Potent Antitumor Effects of a Combination of Three Nutraceutical Compounds

Head and neck squamous cell carcinoma (HNSCC) is associated with low survival, and the current aggressive therapies result in high morbidity. Nutraceuticals are dietary compounds with few side effects. However, limited antitumor efficacy has restricted their application for cancer therapy. Here, we examine combining nutraceuticals, establishing a combination therapy that is more potent than any singular component, and delineate the mechanism of action. Three formulations were tested: GZ17-S (combined plant extracts from Arum palaestinum, Peganum harmala and Curcuma longa); GZ17-05.00 (16 synthetic components of GZ17-S); and GZ17-6.02 (3 synthetic components of GZ17S; curcumin, harmine and isovanillin). We tested the formulations on HNSCC proliferation, migration, invasion, angiogenesis, macrophage viability and infiltration into the tumor and tumor apoptosis. GZ17-6.02, the most effective formulation, significantly reduced in vitro assessments of HNSCC progression. When combined with cisplatin, GZ17-6.02 enhanced anti-proliferative effects. Molecular signaling cascades inhibited by GZ17-6.02 include EGFR, ERK1/2, and AKT, and molecular docking analyses demonstrate GZ17-6.02 components bind at distinct binding sites. GZ17-6.02 significantly inhibited growth of HNSCC cell line, patient-derived xenografts, and murine syngeneic tumors in vivo (P < 0.001). We demonstrate GZ17-6.02 as a highly effective plant extract combination and pave the way for future clinical application in HNSCC.

and nanoparticle encapsulation techniques have been designed to increase bioavailability 16,17 . Further, the combination of curcumin with additional nutraceuticals potentiates efficacy 18 , and the combination of curcumin provides additive benefit to chemotherapeutics 19 .
Arum palaestinum is widely cultivated in Palestine, and has been used in the treatment of cancer in Palestine for many years 20 . Ethanolic extracts of Arum palaestinum have shown antitumor efficacy against breast cancer, leukemia, and prostate cancer 21,22 . Yet, little has been done to characterize Arum palaestinum's mechanism of action.
The alkaloids from the plant Peganum harmala are known to contain a wide spectrum of medicinal properties. The main constituent, harmine, is an inhibitor of monoamine oxidase, and also demonstrates anti-tumor effects 23 . Harmine intercalates and damages DNA 24 , and mitigates chemotherapy resistance by interfering with drug efflux 25 . Further, harmine decreased proliferation of various tumor lines, while having little effect on normal cells 26 .
Chemotherapies are often given to patients in combination. The aim of this study was to determine whether a potentiated effect could be achieved by combining nutraceuticals. Given the documented success of combination therapy with curcumin 18,19 , this was used as a starting point and included Arum palaestinum and Peganum harmala for their proposed anti-cancer activity in HNSCC. We assessed the combined plants, comparing a dried extract of the three plants (GZ17-S), a synthetic version of the extract (GZ17-05.00) and the three major anti-cancer agents found in the original plants (GZ17-6.02). Our results demonstrate a highly effective combination for use in HNSCC, more potent than any component used singularly, when assessed in preclinical models. We delineate the mechanism of action, and provide evidence of a useful biomarker for future clinical study.
To characterize cytotoxic effects of curcumin, harmine, and isovanillin in GZ17-6.02, we treated HNSCC cells with vehicle control (ethanol), single or double combination each in a ratio representative of GZ17-6.02 in a final dose of 50 ug/mL for 48 h. GZ17-6.02 demonstrated improved attenuation of viability, which was superior to single or double combination of agents (Fig. 1B). The cytotoxic effects of the combination are additive in nature (Supplemental Fig. 1C). This is the first report of the additive effects of such a combination, and demonstrates efficacy at lower concentrations than any singular component.
Additionally, to assess this combination across multiple tumor types, we assessed the proliferation of glioblastoma line U87, and lung cancer lines 201 T and A549. GZ17-6.02 was cytotoxic in these cell lines as similarly to the HNSCC lines tested (Fig. 1C). These data demonstrate combination therapy using multiple nutraceutical components decreases cancer cell proliferation. GZ17-6.02 and −05.00 induce apoptosis through caspase-3 activation and PARP cleavage. To understand the mechanism underlying GZ17 anti-proliferative effects, we treated HNSCC cells for 72 h at the ED 50 dose. GZ17-05.00 and GZ17-6.02 induced significant apoptosis (24% and 22%, respectively, P < 0.05) as observed by the sub G0 fraction of cell cycle analysis in treated cells (Fig. 1D). To gain insight into the mechanism of GZ17-induced apoptosis, levels of poly-ADP ribose polymerase (PARP) and caspase-3 were evaluated using immunoblot. Treating cells with GZ17-6.02 and GZ17-05.00 at ED 50 concentrations increased cleaved PARP and decreased caspase-3 ( Fig. 1D), indicators of treatment-induced apoptosis.

GZ17-6.02 mitigates migration and invasion of HNSCC cells, and endothelial cell tubule formation.
No treatment options are available for the metastatic spread of HNSCC, and patients with metastasis have a 5-year survival rate of less than 20% 27 . We assessed the effects of GZ17-formulations on HNSCC migration and invasion, prerequisites to metastasis. GZ17-6.02 (ED 50 dose) was most effective in significantly inhibiting migration (p < 0.0001, Fig. 2A, Supplemental Fig. 2A); and invasion (p = 0.0003, Fig. 2B, Supplemental Fig. 2B). Additionally, we assessed the inhibitory effects of GZ17-6.02 against glioblastoma and lung cancer cell line migration and invasion, and observed significant reductions in U87 (Fig. 2C), 201T (Fig. 2D), and A549 (Fig. 2E) migration and invasion.
Angiogenesis plays a critical role in HNSCC progression, however, no antiangiogenic therapies are approved for use in HNSCC 28 . Previously, curcumin has been reported to limit HUVEC tubule formation 29 , and vascular tubule mimicry 30 . We tested the efficacy of GZ17-6.02 (ED 50 dose) to inhibit endothelial tubule formation, an in vitro indicator of angiogenesis 31 . GZ17-6.02 significantly decreased the length of HUVEC endothelial tubules compared to the vehicle control (p = 0.0001, Fig. 2F).
Macrophage infiltration is frequently associated with an inflammatory reaction, and tumor-associated macrophages facilitate tumor progression 32 . We tested the ability of GZ17-6.02 to attenuate macrophage infiltration into tumors using a 3-dimensional cell culture device called the metastatic mimetic device (MMD) 33 . HNSCC cells embedded in a collagen matrix were subjected to macrophage infiltration in the MMD in the presence of vehicle control or GZ17-6.02. GZ17-6.02 effectively reduced macrophage survival (Fig. 2G) and subsequently infiltration through the collagen plugs (Supplemental Fig. 2C). These cumulative data demonstrate that GZ17-6.02 significantly mitigates in vitro tumor progression. GZ17-formulations modulate key signaling molecules in HNSCC. GZ17-formulations demonstrate significant anti-cancer effects in vitro. In order to assess the mechanism of action, and to understand the additive effects observed, we used a phospho-protein array. Phosphorylation of several proteins including Src, ERK1/2, EGFR, Akt, STAT-2 and Chk-2 were significantly reduced in GZ17-treated HNSCC (ED 50 dose) (Fig. 3A,B). These are important mediators of HNSCC proliferation, migration and survival. ERK1/2 is a regulator of HNSCC proliferation, and increased levels are associated with advanced disease 34 . To validate phospho-protein array results, we evaluated pERK1/2 by immunoblot at both a short time interval when induced by EGF (5 m) and long interval (72 h) (Fig. 3C,D and Supplemental Fig. 2D). These data identify that GZ17-6.02 inhibits phosphorylation of ERK1/2, an important molecular mediator of HNSCC progression. GZ17-6.02 components bind with high affinity to EGFR, ERK1/2, and Akt1. To further understand the mechanism of GZ17-6.02 active components, we carried out molecular docking analyses on the targets identified. Curcumin and harmine, as well as their metabolites, demonstrated high binding affinity to several targets (Supplemental Table 2). Both inhibitors, and their metabolites, demonstrated high affinity towards distinct binding sites on EGFR, ERK1/2, and Akt1 (Fig. 4, and Supplemental Figs 3-6). Curcumin exhibited best binding affinity of −6.6 Kcal/mol to Akt1 and −6.5 Kcal/mol to EGFR. Whereas, harmine showed best binding affinity with −6.3 Kcal/mol to Akt1 and −6.8 Kcal/mol to EGFR. Isovanillin did not have significant binding to the targets assessed. Simultaneous interaction of two inhibitors to the target may explain the improved efficacy of the combination. GZ17-6.02 enhances cisplatin efficacy. Platinum-based chemotherapy is a standard choice of treatment for HNSCC 35 . However, despite new protocols, tumor resistance to cisplatin remains a significant hurdle to HNSCC treatment. We assessed the combinatorial effect of GZ17-6.02 (ED 50 ) with cisplatin (4.0 µM) in HNSCC cell lines. Combination of GZ17-6.02 and cisplatin significantly reduced HNSCC cell survival as compared to cisplatin or GZ17-6.02 treatments alone (p < 0.0008, Fig. 5A-C). Combination of radiation and GZ17-6.02 found GZ17-6.02 to have greater cytotoxicity than 3, 6, or 9 Gy of radiation alone (Supplemental Fig. 7), with no additive benefit of radiation in combination. This demonstrates GZ17-6.02 treatment potentiates effects of standard HNSCC chemotherapy. GZ17 formulations have not been tested in animal models. To determine the efficacy of GZ17 formulations in vivo, we treated subcutaneous OSC19 tumors in athymic nude mice with intratumoral injections (15 mg/kg/day) of GZ17-6.02 formulations. Among the formulations tested, GZ17-6.02 treatment was most efficacious with significant tumor volume reduction (p < 0.001, Fig. 6A). To assess GZ17-6.02 in an immunocompetent host, a syngeneic SCC model was employed. SCC vII/SF syngeneic tumors are highly aggressive requiring higher doses for testing. Thus, in this model, GZ17-6.02 was delivered by oral gavage, and a greater dose (100 mg/kg/day). Again, a significant decrease in tumor volume was demonstrated by GZ17-6.02 administered by oral gavage (p < 0.001, Fig. 6B). GZ17-6.02 receiving mice were active, curious and demonstrated normal, healthy behavior. There were no differences in mouse weights between treatment groups (Supplemental Figure 8), and no adverse effects from treatment were observed. This indicates GZ17-6.02 is well tolerated.

GZ17-6.02 effectively inhibits HNSCC tumor growth in vivo.
Finally, the response of GZ17-6.02 in patient-derived xenograft (PDX) tumors was evaluated. GZ17-6.02 was administered by oral gavage at a lower dose of 30 mg/Kg/d for 8 days. The dose was increased to 50 mg/kg/d for the remainder of the study to improve the antitumoral response. GZ17-6.02 demonstrated significant anti-tumor effects compared to the vehicle control treated mice (p < 0.01, Fig. 6C). In addition, tumors from GZ17-6.02 treated mice demonstrated a significant decrease in pERK1/2 levels (p = 0.0006, Fig. 6D). These data indicate substantial anti-tumor effects of GZ17-6.02 in HNSCC preclinical models.
Although promising anti-tumor effects are observed in preclinical models, a shortcoming of the current study is the lack of a clinical trial. Nevertheless, this study paves the way for future clinical studies.

Discussion
The safety and proven efficacy of nutraceuticals has spurred research enthusiasm 36 . In HNSCC, as well as other tumor types, the usefulness of combination therapy has also gathered great attention as targeting multiple components of a signaling pathway more effectively mitigates tumor progression 37 . However, few studies have examined combining nutraceuticals to potentiate their effects 18 , and none have studied a nutraceutical combination for HNSCC. As such, we hypothesized a combination therapy of known anti-cancer nutraceuticals could potentiate antitumor effects. Nonetheless, overly complex combinations of a plethora of nutraceuticals diminishes, and may even worsen, expected effects 38 . In this work, the nutraceutical curcumin, which had demonstrated putative efficacy in previous combinatorial approaches, was combined with two nutraceuticals proposed to have anti-cancer activity, Arum palaestinum and Peganum harmala. Our results demonstrate the rationally designed combination of GZ17-6.02, a blend of curcumin, harmine, isovanillin, is superior in anti-cancer efficacy compared to any single agent, and demonstrated robust anti-tumor effects in preclinical models.
In HNSCC, curcumin enhances cisplatin efficacy by increasing caspase 3/9 activity 39 , and in breast cancer, curcumin decreases proliferation by inhibiting phosphorylation of ERK1/2 40 . In other tumor types, curcumin demonstrated promising combinatorial effects when combined with both natural compounds and effective drugs 18 . Our study corroborated these findings in that the combination therapy with GZ17-6.02 induced significant apoptosis through caspase 3, as well as inhibited phosphorylation of ERK1/2. Curcumin inhibits the production of reactive oxygen species in macrophages. This is an important step preventing macrophage-mediated inflammation-induced tumor progression 41 . Further, curcumin has been shown to induce the differentiation of macrophages to the tumor inhibitory M1 phenotype 42 . To this end, we demonstrated a significant reduction in macrophages in our MMD model. Harmine, the naturally occurring β-carboline, has anti-tumor and anti-angiogenic effects by intercalating and damaging DNA, and inducing tumor-suppressor p53 24,43,44 . Our studies substantiate these findings as GZ17-6.02 significantly decreased HNSCC proliferation and angiogenesis. Further, it reduced macrophage viability and infiltration into tumors in vitro. Isovanillin is a phenolic aldehyde, an isomer of vanillin that has not been extensively studied in the published literature for anticancer effects. However, isovanillin is known for its antioxidant effects 45 , and has been proposed as a potential antitumor agent 22 . In our preliminary studies, the combination of harmine and curcumin was potentiated by isovanillin, and, as such, we established the use of all three in the combined therapeutic.
As cancer pathogenesis is complex with many oncogenic pathways, the ability of an anti-cancer formulation to target multiple pathogenic steps is greatly desirable 37 . GZ17-6.02 decreases multiple signaling cascades, such as Src, ERK1/2, AKT, Stat-2, EGFR, and Chk-2. With molecular docking analyses, GZ17-6.02 components curcumin and harmine demonstrate high binding affinity with the active domains of EGFR, ERK1/2, and Akt1. Mutations in cancer related genes, including Akt1 and EGFR, and intratumoral heterogeneity can induce chemotherapy resistance 46,47 . The simultaneous binding of two or more ligands to the same target could be helpful in alleviating such resistance. Further, cisplatin resistance remains a significant hurdle for HNSCC treatment. We demonstrate the potentiated effects of standard of care cisplatin 35 , with GZ17-6.02. Previous studies indicate the improvement of cisplatin therapy with inhibition of Chk-2 48 , and we demonstrate the deactivation of Chk-2 with GZ17-6.02 administration, providing a possible explanation to the potentiation we observed.
Whereas nutraceuticals are often limited by bioavailability, we demonstrate the combination therapy GZ17-6.02 decreases growth of HNSCC xenograft tumors, syngeneic SCC tumors in an immunocompetent mouse model, and patient-derived xenografts of HNSCC. In establishing an in vivo dose, 15 mg/kg/day was initially tested by intratumoral injection. When the treatment was delivered via oral gavage, the dose was increased to 100 mg/kg/day in the syngeneic model. To assess a lower dose by oral gavage, 30 mg/kg/day was initially tested in the PDX model, however, to improve the antitumor efficacy, the dose was increased to 50 mg/kg/day on the eighth day. In all models, formulations were well tolerated in mice, mouse weight was consistent between treatment arms, and no adverse effects were observed. This tolerability is supported by our in vitro findings of a high ED 50 in the non-cancerous cell line Het1A.
Therefore, the potentiated effects observed through combination in vitro align with the significant anti-tumoral responses in animal models. Further, GZ17-6.02 significantly decreases phospho-ERK1/2 levels in vivo, which supports the in vitro mechanistic experiments, and the molecular docking analyses. A limitation to our study is the lack of comparing GZ17-6.02 with single agents or combinations of two agents in vivo. However, these agents have had little clinical efficacy as single agents in patients, and no toxicities were observed in combination therapy in the in vivo models we assessed. Thus, we demonstrate GZ17-6.02 to be an effective anti-tumor therapy for the treatment of HNSCC in preclinical models, and pave the way for future clinical trials using this combination.

Materials and Methods
Cells and Reagents. Well characterized HNSCC cell lines were used in this study 49 . HNSCC lines HN5, UM-SCC-1 and OSC19, immortalized esophageal Het1A line from cancer free patient, glioblastoma line U87, and murine SCC/vII maintained in high glucose DMEM (Corning, NY, USA) with 10% heat-inactivated fetal-bovine serum (FBS). Lung cancer lines 201T and A549 grown in Basal Eagle Media (for 201T) and Hams F12K media (for A549) both supplemented with 10% FBS. All cells were grown for no more than 10-12 passages. HUVECs maintained in endothelial basal medium-2 supplemented with endothelial growth media-2 SingleQuot Kit (Lonza, Basel, Switzerland) per manufacturer's instructions. Primary HNSCC tumors were collected under the auspices of the Biospecimen repository core at the University of Kansas Cancer Center with written informed consent from patients, using protocols approved by Institutional Review Boards at the University of Kansas Medical Center (KUMC), and all experiments were performed in accordance with the relevant guidelines and regulations.
Cytotoxicity assay. Cells (4 × 10 3 cells/well, 96-well plate) were treated in triplicate with GZ17-formulations or indicated components at various concentrations, and cell proliferation assessed using CyQUANT assay kit (Life Technologies, Waltham, MA) per manufacturer's instructions. ED 50 calculated using non-linear curve fit with GraphPad Prism software version 6.03 (GraphPad Software Inc., San Diego, CA).
To assess the efficacy of GZ17-06.02 to potentiate the effects of radiation on HNSCC, OSC19 (2 × 10 3 cells/well in 96-well plates) were treated with 3, 6, and 9 Gy of radiation. Plates were exposed to gamma radiation (J.L. Shepherd and Associates Mark I Model 68 A cesium-137 source irradiator; dose rate = 2.9 Gy/min). Media was aspirated, and cells were treated with GZ17-06.02 ED50 concentration or vehicle control in DMEM with 10% FBS.
Migration and invasion assay. Cells (2 × 10 4 cells per insert) were seeded in trans-well inserts with 8 μm pores (Becton Dickinson, Franklin Lakes, NJ). For invasion assay, a layer of diluted Matrigel (2 mg/ml) in DMEM (BD Biosciences, San Jose, CA) was placed in the insert. Cells in serum-free media were seeded onto Matrigel layer for invasion or directly onto insert for migration assay. The inserts were placed in duplicate holding-wells containing GZ17-formulations (ED 50 ) in complete media for 24 h. Treated cells were plated in parallel to assess viability using CyQuant. The number of cells that moved to other side of membrane were counted after fixation and staining with Hema3 kit (Fisher Scientific). The numbers of invading or migrating cells were normalized to cell viability. Additional details provided in supplemental methods.
Tubule Formation Assay. HUVEC cells were plated on Matrigel in triplicate with GZ17-6.02 or vehicle control (15 × 10 3 cells/well of 96-well plate). After 6 h, images were taken from 5 random fields per well. Images analyzed using Pipeline software version 1.4 (Medical College of Wisconsin, Milwaukee, WI) according to published instructions to quantify total tube length 50 .

Macrophage infiltration and viability.
To test the effect of GZ17-formulations in attenuation of macrophage infiltration into the tumor, we used the metastasic mimetic device (MMD) as previously described 33 . Briefly, 2.5 × 10 5 OSC19 cells were embedded in rat tail collagen in the MMD and allowed to gel overnight. 5 × 10 5 Thp1 cells were plated into the outer chamber of the device in serum containing medium. The collagen plug was imaged or processed for immunofluorescence analyses after 48 h. The number of macrophages that invaded the HNSCC containing collagen plugs were quantified using ImageJ software. Viability of Thp1 cells were assessed by counting cells on a hemocytometer with using trypan blue dye exclusion. HNSCC cell viability was assessed by imaging cells stained with 0.1 µM Calcein AM dye for 60 min at 37 °C in the dark. probed with p-ERK1/2 (p44/42-pMAPK; Thr202/Tyr204), total ERK1/2, cleaved PARP, caspase-3 (Cell Signaling, Danvers, MA) and β-tubulin (Fisher). Immunoreactivity detected using anti-mouse or anti-rabbit IgG conjugated to Dylight-680 or −800 (Fisher). Protein bands detected using Li-Cor Odyssey protein imaging system (Li-Cor Biotechnology, Lincoln, NE) and quantified using ImageJ. Additional details provided in supplemental methods.
Computational Molecular Docking. Ligands designed using ACD/ChemSketch software (ACDlabs, Ontario, Canada). Crystal structures of proteins downloaded from RCSB-Protein Data Bank (http://www.rcsb. org/pdb/home/home.do). AutoDock-Vina employed for molecular docking using protein and ligand information. During the docking procedure, proteins were considered as rigid and ligands as flexible. The docking pose with lowest binding energy and highest binding affinity was aligned with the protein structure. Additional details provided in supplemental methods.
Cell cycle analysis. OSC-19 cells (2 × 10 5 ) treated in triplicate with ED 50 concentrations of GZ17-formulations for 72 h. Cells were washed with PBS, fixed with 70% ethanol, stained with propidium iodide (0.02 mg/ml in PBS with 0.1% TritonX-100) (Life Technologies), and subjected to cell cycle analysis using Accuri C6 Flow Cytometer (BD Biosciences). Data were analyzed using the software provided by the manufacturer and samples analyzed in triplicate.
In vivo studies. Animal care was in strict compliance with IACUC guidelines at the University of Kansas Medical Center, and all experimental protocols were approved by the IACUC. Mice were housed in individually ventilated cages in a sterile environment, and all experimental work was conducted during daytime hours. All outcome measures were assessed by a blinded observer. To assess in vivo antitumor efficacy of GZ17-formulations, 1 × 10 6 HNSCC (OSC19) cells were injected subcutaneously into the flank of athymic nude-Foxn1 nu mice (Harlan Sprague Dawley, Indianapolis, IN). Tumor bearing mice were randomized once tumor volume reached 5.0 mm 3 . Mice treated with 15 mg/kg/day of GZ17-formulations or with saline control by 50 µL intratumoral administration once daily, five days/week. Tumor diameters measured in two perpendicular dimensions and volume calculated as previously described, briefly (tumor volume = long dimension x (short dimension)^2 × 0.52) 51 . Animals euthanized by CO 2 asphyxiation followed by pneumothorax.
To assess GZ17-6.02 in immunocompetent host, a syngeneic HNSCC model was used 52 . 1 × 10 6 SCC/vII cells were inoculated in the flank of C3H mice. Tumors were established for 6 days, and mice randomized into treatment groups (n = 10/group). Treatment was delivered suspended in 1% carboxymethylcellulose (CMC) by oral gavage (100 mg/kg/day). Mouse weight and tumor volumes were assessed three times per week.
To assess GZ17-6.02 in patient-derived HNSCC xenografts, 35 mg of primary HNSCC tissue was surgically implanted subcutaneously into flanks of athymic nude-Foxn1 nu mice under inhalant isofluorane anesthesia. Intraoperatively animals were administered analgesic Buprenorphine SR at 0.3 mg/kg and tofacitnib (to reduce natural killer cell counts) at 15 mg/kg via subcutaneous injections. Established tumors were passaged twice through mice before implantation for study. Tumor bearing mice were randomized in treatment groups (n = 10/group). Mice were treated with 30 mg/kg GZ17-6.02 or 1% carboxymethycellulose control via oral gavage once daily, five days/week. The dose was increased to 50 mg/kg/day on treatment day 8 to improve antitumoral response. As PDX models have variable growth rates, data presented for in vivo studies as fractional tumor volume. Tumors were excised, snap frozen in dry ice and analyzed by immunoblotting. Statistical Analysis. All results are cumulative from three independent experiments, unless otherwise indicated. Data analyzed using nonparametric Mann-Whitney tests. Analyses performed using Graphpad Prism Version 6.03. Data Availability. The data generated or analyzed during this study are available from the corresponding author on reasonable request.