Honokiol Decreases Lung Cancer Metastasis through Inhibition of the STAT3 Signaling Pathway

Lung cancer is the leading cause of cancer death worldwide. Brain metastasis is one of the most intractable clinical problems associated with lung cancer and is a major cause of lung cancer mortality (1, 2). It is estimated that approximately 10% of patients possess brain metastases at the time of their lung cancer diagnosis, whereas 40% to 50% of patients develop brain metastasis during a typical course of lung cancer disease (2). Because of the difficulty of drug transport through the blood–brain barrier (BBB), the only available therapies to address central nervous system (CNS) metastases include whole brain/CNS irradiation or surgical resection in eligible patients with non-EGFR–mutant lung cancer. Patients with EGFR-mutant disease are treated with anti-EGFR agents (2). All of these therapies are purely palliative and elicit significant toxicities. Therefore, naturally occurring agents that produce little or no toxicity, and that can be delivered systemically to the original tumor site and to the brain, may prove highly efficacious for lung cancer treatment.

Honokiol (HNK) is a key bioactive compound present in the extracts of magnolia bark. The extracts have been used as a folk remedy in Asian countries to treat gastrointestinal disorders, cough, anxiety, stroke, and allergic diseases for centuries. HNK has a favorable bioavailability profile in rodents with a sustained plasma concentration in mice and a 2-compartment pharmacokinetic profile in rats (3, 4). Our recent study measuring oxygen consumption rate in whole intact cells demonstrated that HNK may directly target mitochondria, leading to rapid and persistent inhibition of mitochondrial respiration. This results in the induction of apoptosis in lung cancer cells and ultimately attenuates lung squamous cell carcinoma (SCC) growth in the N-nitroso-tris-chloroethylurea–induced murine model of lung SCC (5). Remarkably, HNK has been shown to readily cross both the BBB (6, 7) and blood–cerebrospinal fluid barrier to inhibit brain tumor growth in rodent models (6), which prompted us to develop the hypothesis that HNK may not only inhibit lung tumorigenesis but also suppress lung cancer brain metastasis.

Several mechanisms of action have been suggested for HNK as an antitumor agent, including induction of apoptosis by causing mitochondrial dysfunction and endoplasmic reticulum stress (8), cell-cycle arrest (9), and inhibition of tumor invasion via downregulation of EGFR, NF-κB, Ras/ERK, PI3K/AKT, and Akt/mTOR pathways (10–13). One key mechanism of action for HNK is the induction of apoptosis through a mitochondrial-dependent mechanism (5, 7). We demonstrated that HNK suppresses mitochondrial respiration and increases the generation of reactive oxygen species in mitochondria, leading to the induction of apoptosis in lung cancer cells (5). Recently, the mitochondrial proteins SIRT3 and Grp78 (an apoptosis-associated protein) have been suggested as possible targets of HNK. Interestingly, STAT3 is a major downstream mediator of these pathways and is also known to play a major role in regulating mitochondrial activity (14–18). STAT3 is a well-known oncogene that can be regulated by receptor tyrosine kinases (RTK), G-protein–coupled receptors, and interleukin families via phosphorylation. Phosphorylated STAT3 undergoes dimerization and translocalization in either the nucleus or mitochondria to mediate its activity, resulting in enhanced cell proliferation, invasion, and survival for many cancer types (19, 20).

In the current study, we evaluated the ability of HNK to prevent lung cancer metastasis to lymph nodes and brain using well-established murine models. We also explored the potential role of RTKs as targets of HNK in the inhibition of lung cancer brain metastasis. We found that a major effect of HNK is the inhibition of STAT3 phosphorylation. We also determined the role of STAT3 in mediating the anticancer effects of HNK in lung cancer by knocking down endogenous STAT3 in the brain metastatic lung cancer cell lines (PC9-BrM3 and H2030-BrM3). Our results showed that HNK significantly inhibits STAT3^Tyr705 and STAT3^Ser727 phosphorylation in both cell lines. STAT3 knockdown abrogated the antiproliferative and antiinvasive effects of HNK. Understanding this novel mechanism of action for HNK may lead to the development of a new class of chemopreventive agents that not only inhibit lung cancer locally (5), but also have the potential to inhibit distal metastasis, which could directly benefit patients.

Brain metastatic lung cancer cell lines, PC9-BrM3 and H2030-BrM3, were generously provided by Dr. Joan Massagué (Cancer Biology and Genetics Program, Memorial Sloan Kettering Cancer Center, New York, NY), and were not authenticated by the authors. Both cell lines were maintained in RPMI1640 medium (Invitrogen) supplemented with 10% FBS (Invitrogen) at 37°C in a humidified atmosphere of 5% CO₂ and air. HNK was purchased from Sigma-Aldrich.

For cell proliferation assays, cells were seeded onto 96-well tissue culture plates at 2–3 × 10³ cells per well. Twenty-four hours after seeding, cells were exposed to various concentrations of HNK for 48 hours, while the control cells received medium only. The plates were incubated at 37°C and 5% CO₂, and cell growth was monitored by IncuCyte (Essen BioScience). Data analysis was conducted using IncuCyte 2011A software. All assays were performed in triplicate.

Cell invasion was determined using Boyden chamber transwells that were precoated with a growth factor reduced Matrix (Thermo Fisher Scientific). Transwell invasion assays were performed as described in the manufacturer's protocol. Briefly, 3 × 10⁵ cells were seeded into each transwell containing serum-free RPMI1640 media and 10 μmol/L HNK. Bottom wells contained RPMI1640 media supplemented with 10% FBS and 10 μmol/L HNK. After 36 hours, cells that had invaded through the transwell were fixed with 10% formalin, stained with 5% crystal violet in 70% ethanol, and counted in three randomly selected areas of each transwell using an inverted tissue culture microscope at ×10 magnification. The results were normalized to controls.

H2030-BrM3 cells, treated either with DMSO (vehicle control) or HNK at various concentrations for 6 hours, were lysed with 200 μL of 1× NP40 lysis buffer containing proteinase inhibitor cocktails (Thermo Fisher Scientific), sheared 10 times with a 28-gauge needle, spun at 16,000 × g for 30 minutes, and normalized by protein concentration as determined by the Bradford method. Normalized lysate was resolved in PathScan RTK Signaling Array, and the signaling array was examined by Li-COR Odyssey infrared imaging system (Li-Cor).

Cells were lysed with 200 μL of RIPA buffer containing proteinase inhibitor cocktails (Thermo Fisher Scientific), sheared 10 times with a 28-gauge needle, spun at 16,000 × g for 30 minutes, normalized by protein concentration as determined by the Bradford method, and boiled for 5 minutes. Normalized lysate was resolved by 4% to 12% SDS-PAGE (Thermo Fisher Scientific) and immunoblotted with indicated antibodies; p-EGFR (#3777S), p-STAT3 (#9134S), p-AKT (#4060S), EGFR (4267S), STAT3 (9139S), and AKT (9272S), which were all purchased from Cell Signaling Technology. Actin (SC-8432) was purchased from Santa Cruz Biotechnology.

Lentiviral particles against STAT3 were purchased from Santa Cruz Biotechnology. PC9-BrM3 and H2030-BrM3 cells were infected with lentiviral particles using 8 μg/mL polybrene, and the infected cells were selected by treatment with puromycin (2 μg/mL) for 3 days.

Direct interaction between HNK and candidate RTKs was examined via the KINOMEscan binding assay from DiscoveRx.

We conducted an RNA sequencing (RNA-seq) study of human lung tumor metastases in mouse brains. Three brain metastases were sampled from mice without HNK treatment and another three brain metastases were obtained from mice treated with HNK. Total RNA samples were extracted from these six samples using Qiagen RNeasy Mini Kit. We used NEBNext Ultra RNA Library Prep Kit from Illumina to construct the RNA-seq libraries for these samples. Whole transcriptome analysis of the RNA-seq library samples was performed using HiSeq 2500 sequencing platforms (Illumina). The experiment was single-end with 50 nucleotides read length. Coverage for the samples ranged from 15 million to 32 million reads per sample. To identify and unequivocally separate graft (human) and host (mouse) reads, processed sample reads were sequentially aligned to both graft [complete hg19 human genome (UCSC version, February 2009)] and host [complete mm9 mouse genome (UCSC version, July 2007)] genomes using Bowtie-TopHat (21, 22). Read counts were obtained using HTseq (23). Data normalization and differential expression analysis were performed using the statistical algorithms implemented in EdgeR Bioconductor package (24). FDR, corrected P values of less than 0.05, was used as criteria for significantly regulated genes. We used a strategy that efficiently separates human lung tumor sequence data from xenograft mouse (mice with genetically human tumors) sequence data into separate microenvironment and tumor expression profiles (25, 26). Using this tool, we obtained more accurate RNA expression profiles for both metastatic human lung tumors and mouse stromal cells.

Animal procedures were in accordance with the Medical College of Wisconsin Institutional Animal Care and Use Committee. For lung cancer brain metastasis study, 4- to 6-week-old female NOD/SCID mice were used. Brain-seeking H2030-BrM3 cells (2 × 10⁵) were suspended in 0.1 mL PBS and injected into the left ventricle (LV) under ultrasound guidance (ECHO 707, GE). One day after engrafting H2030-BrM3 cells into the arterial circulation, mice were randomly grouped into vehicle treatment group and HNK treatment group (10 mg/kg b.w.). Mice were treated with either solvent control (0.1% DMSO in corn oil) or HNK by oral gavage for 4 weeks, metastasis was monitored over time by bioluminescence with an IVIS 200 Xenogen and confirmed with ex vivo luminescence, GFP fluorescence followed by hematoxylin and eosin (H&E), and GFP staining. For analysis of lung tumor lymph node metastases, H2030-BrM3 (10⁴) cells were suspended in a 1:2 mixture of PBS and growth factor reduced Matrigel (BD Biosciences) and injected into the lung. HNK treatment was initiated one day after orthotopic injection of tumor cells by oral gavage.

We used an orthotopic model of lung adenocarcinoma cells (H2030-BrM3 cells) in athymic nude mice to evaluate the inhibitory effect of HNK on lung tumor growth and lymph node metastasis. Five-week-old male athymic nude mice were used for the experiments. Mice were anesthetized with isoflurane and placed in the right lateral decubitus position. A total of 1 × 10⁶ H2030-Br3 cells in 50 μg of growth factor reduced Matrigel in 50 μL of RPMI1640 medium were injected into the left lung through the left rib cage as described previously (27). One day after injection, mice in the HNK group were treated with 2 or 10 mg/kg b.w. HNK, once a day, 5 days per week for four consecutive weeks. Tumor growth and metastases phenotype was monitored over time by bioluminescence with an IVIS 200 Xenogen. Mice were euthanized at the endpoint; tissues were immediately fixed in optimal cutting temperature and frozen in liquid nitrogen for subsequent Western blot and immunohistochemical analyses.

In vivo imaging system (IVIS) consists of a highly sensitive, charge-coupled digital camera with accompanying advanced computer software for image data acquisition and analysis. This system captures photons of light emitted by reagents or cells that have been coupled or engineered to produce bioluminescence in the living animal. The substrate luciferin was injected into the intraperitoneal cavity of mice at a dose of 150 mg/kg b.w. (30 mg/mL luciferin), approximately 10 minutes before imaging. Mice were anesthetized with isoflurane/oxygen and placed on an imaging stage. Photons emitted from the lung region were quantified using Living Image software (Xenogen Corp.).

Mouse brains were fixed in 10% zinc formalin solution overnight and stored in 70% ethanol for histopathology. Serial tissue sections (5-μm each) were made and stained with H&E or GFP and examined histologically under a light microscope to assess severity of tumor development.

Data were analyzed by one-way ANOVA. * P < 0.05, ** P < 0.01, and *** P < 0.001 were considered statistically significant.

Previous research in our laboratory and in other laboratories demonstrated the anticancer effect of HNK in many cancer types, including lung (4, 5, 9, 11, 12, 28). To evaluate the effects of HNK in brain metastatic lung cancer, we examined the effects of HNK on the proliferation and invasion of PC9-BrM3 and H2030-BrM3 brain metastatic lung cancer cells. Initially, PC9-BrM3 and H2030-BrM3 cells were treated with various concentrations of HNK for 96 hours to examine the antiproliferative effects of the compound. HNK effectively inhibited both PC9-BrM3 and H2030-BrM3 cell proliferation in a dose- and time-dependent manner (IC₅₀ for PC9-BrM3 is 28.4 μmol/L, for H2030-BrM3 is 25.7 μmol/L; Fig. 1A). We also examined the effects of HNK on the invasion of PC9-BrM3 and H2030-BrM3 cells in the Boyden chamber (Fig. 1C and D). As shown in Fig. 1C and D, HNK significantly inhibited the invasion of both PC9-BrM3 and H2030-BrM3 cell lines in a dose-dependent manner. The doses of HNK required to reduce the invasion of PC9-BrM3 and H2030-BrM3 cells were much lower (Fig. 1C and D) than those required to inhibit cell proliferation (Fig. 1A and B). On the basis of previous research (5–7, 29–31) and our current data, HNK is not only an effective chemopreventive/chemotherapeutic agent, but could also be an effective agent to prevent or inhibit invasion of lung cancer.

In the mice implanted orthotopically with H2030-BrM3 cells in the left lung, lung tumors grew and spread within the lung and then to the mediastinum. The incidence of tumor formation was 100%. Representative bioluminescence images of lung orthotopic xenografts are shown in Fig. 2A. Mice did not show any observable side effect when treated with HNK. Higher doses of HNK significantly decreased lung tumor growth when compared with the vehicle control group (Fig. 2B). Comparing with the control group, HNK, at the higher dose (10 mg/kg b.w.), significantly reduced the incidence of mediastinal adenopathy. The incidence of mediastinal lymph node metastasis in the control group was 100%; in the high-dose HNK treatment group, only 2 of 6 mice had lymphatic metastasis. The high-dose HNK also significantly decreased the weight of mediastinal lymph nodes over 80% compared with control group (Fig. 2C).

In this assay, we used an ultrasound-guided procedure to insure the injection of brain-seeking H2030-BrM3 lung cancer cells into the LV of NOD/SCID mice (Fig. 3A). One day after cell inoculation, the mice were randomly grouped to vehicle control and HNK low- (2 mg/kg b.w.) and high-dose (10 mg/kg b.w.) groups. High-dose HNK significantly decreased brain metastasis over 70% when compared with the vehicle control group (Fig. 3B). At necropsy (28 days post-LV injection), the extent of brain metastases was also quantified by ex vivo bioluminescence and GFP imaging as shown in Fig. 3C. HNK treatment decreased brain metastasis to approximately one third of that observed in control mice. Lung tumor cell migration to brain was confirmed by H&E staining, as well as GFP staining (Fig. 3D and E). Collectively, our data suggest that HNK could be effective in preventing the metastasis of lung cancer cells to the brain.

Potential mechanisms of action of HNK in the inhibition of lung cancer cell brain metastasis were examined via RTK assays (Fig. 4A). H2030-BrM3 cells were treated with 10 and 20 μmol/L HNK for 6 hours. PathScan RTK signaling array revealed that HNK treatment dramatically decreased STAT3 phosphorylation (Fig. 4B and C), suggesting that STAT3 is at least one molecular target of HNK. The effects of HNK on STAT3 phosphorylation in PC9-BrM3 and H2030-BrM3 cells were confirmed by Western blot analysis (Fig. 4D). Previously, HNK was found to be effective in the treatment of head and neck squamous cell carcinoma (HNSCC) via targeting the EGFR signaling pathway (32). In the current study, we also observed that HNK targets the EGFR–AKT signaling pathway in PC9-BrM3 cells, which harbor an EGFR mutation, but not in H2030-BrM3 cells, which harbor Kras mutations. In addition, we examined the interaction between HNK and multiple RTKs via the KINOMEScan binding assay. As shown in Supplementary Fig. S1, HNK did not bind directly to any of the RTKs tested.

shRNA knockdown was used to demonstrate the role of STAT3 in mediating the effects of HNK in lung cancer. Knockdown of STAT3 in PC9-BrM3 and H2030-BrM3 cells was validated by Western blot analysis (Fig. 5A). STAT3 knockdown decreases the antiproliferative (Fig. 5B) and antiinvasive (Fig. 5C and D) effects of HNK in both PC9-BrM3 and H2030-BrM3 cell lines. HNK treatment (20 μmol/L) for 48 hours inhibited the proliferation of PC9-BrM3 vector control cells by 30% and H2030-BrM3 vector control cells by 20% but had no significant effect on the proliferation of STAT3 knockdown PC9-BrM3 or H2030-BrM3 cells (Fig. 5B). In addition, HNK treatment (10 μmol/L) significantly inhibited the invasion of both PC9-BrM3 and H2030-BrM3 vector control cells but had no effect on the invasion of STAT3 knockdown PC9-BrM3 or H2030-BrM3 cells (Fig. 5C and D). Finally, we examined the effects of STAT3 on mitochondrial respiratory function in PC9-BrM3 and H2030-BrM3 cells. As shown in Fig. 5E, STAT3 knockdown significantly decreased mitochondrial respiratory function in both PC9-BrM3 and H2030-BrM3 cell lines. The anticancer effects of HNK, therefore, could be through inhibition of STAT3-mediated mitochondrial functions in lung cancer cells that have metastasized to the brain.

The differentially expressed genes identified by RNA-seq analysis software were subjected to IPA analysis (Ingenuity Pathway Analysis; http://www.ingenuity.com/products/ipa) to identify the most significant oncogenic pathways in metastatic lung tumors changed by HNK treatment. Our genome-wide RNA-seq scan showed that STAT3 pathway is the top downregulated one among the oncogenic pathways that were significantly downregulated in the HNK-treated human lung tumor metastases in mouse brains (Fig. 4A). In addition, RNA-seq analyses identified that six key genes involved in the activation of STAT3 pathways were significantly downregulated in metastatic lung tumors in vivo upon HNK treatment (Table 1). They were FGFR4, IGF1R, IGF2R, MAP2K1, MAP3K11, and SRC. These matched the findings from our functional studies and supported that the anti–lung cancer role of HNK was mediated via the STAT3 signaling pathway.

One of the common sites for metastases of lung cancers is the brain. Currently available therapies to address CNS metastases include whole brain/CNS irradiation or surgical resection in eligible patients, treatment with anti-EGFR agents in patients whose tumors contain EGFR mutations, as well as using next-generation ALK TKI that is brain-penetrable such as PF-06463922, to control CNS metastases in lung cancer patients (2, 33). However, these treatment options are available only after the diagnoses of brain metastases, and in many cases, metastatic lesions remain undiagnosed for long periods or they are not amenable to treatment with chemo/radiotherapy or surgery. Therefore, it is necessary to develop prevention strategies to inhibit metastases from primary tumors. Recently, we demonstrated the ability of HNK to potently inhibit the development of lung tumors in mice (5). Analysis of HNK's mechanism of action suggests that its effect is primarily mediated by inducing apoptosis through a mitochondria-dependent mechanism (5, 7, 34). Here, through the use of the well-characterized brain metastases murine model, we report that HNK exerts inhibitory effects on the metastasis of lung cancer cells to the brain, indicating that the compound has chemopreventive potential against both primary lung tumors and on metastasis of lung cancer to the brain.

Direct injection of tumor cells into the LV is the most widely used brain metastasis model in rodents because it bypasses the precolonization steps of dissemination of cancer cells through the bloodstream, homing, and extravasation, and recapitulates the process of cancer cells crossing the BBB and growing within the brain microenvironment. Metastatic brain lesions in mice vary from round, circumscribed lesions typical of that seen on human scans, to infiltrative tumor cells, which over time form typical round lesions that are ideal for evaluating the preventative effect of HNK on lung cancer metastasis. The brain homing H2030 and PC9 lung cancer cell lines were developed to have 100% brain metastatic potential (5–7, 29–31), H2030 cells with a KRAS^G12C mutation (35) and PC9 cell with an EGFR^Δexon19 mutation (36). These cell lines were engineered to stably express GFP–luciferase fusion for real-time monitoring of metastatic tumor growth. In the current study, we monitored metastatic tumor growth using both live animal imaging and endpoint ex vivo imaging. We also validated tumor growth by staining the brain tissues with H&E and GFP, and both stains consistently demonstrated about a 70% inhibition of brain metastases by HNK.

At least one of the mechanisms through which HNK inhibited lung cancer cell metastasis to the brain was through the inhibition of STAT3 phosphorylation. HNK is known to target multiple signaling pathways, including EGFR, MAPK, and PI3K/AKT (10–13). Recently, Sirt3 and GRP78 were also suggested as potential binding targets of HNK in different tissue types (34, 37). Interestingly, STAT3 is a major downstream mediator of multiple RTK pathways (14–17). Our data suggest that STAT3 could be a universal downstream target of HNK treatment. As indicated before, HNK was effective in the treatment of HNSCC via targeting the EGFR signaling pathway (38). The brain homing H2030 and PC9 lung cancer cell lines carry different driver mutations, H2030 with a KRAS^G12C mutation (35) and PC9 with an EGFR^Δexon19 mutation (36). In PC9-BrM3 cells, we observed downregulation of phosphorylated EGFR by HNK, but not in H2030-BrM3 cells, which do not carry an EGFR mutation. Therefore, the effects of HNK on the EGFR–AKT signaling pathway could be cell type- or tissue-specific. RNA-seq data suggest that FGFR4 is the most significant gene that was affected by HNK treatment, and FGFR4 is known to mediate STAT3 signaling pathway (39). Therefore, it will be interesting to investigate the role of FGFR4–STAT3 signaling pathway in mediating the anticancer effects of HNK. STAT3 phosphorylation was reduced in both PC9-BrM3 and H2030BrM lung cancer cell lines by HNK, and knockdown of endogenous STAT3 in these cell lines abrogated the antiproliferative, antimigratory, and anti-invasive effects of HNK, further supporting the concept that STAT3 could be a universal downstream target of HNK regardless of EGFR mutation status of lung cancer cells. Although, with prolonged treatment (over 48 h), HNK will eventually inhibit proliferation of STAT3 knockdown cells (data not shown), which most likely would be due to the off-target effects of HNK, considering it is a polyphenol compound, once its main target has been blocked, it may target other pathways to inhibit tumor growth. STAT3 knocked down PC9-BrM3 and H2030-BrM3 cells exhibit significantly less mitochondrial respiratory function than normal lung cells. HNK inhibition of lung cancer progression via inhibition of mitochondrial respiratory function could, therefore, be due to inhibition of STAT3 phosphorylation which, in turn, leads to inhibition of the metastases of lung cancer cells to the brain.

No potential conflicts of interest were disclosed.

Conception and design: Y. Lee, M. You

Development of methodology: J. Pan, Q. Zhang, T.C. Wan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Pan, Y. Lee, Q. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Pan, Y. Lee, Q. Zhang, D. Xiong

Writing, review, and/or revision of the manuscript: J. Pan, Y. Lee, Q. Zhang, D. Xiong, Y. Wang, M. You

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Lee, Y. Wang

Study supervision: M. You

We thank Dr. John Auchampach for his contribution, especially for the guidance of the left ventricle under ECHO 707.

This work was supported by R01CA208648.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.