Cyclooxygenase-2 Up-regulates CCR7 via EP2/EP4 Receptor Signaling Pathways to Enhance Lymphatic Invasion of Breast Cancer Cells*

Recent studies demonstrate that cyclooxygenase-2 (COX-2) expression is frequently associated with lymph node metastasis. However, the mechanism by which COX-2 increases the invasion of cancer cells to lymph node is unclear. CCR7 is a chemokine receptor that plays important roles in the mediation of migration of leukocytes and dendritic cells toward lymphatic endothelial cells (LECs) that express receptor ligand CCL21. We found that treatment of prostaglandin E2 or ectopic expression of COX-2 in MCF-7 cells up-regulated CCR7 expression. On the contrary, knockdown of COX-2 by small hairpin RNA reduced CCR7 in COX-2-overexpressing MDA-MB-231 cells. Interaction of CCR7 and CCL21 was important for the migration of breast cancer cells toward LECs because antibodies against these two molecules inhibited the migration. We also found that COX-2 increased CCR7 expression via the EP2 and EP4 receptor in breast cancer cells. EP2 and EP4 agonists stimulated CCR7 in MCF-7 cells, whereas antagonists or small hairpin RNA of EP2 and EP4 attenuated CCR7 in MDA-MB-231 cells. Protein kinase A and AKT kinase were involved in COX-2-induced CCR7. Pathological analysis demonstrated that COX-2 overexpression was associated with CCR7, EP2, and EP4 expressions in breast tumor tissues. In addition, CCR7 expression in COX-2-overexpressing tumors was significantly correlated with lymph node metastasis. Collectively, we suggest that CCR7 is a down-stream target for COX-2 to enhance the migration of breast cancer cells toward LECs and to promote lymphatic invasion.

Cyclooxygenases (COXs) 2 are the rate-limiting enzymes that catalyze the conversion of arachidonic acid to prostaglandins (PGs). Two COX isoforms with distinct tissue distributions and physiological functions have been identified (1,2). Cyclooxygenase-1 (COX-1) is constitutively expressed in many tissues and plays important roles in the control of homeostasis (3). Conversely, COX-2 is an inducible enzyme and is activated by extracellular stimuli such as growth factors and pro-inflammatory cytokines (4). Recent investigations indicated that overexpression of COX-2 is frequently found in many types of cancer, including colon, lung, breast, pancreas, and head and neck cancers (5)(6)(7)(8)(9), and is usually associated with poor prognosis and short survival (10 -12).
The contribution of COX-2 to tumorigenesis has been intensively studied. Several mechanisms are considered to mediate the tumorigenic activity of COX-2. First, PGE 2 , the main metabolite of COX-2, is a growth promoter and may directly stimulate proliferation of cancer cells (13,14). Second, COX-2 is an angiogenic stimulator and may increase the production of angiogenic factors and migration of endothelial cells (15,16). Third, COX-2-derived PGE 2 is an anti-apoptotic molecule that may prevent apoptosis induced by anti-cancer drugs (17)(18)(19). Fourth, PGE 2 is an immunoregulatory molecule that may suppress the anti-tumor activity of natural killer cells and macrophages (20). Fifth, COX-2 expression may increase the invasive ability of tumor cells and is closely linked with lymph node metastasis (21,22). Results of xenograft animal studies also indicated that inhibition of COX-2 decreased tumor growth and lymph node metastasis (23). Whereas the functional role of COX-2 in the promotion of tumor angiogenesis is well documented, the mechanism by which COX-2 enhances lymphangiogenesis and lymph node metastasis remains unknown. A possible mediator that participates in the COX-2-induced lymphangiogenesis is vascular endothelial growth factor-C (VEGF-C). Ectopic expression of COX-2 in breast and lung cancer cells up-regulates VEGF-C, which may stimulate VEGF receptor 3 (VEGFR3) on the surface of LECs to promote lymphangiogenesis (24,25).
The interaction between chemokines and their cognate receptors is critical in tumor metastasis. Recent evidence demonstrates that the chemokine receptor CXCR4 and its ligand stroma-derived factor-1 are key regulators for the dissemination of cancer cells (26,27). The chemokine receptor CCR7 is important for the adhesion and chemotaxis of leukocyte and dendritic cell toward lymph nodes. Since LECs express a high level of CCR7 ligands CCL19 and CCL21, up-regulation of CCR7 in cancer cells may promote the migration of cancer cells toward LECs and enhance lymph node invasion. In this study, we addressed the following objectives: (a) correlation of expression of COX-2 and CCR7 in breast cancer cell lines and tumor tissues, (b) whether COX-2 may up-regulate CCR7 expression in breast cancer cells and promote their migration toward LECs, and (c) the receptors and signaling pathways that mediate the induction of CCR7 by COX-2. Tumor Tissues-Seventy nine paired normal and breast tumor tissues were obtained from patients who underwent resection of tumors at the Department of Surgery, Chung-Ho Memorial Hospital, Kaohsiung Medical University. Detailed data about patient-and tumor-related variables were collected by reviewing the patients' medical charts. Before acquisition of these tissues, the investigational nature of this study was explained to patients, and informed consent was obtained. Por- tions of resected tissues were quickly placed into the RNAlater solution (Ambion, Austin, TX) and were subjected for isolation of genomic DNA, total RNA, and proteins by using TRIzol reagent according to the procedures of the manufacturer (Invitrogen).
Immunoblotting-Treated cells were washed with ice-cold phosphate-buffered saline and harvested A, MCF-7 cells were transfected with control (C) or COX-2 expression vector. After 48 h, cells were collected and subjected to in vitro invasion assay as described under "Experimental Procedures." Non-immune immunoglobulin (Ig) or anti-CCR7 antibody (5 g/ml) was added with cells into the upper part of the transwell unit, and the conditioned medium of LECs was added in the lower part. Invaded cell number was determined at 24 h after cell seeding. ૺ, p Ͻ 0.05 when anti-CCR7 group was compared with Ig group. B, MDA-MB-231 cells was incubated with Ig or anti-CCR7 antibody (5 g/ml), and cell invasion was studied as described above. ૺ, p Ͻ 0.05 when anti-CCR7 group was compared with Ig group. C, conditioned medium of LECs was pre-cleared with anti-CCL21 antibody (5 g/ml) for 4 h at room temperature and then added into the lower part of the transwell unit. MCF-7 or MDA-MB-231 (231) cells were seeded on the upper part of the unit, and invaded cell number was counted at 24 h after cell seeding. ૺ, p Ͻ 0.05 when anti-CCL21 group was compared with Ig group. D, MCF-7 and MDA-MB-231 stable transfectants expressing COX-2 or CCR7 shRNA were used for in vitro invasion assay. ૺ, p Ͻ 0.05 when shCOX-2 or shCCR7 group was compared with control vector group. E, inhibition of CCR7 expression by shRNA in two breast cancer cell lines is shown.
in a lysis buffer as described previously (28). Equal amounts of cellular proteins were separated by SDS-PAGE on 10 or 12.5% gels. Proteins were transferred onto nitrocellulose membranes, and the blots were incubated with different primary antibodies. Enhanced chemiluminescence reagents were used to depict the protein bands on the membrane.
Flow Cytometry-For detection of CCR7 protein expression on the cell surface, cells were washed with ice-cold phosphatebuffered saline and incubated with biotinylated anti-CCR7 antibody at 4°C for 30 min. After washing, cells were incubated with fluorescein isothiocyanate-conjugated avidin and subjected to flow cytometric analysis as described previously (29).
In Vitro Invasion Assay-In vitro invasion assay was performed by using 24-well transwell units with polycarbonate filters (pore size 8 m) coated on the upper side with Matrigel (Discovery Labware) (30). MCF-7 and MDA-MB-231 cells were collected, and 3 ϫ 10 3 cells in 100 l of medium with control immunoglobulin or anti-CCR7 antibody (5 g/ml) were placed in the upper part of the transwell unit and allowed to invade for 24 h. The lower part of the transwell unit was filled with LEC-conditioned medium. In some experiments, LEC-conditioned medium was first incubated with anti-CCR21 (5 g/ml) to neutralize CCL21 and then placed on the lower part for invasion study. After incubation, non-invaded cells on the upper part of the membrane were removed with a cotton swab. Invaded cells on the bottom surface of the membrane were fixed in formaldehyde, stained with Giemsa solution, and counted under a microscope. Invasive ability of stable transfectants of MDA-MB-231 cells expressing COX-2, CCR7, EP2, or EP4 shRNA was also investigated.
Statistical Analysis-The associations between COX-2 and CCR7 with clinicopathological parameters were assessed using 2 test and Fisher's exact test. The correlations between COX-2 expression with CCR7, EP2, and EP4 were examined by Spearman rank correlation. Paired t test was performed to test the association of COX-2 and CCR7 up-regulation in tumor tissues. Wilcoxon rank sum test was used for testing COX-2 and CCR7 levels of lymph node-negative tumors with lymph nodepositive tumors. Statistical significance was defined as p Ͻ 0.05. , sulprostone (Sul, 10 mol/liter), butaprost (But, 10 mol/liter), PGE 1 alcohol (E1, 1 mol/liter), and PGE 2 (E2, 100 nmol/liter) for 24 h, and CCR7 expression was studied by quantitative RT-PCR. ૺ, p Ͻ 0.05 when drug-treated group was compared with vehicle group. C, MDA-MB-231 cells were treated with vehicle (V, 0.1% DMSO), AH23848 (10 mol/liter), or AH6809 (50 mol/liter) for 24 h, and quantitative RT-PCR was performed to investigate CCR7 mRNA expression. ૺ, p Ͻ 0.05 when drug-treated group was compared with vehicle group. Flow cytometry was also carried out to study the CCR7 protein level (lower panel). D, MDA-MB-231 stable transfectants expressing EP2 shRNA (EP2i) or EP4 shRNA (EP4i). EP2 and EP4 protein level was studied by Western blot analysis (left upper panel). CCR7 mRNA level of control and EP2i and EP4i was investigated by quantitative RT-PCR (right upper panel). ૺ, p Ͻ 0.05 when shRNA-treated group was compared with control vector group. Flow cytometry was also carried out to study the CCR7 protein level of control (con), EP2i, or EP4i cells (lower panel). E, invasive ability of control (con), EP2i, or EP4i cells was studied. ૺ, p Ͻ 0.05 when shRNA-treated group was compared with control vector group.

Up-regulation of CCR7 by COX-2 in Breast Cancer Cell Lines-
We first compared the expression of COX-2 and CCR7 in nonmetastatic MCF-7 and highly metastatic MDA-MB-231 breast cancer cells. As demonstrated in Fig. 1A, RT-PCR analysis showed that CCR7 was highly expressed in COX-2-overexpressing MDA-MB-231 cells. Conversely, the expression of CCR7 was low in MCF-7 cells that expressed low levels of COX-2. To confirm the RT-PCR results, we performed flow cytometry to investigate the CCR7 protein level on the cell surface. A positive correlation between CCR7 mRNA and protein expression was seen in these two breast cancer cell lines (Fig.  1A). We next used different approaches to clarify whether COX-2 could regulate CCR7 expression in breast cancer cells. Since MCF-7 cells expressed low level of COX-2 and CCR7, we treated these cells with PGE 2 and found that PGE 2 increased CCR7 mRNA and protein expression in MCF-7 cells (Fig. 1B). Ectopic expression of COX-2 also increased CCR7 expression in these cells (Fig. 1C). MDA-MB-231 cells expressed high levels of COX-2; we used shRNA to knock down COX-2 expression, and our data showed that inhibition of COX-2 by shRNA induced down-regulation of CCR7 (Fig. 1D). Consistent with this result, a COX-2-specific inhibitor NS398 also reduced the expression of CCR7 in MDA-MB-231 cells (Fig. 1E). By manipulating the expression of COX-2 in different breast cancer cell lines, we concluded that COX-2 and its metabolite PGE 2 might stimulate CCR7 expression.
CCR7 Expression Enhanced the Migration of Breast Cancer Cells toward LECs-We next addressed whether CCR7 is required for the migration of breast cancer cells toward LECs. Primary cultured LECs used in these assays expressed high levels of CCL21 (data not shown). Conditioned medium of LECs was placed on the lower part of a transwell unit, and breast cancer cells were added to the upper part in the absence or presence of anti-CCR7 antibody. As shown in Fig. 2A, the migration ability of MCF-7 was poor. Ectopic expression of COX-2 significantly enhanced the migration ability of MCF-7 cells that could be specifically blocked by the anti-CCR7 antibody. MDA-MB-231 cells were highly invasive, and the migration ability was also significantly attenuated by the anti-CCR7 antibody (Fig. 2B). Similar results were observed when a co-culture system was used by seeding the LECs on the lower part of the transwell unit (data not shown). This effect was not non-specifically caused by anti-CCR7 antibody because this antibody did not exhibit cytotoxic effect on breast cancer cells (supplemental Fig. 1). To verify the migration of breast cancer cells was induced by the interaction between CCR7 and CCL21, conditioned medium was pre-cleared by anti-CCL21 antibody and then used for the migration assay. As shown in Fig. 2C, we found that migration of MCF-7 and MDA-MB-231 cells into a lower unit was significantly reduced. We also used shRNA to knock down COX-2 or CCR7 expression in these two cell lines and found that migration ability of the stable transfectants was attenuated (Fig. 2, D and E). In addition, the reduction was more significant in MDA-MB-231 cells (Fig. 2D). Our data suggested that up-regulation of CCR7 by COX-2 in breast cancer cells enhanced their migration toward LECs.

PGE 2 and COX-2 Up-regulated CCR7 via EP2 and EP4
Receptor-Four types of G protein-coupled EP receptor have been identified in mammalian cells. Therefore, we addressed which type of receptor was involved in PGE 2 -induced increase of CCR7. RT-PCR analysis demonstrated that MCF-7 and MDA-MB-231 expressed all types of EP receptor (Fig. 3A). Both cell lines expressed similar levels of EP1 and EP3. Conversely,  APRIL 25, 2008 • VOLUME 283 • NUMBER 17

JOURNAL OF BIOLOGICAL CHEMISTRY 11159
EP2 and EP4 were up-regulated in MDA-MB-231 cells. Since MCF-7 cells expressed very low levels of endogenous COX-2, we used ligand stimulation assay to study the specific receptors that mediated the induction of CCR7 by COX-2. Addition of PGE 2 , butaprost (EP2 agonist), or PGE 1 alcohol (EP4 agonist) up-regulated CCR7 expression by 2-3-fold in MCF-7 cells (Fig.  3B). Conversely, the EP1/EP3 agonist sulprostone had little effect. We next performed antagonist inhibition assay to test whether EP2 and EP4 receptors were involved in the autocrine induction of CCR7 in COX-2-overexpressing MDA-MB-231 cells. Our data demonstrated that the EP2 antagonist AH6809 and the EP4 antagonist AH23848 suppressed CCR7 expression in MDA-MB-231 cells (Fig. 3C). To clarify the role of EP2 and EP4 more clearly, we used shRNA to knock down EP2 and EP4 expression. As shown in Fig. 3D, knockdown of EP2 and EP4 protein led to down-regulation of CCR7 mRNA and protein level in MDA-MB-231. In addition, the invasive ability of EP2 or EP4 knockdown cells was significantly reduced (Fig. 3E). These results strongly suggested that COX-2 up-regulated CCR7 via EP2 and EP4 receptor in breast cancer cells.
PKA and AKT Were Involved in the Induction of CCR7 by COX-2-The common signaling effector activated by EP2 and EP4 receptor is PKA. Therefore, we investigated the functional role of PKA in COX-2-induced CCR7. Stimulation of MCF-7 cells by dibutyryl cAMP, a potent PKA activator, up-regulated CCR7 mRNA and protein expression (Fig. 4A). Forskolin, an adenylyl cyclase and PKA activator, also increased CCR7 expression in MCF-7 cells (supplemental Fig. 2). Pretreatment of a PKA inhibitor KT5720 potently suppressed dibutyryl cAMP-induced CCR7 (Fig.  4B). Similarly, KT5720 also attenuated CCR7 expression in MDA-MB-231 cells (Fig. 4C). To address the functional role of PKA more specifically, we ectopically expressed the regulatory subunit of PKA to inhibit its activity in MDA-MB-231 cells and found that it caused a decrease of CCR7 expression (supplemental Fig. 3). Moreover, the increase of PKA regulatory subunit in MCF-7 cells also reduced butaprost (EP2 agonist)-induced and PGE 1 alcohol (EP4 agonist)-induced CCR7 expression (supplemental Fig. 4). EP4 receptor has been found to activate ERK via phosphatidylinositol 3-kinase (31). However, treatment of phosphatidylinositol 3-kinase and ERK inhibitor (LY294002 and U0126) did not affect CCR7 expression in MDA-MB-231 cells (Fig. 4D). A recent study demonstrated that stimulation of the EP2 receptor induced PKA-dependent and phosphatidylinositol 3-kinase-independent activation of AKT (31). In addition, PKA phosphorylated Thr-308 but not Ser-473 (two major activating phosphorylation sites) on AKT (32). Therefore, we studied whether AKT was involved in the induction of CCR7 by COX-2. Fig. 5A showed that PGE 2 and dibutyryl cAMP increased the phosphorylation of Thr-308 of AKT in MCF-7 cells. Pretreatment of PKA inhibitor KT5720 totally abolished this effect. Our data also showed that PGE 2 -increased CCR7 expression in MCF-7 cells was attenuated by dominant-negative AKT (Fig. 5B). In addition, dominant-negative AKT also reduced CCR7 expression in MDA-MB-231 cells (Fig. 5C). PKA inhibitor KT5720 reduced Thr-308 but not Ser-473 phosphorylation of AKT in MDA-MB-231 cells (Fig. 5D). More importantly, EP2 and EP4 receptor antagonists AH23848 and AH6809 also inhibited Thr-308 phosphorylation. These data suggested that COX-2 induced CCR7 expression via a PKA/ AKT-dependent pathway.
COX-2, CCR7, EP2, and EP4 Were Co-expressed in Breast Tumor Tissues-Our study on breast cancer cell lines clearly indicated that CCR7 was up-regulated by COX-2. In addition, we also found that EP2 and EP4 receptors were up-regulated in COX-2-overexpressing MDA-MB-231 cells. Therefore, we investigated whether COX-2 and CCR7 were co-expressed in breast tumor tissues as seen in breast cancer cell lines. Breast tumor tissues and their adjacent normal parts from 79 patients were studied. RT-PCR analysis of five tumor tissues is shown in   6. The expression levels of COX-2 and CCR7 of normal and tumor tissues were normalized with GAPDH and compared. We found that COX-2 and CCR7 were increased in 44.3% (35/ 79) and 51.9% (41/79) of breast tumor tissues, respectively. 2 analysis indicated that CCR7 expression was significantly asso-ciated with COX-2 expression (p ϭ 0.008; Table 1). COX-2 expression also significantly correlated with EP2 (p ϭ 0.024) and EP4 (p ϭ 0.008) expression. Spearman rank correlation test confirmed that COX-2 expression strongly associated with CCR7 (coefficient ϭ 0.528, p ϭ 0.000), EP2 (coefficient ϭ 0.235, p ϭ 0.037), and EP4 (coefficient ϭ 0.229, p ϭ 0.042). Our data indicated that COX-2 overexpression was associated with up-regulation of CCR7, EP2, and EP4 in breast cancer tissues as found in breast cancer cell lines.

COX-2 and CCR7 Were Up-regulated in Breast Tumor Tissues and Associated with Lymph Node
Metastasis-Paired t test clearly indicated that expression of COX-2 and CCR7 was increased in tumor tissues (p ϭ 0.031 and 0.047, respectively). So, we investigated the correlation of COX-2 and CCR7 with clinicopathological parameters. As shown in Table 2, COX-2 expression was not associated with tumor size, histological grade, progesterone receptor, HER-2/neu, and histological type. On the contrary, COX-2 expression was positively associated with lymph node metastasis (p ϭ 0.030) and negatively correlated with estrogen receptor (p ϭ 0.044). CCR7 expression was not correlated with tumor size, histological grade, HER-2/neu, and histological type. However, CCR7 was negatively correlated with estrogen receptor and progesterone receptor. The lack of association of CCR7 with lymph node metastasis was unexpected. We focused on the expression of COX-2 and CCR7 in tumor tissues and compared them in lymph node-negative and lymph node-positive tumors by using Wilcoxon rank sum test. Our data showed that COX-2 expression was more frequently found in tumors with lymph node metastasis (p ϭ 0.040). Similar to the results of 2 test, no significant association of CCR7 and lymph node metastasis was observed (p ϭ 0.355). However, we found that CCR7 expression was significantly associated with lymph node metastasis (p ϭ 0.048) in COX-2-overexpressing tumors when a serial testing was carried out. These data strongly suggested that CCR7 was a critical determinant to predict lymph node metastasis in COX-2-overexpressing tumors. Collectively, our results indicated that COX-2 and CCR7 were up-regulated in breast cancer and were associated with enhancement of lymph node metastasis.

DISCUSSION
The importance of our study is that we provide a mechanistic insight by which COX-2 promotes breast cancer cells to invade  into the lymphatic system. Previous studies have demonstrated that COX-2 expression was associated with lymph node metastasis in breast cancer (21,22). In addition, COX-2 is a negative prognostic factor for disease-free survival and overall survival in patients with breast cancer (22). Therefore, it is predictable that COX-2 may contribute to enhancement of lymph node metastasis. However, the underlying mechanism is unknown. Recently, a candidate gene for the mediation of lymph node invasion or lymphangiogenesis by COX-2 has been suggested. Timoshenko et al. (25) showed that COX-2 expression was strongly correlated with VEGF-C expression in breast cancer. VEGF-C is one of the most potent growth factors for the induction of lymphangiogenesis because the LECs express high amounts of its cognate receptor VEGFR3. The authors showed that COX-2 acted via EP1 and EP4 receptor to increase VEGF-C expression. Furthermore, a selective COX-2 inhibitor celecoxib potently decreased tumor growth and recurrence by inhibiting COX-2-induced lymphangiogenesis (23). Similar results have also been found in human lung cancer cell lines and tumor tissues (24). Because COX-2-induced lymph node metastasis is a complex process, we think VEGF-C is not the only effector of COX-2. In this study, we identified another key player CCR7. A pioneer study demonstrated that the chemokine receptors were involved in breast cancer metastasis, and CCR7 was found to be one of the most consistent up-regulated chemokine receptors in breast tumor tissues (26). In agreement with these data, we found that CCR7 was up-regulated in 51.9% of breast tumor tissues. Another recent study showed that CCR7 was a novel biomarker for predicting lymph node metastasis in T1 breast cancer (33). Although the association of COX-2 and CCR7 with lymph node metastasis of breast cancer has been separately reported, the functional cross-link between these two genes in the mediation of lymph node metastasis was missing. We provide the first evidence that COX-2 acts via EP2 and EP4 receptor to up-regulate CCR7 expression in breast cancer cells. The increase of CCR7 was important for the migration and invasion of breast cancer cells toward LECs because anti-CCR7 antibody significantly attenuated this effect. Moreover, knockdown of CCR7 in COX-2-overexpressing MDA-MB-231 cells also resulted in reduction of LEC-induced migration (Fig. 2D). These data strongly suggest that CCR7 is a critical mediator for COX-2 to promote lymph node metastasis.
The signaling pathways that participated in COX-2-induced CCR7 is another important issue. We found that PKA and AKT were involved in the up-regulation of CCR7 by COX-2. Because activation of EP2 and EP4 increases intracellular cAMP, it is not surprising that PKA is a downstream effector for COX-2 to increase CCR7. Interestingly, similar observations had also been reported in dendritic cells (34). Dendritic cells are antigen-presenting cells and are able to migrate from peripheral tissues to lymph node (35). Scandella et al. (34) demonstrated that PGE 2 increased cAMP level and CCR7 expression in dendritic cells. We extend these results by showing that AKT is a downstream mediator of PKA because PGE 2 , dibutyryl cAMP,  and forskolin increased phosphorylation of Thr-308 of AKT, whereas PKA inhibitor KT5720 totally abolished this effect. In addition, ectopic expression of the regulatory subunit of PKA or dominant-negative AKT potently attenuated CCR7 expression. These results suggest that COX-2 acts via a PKA/AKT-dependent pathway to up-regulate CCR7. The most well defined transcription factor that controls CCR7 expression is NF-B (36). The functional link between AKT and NF-B is now under investigation in our laboratory. An important finding of our pathological analysis is the association of COX-2 and CCR7 with lymph node metastasis. Our cell-based study indicates that CCR7 is one of the critical mediators for COX-2 to enhance lymph node invasion. However, the clinical relevance needs to be verified. We studied the expression of COX-2 and CCR7 in breast tumor tissues, and our results confirmed a strong correlation between COX-2 and lymph node metastasis. Since COX-2 exhibits proliferative, angiogenic, and anti-apoptotic activities, it is not surprising that COX-2 is associated with lymph node metastasis. However, we found that CCR7 alone is not significantly linked with lymph node metastasis. Conversely, CCR7 expression in COX-2-expressing tumors is an important predictor for lymph node metastasis. The results suggest that COX-2 may increase CCR7 expression and cooperate with CCR7 to promote lymph node metastasis. Interestingly, we also found that the CCR7 level was negatively associated with progesterone receptor and estrogen receptor. Whether steroid hormones like estrogen and progesterone may regulate CCR7 expression in dendritic cells or cancer cells is still unknown. It is a critical issue for future studies.
The results of this study are of great clinical significance. Several strategies for the prevention or treatment of lymph node metastasis in breast cancer are suggested. Because COX-2 is an upstream regulator for CCR7, targeting of COX-2 may possibly reduce CCR7 expression and attenuate metastasis. This hypothesis is strongly supported by our results (Fig. 1, D  and E). We think non-steroidal anti-inflammatory drugs and COX-2 inhibitors are good candidates for the prevention or treatment of COX-2-induced lymph node metastasis because these drugs may simultaneously repress the expression of VEGF-C and CCR7 stimulated by COX-2. Another choice is the antibody or specific inhibitors that can block the interaction between CCR7 and its cognate ligand CCL21 and CCL19. Identification of the functional importance of stroma-derived factor-1/CXCR4 interaction in tumor metastasis has led to the intense development of drugs that may inhibit this interaction. Similar concepts can be adapted to the prevention of CCL21/ CCR7 interaction. This hypothesis is supported by our results that anti-CCL21 and CCR7 antibodies potently suppress the migration of breast cancer cells toward LECs (Fig. 2, B and C). Collectively, we identify CCR7 as a downstream mediator for COX-2 to promote lymphatic invasion and suggest that inhibition of COX-2 and CCL21/CCR7 interaction may be helpful for the treatment of lymph node metastasis.