Conophylline Inhibits Hepatocellular Carcinoma by Inhibiting Activated Cancer-associated Fibroblasts Through Suppression of G Protein–coupled Receptor 68

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and is challenging to treat (1). Cancer-associated fibroblasts (CAFs) are reported to promote the malignancy of HCC cells (2, 3). Treatment of HCC has advanced with the availability of the multikinase inhibitor, sorafenib; however, CAFs increase the resistance of HCC cells to this drug via secretion of cytokines such as IL6 (4–7). Many cytokine regulation mechanisms are reported for cancer and hematopoietic cells (8), and IL6 is reported to be regulated by G protein–coupled receptor 68 (GPR68; ref. 9).

Conophylline (CnP) is a vinca alkaloid obtained from the leaves of Ervatamia microphylla (Supplementary Fig. S1A). The compound is reported to have a effect of β-cell differentiation in pancreatic precursor cells and quell liver fibrosis by suppressing the activation of hepatic stellate cells (10, 11). Recently, we reported that CnP can suppress pancreatic cancer cell growth by inhibiting CAFs (12). However, to date, no investigation into the effects and mechanisms of action of CnP on CAFs in HCC has been reported.

This study aimed to investigate whether CnP can reduce HCC aggressiveness by suppressing activated CAFs. We thus evaluated the effects of CnP on CAFs derived from HCC. In addition, we evaluated the therapeutic benefit of combining the conventional anti-HCC agent sorafenib with CnP to target activated CAFs using mouse models. Subsequently, we elucidated the fundamental mechanism of the effect of CnP-altered CAFs using cap analysis of gene expression (CAGE, RRID:SCR_007574).

We established primary CAFs from HCC tissues, which were surgically resected as reported by Lau and colleagues (6). Specifically, HCC tissues with their surrounding capsules were minced and incubated in DMEM containg 1% penicillin–streptomycin (Thermo Fisher Scientific), 20% FBS, and 1 ng/mL basic FGF (bFGF, Wako). To allow cells to attach to the culture plates, the samples were cultured in a humidified incubator with 5% CO₂ at temperature of 37°C. We replenished the medium and removed unattached cells twice a week. The cells with myofibroblast-like morphology were defined as CAFs. Expression of α-smooth muscle actin (α-SMA) of cells was evaluated using Western blotting. CAFs were used between passage numbers 4 and 8. We were approved to establish CAFs from surgically resected HCC tissues by the Clinical Ethic Committee in our University (approval number 2016–118). CAF1 and CAF2 were isolated from different patients with HCC.

The human hepatoblastoma cell line HepG2 (RRID: CVCL_0027) and the human HCC cell line PLC/PRF/5 (RRID:CVCL_0485) were also used in this study. Both were purchased from the JCRB Cell Bank and authenticated within the last 3 years by short tandem repeat DNA profiling by BEX. All experiments were performed with Mycoplasma-free cells. Cells were cultured in DMEM containg 1% penicillin–streptomycin and 10% FBS and cultured at 37°C in a humidified incubator with 5% CO₂ atmosphere. We used cells between passage numbers 4 and 8.

The CnP used in the in vitro study was isolated from leaves of E. microphylla grown in Thailand and purified as described previously (Supplementary Fig. S1A; ref. 13). Specifically, a chloroform extract of the leaves of E. microphylla (4.97 g) was chromatographed on a silica gel column with hexane:ethylacetate (1:2) to yield 99.7 mg of the active fraction. This fraction was dissolved in MeOH, and the insoluble fraction was removed. The solution was then applied to a Toyopearl HW-40 column and eluted with methanol to yield 20.5 mg of active material. Finally, the crude material was purified by centrifugal partition chromatography and freeze-dried from t-butyl alcohol to yield 15.0 mg of white powder. The crude CnP preparation II (CCP-II) used for the in vivo study was obtained from leaves of Tabernaemontana divaricata. The method for extraction was described previously (14). Dry leaves (5 kg) were added to 500 L of 0.025 N hydrochloric acid and heated to 60°C for 30 minutes with gentle mixing. A clear supernatant was obtained by centrifugal filtration. The supernatants obtained from three serial extractions were pooled and cleaned with column chromatography using a synthetic adsorbent resin, Diaion HP20 (Mitsubishi Chemical). The column was washed with water followed by 30% (colume for volume) ethanol, and CnP was then eluted with pure ethanol. The elute was concentrated in vacuo and lyophilized. The CCP-II thus obtained contained 14 mg/g of CnP. Sorafenib was provided by Bayer HealthCare Pharmaceuticals Inc.

Protein extraction was performed using RIPA buffer (Wako), following the manufacturer's protocol. SDS-PAGE with 10% Bis–Tris gels was used for separating proteins. They were transferred to nitrocellulose sandwiches (No. 12369; Cell Signaling Technology). After blocking with 5% BSA or 5% skim milk, the membranes were incubated with primary antibodies overnight at 4°C. Anti-α-SMA mouse mAb (A2547; 1:1,000; Sigma-Aldrich, RRID:AB_476701), GPR68 rabbit antibody (No. ab133818; 1:1,000; Abcam), anti–β-actin mouse mAb (A5316; 1:1,000; Sigma-Aldrich, RRID:AB_476743) were used as primary antibodies. We treated the membranes with horseradish peroxidase–linked secondary antibodies. We detected the bands of protein on it with an Image Quant LAS 4000 instrument (GE Healthcare) and ECL Prime Western Blotting Detection System.

We cultured CAFs in DMEM containg 1 ng/mL bFGF and 20% FBS till they proliferated 90% confluency. Next, we changed the medium and the CAFs were cultured with DMEM (serum-free) for an additional 5 days. CAF-CM (conditioned medium) was then collected and centrifuged for 10 minutes at 1,500 g. The CnP effect on the CAF-CM was evaluated in CAFs cultured in DMEM (serum-free) containing CnP (0.3 μg/mL) for 5 days, and the CM was collected as CnP-treated CAF-CM.

The Cell Counting Kit-8 (CCK-8; Dojindo Laboratories) was used to evaluate cell proliferation. We seeded and cultured HCC cell lines and CAFs onto 96-well plates. We removeded the culture medium and change to DMEM (serum-free) with varying concentrations (from 0 to 1 μg/mL) of CnP after overnight incubation. Proliferation of HCC cell lines was also assessed by replacing HCC cell medium with CAF-CM and CnP-treated CAF-CM. We evaluated them by measuring the absorbance of each well after 48 hours using a spectrophotometerat 450 nm with a reference of 650 nm (Bio-Rad).

We evaluated the invasion ability of cells using 24-well Corning BioCoat Matrigel Invasion Chambers (Corning). The lower chamber was filled with medium supplemented with 3% FBS containing CnP (0.3 μg/mL), CAF-CM, or CnP-treated CAF-CM, and HCC cell lines in DMEM (serum-free) were seeded into the upper chamber. After 48 hours, Diff-Quik stain (Sysmex) were used for fixity and staining the cells. We evaluated five randomly selected fields under a microscope and counted the cells that invaded and migrated through the membrane.

Cytokine Array Kit (ARY022B; R&D Systems) was used for evaluation of cytokine profiles in CAF-CM and CnP-treated CAF-CM, following the manufacturer's instructions. An Image Quant LAS 4000 imager was used for quantification of spots on the array. An ELISA was used for measuring concentrations of cytokines in CAF-CMs. ELISA kits of IL6 (ab46027; Abcam), IL8 (ab46032; Abcam), and chemokine (C-C motif) ligand 2 (CCL2; ab179886; Abcam) were used following the manufacturer's instructions.

We extracted total RNA using a miRNeasy Mini Kit (No. 217004, Qiagen) and quantified it with an ND-1000 (NanoDrop Technologies). Reverse transcription and quantitative (RT-q) PCR were performed using an RT Kit (Toyobo) and Power SYBR Green PCR Master Mix (Life Technologies), respectively, following the manufacturer's instructions. Quantitative real-time PCR was performed with a Step-One-Plus PCR system (Applied Biosystems) by the 2^−ΔΔC_t method. The following primers were used: IL6 forward, 5′-TGCAATAACCACCCCTGACC-3′; reverse, 5′-CCCAGTGGACAGGTTTCT GA-3′; IL8 forward, 5′-TCCAAACCTTTCCACCCC-3′; reverse, 5′-CACAACCCTCTGCACCCA-3′; CCL2 forward, 5′-GAAGAATCACCAGCAGCAAGT-3′; reverse, 5′-TCCTGAACCCACTTCTGCTT-3′; Angiogenin forward, 5′-TGGGCGTTTTGTTGTTGGTCTTC-3′; reverse, 5′-CGTTTCTGAACCCCGCTGTGG-3′; Osteopontin(OPN) forward, 5′-ATCTCCTAGCCCCACAGAAT-3′; reverse, 5′-GTGGGTTTCAGCACTCTGGT-3′; GPR68 forward, 5′-CCTTCCGCTTCCACCAGTT-3′; reverse, 5′-TCGTCCTCGATGACCTCCT-3′; and β-actin forward, 5′-CACCATTGGCAATGAGCGGTTC-3′; reverse, 5′-AGGTCTTTGCGGATGTCCACGT-3′. β-actin was used to normalize RNA input for all RT-qPCR analyses.

First, we evaluated the effect of CAFs in a xenograft model by comparing tumor volume rates of PLC alone and PLC plus CAF in vivo. We used female mice in this study, following previous study (15). Female NOD-SCID mice which were 7 weeks old (CLEA Japan) were injected 5 × 10⁶ PLC cells and 5 × 10⁶ PLC plus 1 × 10⁶ CAF (CAF1) cells bilaterally and subcutaneously. Each group contained four xenografts. Moreover, we analyzed the CnP effects on PLC plus CAF xenograft in vivo. A week after injection the cells into mice, mice were divided into four groups (control, CnP, sorafenib, and CnP and sorafenib) randomly. Mice were treated with 1.0 μg/g CnP via subcutaneous injection twice weekly and 10 mg/kg sorafenib oral administration five times weekly. Each group had six xenografts. Tumor volumes and body weights of mice were measured twice weekly. The method to calculate tumor volumes was following formula: L/2 × S × S (L and S are the long and short diameters of the tumors). Protein levels of xenografts were measured immunohistochemically using tumors collected 5 days after the treatment started. We stained xenografts with hematoxylin and eosin staining and IHC and evaluated them microscopically. The serum biochemical tests of blood samples of mice were performed (Oriental Yeast Co) as evaluation of adverse events of the treatment. All in vivo experiments were conducted in compliance with the guidelines of our laboratory for animal experiments (approval number 18-024).

A paraffin-embedded block of xenografts was cut into 2-μm-thick sections and put on glass slides. Xylene was used for deparaffinization of each section and alcohol was used for dehydration. They were soaked in 0.3% H₂O₂/methanol for 30 minutes at room temperature to inhibit endogenous peroxidase. Next, they were soaked in heated water with 0.5% Immunosaver (Nisshin EM Co. Ltd.) at 98°C for 45 minutes. Nonspecific antigens were blocked by incubation with Protein Block Serum-Free (Dako) for 30 minutes at room temperature. The sections were then incubated with primary antibodies: anti–α-SMA mouse mAb (A2547; 1:1,000; Sigma-Aldrich, RRID:AB_476701), anti–caspase-3 antibody (No. 9662; 1:1,000; Cell Signaling Technology, RRID:AB_331439) and anti–Ki-67 antibody (M7240; 1:200; Dako, RRID:AB_2142367) for 24 hours at 4°C. After washing with PBS, they were incubated with Histofine Simple Stain MAX-PO (MULTI) kit (Nichirei Co.) for 45 minutes at room temperature. The chromogen 3, 3′-diaminobenzidine tetrahydrochloride was then applied as a 0.02% solution containing 0.005% H₂O₂ in 50 mmol/L ammonium acetate-citrate acid buffer (pH 6.0). Finally, the nucleus was counterstained with Mayer hematoxylin solution. A negative control was evaluated by replacing the primary antibodies with PBS in 0.1% BSA, and no staining was detected (Supplementary Fig. S1B). ImageJ (RRID:SCR_003070) version 1.51 was used for quantification of the positivity of α-SMA, Ki-67, and caspase-3. Masson Trichrome staining reagents (Muto Pure Chemical Ltd.) were used following the manufacturer's instructions.

CAFs were seeded in 10-cm dishes and cultured until they grew to their 90% confluency. The medium was changed to with or without 0.3 μg/mL CnP. After 6 hours, CAFs were collected and washed with PBS, snap frozen in nitrogen, and stored at −80°C. DNAFORM performed CAGE library preparation, sequencing, mapping, and gene expression and motif discovery analysis. Briefly, RNA quality was assessed using a Bioanalyzer (Agilent). First-strand cDNAs were transcribed to the 5′ ends of capped RNAs and attached to CAGE “bar code” tags. The sequenced CAGE tags were then mapped to mouse mm9 genomes using BWA (SCR_010910) software (v0.5.9) after discarding ribosomal or non-A/C/G/T base containing RNAs. Finally, the CAGE-tag 5′ coordinates were input for CAGEr clustering (16) using the Paraclu algorithm (17) with default parameters.

We used nontargeted control siRNA (ON-TARGETplus Non-targeting Pool; Dharmacon GE Healthcare) and GPR68-specific siRNA (MISSION esiRNA, EHU027261; Sigma). CAFs were suspended at a density of 5.0 × 10⁵ cells in 100-μL Opti-MEM I Reduced Serum Media (Thermo Fisher Scientific) and then mixed with GPR68-specific siRNA or nontargeted control siRNA as a negative control. We performed transfection of siRNA with a T CUY21 EDIT II electroporator (BEX) with 175 V as a poring pulse and 10 V as atransfer pulses. The mRNA levels in CAFs with GPR68-specific siRNA transfection are expressed relative to those in CAFs with nontargeted control siRNA transfection. We checked mRNA and protein levels in CAFs 48 hours after transfection with siRNA. We collected GPR68-knockdown CAF-CM, as described above.

Data are presented as means ± SD. To compare defferences between two groups was analyzed with t tests. ANOVA with Tukey multiple comparison tests was used to evaluated to compare the differences among four groups. For the in vivo study, differences between groups were evaluated using repeated-measures ANOVA with Bonferroni post hoc tests. Results were considered statistically significant when the relevant P < 0.01 or 0.05. Data were analyzed using EZR.

CAGE data in this study have been deposited to the Gene expression Omnibus and are available (accession number: GSE140530).

To evaluated the CAFs effect on HCC progression, we performed proliferation and invasion assays with HCC cell lines incubated in CAF-CM. The proliferation and invasion of HCC cells were promoted by CAF-CM treatment (Fig. 1A and B). Furthermore, the volumes of tumors formed in vivo by HCC cells coimplanted with CAFs increased in size compared with tumors formed by HCC cells alone (Fig. 1C). Tumors with HCC cells plus CAFs had α-SMA staining and Masson Trichrome staining in the stroma (Fig. 1C).

The effects of CnP were examined by evaluating the proliferation of CAFs and HCC cells. Proliferation of CAFs were significantly suppressed in a concentration-dependent manner of CnP treatment (Fig. 2A). However, HCC cell lines proliferation were suppressed by only high concentration of CnP (1 μg/mL; Fig. 2B). The expression of α-SMA, a marker of CAF activation, in CAF1 and CAF2 cells were suppressed by CnP concentration dependently (Fig. 2C).

To clalify the CnP effects on the malignant behavior of HCC cells promoted by CAFs, we performed proliferation assay and invasion assays of HCC cells treated with CnP-treated CAF-CMs. As mentioned above, direct treatment of low (0.3 μg/mL) CnP concentrations did not have significant effect on those of HCC cells (Fig. 3A and B), incubation in CAF-CM alone promoted both effects. Interestingly, CnP treatment eliminated the proliferation-promoting effects of CAF-CM (Fig. 3A), and CnP decreased the CAF-CM–enhanced invasive ability of HCC cells (Fig. 3B). The same result was obtained with Hep3B cells (Supplementary Fig. S2).

To clarify the differences between control CAF- and CnP-treated CAF-CM, we performed cytokine array analysis. IL6, IL8, CCL2, angiogenin, and OPN levels were markedly decreased in CnP-treated CAF-CM compared with those in CAF-CM (Fig. 4A). These cytokines were confirmed to be also suppressed at the mRNA and protein levels using qRT-PCR and ELISA-based analysis (Fig. 4B; Supplementary Fig. S3).

Genes of CAF that are strongly upregulated and downregulated by CnP were identified (Supplementary Fig. S4). Twenty cancer-promoting genes in CAFs strongly enhance the viability of cocultured cancer cells (18). However, whether CnP treatment can regulate these genes in HCC CAFs is not known. Therefore, we evaluated the expression of these 20 genes in CAFs using CAGE analysis. Notably, the expression of GPR68 was suppressed the most in CnP-treated CAFs relative to that in control CAFs (Fig. 5A). In addition, we validated that CnP treatment suppressed the expression of GPR68 in CAFs (Fig. 5B) and investigated the proliferative capabilities of HCC cells using CM derived from GPR68-suppressed CAFs (siGPR68 CAF-CM; Fig. 5C). Enhanced proliferation induced by CAF-CM was eliminated by GPR68 knockdown in CAFs (Fig. 5D). Furthermore, we confirmed using RT-PCR that the mRNA levels of IL6, IL8, angiogenin, and OPN were decreased but that CCL2 was not (Fig. 5E). The significant IL6 and IL8 suppression in GPR68-knockdown CAF-CM were validated with ELISA (Supplementary Fig. S5). Furthermore, we evaluated GPR68 expression levels of HCC cells and CAFs. GPR68 expression levels of HCC cells were higher than those in CAFs (Supplementary Fig. S6A). CnP had no effect on the expression of GPR68 in HCC cells (Supplementary Fig. S6B).

We investigated the combination therapy with CnP and the standard HCC treatment drug, sorafenib in vivo. Single treatment with CnP or sorafenib inhibited tumor growth, but treatment with CnP and sorafenib combined inhibited tumor growth to a greater extent (Fig. 6A). The tumors treated with this combination showed a reduction in Ki-67–positive HCC cells and α-SMA staining areas, along with an increase in caspase-3–positive HCC cells (Fig. 6B). We evaluated body weight and organ toxicity related to the treatment in this study, but there were no differences between the treatment groups and controls (Supplementary Fig. S7A and S7B).

We showed that CAFs promote HCC malignancy and CnP suppressed this interaction. We also demonstrated that CnP treatment strongly suppressed various cytokines such as IL6, IL8, CCL2, angiogenin, and OPN producted from CAFs. In vivo experiments showed that combined therapy with CnP and sorafenib inhibited tumor growth to a greater extent than single-agent treatment. In addition, our CAGE analysis and in vitro data indicated that CnP suppressed the expression of GPR68, which functions to promote cytokine production in CAFs. This study is the first report to show the significance of this new therapeutic compound for targeting CAF involvement in HCC.

A low concentration (0.3 μg/mL) of CnP did not suppress proliferation of HCC cells (Fig. 2B). It was suggested that CnP was more effective against CAFs than cancer cells because at low concentrations, it could specifically suppress CAFs in in vitro experiments (Figs. 2C, 4A, and B). However, sorafenib is effective against HCC and is widely used in clinical practice (19). In our in vivo study, the concurrent administration of CnP and sorafenib for targeting CAFs and HCC cells, respectively, strongly inhibited both stromal α-SMA staining in CAFs and the proliferation ability of HCC cells in xenograft tumors (Fig. 6A and B). No significant difference between the control and CnP-treated group was observed; however, xenografts in the CnP-treated group were a little smaller than those in the control group (Fig. 6A). Moreover, the expression of α-SMA, a marker of activated CAFs, in CnP-treated animals was suppressed. Therefore, we surmised that CnP suppressed the cancer-promoting effect via regulation of CAF activity. Combination therapy with CnP and existing anticancer agents, such as sorafenib, may be effective for HCC in patients with activated CAFs. The significance of combined therapy is anticipated to be verified in clinical trials. No apparent side effects were observed in the experimental mice used (Supplementary Fig. S7).

We demonstrated that CnP suppressed the production of multiple cytokines, particularly IL6, IL8, CCL2, angiogenin, and OPN. They were reported to play key roles in tumor progression. For example, IL6 regulates JAK-STAT3 signaling and the mammalian target of rapamycin signaling. These pathways promote the proliferation, invasion, and immunosuppression and invasion of tumor cells (3, 20). Furthermore, Zhao and colleagues showed that activated CAF effects on cancer cells were suppressed by an IL6-neutralizing antibody (3). IL8 is thought to promote epithelial-to-mesenchymal transition (EMT) of HCC cells (21). CCL2 is reported to activate Hedgehog signaling in HCC cells and suppress T-cell antitumor immune responses (22, 23). Angiogenin is believed to promote neovascularization of HCC and EMT of HCC cells (24, 25). OPN may promote migration and metastasis of HCC cells (26, 27). CnP might inhibit the HCC-promoting effects of CAFs via suppression of cytokine production and release.

In this study, we showed that CnP regulates IL6, IL8, CCL-2, angiogenin, and OPN produced by CAFs. These cytokines, except for angiogenin, were found to be regulated by GPR68 expression in CAFs. Previously, GPR68 was reported to regulate IL6 expression in CAFs in pancreatic cancer cells and IL8 in β cells in the pancreas (9, 28). Production of multiple cytokines, including IL6, IL8, angiogenin, and OPN, in our study, was weakened by GPR68 knockdown in CAFs. CnP downregulated CCL-2, but this factor is not regulated by GPR68. We hypothesized that CnP might regulate several cytokines via GPR68 and also CCL-2, which are not dependent on GPR68. However, the cancer-promoting activity of CAFs was weakened by the knockdown of GPR68 by siRNA. GPR68 is important in regulating the production of cytokines in CAFs. CAFs mainly promote HCC by secretion of IL6, IL8, angiogenin, and OPN, as induced by GPR68.

GPR68 expression is induced by activation of TNFα/NFκB signaling (9). We evaluated NFκB signaling after CnP treatment in CAF cells. However, the NFκB signaling was not altered by CnP treatment (Supplementary Fig. S8A). Moreover, GPR68 did not regulate α-SMA of CAF (Supplementary Fig. S8B). Further studies are necessary to clarify the effects of CnP on GPR68 regulation.

In conclusion, CnP was demonstrated to inhibit the HCC-promoting effects of CAFs through suppressing activation and cancer-promoting cytokines production of CAFs. The combination of a multikinase inhibitor, sorafenib, and CnP against HCC and CAFs, respectively, completely suppressed tumor growth in our in vivo analysis. Therefore, combined therapy with CnP and existing anticancer agents may be a promising therapeutic strategy in overcoming refractory HCC with activated CAFs. Moreover, we showed that GPR68 in CAFs is important in regulating the secretion of cytokines from CAFs. Therefore, CnP-based alteration of GPR68 in CAFs may be one of the fundamental mechanisms of preventing HCC promotion.

No disclosures were reported.

T. Yamanaka: Formal analysis, validation, investigation, visualization, writing–original draft, project administration, writing–review and editing. N. Harimoto: Project administration. T. Yokobori: Writing–original draft. R. Muranushi: Validation, investigation. K. Hoshino: Validation, investigation. K. Hagiwara: Validation, investigation. D. Gantumur: Validation, investigation. T. Handa: Validation, investigation. N. Ishii: Supervision. M. Tsukagoshi: Supervision. T. Igarashi: Supervision. A. Watanabe: Supervision. N. Kubo: Supervision. K. Araki: Supervision. K. Umezawa: Resources, investigation. K. Shirabe: Supervision, funding acquisition, project administration.

The authors would like to thank Bayer HealthCare Pharmaceuticals Inc. for providing sorafenib. The authors would like to thank Enago (www.enago.com) for the English language review.

This study was supported by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JP18H02877 and JP18K16299) and Research Program on Hepatitis from Japan Agency for Medical Research and development, AMED.

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.