Dioscin elicits anti‐tumour immunity by inhibiting macrophage M2 polarization via JNK and STAT3 pathways in lung cancer

Abstract Tumour‐associated macrophage (TAM) is an important component in tumour microenvironment. Generally, TAM exhibits the function of M2‐like macrophage, which was closely related to angiogenesis and tumour progression. Dioscin, a natural steroidal saponin, has shown its powerful anti‐tumour activity recently. However, the mechanism of dioscin involved in immune regulation is still obscure. Here, we observed dioscin induced macrophage M2‐to‐M1 phenotype transition in vitro and inhibited IL‐10 secretion. Meanwhile, the phagocytosis of macrophages was enhanced. In subcutaneous lung tumour models, dioscin inhibited the augmentation of M2 macrophage populations. Furthermore, dioscin down‐regulated STAT3 and JNK signalling pathways in macrophages in vitro. In BMDMs, activating JNK and inhibiting STAT3 induce macrophages to M1 polarization while inhibiting JNK and activating STAT3 to M2 polarization. Additionally, condition mediums from dioscin‐pre‐treated macrophages inhibited the migration of 3LL cells and the tube‐formation capacity of HUVECs. What's more, dioscin‐mediated macrophage polarization inhibited the in vivo metastasis of 3LL cells. In conclusion, dioscin may act as a new anti‐tumour agent by inhibiting TAMs via JNK and STAT3 pathways in lung cancer.

non-small-lung cancers (NSCLCs). 14 Overexpressed negative immune regulatory molecules of M2-like TAMs, such as Arg-1, IL-10, programmed cell death 1 ligand (PD-L1) and cytotoxic T lymphocyte antigen 4 (CTLA-4), inhibited the effect of CD4 + and CD8 + T cells to induce immunosuppressive microenvironment for tumour development. [16][17][18][19][20] Many other factors also exist in the tumour microenvironment, such as platelet-derived growth factor (PDGF), vascular endothelial growth factor (VEGF), matrix metallopeptidase (MMP) and CCL. 2,9,15 And the regulator network of these factors leads to angiogenesis, proliferation of malignant cells, tumour invasion and metastasis. 9,15,21 What's more, in NSCLCs, the high ratio of M1/M2 TAMs and M1 TAMs is positively associated with patients' survival while M2 TAM accumulation with poor outcome. 12,14,18 Therefore, intervention of M2 polarization may become a promising new strategy for lung cancer treatment. 13 Dioscin, a natural steroidal saponin, is extracted from the roots of dioscorea plants, such as dioscorea zingiberensis and dioscorea nipponica. 22 During recent years, the anti-tumour effect of dioscin has been reported progressively. [23][24][25][26][27] In human lung cancer cells, dioscin could inhibit TGF-β1-mediated epithelial-mesenchymal transition, induce cell apoptosis and suppress tumour invasion. 27,28 Interestingly, some studies detect dioscin has the potential effect to reverse drug resistance. [29][30][31] However, there are few studies focused on the effects of dioscin in immune regulation. It has been confirmed dioscin could induce Raw264.7 cells to M1 polarization and then up-regulate connexin 43 expression to inhibit melanoma progression. 10 But whether the anti-tumour influence of dioscin is related to the effect on macrophage polarization and the detail mechanism has yet to be determined.
In the current study, we try to explore the impact of dioscin on phenotypes and functions of macrophages. We utilized in vitro cell culture systems (BMDMs and Raw264.7 cells) to elucidate dioscin-induced phenotype transition from M2 to M1 with the down-regulation of STAT3 and JNK. Then, we constructed a subcutaneous lung cancer model to confirm the inhibition of dioscin on macrophage M2 polarization in vivo. What's more, the phagocytosis of BMDMs was enhanced with dioscin treatment. With condition medium treated, we discovered dioscin could inhibit the migration of 3LL cells and the tube-formation capacity of HUVECs. And our lung metastases models in vivo indicated dioscin-mediated macrophage polarization inhibited the metastasis of 3LL cells. In conclusion, our results suggested dioscin elicits anti-tumour immunity by inhibiting macrophage M2 polarization through JNK and STAT3 pathways in lung cancer.

| Animal experiment
Male C57BL/6 mice (6-8 weeks) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. 2 × 10 5 suspended 3LL cells were injected into the right axilla of each mouse. After 4 days, the mice were randomized into 3 groups. Each group of mice was given with 0.5% sodium carboxymethylcellulose (Sangon Biotech, Shanghai, China) or dioscin (30 and 60 mg/kg/d) by gavage for 17 days. During the experiment, mice were sacrificed when the tumour diameter exceeded 2 cm. At the endpoint, blood, tumours and spleens were collected for further analysis.
Then, we established lung metastases mouse model: C57BL/6 mice were divided into 4 groups: control, dioscin treatment, clodronate liposome treatment and dioscin combined with clodronate liposome treatment (n = 4 per group). Mice were injected intravenously with 1 × 10 6 3LL cells in 0.1 mL PBS. Clodronate liposome or control liposomal (Liposoma, Amsterdam, Netherlands) was given to the mice by intraperitoneal injection at a dose of 100 μL/10 g daily for 3 days before tumour-cell injection, followed by repeated injections of 50 μL/10 g every fourth day. Dioscin was given at 60 mg/kg by gavage 24 hours before tumour-cell injection, and once a day thereafter for prolonged treatments.
We also established co-inoculated lung metastasis mouse model:

| Flow cytometry
BMDMs were pre-treated and then stained in PBS with cell surfacespecific antibodies for 30 mins at room temperature in the dark.

| Quantitative real-time PCR
Cells were extracted using RNA-Quick Purification Kit according to the manufacturer's instructions (Yishan Biotech, Shanghai, China).
The RNA concentration was measured using a NanoDrop 2000 (Thermo Scientific, Waltham, MA, USA) and then was converted into cDNA using a reverse transcription system (Takara, Shiga, Japan).

| Western blot analysis
After the different treatment, cells were lysed in RIPA lysis buffer

| Phagocytosis assay
Phagocytosis is determined by the amount of fluorescence-labelled latex beads (Sigma-Aldrich, St. Louis, MO, USA) internalized by BMDMs. BMDMs were seeded in 6-well plates and treated with dioscin for 48 hours. Then, BMDMs were incubated with fluorescent latex beads for 1-2 hours. After washing, BMDMs were analysed by microscopy (Olympus, Tokyo, Japan) or were detached by trypsin and centrifuged in PBS for flow cytometry.

| Inhibitors and activators
SP600125 (a JNK inhibitor), Stattic (a STAT3 inhibitor) and Anisomycin (a JNK activator) were purchased from Selleck Chemicals (Houston, TX, USA). Colivelin, a STAT3 activator, was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). BMDMs were treated with different inhibitors and activators alone or with dioscin, then were analysed by quantitative real-time PCR and Western blot.

| Preparation of conditioned medium
Raw264.7 cells and BMDMs were seeded in 6-well plates with dioscin for 48 hours, then replaced with fresh serum-free DMEM medium for another 24 hours and collected these media as conditioned medium. After centrifugation, supernatant was collected and stored at −20℃.

| Transwell assay
3LL cells were starved for 12 hours and then cultured in the upper chamber (24-well Transwell chambers, 8-μm pore size, Corning, NY, USA) with serum-free medium, while high-serum medium (5% FBS) with condition medium was added to the lower chamber. After 24 hours of incubation at 37℃, cells were fixed with methanol and stained with 0.1% crystal violet for 30 minutes. Images were taken using microscope (Olympus, Tokyo, Japan).

| Tube formation assay
96-well plates were pre-treated with 50 μL Matrigel (BD Bioscience, SFO, USA) per well and were placed at 37℃ for 1 hour. HUVECs were resuspended in serum-free DMEM medium with condition medium and then cultured in 96-well plate with Matrigel at 37℃ for 4-6 hours. Images were taken using microscope (Olympus, Tokyo, Japan).

| Statistical analysis
All data were presented as the means ± SD. The statistical significance was assessed by Student's t test using Prism 6.04 software (GraphPad Software Inc, San Diego, CA, USA). *P < 0.05, **P < 0.01, ***P < 0.001 are determined as significance.

| Determination of BMDMs and detection of the non-cytotoxic concentration of dioscin in Raw264.7 cells and BMDMs
On the 7th day of culture, image of BMDMs was obtained under a light microscope ( Figure 1A). And the per cent of F4/80 + cells by flow cytometry was over 90% ( Figure 1B). We then evaluated the cytotoxicity of dioscin ranging from 0.1 to 12.8 μmol/L in Raw264.7 cells and BMDMs. Compared with the control treatment, dioscin began to inhibit the viability of RAW264.7 cells and BMDMs from the concentration of 3.2 μmol/L at 48 hours ( Figure 1C,D). The half maximal inhibitory concentration (IC50) of RAW264.7 cells and BMDMs was 3.809 μmol/L and 4.319 μmol/L, respectively.
The M1 and M2 phenotypes were further confirmed by gene expression analysis in Raw264.7 cells and BMDMs. We measured the expression of M1 genes (IL-6 and NOS2) and M2 genes (Arg-1 and CD206) by qPCR. As shown in Figure 2C, dioscin-treated groups showed significant up-regulation of M1 and down-regulation of M2 relevant gene expression in Raw264.7 cells. BMDMs treated with 1 μmol/L dioscin also had higher IL-6 and NOS2 mRNA levels but lower Arg1 and CD206 mRNA levels ( Figure 2D).

| Dioscin decreased the IL-10 secretion of macrophages in vitro
To investigate the effect of dioscin on macrophage secretions, we used flow cytometry to analyse the intracellular cytokines. Compared with the control group, we found 1 μmol/L dioscin could significantly inhibit the level of IL-10 secreted by BMDMs (0.23 ± 0.03% vs 1.65 ± 0.86%, P < 0.05), while could not influence IL-12 secretion evidently. Interestingly, the ratio of IL-12/IL-10 was highly increased from 4.09 ± 2.08 in the control group to 19.96 ± 6.44 in 1 μmol/L dioscin group (P < 0.05) ( Figure 4A,B).

| Dioscin enhanced the phagocytosis of BMDMs
To study the effect of dioscin on phagocytosis, we used microscopy and flow cytometry to measure the amount of fluorescence-labelled latex beads internalized by BMDMs, as described in Materials and Methods. As shown in Figure 4, 1 μmol/L dioscin treatment obviously raised the mean number of latex beads internalized by BMDMs (6.9 ± 1.6 vs 2.6 ± 1.1, P < 0.05). Compared with the control group, 1 μmol/L dioscin apparently increased the population of BMDMs internalizing latex beads (14.53 ± 0.78% vs 10.10 ± 1.44%, P < 0.001) ( Figure 4E,F).

F I G U R E 1 Determination of BMDMs and detection of the non-cytotoxic concentration of dioscin in macrophages. A-B, Image of
BMDMs was obtained under a light microscope. Scale bars: 100 μm. The percentage of F4/80 + cells as obtained by flow cytometry. C-D, Detection of the non-cytotoxic dose of dioscin. BMDMs and Raw264.7 cells were incubated with dioscin at range from 0 to 12.8 μmol/L for 24 h and 48 h. Cell viability was determined by CCK-8 assay and was performed in triplicate, thrice independently. The data were presented as mean ± SD. *P < 0.05, **P < 0.01 and ***P < 0.001 vs control F I G U R E 2 Dioscin induced macrophage phenotype transition from M2 to M1 in vitro. BMDMs and RAW264.7 cells were treated with dioscin for 48 h. A-B, The numbers of F4/80 + CD86 + (M1), F4/80 + NOS2 + (M1), F4/80 + CD206 + (M2) and F4/80 + CD209 + (M2) cells among the total F4/80 + cells were quantified. C-D, Relative mRNA expressions of the M1 genes IL-6, NOS2 and the M2 genes Arg-1, CD206 were measured by qPCR. The data were representatives of three independent experiments and presented as the mean ± SD. *P < 0.05, **P < 0.01 and ***P < 0.001 vs control

| Dioscin down-regulated STAT3 and JNK signalling pathways in Raw264.7 cells and BMDMs
We further evaluated the effect of dioscin on activation of STAT3 and the MAPK pathway in RAW264.7 cells and BMDMs. As shown in Figure 5A, compared with the control group, dioscin treatment (0.1 μmol/L and 1 μmol/L) significantly inhibited the phosphorylation of STAT3 and JNK in a low dose-dependent manner.
However, dioscin had no obvious effect on the p38 and ERK signalling pathway. In order to discover the role of STAT3 and JNK signalling pathways during dioscin-mediated macrophage polarization, BMDMs were treated with different inhibitors and activators ( Figure 5B). We found activating JNK and inhibiting STAT3 induced higher IL-6 mRNA levels but lower Arg1 mRNA levels in BMDMs, while inhibiting JNK and activating STAT3 induced opposite results ( Figure 5C,D). What's more, the effect of dioscin in macrophage polarization could be enhanced by Anisomycin and weakened by Colivelin ( Figure 5E,F).

| Dioscin-mediated macrophage polarization suppressed the migration of 3LL cells
Previous results exhibit dioscin could regulate macrophage polarization. However, whether dioscin could influence the metastasis of tumour by regulating macrophage polarization still unknown. We then collected condition medium from dioscin pre-treated BMDMs and Raw264.7 cells to study the impact on 3LL cells' migration. And we found dioscin (0.1 μmol/L and 1 μmol/L) pre-treated condition medium suppressed the migration of 3LL cells in BMDMs (188 ± 3 and 88 ± 15 vs 316 ± 51, respectively, P < 0.05 and P < 0.01) ( Figure 6A).

| Dioscin-mediated macrophage polarization inhibited the tube-formation capacity of HUVECs
Similarly, we used the collected conditioned medium to treat the HUVECs. As shown in Figure 6C, 1 μmol/L dioscin pre-treated condition medium from BMDMs could inhibited the tube formation.
The number and total size of tubes relative to the control group was 0.35 ± 0.12 and 0.14 ± 0.05 (both P < 0.001). And in Raw264.7 cells, condition medium from 2 μmol/L dioscin treatment could apparently decreased the number and total size of tubes formed by HUVEC (0.40 ± 0.02 and 0.34 ± 0.11, relative to the control group, both P < 0.001) ( Figure 6D). So, we believed dioscin inhibited the tube-formation capacity of HUVECs by regulating macrophage polarization.

| Dioscin inhibited the in vivo metastasis of 3LL cells through mediating macrophage polarization
To verify the effect of dioscin on anti-tumour metastasis, we constructed an intravenous injection lung metastasis mouse model. We also used clodronate liposomes to chemically eliminate macrophages to further confirm the anti-tumour metastasis effect was related to macrophages. The total metastases per lung of dioscin-treatment and clodronate liposome-treatment were reduced dramatically (both P < 0.01) ( Figure 7A). What's more, there were no difference between clodronate liposome combined with dioscin treatment and clodronate liposome treatment. This means after eliminating macrophages, the effect of dioscin on anti-tumour metastasis was not obvious. Another co-inoculated lung metastasis mouse model indicated dioscin-treated Raw264.7 cells could inhibit the lung metastasis of 3LL cells (P < 0.05) ( Figure 7B). Therefore, we demonstrated dioscin inhibited the lung metastasis of 3LL cells through regulating macrophage polarization in vivo.

| D ISSCUS I ON
In the recent years, dioscin has shown its powerful anti-tumour activity. Many studies illustrate dioscin could induce autophagy, increase mitochondrial injury, enhance cell apoptosis, promote ROS accumulation and Ca 2+ release to suppresses malignant activities of cancer cells. [24][25][26]32 And in human lung cancer cells, dioscin could inhibit proliferation and promote apoptosis by inducing DNA damage, cell cycle arrest and activating mitochondrial signalling pathways. 27 markers in BMDMs and Raw264.7 cells. CD206 and CD209 are two classical markers to identify the M2 macrophages with confirmed specificity and reliability in lung cancers. 12,[33][34][35][36] Arg-1 and NOS2 are two opposite markers to catabolize L-arginine into different ingredients with different effects. 15 It is worth noting that Arg-1 may not be suitable for marking human M2 macrophages with not regulated by M2-inducing cytokines in human. 33 Besides our results even showed the phagocytosis of BMDMs was enhanced by dioscin. What's more, we constructed a subcutaneous lung tumour model to further verify the effects of dioscin in vivo. And we found dioscin inhibited the F I G U R E 3 Dioscin restrained the augmentation of M2 macrophage populations in vivo. PBMCs, splenocytes and tumour tissue cells were collected from 3LL xenograft model. F4/80 + CD206 + cells and F4/80 + CD209 + cells were identified as M2 macrophages. The percentages of M2 macrophages in PBMCs (A-B), splenocytes (C-D), tumour tissue (E-F) were shown. The data were representatives of three independent experiments and presented as the mean ± SD. *P < 0.05, **P < 0.01 and ***P < 0.001 vs control which finally caused TAMs M2 polarization. 43 By activating STAT3, TAMs and small-cell lung cancers could jointly promote tumour progression. 44 What's more, the activation of STAT3 up-regulated angiogenic, metastatic and pro-proliferation relevant gene expression. 43 MAPK pathway, including ERK, p38 and JNK, is disputable in macrophage polarization. Previous studies confirm the activation of ERK and p38 induces M2 macrophage polarization while JNK for M1 polarization. 12,41,45,46 Oppositely, others believe activated-JNK leads to M2 polarization. 16 Moreover, IL-4, a M2-inducing mediator, could activate JNK and promote proliferation of cancer cells simultaneously. 47 Our study showed dioscin, may as a potential IL-10 inhibitor, down-regulated the expression of activated STAT3 based on phenotype transform. And STAT3 inhibitor Stattic induced macrophages to M1 polarization, and STAT3 activator Colivelin induced macrophages to M2 polarization. What's more, the effect of dioscin F I G U R E 6 Dioscin-mediated macrophage polarization suppressed the migration of 3LL cells and the tube-formation capacity of HUVECs. Condition mediums were collected to treat 3LL cells and HUVECs. A-B, Migration of 3LL cells was assessed by Transwell assays. Cells were counted by ImageJ. Scale bars: 50 μm. C-D, The numbers and total size of tubes formatted by HUVEC were summed up by ImageJ. Scale bars: 100 μm. The data were presented as the mean ± SD of triplicate independent samples. *P < 0.05, **P < 0.01 and ***P < 0.001 represent a significant difference in macrophage polarization could be reversed by Colivelin. In accord with our findings, other study has shown anti-IL-10 receptor antibodies and STAT3 inhibitors induced macrophage phenotype from M2 to M1 switch. 2 We also found the expression of activated JNK was down-regulated by dioscin with no significant change in p38 and ERK. A recent study demonstrated Arg-1 expression varies as the JNK expression changes simultaneously in IL-4-induced M2 macrophage polarization. 16 However, our results illustrate JNK activator F I G U R E 7 Dioscin-mediated macrophage polarization suppressed the metastasis of 3LL cells in vivo. In lung metastases mouse model (A), C57BL/6 mice (n = 4) were injected intravenously with 3LL cells and treated with dioscin (60 mg/kg), clodronate liposome or both of them. In co-inoculated lung metastasis mouse model (B), C57BL/6 mice (n = 4) were injected intravenously with 3LL cells alone or in combination with RAW264.7 cells not treated or exposed to 2 μmol/L dioscin for 48 h. After 21 d, the mice were sacrificed and their lungs were removed and histologically analysed for metastases. The number of metastases per lung and the number of small (diameter: <200 μm), medium-sized (diameter: 200-400 μm) and large (diameter: >400 μm) lung metastases were determined. C-E, Representative images of the lungs. Scale bars: 200 μm. *P < 0.05, **P < 0.01 and ***P < 0.001 represent a significant difference Anisomycin down-regulated relative expression of Arg-1 mRNA and enhanced the effect of dioscin.
Tumour metastasis is a complex invasion-metastasis cascade requiring angiogenesis, damage of the basement membrane and remodelling of the extracellular matrix for malignant cell migration, invasion and extravasation. 48,49 It is known that M2-like TAMs induce immunosuppressive tumour microenvironment to promote dissemination of malignant cells in the early stage of tumour metastasis. 9,19,48 A notably study illustrated the augmentation of M2-like TAMs in lung adenocarcinoma was associated with tumour metastasis. 12 This study confirmed condition medium from dioscin pre-treated macrophages (BMDMs and Raw264.7 cells) inhibited the migration of 3LL