Main

Myxoid liposarcoma (MLS) is the second most common subtype of liposarcoma, comprising ∼30% of all liposarcomas and about 10% of all adult soft tissue sarcomas (Antonescu and Ladanyi, 2013). The morphologic spectrum of tumours ranges from purely myxoid tumours to the more aggressive highly cellular round cell liposarcoma. Primary myxoid liposarcoma with more than a 5% round cell component is associated with a poor clinical prognosis (Antonescu et al, 2001; Oda et al, 2005). The most commonly affected regions tend to be the deep compartments of the extremities. Metastasis occurs in up to one-third of cases and is characterized by a propensity to spread to unusual extra-pulmonary locations (Haniball et al, 2011; Asano et al, 2012). However, the mechanism of tumour progression in MLS remains unclear. Therefore, it is important to elucidate the mechanism and to discover new prognostic factors and medical treatment targets.

Recent reports have shown that tumour malignancy is dependent upon interactions in the tumour microenvironment between tumour cells and stromal cells such as immune cells, fibroblasts, and endothelial cells (Kamura et al, 2010; Pietras and Ostman, 2010). In particular, tumour-associated macrophages (TAMs) are stromal cells that are known to promote tumour invasion, metastasis, and angiogenesis in various tumours (Condeelis and Pollard, 2006; Qian and Pollard, 2010). Tumour-associated macrophages produce and secrete growth factors, cytokines, and other inflammatory mediators that may have important roles in cancer progression (Condeelis and Pollard, 2006; Baay et al, 2011). These tumour-promoting functions are consistent with clinical studies showing that high macrophage density in many human cancer types is associated with increased tumour angiogenesis and metastasis, and/or a poor prognosis.

In carcinomas, the majority of studies show that greater TAM infiltration is associated with poor prognosis, suggesting that TAMs may promote tumour progression (Leek et al, 1996; Komohara et al, 2013). With regard to sarcomas, we previously reported that TAM infiltration predicts poor prognosis in Ewing sarcoma (Fujiwara et al, 2011). Lee et al (2008) reported that an increased density of infiltrating macrophages was associated with poor outcomes in non-gynaecologic leiomyosarcomas. In contrast, Buddingh et al, (2011) showed that TAMs were associated with reduced frequency of metastasis and improvement in overall survival in. Thus, the role of TAMs in sarcomas differs depending on sarcoma subtypes, and there are few findings regarding their role in MLS. In this study, we investigated the role of TAMs in MLS.

Materials and methods

Cell lines

The myxoid liposarcoma cell lines MLS1765 and MLS402-91 were kindly provided by Dr Aman (Department of Clinical Genetics, University Hospital, Lund, Sweden) (Aman et al, 1992). The murine macrophage RAW264.7 cell line was obtained from the European Collection of Cell Cultures (Salisbury, Wiltshire, England), and the human monocyte cell line U937 was obtained from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured in RPMI-1640 (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (HyClone Laboratories, Inc., Logan, UT, USA), 100 units per ml penicillin, and 100 μg ml−1 streptomycin at 37 °C in an atmosphere of 5% CO2. To induce U937 differentiation into macrophage-like cells, 10 ng ml−1 phorbol 12-myristate 13-acetate (Sigma-Aldrich, St Louis, MO, USA) dissolved in dimethyl sulfoxide was added for 72 h, and lipopolysaccharide 1 μg ml−1 was added for an additional 48 h. In all, 1 × 106 cells from each cell line were placed in 10-cm dishes. After cell adhesion, the medium was changed to 10 ml per well serum-free RPMI, and incubated for 24 h. The medium was collected and the cellular debris was pelleted by centrifuging. The undiluted supernatant was used as the conditioned medium (CM) of each cell line.

Reagents

Recombinant human HB-EGF and anti-HB-EGF antibodies were purchased from R&D Systems (Minneapolis, MN, USA). Gefitinib was purchased from Cayman Chemical (Ann Arbor, MI, USA). Anti-EGF receptor antibody (D38B1) and anti-phosphorylated EGF receptor (pEGFR) antibody (Tyr1068, D7A5) were purchased from Cell Signaling Technology (CST; Danvers, MA, USA). Anti-CD68 antibody was purchased from Dako (Glostrup, Denmark). Anti-CD163 antibody was purchased from Leica Biosystems (Newcastle, UK). PD98059, LY294002, and U-73122 were purchased from Calbiochem (San Diego, CA, USA).

Clinical samples

The study protocol was approved by the Institutional Review Board at Kyushu University. A total of 93 MLS cases that were diagnosed between 1981 and 2012 were retrieved from the archives of the Department of Anatomical Pathology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan. In each case, MLS was diagnosed on the basis of histological features. All patients had been treated by marginal or wide resection. Tissues were collected during primary tumour biopsy or surgical resection; those obtained after radiation or chemotherapy treatment were omitted. Of the 93 cases identified, 15 were excluded, 4 because of lack of availability of adequate tissue, and 11 because of lack of follow-up data. Thus, 78 patients were included in the present study. Clinical data were obtained by reviewing patient records, and survival data were collected during the winter of 2013. Each case was evaluated according to the American Joint Committee on Cancer (AJCC) grading system.

Immunohistochemistry

Whole-section samples were fixed in 10% neutral buffered formalin and embedded in paraffin, and the sections were deparaffinised in xylene and rehydrated in ethanol. Antigen was retrieved by microwave pretreatment with 10 mM citrate buffer (pH 6.0) for CD68 and CD163, and 1 mM EDTA buffer (pH 8.0) for pEGFR. After peroxidase blocking with hydrogen peroxide, the sections were incubated with the following primary antibodies: anti-CD68 (1 : 100, Dako) and anti-CD163 (1 : 100, Leica Biosystems) 1 h at room temperature and anti-pEGFR (1 : 200, CST) overnight at 4 °C. EnVision System HRP Labelled Polymer (Dako) was used as a secondary antibody. The reaction products were visualized with 3-3 diaminobenzidine tetrahydrochloride and counterstained with haematoxylin. Assessment of macrophage number was performed by counting CD68-positive cells and CD163-positive cells in 10 random high power field (HPF) profiles. Expression of pEGFR was evaluated and samples were classified as negative when there was either no staining or weak staining in the cytoplasm and/or nucleus compared with the stromal elements, and as positive when there was moderate to strong staining. Images were acquired using an AX70 microscope equipped with a DP72 camera (both from Olympus Corp., Tokyo, Japan).

Double fluorescent immunohistochemistry

To identify the co-expression of CD68 and CD163, double fluorescent immunohistochemical staining was performed in 10 different MLS clinical samples. Tissue samples were deparaffinised, rehydrated, and antigen was retrieved by microwave with 10 mM citrate buffer (pH 6.0). After blocking with normal goat serum for 1 h, the samples were incubated with anti-CD68 (mouse IgG) and anti-CD163 (rabbit IgG, Santa Cruz) for 2 h. The reactions were visualised using secondary fluorescently labelled antibodies including goat anti-mouse Alexa Fluor 488 and goat anti-rabbit Alexa Fluor 546 (both from Invitrogen Corp.), mounted in Vectashield mounting medium containing DAPI (Vector Laboratories, Burlingame, CA, USA), and imaged using BZ-X710 fluorescence microscopy (Keyence Corp., Osaka, Japan).

Transwell migration assay

A migration assay was performed using Transwell chambers (Corning Costar Corp., Cambridge, MA, USA) as described previously (Kamura et al, 2010). In brief, MLS cell lines or RAW264.7 cells (1.0 × 105 per well) were suspended in 100 μl serum-free RPMI and seeded in the upper chambers. The lower chambers were placed in 600 μl serum-free RPMI containing CM or the indicated growth factors, antibodies. The indicated inhibitors were added to both upper and lower chambers. RAW264.7 cells were permitted to migrate for 6 h, and MLS cells migrated for 12 h. The cells that migrated to the lower side of the filter were fixed and stained using a Diff-Quik kit (Sysmex, Kobe, Japan), and counted by examining six fields per filter under a microscope.

Wound-healing assay

Wound-healing assays were performed as described previously (Kamura et al, 2010). Confluent cell monolayers in 35-mm dishes were wounded by scraping with a micropipette tip. The cells were washed and then cultured in serum-free RPMI containing CM or the noted reagents and inhibitors. The degree of wound closure was assessed by measuring the distance between the wound edges after 0 and 12 h.

Invasion assay

Invasion assays were performed using the Biocoat Matrigel invasion chamber (BD Biosciences, Bedford, MA, USA) as described previously (Harimaya et al, 2000). Myxoid liposarcoma cell lines (1.0 × 105 per well) were suspended in 100 μl serum-free RPMI and seeded in the Matrigel-coated upper chambers. The lower chambers were placed in 600 μl serum-free RPMI containing CM or the indicated growth factors and antibodies. The indicated inhibitors were added to both upper and lower chambers. After 24-h incubation, the filters were fixed and the number of cells was determined as described for the transwell migration assay.

Western blot analysis

Myxoid liposarcoma cells were stimulated with macrophage-conditioned medium (M-CM: RAW-CM and U937-CM) or HB-EGF 20 ng ml−1 for 5 min. To inhibit the stimulation, various concentrations of gefitinib and anti-HB-EGF antibody were added for 2 h before the stimulation. After stimulation, the cells were lysed using CelLytic (Sigma-Aldrich) with a protease and phosphatase inhibitor cocktail (Complete Mini, PhosSTOP; Roche Diagnostics, Mannheim, Germany). Western blot analysis was performed as described previously (Fujiwara-Okada et al, 2013; Iida et al, 2013) with the following primary antibodies: pEGFR (1 : 500), EGFR (1 : 1000), and β-actin (1 : 2000, Santa Cruz Biotechnology, Santa Cruz, CA, USA).

RNA preparation and PCR

Total RNA from MLS clinical samples and cell lines was extracted using the Rneasy Lipid Tissue Mini kit and Rneasy kit (Qiagen, Hilden, Germany), respectively. PCR was performed as described previously (Fujiwara et al, 2011). The primers are summarized in Supplementary Table S1.

Phospho-receptor tyrosine kinase analysis

Phosphorylated receptor tyrosine kinases (RTKs) of MLS cell lines after stimulation by M-CM were detected using the Human Phospho-RTK Array kit (R&D Systems) according to the manufacturer’s instructions. Myxoid liposarcoma cells were stimulated with or without M-CM for 5 min. The cell lysate was added to a nitrocellulose membrane that contained 49 different anti-RTK antibodies printed in duplicates, and incubated overnight at 4 °C. After washing, the membrane was incubated with anti-phospho-Tyrosine-HRP for 2 h at room temperature, and immunoreactivity was detected using Chemi Reagent Mix.

EGFR signalling analysis

Phosphorylated downstream signalling modules of EGFR after stimulation by HB-EGF were detected using the PathScan EGFR Signaling Antibody Array kit (CST) according to the manufacturer’s instructions. In brief, MLS cells were stimulated with or without HB-EGF (20 ng ml−1) for 5 min. The cell lysate was arrayed on a nitrocellulose-coated glass slide which antibodies were spotted in duplicates, and incubated overnight at 4 °C. After washing, the glass slide was incubated with a detection antibody cocktail for 1 h and HRP-linked streptavidin for 30 min at room temperature, and immunoreactivity was then detected using Chemi Reagent Mix.

Statistical analysis

Survival curves were calculated using the Kaplan–Meier method, and the log-rank test was used for survival analysis. Fisher’s exact test was used to compare categorized variables. The hazard ratios for risk factors for death were evaluated using the Cox proportional hazards regression model. The Student’s t-test was used for two-group comparisons. P<0.05 was considered as statistically significant. Data in graphs are given as mean±s.d. All data analysis was performed using a statistical software package (JMP9, SAS Institute Inc., Cary, NC, USA).

Results

Infiltrating macrophages into MLS associated with poor clinical outcome

To investigate whether infiltrating macrophages were associated with clinical outcome in patients with MLS, anti-CD68 antibodies were used to quantify the number of macrophages in 78 primary MLS clinical samples (Figure 1A). Clinicopathological data are summarized in Supplementary Table S2. Kaplan–Meier survival analysis was performed to determine the prognostic significance of macrophages, and showed that greater macrophage infiltration (⩾100 CD68 cells/10 HPF) was associated with poorer overall survival (P=0.0002, Figure 1B). Macrophages have a wide phenotypic diversity and can be classified into two activation phenotypes, M1 and M2 (Goerdt and Orfanos, 1999). Classically, M1 macrophages are pro-inflammatory and can exert cytotoxic activity. In contrast, M2 macrophages are anti-inflammatory and promote angiogenesis and tissue remodelling. Tumour-associated macrophages often exhibit features of M2 macrophages and produce a number of cytokines and growth factors that promote tumour progression. To evaluate the M2 phenotype in infiltrating macrophages, we performed immunostaining with antibodies against CD163, a marker of M2 macrophages (Figure 1A), as well as Kaplan–Meier survival analysis (Figure 1B). Co-expression of CD68 and CD163 was assessed by double fluorescent immunohistochemical staining (Figure 1C). Higher levels of CD163-positive macrophages were also associated with poorer overall survival (P=0.0213, Figure 1B). On the basis of these results, we suggest that a high percentage of the TAMs in MLS are M2 macrophages, and we hypothesized that MLS might have the ability to recruit macrophages that have a role in the malignant transformation in MLS. To investigate the potential mechanisms underlying recruitment of macrophages to MLS, a Transwell migration assay was performed. Myxoid liposarcoma cell line-conditioned medium (MLS-CM) significantly promoted cell migration of RAW264.7 cells, a murine macrophage cell line (Figure 1D), suggesting that the factors that promote macrophage recruitment were contained in MLS-CM. Next, expression profiles of cytokines and chemokines secreted by MLS cells were examined using the Luminex multiplex assay system. Remarkably, CCL2, CCL5, CXCL8, and VEGF, which are known to stimulate monocyte chemotaxis (Carr et al, 1994; Matsumoto et al, 2002), were secreted in MLS-CM (Supplementary Figure S1). Therefore, these MLS-secreted cytokines and chemokines are likely to be critical factors in recruiting macrophages into the tumour microenvironment of MLS.

Figure 1
figure 1

Association between macrophage infiltration and poor prognosis in MLS. (A) Immunohistochemical staining of human MLS sections using anti-CD68 and anti-CD163 antibodies. Upper panel shows cases with higher levels of macrophage infiltration (left, ⩾100 CD68 cells/10 HPF; right, ⩾80 CD163 cells/10 HPF), and lower panel demonstrates cases with lower levels of macrophage infiltration (left, <100 CD68 cells/10 HPF; right, <80 CD163 cells/10 HPF). Scale bars: 20 μm. (B) Kaplan–Meier survival curves for overall survival of all patients based on CD68-positive (upper) and CD163-positive (lower) macrophage infiltration. Log-rank test was performed to determine statistical significance, with P<0.05 defined as significant. (C) Co-expression of CD68 and CD163 in MLS clinical samples. Sections were immunostained for CD68 (green) and CD163 (red). Nuclei were visualized using DAPI (blue). Scale bar: 10 μm. (D) Migration of macrophages was examined using the Transwell system. The lower wells were filled with serum-free medium or medium conditioned by MLS1765 or MLS402-91, and RAW264.7 cell migration to the bottom surface of the Transwell was assessed. Results are presented as mean±s.d. ***P<0.001 vs control.

Macrophages upregulated the motility and invasion of MLS cells by phosphorylation of EGFR

To elucidate the role of TAMs in the development and progression of MLS, we first performed a cell proliferation assay to measure the effect of RAW264.7 and U937 macrophages on the growth of MLS cells. The presence of M-CM did not affect MLS cell growth (Supplementary Figure S2). Next, we investigated the effect of M-CM on motility of MLS cells via Transwell migration assay and wound-healing assay. M-CM significantly upregulated chemotaxis and chemokinesis of MLS cells (Figure 2A and B). We also found that M-CM significantly promoted MLS cell invasion (Figure 2C), an important result as cell motility is a critical factor in tumour cell invasion. One of the mechanisms known to upregulate cell motility is activation of the RTK pathway (Hubbard and Miller, 2007). We used a phospho-RTK antibody array to determine which RTKs were activated in M-CM-stimulated MLS cells. Of 49 MLS cell RTKs, only EGFR was phosphorylated by M-CM (Figure 2D), which was also confirmed by western blot (Figure 2E; Supplementary Figure S3). Gefitinib, a tyrosine kinase inhibitor of EGFR, dose-dependently reduced M-CM-induced EGFR phosphorylation and also significantly suppressed MLS cell motility (Figure 2E and F; Supplementary Figure S3). These data suggested that activation of EGFR by macrophages could have a critical role in MLS motility. To investigate the correlation between macrophage infiltration and activation of EGFR in MLS, MLS clinical samples were immunostained with anti-pEGFR (Figure 2G and H). Twenty-seven cases (35%) were positive and fifty-one cases (65%) were negative. The results of immunostaining with anti-CD68 and anti-pEGFR clearly demonstrated that greater macrophage infiltration was significantly associated with positive pEGFR (odds ratio, 8.07; 95% confidence interval, 2.25–28.97; P=0.0016) (Figure 2I). In addition, greater macrophage infiltration was also significantly associated with metastasis (odds ratio, 16.85; 95% confidence interval, 4.02–70.57; P=0.0001) and higher AJCC stage (odds ratio, 6.67; 95% confidence interval, 1.68–26.39; P=0.0072) (Supplementary Table S3).

Figure 2
figure 2

Macrophages promoted the motility and invasion of MLS cells by phosphorylation of EGFR. (A) The chemotaxis of MLS1765 and MLS 402-91 cells was assessed by Transwell migration assay when the lower chambers were filled with serum-free medium, RAW-conditioned medium (CM), and U937-CM. Data are depicted as mean±s.d. **P<0.01, ***P<0.001 vs control. (B) Wound-healing assay was performed to assess the chemokinesis of MLS cells treated with RAW-CM and U937-CM. The percentage of wound closure corresponds to the distance between the wound edges. Data are shown as mean±s.d. ***P<0.001 vs control. (C) Invasion of MLS cells in response to RAW-CM and U937-CM was assessed by invasion assay. Data are depicted as mean±s.d. ***P<0.001 vs control. (D) Human Phospho-RTK Array Kit used to screen phosphorylated receptor tyrosine kinase of MLS1765 cells activated by RAW-CM and U937-CM. P, positive control. (E) Macrophage-CM-induced phosphorylation of EGFR in MLS1765 cells was assessed by western blot analysis with anti-EGFR and anti-pEGFR antibodies. MLS1765 cells were pre-treated with gefitinib (0.1 μ M and 0.5 μ M) for 2 h. Actin is shown as a loading control. (F) The effect of gefitinib on macrophage-CM-induced motility of MLS1765 cells was assessed by Transwell migration assay. Gefitinib (0.1 and 0.5 μ M) was added to the upper and lower chambers. Data are shown as mean±s.d. ***P<0.001. (G, H) Immunohistochemically stained human MLS sections using anti-pEGFR antibody. Positive staining for pEGFR (G), and negative (H). Scale bars: 20 μm. (I) Association between macrophage infiltration and expression of pEGFR in MLS clinical samples was analysed statistically. **P<0.01 (Fisher’s exact test); OR, odds ratio; CI, confidence interval.

HB-EGF secreted by macrophages was a key ligand in activating EGFR of MLS cells

Another important issue was identifying which EGFR ligands were dominant in M-CM-induced motility of MLS cells. The major ligands of EGFR are EGF, HB-EGF, transforming growth factor-α (TGFα), amphiregulin (AREG), epiregulin (EREG), epigen (EPGN), and betacellulin (BTC). Of these, HB-EGF and AREG were commonly expressed in RAW264.7 and U937 (Figure 3A). Importantly, HB-EGF significantly upregulated MLS cell motility (Figure 3B and C) and invasion (Figure 3D). In contrast, AREG did not promote MLS cell motility (Supplementary Figure S4). Anti-HB-EGF-neutralizing antibody inhibited the M-CM-induced motility of MLS cells and reduced their M-CM-induced EGFR phosphorylation (Figure 3E–G; Supplementary Figure S5). In contrast, anti-AREG neutralizing antibody had no effect on M-CM-induced cell motility (Supplementary Figure S4). Taken together, these results showed that macrophage-secreted HB-EGF was the key ligand in activating EGFR of MLS cells, and it strongly enhanced MLS cell motility and invasion. In addition, the expression of EGFR ligands in MLS clinical samples was evaluated by RT-PCR (Figure 3H). Among 10 MLS clinical samples, HB-EGF was the most commonly expressed ligand (90%), suggesting a clinical role of HB-EGF in the progression of MLS.

Figure 3
figure 3

HB-EGF secreted by macrophages was the key ligand in activating EGFR of MLS cells. (A) The expression profiles of EGFR ligands in RAW264.7 and U937 cells were detected by RT-PCR. GAPDH was used as an internal control. P, positive control; N, negative control. (B) The chemotaxis of MLS1765 and MLS402-91 cells was assessed by Transwell migration assay with various concentrations of HB-EGF added to the lower chamber. Data are depicted as mean±s.d. ***P<0.001 vs HB-EGF 0 ng ml−1. (C) Wound-healing assay was performed to assess the chemokinesis of MLS cells treated with HB-EGF (20 ng ml−1) for 12 h. Data are shown as mean±s.d. **P<0.01, ***P<0.001. (D) Invasion of MLS cells in response to HB-EGF (20 ng ml−1) was assessed by invasion assay. Data are depicted as mean±s.d. ***P<0.001. (E) The effect of anti-HB-EGF–neutralizing antibody on macrophage-CM-induced motility of MLS1765 cells was assessed by Transwell migration assay. Anti-HB-EGF antibody (10 μg ml−1) or normal IgG was added to lower chamber. Data are shown as mean±s.d. **P<0.01, ***P<0.001. (F) The effect of anti-HB-EGF-neutralizing antibody on macrophage-CM-induced phosphorylation of EGFR in MLS1765 cells was assessed by western blot analysis with anti-EGFR and anti-pEGFR antibodies. Actin was used as a loading control. (G) Invasion assay was performed to assess the effect of anti-HB-EGF-neutralizing antibody on macrophage-CM-induced invasion. Data are depicted as mean±s.d. *P<0.05, **P<0.01, ***P<0.001. (H) The expression profiles of EGFR ligands from 10 MLS clinical samples were detected by RT-PCR. GAPDH was used as an internal control. P, positive control.

The PI3K/Akt pathway was the key downstream signalling pathway in HB-EGF-induced motility of MLS cells

Activated EGFR stimulates a wide range of signalling modules, including PI3K/Akt, mitogen-activated protein kinase (MAPK) cascade, phospholipase C (PLC) γ, and signal transducer and activator of transcription 3 (STAT3). The PathScan EGFR Signalling Antibody Array was used to accurately detect the activated EGFR-stimulated downstream signalling modules in HB-EGF-induced MLS cells. HB-EGF stimulated EGFR, MEK1/2, Erk1/2, Akt, and PLCγ (Figure 4A). STAT3 was not phosphorylated by HB-EGF stimulation. To identify the dominant downstream EGFR signalling pathway required for HB-EGF-induced cell motility, we used various pharmacological inhibitors, including LY294002 for PI3K/Akt, PD98059 for MEK and Erk, and U-73122 for PLCγ. LY294002, but not PD98059 or U-73122, remarkably suppressed HB-EGF-induced migration of MLS cells (Figure 4B). These results confirmed that the HB-EGF-EGFR-PI3K/Akt axis was the dominant pathway involved in the HB-EGF-induced motility of MLS cells.

Figure 4
figure 4

The PI3K/Akt pathway was a key downstream signalling pathway in HB-EGF-induced motility of MLS cells. (A) Analysis of EGFR-stimulated intracellular signalling modules using PathScan EGFR Signalling Antibody Array system. MLS1765 and MLS402-91 cell lysates either treated or not treated with HB-EGF (20 ng ml−1) for 5 min were assessed. Each antibody was spotted in duplicate. (B) The effects of each inhibitor on HB-EGF-induced motility of MLS1765 cells were assessed using a Traswell migration assay with HB-EGF (20 ng ml−1). HB-EGF was added to the lower chambers and each inhibitor was added to upper and lower chambers at the same time. Data are shown as mean±s.d. ***P<0.001, #not significant.

Macrophage infiltration is a prognostic factor in MLS

Finally, we investigated in detail the prognostic significance of macrophage infiltration into MLS. Univariate and multivariate analyses were performed using the following variables: number of CD68 cells, age, sex, location, tumour size, and round cell component (Table 1). Phosphorylated EGFR, metastasis, and AJCC stage were excluded because of their strong association with the number of CD68 cells. In univariate analysis, the number of CD68 cells and sex were significantly associated with poor prognosis. Multivariate analysis revealed that only the number of CD68 cells was an independent and significant factor for poor prognosis.

Table 1 Results of univariate and multivariate analyses for overall survival

Discussion

Metastasis occurs in up to one-third of MLS cases, and causes unfavourable outcomes. Tumour-associated macrophages have important roles in tumour invasion and metastasis (Condeelis and Pollard, 2006), and infiltration of TAMs into tumours results in tumour progression and poor prognosis. The presence of a high number of TAM is associated with various types of tumours, including bone and soft tissue tumours (Komohara et al, 2013). We previously reported that higher TAM infiltration was associated with poor prognosis in Ewing sarcoma (Fujiwara et al, 2011). In this study, we showed that greater TAM infiltration was associated with poor prognosis in MLS (Figure 1B), and its prognostic importance was confirmed in multivariate analysis (Table 1). On the basis of these results, TAMs are potentially a new prognostic factor for MLS. To our knowledge, this is the first report demonstrating an association between poor MLS prognosis and the role of TAMs.

In order for metastasis to occur, tumour cells must invade the extracellular matrix to separate from the primary lesion, intravasate to circulate towards distant organs, and then extravasate and engraft. In this study, we showed that macrophages promoted cell motility and invasion of MLS cells (Figure 2A–C). Furthermore, greater TAM infiltration was significantly correlated with the presence of metastasis (Supplementary Table S3). These results suggest that TAMs may have an important role in invasion and metastasis of MLS, especially during the first step involving dispatch from the primary lesion and intravasation. On the contrary, roles of TAMs in the later stages in MLS metastasis, such as during extravasation and homing to target organs, remain obscure. Thus further investigation, including in vivo experiments, is required.

It will be beneficial and interesting to define the mechanisms involved in TAM-induced motility and invasion of MLS cells. One possible mechanism is activation of the RTK pathway in MLS, since various RTKs, such as MET, IGF-1R, PDGFR, and EGFR, have been identified in MLS (Cheng et al, 2009; Negri et al, 2010). In MLS cells in this study, only EGFR, and not MET, IGF-1R, or PDGFR, was phosphorylated by M-CM in vitro (Figure 2D). Importantly, greater TAM infiltration in MLS was associated with higher expression of pEGFR in clinical samples (Figure 2I). These results offer solid evidence that TAMs enhance invasion and metastasis of MLS by activating the EGFR signalling pathway.

Identifying the key factors involved in TAM-induced activation of EGFR in MLS is the next important task that should be undertaken. Heparin-binding EGF-like growth factor is a member of the EGF family of molecules, which has a role in wound healing, cardiac hypertrophy, and heart development and function (Nanba and Higashiyama, 2004). In cancer, HB-EGF can contribute to tumour aggressiveness, including tumour proliferation, invasion, and metastasis, and there is growing evidence of the association between HB-EGF expression and poor cancer prognosis (Miyamoto et al, 2006). High expression of HB-EGF has also been detected in some soft tissue sarcomas and it is significantly associated with poor overall survival (Hoffmann et al, 2009). Furthermore, recent reports have shown that HB-EGF secreted by macrophages is a key factor in tumour progression in breast cancer (Vlaicu et al, 2013) and colon cancer (Rigo et al, 2010). In this study, we showed that HB-EGF enhanced migration and invasion of MLS cells via strong activation of EGFR (Figure 3B–F). On the basis of our findings and those of previous reports, we propose that HB-EGF is an important factor in TAM-induced invasion, metastasis, and activation of the EGFR pathway. Of the EGFR-stimulated downstream signalling modules in MLS, HB-EGF activated MEK1/2, Erk1/2, Akt, and PLCγ, but not STAT3 (Figure 4A). In this study, we confirmed that PI3K/Akt was the key pathway in MLS (Figure 4B). This result corresponded the previous findings that activation of PI3K/Akt pathway is associated with tumour survival, invasion, and poor prognosis in a variety of tumours (Engelman, 2009). In MLS, Demicco et al (2012) demonstrated that activation of the PI3K/Akt pathway was significantly associated with round cell change. Thus, HB-EGF secreted by TAMs in MLS might be one of the critical factors in activating the PI3K/Akt pathway that drives round cell transformation of MLS cells, therefore resulting in poor prognosis. This notion is intriguing, however, we were not able to confirm the association between macrophage infiltration and the round cell component in this study. Therefore, further studies are required to better understand the role of TAMs during the progression from myxoid to round cells.

From the findings of this study, we consider TAMs and HB-EGF to be potential molecular targets for treatment of MLS. In particular, CRM197, a non-toxic mutant of diphtheria toxin and a specific inhibitor of HB-EGF, has shown anticancer efficacy in ovarian cancer both in vitro and in vivo, and clinical trials are now underway. Because this study confirmed the expression of HB-EGF in clinical samples of MLS as well as the biological effects of HB-EGF on malignant phenotypes of MLS cells (Figure 3), MLS could be a promising candidate for anti-HB-EGF agents.

In conclusion, the present study showed that macrophage infiltration predicted poor prognosis in MLS, and a possible underlying mechanism is that HB-EGF secretion by macrophages enhanced MLS cell motility and invasion by activating EGFR. The HB-EGF-EGFR-PI3K/Akt downstream signalling pathways had a dominant role in TAM-induced cell motility and invasion of MLS. Tumour-associated macrophages, HB-EGF, EGFR, and PI3K/Akt could be new candidates for therapeutic targets of MLS and thus further investigation is warranted.