Metastasis proceeds through interaction between cancer cells and resident cells such as leukocytes and fibroblasts. An i.v. injection of a mouse renal cell carcinoma, Renca, into wild-type mice resulted in multiple metastasis foci in lungs and was associated with intratumoral accumulation of macrophages, granulocytes, and fibroblasts. A chemokine, CCL3, was detected in infiltrating cells and, to a lesser degree, tumor cells, together with an infiltration of leukocytes expressing CCR5, a specific receptor for CCL3. A deficiency of the CCL3 or CCR5 gene markedly reduced the number of metastasis foci in the lung, and the analysis using bone marrow chimeric mice revealed that both bone marrow- and non-bone marrow-derived cells contributed to metastasis formation. CCL3- and CCR5-deficient mice exhibited a reduction in intratumoral accumulation of macrophages, granulocytes, and fibroblasts. Moreover, intratumoral neovascularization, an indispensable process for metastasis, was attenuated in these gene-deficient mice. Intrapulmonary expression of matrix metalloproteinase (MMP)-9 and hepatocyte growth factor (HGF) was enhanced in wild-type mice, and the increases were markedly diminished in CCL3- and CCR5-deficient mice. Furthermore, MMP-9 protein was detected in macrophages and granulocytes, the cells that also express CCR5 and in vitro stimulation by CCL3-induced macrophages to express MMP-9. Intratumoral fibroblasts expressed CCR5 and HGF protein. In vitro CCL3 stimulated fibroblasts to express HGF. Collectively, the CCL3-CCR5 axis appears to regulate intratumoral trafficking of leukocytes and fibroblasts, as well as MMP-9 and HGF expression, and as a consequence to accelerate neovascularization and subsequent metastasis formation.
Both carcinomas and sarcomas arising anywhere in the body can spread to the lungs via the blood or lymphatics. The occurrence of pulmonary metastasis has profound adverse effects on the prognosis and the quality of life of the patients (1). Treatment for lung metastasis is in most cases palliative but not curative, particularly for those with multiple and/or disseminated foci. Thus, it is necessary to develop novel preventive and/or therapeutic measures based on the elucidation of molecular and cellular mechanisms underlying lung metastasis.
Metastasis occurs in multiple steps. Tumor cells degrade basement membrane, mainly with the use of matrix metalloproteinase (MMP)3 (2), to enter into bloodstream or lymphatics. Within the circulation, tumor cells tend to form emboli by interacting with blood cells, particularly platelets (3). Tumor emboli arrest and extravasate at distant sites, followed by egress through the basement membrane, with the use of MMPs and adhesion molecules (4, 5). Following entry into the tissue, tumor cells proliferate and develop a vascular supply based on the production of various angiogenic factors, including vascular endothelial growth factor (VEGF) (6, 7, 8), basic fibroblast growth factor (bFGF) (7, 9, 10), and hepatocyte growth factor (HGF) (11). Accumulating evidence indicates that MMPs and various angiogenic factors are produced by resident cells, including leukocytes, fibroblasts, and endothelial cells, in addition to tumor cells (12, 13, 14). Thus, the metastatic process is regulated by the coordinated interaction between tumor cells and host resident cells.
An i.v. injection of certain types of tumor cells into mouse tail veins can cause multiple lung metastases (3, 15, 16). This procedure recapitulates the process of lung metastasis following the entrance of tumor cells into the circulation and therefore is frequently used to elucidate the molecular and cellular mechanisms underlying lung metastasis. An i.v. injection of a mouse renal cell carcinoma cell line, Renca, can cause multiple lung metastasis foci (16, 17). We previously observed that macrophages accumulate in metastasis foci in this lung metastasis model (17). Moreover, in our preliminary experiments, the expression of CCL3, a chemokine with a potent chemotactic activity for macrophages, was progressively enhanced in lungs after Renca cell injection. CCL3 utilizes two distinct receptors, CCR1 and CCR5 (18, 19, 20). The important roles of CCL3 in lung pathology are supported by our previous observations that the deficiency of either the CCL3 or CCR5 gene markedly attenuated bleomycin-induced lung fibrosis, together with reduced intrapulmonary macrophage infiltration (21). Hence, we evaluate the roles of CCL3 and its receptors in lung metastasis induced by an i.v. injection of Renca cells. We provided herein definitive evidence on the crucial involvement of the CCL3-CCR5 axis, but not the CCL3-CCR1 interactions, in this lung metastasis model.
Materials and Methods
Abs and other reagents
Rabbit anti-mouse CCR5 polyclonal Abs (pAbs) were prepared as described previously (22). The following mAbs and pAbs were commercially obtained: goat anti-mouse CCR1 pAb and rabbit anti-mouse HGF pAb (Santa Cruz Biotechnology); goat anti-mouse CCL3, CCL4, and CCL5 pAbs, goat anti-mouse MMP-9 pAb (R&D Systems); rat anti-mouse F4/80 mAb and rat anti-CD3 mAb (Serotec); rat anti-mouse Gr-1 mAb, and rat anti-mouse CD31 mAb (BD Pharmingen); goat anti-human type I collagen (SouthernBiotech); Alexa Fluor 488-labeled donkey anti-rat IgG Ab, Alexa Fluor 546-labeled donkey anti-goat IgG Ab, Alexa Fluor 488-labeled donkey anti-goat IgG Ab, and Alexa Fluor 594-labeled donkey anti-rabbit Ab (Molecular Probes/Invitrogen). ELISA assay kits for CCL3, CCL4, VEGF, and HGF were obtained from R&D System.
Specific pathogen-free 8- to 10-wk-old female BALB/c mice were purchased from Charles River Laboratories, which are designated as wild-type (WT) mice hereafter. CCL3-deficient (CCL3KO) mice were obtained from The Jackson Laboratory. CCR5-deficient (CCR5KO) mice were generated as previously described (23). CCR1-deficient (CCR1KO) mice were a generous gift from Drs. P. M. Murphy and J.-L. Gao (National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD) (24). These gene-deficient mice were backcrossed with BALB/c for more than eight generations. All mice were kept under the specific pathogen-free conditions, and all of the animal experiments in this study complied with the Guidelines for the Care and Use of Laboratory Animals of Kanazawa University.
Induction of lung metastasis
A murine renal cell carcinoma, Renca (16, 17), was maintained in RPMI 1640 medium supplemented with 10% heat-inactivated FBS at 37°C in a humidified atmosphere with 5% CO2. Subconfluent cells were collected and resuspended in HBSS at a cell density of 2.5 × 105 cells/ml. The cell viability was always >95% by a trypan blue exclusion test. Two hundred microliters of cell suspensions was injected into the tail veins of mice. At the indicated intervals after tumor injection, lungs were removed for the determination of the numbers of tumor foci on the lung surfaces, lung weights, and the following analyses.
Generation of bone marrow (BM) chimeric mice
The following BM chimeric mice were prepared as described previously (21). BM cells were collected from the femurs and tibia of either WT or CCR5KO male mice, by aspiration and flushing, and were suspended in PBS at the concentration of 5 × 107/ml. Recipient WT or CCR5KO female mice were lethally irradiated to 8 Gy by using an x-ray irradiator (Hitachi, MBR-1520R). Then, the animals were given 1 × 107 BM cells i.v. under anesthesia. Thereafter, the mice were housed under specific pathogen-free conditions and were fed normal chow and autoclaved hyperchlorinated water for 60 days. To verify successful engraftment and reconstitution of the BM in recipient mice, genomic DNA was isolated from peripheral blood of each chimeric mouse 30 days after BM transplantation, with a NucleoSpin blood kit (Macherey-Nagel). Quantitative PCR was performed to detect the Sry gene present in the Y chromosome (forward primer, 5′-TGGGACTGGTGACAATTGTC-3′ and reverse primer, 5′-CTCTGTATTTTGCATGCTGG-3′) and GAPDH gene as an internal control with the help of an Applied Biosystems StepOne real-time PCR system and 2× QuantiFast SYBR Green PCR Master mix (Qiagen). We calculated the chimeric rates on the assumption that the ratio of Sry to GAPDH gene was 100% in male mice. We confirmed that the chimeric rates of CCR5KO to WT and WT to CCR5KO were 84.1 ± 14.9% and 80.1 ± 12.1%, respectively. After BM reconstitution was confirmed, mice were i.v. injected with Renca cells as described above.
The lung tissues were fixed in 10% formalin buffered with PBS (pH 7.2) and embedded in paraffin. Five-micrometer-thick sections were stained with H&E. Hisotopathological changes were evaluated by an examiner without a prior knowledge of the experimental procedures. Immunohistochemical analysis was also performed using anti-CCL3, anti-CCR1, anti-CCR5, anti-Gr-1, anti-F4/80, anti-CD3, or anti-MMP-9 Abs as described previously (21). The positive cell numbers were enumerated in 10 randomly chosen visual fields at ×400 magnification from three mice in each group. The incubation of isotype-matched control Ab did not give rise to any positive reaction. Frozen sections were prepared by embedding lung tissue in OCT compound (Sakura Finetec) and snap-frozen in liquid nitrogen. Then, 8-μm-thick sections were prepared and dried at room temperature until the sections were firmly adherent to the slides, followed by fixation in cold acetone for 10 min and stained with anti-CD31 or anti-type I collagen Ab. The immune complexes were visualized by a catalyzed signal amplification system (Dako) or Elite ABC kit and DAB substrate kit (Vector Laboratories) according to the manufacturers’ instructions. CD31-postive areas in the tumor tissue were defined as the intratumoral vascular areas (25). Areas of active neovascularization (hot spots) were found inside metastasis foci by scanning the section at a lower magnification, and the pixel numbers of CD31-positive areas were then determined on five randomly chosen fields in hot spots of each animal at ×400 magnification with the aid of Photoshop version 7.0. The density of neovascularization was expressed as a percentage of the whole metastasis area. Type I collagen-positive areas in tumor foci were also determined on five randomly chosen fields in hot spots of each animal at ×400 magnification with the aid of Photoshop version 7.0.
Double-color immunofluorescence analysis
After fixation in cold acetone, frozen sections were incubated with the combination of anti-MMP-9 and anti-F4/80, anti-MMP-9 and anti-Gr-1, anti-type I collagen and anti-CCR5, or anti-type I collagen and anti-HGF Abs overnight at 4°C. After washing with TBS-T, the sections were further incubated with Alexa Fluor 546-labeled donkey anti-goat IgG and Alexa Fluor 488-labeled donkey anti-rat IgG Abs or Alexa Fluor 594-labeled donkey anti-rabbit IgG and Alexa Fluor 488-labeled donkey anti-goat IgG Abs for 30 min at room temperature in the dark. Finally, the section was washed with TBS-T and immunofluorescence was visualized in a dual channel mode on a fluorescence microscope.
Isolation and culture of murine peritoneal macrophage
Eight- to 12-wk-old WT or CCR5KO female mice were injected i.p. with 2 ml of sterile 3% thioglycolate broth (Difco). Three days later, peritoneal exudate macrophages were harvested by lavage of the peritoneal cavity with 7–8 ml HBSS and centrifuged, washed, and plated in 6-well plates at a density of 1 × 106/ml in RPMI 1640 containing 10% FBS and incubated in a humidified incubator at 37°C in 5% CO2. After 1 h of attachment, the medium was aspirated and rinsed twice with PBS. The resultant cell population was judged to be composed of >95% macrophages by flow cytometry analysis using F4/80 Abs. The cells were then stimulated with the indicated concentrations of mouse CCL3 (R&D Systems) for 12 h to obtain total RNA for a quantitative real-time RT-PCR.
Isolation and culture of murine lung fibroblasts
Lung tissues obtained from 8- to 12-wk-old WT or CCR5KO mice were divided into small pieces before the digestion in HBSS containing 1 mg/ml collagenase and 2.5 mg/ml trypsin at 37°C on a shaking water bath. Three cycles of digestion were conducted each for 30 min. At the end of each cycle, the digested cell suspension was mixed with an equal volume of DMEM containing 10% FBS and the medium was replaced with fresh medium containing the same enzymes for the next cycle of digestion. Pooled cells were filtered through sterile gauze, centrifuged at 1000 rpm for 10 min at 4°C, and finally resuspended in 10% FBS-DMEM medium. After the cells were incubated in a humidified incubator at 37°C in 5% CO2 for 48 h, unattached cells were removed by washing, and fresh medium was added. Two days later when the cells became confluent, the cells were detached by the treatment with 0.25% trypsin for 2–3 min at 37°C and split at a ratio of 1:2 or 1:3. Cells were maintained in 10% FBS-DMEM and subcultured every 3–4 days. By the third passage, the cells formed homogeneous monolayers morphologically consistent with fibroblast-like cells and were confirmed to consist of >90% collagen type I-positive cells. Lung fibroblasts were used at a confluence and between passages three and six. Cells were stimulated with the indicated concentrations of mouse CCL3 for 12 h to obtain total RNA for a quantitative real-time RT-PCR. In another experiment, the cells were stimulated with the indicated concentrations of CCL3 for 72 h to determine HGF content in supernatants using a specific ELISA kit (R&D Systems).
Fibroblast migration assay
Cell culture insert wells with a 8-μm pore size filter (BD Biosciences) were precoated with 0.1% gelatin (Sigma-Aldrich). The indicated concentrations of CCL3 and 50,000 fibroblasts were placed in the lower and upper chambers, respectively, and incubated at 37°C for 18 h in 5% CO2. Then, after cells remaining on the upper surface of filters were mechanically removed, those that had migrated to the lower surface were fixed with methanol and stained with Giemsa solution (Merck). The cell numbers were counted at >10 randomly chosen fields for each filter.
Peritoneal macrophages were incubated in serum-free RPMI 1640 medium in the presence of the indicated concentrations of CCL3 for 24 h. Ten microliters of the resultant supernatants was subjected to zymography as described previously (26). The intensities of the resultant clear bands were measured as gelatinolytic activities with the help of NIH Image analysis software version 1.62.
RNA isolation and cDNA synthesis
Total RNAs were extracted from a fraction of lung tissue with RNA-Bee (Tel-Test) according to the manufacturer’s instruction. After the treatment with RNase-free DNase I (Promega), the RNA preparations were further purified with the use of TRIzol LS reagent (Invitrogen) according to the manufacturer’s instructions. For primary cultured peritoneal macrophages and lung fibroblasts, total RNAs were extracted using RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Two micrograms of total RNA was reverse-transcribed at 42°C for 1 h in 20 μl reaction mixture containing Moloney murine leukemia virus reverse transcriptase (Toyobo) and hexanucleotide random primers (Qiagen) to obtain cDNA.
Serially 2-fold diluted cDNA products were amplified for GAPDH using a specific set of primers (Table I) with 25 cycles of 94°C for 30 s, 58°,C for 30 s and 72°C for 1 min in 25 μl of reaction mixture containing Taq polymerase (Takara Bio) to evaluate the amount of the transcribed cDNA. Equal amounts of cDNA products were then amplified for the indicated genes using specific sets of primers (Table I) with 32 cycles of 94°C for 30 s, 55°C for 1 min, and 72°C for 1 min. The resultant PCR products were fractionated on 1.5% agarose gel and visualized by ethidium bromide staining under UV light transillumination. The band intensities were measured using NIH Image analysis software version 1.62, and the ratios to GAPDH were calculated on the assumption that the ratios of untreated animals are 1.0.
|Gene Name .||Forward .||Reverse .||Cycles .||Products (bp) .|
|Gene Name .||Forward .||Reverse .||Cycles .||Products (bp) .|
Quantitative real-time PCR
Real-time RT-PCR was performed on an Applied Biosystems StepOne real-time PCR system using 2× Quantifast SYBR Green PCR Master mix, 1 μM primers (Table II), and <100 ng cDNA in a 25 μl reaction mixture. Each target and standard GAPDH gene was analyzed in duplicate in three independent real-time RT-PCR assays. Thermal cycling was initiated with an initial activation step for 5 min at 95°C, followed by 40 cycles of 95°C for 10 s and 60°C for 30 s. Immediately after the amplification, melt curve protocols were performed to ensure that primer-dimers and other nonspecific products had been minimized. Relative expression of target gene was analyzed by the ΔΔCt method. The ratios of mRNA levels were expressed relative to those of the untreated group. Results are expressed as means ± SEM.
|.||Forward .||Reverse .|
|.||Forward .||Reverse .|
Determination of intrapulmonary cytokine levels
Lung tissues were obtained and homogenized with RIPA buffer (Santa Cruz Biotechnology) containing proteinase inhibitor cocktail (Hoffmann-La Roche) and centrifuged to obtain supernatants. After determining total protein contents with a BCA kit (Pierce), CCL3, CCL4, VEGF, or HGF levels were determined using specific ELISA kits against these cytokines (R&D Systems), according to the manufacturer’s instructions. The data are expressed as the target molecule (pg or ng) per total protein (mg) for each sample.
Data were analyzed statistically using one-way ANOVA followed by the Fisher protected least significant difference test or the Mann-Whitney U test. p < 0.05 was considered statistically significant.
Enhanced CCL3 and CCR5 expression by lung metastases
In line with our previous observation (17), an i.v. injection of Renca cells caused microscopic and macroscopic metastases to develop in lungs 7 and 14 days after the injection, respectively. We previously observed that macrophages accumulated in the lung metastases in this model (our unpublished data). These observations prompted us to evaluate the expression of macrophage-tropic chemokines CCL3, CCL4, and CCL5 in this process. Intrapulmonary CCL3 and CCL4 mRNA expression was increased by 14 days after the injection, whereas CCL5 mRNA expression was not augmented significantly (Fig. 1,A). We next evaluated intrapulmonary mRNA expression of CCR1 and CCR5, the receptors for CCL3 and CCL4. CCR5, but not CCR1, mRNA expression was enhanced in lungs 21 days after Renca cell injection (Fig. 1,A). Moreover, the intrapulmonary content of CCL3 and, to a lesser degree, CCL4 was increased after Renca cell injection (Fig. 1,B). Immunohistochemical analysis detected CCL3 proteins in resident mononuclear cells before the injection, and infiltrating inflammatory cells and some tumor cells after the injection (Fig. 1,C). Similar but weak staining pattern was observed for CCL4 but not for CCL5 (supplemental Fig. 1).4 Moreover, CCL3 deficiency abrogated increases in both the intrapulmonary contents of CCL3 and CCL4 (Fig. 1,B). These observations suggest that host resident cell-derived CCL3 is essential for additional CCL3 and CCL4 production observed after tumor cell injection. CCR1 proteins were detected mainly in the large vessels or bronchi, but the expression pattern did not change after Renca cell injection (Fig. 1,C). CCR5 proteins were detected in the large vessels and bronchi and some macrophages before Renca injection. As metastases formed, CCR5-positive cells appeared around and inside tumor foci, particularly in the tumor periphery (Fig. 1 C). These observations indicate that CCL3 and CCL4 were produced in foci of lung metastases and presumably the produced chemokines attracted CCR5-positive cells into the tumor metastases.
Reduced lung metastasis in CCR5KO and CCL3KO mice
To establish the pathogenic roles of CCL3 and its receptors, CCR5 and CCR1, in this lung metastasis model, we injected the same number of Renca cells i.v. into CCL3KO, CCR1KO, and CCR5KO mice. Compared with WT mice, CCL3KO and CCR5KO, but not CCR1KO, mice developed significantly fewer metastases 21 days after the injection (WT, 77.2 ± 10.47; CCR5KO, 41.75 ± 7.01; CCL3KO, 38.25 ± 10.01; CCR1KO, 70.25 ± 11.93: Fig. 2, A and B). Similar reductions were also observed for lung weights (Fig. 2,C). Histological analysis consistently demonstrated that pulmonary architecture was compressed by large numbers of metastatic tumors in WT mice 21 days after the injection (Fig. 2 D). In contrast, tumors were smaller in CCL3KO and CCR5KO mice and pulmonary structures were maintained 21 days after the injection. These observations indicate that the CCL3-CCR5, but not the CCL3-CCR1, axis was crucially involved in the development of lung metastases.
Contribution of both BM- and non-BM-derived cells to lung metastasis
CCR5 is expressed by non-BM-derived cells such as fibroblasts, in addition to macrophages and granulocytes (21). To address the contribution of non-BM- and BM-derived cells, we injected Renca cells i.v. into various BM chimeric mice. WT mice reconstituted with WT-derived BM developed similar numbers of tumor foci as did untreated WT mice (Fig. 2,E), thereby precluding the possibility of apparent effects of irradiation on morphology and functions of lungs. Both WT mice reconstituted with CCR5KO mouse-derived BM and CCR5KO mice reconstituted with WT mouse-derived BM developed fewer metastatic foci than did WT mice (Fig. 2 E). These observations suggest that both BM and non-BM cell-derived CCR5 cells contributed to the process of lung metastases in this model.
Reduced accumulation of inflammatory cells and fibroblasts in CCR5KO and CCL3KO mice
Because CCR5 is expressed by various types of leukocytes, including granulocytes, monocytes/macrophages, and T lymphocytes (22, 27, 28), we next evaluated the effects of CCL3 or CCR5 gene deficiency on the intratumoral infiltration of these cells. Gr-1-positive granulocytes and F4/80-positive macrophages started to infiltrate around tumor foci in WT mice by 7 days after Renca cell injection, and their numbers increased progressively thereafter (Fig. 3,A–C). The deficiency of either the CCL3 or CCR5 gene markedly reduced the numbers of intratumorally infiltrated granulocytes and macrophages following 7 days after Renca cell injection. CCR5KO mice appeared to have fewer leukocytes than CCL3KO mice, although we failed to detect significant differences between these two strains. CCL4, another CCR5 ligand, was produced at a low level in the lungs of CCL3KO mice (Fig. 1,B) and might be responsible for these differences. Moreover, a substantial proportion of type I collagen-positive cells in metastatic foci expressed CCR5 (Fig. 3,D), as previously reported (29). Furthermore, fibroblasts could migrate in response to CCL3 in vitro (Fig. 3,E). Hence, we determined the changes with time in the numbers of intratumoral collagen type I-positive fibroblasts. Fibroblasts were evident at tumor sites by 14 days after the tumor cell injection and increased thereafter (Fig. 3,F and supplemental Fig. 2A). The increases in fibroblast numbers were depressed in CCR5KO and CCL3KO mice by 17 and 21 days after the tumor injection (Fig. 3 G and supplemental Fig. 2A). Similar results were observed by using another fibroblast marker, α-smooth muscle actin (data not shown). These observations would indicate the essential involvement of the CCL3-CCR5 axis in the intratumoral accumulation of granulocytes, macrophages, and fibroblasts.
Because neovascularization is a prerequisite for metastasis (30), particularly in the growth of metastatic foci, we evaluated angiogenesis in the site of lung metastatic tumor foci by immunohistochemistry with anti-CD31. Intratumoral CD31-positive vascular areas became evident in WT mice by 7 days after Renca cell injection, increasing thereafter and reaching a maximal level by 17 and 21 days after the injection (Fig. 4,A and supplemental Fig. 3A). Moreover, vascularity was decreased in CCL3KO and CCR5KO mice by 17 and 21 days after Renca cell injection (Fig. 4,B and supplemental Fig. 3B). Of note, the increases in neovascularization coincided with those in intratumoral accumulation of granulocytes, macrophages, and fibroblasts. The intrapulmonary content of the major angiogenic factor, VEGF, was not increased after Renca cell injection (Fig. 5,A). In contrast, intrapulmonary mRNA expression of HGF, bFGF, and placenta-derived growth factor (PlGF) but not stromal cell-derived factor (SDF)-1 was significantly increased in WT mice after Renca cell injection (Fig. 5,B–E). Among these genes, HGF gene expression was markedly depressed in CCL3KO and CCR5KO mice as compared with WT mice. Similar reductions were observed in the intrapulmonary HGF contents (Fig. 5,F). Moreover, tumor angiogenesis depends on the activity of MMPs, particularly, MMP-2 and MMP-9, the MMPs with gelatinase activity (31, 32, 33, 34), in addition to angiogenic factors. Hence, we examined MMP-2 and MMP-9 mRNA expression by a quantitative RT-PCR analysis. MMP-9, but not MMP-2, mRNA was increased markedly by 17 days after Renca cell injection and the increases were depressed in CCR5KO and CCL3KO mice (Fig. 6, A and B). Furthermore, by an immunohistochemical analysis, MMP-9-positive cells became evident by 17 days after the tumor injection, and the numbers were decreased in CCR5KO and CCL3KO mice (Fig. 6 C and supplemental Fig. 4). These observations would indicate that either the CCR5 or CCL3 gene deficiency attenuated neovascularization in tumor foci by depressing the expression of two potent angiogenic molecules, HGF and MMP-9.
Interactions between the CCL3-CCR5 axis and MMP-9
In line with previous reports (34), double-color immunofluorescence analysis demonstrated that a substantial portion of F4/80-positive macrophages and Gr-1-positive granulocytes expressed MMP-9 proteins (Fig. 7,A), whereas CD3-positive cells did not express MMP-9 protein (data not shown). CCL3 enhanced MMP-9 mRNA expression in peritoneal macrophages of WT mice (Fig. 7,B), but not MMP-2 or MMP-13 (data not shown). Moreover, CCL3 increased gelatinolytic activity of MMP-9 in WT but not in CCR5KO mouse-derived macrophages (Fig. 7 C). These observations suggest that the CCL3-CCR5 interactions could enhance MMP-9 expression and activity in macrophages, thereby inducing tumor angiogenesis.
Interactions between the CCL3-CCR5 axis and HGF
Given that fibroblasts are a major cellular source of HGF (11, 35), we next examined whether fibroblasts could express HGF proteins. A double-color immunofluorescence analysis detected HGF proteins in type I collagen-positive cells in tumor sites (Fig. 8,A). Moreover, in vitro stimulation by CCL3 markedly augmented HGF expression at mRNA and protein levels by WT-derived fibroblasts, but not by CCR5KO mouse-derived fibroblasts (Fig. 8, B and C). Thus, the CCL3-CCR5 interactions can directly regulate the intratumoral expression of HGF, a major angiogenic factor by fibroblasts.
Chemokines can recruit and activate a selective set of leukocytes and nonleukocytic cells such as endothelial cells (36, 37, 38). We previously demonstrated that a chemokine, CCL3, was abundantly expressed in human hepatocellular carcinoma tissues and proved the crucial involvement of CCL3 and its receptor, CCR1, in chemical carcinogen-induced murine hepatocarcinogenesis (39). Our preliminary experiments demonstrated that i.v. injection of Renca cells caused multiple lung metastasis foci, together with enhanced intrapulmonary expression of CCL3. CCL3 utilizes two distinct receptors, CCR1 and CCR5 (20, 40), but these two receptors exhibit different expression patterns, and it therefore can have distinct pathophysiological roles in a context-dependent manner (18, 39, 41, 42). These observations prompted us to investigate the roles of CCL3 and its receptors in the present lung metastasis model. CCR5-positive but not CCR1-positive cells accumulated in metastasis foci, and the deficiency of the CCL3 or CCR5 gene, but not of the CCR1 gene, markedly reduced the numbers of lung metastases. Similar results were observed in another lung metastasis model in C57BL/6 mice injected with a melanoma cell line, B16-F10 (43). This melanoma-induced lung metastasis process requires stromal cells but not leukocytes (43), whereas our model requires both BM- and non-BM-derived cells. This discrepancy may be explained by differences in mouse strains and cell lines.
Before tumor injection, intrapulmonary mononuclear cells, particularly macrophages, expressed constitutively a low level of CCL3. In contrast, unstimulated Renca cells in vitro expressed undetectable levels of CCL3 and CCL4 (our unpublished data). Considering that Renca cell injection induced infiltrating cells and tumor cells to express CCL3 and CCL4 protein in WT but not in CCL3-deficient mice, it is reasonable to speculate that macrophage-derived CCL3 could induce additional CCL3 and CCL4 production by infiltrating cells as well as tumor cells in a paracrine manner. Thus, the absence of macrophage-derived CCL3 abrogated the enhanced CCL3 and CCL4 production induced by tumor cell injection. This conclusion was supported by the observation that intrapulmonary CCL3 and CCL4 levels were markedly lower in CCL3-deficient mice than in WT mice. Thus, although CCL3-deficient mice exhibited CCR5 on responding cell populations to a similar extent as WT mice, a marked reduction in intrapulmonary CCL3 and CCL4 levels attenuated the response of responding cell populations, such as macrophages and fibroblasts, which are a major source of MMP-9 and HGF, respectively.
Tumor cells extravasate from a primary site, circulate through the bloodstream and/or lymphatics, and invade other tissues and grow to establish metastatic foci (44). Our present lung metastasis model is frequently employed to elucidate the molecular mechanisms underlying development of metastases. Neovascularization is indispensable for tumor growth, particularly at metastasis sites (30). CD31-positive vascular areas were diminished in CCL3KO and CCR5KO mice compared with WT mice. Thus, reduced neovascularization may account for reduced metastasis formation in these gene-deficient mice.
The first step of neovascularization is the degradation of basement membrane surrounding an existing vessel (45). This process is presumed to be regulated by MMPs, particularly MMP-2 and MMP-9 (31, 32, 33), which can efficiently degrade type IV collagen, a major extracellular component surrounding vessels. Intrapulmonary MMP-9, but not MMP-2, mRNA expression was enhanced progressively in WT mice as lung metastasis foci formed. Moreover, the increase in MMP-9 mRNA expression was considerably reduced in both CCL3KO and CCR5KO mice, as were MMP-9-expressing cells. Several lines of evidence indicated that inflammatory cells are a major source of MMP-9 in developing tumors (13, 33). MMP-9 protein was consistently detected in granulocytes and macrophages, the cells that express CCR5. Furthermore, in vitro stimulation by CCL3 could augment MMP-9 expression by macrophages in a CCR5-dependent manner. Thus, the CCL3-CCR5 axis can induce the accumulation of inflammatory cells and their MMP-9 expression, thereby contributing to neovascularization.
As neovascularization proceeds, migration and proliferation of endothelial and mural cells, as well as lumen formation, ensue after the degradation of basement membrane. These processes are regulated by several angiogenic factors including HGF, bFGF, and VEGF (46, 47, 48). Indeed, the mRNA expression of HGF and bFGF but not VEGF was augmented in lungs after Renca cell injection and, therefore, these two angiogenic factors may be mainly responsible for endothelial cell proliferation and lumen formation in this metastasis process. However, the deficiency of either the CCL3 or CCR5 gene attenuated only HGF, but not bFGF, expression. Thus, the bFGF-mediated signals persisted in the lungs of CCL3KO and CCR5KO mice to a similar extent as in WT mice. Consequently, the persistent bFGF-mediated signals may account for residual neovascularization and subsequent metastasis formation in these deficient mice.
Tumor-associated fibroblasts are presumed to be crucially involved in tumor progression (29, 49, 50). In this model, type I collagen-positive cells expressed CCR5 and accumulated progressively in tumor foci. A deficiency of either the CCL3 or CCR5 gene reduced the numbers of type I collagen-positive cells in tumor foci. Thus, the CCL3-CCR5 axis regulated the intratumoral accumulation of fibroblasts in this model. Tumor-associated fibroblasts can produce SDF-1α/CXCL12 and HGF (11, 50, 51), the mediators that can promote tumor progression. Renca cell injection, however, failed to augment intrapulmonary CXCL12 mRNA expression in WT mice. In contrast, intrapulmonary HGF expression was enhanced progressively in WT mice after Renca cell injection, and a deficiency of either the CCL3 or CCR5 gene attenuated HGF expression. Moreover, HGF was detected mainly in type I collagen-positive fibroblasts, in line with previous observations that fibroblasts are a major source of HGF (11). Furthermore, in vitro stimulation with CCL3 can augment HGF expression by lung fibroblasts at mRNA and protein levels. Thus, the CCL3-CCR5 axis is involved in HGF expression by fibroblasts as well as their intratumoral accumulation.
Fibrocytes have recently been identified as a minor component of circulating pools of leukocytes based on the characteristic patterns of markers (21, 52, 53), including type I collagen and CD45. Subsequent studies demonstrated that fibrocytes traffic to injured organs, including skin and lung, and contribute to the development of fibrosis (54). We previously detected CCR5 on fibrocytes present in peripheral blood and lungs damaged by bleomycin treatment (21). However, the origin of tumor-associated fibroblasts in lung metastasis foci remains elusive. We hypothesized that BM transplantation from WT to CCR5KO mice could abrogate a reduction in the formation of metastasis in CCR5KO mice if CCR5 expression was restricted to BM-derived cells, including leukocytes and fibrocytes. In contrast to this assumption, BM transplantation from WT to CCR5KO mice failed to abolish the reduction in metastasis formation. Thus, it is more probable that CCR5-expressing tumor-associated fibroblasts were not derived from circulating fibrocytes.
Our present observations suggest that the CCL3-CCR5 axis has important roles in the accumulation of MMP-9-expressing leukocytes and HGF-expressing fibroblasts. Because both mediators have profound effects on various aspects of tumor cell functions, in addition to neovascularization, the CCL3-CCR5 axis may augment tumor metastasis by augmenting various tumor cell functions, particularly their motility.
We express our gratitude to Drs. P. M. Murphy and J.-L. Gao (National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD) for providing us with CCR1KO mice. We thank Dr. Joost J. Oppenheim (National Cancer Institute, Frederick, MD) for his critical comments on the manuscript.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work is supported in part by a Grant-in-Aid for Scientific Research (B) (No. 19390112) from the Ministry of Education, Culture, Sports, Science and Techology of the Japanese government. Y.-Y.L. is a recipient of a Postdoctoral Fellowship for Foreign Researcher, supported by Japan Society for the Promotion of Science.
Abbreviations used in this paper: MMP, matrix metalloproteinase; bFGF, basic fibroblast growth factor; BM, bone marrow; HGF, hepatocyte growth factor; KO, knockout; pAb, polyclonal antibody; PlGF, placenta-derived growth factor; SDF, stromal cell-derived factor; VEGF, vascular endothelial growth factor; WT, wild type.
The online version of this article contains supplemental material.