Background
Hepatocellular carcinoma (HCC) is one of the fastest-increasing causes of cancer-related deaths in the United States and is one of the most common cancers in China [
1,
2]. Although remarkable improvements of therapeutic strategies, such as local ablation, surgery and liver transplantation, have been achieved for early-staged HCC, patients suffering advanced-staged HCC are still deficient in effective strategies to control the malignant progression. One of the main reasons for this bleak prognosis associated with advanced HCC is the high incidence of metastasis [
3]. Therefore, it is urgent to clarify novel molecular pathogenesis involved in the invasiveness and metastasis of HCC and to identify new effective therapeutic targets.
Cancer-associated fibroblasts (CAFs) are the most abundant cells in the tumor microenvironment (TME), a key source of the extracellular matrix that contributes to the desmoplastic stroma, and play crucial roles during cancer malignant progression and metastasis [
4,
5]. CAFs are found in stroma-rich primary HCC and facilitate proliferation, migration, invasion, epithelial-mesenchymal transition (EMT), therapeutic resistance, and induce cancer stem cell (CSC)-like phenotypes of HCC cells by reshaping the tumor microenvironment and paracrine of a variety of cytokines [
6,
7]. For example, peri-tumor tissue-sourced fibroblasts secrete various cytokines, including IL-6, CXCL1, CCL2, SCGF-β, CXCL8 and HGF, to recruit cancer stem cells, maintain cancer stemness and promote intrahepatic metastasis of HCC [
8]. Moreover, HCC-derived exosomal miR-1247-3p trans-differentiates lung fibroblasts into CAFs and then educates CAFs to secrete IL-6 and IL-8 to promote cancer stemness, chemoresistance, EMT, and tumorigenicity of HCC cells, leading to lung metastasis of HCC [
9]. Nevertheless, the explicit mechanism accounting for the interactions between CAFs and HCC cells is complex and still obscure.
EMT is a key process involved in cancer invasion and distant metastasis, featuring the loss of intercellular adhesion with the downregulation of epithelial markers (such as E-cadherin) and upregulation of mesenchymal markers (such as N-cadherin and vimentin). Cancer stem-like properties are distinguished by the propensity to exhibit high self-renewal, differentiation and tumorigenicity capacities [
10]. Although molecular markers of cancer stemness are still emerging, transcription factors including Nanog, Sox2 and Oct4 have been strongly identified as master mediators of pluripotency [
11]. In fact, a direct link between the EMT process and the acquisition of stem-like properties by neoplastic cells has been previously reported [
12]. Furthermore, cancer stem cells possess a propensity to express lower levels of epithelial marker and higher levels of mesenchymal markers during metastasis [
12].
Cartilage oligomeric matrix protein (COMP), a valuable maker of cartilage turnover, is a 524-KDa soluble pentameric glycoprotein [
13]. COMP expression in the fibrotic liver is increased by reactive oxygen species (ROS), chemokines, growth factors, matrix stiffness, and matricellular proteins [
14,
15]. However, the mechanisms underlying the regulation of COMP expression by ROS and other factors are still not clear. Moreover, COMP enhances the synthesis of type 1 collagen in hepatic stellate cells (HSCs) via CD36 receptor-mediated activation of MEK1/2-pERK1/2 signaling to regulate liver fibrosis [
16]. The potential role of COMP in tumors has been reported in breast cancer, prostatic cancer and colon cancer [
17‐
19], and a high expression level of COMP has been detected both in tumor cells and the surrounding stroma. Our previous study has also shown that HSCs-derived COMP facilitates invasion and metastasis of HCC by activating PI3K-AKT and MEK-ERK signaling in a CD36-dependent manner [
20]. However, little is known about the regulatory mechanism of COMP expression and its role in maintaining cancer stem-like phenotypes.
Forkhead box M1 (FOXM1), a member of FOX transcription factor family, has a crucial role in cell-cycle progression [
21]. Previous studies have shown that FOXM1 is overexpressed in a variety of human malignancies, and most of research has focused on tumor cells including HCC [
22]. Recently, studies have unveiled that FOXM1 plays a pivotal role in bleomycin-induced pulmonary fibrogenesis and serves as a driver of lung fibroblast activation [
23]. Analogously, deletion of FOXM1 in postnatal cardiomyocytes results in cardiac fibrosis [
24], and pulmonary artery smooth muscle cells-specific FOXM1 regulates hypoxia-induced pulmonary hypertension [
25]. However, its potential role in the stroma of HCC requires further elucidation. Intriguingly, there seems to be potential links between ROS and FOXM1. FOXM1 can exert its role in a ROS-dependent manner and has been recognized as a critical regulator of oxidative stress during oncogenesis [
26].
Resolvins, a family of endogenous proresolving and anti-inflammatory lipid mediators derived from ω-3 polyunsaturated fatty acid, have been generated from EPA and DHA [
27,
28]. The potent in vivo actions of RvD1 have been reported in many pathologies, such as obesity and those affecting the vascular, airway, dermal, renal and ocular systems, and in processes including pain, fibrosis and wound healing. Its role in governing neutrophil influx, macrophage resolution and reducing pro-inflammatory mediators seems to be fundamental in all organs [
27]. The illuminating view of tumors as “wounds that do not heal” provides insights that RvD1 may exert a critical role in carcinogenesis. Among this family, RvD1 is specially derived from DHA by biosynthetic pathways involving lipoxygenase (LOX) [
29], and RvD1 can exert effects through binding to its two receptors, lipoxin A4 receptor/formyl peptide receptor 2 (ALX/FPR2) and a GPCR denoted GPR32 [
30]. Resolvins differ from classic anti-inflammatories in that they stimulate, as agonists, the resolution of inflammation, act at significantly lower doses and are not immunosuppressive [
27,
31,
32]. Previous studies have confirmed that resolvins facilitate macrophage phagocytosis of debris from apoptotic tumor cells and counterregulate macrophage secretion of pro-inflammatory cytokines in a restricted receptor-dependent manner to suppress debris-stimulated tumor growth, and the antitumor activity of resolvins is mediated by stromal cells rather than a direct action on tumor cells [
31]. Interestingly, CAFs, which are a major component of the tumor stroma, play a profound role in forming the inflammatory microenvironment to regulate stroma-tumor interactions in HCC. However, the effects of resolvins on CAFs in HCC are still not clear.
In this study, we postulated that RvD1 could inhibit COMP secretion in CAFs in a paracrine manner via FPR2/ROS/FOXM1 signaling to block stoma-tumor cells interactions and then suppress EMT and cancer stemness to alleviate malignant progression of HCC.
Methods
Reagents and antibodies
The COMP ELISA kit, recombinant human COMP protein (rh-COMP) and human COMP neutralization antibody were purchased from R&D systems (Minneapolis, MN USA). RvD1 was obtained from Cayman Chemical Corporation (Ann Arbor, MI, USA). Detailed information about the antibodies utilized in this study is presented in Additional file
1: Table S1. N-acetyl-L-cysteine (NAC) and H
2O
2 were purchased from Sigma (St. Louis, MO, USA). Scrambled siRNA (si-Control: sense 5′-UUCUCCGAACGUGUCACGUTT-3′; antisense 5′-ACGUGACACGUUCGGAGAATT-3′) and ALX/FPR2 siRNA (si-FPR2: sense 5′-CGGUUUGUCAUUGGCUUUATT-3′; antisense 5′- UAAAGCCAAUGACAAACCGTT-3′) were purchased form GenePharma (Shanghai, China) as we previously described [
33]. Si-Control and si-FOXM1 were obtained from Santa Cruz Biotechnology. The pcDNA3.1-Control and pcDNA3.1-FOXM1 (pcDNA/ FOXM1) were purchased from Invitrogen (USA). Luciferase-expressing lentiviruses were purchased from GeneChem Co, Ltd. (Shanghai, China). All these reagents were stored following the manufacturer’s instructions.
Cell culture and intervention
Two HCC cell lines (SMMC-7721, Hep3B) were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and cultured in DMEM (HyClone, Logan, UT, USA) with 10% fetal bovine serum (FBS, Gibco, USA), 100 μg/mL streptomycin and 100 U/mL penicillin (Sigma, USA) in a humidified atmosphere at 37 °C containing 5% CO2.
Isolation of CAFs
The methods used to isolate CAFs from HCC tissue have been previously described [
34]. Briefly, surgically resected HCC tissues and peritumoral tissues were minced into 2–3-mm fragments, washed in phosphate-buffered saline (PBS), and cultured in 6-well plates supplemented with F12/DMEM containing 10% fetal bovine serum and 1% penicillin/streptomycin. Peritumoral fibroblasts (PTFs), and the CAFs were allowed to grow out of the tissue fragments at 37 °C with 5% CO
2. After purification, the expression of α-SMA in these cells was determined by immunofluorescence and western blotting. We used CAFs within 6 passages for the various experiments.
HCC-CAFs co-culture model
Both direct and indirect co-culture models were established to analyze the interactions between hepatocellular carcinoma cells and CAFs in the presence of RvD1. For the indirect HCC-CAFs co-culture model, CAFs were treated with or without RvD1 (400 nM) for 24 h; the cells were then serum-starved (1% serum in fresh medium) for an additional 48 h, and the conditioned medium (CM) of CAFs was subsequently collected and filtered. Then, the CM of CAFs was added to serum-starved HCC cells (SMMC-7721, Hep3B). For the direct HCC-CAF co-culture model, HCC cells and CAFs were proportionally mixed and seeded on the slides in 24-well plates. After the cells were serum-starved for 6–8 h, the cells were treated with RvD1 (400 nM) for 48 h.
Immunofluorescence staining
After completion of the designated treatment, immunofluorescence staining was performed according to a standard protocol as described in our previous study [
35]. The cells on the slides were imaged and recorded with appropriate excitation and emission spectra at a magnification of 400 × using a Zeiss Instruments confocal microscope (Zeiss, Oberkochen, Germany).
Enzyme-linked immunosorbent assay (ELISA)
CAFs were treated RvD1 at different concentrations (0, 200, 400, and 800 nM) for 24 h; then, the cells were serum-starved (1% serum in fresh medium) for an additional 48 h and the supernatant collected and centrifuged (1500 rpm for 5 min). Subsequently, the secretion of COMP into the conditioned medium was detected using ELISA kits following the manufacturer’s instructions (R&D Systems, USA). The contents of RvD1 in 25 HCC and adjacent nontumor samples were examined using an ELISA kit purchased from Cayman Chemical Corporation (Ann Arbor, MI, USA) as previously described [
31].
Transwell migration and invasion assays
The migration assay was performed using Transwell chambers (BD Biosciences, Franklin Lakes, NJ), in which 2 × 104 cells in 200 μL serum-free medium were plated in the upper chambers. For the invasion assay, the basement membranes of the filters were coated with 50 μL Matrigel (Matrigel; BD Biosciences, Bedford, MA). After completion of the designated intervention, the CAFs were utilized for the migration assay, and HCC cells (SMMC-7721, Hep3B) were used for the invasion assay. Finally, the cells that had migrated or invaded to the lower surface of the membrane were fixed and stained with crystal violet. The results were analyzed by counting the stained cells using optical microscopy (100 × magnification) in five randomly selected fields. Each experiment was carried out in triplicate wells and repeated at least three times.
After co-culturing with the CM collected from CAFs, HCC cells were seeded at a density of 5000 cells per well in six-well ultralow attachment plates (Corning, Corning, NY, USA) and then incubated with serum-free DMEM/F12 medium (Gibco) supplemented with 1% B27 (Invitrogen, Carlsbad, CA, USA), 20 ng/mL human FGF and 20 ng/mL human EGF. The cells were subsequently cultured at 37 °C with 5% CO2 for two weeks. A microscope (Nikon Instruments Inc.) was utilized to count the number and measure the diameter of the tumor sphere at a magnification of 200 × .
Intracellular ROS measurement
Intracellular ROS production in CAFs was determined using the ROS probe 2,7-dichlorofluoresceindiacetate (DCF-DA). Briefly, after treatment with RvD1, NAC or H2O2, CAFs were incubated with F12/DMEM containing DCF-DA (10 μmol/L) for 30 min at 37 °C, and the DCF-DA fluorescence intensity was measured by flow cytometry using a FACSCalibur (BD Biosciences, San Diego, CA, USA), or using a Zeiss Instruments confocal microscope to visualize the ROS level in CAFs.
Cell viability assay
SMMC-7721, Hep3B were seeded into 96-well plates at a density of 5 × 103 cells per well and treated with conditioned medium (CMCAFs, CMCAFs + RvD1), CMCAFs + anti-COMP and CMCAFs + RvD1 + rh-COMP for 24, 48, 72 and 96 h. Subsequently, the MTT assay was applied to assess the cell viability, and the absorbance at 490 nm detected by using a multiwall microplate reader (BIO-TEC Inc., VA) was used for the assessment.
After finishing the designated treatment, the cells were plated in a six-well plate (1000 cells per well) and cultured in fresh medium for 14 days. The plates were washed with phosphate-buffered saline (PBS), fixed in 4% formalin, stained with crystal violet solution for 15 min and then washed with PBS to remove excess dye. The number of colonies was counted for each sample.
Quantitative real-time PCR (qRT-PCR)
Total cellular RNA was extracted using TRIzol reagents (Takara Bio, Dalian, China) and quantitated by the absorbance at 260 nm. The RNA (1 μg) sample was reverse-transcribed with PrimeScript RT Master Mix, and quantitative real-time PCR was conducted with SYBR-Green PCR Master Mix (Takara Bio, Dalian, China) using gene-specific primers. The sequences of the specific primers are presented in Additional file
2: Table S2. GAPDH was used as a loading control, and the results were calculated using the 2
-ΔΔCt method.
Western blot analysis
Total proteins of CAFs, SMMC-7721 and Hep3B (1 × 10
6) grown under our experimental conditions were extracted using RIPA Lysis Buffer (Beyotime, Guangzhou, China). The BCA protein assay kit (Pierce, Rockford, USA) was utilized to measure the concentration of the proteins based on the manufacturer’s instructions. The details of the western blot assay have been previously described [
36]. The immunoreactive bands to visualize the expression of designated proteins were evaluated using the chemiluminescence detection system through the peroxidase reaction, and the images of the bands were recorded with the ChemiDoc XRS imaging system (Bio-Rad, USA). β-actin was used as the internal loading control.
Transfection
Plasmids and siRNAs were transiently transfected into CAFs cultured in a 6-well plate at a density of 2 × 105 per well using lipofectamine 2000 (Invitrogen, CA, USA). For transient transfection, CAFs were transfected with plasmids or siRNAs at different concentrations according to the manufacturers’ instructions for 48 h prior to further experiments. The luciferase-expressing lentiviruses were transfected into Hep3B cells following the standard protocols from GeneChem Co, Ltd.
The final full-length reporter plasmid containing multiple FOXM1-binding sites was designated pLuc–COMP-#1 and purchased from Genecopoeia (#HPRM30080 Guangzhou, China). The deletion mutation reporter plasmid, which did not have FOXM1-binding sites, designated as pLuc–COMP-#2, was then generated. Both constructs were verified by sequencing the inserts and flanking regions of the plasmids.
Dual luciferase reporter assay
CAFs were transfected with the indicated COMP promoter reporter (pLuc–COMP-#1 and pLuc–COMP-#2 respectively), siFOXM1, or overexpression-FOXM1 plasmid in different groups. The COMP promoter activity was normalized via co-transfection of a β-actin/Renilla luciferase reporter containing the full-length Renilla luciferase gene. The luciferase activity in the CAFs was quantified using a dual luciferase assay system (Promega) for 24 h after transfection.
Chromatin immunoprecipitation (ChIP)
The ChIP assay were performed as previously described [
37]. Briefly, the ChIP assay was conducted using a commercial kit (Upstate Biotechnology) based on the manufacturer’s instructions. The PCR primers are indicated in Additional file
2: Table S2.
In vivo tumorigenesis assays
All animal experiments were performed according to the protocols sanctified by the ethical committee of Xi’an Jiaotong University. For the subcutaneous tumor formation assay, 5 × 10
5 of Hep3B cells were suspended in 100 μL PBS alone in the control group, and 100 μL of Hep3B cells (5 × 10
5) and CAFs (5 × 10
5), mixed in a single cell suspension, was subcutaneously injected into the left flanks of 4-week-old male BALB/c nude mice (obtained by and housed in the Animal Center at Medical College, Xi’an Jiaotong University). Then, the animals that received the Hep3B and CAF co-injection were randomly divided into two groups (six mice per group). One group received RvD1 (6 μg/kg/d for 4 weeks), and the other group was treated with vehicle control. Tumor growth was continuously monitored by calculating the tumor volume according to the following formula: V (tumor volume) =0.5 × s (shorter diameter)
2 × L (longer diameter). The mice were sacrificed at day 28, and the tumor samples were weighed, measured and then stained by immunohistochemistry for histological analyses. The immunohistochemistry procedure was performed as we previously reported [
38].
An orthotopic liver tumor model in nude mice was established to assess metastasis according to our previous report [
39]. Briefly, Hep3B cells were transfected with luciferin lentiviruses, and subsequently, 5 × 10
5 transfected-Hep3B cells mixed with 5 × 10
5 CAFs and suspended in 100 μL PBS were injected into nude mouse liver. The mice (six mice per group) were then treated with RvD1 (6 μg/kg/d for 3 weeks) or vehicle through intraperitoneal injection. After 3 weeks, bioluminescence imaging (BLI) was conducted to monitor the tumor volume and metastases after injection of 450 mg/kg D-luciferin substrate (Biosynth, Naperville, IL, USA) in PBS into anesthetized mice.
Statistical analysis
All the data are displayed as the mean ± standard deviation (SD) of three independent experiments. The Student’s t-test was applied to compare two groups. Statistical analyses for multiple comparisons were conducted using one-way ANOVA followed by the LSD post hoc test with SPSS 18.0. *, P < 0.05 and **, P < 0.01 were considered to indicate a statistically significant difference.
Discussion
Cancer stem cells require stromal signals to maintain pluripotency and self-renewal capacities to confer successful metastatic colonization [
42,
43]. Previous studies have identified stromal fibroblasts as the stem cell niche responsible for the production of periostin, a component of the extracellular matrix, to recruit Wnt1 and Wnt3A and then activate Wnt signaling in breast cancer stem cells. These infiltrating breast cancer cells induce lung fibroblasts expressing periostin to maintain stemness to achieve initial metastasis colonization [
44]. Our study also demonstrated that CAFs-derived COMP induced EMT and cancer stem cell-like properties to promote invasion and metastasis of HCC, which was in accord with previous findings that IL-6 secreted by CAFs confers stem-like properties in HCC via the upregulation of stemness-correlated transcription factors including Sox2, Oct4 and Nanog [
34]. Thus, educational cancer-associated fibroblasts in the tumor microenvironment play crucial roles in cancer metastasis, and targeting the CAF-provided stem niche may represent a novel strategy for the treatment cancer metastasis.
Most cases of HCC initiate from cirrhotic livers, which contain an abundance of trans-differentiated myofibroblasts from quiescent fibroblasts or hepatic stellate cells [
6]. Previous researchers have demonstrated that COMP plays a key role in modulating liver fibrosis through activating MEK1/2 –pERK1/2 signaling and enhancing the synthesis of type 1 collagen in hepatic stellate cells (HSCs), and HSCs-derived COMP facilitates invasion and metastasis of HCC [
20]. These results suggest that COMP has an essential role in HCC tumorigenesis initiated in a fibrotic or cirrhotic background. In fact, a positive feedback loop exists between COMP expression and TGF-β signaling [
13,
45], in which onset of the fibrotic or cirrhotic process leads to a self-perpetuating cycle with COMP promotion of TGF-β activity and TGF-β facilitation of COMP expression [
45,
46]. Previous findings have highlighted that lipoxin A4, another endogenous anti-inflammatory lipid mediator derived from the metabolite of arachidonic acid, can mitigate the invasion and metastasis of pancreatic cancer by inducing inhibition of autocrine TGF-β1 signaling [
33]. Interestingly, herein we found that RvD1 inhibited COMP in CAFs in a paracrine manner to modulate EMT and cancer stemness in HCC, providing a promising therapeutic strategy for patients with HCC. However, whether RvD1 inhibits COMP expression by interrupting the positive feedback loop between COMP expression and TGF-β signaling in CAFs requires further clarification.
Our further investigations demonstrated that RvD1 inhibited COMP in CAFs in a paracrine manner via FPR2/ROS/FOXM1 cascades. First, RvD1 suppressed the expression of COMP, FOXM1 and ROS in a receptor-dependent manner; then, manipulation of the ROS level was found to influence the inhibitory efficacy of RvD1 toward the expression of COMP and FOXM1; furthermore, as a critical regulator of oxidative stress, FOXM1 could directly bind to the COMP promoter and regulate COMP expression at the transcriptional level, and abrogation of FOXM1 recruitment through FPR2/ROS signaling was an important mechanism of RvD1-mediated inhibition of COMP expression. These results implicated that oxidative stress-mediated FOXM1 directly modulated COMP expression, and targeting ROS/FOXM1/COMP cascades may provide new avenues to inhibit CAF-induced cancer stemness in HCC.
In this study, we mainly concentrated on the upstream mechanism by which RvD1 provided paracrine inhibition of CAFs-derived COMP via FPR2/ROS/FOXM1 cascades to prevent EMT and cancer stemness in HCC. Nevertheless, further mechanisms by which CAFs-derived COMP regulates stem-like phenotypes in HCC remain unknown. In conjunction with our previous studies [
20], we postulated that CAFs-derived COMP may function through binding to its potential receptor on HCC cells and subsequently activating downstream signaling to regulate stem-like properties. It would be of great interest to explore its explicit mechanisms in future studies.
Previous studies have illustrated that conventional chemotherapy, irradiation and targeted therapy are a double-edged sword, as these methods not only result in the killing of cancer cells to reduce tumor burden but also generate tumor-promoting effects via treatment byproducts, i.e., apoptotic tumor cells or tumor cell debris [
31]. Moreover, these therapies can trigger a cytokine storm in the tumor stroma, such as the release of IL-6 and TNF-α, as well as activate macrophages to generate pro-inflammatory mediators to stimulate tumor growth and contribute to recurrence [
31,
47,
48]. The tumor stroma is also regarded as a niche for the maintenance of cancer stem-like properties and, thus, participates in resistance to conventional therapy [
49]. Therefore, overcoming the dilemma between killing tumor cells and stroma resistance is paramount to preventing tumor recurrence after therapy. Intriguingly, RvD1, a novel endogenous anti-inflammatory lipid mediator governing neutrophil influx, macrophage resolution and pro-inflammatory cytokine reduction, promotes the clearance of tumor debris and subsequent inhibition of tumor growth. Furthermore, our study also showed that RvD1 could inhibit CAF-derived COMP in a paracrine manner to suppress EMT and stemness in HCC. Consequently, RvD1, as a novel anti-inflammatory mediator targeting the tumor stroma, may be a useful agent in conjunction with conventional therapy to promote treatment outcomes in HCC patients.