Tumor-Associated Macrophage Promotes Tumor Progression via STAT3 Signaling in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the fifth most common cause of cancer in the world [1]. Although surgical therapies for HCC have progressed and outcomes of HCC have improved, HCC still often recurs after surgery [2,3]. Sorafenib, one of the molecular targeted therapies, was reported to show activity against unresectable HCCs; however, its survival advantage is only 3.7 months [4]. New therapeutic targets are required to improve the survival of patients with HCC.

Signal transducer and activator of transcription 3 (STAT3) is an important molecule in tumor progression [5]. STAT3 activation occurs via phosphorylation and dimerization of tyrosine residue (Tyr705), leading to nuclear entry, DNA binding and gene transcription. STAT3 was regarded as a critical transcription activator for cell cycle- or cell survival-related genes. Bcl-XL is an antiapoptotic protein transcribed by STAT3 activation [6]. Some cytokines such as interleukin (IL)-6 or IL-10 activate STAT3 signaling via their receptors [7]. Constitutive activation of STAT3 has been demonstrated to contribute to tumorigenesis in breast cancer [8], colon cancer [9], lung cancer [10], pancreatic cancer [11], prostate cancer [12], and melanoma [13]. In human HCC, STAT3 phosphorylation was also detected and related to tumor progression [14], angiogenesis [15] and tumorigenesis [16]. The tumor microenvironment is closely associated with the growth of tumor cells, and tumor-associated macrophages play an important role in tumor progression [17]. Macrophages are major inflammatory cells that infiltrate tumors; several studies have shown that high infiltration of tumor-associated macrophages was associated with tumor progression and metastasis [17,18,19,20] and predicts poor prognosis in patients with HCC [21]. Tumor-associated macrophages activate STAT3 in ovarian cancer [22] and glioblastoma [23]. However, the correlation between tumor-associated macrophages and STAT3 activation of HCC tumor cells is unknown. Therefore, we examined STAT3 activation, cytokine expression and infiltration of tumor-associated macrophages in resected HCCs and analyzed their association with clinicopathological findings. Alterations in cell growth and migration by cytokine stimulation and STAT3 inhibitor were also analyzed in HCC cell lines.

One hundred and one available paraffin-embedded specimens from patients with HCC who underwent hepatectomy between January 1997 and December 2001 in our institute were selected by reviewing their pathology data. Any patients undergoing previous or noncurative surgery were excluded. After the surgery, monthly measurement of the serum α-fetoprotein (AFP) level was performed. In addition, ultrasonography and dynamic CT were performed every 3 months. The postoperative survival period or recurrence was entered into the database immediately when a patient died or if recurrence was strongly suspected on diagnostic imaging such as CT or magnetic resonance imaging.

This study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethics committees of Kyushu University Hospital (grant No. 21-117). Informed consent was obtained from each patient included in the study.

Sections of resected specimens were fixed in 10% buffered formalin, embedded in paraffin and stained by Envision+ system and DAB kit (Dako, Glostrup, Denmark). Immunohistochemical stains were performed with antibodies of phosphorylated STAT3 (pSTAT3; Tyr 705; D3A7, 1:50; Cell Signaling Technology), CD163 (10D6, 1:200; Novocastra), IL-6 (rabbit polyclonal, 1:1,000; Abcam), Ki-67 (MIB-1, 1:200; Dako), and Bcl-XL (rabbit polyclonal, 1:200; Santa Cruz Biotechnology, Santa Cruz, Calif., US). Sections were pretreated before being incubated with primary antibodies in a microwave oven at 99°C for 20 min for pSTAT3, CD163, IL-6 and Bcl-XL or in a pressure cooker for 25 min for Ki-67.

Each slide was stained in serial sections and examined by two pathologists (Y.M. and S.A.). In nuclear staining of pSTAT3 and Ki-67 and in cytoplasm staining of Bcl-XL, the percent positive cells was estimated by count of 1,000 tumor cells in most staining areas (hot spots). Staining of CD163, a marker of tumor-associated macrophages [19,22,23,24,25], and IL-6 was evaluated by estimating the total counts of cytoplasm or membrane at 3 high-power fields. The mean of nuclear pSTAT3-positive cells in HCCs was 10.7% (range 0–82.0), and pSTAT3 stain was classified into a positive (≥10.7% of tumor cell nuclei) and a negative group (<10.7% of tumor nuclei). Furthermore, in the cases of the pSTAT3-positive group (n = 36), the CD163-positive cells were counted separately in areas of pSTAT3-positive and pSTAT3-negative HCC cells.

For double staining of IL-6 and CD163, HCC specimens were boiled in 10 mM citrate buffer (pH 6.0) for 20 min and incubated with IL-6 primary antibody (1:1,000) at room temperature for 15 min. The sections were washed three times and incubated with anti-rabbit horseradish peroxidase-conjugated polymer at room temperature for 45 min; IL-6 was visualized by DAB kit. Next, the sections were boiled in 10 mM citrate buffer (pH 6.0) for 10 min, incubated with CD163 primary antibody (1:200) for 90 min and incubated with anti-mouse alkaline phosphatase-conjugated polymer at room temperature for 45 min. CD163 of the sections was visualized by New Fuchsin Substrate kit (Nichirei, Tokyo, Japan).

Human HCC cell lines PLC/PRF/5 and Huh7 were obtained from Riken Bioresource Center, Tsukuba, Japan, and cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 1 or 10% fetal bovine serum (FBS). PLC/PRF/5 and Huh7 cells were maintained in DMEM containing 1% FBS for 24 h prior to IL-6 (Peprotech, Rocky Hill, N.J., USA) stimulation. All in vitro experiments were done in triplicate.

Cellular proteins were solubilized in lysis buffer containing protease inhibitor and phosphatase inhibitor 30 min after stimulation with IL-6 (20 µg/ml). Equal amounts of protein were separated by SDS-PAGE and then transferred to the polyvinylidene fluoride membrane. Following blocking in Tris buffer containing 2% BSA, the membrane was stained with 1:1,000 dilution of anti-STAT3 (Cell Signaling Technology, Danvers, Mass., USA) and anti-pSTAT3 (Cell Signaling Technology) antibodies, then washed and incubated with horseradish peroxidase-conjugated secondary antibody (Cell Signaling Technology). Bands were visualized by the enhanced chemiluminescence system (GE Healthcare, UK).

PLC/PRF/5 and Huh7 cells were seeded at a density of 5 × 10⁴ cells/24-well plates and maintained in conditioned medium for 24 h before stimulation. Viable cells were counted by trypan blue stain 48 h after stimulation with IL-6 (25 ng/ml).

PLC/PRF/5 and Huh7 cells were seeded at a density of 5 × 10⁴ cells/6-well plates. Approximately 24 h later, when the cells were 100% confluent, a sterile 100-µl pipette tip was used to longitudinally scratch a constant-diameter strip in the confluent monolayer. The medium and cell debris were aspirated away and replaced by 2 ml of fresh DMEM containing 1% FBS with or without IL-6 (25 ng/ml). Photographs were taken at 0 and 48 h after wounding by phase-contrast microscopy. For statistical analysis, three randomly selected points along each wound were marked, and the horizontal distance between the migrating cells and the initial wound was measured 48 h later.

In both cell growth and wound-healing assays, PLC/PRF/5 and Huh7 cells were cultured in DMEM containing 1% FBS and IL-6 (25 ng/ml) with or without 100 nM S3I-201 (NSC 74859; Santa Cruz Biotechnology). S3I-201 was treated 30 min before IL-6 stimulation. DMSO was used for control.

Statistical analysis was carried out using Microsoft Excel software and JMP software (SAS Institute, Cary, N.C., USA). Comparison between pSTAT3 staining and clinicopathological findings or staining of other antibodies was evaluated by Pearson’s χ², Fisher’s exact tests and the Mann-Whitney U test. Patient survival analysis including overall survival (OS) and disease-free survival (DFS) was calculated by the Kaplan-Meier method; differences were evaluated by the log-rank test. For multivariate analysis, the Cox proportional hazard model was used. Two-sided Student’s t test was applied for analysis of in vitro data. Statistical analyses were considered significant at a p value <0.05.

pSTAT3 was stained in the nuclei of HCC cells, normal endothelial cells, some bile duct epithelial cells and inflammatory cells. pSTAT3 nuclear staining in HCC cells is displayed in figure 1a. The mean percentage of nuclear pSTAT3-positive cells in HCCs was 10.7% (range 0–82.0). The number of pSTAT3-positive and pSTAT3-negative samples was 36 and 65, respectively. We also examined pSTAT3 staining at the lesions of 19 portal venous invasions (PVIs) and 12 intrahepatic metastases (IMs) in 101 cases. Fifteen of 19 PVIs (78.9%) and 9 of 12 (75.0%) IMs were defined as pSTAT3-positive cases (fig. 1b). Positive rates in both lesions were significantly higher than those in the primary lesions (35.6%; p < 0.05; fig. 1c).

A comparison of clinicopathological findings in pSTAT3-positive and pSTAT3-negative groups is summarized in table 1. The pSTAT3-positive group showed higher AFP (p = 0.0276), larger tumor size (p = 0.0092), more frequent IMs (p = 0.0214), a higher Ki-67 labeling index (LI; p = 0.0002), and more Bcl-XL-positive cells (p = 0.0001) than the pSTAT3-negative group, whereas no significant differences were noted with respect to sex, age, infection of hepatitis viruses, liver cirrhosis, PIVKA II (proteins induced by vitamin K absenceor antagonist II), histological differentiation, capsule formation, infiltration to the capsule, and vessel invasion.

The median follow-up period was 1,391 days (range 36–3,289). pSTAT3 expression was significantly correlated with OS and DFS (p = 0.0234 and 0.0003, respectively; fig. 1d, e). Univariate analyses indicated that high AFP (>100 ng/ml), large tumor size (>5 cm), PVIs and IMs were prognostic factors for OS and male sex, hepatitis C virus infection, high AFP (>100 ng/ml) and IMs for DFS (table 2). Multivariate proportional hazard models revealed that high AFP and IMs were independent prognostic factors for OS and pSTAT3 expression and high AFP for DFS (table 2).

CD163-positive cells were localized around the pSTAT3-positive HCC cells (fig. 2a). Figure 2b shows the boxplots of CD163-positive cells (mean ± SD: pSTAT3-negative group, 28.5 ± 15.4; pSTAT3-positive group, 42.6 ± 26.6). The pSTAT3-positive group (n = 36) showed statistically higher CD163-positive cells (p = 0.0013; fig. 2b) than the pSTAT3-negative group (n = 65). Furthermore, we analyzed the localization of CD163-positive cells in areas where pSTAT3-positive and pSTAT3-negative HCC cells existed in the pSTAT3-positive group (n = 36), and figure 2c shows the boxplots of the analyses (mean ± SD: pSTAT3-negative area, 27.7 ± 17.9; pSTAT3-positive area, 42.6 ± 26.6). In the pSTAT3-positive group, CD163-positive cells in areas where pSTAT3-positive HCC cells existed were statistically higher than in those where pSTAT3-negative HCC cells existed (p = 0.0064; fig. 2c).

IL-6 was stained in some macrophages, HCC cells and normal hepatocytes (fig. 3a). According to the double staining of CD163 and IL-6, CD163-positive cells (tumor-associated macrophages) expressed IL-6 (fig. 3b). We divided them into two by the median values of positive macrophages of IL-6 and CD163, and thereby classified the 101 cases into four groups such as CD163^low and IL-6^low, CD163^low and IL-6^high, CD163^high and IL-6^low, and CD163^high and IL-6^high. HCCs containing high infiltration of IL-6- and CD163-positive macrophages exhibited a significantly higher rate of positive staining for pSTAT3 (fig. 3c).

IL-6 stimulation increased the levels of pSTAT3 in both PLC/PRF/5 and Huh7 HCC cell lines (fig. 4a). IL-6 stimulation resulted in higher proliferation (fig. 4b) and greater migration distance (fig. 4c) versus control. S3I-201, a STAT3 inhibitor, inhibited IL-6-induced STAT3 phosphorylation (fig. 4a) and decreased proliferation and migration of HCC cell lines (fig. 4b, c).

Our results suggest that macrophage infiltration into HCC tissue stimulates tumor cells via STAT3 signaling. In the present study, pSTAT3-positive HCCs show malignant behavior and confer poor prognosis because of their high abilities of cell proliferation and migration. We found that high pSTAT3 expression was significantly correlated with larger tumor size, higher Ki-67 LI, higher Bcl-XL expression and greater frequency of IMs, and higher pSTAT3 expression was observed in the lesions of IMs and PVIs than in the primary lesions. STAT3 activation upregulates cell cycle-related, antiapoptotic and invasion genes [8,9,10,11,12,13,26,27]. In our results, large tumor size and high Ki-67 LI indicated cell cycle progression, high Bcl-XL expression indicated antiapoptotic function, and frequent IMs indicated invasive capacity. Furthermore, high pSTAT3 expression in the lesions of IMs and PVIs suggests that the tumor cells with STAT3 activation in the primary lesions tended to invade the vessels and metastasize to the other liver sites. Xie et al. [26,27] reported that activated STAT3 regulated tumor invasion of melanoma cells by regulating the gene transcription of matrix metalloproteinase 2. These results suggest that pSTAT3 expression plays an important role for cell survival and migration in HCC, consistent with previous studies in HCC [14,17,28,29] and other tumors [5,8,9,10,11,12,26,27,30].

In recent years, it has been recognized that the balance between tumor immunity and tumor progression is important [31]. The present study revealed that tumor-associated macrophages are important for pSTAT3 expression of tumor cells. First, CD163-positive cells around pSTAT3-positive HCC cells were statistically more prevalent than around pSTAT3-negative HCC cells. Some of the CD163-positive cells expressed IL-6 in HCC tissue, and STAT3 was phosphorylated by IL-6 stimulation in vitro. These results suggest that tumor-associated macrophages can activate HCC cells via STAT3 signaling by IL-6 expression. However, CD-163-positive cells were detected not only in the pSTAT3-negative tumor area but also in the pSTAT3-positive tumor area and in noncancerous liver tissue. IL-6-secreting tumor-associated macrophages may be part of the CD163-positive cells, and the CD163-positive cells in the pSTAT3-positive tumor area were more stained by IL-6 than in the pSTAT3-negative tumor area and normal liver tissue (data not shown). Tumor-associated macrophages express immunosuppressive cytokines including IL-4, IL-6, IL-10, IL-17, and IL-23 [32,33]. These cytokines activate immunosuppressive inflammatory cells such as other tumor-associated macrophages, helper T cells and regulatory T cells and suppress antitumor inflammatory cells such as lymphocytes, natural killer cells and dendritic cells [34,35,36]. Kuang et al. [32] showed that tumor-associated macrophages expressed IL-6 in vitro, whereas Ding et al. [21] reported that tumor-associated macrophage was correlated with poor prognosis in HCC. Our results are consistent with these previous reports.

Both proliferation and migration of PLC/RPF/5 and Huh7 were increased following IL-6 stimulation and STAT3 phosphorylation. On the other hand, IL-6 was expressed in not only macrophages but also in HCC cells. STAT3 can be activated through autocrine signaling in HCC cells; moreover, other cytokines and growth factors might activate STAT3 of tumor cells [22,23,24]. It is very difficult to exclude activation of STAT3 by the autocrine manner. In our data, STAT3 activation of HCC cells was not correlated with surrounding IL-6-positive normal hepatocytes and HCC cells but it was correlated with the infiltration of CD-163-positive cells (fig. 2). Thus, we thought that the IL-6 secretion of tumor-associated macrophages is more important for STAT3 activation of HCC cells than the IL-6 secretion of other cells.

Recently, STAT3 phosphorylation inhibitors such as S3I-201 have been investigated in vitro and in vivo [28,29,30]. Avella et al. [37] reported that STAT3 can be one of the targets of chemoimmunotherapies. In our study, S3I-201 inhibited IL-6-induced STAT3 phosphorylation in vitro and decreased cell proliferation and migration. The inhibition of tumor-associated macrophages as therapeutic strategy of malignancy has been investigated, too [38,39,40,41]. Therefore, it is very important to suppress tumor-associated macrophage activation and STAT3 signaling in the treatment of HCC. Furthermore, tumor-associated macrophage activation requires STAT3 signaling [22]. We consider that the STAT3 inhibitor may suppress STAT3 activation in both tumor cells and tumor-associated macrophages, release antitumor immune systems from suppression by tumor-associated macrophages and thereby control tumor progression of HCC. Therefore, STAT3 signaling is a feasible therapeutic target for HCC.

In conclusion, STAT3 activation is one of the prognostic factors in HCC. Tumor-associated macrophage expresses IL-6 and can activate STAT3 signaling of HCC cells, resulting in their cell proliferation, antiapoptosis and migration. In the future, HCC may be suppressed by inhibition of STAT3 signaling of tumor cells and tumor-associated macrophages.

The authors have no conflicts of interest.