Int. J. Cancer: 125, 1276–1284 (2009) ' 2009 UICC
Stromal MCP-1 in mammary tumors induces tumor-associated macrophage infiltration and contributes to tumor progression Hiroshi Fujimoto1,2, Takafumi Sangai1,2, Genichiro Ishii1, Akashi Ikehara1, Takeshi Nagashima2, Masaru Miyazaki2 and Atsushi Ochiai1* 1 Pathology Division, Research Center for Innovative Oncology, National Cancer Center Hospital East, Kashiwa-City, Chiba, Japan 2 Department of General Surgery, Graduate School of Medicine, Chiba University, Chiba, Japan There is growing evidence that tumor-associated macrophages (TAMs) promote tumor growth and dissemination. Many individual reports have focused on the protumor function of molecules linked to the recruitment of macrophages, but little is known about which factor has the strongest impact on recruitment of macrophages in breast cancer. To elucidate this question, we performed RT-PCR using species-specific primers and evaluated tumoral and stromal mRNA expression of macrophage chemoattractants separately in human breast tumor xenografts. The correlation between the tumoral or stromal chemoattractant mRNA expression including monocyte chemoattractant protein-1 (MCP1) (CCL2), MIP-1a (CCL3), RANTES (CCL5), colony-stimulating factor 1, tumor necrosis factor a, platelet-derived growth factor (PDGF)-BB and macrophage infiltration were compared. There was significant positive correlation between stromal MCP-1 expression and macrophage number (r 5 0.63), and negative correlation between tumoral RANTES expression and macrophage number (r 5 20.75). However, no significant correlation was found for the other tumoral and stromal factors. The interaction between the tumor cells and macrophages was also investigated. Tumor cell–macrophage interactions augmented macrophagederived MCP-1 mRNA expression and macrophage chemotactic activity in vitro. Treatment of immunodeficient mice bearing human breast cancer cells with a neutralizing antibody to MCP-1 resulted in significant decrease of macrophage infiltration, angiogenetic activity and tumor growth. Furthermore, immunohistochemical analysis of human breast cancer tissue showed stromal MCP-1 had a significant correlation with relapse free survival (p 5 0.029), but tumoral MCP-1 did not (p 5 0.105). These findings indicate that stromal MCP-1 produced as a result of tumor– stromal interactions may be important for the progression of human breast cancer and macrophages may play an important role in this tumor–stroma interaction. ' 2009 UICC
(PDGF).12 There are both clinical and experimental evidence, which suggests that these molecules are associated with tumor progression and poor prognosis of the associated cancers.13 Although there are many reports describing the functions of each molecule, little is known about which factor has the strongest impact on recruitment of macrophages in vivo. In addition, most these studies have focused on the molecule produced by the tumor cells without paying attention to the influence of stromal cells. To clarify these questions, human breast cancer xenografts were analyzed by RT-PCR using species-specific primers and separately evaluated for tumoral and stromal mRNA expression of macrophage chemoattractants. We developed an in vitro coculture model and an in vivo treatment model to investigate the interaction between the tumor cells and macrophages. Finally, these results were confirmed by immunohistochemical analysis in human breast cancer.
Key words: macrophage; breast cancer; stroma; monocyte chemoattractant protein-1 (MCP-1); tumor-associated macrophages (TAMs)
Animal experiments Six to 8-week-old female severed combine immune deficient (SCID) mice were purchased from CLEA JAPAN (Tokyo, Japan). All animals were maintained and treated in accordance with institutional guidelines under approved protocols. Four human breast cancer cell lines (5 3 106 cells each) were injected into the intramammary tissue of the SCID mouse as a suspension in 200 lL of serum-free medium. The mice underwent estrogen replacement using 0.72 mg slow-release pellets (Innovative Research of America, FL) or a sham operation. Mice were sacrificed at the indicated time points after inoculation of the tumors. All tumors were excised and the diameter of each was measured using calipers. Tumor volume (TV) was calculated with following formula: TV 5 length 3 width 3 depth 3 0.52 (mm3).
Tumors consist of both tumor cells and stromal cells surrounded by an extracellular matrix. Recent studies have shown that interactions between tumor and stromal cells create a unique microenvironment which is essential for tumor progression.1,2 Tumor-associated macrophages (TAMs) are recruited to most tumors and represent one of the major components of the tumor stroma. The role of macrophages in tumor growth and development is complicated and wide ranged. Although activated macrophages may have anti-tumor activity, there is contrasting evidence that tumor cells escape from the tumoricidal activity of TAMs.3 There are some reports that TAM infiltration has been shown to correlate with angiogenesis and poor prognosis in breast cancer patients.4,5 In animal models, removal of macrophages by genetic mutation reduced the rate of tumor progression and metastasis.6 TAMs are recruited from circulating monocytes into tissues in response to chemoattractants and interact with tumor cells to make an environment that has abundant chemoattractants and growth factors.7 These include monocyte chemoattractant protein1 (MCP-1/CCL2),8,9 macrophage inflammatory protein 1a (MIP1a/CCL3), regulated on activation, normal T expressed and secreted (RANTES/CCL5), colony-stimulating factor 1,10 tumor necrosis factor a (TNF-a)11 and platelet-derived growth factor Publication of the International Union Against Cancer
Material and methods Cell culture Four human breast cancer cell lines: MCF7, T47D, MDA-MB231 (estrogen dependent) and MDA-MB-468 (estrogen independent) were used in this study. All cell lines were obtained from the American Type Culture Collection (Manassas, VA). Cells were cultured in RPMI1640 medium containing 10% heat-inactivated fetal bovine serum (Sigma-Aldrich, St. Louis, MO) at 37°C in humidified air under 5% CO2. On the days of inoculation, the cells were harvested by incubation with 0.5% trypsin and washed in PBS twice before injection in vivo. The concentration of the cells was determined with a hemocytometer.
Additional Supporting Information may be found in the online version of this article. Grant sponsor: The Ministry of Health and Welfare of Japan. *Correspondence to: Pathology Division, Research Center for Innovative Oncology, National Cancer Center Hospital East, 6-5-1 Kashiwanoha, Kashiwa-City, Chiba 277-8577, Japan. Fax: 81-4-7134-6865. E-mail:
[email protected] Received 4 May 2008; Accepted after revision 22 January 2009 DOI 10.1002/ijc.24378 Published online 18 February 2009 in Wiley InterScience (www. interscience.wiley.com).
STROMAL MCP-1 IN MAMMARY TUMORS
All tumors were fixed in periodate-lysine-4% paraformaldehyde0.05% glutaraldehyde, pH 7.4 for 6 hr at 4°C and then embedded in paraffin wax. Sections of 4 lm thickness were prepared for immunohistochemistry. Immunohistochemical staining TAMs were stained with an anti-mouse F4/80 antibody (Serotec, Oxford, UK; 1:100 dilution) and with an anti-human CD68 monoclonal antibody (PG-M1, DAKO, Copenhagen, Denmark; 1:100 dilution). CD11b and Gr-1 expression was investigated using an anti-mouse CD11b rat monoclonal antibody (Serotec; 1:50 dilution) and an anti-mouse Gr-1 rat monoclonal antibody (R&D Systems, Minneapolis, MN; 1:50 dilution). Microvessel densities (MVDs) were counted by scanning sections stained with anti-mouse CD31 (platelet endothelial cell adhesion molecule 1) rat monoclonal antibody (BD PharMingen, San Diego, CA; 1:50 dilution). MCP-1 staining was carried out using anti-mouse MCP-1 polyclonal goat antibody (R&D systems; 1:50 dilution) and anti-human MCP-1 monoclonal antibody (R&D Systems; 1:20 dilution). After deparaffinization and rehydration, sections were pretreated by using either a microwave-based antigen retrieval technique for 20 min with 10 mmol/l citrate buffer, at pH 6.0 and 90°C (F4/80), or a trypsin (0.005%) antigen retrieval technique for 5 min at 37°C in a humidified chamber (CD11b, CD31, CD68, Gr-1, MCP-1) and incubated with antibody at 4°C overnight. Endogenous peroxidases were inactivated with 3% H2O2 in methanol. Immunohistochemical staining was done with DAKO EnVision Plus System HRP. The reacted products were stained with diaminobenzidine. Immunofluorescene staining and confocal microscopy Double immunofluorescence analysis was performed as previously described.14 For immunofluorescence staining, we used the rat anti mouse CD11b:Alexa Fluor 488 (Serotec; 1:50 dilution) and the rat anti mouse Gr-1:Alexa Fluor 647 (Serotec; 1:50 dilution). Immunoquantification Immunohistochemical assessments were conducted by two investigators (HF and GI) who were completely blinded from any clinical information. The number of macrophages and Cd11b1 Gr-11 cells were assessed in the 3 areas of densest cell infiltration of the maximum cross sections with the use of a square micrometer eyepiece (counting grid, 10 mm square divided into 25 (2 mm) squares) at 2003 magnification following a brief scan of the maximum cross sections at low power. The microvessels were counted at 1003 magnification in a manner similar to TAM. The mean count of three areas was calculated as the cell count (per mm2) and MVD (per mm2). In human breast cancer tissue, MCP-1 staining was detected in both tumor cells and stromal cells. The percentages of cells staining for MCP-1 were counted and grouped as ‘‘not stained’’ (0–25%), ‘‘slightly stained’’ (25–50%), ‘‘partially stained’’ (50–75%) and ‘‘diffusely stained’’ (75–100%). In the present study, ‘‘partially stained’’ and ‘‘diffusely stained’’ were defined as positive, and ‘‘not stained’’ and ‘‘slightly stained’’ were defined as negative. RT-PCR Total RNA was extracted from human mammary tumor xenografts using 1 ml of Trizol reagent (Life Technologies, Gaithersburg, MD) as described previously.15 RNA from the xenografts was reverse-transcribed into cDNA using TaKaRa RT Reagents kit (TaKaRa, Shiga, Japan). Quantitative RT-PCR was performed by a Smart CyclerTM (Cepheid, Sunnyvale, CA) with SYBR Premix Ex TaqTM (TaKaRa) using human or mouse mRNA specific primers, according to the manufacturer’s instructions. A volume of 1 ll of cDNA solution corresponding to 500 ng of total RNA was subjected to 40 PCR cycles using a 2 tempera-
1277
ture method (5 sec at 95°C, 20 sec at 60°C) in a 25 ll mixture containing 12.5 ll SYBR premix Ex Taq and 5 lM each of forward and reverse gene-specific primers. Primers are listed in supporting Information Table 1. The housekeeping GAPDH transcripts were used to allow normalization among the samples. Preparation of peritoneal macrophages Peritoneal cells were obtained 3 days after the intraperitoneal injection of 2 ll of 3% thioglycollate into each mouse. These cells were suspended in RPMI 1640 medium containing 50 U/ml penicillin and 100 U/ml streptomycin. The cells were incubated for 90 min at 37°C in a CO2 incubator to allow them to adhere to the plates. The medium was then removed and the nonadherent cells were removed by washing twice with prewarmed PBS solution. Conditioned media system The conditioned media was collected from the macrophage and cancer cell culture media using the following method. After macrophages were cultured for 48 hr, the plates were washed twice with sterile PBS. MDA-MB-231 or MCF7 in serum free RPMI1640 containing 0.1% BSA were added to the plates and incubated for 48 hr. Tumor cells and macrophages were prepared so that the total cell numbers were 3 3 106 cells in 3 ml of medium. After the incubation, the medium was collected and centrifuged at 14000g to remove cell debris. Chemotaxis assay Chemotaxis assays were performed using 8 lm pore size cell culture inserts within a 24-well-plate (BD falcon, Franklin Lakes, NJ). Peritoneal macrophages (5 3 104 cells/100 ll RPMI) were seeded into the upper well of the insert, whereas the lower well was filled 600 ll of various conditioned media. Neutralizing antibody to mouse MCP-1 (AF-479-NA: 1 lg/ml; R&D Systems) was added to block chemotaxis. Recombinant murine MCP-1 (479-JE: 10 ng/mL; R&D Systems) was used for the control. Chambers were incubated at 37°C in a CO2 incubator for 5 hr and cells that migrated through and adhered to the bottom of the filter were stained with Hematoxilin. The number of cells in 9 high-power fields were counted per well, and the results were expressed as the mean number of cells in the high power field. Each experiment was determined in triplicate. MCP-1 depletion with antibody and macrophage depletion with clodronate liposome A total of 5 3 106 MDA-MB231 cells were injected intramammary into a SCID mouse on day 0. To study MCP-1 depletion, neutralizing antibody to mouse MCP-1 or control goat IgG (AB108-C: 15 lg per mouse; R&D Systems) were given intravenously to the mice on days 2, 6, 10, 14 and 18. Systemic depletion of macrophages was accomplished as described by Van Rooijen and Sanders.16 A volume of 200 ll of clodronate liposome (Roche Diagnostics GmbH) was injected intravenously into the SCID mice on days 2, 5, 12 and 19. Control mice were injected with liposome containing PBS at the same time points. All mice were sacrificed and tumors were evaluated on day 21. In vitro growth effect on tumor cells MDA-MB-231 cells were resuspended at 1.5 3 105 cells per milliliter in RPMI with 10% FBS. Cell suspensions (2 ml per dish) were placed in a 3 cm dish and stimulated with neutralizing antibody to mouse MCP-1 (100 lg or 1 mg/ml) or rabbit IgG (1 mg/ml). Plates were incubated at 37°C in 5% CO2 for 48 hr. The cells were then cultivated and counted microscopically using a hemocytometer. Each experiment was performed in triplicate and compared.
1278
FUJIMOTO ET AL.
Assessment in human breast cancer tissue Human breast cancer tissue was obtained from surgical specimens of invasive ductal carcinomas submitted to the Pathology division, National Cancer Center Hospital East between January 1993 and August 1995. A total of 128 specimens were available for the study. All signed an Institutional Review Board-approved informed consent. The stage of cancer was assessed using the UICC classification. Clinical information was obtained from the medical records. All patients were Japanese women, ranging from age 28 to 77 years (mean, 53 years). All had a solitary lesion. A total of 62 patients were premenopausal and 66 patients were postmenopausal. A partial mastectomy had been performed in 21 and a modified radical mastectomy had been performed in 107. Axiliary lymph node dissection of level I, II and 6 III was carried out in all patients. Lymph node metastasis was observed in 72 patients. According to the evaluation of the risk category, the patients were offered adjuvant chemotherapy and/or hormonal therapy and radiotherapy after surgery, as described previously.17 None of the patients had received radiotherapy or chemotherapy before surgery, and 99 patients received adjuvant therapy. Of the 56 patients without nodal metastasis, 27 received no adjuvant therapy, 8 received tamoxifen alone, 8 received chemotherapy (cyclophosphamide, methotrexate and 5-fluorouracil, adriamycin and cyclophosphamide or oral uracil/tegafur), whereas 13 received chemotherapy plus tamoxifen. Of the 72 patients with nodal metastasis, 2 received no adjuvant therapy 4 received tamoxifen alone, 16 received chemotherapy and 50 received chemotherapy plus tamoxifen. MCP-1 status (positive or negative) was compared with macrophage infiltration (less than or at least 320 cells/mm2) and clinicopathological variables including tumor size (3 cm versus < 3 cm), nodal status (node positive or negative), menopausal status (premenopausal or postmenopausal), venous invasion (VI) (presence or absence of VI), lymphatic invasion (LI) (presence or absence of LI), histological grade (HG) 1, 2 or 3), estrogen receptor (ER) status (positive or negative) and progesterone receptor (PR) status (positive or negative). Values in parentheses are the mean cutoff values. Statistical analysis Statistical analysis was performed using SPSS 11.0J (SPSS, Chicago, IL). Comparisons were performed by means of a Student’s t-test. The relationship between mRNA expression and macrophage infiltration was assessed by linear regression analysis (Mann-Whitney U-test). A v2 test was used to assess the relationship among MCP-1 staining, macrophage infiltration and clinical and pathologic variables. Kaplan-Meier curves were plotted, and the difference between the 2 groups was analyzed by the log-rank test. Multivariate analyses were performed using the Cox proportional hazards model. Statistical significance was assumed when p < 0.05 was obtained. Results Macrophage infiltration in human mammary tumor xenografts To investigate the relationship between tumor cells and macrophage infiltration, we implanted four human breast cancer cell lines into mammary fat pads with or without estrogen pellets in mice. After 4 weeks, the mice were sacrificed and the tumors were removed and examined. Figure 1a—a shows immunohistochemical staining of F4/80 in the tumor. Large amounts of F4/80 positive cells were observed in the intratumoral and peritumoral lesion, and they occupied a large part of the stroma. Intratumoral macrophage infiltration was homogenous, but peritumoral macrophage infiltration was highly heterogenous in the same section. Thus, in this study, intratumoral macrophages were defined as TAMs. Recent research shows CD11b1Gr-11 cells have the potential to promote tumor growth.18 To investigate the involvement of
these cells, we also performed immunohistochemical and immunofluorescence analysis using CD11b and Gr-1 antibodies(Figs. 1b0 –1d0 ). CD11b and Gr-1 positive cells are mainly present in peritumoral lesions. The results are shown in Figure 1b. In estrogen-treated mice, MDA-MB-231 tumors had the largest number of infiltrating macrophages, followed by T47D, MDA-MB-468 and MCF-7. The MCF-7 and T47D tumors did not grow without estrogen. The MDA-MB-231 tumors from nonestrogen mice had a larger number of infiltrating macrophages than of the MDA-MB-231 tumor of estrogen-treated mice. However, the number of macrophages in the MDA-MB-468 tumor was not affected by estrogen treatment. CD11b1Gr-11 cells are present in much lower numbers than F4/80 positive cells. There is no significant difference in the number of CD11b1Gr-11 cells among cell lines. In this study, we focused on revealing the mechanism of recruitment of F4/80 positive macrophages which dominates the tumor stroma. Correlation between macrophage infiltration and tumor or stromal derived factor To identify the molecules responsible for macrophage infiltration in the human breast cancer xenografts, we investigated the expression of 6 chemoattractants and growth factors in tumor xenografts. Using a species-specific primer, we were able to separately evaluate the mRNA expression of human breast cancer cells and murine stromal cells. The correlation coefficients of all factors to the number of TAMs are listed in Table I. The correlation between the number of TAMs and the level of human and mouse MCP-1 is shown in Figures 1d and 1e. Mouse MCP-1 expression had a positive correlation to the level of TAMs (r 5 0.63, p 5 0.03). Human RANTES had a negative correlation to the level of TAMs (r 5 20.75, p < 0.01). In vitro interaction of tumor cells and macrophages Cancer stroma contains endothelial cells, fibroblasts and immune cells including macrophages. To investigate the source of MCP-1, we examined MCP-1 expression using immunohistochemical methods. MCP-1 expression was found in macrophages, fibroblasts and endothelial cells. Among these cells, MCP-1 staining was more evident for macrophages (Fig. 1c). In human breast carcinoma, there are some reports that the tumor cell population is outnumbered by TAMs, which can comprise more than 50% of the total tumor mass.19 Thus we focused on the tumor cell–macrophage interactions in the tumor microenvironment. To test whether these interactions changed the level of MCP-1 expression, MDA-MB-231 cells and macrophages were cocultured with a fixed number of cells (3 3 106) in the dish. The ratio of macrophages to MDA-MB-231 was then varied. Both human and mouse MCP-1 expression were measured after 48 hr of incubation using the species-specific primers. When macrophages or MDA-MB231 cells were cultured alone, MCP-1 expression was low. However, when macrophages were cocultured with MDA-MB231 cells, MCP-1 expression in macrophages was upregulated. Accordingly, as the number of macrophages to cocultured MDAMB-231 cells increased, MCP-1 expression in macrophages was significantly upregulated (Fig. 2a). To confirm whether MCP-1 secreted into the media containing the coculture of tumor cells and macrophages would increase the recruitment of macrophages, chemotaxis assays were performed (Fig. 2b). In fact, tumor cell-macrophage conditioned media stimulated macrophage chemotactic activity in comparison with the media collected from the culture of macrophages or tumor cells alone. As the percentage of macrophage to tumor cells increased, the corresponding chemotactic activity of macrophages was stimulated. The coculture medium of macrophages with MDA-MB-231 cells which had the highest macrophage density in the xenotransplanted tumor, showed higher
STROMAL MCP-1 IN MAMMARY TUMORS
1279
FIGURE 1 – Macrophage infiltration in mammary tumor xenografts. (a) Immunohistochemical staining of F4/80 (a0 ), CD11b (b0 ) and Gr-1(c0 ) in the MDA-MB-231 tumor. Bar: 100 lm. Immunofluorescent staining of CD11b (green) and Gr-1 (blue) in the MDA-MB-231 tumor (d0 ). T: tumor. (b) The number of F4/801 cells, CD11b1 cells and CD11b1Gr-11 cells infiltrated in MCF7, T47D, MDA-MB-231 and MDA-MB-468 xenografted tumors with or without E2 pellet obtained 28 days after inoculation (n 5 5). The number of tumor associated macrophages was quantified microscopically at 2003 magnification. MCF7 and T47D tumor did not grow without estrogen. Statistical significance compared with MCF7 with estrogen treatment is shown as *p < 0.05. MDA-MB-231 tumor without estrogen treatment has the highest macrophage infiltration. (c) Immunohistochemical staining of MCP-1 in the MDA-MB-231 tumor. Macrophage (arrow), fibroblast (blank arrowhead) and endothelial cells (arrowhead) showed a positive staining for MCP-1. Bar: 50 lm. (d) Correlation between the number of macrophages and the human (tumor cell) MCP-1 mRNA expression. (e) Correlation between the number of macrophages and the mouse (stroma) MCP-1 mRNA expression.
chemotactic activity than the medium with the MCF-7 cells which demonstrated the lowest macrophage density. Addition of murine MCP-1 neutralizing antibody to the media reduced the enhanced chemotactic activity of macrophages to the control levels.
Effect of MCP-1 on macrophage infiltration and tumor development of mammary tumor xenografts The growth effect of anti-MCP-1 antibodies on MDA-MB-231 cells in vitro was investigated and anti-MCP-1 antibody treatment did not change the growth of MDA-MB-231 cells.
1280
FUJIMOTO ET AL. TABLE I – CORRELATION BETWEEN EXPRESSION OF CHEMOATTRACTANTS mRNA EXPRESSION AND NUMBER OF TUMOR-ASSOCIATED MACROPHAGES
Chemoattractants
MCP-1 MIP1-a RANTES CSF-1 TNF-a PDGF-BB
Human Mouse Human Mouse Human Mouse Human Mouse Human Mouse Human Mouse
r
p-value
–0.26 –0.63 –0.47 0.38 –0.75 0.20 0.13 0.21 –0.22 0.39 –0.29 0.42
0.4227 0.0297 0.1193 0.2277 0.0052 0.5345 0.6881 0.5048 0.4949 0.2039 0.3611 0.8236
To investigate the effect of MCP-1 on in vivo macrophage infiltration and tumor development, MDA-MB-231 cells were implanted into the mammary fat pad in SCID mice and the mice were treated with anti-MCP-1 antibodies. A smaller number of macrophages were observed in the mice treated with anti-MCP-1 compared with the control mice (Figs. 3a and 3b). The TV of the control mice was 1.5 times larger than that of the anti-MCP-1 antibody-treated mice (Fig. 3c). Additionally, MVD, defined with a staining of CD31, was reduced in mice treated with anti-MCP-1 antibody (Fig. 3d). We then depleted the macrophages from the tumor xenografts by administration of clodronate liposomes to confirm the role of tumor associated macrophages. Immunohistochemical analysis showed higher numbers of macrophages infiltrated in the control tumor compared with the clodronate-treated tumor (Figs. 4a and
FIGURE 2 – Coculture of tumor cells and macrophages. (a) Human and mouse MCP-1 mRNA expression in a coculture of tumor cells and macrophages. ND (not detected), H (human), M (mouse) *p < 0.01. (b) Influence of the interaction between tumor cells and macrophages for the chemotaxis of macrophages. Quantitative determination of the number of macrophages migrating through the transwell membrane in response to conditioned medium from the coculture of cancer cells and macrophages in the lower chamber is depicted. The migration toward medium alone (basal migration) was subtracted. Results are shown as mean 6 SD (n 5 3). *p < 0.001.
STROMAL MCP-1 IN MAMMARY TUMORS
1281
FIGURE 3 – Effect of MCP-1 on macrophage infiltration and growth of mammary tumors. MDA-MB231 cells were injected intramammary into SCID mice. Control goat IgG and neutralizing mouse MCP-1 antibody were given intravenously to the mice. (a) Immunohistochemical images of F4/80 staining in a mouse treated with goat IgG antibodies and anti-MCP-1 antibodies (Bar: 100 lm). (b) Quantification of macrophage infiltration in control IgG and anti MCP-1 antibody. (*p 5 0.02). Macrophage infiltration was determined by counting the number of F4/ 80 -stained cells. (c) Quantification of mammary tumor xenograft volume in control IgG and anti MCP-1 antibody (*p < 0.001). (d) Microvessel density (MVD) in control IgG and anti MCP-1 (*p 5 0.0011). Microvessel density was determined by counting CD31-stained cells.
4b). Mouse MCP-1 mRNA expression in the xenograft was decreased in clodronate-treated mice (Fig. 4c). The TV was significantly smaller in the clodronate treated mice than in the control mice (Fig. 4d). Furthermore, depletion of macrophages resulted in decreases in MVD and mouse VEGF mRNA expression (Figs. 4e and 4f). Immunohistochemical staining of MCP-1 in human breast cancer tissue The clinical importance of MCP-1 expression for macrophage infiltration and tumor development was evaluated in human breast cancer tissue. Immunohistochemical analysis in breast cancer tissue showed that MCP-1 was expressed both in the tumor cells and the stroma (Fig. 5a). Monocytic cells, which were immunostained with CD68, were the major components of the stromal cells that expressed MCP-1. We assessed the staining of MCP-1 in the tumor cells and the stroma separately. Each staining of MCP-1 was compared with macrophage infiltration and with clinicopathological variables including tumor size, nodal status, menopausal status, VI, LI, HG, ER status and PR status. Although there was no significant correlation between tumoral MCP-1 expression and macrophage infiltration (p 5 0.123), there was a significant correlation between stromal MCP-1 expression and macrophage infiltration (p 5 0.006). Although tumoral MCP-1 did not have a significant correlation with any other clinicopathological variables, stromal MCP-1 had a significant correlation with lymphatic invasion (p 5 0.023). The prognostic significance of tumoral or stromal MCP-1 expression was then assessed. Univariate analysis showed significant prognostic values in stromal MCP-1 (p 5 0.029, Table II, Fig. 5c), tumor size (p < 0.001), nodal status (p < 0.001), HG (p < 0.001), ER status (p 5 0.016) and PR status (p 5 0.042). However, tumoral MCP-1 expression did not have significant correlation with relapse-free survival (p 5 0.105; Table II, Fig. 5b). All variables with p < 0.05 in the univariate analysis were included in the multivariate analysis. Stromal MCP-1 had an inde-
pendent prognostic significance after tumor size and nodal status (p 5 0.005; Table III). Discussion In this study, we elucidated that stromal MCP-1 is primarily responsible for recruiting macrophages into the tumor area and clarified that this effect is augmented by the interaction between the tumor cells and the macrophages in vitro and in vivo. In our study, none of the tumor-derived factors that were studied had positive correlations with macrophage infiltration. Only the stromal MCP1 had a positive correlation. Several studies have reported that tumor-derived MCP-1 correlates with enhanced tumor progression and angiogenesis.20,21 However, these studies used cancer cell lines which highly express MCP-1 mRNA. Overexpressed tumoral MCP-1 may also contribute to macrophage recruitment, but our data showed that stromal MCP-1 mainly contributes to macrophage recruitment rather than tumoral MCP-1. We determined that the high density of macrophages in coculture increased MCP1 mRNA expression. This result indicates that once macrophages are recruited into the tumor environment, the stromal MCP-1 induced through tumor cell–macrophage interactions recruits additional macrophages. There are many reports that MCP-1 plays an important role in tumor progression. However, based on several in vitro studies, the current belief is that MCP-1 does not promote proliferation or prosurvival characteristics in mammary tumor cells.22 One of the important roles of MCP-1 in tumor progression is through its involvement in angiogenesis. There are 2 ways for MCP-1 to promote tumor angiogenesis. First, MCP-1 can act directly on endothelial cells to promote angiogenesis. Salcedo et al.22 has reported that treatment of immunodeficient mice bearing mammary tumors with anti-MCP-1 antibodies resulted in a significant increase of survival and inhibition of lung metastases. They showed that these results were attributed to the direct angiogenic effect of MCP-1. There have been similar reports for murine sarcomas23 whereas
1282
FUJIMOTO ET AL.
FIGURE 4 – Effect of macrophage depletion on macrophage infiltration and growth of mammary tumors. Effect of macrophage depletion on the growth of tumor xenografts was investigated using liposome injection. MDA-MB 231 cells were injected intramammary into SCID mice. Control liposomes and clodronate liposomes were given intravenously to the mice. (a) Immunohistochemical images of F4/80 staining in the control liposome and clodronate liposome-treated samples (Bar: 100 lm). (b) Quantification of macrophage infiltration (*p < 0.0001). (c) Mouse MCP-1 mRNA expression in control liposome and clodronate liposome-treated samples (*p 5 0.0017). (d) Quantification of tumor xenograft volume in control liposome and clodronate liposome-treated samples (*p 5 0.0027). (e) Microvessel density (MVD) in control liposome and clodronate liposome-treated samples (*p 5 0.0002). (f) Mouse VEGF mRNA expression in control liposome and clodronate liposometreated samples (*p 5 0.0032).
other reports showed that hemangiendothelioma in the presence of highly expressed MCP-1, can induce angiogenesis by autocrine disruption.24 Second, MCP-1 can increase the amount of TAM in mammary tumors, followed by the release of a wide array of proangiogenic factors and enzymes. Two distinct polarization states have been described for macrophages: the M1 and M2 macrophage.25 The M1 type is proinflammatory and characterized by the release of inflammatory cytokines and microbicidal/ tumoricidal activity. M2 macrophages, by contrast, have an immunosuppressive phenotype and are associated with tissue remodeling and angiogenesis. A number of studies have shown that TAMs predominantly exhibit an M2-like phenotype. There is now growing evidence that the macrophage can switch phe-
notype depending on the stage of tumor development. TAMs can produce a series of angiogenic factors including VEGF, FGF2, MMP9 and TP.13 In our study, administration of anti-MCP-1 antibodies reduced the mammary tumor growth, angiogenesis and macrophage infiltration. Furthermore, selective depletion of macrophages by administration of clodronate liposomes suppressed the mammary tumor growth, angiogenesis and stromal MCP-1 and VEGF mRNA expression. These results support the idea that macrophages are a major source for stromal MCP-1 which in turn, promotes macrophage infiltration and tumor growth. Gazzaniga et al.26 also reported that selective depletion of TAMs reduced tumor growth and angiogenesis in melanomas. It is difficult to distinguish the direct effect of MCP-1 from indirect studies on tumor angiogenesis.
1283
STROMAL MCP-1 IN MAMMARY TUMORS TABLE II – UNIVARIATE ANALYSIS FOR RFS IN BREAST CANCER PATIENTS Prognostic indicator
No. of patients (n 5 128)
v2
p value
Tumoral MCP-1 (2 vs. 1) Stromal MCP-1 (2 vs. 1) Tumor size (3 cm vs. 3 cm 0.001> 0.778 0.117 0.078 0.001> 0.016 0.042
TABLE III – MULTIVARIATE ANALYSIS FOR RFS IN BREAST CANCER PATIENTS
FIGURE 5 – Immunohistochemical analysis of MCP-1 in human breast cancer. Immunohistochemical staining for MCP-1 was analyzed in samples of primary human breast cancer. Bar: 50 lm. (a) Tumor cells (arrow) and stromal cells (arrowheads) showed a positive staining for MCP-1 in breast cancer. (b) RFS according to cancer MCP-1 expression in breast cancer patients (p 5 0.104). (c) RFS according to stromal MCP-1 expression in breast cancer patients (p 5 0.029).
However, there is little doubt that MCP-1 is one of the key molecules which recruit macrophages and produce a microenvironment filled with angiogenic factors. Stromal MCP-1 expression may be one of the final outcomes of the tumor cell–stroma interaction. From the data obtained in our study, we could not find the trigger molecule responsible for stimulating MCP-1 expression in cancer stromal cells. There are reports that tumor-producing growth factors like PDGF, IL-1 and TNF-a induce MCP-1 gene expression in monocytes.27 These factors are associated with the malignant potential of human breast cancer.28–30
Prognostic indicator
Hazard
95% CI
p value
Stromal MCP-1 (2 vs. 1) Tumor size (3 cm vs. 3 cm