivity in parallel to reduced side effects. We hypothesize, that CDDO-Me mediates anticancer effects in part via cellular ROS accumulation due to enhanced mitochondrial metabolism.