Tumor-stromal crosstalk in invasion of oral squamous cell carcinoma: a pivotal role of CCL7

Authors

  • Da-Woon Jung,

    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    Search for more papers by this author
  • Zhong Min Che,

    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    Search for more papers by this author
  • Jinmi Kim,

    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    Search for more papers by this author
  • Kyungshin Kim,

    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    Search for more papers by this author
  • Ki-Yeol Kim,

    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    Search for more papers by this author
  • Darren Williams,

    1. Department of Life Science, Gwangju Institute of Science and Technology, Gwangju, Republic of Korea
    Search for more papers by this author
  • Jin Kim

    Corresponding author
    1. Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, Seoul, Republic of Korea
    • Department of Oral Pathology, Oral Cancer Research Institute, Brain Korea 21 Project, Yonsei University College of Dentistry, 134 Shinchon-Dong, Seoul 120-752, Republic of Korea
    Search for more papers by this author
    • Tel.: +82-2-2228-3042, Fax: +82-2-392-2959


Abstract

Recent studies have shown that stromal fibroblasts have a more profound influence on the initiation and progression of carcinoma than was previously appreciated. This study aimed at investigating the reciprocal relationship between cancer cells and their associated fibroblasts at both the molecular and cellular level in oral squamous cell carcinoma (OSCC). To identify key molecular regulators expressed by carcinoma-associated fibroblasts (CAF) that promote cancer cell invasion, microarrays were performed by comparing cocultured OSCC cells and CAF with monoculture controls. Microarray and real-time PCR analysis identified marked upregulation of the chemokine (C-C motif) ligand 7 (CCL7) in cocultured CAF. ELISA showed an elevated level of CCL7 secretion from CAF stimulated by coculture with OSCC cells. CCL7 promoted the invasion and migration of OSCC cells, and the invasiveness was inhibited by treatment with CCL7 neutralizing antibody. OSCC cells were shown to express CCR1, CCR2 and CCR3, receptors for CCL7, by RT-PCR. In addition, treatment with anti-CCR1 or anti-CCR3 antibody inhibited CCL7-induced OSCC cell migration, implicating that CCL7 promotes cancer cell migration through CCR1 and CCR3 on OSCC cells. Cytokine antibody array analysis of the supernatant from OSCC cell culture revealed that interleukin-1α was an inducer of CCL7 secretion by CAF. This study confirms the reciprocal relationship of the molecular crosstalk regulating the invasion of OSCC and describes new potential targets for future therapy.

Carcinomas are malignant neoplasms derived from epithelial cells and are surrounded by specialized stroma, which orchestrate with cancer cells to regulate disease progression.1–3 Carcinoma-associated fibroblasts (CAF) have been recognized as prominent modifiers of cancer initiation and progression.4, 5 For instance, it has been previously demonstrated that human prostatic CAF induce tumor formation from initiated but nontumorigenic human prostatic epithelial cells.6 CAF also facilitate the invasiveness of otherwise noninvasive cancer cells when coinjected into mice.7 The putative proinvasive effects of CAF may be mediated through either direct heterotypic cell−cell contacts8 or diffusible molecules, such as inflammatory mediators, cytokines and chemokines.9, 10

Chemokines have been shown to play an important role in tumor biology by influencing tumor growth, invasion and metastasis.11 Chemokines are a family of small, structurally related cytokines with chemoattractant and activation properties that are involved in inflammatory reactions.12 They are classified mainly into the CC and CXC subfamilies, according to the location of the first 2 cysteine residues, and are produced by a range of cell types, including fibroblasts. Various types of cancer cells also express chemokines and chemokine receptors,11, 13–15 and their autocrine and paracrine roles in cancer progression are receiving increasing attention. For example, the CXC chemokine, CXCL12 (stromal cell-derived factor 1), secreted by CAF, recruits endothelial progenitor cells, enhancing angiogenesis and stimulates tumor growth in a xenograft model and increases tumor cell invasiveness.16

Head and neck squamous cell carcinoma (HNSCC) is the fifth most common cancer and about 40% develop within the oral cavity. Although oral squamous cell carcinoma (OSCC) can be noticed by patients or clinicians, most of them are detected at stage III or IV, which predict severe facial deformity or poor outcome.17 Consequently, research has focused on the early detection and prevention of OSCC. Recently, we demonstrated that the presence of fibroblasts is essential for OSCC cell invasion.18 In our organotypic coculture model, OSCC cells separated from collagen embedded stromal fibroblasts by a collagen filter showed invasive growth into a dermal equivalent, whereas cancer cells failed to invade unless the stromal fibroblasts are present. From this result, we presumed that the initial step of cancer invasion is induced by soluble mediators derived from stromal fibroblasts. The study presented here analyzed the proinvasive molecular crosstalk between OSCC cells and CAF using a coculture model employing collagen-coated transwells. We identified chemokine (C-C motif) ligand 7 (CCL7/MCP-3) as a key regulator of OSCC cell invasiveness and migration and showed that this corresponds to its expression pattern in vivo. In addition, we have demonstrated that interleukin-1α (IL-1α) and vascular endothelial growth factor (VEGF) secreted by OSCC cells regulated CCL7 expression by CAF. This study showed a molecular-based dialogue between OSCC cells and CAF in the progression of OSCC and provides a new strategy for the development of novel therapies targeting the expression and/or activity of these cytokines and their receptors.

Materials and Methods

All studies involving human subjects were approved by the Institutional Review Board of Yonsei Dental Hospital, Yonsei University Health System, Seoul.

Chemokines and antibodies

Human recombinant monocyte chemotactic protein CCL7, human recombinant IL-1α and VEGF were obtained from R&D Systems (R&D, Minneapolis, MN). Anti-human CCL7, anti-human IL-1α, anti-CCR1, anti-CCR2, anti-CCR3 and anti-CCR5 antibodies and isotype control antibodies (murine IgG1, IgG2A and IgG2B and rat IgG2A) were also purchased from R&D Systems.

Fibroblast and oral cancer cell culture

CAF were derived from the surgical specimens of 3 OSCC patients, and normal oral mucosal cells were derived from 3 patients who underwent wisdom tooth extraction without mucosal disease. The informed consent was given by the 6 patients for this study. CAF were selected by Versene solutions (0.1 g EDTA and 2 mL glucose in 500 mL PBS buffer) from explanted cancer tissues. The fibroblasts were maintained in Dulbecco's Modified Eagles Medium (DMEM; Gibco BRL, USA) and Ham's Nutrient Mixture-F12 (Gibco BRL, USA) culture media at a ratio of 3:1, supplemented with 10% FBS and 1% penicillin/streptomycin. The isolated fibroblasts were characterized by immunohistochemical staining with antivimentin (1:100, DAKO, Tokyo, Japan) and anti-α-smooth muscle actin (1:50, Sigma, St. Louis, MO), along with comparison to normal fibroblasts (NF) (Supporting Information Figure S1). YD-10B and YD-38 OSCC cells were obtained from the Korean Cell Line Bank (Seoul, Korea) and cultured as described.18 The oral cancer cell lines HSC-2, HSC-3 and Ca9.22 were gifts from Prof. Takashi Muramatsu, Tokyo Dental College, Japan, and maintained in DMEM: Ham's-F12 (3:1) culture media, supplemented with 10% FBS and 1% penicillin/streptomycin. The breast cancer cell lines MDA-MB-231 and MCF-7 were obtained from the American Type Culture Collection (ATCC) and cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. PC-3 prostate cancer cell line was a gift from Prof. Sun-Young Ra, Yonsei University College of Medicine, Republic of Korea, and maintained in Ham's-F12 supplemented with 10% FBS and 1% penicillin/streptomycin. All cells were cultured at 37°C in an atmosphere containing 5% CO2.

Invasion assay

To evaluate invasive activity, we used a modified transwell invasion assay.19 Using 24-transwell plates (Corning, NY), the inserts containing 8 μm pore size filters were coated with collagen Type I (45 μg/30 μL/well) for invasion assay. Briefly, 2 × 104 OSCC cells were placed in the transwell chambers with porous filters in the upper wells. CAF (2 × 104) or CAF-CM was added into the lower well to evaluate whether CAF or their soluble factors induce invasive growth of cancer cells. The cells that penetrated the filter were fixed, stained with 0.25% crystal violet and counted by light microscopy.

Microarray data analysis

As shown in Figure 1b, we designed a coculture model to identify molecules showing different expression in CAF when they were exposed to cancer cells. YD-10B OSCC (1.7 × 105) cells and 1.7 × 105 CAF were seeded in the upper chamber and lower chamber, respectively, of 6-transwell plates containing collagen-coated 1 μm pore transmembrane filters (Becton Dickinson, Franklin Lakes, NJ). Monoculture control samples were generated by culturing only CAF or OSCC on the same side of the filter as in the coculture design. Total RNA was extracted using the TRI Reagent (MRC, Cincinnati, OH) according to the manufacturer's instructions. Each total RNA sample (30 μg) was labeled with cyanine 3 (CY3) or cyanine (Cy5)-conjugated dCTP (Amersharm, Piscataway, NJ) by a reverse transcription reaction using reverse transcriptase, SuperScrip ll (Invitrogen, Carlsbad, CA). More details of the microarray method are described in the supplementary method section.

Figure 1.

CAF induces invasive growth of cancer cells, and CCL7 mRNA expression is increased in CAF by coculturing with OSCC cells. (a) The numbers of cancer cells that penetrated the transwell membrane were stained and counted under light microscopy. Data shown are representative of multiple repeats (means ± SD of triplicates, **p value of <0.01). (b) Coculture system was created to identify gene expression profiling, and microarray was performed. The numbers of upregulated genes are shown in the Venn diagram (full microarray data can be found at Gene Omnibus Online; accession number GSE18532). (c) Microarray data were confirmed by real-time PCR. Chemokines, such as CXCL8, CXCL1, CXCL2, CXCL3 and CCL7, showed high expression in cocultured CAF. (d) CCL7 mRNA expression in CAF cocultured with OSCC cells (YD-10B, HSC-2 and Ca9.22) was quantified using SYBR Green, normalized to GAPDH mRNA and compared with CAF monoculture. The real-time PCRs were carried out in triplicates (mean ± SD) using equal amounts of cDNA sample equivalent to 50 ng of starting total RNA.

Quantitative real-time PCR

For real-time PCR analysis, 1.7 × 105 OSCC cells (YD-10B, HSC-2 or Ca9.22) and 1.7 × 105 CAF were seeded in the upper chamber and lower chamber, respectively, of 6-transwell plates containing collagen-coated, 1 μm pore transmembrane filters (Becton Dickinson, Franklin Lakes, NJ). Monoculture control samples were generated by culturing only CAF on the same side of the filter as in the coculture design. Cells were incubated in serum-free media for 24 hr at 37°C, and total RNA was isolated from CAF using the Trizol reagent (Invirogen, Carlsbad, CA).

The sequences of the primers were described in the Supporting Information Table S1. Reverse transcription was carried out using the Taqman Reverse Transcription Reagents (Applied Biosystems, Foster City, CA). Each 20 μL PCR reaction was carried out by adding 0.8 μL of 10 pM forward and reverse primers for CCL7 and GAPDH, 10 μL SYBR Green Master Mix (Applied Biosystems, Foster City, CA), 8.7 μL deionized water and 0.5 μL of cDNA. The fluorescent signal detection was detected using the Applied Biosystems Prism 7900HT Sequence Detection System.

Immunohistochemistry

Immunohistochemistry was performed as described previously.18 Paraffin-embedded tissue sections obtained from 25 OSCC patients were used for CCL7 detection. Anti-CCL7 polyoclonal antibody (Gen Way Biotech, San Diego, CA) and anti-IL-1α monoclonal antibody (R&D, Minneapolis, MN) were applied at a dilution of 1:50 for 30 min at 37°C. For immunohistochemical staining of cultured cells, CAF and NF were seeded in Lab-Tek chamber slides (Nalge Nunc, Naperville, IL) and grown until 70–80% confluence. They were then fixed in 10% formaldehyde and washed with PBS and absolute ethanol, followed by 3% hydrogen peroxide treatment. Anti-vimentin (1:100, DAKO, Tokyo, Japan) or anti-α-smooth muscle actin (1:50, Sigma St. Louis, MO) antibodies were incubated overnight at room temperature. All immunohistochemical staining were then incubated with biotin-labeled horse anti-mouse antibody (Vector Lab, Burlingame, CA) for 30 min at 37°C, followed by horseradish peroxidase-streptavidin conjugate for another 30 min at 37°C. Visualization was performed by 3′,3-diaminobenzidine tetrachloride (DAB, Vector Lab, Burlingame, CA), and cells were counterstained with Meyer's hematoxylin.

Conditioned medium

YD-10B OSCC cells (1 × 106) or 1 × 106 CAF cells were cultured for 2 days in 10 cm culture plates containing 5 mL DMEM: Ham's-F12 (3:1). The supernatant was collected and used as conditioned media after filtration.

Measurement of secreted CCL7 by ELISA

Cancer cells (2 × 104) were cultured for 24 hr in 24-transwell plates with or without CAF (2 × 104), and their supernatants were obtained. CCL7 was then measured using the commercially available CCL7 ELISA kit (R&D, Minneapolis, MN). To test the effects of IL-1α and/or VEGF on CCL7 secretion, 2.1 × 104 CAF were seeded in 24-well plates for 24 hr in DMEM: Ham's-F12 (3:1) supplemented with 10 % FBS, washed 3 times with serum-free media and further incubated for 24 hr in the serum-free media containing cytokine. The supernatant was collected for the ELISA.

Cell migration and wound healing assay

Migration assay was carried out using the same procedure as the invasion assay, except filters were coated with fibronectin (10 μg/50 μL, R&D, Minneapolis, MN) instead of collagen. For wound healing assay, OSCC cells (YD-10B, Ca9.22 and HSC-2) in DMEM: Ham's-F12 (3:1) supplemented with 1% FBS were seeded onto 6-well plates and allowed to adhere overnight (>90% confluence). The monolayer was wounded with a 200 μL plastic pipette tip. Cells were washed once, and fresh culture media was added with or without CCL7 (10 and 100 ng/mL). The cells were allowed to invade the wound and counted by light microscopy.

F-actin stain

FITC-labeled phalloidin (Sigma St, Louis MO) was used to visualize the cytoskeletal changes in OSCC cells after incubation with CCL7. HSC-2 cells (3 × 104) were seeded in chamber slides (Nalge Nunc) and incubated with CCL7 (10 and 100 ng/ml) or CAF-CM in serum-free media at 37°C. After 6 hr incubation, cells were washed in PBS and fixed with 4% paraformaldehyde for 5 min at room temperature. Cells were dehydrated with acetone, permeabilized with 0.1% Triton X-100 in PBS and incubated for 40 min at room temperature with the fluorescent phalloidin conjugate (50 ng/mL). Cells were rinsed in PBS, mounted and visualized by fluorescence microscopy (Zeiss Pascal laser scanning microscope). Nuclei were labeled with 50 ng/mL propidium iodide (Sigma St, Louis MO). To address whether the effect of CCL7 on cells was specific, we added anti-CCL7 antibody (2 μg/mL) to CCL7- and CAF-CM-treated cells.

Three-dimensional culture

For 3-dimensional culture, the dermal equivalent was prepared as described.18 Briefly, 300 μL Type I-A collagen mixture (Nitta Gelatin Inc, Osaka, Japan) was generated by mixing 8 volumes of an ice-cold collagen solution with 1 volume 10× reconstitution solution (0.022 g/mL NaHCO3, 0.0477g/mL HEPES, 0.05 N NaOH) and 1 volume 10× DMEM. HSC-2 cells were cultured on the dermal equivalents alone or mixed with CAF. CAF (1 × 105) was added, and the mixture was dispensed into a 12 mm Millicell (Millipore Corp, Bedford, MA) inserted into a 6-well culture plate. After 1 day of incubation at 37°C, HSC-2 cells (1 × 105) were plated onto the dermal equivalents and submerged into the culture medium for 4 days. The cancer cells were then exposed to the air by removing the medium from the surface and cultured for 3 days. The cultured tissue was fixed in 10% neutral formalin and embedded with paraffin. For histological examination, the tissue was sectioned and stained with hematoxylin and eosin.

RT-PCR

OSCC cells (YD-10B, HSC-2 and Ca9.22) and PC-3 prostate cancer cells were seeded in 6 cm dishes and incubated in serum-free media for 24 hr at 37°C. CCL7 (100 ng/mL) was treated for 2 hr before harvest. Total RNA was isolated from each cell type using the RNeasy kit (Qiagen, GmbH, Germany) and 1 μg of the total RNA was used for first-strand cDNA synthesis.

The following primers were used for RT-PCR: CCR1:

5′-CTACGCCTTCGTTGGTGAGA-3′ (forward) and 5′-GCTTATTTTGGGTTGGCCTC-3′ (reverse), CCR2: 5′-AGG TGGAGAAGCTCCCTGAA-3′ (forward) and 5′-TTAGCC ATGTGGCCTGAAAG-3′ (reverse), CCR3: 5′-CCGGACTGT CACTTTTGGTG-3′ (forward) and 5′-ACCTCAGCAGCG TTTTGATG-3′ (reverse), GAPDH: 5′-GAAGGTGAAGG TCGGAGT-3′ (forward) and 5′-GAAGATGGTGATGGGAT TTC-3′ (reverse)

The PCR protocol for CCR1, CCR2 and CCR3 consisted of 28 thermocycles of 30 sec denaturation at 95°C, 30 sec annealing at 60°C and 1 min renaturating at 72°C.

Cytokine antibody array analysis

The sandwich ELISA human cytokine antibody array was purchased from Panomics Inc (CA). Ca9.22 (2.5 × 106) or YD-38 cells were seeded in T75 tissue culture flasks overnight, and the media were changed to 5 mL of serum-free DMEM: Ham's-F12 (3:1) after 3 times washing with PBS. After 24 hr incubation, the supernatants were collected and normalized for cell number between samples by dilution with serum-free media before binding to the antibodies immobilized on the array. The changes among 36 cytokines examined in the culture medium could be profiled at a protein level of pg/mL, in accordance with the manufacturer's instructions. The relative expression level of the cytokines was determined by comparing signal intensities.

siRNA and Western blot

IL-1RI and control siRNAs were purchased from Santa Cruz Biotechnology, Inc., CA. siRNA-mediated inhibition of gene expression was carried out according to manufacturer's instructions. CAF were seeded at 1 × 105 cells/well in a 6-well culture plate 24 hr before transfection. For western blot analysis of IL-1RI expression, CAF protein was harvested 48 hr after transfection from triplicate wells of a 6-well culture plate. The proteins were separated on a 10% SDS-PAGE and transferred on polyvinylidene difluoride membranes (Millipore Corp, Bedford, MA).

The CAF were incubated for 24 hr with or without 50 pg/mL IL-1a(after transfection) and the conditioned media were collected for ELISA specific to CCL7.

Zymography

The effect of CCL7 on the activity of MMP-2 and MMP-9 proteins in OSCC cells was analyzed by gelatinolytic zymography using SDS-PAGE. Briefly, HSC-2 (1 × 105) cells were seeded in a 6-well plate and treated with 100 ng/mL of CCL7 in serum-free media for 24 hr. CXCL8 (interleukin-8) was treated separately or together with CCL7. More details of the zymography are described in the supplementary method section.

Statistical Analysis

Data were analyzed with the SPSS statistical program (version 17.0). One-way ANOVA was used for comparison between multiple groups. Student's t test was used for comparison between 2 groups. p values of <0.05 were considered significant.

Results

Expression of CCL7 mRNA was increased in CAF cocultured with OSCC cells

We established a modified transwell invasion assay, using collagen type I-coated filters, to assess the effect of CAF on OSCC cell invasion in vitro. OSCC cell invasion (YD-10B, Ca9.22, HSC-2, HSC-3 and YD-38) was found to be significantly increased by coculture with CAF (Fig. 1a). We then speculated that CAF stimulated by OSCC cells release paracrine factor(s) that increase cancer cell invasion. To understand this crosstalk at the molecular level, the gene expression profile of the cocultured CAF was analyzed by microarray (the complete microarray data are deposited at Gene Expression Omnibus (GEO) accession number GSE18532). A total of 189 genes were upregulated (≥2.0-fold increase) in cocultured CAF compared with monocultured CAF (Fig. 1b). Among them, 64 genes were selectively upregulated in cocultured CAF but not in cocultured OSCC cells. These 64 genes were taken into consideration for further study to find soluble factor(s), which are released from CAF in response to cocultured OSCC cells, and promote OSCC cell invasion in a paracrine manner. A list of the 64 genes selectively upregulated in CAF after coculture with OSCC cells is provided in Supporting Information Table S2. CXCL1, CXCL2, CXCL3, CXCL8 and CCL7 were chemokines showing the greatest relative fold change in cocultured CAF (compared with cocultured OSCC cells) and were chosen for quantitative real-time PCR analysis. CCL7 was found to exhibit the highest fold change (55.55 fold) in gene expression, as measured by quantitative real-time PCR (Fig. 1c).

CCL7 mRNA expression in CAF was further analyzed in cocultures of CAF and 3 human OSCC cell lines (HSC-2, YD-10B and Ca9.22) and compared with control monocultured CAF. CCL7 was markedly upregulated (43.6–158.8 fold) in CAF by coculturing with OSCC cells (Fig. 1d), showing good concordance with our microarray data.

The protein level of CCL7 secreted by CAF was highly enhanced by coculturing with OSCC cells

To determine whether coculture of CAF with OSCC cells increases CCL7 protein secretion, we used CCL7-ELISA. CCL7 protein was detected in the supernatants of both OSCC and CAF single cultures. However, coculturing the OSCC cells and CAF significantly increased CCL7 release into the culture media within 24 hr of incubation (Fig. 2a). Interestingly, breast cancer cell lines did not elevate the level of CCL7 released by CAF in coculture. In addition, CCL7 secretion was elevated by coculturing CAF with normal epithelial cells (NE), although no detectable amount of CCL7 was secreted by NE. However, this increased CCL7 secretion was lower compared with CAF cocultured with OSCC cells, suggesting that CCL7 can be a unique tool for molecular communication between CAF and OSCC cells.

Figure 2.

CAF dramatically increased the release of CCL7 by coculturing with OSCC cells. (a) Five OSCC cell lines (YD-10B, HSC-2, HSC-3, Ca9.22 and YD-38), 2 human breast cancer cell lines (MCF-7 and MDA-MB-231) and primary cultured normal oral epithelial cells (NE) were tested for their ability to enhance the secretion of CCL7 from CAF. Cancer cells were seeded in 24-transwell chambers with porous filters (8 μm pore size) in the upper well, and CAF was added into the lower well. After 24 hr incubation, the supernatant was collected and measured using the commercially available CCL7 ELISA kit. The statistical difference for all 5 OSCC cell lines compared with CAF monoculture (designated as CAF alone) is **p < 0.01. Data shown are representative of multiple repeats (means ± SD of triplicates). (b) Conditioned media were collected from CAF (designated as CAF-CM) or YD-10B (designated as 10B-CM) and added to YD-10B culture or CAF culture, respectively, in a 24-well plate. Twenty-four hours later, the supernatant was collected from each culture, and the CCL7 content was measured. Results were reported as means ± SD of 3 independent assays (**p value of <0.01 compared with CAF monoculture). The minimum detectable dose of the CCL7 ELISA ranged 0.29–8.52 pg/mL. (c) Immunohistological staining of CCL7 was performed in the specimens from OSCC patients. a: negative control of murine IgG1 staining (×200). b: CCL7 was expressed in the cytoplasm of cancer cells, stromal fibroblasts and inflammatory cells (×200). c: the strongest expression of CCL7 was found in the fibroblasts (×1,000). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

As shown in Figure 2b, we also found that 10B-CM increased CCL7 secretion by CAF to a similar level to that observed in the coculture of YD-10B cells and CAF, whereas CAF-CM did not induce CCL7 release into the YD-10B supernatant. This observation suggests that CAF exposed to OSCC were the main source of the increased CCL7 in the coculture.

We observed the CCL7 expression in human tissues from 25 OSCC patients (Fig. 2c). Immunohistochemical staining revealed that CCL7 was expressed in the cytoplasm of cancer cells, fibroblasts and inflammatory cells. In particular, the strongest CCL7 expression was found in the cytoplasm of stromal fibroblasts (Fig. 2c-c).

We also compared CAF with NF for their abilities to release CCL7 into the coculture media. Fibroblasts were cultured with or without YD-10B OSCC cells, and the culture media were collected 24 hr later for CCL7-ELISA. CCL7 release from the cocultured CAF was much higher than that from the cocultured NF, whereas the release of CCL7 from both CAF and NF was significantly increased by coculture with OSCC cells (Supporting Information Table S3), indicating that CAF were more responsive than NF to OSCC cell stimulation for enhancing CCL7 release. In addition, CAF was found to be a stronger inducer of OSCC cell invasion (Supporting Information Figure S1b).

CCL7 enhanced OSCC cell migration and induced cytoskeletal modification in the OSCC cells

We examined the effect of CCL7 on OSCC cell migration by wound healing assays in 6-well culture plates. Confluent monolayers of OSCC cells (YD-10B, HSC-2 and Ca9.22) were injured and then treated with CCL7. Cell migration into the wound was significantly increased by CCL7 dose dependently (Fig. 3a). Migration induced by 100 ng/mL CCL7 was inhibited by pretreatment with a neutralizing antibody for CCL7. Cell migration assay using fibronectin-coated transwells showed similar results compared with the wound healing assay (Fig. 3b)

Figure 3.

CCL7 promoted OSCC cell migration and induced membrane ruffling and cell spreading in OSCC cells. (a) Confluent monolayers of OSCC cells (YD-10B, HSC-2 or Ca9.22) were scraped with a sterile pipette tip, and the cells that migrated into the wounded monolayer were counted after 12 hr in the presence or absence of CCL7 (10 and 100 ng/ml). Anti-CCL7 antibody (1 μg/ml) was tested for its antagonizing effect on cancer cell migration enhanced by 100 ng/mL CCL7. Murine IgG1 antibody (1 μg/ml) was used as a control. Data shown are representative of multiple repeats (means ± SD of triplicates; **p value of <0.01; *p value of <0.05). (b) YD-10B cells were plated on fibronectin-coated transwell filters (8 μm pore) with or without CCL7 (1, 10 and 100 ng/ml) and incubated for 24 hr. Cell migration was quantified by counting the invaded cells, which were counted in 5 separate microscopic fields per filter, and the mean values per filter (±SD) were calculated from 4 replicate filters. Data shown are representative of multiple repeats (**p value of <0.01). (c) HSC-2 cells were incubated with or without CCL7 (10 and 100 ng/ml) for 6 hr and were stained with FITC-conjugated phalloidin to determine the changes in cell morphology. Cells were treated as follows: a, serum-free media; b, 10 ng/ml CCL7; c, 100 ng/ml of CCL7; d, CAF-CM; e, 2 μg/mL of CCL7 neutralizing antibody plus CAF-CM and f, 2 μg/mL of CCL7 neutralizing antibody plus 100 ng/ml CCL7. Size bar: 50 μm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Importantly, CCL7 influenced cell migration and not cell proliferation, because no significant changes were observed in the number of OSCC cells (HSC-2 and YD-10B) treated with CCL7 (10 and 100 ng/mL) for 72 hr (Supporting Information Figures S2a and b).

In addition, we speculated that CCL7 may induce cytoskeletal changes in OSCC cells to enhance cell motility. Comparison of untreated and CCL7-treated cells revealed that CCL7 stimulated membrane ruffling and cell spreading (Figs. 3c-b and 3c-c), which are considered important events in cell locomotion.20, 21 CAF-CM induced similar changes to those induced by 100 ng/mL CCL7 (Fig. 3c-d). Neutralization by anti-CCL7 monoclonal antibody efficiently antagonized the cytoskeletal changes induced by either CCL7 or CAF-CM (Figs. 3c-e and 3c-f).

Anti-CCL7 antibody treatment inhibited OSCC cell invasion

To determine whether CCL7 is involved in the induction of OSCC cell invasion by CAF, we assessed the ability of OSCC cells (YD-10B and HSC-2) to invade through a collagen I matrix toward CAF-CM in the presence and absence of an anti-CCL7 antibody. After 24 hr of culture, a significant reduction of cancer cell invasion was observed and compared with controls (Fig. 4a).

Figure 4.

Invasion of OSCC cells was inhibited by pretreatment with CCL7 neutralizing antibody. (a) YD-10B or HSC-2 cells in serum-free media were placed in the upper well of a 24-transwell plate with collagen-coated filters (8 μm pore). CAF-CM was added into the lower well to induce invasion. For inhibition, 1 μg/mL of the anti-CCL7 antibody was added to the lower wells before OSCC cell placement. i, After 24 hr incubation, the number of cells that transmigrated through the filter were stained with crystal violet and visualized by light microscopy (scale bar: 150 μm). Representative pictures of crystal violet stained filters are shown. Arrow indicates the transmigrated cancer cell. ii, Cell invasion was quantified by counting the invaded cells, which were counted in 5 separate microscopic fields per filter, and the mean values per filter (±SD) were calculated from 4 replicate filters. Murine IgG1 was used as a control antibody. Data shown are representative of multiple repeats (means ± SD of quadruple assays, **p value of <0.01). (b) HSC-2 cells were seeded onto dermal equivalents without CAF (a) or with CAF (b–d) and grown over a period of 7 days. The anti-CCL7 antibody (1 μg/mL) was added to the culture media for inhibition study (c). Murine IgG1 was used as a control antibody (d). Representative pictures of hematoxylin & eosin stained sections are shown. Arrow head indicates cancer cells invaded into the dermal equivalents with CAF. Scale bar: 150 μm. (c) HSC-2 (1 × 105) cells were seeded in a 6-well plate and treated with or without 100 ng/mL CCL7 in serum-free media for 24 hr. CXCL8 (100 ng/mL) was treated separately or together with CCL7. Samples (15 μg protein) were loaded onto a 10% electrophoresis gel copolymerized with 0.2% gelatin. i, A representative picture of Coomassie blue-stained gel is shown. ii, Quantification analysis was done by densitometric measurements (means ± SD of 2 independent assays). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

We also tested whether the anti-CCL7 antibody can inhibit OSCC cell invasion in 3 dimensional cultures that allow studies of invasion with respect to cancer cell-fibroblast interaction; a closer similarity to in vivo conditions. The section of collagen gel devoid of CAF showed no infiltrative growth (Fig. 4b-a), whereas the section cultured with CAF embedded in a collagen gel showed definite infiltrative growth of cancer cells (Fig. 4b-b). However, the addition of anti-CCL7 antibody to the culture media (Fig. 4b-c) resulted in a marked reduction of cell invasion. These results demonstrate that CCL7 is involved in CAF-induced OSCC cell invasion.

Additionally, zymography showed that CCL7 treatment increased MMP-2 expression, which was further increased by cotreatment of CXCL-8 (Fig. 4c). This suggests that CXCL8 has a role in the increased OSCC invasion by CCL7 treatment.

Anti-CCR1 and anti-CCR3 antibodies inhibited OSCC cell migration induced by CCL7

OSCC cells must express receptors for CCL7 to undergo invasive growth induced by CCL7 released from CAF. CCL7 has been known to act through G-protein-coupled receptors, termed CCR1, CCR2, CCR3 and CCR5.11, 22, 23 We confirmed that CCR1, CCR2 and CCR3 were constitutively expressed by OSCC cells (HSC-2, Ca9.22 and YD-10B), including PC-3 prostate cancer cells, which served as a positive control (Fig. 5a). Interestingly, the expression of these receptors was increased in the PC-3 and OSCC cell lines, except YD-10B, by incubation with 100 ng/mL CCL7.

Figure 5.

Receptors for CCL7 in OSCC cells were shown to be CCR1 and CCR3. (a) Expression of CCR1, CCR2 and CCR3 in OSCC cells, and PC-3 prostate cancer cells was analyzed by RT-PCR. The specific 213, 200 and 268 bp PCR fragments detected CCR1, CCR2 and CCR3, respectively. PC-3 cells were used as controls. GAPDH was used for equal loading. (b) Confluent monolayers of HSC-2 cells were scraped with a sterile pipette tip, and the cells that migrated into the wounded monolayer were counted after 14 hr in the presence of 100 ng/ml CCL7. Anti-CCR1, anti-CCR2, anti-CCR3, or anti-CCR5 antibody (10 μg/ml) was tested for its antagonizing effect on cancer cell migration enhanced by CCL7. Control antibodies: Murine IgG2B antibody: anti-CCR1 and anti-CCR2; Rat IgG2A: anti-CCR3; murine IgG1: anti-CCR5. Data shown are representative of multiple repeats (means ± SD of quadruple assays; **p value of <0.01; *p value of <0.05).

To confirm that the observed CCL7-induced OSCC cell migration required the interaction between CCL7 and G-protein-coupled receptors, we used the wound healing assay to test the effect of anti-CCR1, anti-CCR2, anti-CCR3 and anti-CCR5 neutralizing antibodies on OSCC cell migration stimulated by CCL7 (Fig. 5b). Indeed, OSCC cell migration increased by 100 ng/mL CCL7 was inhibited by anti-CCR1 and anti-CCR3 antibodies, indicating that CCL7 may enhance migration through CCR1 and/or CCR3 in OSCC cells.

IL-1α secreted by OSCC cells stimulated CAF to induce CCL7 secretion

To find the soluble factor(s) secreted by OSCC cells that induces CCL7 secretion by CAF, we analyzed the expression levels of 36 cytokines (Supporting Information Figure S3) comparing 2 cell lines; Ca9.22 cells (showing the highest induction of CCL7 expression) and YD-38 cells (with the lowest induction as shown in Fig. 2a). This cytokine array showed that the expression levels of VEGF and IL-1α were higher in Ca9.22 cells than in YD-38 cells (Fig. 6a). Therefore, we speculated that VEGF and/or IL-1α might stimulate CAF to increase CCL7 secretion. In contrast, the cytokines IL-4, IL-12 and CCL5 showed higher expression levels in the YD-38 control cell line. However, except IL-4, IL-12 and CCL5 induced no significant increase of CCL7 secretion by CAF (Supporting Information Figure S4).

Figure 6.

IL-1α increased CCL7 secretion by CAF. (a) Culture medium collected from Ca9.22, which induces the highest level of CCL7 secretion among the OSCC cell lines tested, was analyzed by Western dot blot on 36 cytokine antibody arrays (Supporting Information Figure S3) to find the soluble factor(s) responsible for increasing CCL7 secretion. Analysis of culture medium collected from YD-38 served as a control. (b) CCL7 concentration in CAF culture media was measured by ELISA 24 hr after IL-1α treatment (at a concentration range of 1, 10, 100, 1,000 and 10,000 pg/mL). A specific antibody to IL-1α (1 and 10 μg/mL) was incubated with 10B-CM for 15 min before treating CAF with the CM. Data shown are representative of multiple repeats (means ± SD of triplicate assays; **p value of <0.01; *p value of <0.05). (c) Transient siRNA-mediated knockdown of the IL-1 receptor I protein in CAF. i, CCL7 release from control siRNA-treated and IL-1RI siRNA-treated CAFs was analyzed by ELISA. Fold change was calculated from noninfected control. The results are reported as the mean ± SD of quadruple assays (**p value of <0.01). ii, Expression of IL-1RI protein in noninfected, control siRNA-treated and IL-1RI siRNA-treated CAFs was analyzed by western blot analysis. β-Actin was used for equal loading. (d) Invasion of OSCC cells (YD-10B and Ca9.22) was dramatically reduced by anti-IL-1α neutralizing antibody. Anti-IL-1α antibody (1 μg/mL) was added to OSCC cells cocultured with CAF. Cell invasion was quantified by counting the invaded cells, which were counted in 5 separate microscopic fields per filter, and the mean values per filter (±SD) were calculated from 4 replicate filters. Murine IgG2A was used as a control antibody. Data shown are representative of multiple repeats (means ± SD of quadruple assays, **p value of <0.01). (e) a: negative control of murine IgG2A antibody (×200). b: Immunohistochemical staining showed IL-1α expression in cancer cells of human OSCC tissue (×200). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

When comparing the effects of VEGF and IL-1α by adding each cytokine or both to CAF cultures, we found that IL-1α induced a similar high level of CCL7 expression (pg/mL) in CAF to that stimulated by 10B-CM, whereas VEGF failed to enhance CCL7 expression (Fig. 6b). Interestingly, 1 ng/mL of VEGF has a synergic effect on CCL7 secretion induced by IL-1α (1 ng/mL), increasing the level of CCL7 secretion as high as that stimulated by 10B-CM. We noted that preincubation with an anti-IL-1α monoclonal antibody resulted in a marked reduction of CCL7 secretion induced by 10B-CM.

Furthermore, CAF subjected to siRNA-mediated knockdown of IL-1 receptor I (IL-1RI), a receptor for IL-1α, failed to increase CCL7 release by IL-1α, when compared with noninfected or control siRNA treated CAF (Fig. 6c). In addition, we demonstrated that an anti-IL-1α neutralizing antibody inhibited OSCC cell invasion induced by cocultured CAF (Fig. 6d). Immunohistochemical staining revealed that IL-1α expression was found in all cases of human OSCC tissue (Fig. 6e).

Discussion

Chemokine/chemokine receptor interactions play an essential role in leukocyte chemotaxis during inflammatory process. Recently, chemokine/chemokine receptor interactions are acknowledged to mediate the invasive response of cancer cells and further function in colonizing tumor foci in distant metastasis.23 Herein, we attempted to prove that chemokines and their receptors acts as a communication tool for crosstalk between cancer cells and stromal fibroblasts during the invasive growth of cancer cells.

Accordingly, we used a modified transwell invasion assay using a collagen gel-based coculture model of CAF and cancer cells to dissect the reciprocal molecular dialogue between cancer cells and stromal fibroblasts in OSCC. Gene expression profiling and real-time PCR analysis were used to identify the proinvasive soluble factors upregulated by the cocultured CAF that promoted OSCC cell invasion in the coculture model. CAF cocultured with OSCC cells showed prominent changes in the expression of a number of proinflammatory cytokines and chemokines, such as upregulation of CCL7, CXCL1, CXCL2, CXCL3 and CXCL8 precursor, when compared with CAF cultured alone. CCL7 was selected for further investigation because real-time PCR analysis showed that it exhibited the highest fold change in gene expression among the upregulated cytokines.

CCL7 was first isolated and characterized as a tumor-derived monocyte chemotactic protein from human osteosarcoma cells.24 Our data illustrate a role for stromal fibroblast-derived CCL7 in promoting the invasiveness of OSCC. To evaluate the role of CCL7 in cancer invasion, we demonstrated that CCL7 caused increased migration and cytoskeletal changes in OSCC cells. Neutralization of CCL7 by antibody treatment markedly attenuated the invasiveness of OSCC, evidenced by a modified transwell invasion assay and 3-dimentional raft culture. Corroborating our data, CCL7 showed no significant effect on cell growth (Supporting Information Figure S2), suggesting that the invasive growth induced by CCL7 is not due to cell proliferation.

In support of the role for CCL7 as a promoter of carcinoma progression, CCL7 derived from bone marrow stromal cells has been shown to act as a chemoattractant for homing human multiple myeloma cells to bone marrow.25 In contrast, the infection of tumor cells with CCL7-transducing parvovirus vector in syngeneic mouse melanoma model enhanced cell-mediated immunity through its ability to attract and activate a large panel of leukocytes, including natural killer cells and T lymphocytes.26 Taken together, the function of CCL7 in cancer tissue might be dependent on the nature of the target cells. In view of the fact that tumor cells producing leukocyte chemoattractants show a high invasive capacity,24 we envisaged that CCL7 in vivo has a Janus-like dual role in OSCC, i.e., inducing the invasive potential of cancer cells and chemoattracting immunocompetent cells.

Presumably, the invasiveness of OSCC promoted by CCL7 released from CAF can be inhibited by selective blockage of the receptors for CCL7. CCL7 acts on responsive cell types through G-protein coupled receptors such as CCR1, CCR2, CCR3 and CCR5.11, 22, 23 Importantly, we found that inhibiting CCR1 and CCR3 reduces CCL7-induced OSCC cell migration, implicating that CCL7 promotes cancer cell migration through those receptors in OSCC cells. CCR1, CCR2 and CCR3 were constitutively expressed in OSCC cell lines tested. Therefore, targeting the cognate receptor(s) of CCL7 on tumor cells could be beneficial in antitumor therapy.

The antibody cytokine array data showed that the soluble factors secreted from OSCC cells to elevate the expression of CCL7 in the stromal fibroblasts are IL-1α and VEGF, with a major role for IL-1α. IL-1α is among the most potent proinflammatory cytokines (together with TNFα) and has been shown to induce the expression of proinflammatory molecules and adhesion molecules in diverse stromal/inflammatory cells.27 HNSCC constitutively expresses IL-1α and have an autocrine effect, inducing the expression of other cytokines and enhancing cell proliferation.28–30 Interestingly, in our study, the IL-1α secreted from OSCC cells induces CCL7 secretion in CAF in a paracrine manner. In addition, VEGF interacted synergistically with IL-1α to increase the level of CCL7 release from CAF, whereas VEGF itself showed no effect on CCL7 secretion. siRNA-mediated knockdown of the IL-1α receptor in CAF inhibited their ability to produce CCL7. Moreover, inhibition of IL-1α by antibody treatment decreased the invasive growth of OSCC. From these results, we can suggest that inhibition of IL-1α secreted by OSCC could reduce carcinoma invasiveness, pointing to its feasibility in cancer therapy.

Although this study focused on the role of CCL7 and CCL7-related cytokines, other chemokines, such as CXCL1, CXCL2, CXCL3 and CXCL8, were also highly expressed by coculture of CAF and OSCC cells, as shown in Figure 1c. Considering the result that both treatment of CCL7 and CXCL8 increased MMP-2 expression, we can envisage that besides CCL7, other cytokines cooperate to promote cancer progression. For example, CXCL8 secreted from cancer cells and CXCL12 secreted from fibroblasts cooperatively promote invasive growth in pancreatic cancer.31 Also, it should be noted that other CAF-linked factors, for example, protease- and force-mediated matrix remodeling can regulate carcinoma invasion.32

In this study, we attempted to show a reciprocal dialogue between OSCC cells and stromal fibroblasts in promoting cancer progression. As a communication tool between OSCC cells and CAF, we demonstrated that IL-1α released from OSCC cells induced CCL7 secretion from CAF, resulting in cancer progression. Furthermore, we noted that blocking the cognate receptor (IL-1RI) in CAF abrogated OSCC cell invasion. Although we have focused on CCL7 in characterizing the interaction between CAF and OSCC cells in this study, our methodology for characterizing the interaction between carcinoma and stromal fibroblasts provides supportive evidence that stromal therapy can be a rewarding approach for cancer prevention and intervention.33

Ancillary