Imipramine, MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide), RNase, dimethyl sulfoxide (DMSO), and Hygromycin B were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12), fetal bovine serum (FBS), penicillin-streptomycin were purchased from GIBCO^®/Invitrogen Life Technologies (Carlsbad, CA, USA). Primary antibodies against Matrix metalloproteinase-9 (MMP-9) (AB19016, Millipore), vascular endothelial growth factor (VEGF) (ab1316, Abcam, Cambridge, UK), EGFR (Try 1068) (#2234, Cell signaling, Danvers, MA, USA), EGFR (E-AB-63555, Elabscience, Houston, TX, USA), PKC-δ (Thr507) (E-AB-20968, Elabscience), PKC-δ (E-AB-14675, Elabscience), NF-κB p65 (Ser536) (E-AB-70335, Elabscience), NF-κB p65 (E-AB-22066, Elabscience), cell leukemia-1 (MCL-1) (BV-438, BioVision), cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein (cFLIP) (D16A8, Cell signaling), X-linked inhibitor of apoptosis protein (XIAP) (PA5-29253, Thermo Fisher Scientific), Fas (E-AB-40063, Elabscience), Fas ligand (FasL) (E-AB-31410, Elabscience), cleaved caspase-3 (E-AB-30004, Elabscience), cleaved caspase-8 (E-AB-22107, Elabscience), cleaved caspase-9 (#9505, Cell Signaling Technology), PARP-1 (#9532, Cell Signaling Technology) and β-actin (sc-47778, Santa Cruz Biotechnology, Dallas, Texas, Waltam, MA, USA) for Western blotting were purchased from different companies as listed. Secondary antibodies, peroxidase affiniPure Goat Anti-Mouse IgG and Goat Anti-Rabbit IgG were purchased from Jackson Immunoresearch Laboratories Inc. (West Grove, PA, USA). D-luciferin was obtained from Caliper Life Science (Hopkinton, MA, USA).

CL1-5-F4 cells, non-small cells lung cancer (NSCLC), was kindly provided by Dr. Chia-Lin Hsieh (Taipei Medical University, Taiwan), A549, and NCI-460 provided by Professor Jing-Gung Chung (China Medical University, Taiwan) were used for this present study. Cells were maintained in DMEM/F-12 and F12-K containing 10% FBS, 1% PS and incubated at 37°C in a 95% air and 5% CO₂ humidified atmosphere.

Cell viability was performed by MTT assay. Cells were seeded into 96-well plates at a density of 1.5x10⁴ cells/well and incubated overnight. Then treated with different concentrations of imipramine for 24 and 48 hr. After treatment, the medium was replaced with 0.5 mg/ml of MTT solution and maintained in the incubator for another four hr. Before the ELISA reader (Thermo Fisher Scientific, Fremont, CA, USA) analysis, the MTT solution was replaced by 100 μl DMSO. The absorbance wavelength of MTT is 570 nm, the blank value was defined as zero (+/–0.1).

CL1-5-F4 cells were transfected with pNF-κB-luc2 (Promega, Madison, WI, USA) vector using JetPEI™ transfection agent (Illkirch, Bas-Rhin, France). One μg pNF-κB-luc2 plasmid dissolved in 150 mM NaCl and mixed with JetPEI™ agent, then incubated for 30 min at 25°C. The DNA mixture was then added to the CL1-5-F4 cells in a 24 well plate dish and incubated for another 24 hr. After transfection, cells were cultured in a medium containing 200 µg/ml of hygromycin B for two weeks. The surviving clones were subsequently seeded into 96-well plates and imaged by IVIS Lumina LT Series (PerkinElmer, Boston, MA, USA). CL1-5-F4 cells with stable NF-κB expression were renamed as CL1-5-F4/NF-κB-luc2 cells (17).

CL1-5-F4/NF-κB-luc2 cells were plated into 96-well (1.5×10⁴/well) overnight and treated with various concentrations of imipramine (0, 50, 100, 150 and 200 μM), erlotinib (10 μM), hEGF (30 ng/ml), rottlerin (20 nM) and Phorbol myristate acetate (PMA) (75 nM) alone or combined with 100 and 150 μM imipramine for 48 hr. After treatment, 100 μl D-luciferin (Promega, Madison, WI, USA) solution (500 μM D-luciferin) was added into each well and incubated for 5 min in the dark at room temperature before image acquisition. NF-κB activation signal was collected for 3 min by IVIS Lumina LT Series and normalized with cell viability.

CL1-5-F4 cells were seeded into 6-well plates (2×10⁵/well) overnight and treated with 0, 100, and 150 μM of imipramine for 48 hr. Cells were harvested and stained by Annexin V-FITC Apoptosis Detection Kit (Vazyme Biotech Co. Lt, Nanjing, China), DiOC₆ dye (4 μmol/L), FITC-DEVD-FMK (cleaved caspase-3), Red-IETD-FMK (cleaved caspase-8), FITC-VAD-FMK (cleaved caspase-9), ROS peroxide-sensitive fluorescent probe 2′, 7′-dichlorofluorescein diacetate (DCFH-DA 500 μl at 10 μM, Molecular Probes), and fluo-3-acetomethoxyester (Fluo-3/AM, 2.5 μg/mL) for apoptosis detection, respectively. For cell cycle analysis, cells were fixed by 75% ethanol overnight and followed with 40 μg/ml Propidium iodide (6) stained (contained with 100 μg/ml RNase and 1% Triton X-100). For cleaved PARP-1, cells were fixed by 4% paraformaldehyde and permeabilized by ice-cold 100% methanol before PARP-1 stained (3). After staining, cells were resuspended with 300 μl PBS and the signal intensity of each marker was detected by flow cytometry (BD Biosciences, FACS Calibur, San Jose, CA, USA). Quantification results of fluorescence intensity were measured by FlowJo software (version 7.6.1; FlowJo LLC, Ashland, OR, USA).

CL1-5-F4 cells were seeded on coverslips overnight and treated with different concentrations of imipramine. Then, the coverslips were fixed with 3.7% paraformaldehyde for 15 min. Then, slices were permeabilized by 0.1% Triton-X100 for 15 min. Block the coverslips with a blocking buffer (Axel Biotechnology Inc.) at 25°C for one hour. The coverslips were then incubated with primary antibodies (1:250 anti-NF-κB) at room temperature for 1 hr. The coverslips were washed by PBS-T (PBS add 0.1% Tween 20) once and PBS twice. Coverslips were incubated with FITC-labeled secondary antibody (Jackson Immunoresearch Laboratories Inc.) at 25°C and protected from light for 1 hr. Finally, slides were fixed with DAPI mounting buffer and allowed it dry in dark. The fluorescence signal from NF-κB in cells were observed and photographed by Zeiss Axio Scope A1 fluorescence microscope (18).

The 8 μm pore trans-wells were purchased from BD Biosciences (Franklin Lakes, NJ, USA). CL1-5-F4 cells were treated with different concentrations of imipramine for 48 hr and then cells were collected into the upper channel of trans-wells at the number of 5×10⁵, allow 24 hr migration period of CL1-5-F4 cells. Alternatively, to test invasion ability, the upper channel of the trans-well was pre-coated with matrigel. Transwell membranes were then fixed (methanol and acetic acid 3:1) and stained by 0.5% crystal violet. The light microscope (Nikon ECLIPSE Ti-U) was used to photograph the migration and invasion cells at ×100 on the trans-well membrane. The number of migration and invasion cells was calculated using Image J software version 1.50 (National Institutes of Health, Bethesda, MD, USA) (19).

Cells were pre-treated imipramine for 48 hr, then seeded the pre-treated cells into 6 well with ibidi culture-inserts (cat: 80241, ibidi GmbH, Gräfelfing, Germany) insert overnight. The 2-well insert was then removed and a cell migration pattern was observed by microscope at 0, 6, 12 and 24 hr. The migration gap area was quantified by Image-J.

One million CL1-5-F4 cells were seeded into 10 cm dishes overnight and treated with different concentrations (0, 100 and 150 μM) of imipramine, erlotinib (10 uM) or hEGF (100 ng/ml), respectively, for 48 hr. After treatment, the protein was extracted from each group by NP40 lysis buffer. Total protein was measured using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Proteins were separated by 8-15% SDS-PAGE, transferred on polyvinylidene difluoride (PVDF) membrane (FluoroTrans^® Pall Corporation, Port Washington, NY, USA) by the electroblotting system and blocked with 5% fetal calf serum (FCS) in Tris-buffered saline (TBS) containing 0.05% Tween-20. Then, the PVDF membrane was stained by primary antibodies overnight at 4˚C and secondary antibodies for 1 hr at 25°C. Membranes were probed with Immobilon Western Chemiluminescent HRP Substrate (Pierce, Rockford, IL, USA). The chemiluminescence signal from each sample was then detected by the UVP ChemiDoc-It™ (Analytik Jena, Jena, Germany) and their specific band intensities were quantified by VisionWorks (AnalytikJena). Quantification data were all normalized by β-actin expression and averaged by three repeated experiments.

Six-week-old nude mice (N=15) were brought from the National Laboratory Animal Center, Taipei, Taiwan. One million CL1-5-F4/NF-κB-luc2 cells was suspended 100 μl PBS and subcutaneously inoculated in the right legs of nude mice. When average tumor volume reached about 130 mm³, mice were randomly separated into three groups (n=5/group), vehicle group [treated with 0.1% dimethyl sulfoxide (DMSO) by gavage daily for 14 days] and erlotinib group (treated with 20 mg/kg/day by gavage for 10 days), imipramine group (treated with 20 mg/kg/day by gavage for 10 days). Treatment was initiated on day 1, tumor growth and body weights of mice were measured on day 0, 2, 4, 6, 8, 10 after treatment. Mice were sacrificed on day 10 and tumor tissues were isolated for IHC staining. All experiments were repeated at least three times and complied with the guidance of institutional animal care (IACUC number approval by China Medical University: CMUIACUC-2018-323).

Mice were sacrificed on day 10, tumors and livers were fixed with 4% PFA at 4°C, respectively. Paraffin-embedded tumor tissue sections were prepared as 5 μm thickness slices by Bio-Check Laboratories Ltd. (New Taipei City, Taiwan) (20). According to the instructions provided with the EMD Millipore’s IHC Select^® kit, the sections were stained with EGFR (Try1068), PKC-δ (Thr507) and NF-κB (Ser536), cleaved caspase-3, -8, and -9, MMP-9, XIAP, and MCL-1 antibodies, respectively. The stained sections were scanned at 200× magnification by using the microscopy-based TissueFAXS platform (TissueGnostics, Vienna, Austria). Positive expression of IHC indices in tumor tissues was quantified with ImageJ software version 1.50 (National Institutes of Health, Bethesda, MD, USA) (21).

Mice were sacrificed on day 10 and livers were fixed with 4% PFA at 4°C. Heart, liver, kidney, spleen, lung and intestine tissue for mice were extracted to perform H&E staining (20). Paraffin-embedded tumor tissue section with 5 μm thickness and H&E staining was performed by Bio-Check Laboratories Ltd. (New Taipei City, Taiwan).

In vivo, BLI was conducted on a cryogenically cooled IVIS Lumina LT Series using Living Imaging (Caliper Lifesciences). Mice have received an intraperitoneal injection of 150 mg/kg of luciferin potassium salt dissolved in 100 μl PBS (Gold Biotechnology, St. Louis, MO) and subsequently anesthetized with 5% isoflurane (Abbott, North Chicago, IL). The animals were then placed into the BLI chamber and the anesthesia was continually provided with 2% isoflurane by the nose cone. Images were acquired 15 min after luciferin administration. An integration time of 1 min with a binning of 100 pixels was used for luminescence image acquisition. Signal intensity was quantified as total photon flux (p/s/cm²/sr) within the region of interest after subtraction of the background luminescence (22).

One-way ANOVA was performed in this study to compare the difference between control, erlotinib and imipramine treatment groups by Microsoft excel 2017. The P-value that smaller than 0.05 was defined as a significant difference. Each value in this study was displayed as mean ± standard error. Statistical differences between groups were mentioned in each figure legends.

To evaluate the cytotoxicity of imipramine on NSCLC, we used three NSCLC cell lines (CL1-5-F4, A549, and NCI460) and treated them with 0-200 μM imipramine for 48 hr. CL1-5-F4 cells were showed the cell shrinkage morphology after imipramine treatment ( Figure 1A ). Imipramine also decreased cells viability of three NSCLC cell lines a dose-dependent manner ( Figure 1B ). The Annexin-V⁺ cells were all increased along with dosage rising after imipramine treatment in three NSCLC cells ( Figure 1C ). To further investigated whether imipramine-induced apoptosis was associated with caspase-dependent signaling activation, CL1-5-F4 cells were co-treated with imipramine with Z-VAD, a caspase family inhibitor, and tested by Annexin-V/PI staining. As indicated in late apoptosis population, apoptosis induced by imipramine was markedly reversed by Z-VAD ( Figure 1D ). The activation of cleaved caspase-3 was validated by flow cytometry and IF staining. Results in Figures 1E, F indicated that imipramine may effectively induce the activation of cleaved caspase-3. The sub-G1 phase accumulation that represents as apoptotic cells was also found in imipramine-treated CL1-5-F4 cells ( Figure 1G ). Moreover, cleaved PARP-1, an apoptosis marker, were also significantly increased by imipramine ( Figure 1H ). The protein expression of cleaved caspase-3 and cleaved PARP-1 were all increased by imipramine as indicated in Figure 1I . Comet assay which revealed DNA damage pattern was effectively increased in imipramine treated CL1-5-F4 cells ( Figure 1J ). All above, we confirmed that imipramine may induce cytotoxicity and apoptosis of NSCLC cells.

To further investigated if both extrinsic and intrinsic apoptosis pathways were affected by imipramine, flow cytometry and Western blotting assays were utilized to test the activation of the following related markers. In the extrinsic apoptosis pathway, Fas, Fas ligand (Fas-L) and cleaved caspase-8 were all increased by imipramine-treated CL1-5-F4 cells ( Figures 2A–C ). Intrinsic apoptotic pathway markers, including accumulation of cellular calcium (Ca²⁺), loss of mitochondria membrane potential (MMP, ΔΨm), cleaved caspase-9 and production of reactive oxygen species (20) were all investigated induced by imipramine ( Figures 2D–F ). Nevertheless, the highest induction percentage of ROS was found after 12-24 hrs treatment of imipramine but not 48 hrs treatment. Additionally, the protein expression of Fas, Fas-L, cleaved caspase-8 and cleaved caspase-9 were also increased by imipramine ( Figure 2G ). Taken together, we indicated that imipramine markedly triggered both extrinsic and intrinsic apoptosis pathways in CL-1-5-F4 cells.

To investigate the activation of NF-κB after imipramine treatment in CL1-5-F4 cells, we transfected CL1-5-F4 cells with pNF-κB-luciferase reporter gene (pNF-κB-luc2 vector) and performed nuclear translocation assay of NF-κB. Luminescence images indicated the obvious NF-κB activation was suppressed approximately 50-60% by 100-150 μM of imipramine ( Figure 3A ). In addition, we found that the nuclear translocation of NF-κB was reduced by imipramine treatment ( Figure 3B ). In Figure 3B , fewer translocated cells were found in 150 μM as compared to 100 μM of imipramine. Then, we try to identify the upstream regulator of NF-κB, we performed NF-κB reporter gene assay after treated with hEGF and PKC-δ inducers and inhibitors in CL1-5-F4/NF-κB-luc2 cells. Both EGFR inhibitor (erlotinib) and PKC-δ inhibitor (rottlerin) may suppress NF-κB activation ( Figure 3C ). NF-κB which activated by EGFR (hEGF) and PKC inducer (PMA) were be diminished when combining with imipramine (100 and 150 μM). We also identified other upstream regulator of NF-κB including ERK (PD98059) and AKT (LY294002) on NSCLC; however, PKC-δ inhibitor (rottlerin) showed greater NF-κB suppression capacity ( Supplementary Figure ). Furthermore, the protein expressions of EGFR (Try1068), PKC-δ (Thr507), and NF-κB (Ser536) were all downregulated after imipramine treatment ( Figure 3D ). In Figure 3E , erlotinib may suppress phosphorylation of EGFR and PKC-δ, which imply the inactivation of NF-κB is associated with the suppression EGFR/PKC-δ. Additionally, imipramine may successfully suppress hEGF-induced EGFR/PKC-δ ( Figure 3F ). According to the above results, we suggested that imipramine suppressed NF-κB activation of CL1-5-F4 cells was associated with the inhibition of the EGFR/PKC-δ pathway.

The activation of NF-κB was also recognized as a crucial factor in metastasis (23); thus, we investigated the migration and invasion characteristic of CL1-5-F4 cells by wound healing assay and transwell invasion assay. In 20 hours’ endpoint of wound healing assay, non-treated CL1-5-F4 cells only remained 30% of the gap area, while 150 μM imipramine remain 75% of the gap area ( Figures 4A–C ). In migration assay, the number of migrated CL1-5-F4 cells was decreased by imipramine treatment after 48 hr treatment ( Figures 4D, E ). In the invasion assay, the number of invaded CL1-5-F4 cells was also suppressed by imipramine ( Figures 4D, F ). In addition, we used Western blotting to investigate NF-κB related downstream proteins expression, including VEGF, MMP-9, cFLIP, MCL-1, and XIAP (17). Angiogenesis [vascular endothelial growth factor (VEGF), and metastasis-associated proteins (matrix metalloproteinase-9 (MMP-9)], anti-apoptotic protein [(Cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein (cFLIP), myeloid cell leukemia-1 (MCL-1)], and proliferation protein [X-linked inhibitor of apoptosis protein (XAIP)], were all reduced by imipramine as showed in Figure 4G . In sum, our result suggested that NF-κB mediated metastasis ability and tumor progression associated proteins expression was dose-dependently reduced by imipramine.

In an animal experiment, we established subcutaneous CL1-5-F4/NF-κB-luc2 cells bearing model on BALB/c nude mice to validate the treatment efficacy of imipramine. As illustrated in Figure 5A , erlotinib and imipramine groups markedly suppressed tumor growth as compared to control. Tumor growth of each mouse was demonstrated similar growth inhibition pattern after treated with imipramine and erlotinib. The smallest tumor was found in the imipramine treatment group on day 10 as compared to control groups ( Figures 5B, C ). Moreover, the mean tumor growth time of imipramine is 10.8 day, which is markedly smaller than control group ( Table 1 ). The mean tumor growth delay time of imipramine treated group is 5.5 day less than control. The mean growth inhibition rate of imipramine treated group is 2.04 times smaller than control group. Imipramine also markedly decreased the weight of tumor ( Figure 5D ). The statistical result of mice body weight showed no significant difference between the 3 groups ( Figure 5E ). The level of AST, ALT and γGT in serum was used to identify the function of liver. As showed in Figure 5F , the level of AST and ALT remained unchanged after imipramine treatment as compared to vehicle. The level of γGT which represent as liver function was all smaller than 4 ( Table 2 ). However, in heart, liver, kidney, lung, spleen, and intestine pathology results, the chromatin precipitation in liver tissue was only found in the erlotinib treatment group but not in the control group and imipramine group ( Figure 5G ). This result indicated that erlotinib showed noticeable liver toxicity as compared to imipramine. In addition, the NF-κB activation within the tumor was effectively decreased by erlotinib and imipramine ( Figures 5H, I ). In sum, tumor growth and NF-κB activation were both effectively suppressed by imipramine treatment.

Here, we also used IHC staining to observe whether EGFR/PKC-δ/NF-κB signaling was suppressed by imipramine in CL1-5-F4/NF-κB-luc2 bearing tumor. The phosphorylation of EGFR, PKC-δ, and NF-κB was significantly decreased 50-80% by imipramine ( Figure 6A ). Three apoptotic markers included general apoptosis protein-activated caspase-3, extrinsic apoptosis protein-activated caspase-8, and intrinsic apoptosis protein-activated caspase-9 in the tumor section were all markedly increased by imipramine and erlotinib ( Figure 6B ). Moreover, NF-κB downstream protein expressions such as MMP9, MCL-1 and XIAP were all reduced by imipramine as displayed in Figure 6C . Additionally, the protein expression from extracted tumor also demonstrated the induction of cleaved caspase-3, 8, -9 and PARP-1 by imipramine ( Figures 6D, E ). In conclusion, these results revealed that imipramine not only suppressed the phosphorylation of EGFR/PKC-δ/NF-κB signaling and protein expression of NF-κB-mediated downstream proteins but also significantly induced the protein expression of apoptotic related proteins in CL1-5-F4/NF-κB-luc2 bearing mice tumor.