Fruticuline A, a chemically-defined diterpene, exerts antineoplastic effects in vitro and in vivo by multiple mechanisms

Natural products have been recognized as important bioactive compounds on the basis of their wide biological properties. Here we investigated the antitumor effect and molecular mechanisms of the diterpene Fruticuline A (fruti) from Salvia lachnostachys, in human cancer cell lineages and Solid Ehrlich Carcinoma in mice. Fruti reduced MCF-7 and HepG2 proliferation by the reduction of Cyclin D1 levels and decreased NF-κB gene levels in both cell types. Furthermore, fruti also induced apoptosis in HepG2 cells, reduced Bcl-2 gene expression and induced necroptosis by increasing Ripk in MCF-7 cells. In mice, fruti prevented tumor development and reduced Cyclin D1, Bcl-2 and Rela gene levels, and reduced the p-NF-κB/NF-κB ratio in tumor tissue. Furthermore, fruti induced necrosis and apoptosis, increased N-acetyl-β-D-glucosaminidase and TNF-α levels and reduced IL-10 and Vegf levels in tumor tissue. Collectively, fruti exerts antitumor effects through the inhibition of the NF-κB pathway, reducing Cyclin D1 and Bcl-2 levels. In vitro the apoptosis and necroptosis pathways are involved in the cellular death, whereas in vivo, cells undergo necrosis by increased tumor inflammation and reduction of angiogenesis. Thus, fruticuline A acts in tumor cells by multiple mechanisms and represents a promising molecule for drug development in cancer treatment.


Results
Isolation and identification of fruti. Pure fruticuline A was obtained by preparative HPLC and analyzed by NMR ( 1 H, HSQC, HMBC). The NMR data were identical with those previously reported 14,15 . The 1 H NMR ( Supplementary Fig. S1) showed the typical signals of fruticuline A: two doublets at δ H 1.26 (six hydrogens) and δ H 3.14 (two hydrogens); two singlets at δ H 2.44 and 3.79 (three hydrogens each one); an septet at δ H 3.37 (one hydrogen); a multiplet at δ H 6.86-6.98 (three hydrogens); two singlets at δ H 7.66 (hydroxyl group) and 8.16 (one hydrogen). Besides the solvent signal (δ H 7.27), no other signals were observed, confirming the absence of impurities, as aliphatic compounds or other aromatic diterpene, such as fruticuline B. The yield of fruticuline A from EES was around 3%.

Fruti has cytotoxicity on human cancer cells but does not induce oxidative stress in MCF-7 cells.
Fruti at the concentration of 62.5, 125 and 250 µM inhibited MCF-7 viability in 33, 44 and 69% respectively, when compared to the vehicle group. In HepG2 cells the inhibition occurred with Fruti at 125 and 250 µM (17 and 50%, respectively), whereas in fibroblasts, only the concentration of 250 µM reduced viability by 46%, compared to vehicle group ( Fig. 2A). The IC 50 (Fig. 3A,B). In addition, fruti at 25 and 50 µM reduced BCL-2 expression in MCF-7 cells by 36 and 66%, respectively (Fig. 3A). However, at 50 µM there was an increase in VEGFA gene in MCF-7 cells (Fig. 3A). No differences were found in fibroblasts (Fig. 3C). Furthermore, the expression of TNF-α was not detected in these cell lines.

Treatment with fruti inhibits Ehrlich tumor development, induces inflammation, apoptosis
and necrosis in tumor tissue. In order to test the effect of purified fruti on tumor development in vivo, mice were inoculated with Ehrlich ascites cells and either treated with fruti or vehicle. Treatment with fruti (3 mg kg −1 ) significantly inhibited tumor growth from day 10 until the end of experiment, when compared to the vehicle group (Fig. 6A). MTX, used as positive control, also inhibited tumor growth until day 21. In addition, the tumor weight on day 21 was significantly reduced (49%) by fruti treatment (Fig. 6B).
Tumor tissue of animals treated with vehicle displayed viable Ehrlich cells, small areas of coagulative necrosis and discrete inflammation observed by polymorfonuclear cells infiltration and discrete apoptosis (Fig. 6C). Tumor tissue of animals treated with fruti had more coagulative necrosis area, less viable Ehrlich cells, higher inflammation and moderate apoptosis when compared to the vehicle group (Fig. 6D). Furthermore, MTX treatment also induced discrete inflammation and coagulative necrosis, but no apoptosis in tumor tissue (Fig. 6E). The tumor histology quantification is summarized in Fig. 6F.  Fruti reduces NF-κB levels in tumor tissue. After 21 days of treatment, despite the total protein level of phosphorylated NF-κB and total NF-κB did not differ among groups, fruti reduced in 49% the p-NF-κB/NF-κB ratio in tumor tissue (Fig. 8).

Fruti induces oxidative stress and inflammation in tumor tissue. Fruti treatment increased ROS,
N-acetyl-β-D-glucosaminidase (NAG) and TNF-α levels in tumor tissue by 15, 29 and 79%, respectively, and reduced IL-10 by 68%, compared to vehicle group. No differences were observed in the other inflammatory parameters (Table 1).

Fruti does not induce hematological toxicity and increases monocytes.
As depicted on Supplementary Table S1, fruti treatment induced a slight decrease in ALT and increased in monocyte numbers, compared to the vehicle group. MTX treatment reduced total protein, globulin, lymphocytes and hemoglobin levels and increased monocytes compared to vehicle and naïve group. In addition, no toxicity was observed in liver tissue upon Fruti treatment ( Supplementary Fig. S8).

Discussion
Fruticuline A is one of the major constituents of the ethanolic extract of Salvia lachnostachys (EES) leaves, which was demonstrated to have antitumor effect against Solid Ehrlich Carcinoma (SEC) in mice 11 . Fruti is a diterpene quinone and has some biological properties already described, such as antinociceptive, anti-inflammatory, antidepressive and antimicrobial 12,13,16 . Cytotoxicity of fruti against a range of human cancer cell lines has also been demonstrated, including U251 (glioma, central nervous system), UACC-62 (melanoma), MCF-7 (breast), NCI-ADR/RES (ovarian-resistant), NCI-H460 (lung, non-small-cell), PC-3 (prostate), OVCAR-3 (ovarian) and HT-29 www.nature.com/scientificreports/ (colon) 10 . However, none of these studies investigated the antitumor effect of fruti and its action mechanism. Thus, the present study was the first one to investigate the antitumor effect of pure fruti in vivo and to elucidate the mechanism of action in different tumor cell lineages summarized in Fig. 9.
Fruti was capable of inhibiting proliferation of tumor cells (MCF-7 and HepG2) indicating its antiproliferative effect. As fibroblasts are non-tumor cells that surround tumor tissue, we then investigate whether fruti could inhibit the proliferation of this cell type. A reduction of proliferation was only observed at high concentration of fruti (250 µM), but no alterations in expression of genes related to proliferation, apoptosis and angiogenesis in these cells. Thus, we hypothesized that instead of damage to normal cells, fruti not only inhibits the proliferation but also induces cell death only in tumor cells. For this reason, we focused to investigate its antitumor effect and its mechanism in tumor cells.
Fruti inhibited CCND1 (CYCLIN D1) and BCL-2 gene expression in MCF-7 cells. Cyclin D1 is a protein that regulates proliferation and is found in cytoplasm and nuclei 17 , whereas Bcl-2 is a regulatory anti-apoptotic proteinthat determines the response of cancer cells to chemotherapeutics 18 . Both genes can be transcriptionally regulated by STAT3 and/or NF-κB, which play an important role in the development of carcinogenesis 3,19,20 . Despite the fact that fruti did not change STAT3 levels, it decreased NF-κB1 gene expression. NF-κB comprises a family of transcription factors, including RELA, RELB, NF-κB1, NF-κB2, and IκBα, which are inducible by a variety of stimuli. Protein products of these genes lead to activation of p65, p50 and p52. The nuclear translocation of NF-κB complexes p65/p50 and p52/RELB is triggered by degradation of IκBs, particularly IκBα, leading to the nuclear translocation of several NF-κB complexes, predominantly the p50/RELA dimer 21 . Consequently, www.nature.com/scientificreports/ these complexes regulate the transcription of genes in the nuclei thereby mediating several cellular functions such as proliferation (e.g. Cyclin D1, Cyclin E, IL-6), tumor promotion (e.g. VCAM, ICAM, iNOS, COX 2), apoptosis (e.g. Bcl-2, Bcl-xL, Survivin) and inflammation (e.g. TNF-α, COX2, iNOS) 3,22 . Thus, our results suggest the involvement of the NF-κB pathway on the regulation of Cyclin D1 and Bcl-2 gene expression induced by fruti. However, a global inhibition of NF-κB is not possible, as it would induce genetic instability 6 . In fact, we observed a partial decrease in NF-κB, in the order of 63 and 65% in MCF-7 and HepG2 cells, respectively, compared to the control group. In addition, it was already described that some diterpenes such as excisanin A and kamebakaurin from Isodon japonicus, inhibit NF-κB in inflammatory processes 23,24 . Therefore, it is possible that the antitumor effect of fruti is dependent of NF-κB inhibition. The mechanism of cellular death in neoplastic cells can be the result of programmed cellular death (apoptosis), necrosis or necroptosis. These pathways are induced by different signals, such as DNA damage, oxidative stress, hypoxia and/or ligands of death receptor family (i.e. TNF-α) 25 . As fruti is a quinone compound, which are known inducers of GSH adduct formation and oxidative stress 16,26 , we measured GSH content and ROS production on MCF-7 cells. However, curiously, neither short nor longer incubation times depleted GSH, showing no involvement of cellular death by oxidative stress. On the other hand, the c-Jun N-terminal kinase (JNK) pathway is also involved in the regulation of cell death. JNK forms an important subgroup of the mitogen-activated protein kinases (MAPK) superfamily and are mainly activated by oxidative stress and/or TNF-α, being a key promoter for cellular suicide by apoptosis or necrosis 27 . The programmed cell death, on the other hand, has a variety of pathways and proteins that regulate or trigger apoptosis, such as the anti-(e.g. Bcl-2, Bcl-xL) or pro-(e.g. Bax, Bid) apoptotic proteins and caspases effectors (e.g. caspase 3 and 7) 28 . While TNF-α was not detected in the cells and no difference was observed in pJNK levels, apoptosis induced by fruti, was probably due to the observed reduction on Bcl-2 levels. In addition to apoptosis, we also observed that fruti enhanced RIPK1 gene expression at higher concentrations. RIPK1 is the target for necroptosis pathway, a form of regulated necrotic cell death. Nec-1 has been used as a tool to investigate necroptosis pathway due to its ability in inhibit RIPK1 activity 29 . In fact, the antiproliferative effect of fruti in MCF-7 cells was reversed in the presence of Nec-1, confirming the    www.nature.com/scientificreports/ reduction of angiogenesis and nutrition to tumor tissue, which induces cellular death by starvation resulting in coagulative necrosis 31 . Therefore, beyond the tumor cell death through apoptosis induced by fruti in SEC, it also has anti-angiogenic properties in solid tumors inducing to cell death by coagulative necrosis. As inflammation is intrinsically correlated with cancer development 32 , we investigated some cytokines in tumor tissue. Interestingly, there was an increase in tumor TNF-α and decrease in IL-10 levels in animals treated with fruti, showing a pro-inflammatory effect of fruti. The increase in TNF-α levels can be correlated with an increase in blood monocytes, which could be a source for TNF-α production 33 . In addition, Rela expression in tumor tissue was partially inhibited by fruti, despite no changes were observed in IkBα expression, suggesting that the inflammation induced by fruti is not dependent of NF-κB pathway. This indicates a compensatory mechanism of fruti treatment in the steady state of IkBα/NF-κB activation, disturbing only the expression of the active form Rela 34 . In addition to the enhanced inflammatory environment in tumor tissue, we observed a slight increase in ROS levels, which can also be triggered by inflammatory mediators. Indeed, the tumor histology showed more inflammatory cells, coagulative necrotic cells and apoptosis, and less Ehrlich viable cells, in tumor of fruti-treated mice compared to the control group. Thus, the increase in tumor inflammation can also induce tumor cell death or can be a result of the necrosis 35,36 induced by the fruti.
The inflammatory tumor microenvironment can be the explanation for some different results observed in in vitro and in vivo experiments with fruti, for instance the VEGF. Due to the pro-inflammatory effect of fruti we hypothesized that in vivo the tumor microenvironment changes the response to antitumor agents, compared to in vitro observations. The long-lasting treatment in vivo can also change the mechanism of cellular death. Furthermore, the difference in tumor cell lineages studied in vitro (MCF-7 and HepG2) and in vivo (Ehrlich) should be taken into account in the interpretation of the results.
Despite the cytotoxic effects of fruti in tumor cells, it did not induce systemic toxicity in mice, observed by the hematological and biochemical parameters. Moreover, no difference among treatments was observed in body and organ weight (data not shown) and in liver histology. Previously, our group showed that the source of fruti, the S. lachnostachys ethanolic extract, had no toxicity, with a LD 50 higher than 2000 mg kg −111 . On the other hand, MTX, the chemotherapy compound used as positive control, has oral LD 50 in mice of 146 mg kg −137 . MTX treatment in mice bearing Ehrlich tumor produced systemic and hematologic adverse effects. As fruti also had antitumor effect, this can make fruti, or an improved derivative of fruti, a good candidate to treat solid cancers with less toxic effects.
In conclusion, fruticuline A exerts antitumor effects through inhibition of the NF-κB pathway, reducing Cyclin D1 and Bcl-2 levels. The effects depend on both the used cell lineages and the experimental conditions. In MCF-7 and HepG2 cells the apoptosis and necroptosis pathways are involved in cell death, whereas in the www.nature.com/scientificreports/ Ehrlich tumor in mice, due to the long-term treatment and tumor microenvironment, cells undergo necrosis by the increase in tumor inflammation and reduction of angiogenesis. Thus, fruticuline A acts on tumor cells by a combination of two mechanisms and represents a promising molecule for being used directly or in a modified formulation in cancer therapy, especially against solid mammary and hepatic tumors.

Isolation of fruticuline A by HPLC and NMR.
A detailed procedure of isolation and identification of fruticuline A from Salvia lachnostachys (SisGen AA49C69) was previously reported 10 . Briefly, dried and powdered leaves of S. lachnostachys (415.3 g) were extracted with hexane and ethanol, at room temperature. The ethanol extract (EES, 45.0 g) was submitted to silica gel vacuum liquid chromatography and eluted with hexane, hexane:dichloromethane 1:1, pure dichloromethane, acetone, and methanol. The fraction eluted with hexane (1.8 g) was subjected to silica gel column chromatography, eluted with mixtures of hexane:dichloromethane in increasing polarity, to give 14 subfractions. Subfraction 6 revealed to contain mainly impure fruticuline A (458.4 mg). An aliquot (80 mg) was purified by HLPC (Rt = 8.21 min, methanol:acetonitrile 50:50, isocratic, semipreparative Nucleosil 100-5 C18 column, 250 × 10 mm), yelding pure fruticuline A (58.8 mg). Fruticuline A was identified by 1D and 2D Nuclear Magnetic Resonance [Bruker Avance 400 spectrometer, using deutered chloroform (CDCl 3 ) as solvent], and compared with previously reported data 14,15 . In vitro experiments. Culture cells. Human MCF-7, HepG2 and Fibroblasts cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 10% fetal bovine serum, 4 mmol L −1 of L-glutamine and a mixture of antibiotics (5 mg mL −1 of streptomycin and 5 mg mL −1 penicillin). Cells were incubated at 37 °C in humidified atmosphere containing 5% CO 2 . After cells reached ~ 80% of confluence, cells were harvested and seeded in a 96 well plate or 12 well plate, according to the experimental protocol.
WST-1 viability assay. The WST-1 viability assay is a colorimetric assay for the quantification of cell viability, based on the cleavage of the tetrazolium salt WST-1 by mitochondrial dehydrogenase in viable cells producing a colorimetric product, formazan (CELLPRO-Roche, Mannheim, Germany). Briefly, 5 × 10 3 of MCF-7 cells/ well were added in 96 wells plate. Cells were incubated with vehicle (0) or fruti (31.2, 62.5, 125 and 250 µM) dissolved in 0.1% DMSO/DMEM for 24 h. Then, the medium was replaced by new medium to which 10 µL of WST-1was added and the cells were incubated for 2 h at 37 °C in humidified atmosphere 5% CO 2 . Additionally, fibroblasts were also incubated with fruti to evaluate the cytotoxicity of fruti in non-transformed cells. Viability was measured by spectrophotometer at 450 and 690 nm, after 1 min shaking the plate to equilibrate the intracellular formazan. The results were expressed as cell viability (%). In addition, to evaluate the necroptosis pathway, the same procedure described above was performed, including the incubation of cells with 50 µM necrostatin-1 (Nec-1) (Sigma-Aldrich, St. Louis, Missouri, EUA), an inhibitor of receptor-associated kinase 1 (RIPK1).
Oxidative stress parameter assays. To perform reduced glutathione (GSH) measurement, MCF-7 cells were seeded in 96 wells plate until reaching 80% of confluence. Vehicle (0) or fruti (12.5, 25, 50 and 100 µM) was incubated for 0.5, 2, 16 and 24 h to better evaluate if alterations in quinone concentration over a period of time would deplete this peptide. After the incubation, cells were scratched in the presence of iced cold 10% perchloric acid and centrifuged for 10 min at 14,000×g 4 °C. To increase the pH to 8, 3.0 M K 2 PO 4 was added in the supernatant and samples were immediately frozen in liquid nitrogen. After 30 min samples were thawn and incubated with work solution (DTNB and NADPH) diluted in GSH buffer (0.1 M Na 2 HPO 4 .2H 2 O, 0.1 M NaH 2 PO 4 .2H 2 O and 0.05 M Na 2 EDTA.2H 2 O) in a 96 wells plate. The reaction was started by addition of glutathione reductase (GR) and the absorbance was monitored at 412 nm at 37 °C on a Clariostar (BMG LabTech, Cary, USA) analyzer. The results were interpolated with GSH standards and were expressed in nmol GSH/well 38 .
Total RNA isolation and RT-qPCR. Human MCF-7, HepG2 and Fibroblast cells were seeded in 12 wells plate. After reaching confluence, cells were incubated with vehicle (0) or fruti at 10, 25, 50, 75 or 100 µM for 24 h (concentrations previously chosen by WST1 assay screening). Total RNA was extracted from cultured cells using Trizol reagent (Sigma-Aldrich, St. Louis, Missouri, USA) and measured by Nanodrop 100 (Thermo Scientific, Waltham, Massachusetts, USA). Complementary DNA was synthesized from 2 µg total RNA with an oligo-dT, random hexamers primers and deoxythymidine oligomer, Ribolock RNAse inhibitor and RevertAid reverse transcriptase. Real-time polymerase chain reaction was performed in a Lightcycler apparatus (Roche, Mannheim, Germany) using the sensiFAST SYBR No-ROX kit (Bioline, London, UK). Initial fluorescent values were calculated by LinRegPCR3 (version 2013.0; Academic Medical Center, Amsterdam, The Netherlands). Primer sequences were described in Supplementary Table S2. The most stable reference genes were calculated using geNorm 40 . Transcript levels were normalized to housekeeping genes ratio GAPDH and Cyclophilin (CyP) for MCF-7 and Fibroblasts cells, and the genes ratio HPRT and Actin for HepG2 cells. www.nature.com/scientificreports/ Western blot. MCF-7 cells were seeded in a 12 wells plate until reaching confluence. Then the cells were incubated with vehicle (0) or fruti at 10, 25 and 50 µM for 24 h (concentrations previously chosen by WST1 assay screening), harvested in iced cold PBS and homogenized in RIPA buffer freshly supplemented with PhosSTOP phosphatase inhibitor (Roche, Mannheim, Germany) and 5 mM EDTA-free protease inhibitor cocktail (Roche, Mannheim, Germany). The samples were centrifuged for 10 min at 4 °C and the supernatant was used for protein measurement by the bicinchonic acid assay. Forty micrograms of protein were electrophoresed on 8-10% sodium dodecyl sulfate-polyacrylamide gel and transferred to a polyvinylidine difluoride (PVDF) membrane by semidry blotting. The membrane was blocked overnight in 5% nonfat milk/TBST followed of overnight incubation with primary antibody in a cold room at 4 °C. The PVDF membrane was washed 3 times with TBST and incubated with secondary antibody with horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse immunoglobulin G antibody (BioRad, California, USA) for 1 h at room temperature. All antibodies were diluted in 1% nonfat milk/TBST. The lists of antibodies and dilutions used were described in Supplementary Table S3 42 . Efforts were made to avoid animal suffering, and environmental enrichment was used in the cages during all of the experimental protocols.
Ehrlich Carcinoma manipulation and groups of treatment. Ascites Ehrlich cells (2 × 10 6 cells/animal) were maintained by weekly intraperitoneal passage. After cells reach ≥ 98% of viability they were inoculated subcutaneously (2 × 10 6 cells/animal) in the right pelvic member of mice for SEC development. Animals were orally and daily treated with vehicle (0.1% Tween 20 in distilled water, 10 mg mL −1 ) or fruti 3 mg kg −1 for 21 days. This dose of fruti was based on the yield of the ethanolic extract effect, previously prepared by our group 11 . Additionally, a separate positive control group was treated with methotrexate (MTX) 2.5 mg kg −1 every 3 days by intraperitoneal route during 21 days. The tumor volume was measured from day 7 until day 21 and was calculated according to Mishra et al. 43 as V cm 3 = L × W 2 × 0.52 , where L is the largest tumor diameter and W is the smallest tumor diameter (in centimeters). On day 22, the mice were fasted for 16 h and anesthetized with ketamine (90 mg kg −1 ) and xylazine (10 mg kg −1 ) (Vetnil, São Paulo, Brazil) by intraperitoneal route, and blood, tumor and organs were collected for further analysis. www.nature.com/scientificreports/ titative densitometric analysis was then performed using Scion Image Beta 4.03 software (Scion Corporation, USA) 44 . The band intensities were normalized to GAPDH (1:5000; sc-32233, Santa Cruz Biotechnologies, Dallas, TX, USA).
Oxidative stress and inflammatory parameters. To measure oxidative stress parameters, tumor tissue was homogenized in phosphate buffer pH 6.5 (1:10) and centrifuged at 10,000×g at 4 °C for 20 min. The homogenate was used to measure reduced glutathione (GSH) 46 and the supernatant was used to measure reactive oxygen species (ROS) 39,47 . For inflammatory parameters, tumor tissues were homogenized in PBS and centrifuged at 10,000×g at 4 °C for 20 min. The supernatant was used to measure NO according to Green et al. 48 and cytokines (TNF-α, IL-6, IL-10, IL-4) according to the manufacturer's instructions (Kits BD Bioscience, Califórnia, USA). The pellet was ressuspended in 0.1% Saline-Triton X and centrifuged at 11,000×g at 4 °C for 20 min. The supernatant was used to measure myeloperoxidase (MPO) 49 and N-acetyl-β-D-glucosaminidase (NAG) levels 50 .
Plasmatic and hematologic parameters. Hematological parameters were analyzed through complete hemograms using a BC2800-Vet (BioBrasil, São Paulo, Brazil) automated device. Plasma samples were obtained after blood centrifugation at 1344×g for 5 min. The samples were used to determine creatinine, glucose, total protein, albumin, globulin, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, according to commercial kits instructions (Kovalent, São Gonçalo, Brazil). All parameters were analyzed by an automated system (Mindray BS-200, Shenzhen, China).
Data analysis. Data were expressed as means ± standard error of mean (S.E.M.). Results were subjected to t-test or one-or two-way analysis of variance (ANOVA), followed by post hoc Newman Keuls test, when applicable. The results were analyzed using GraphPad Prism (v. 5.0, San Diego, CA, USA). The level of significance was set at 95% (p < 0.05).

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.