The role of CXCR3 and CXCR4 in colorectal cancer metastasis

Authors


Abstract

Chemokines and their receptors play key roles in leukocyte trafficking and are also implicated in cancer metastasis. We previously demonstrated that forced expression of CXCR3 promotes colon cancer metastasis preferentially to the draining lymph nodes (LNs), with poor prognosis. Using clinical colorectal cancer (CRC) samples, here, we show that expressions of CXCR3 and CXCR4 are significantly higher in metastatic foci within LNs and liver compared to primary tumors, whereas ligands for CXCR3 and CXCR4 are not. We also have demonstrated that some human CRC cell lines constitutively express both CXCR3 and CXCR4, and that activation of CXCR3 strengthens the CXCR4-mediated cell migration in vitro in a synergistic manner. By constructing SW620 cell lines with reduced expression of CXCR3 and/or CXCR4 using microRNA, we investigated in vivo metastatic activities in a mouse rectal transplantation model. Six weeks after inoculation, CXCR3-, CXCR4-, and CXCR3/CXCR4 double-knockdowns significantly reduced metastasis to LNs, liver and lungs, compared to the control (p < 0.05). Importantly, its suppressive effect on LN metastasis was significantly stronger in CXCR3- and CXCR3/CXCR4 double-knockdowns. In addition, CXCR3- and CXCR3/CXCR4 double-knockdowns significantly decreased the dissemination of cancer cells to liver and lungs, even after 2 weeks. These results indicate that targeting CXCR3 and CXCR4 can be a promising therapy against CRC metastasis.

Colorectal cancer (CRC) is the third commonest cancer in the world and the fourth of cancer-related deaths,1 with the majority attributable to distant metastases. The 5-year survival rate for patients with local CRC is ∼80–90%, whereas that of those with distant metastasis is only 10–20%.2 The addition of molecular targeted agents, such as anti-vascular endothelial growth factor (VEGF)-based and anti-epidermal growth factor receptor (EGFR)-based antibodies, to the existing cytotoxic chemotherapy have changed clinical practice of CRC. However, the overall survival benefits remain modest despite exceedingly high financial costs. Clearly, development of novel approaches to maximize the efficacy of available treatments remains a major priority in metastatic CRC.

Chemokines are chemotactic cytokines that direct the migration of chemokine receptor-expressing cells. They are structurally related, small-polypeptide signaling molecules that bind to, and activate a family of G-protein-coupled receptors. Besides their functions in the immune system, chemokines and their receptors also play a critical role in tumor initiation, progression and metastasis. Of the different components that affect the organ selectivity, instrumental roles have been attributed to interactions between chemokine receptors on cancer cells and their ligands in the target organs.3 CRC cells have been found to express the chemokine receptors CXCR4 and CXCR3.4–9

Chemokine receptor CXCR4 is the first identified chemokine receptor to play a critical role in determining the metastatic destination of breast cancer to the bone and lungs where its ligand CXCL12 is abundant.10, 11 It was also reported that metastasis of mouse colon cancer cell line CT-26 to the liver and lungs was greatly reduced when CXCR4 function was blocked.4 In clinical studies, CXCR4 expression in CRC was shown to increase the recurrence, poor survival and liver metastasis.5 In addition, we recently demonstrated that CXCL12 secreted by myofibroblasts derived from hepatic stellate cells promotes liver metastasis of CRC through the CXCL12/CXCR4 axis, and that a CXCR4 antagonist, AMD3100, significantly suppressed tumor formation in the liver.12 This shows preclinical evidence that blockade of the CXCL12/CXCR4 axis is a possible therapeutic target in the treatment of metastatic CRC.

CXCR3 also plays an important role in metastasis of several types of cancers including melanoma, CRC and breast cancer.6–9, 13–15 In CRC, it was reported that CXCR3 expression is upregulated in metastatic colon cancer cells, but not in their primary lesions, and that the CXCL10/CXCR3 axis significantly upregulated invasion-related properties.7 In addition, expression profiles of chemokine receptors reported that both CXCR3 and CXCR4 were significantly upregulated in colorectal liver metastases compared to the corresponding primary CRC, whereas CXCR1 and CXCR2 were downregulated.16 It is notable that systemic administration of a CXCR3 inhibitor, AMG487, was recently reported to inhibit lung metastasis of CRC and breast cancer in a mouse model.8, 14 We previously demonstrated that forced expression of CXCR3 promotes colon cancer metastasis preferentially to the draining lymph nodes (LNs) with poor prognosis.6 In 92 human colon cancer specimens, CXCR3 was expressed in 31 (34%) cases, and correlated with the metastatic frequencies to LNs and distant organs. Importantly, the patients with CXCR3-positive colon cancer exhibited shorter survivals than those without CXCR3. The patients with CXCR4-positive tumors also had a significantly poorer prognosis, but its effect was less than that of CXCR3. In addition, the patients with tumors double positive for CXCR3 and CXCR4 had a significantly poorer prognosis than those with tumors positive only for CXCR4 or double negatives.

It remains unclear how the organ-specific metastasis is regulated, despite increasing knowledge about the involvement of CXCR3 and CXCR4 in the invasion and metastasis of several cancers. This is the first study to simultaneously investigate the roles of CXCR3 and CXCR4 in cancer metastasis. Here, we have demonstrated that both CXCR3 and CXCR4 are involved in metastasis of CRC to LNs, lungs and liver using CXCR3-, CXCR4- and CXCR3/CXCR4 double-knockdown clones, and that activation of CXCR3 pathway strengthens the function of CXCR4 in a synergistic manner. These studies further indicate that both CXCR3 and CXCR4 can be potential therapeutic targets for suppressing CRC metastasis.

Material and Methods

Cell lines, reagents and tissue samples

DLD-1 and SW480 were supplied from Cell Resource Center for Biomedical Research, Tohoku University, HCT116 and HT29 were from American Type Culture Collection and SW620 were from European Collection of Cell Cultures. Five- to six-week-old male KSN/Slc nude mice were obtained from Japan SLC (Hamamatsu, Japan). Recombinant chemokines were purchased from R&D Systems (Minneapolis, MN). Anti-CXCR3, anti-CXCR4 and anti-MMP9 monoclonal antibody (mAbs) were obtained from R&D Systems. Anti-AKT, anti-phospho-AKT (Ser473), anti-p44/42 and anti-phospho-p44/42 MAPK Abs were obtained from Cell Signaling Technology (Beverly, MA). Surgical specimens of CRC were collected, with informed consents, upon surgery at Kyoto University Hospital. Histopathological diagnosis was confirmed for each specimen. This study protocol was approved by the institutional review board of Kyoto University, and patients provided their consent for data handling.

Quantitative Reverse transcription polymerase chain reaction (RT-PCR) analysis

Total RNAs were extracted from frozen surgical specimens using ISOGENE (Nippon Gene, Tokyo, Japan). Complementary DNA generated by reverse transcription was quantified using LightCycler 480 and FastStart Universal SYBR Green Master (Roche Applied Sciences, Indianapolis, IN). mRNA levels of tested genes were normalized to GAPDH levels using a ΔCt method. To account for low background expression by normal colonic tissues, the ratio for primary and metastatic tumor specimens were normalized with respect to the mean expression ratios for normal colonic tissues. The primers for human CXCR3 were 5′-ACCACAAGCACCA AAGCAG-3′ and 5′-GGCGTCATTTAGCACTTGGT-3′; for human CXCR4, 5′-CAGCAGGTAGCAAAGTGACG-3′ and 5′-ATAGTCCCCTGAGC-CCATTT-3′; for human CXCL9, 5′-GTGGTGTTCTTTTCCTCTTG-3′ and 5′-GTAGG-TGGA TAGTCCCTTGG-3′; for human CXCL10, 5′-AGCAAGGA AAGGTCTAAAAG-ATCTCC-3′ and 5′-GGCTTGACATAT ACTCCATGTAGGG-3′; for human CXCL11, 5′-GCTATAG CCTTGGCTGTGATA-3′ and 5′-CTGCCACTTTCACTGCT TTTA-3′; for human CXCL12, 5′-CCCGTCAGCCTGAGCT ACAG-3′ and 5′-CGTTGGCTCTGG-CAACATG-3′ and for human GAPDH, 5′-ATGGGGAAGGTGAAGGTCG-3′ and 5′-GGGGTCATTGATGGCAACAATA-3′.

Immunohistochemistry

Formalin-fixed, paraffin-embedded sections were stained with anti-human CXCR3 mAb and anti-human CXCR4 mAb by the avidin–biotin immunoperoxidase method. Microwave antigen retrieval was performed. To investigate the number of CXCR3- or CXCR4-positive cells, we first chose three arbitrary sites under the microscope (original magnification, 400×) and counted the cancer cells, as described previously.12 We then calculated the proportion of CXCR3- or CXCR4-positive cells within the counted region.

Western blot analysis

Cells incubated on dishes were lysed with lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 50 mM NaF, 1% NP40 and protease inhibitors). For stimulation experiments, cells were serum-starved for 24 hr and then stimulated for the indicated times with 100 ng/ml of CXCL10 or/and CXCL12. Cell lysates were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE), immunoblotted with the respective Abs, followed by horseradish peroxidase (HRP)-conjugated secondary Abs, and analyzed.

Matrix metalloproteinase expression

Cells were serum-starved for 24 hr, rinsed with phosphate buffered saline (PBS), and then incubated with 100 ng/ml of CXCL10 or CXCL12 for 48 hr. Cell supernatants were collected and 15 μg of samples were subjected to Western blot analysis using anti-MMP9 mAb.

Chemotaxis assay

Migration was assayed in 24-well Transwell cell culture chambers (8-μm pore; Coster, Cambridge, MA). After cells were added to the upper chamber (4 × 104 cells/well) and incubated for 24 hr, cells attached on the lower surface of the membrane were counted in at least ten different fields (original magnification, 200×). Chemotaxis index was defined as the ratios of migrating cell numbers in the experimental groups divided by those in the controls. At least three experiments were performed for each set. Chemokinesis was tested in checkerboard assays and was negative. For neutralization studies, cells were incubated with anti-CXCR3 mAb in the upper chamber.

Proliferation assay

Cell proliferation was measured by WST-8 colorimetric assay (Cell Counting Kit-8; Dojindo, Tokyo, Japan). In brief, 1 × 104 cells were seeded in 96-well plates and cultured in serum-free medium containing chemokine for 24–48 hr. Absorbance at 450 nm was measured using a microtiter plate reader. Proliferation index was calculated as the ratio of absorbance values for the test sample incubated for 24 hr or 48 hr divided by those for 6 hr.

Generation of CXCR3- and CXCR4-knockdown transfectants

To knockdown the endogenous expressions of CXCR3 and CXCR4 in SW620 cells, we used RNAi vectors (BLOCK-iT™ Pol II miR RNAi Expression vector; Invitrogen, Carlsbad, CA). The microRNA (miRNA)-expressing plasmids were constructed according to the manufacturer's protocol. Target sequences were as followed: miR-CXCR3-1, 5′-GAGAACTTC AGCTCTTCCTAT-3′; miR-CXCR3-2, 5′-CCCTCTTCAACA TCAACTTCT-3′; miR-CXCR4-1, 5′-ATGGATTGGTCATCC TGGTCA-3′ and miR-CXCR4-2, 5′-TGTTGGCTGCCTTACT ACATT-3′. We used the pcDNA6.2-GW/EmGFPmiR −negative control vector as a negative control. To establish clones stably expressing the miR-CXCR3, miR-CXCR4, miR-CXCR3/4 and control plasmid, the transfected cells were cultured with blasticidin (20 μg/ml) for 3 weeks. Several blasticidin-resistant clones were selected [green fluorescent protein (GFP)-expressing cells were selected by fluorescence-activated cell sorting (FACS) sorting when necessary] and maintained with the same concentration of blasticidin.

In vivo metastasis studies

For the rectal xenograft model, colon cancer cells (1 × 106 cells in 50 μl of PBS) were injected submucosally into the rectum of nude mice at day 0. At day 42, para-aortic LNs, liver and lungs were examined for metastatic foci using GFP fluorescence. At day 14, para-aortic LNs, liver and lungs were dissected and pooled for each mouse to quantify the metastasized cells by quantitative PCR, as previously described.6, 17 All animal experiments were approved by the Animal Care and Use Committee of Kyoto University.

Statistical analysis

All values are expressed as means ± standard deviation (SD). The statistical significance of differences was determined by Mann-Whitney U test, Fisher's exact test or Student's t-test. Differences with p < 0.05 were considered significant. Statistical analyses were performed using The SPSS software, version 11.50 (SPSS, Chicago, IL).

Results

Expression of CXCR3 and CXCR4 in CRC clinical samples

To study the expression of CXCR3 and CXCR4 in CRC, we first analyzed their mRNA levels with frozen samples of 25 primary tumors (Stage I/II, n = 8; Stage III/IV, n = 17) and 12 metastatic LNs. Quantitative RT-PCR demonstrated that CXCR3 expression was significantly higher in Stage III/IV primary tumors and metastatic LNs than Stage I/II tumors (Fig. 1a, left; p < 0.05). On the other hand, as for CXCR4 expression, there was no significant difference between the three groups (Fig. 1a, right). Regarding their ligands, CXCR3 ligands (CXCL9, CXCL10 and CXCL11) and CXCR4 ligand (CXCL12) were not associated with the stage and LN metastasis (Supporting Information Fig. 1). We then investigated protein expression by immunohistochemical analysis with the paraffin sections obtained from 49 primary tumors (Stage I/II, n = 23; Stage III/IV, n = 26), 23 LN metastases and 20 liver metastases. We detected CXCR3 and CXCR4 proteins in the plasma membrane and cytoplasm of cancer cells. CXCR3 and CXCR4 were heterogeneously expressed in the primary tumors, whereas their expressions were almost homogeneous in metastatic tumors of LNs and liver (Fig. 1b). When we quantified the proportion of CXCR3- and CXCR4-positive CRC cells, the expression of CXCR3 in the Stage III/IV primary tumors was significantly higher than that in Stage I/II primary tumors (mean, 50.4% vs. 31.3%; p < 0.05). Moreover, CXCR3 was expressed at a much higher proportion in LN and liver metastases than Stage III/IV primary tumors (mean, 80.9% vs. 50.4%; p < 0.05, and 72.8% vs. 50.4%; p < 0.05, respectively; Fig. 1c, left). Similarly, the expression of CXCR4 in the LN and liver metastases were significantly higher than that in Stage III/IV primary tumors (mean, 68% vs. 36.6%; p < 0.05, and 66.8% vs. 36.6%; p < 0.05, respectively), although there was no significant difference between Stage I/II and Stage III/IV primary tumors (mean, 26.5% vs. 36.3%; p = 0.271; Fig. 1c, right). These findings suggest that the CXCR3- and CXCR4-positive CRC cells selectively proliferate to form metastatic foci within LNs and liver, and that CXCR3 and CXCR4 may play key roles in CRC metastasis.

Figure 1.

Expression of CXCR3 and CXCR4 in human CRC samples. (a) Quantitative RT-PCR analysis of CXCR3 and CXCR4 in frozen samples of 25 primary tumors and 12 metastatic LNs. Mean; bars (Student's t-test; *p < 0.05 compared to Stage I/II primary tumor). (b) Immunohistochemical evaluation of CXCR3 and CXCR4 in primary tumor, LN metastasis and liver metastasis. Scale bars, 100 μm. (c) Comparison of the proportion of CXCR3- and CXCR4-positive cells in primary tumors, LN metastasis and liver metastasis. Mean; bars (Student's t-test; *p < 0.05 compared to Stage I/II primary tumor. **p < 0.05 compared to Stage I/II and Stage III/VI primary tumors).

Expression of CXCR3 and CXCR4 in CRC cell lines

Human CRC cell line SW480 was derived from a primary Dukes B CRC, whereas SW620 was from a metastatic LN of the same patient.18 We mainly used this pair of cell lines in this study, because they had the same genetic background with different metastatic potential.19 To evaluate their metastatic activities in vivo, we performed the orthotopic transplantation model of CRC metastasis in mice6, 17, 20 with the use of GFP marker. After injecting the cells into the rectal smooth muscle layer of nude mice, we macroscopically examined metastatic foci using GFP fluorescence. The SW620 cells produced metastatic foci within para-aortic LNs, liver and lungs at high frequencies (11/15, 7/15 and 10/15, respectively), whereas SW480 metastasized only at low frequencies (4/20, 2/20 and 2/20, respectively; Table 1).

Table 1. Metastasis frequencies of SW620 clones to LNs, lungs and liver
inline image

We evaluated the protein expression levels of CXCR3 and CXCR4 by Western blot analysis, and found that both CXCR3 and CXCR4 were expressed at more than 4.0-fold higher levels in SW620 cells compared to SW480 (Fig. 2a; p < 0.05). In addition, we examined three other CRC cell lines (HT29, HCT116 and DLD-1). The HT29 and HCT116 cells are highly metastatic, whereas DLD-1 are not in mouse xenograft models.6, 17, 21 We found that both CXCR3 and CXCR4 were highly expressed in HT29 and HCT116 cells, whereas both receptors were hardly expressed in DLD-1 (Fig. 2a). These results suggest that upregulation of both CXCR3 and CXCR4 may be associated with CRC metastasis.

Figure 2.

Expression and function of CXCR3 and CXCR4 in human CRC cell lines. (a) Western-blot analysis of CXCR3 and CXCR4 (top), and quantified data for CXCR3 and CXCR4 (bottom). (Normalized with β-actin.) Mean; bars, ± SD, n = 3 (Student's t-test; *p < 0.05 against SW620 cells). (b) Chemotactic responses to CXCL10 (100 ng/ml) and CXCL12 (100 ng/ml) are shown with or without neutralizing CXCR3 antibody (20 μg/ml). Experiments were performed as a typical migration assay except that the upper well can also contain a chemokine. Mean; bars, ± SD. (*p < 0.05, **p < 0.05 by Student's t-test). (c) Cell proliferation assay on SW620 cells cultured in serum-free medium, with or without CXCL10 (100 ng/ml) and CXCL12 (100 ng/ml). Mean; bars, ± SD (Student's t-test; *p < 0.05). (d) MMP9 expression was determined by Western blotting after incubation with CXCL10 or CXCL12 for 48 hr. Representatives of three independent experiments done. Mean; bars, ± SD (Student's t-test; *p < 0.05 against control).

The two splice isoforms of CXCR3, CXCR3-A and CXCR3-B have been reported to play different roles in cancer progression.7, 15 Therefore, we examined the expression of the different CXCR3 variants by RT-PCR. None of the CXCR3 antibodies distinguishes CXCR3-A from CXCR3-B due to their near complete overlap in sequence, and so we resorted to RT-PCR analysis (Supporting Information Fig. 2a). We found that CXCR3-A expression in SW620 cells was much higher than that in SW480, and that CXCR3-A was expressed in four (SW620, SW480, HT29 and HCT116) of five CRC cell lines, which was consistent with our Western blot data (Fig. 2a). In contrast, CXCR3-B was not expressed in SW480 and SW620 cells, although it could be detected at low levels in HT29, HCT116 and DLD-1 cells.

It was reported that CCR6 and CCR7 were expressed in CRC.22–24 Therefore, we then evaluated the expression of these two receptors by RT-PCR, and found that CCR6 was expressed at high levels in HCT116 and DLD-1 cells, and at moderate levels in SW620 and HT29 cells, whereas CCR7 was detected only in SW480 cells (Supporting Information Fig. 2a). Taken together, these results might suggest that CRC metastasis is associated with CXCR3-A and CXCR4 rather than CXCR3-B, CCR6 or CCR7.

CXCR3- and CXCR4-mediated cellular responses of CRC cell lines

To determine whether CXCR3 and CXCR4 could induce cell migration, we performed an in vitro chemotaxis assay using SW620 cells that express both CXCR3 and CXCR4 (Fig. 2b). As anticipated, CXCL10 (ligand for CXCR3) and CXCL12 (ligand for CXCR4) caused directional migration. In addition, we found that CXCL12-induced cell migration was significantly enhanced by adding CXCL10 to cancer cells in the upper well (Fig. 2b, sample 5 vs. 6), and that neutralizing anti-CXCR3 antibody significantly suppressed the migratory response (Fig. 2b, sample 8). In contrast, CXCL10-induced cell migration was not affected by adding CXCL12 to cancer cells in the upper well (Fig. 2b, sample 3 vs. 4). Similar data were obtained with SW480 cells (data not shown). These results suggest that activation of CXCR3 can strengthen the CXCR4-mediated cell migration in a synergistic manner, which is consistent with published data of other cell types.25, 26

In addition to induction of cell migration, activation of a chemokine receptor by its ligand stimulates cell proliferation in some cell types.27 Therefore, we analyzed the effects of CXCL10 and CXCL12 on cell proliferation in SW620 cells (Fig. 2c). Interestingly, CXCL10 significantly enhanced cell proliferation compared to the untreated control, whereas CXCL12 did not. When CXCL10 and CXCL12 were treated together, no additional effects were observed when compared with cells treated only by CXCL10.

Matrix metalloproteinases (MMPs), particularly MMP-9, are implicated in CRC progression and metastasis.7 We next analyzed the effects of CXCL10 and CXCL12 on MMP9 secretion (Fig. 2d). Although CXCL10 induced upregulation of MMP-9 secretion in a dose-dependent manner, CXCL12 did not. Similar results were obtained with SW480 cells (data not shown).

Activation of intracellular signaling in CRC cell lines by CXCR3 and CXCR4

As CXCL10 and CXCL12 promoted malignancy-related properties in vitro, we examined the activation of intracellular signaling pathways. Extracellular signal-related kinase (ERK) MAP kinase and phosphatidylinositol 3-kinase are regulated by G protein coupled receptors, including chemokine receptors.27 Therefore, we investigated the phosphorylation of ERK1/2 and Akt/PKB after treatment with CXCL10 and CXCL12. Upon exposure of SW620 cells to CXCL10 and CXCL12, phosphorylation of both ERK1/2 and Akt/PKB took place 10 min after stimulation (Fig. 3a). In addition, pretreatment with CXCL10 further increased the CXCL12-induced phosphorylation of ERK1/2 and Akt/PKB (Fig. 3b), which was consistent with the synergistic effect between CXCR3 and CXCR4 in our cell migration data (Fig. 2b). However, such an increase in phosphorylation was not observed when pretreated, in the reverse order, with CXCL12 followed by CXCL10 stimulation (Supporting Information Fig. 2c). We confirmed that CXCR4 expression was not affected by pretreatment with CXCL10 (Supporting Information Fig. 2b), which suggests that the synergistic effect between CXCR3 and CXCR4 is due to an increase in CXCR4 responsiveness to CXCL12, but not due to upregulation of CXCR4 expression.

Figure 3.

Activation of intracellular signaling by stimulation of CXCR3 and CXCR4. (a) Western-blot analysis of SW620 cells treated by CXCL10 (100 ng/ml) or CXCL12 (100 ng/ml) for the levels of phospho-ERK1/2, ERK1/2, phospho-Akt/PKB (Ser473) and Akt/PKB. (b) Western-blot analysis of SW620 cells treated by CXCL12 (100 ng/ml) with or without 10 min pretreatment of CXCL10 (100 ng/ml) for the levels of phospho-ERK1/2, ERK1/2, phospho-Akt/PKB (Ser473) and Akt/PKB. Mean; bars, ± SD. (*p < 0.05, **p < 0.05 by Student's t-test).

Construction of SW620 transfectants with reduced CXCR3 and/or CXCR4 expression

To investigate the role of CXCR3 and CXCR4 in CRC metastasis in vivo, we established stable SW620 transfectant lines in which CXCR3 and/or CXCR4 were reduced by miRNA expression targeting CXCR3 and/or CXCR4. Two independent transfectant lines for each CXCR3-, CXCR4- and CXCR3/CXCR4 double-knockdowns were established (referred as miCXCR3, miCXCR4 and miCXCR3/CXCR4, respectively). The presence of miRNA vectors in these transfectant clones were verified by GFP fluorescence, because EGFP gene sequence was encoded into the same miRNA vectors. A Western blot analysis showed that the protein levels of CXCR3 and/or CXCR4 in each clone were decreased to 15–25% of those in the control clones (Fig. 4a). To examine the function of CXCR3 and CXCR4, we performed chemotaxis assays. Consistent with the protein levels, ligand-induced migratory responses were virtually eliminated in the targeting miRNA-transfectant clones (Supporting Information Fig. 3a). To exclude possible effects of CXCR3 and CXCR4 knockdown on cell proliferation, we determined the growth rates of the parental and transfectant clones without significant difference among them (Supporting Information Fig. 3b).

Figure 4.

SW620 stable transfectants for each CXCR3-, CXCR4- and CXCR3/CXCR4 double-knockdowns. (a) Western-blot photographs of CXCR3, CXCR4 and β-actin (top), and quantified data for CXCR3 and CXCR4 (bottom). (Normalized with β-actin.) Mean; bars, ± SD, n = 3 (Student's t-test; *p < 0.05 against control clones). (b) Quantification of the metastasized cancer cells to the para-aortic LNs, liver and lungs 2 weeks after inoculation. Eight mice from each group were analyzed. Para-aortic LNs, liver and lungs were pooled for each mouse, and extracted DNA was analyzed for the numbers of metastasized tumor cells using the human β-globin gene (HBB). The estimation was based on a calibration curve of given numbers of cultured SW620 cells. Mean; bars (exact Mann-Whitney U test; *p < 0.05 compared to parental SW620 and control clone).

Next, we assessed the metastatic activities of these transfectant clones using the orthotopic transplantation mouse model. Six weeks after rectal injection, we macroscopically examined metastatic foci within para-aortic LNs, liver and lungs using GFP fluorescence (Figs. 5a and 5b, Table 1). Although the control clones formed metastatic foci in 89% (26 of 29), 34% (10 of 29) and 65% (19 of 29) of LNs, liver and lungs, respectively, the CXCR3-knockdown clones (miCXCR3-1 and −2) metastasized to these organs at only 21% (6 of 28), 4% (1 of 28) and 25% (7 of 28) in total, respectively (p < 0.05). The CXCR4-knockdown clones (miCXCR4-1 and −2) also decreased the metastatic frequencies to 52% (17 of 33), 0% (0 of 33) and 33% (11 of 33) to the three organs in total, respectively (p < 0.05), although its effect on LN metastasis was less than that of the CXCR3-knockdowns (21% vs. 52%; p < 0.05). In addition, CXCR3/CXCR4 double-knockdown clones (miCXCR3/CXCR4-1 and −2) metastasized at only 11% (4 of 35), 3% (1 of 35) and 20% (7 of 35) to these organs in total, respectively (p < 0.05), and the effect was not significantly different from that observed in the CXCR3-knockdowns. In addition, the numbers of metastatic foci within para-aortic LNs, liver and lungs were significantly decreased in the miCXCR3, miCXCR4 and miCXCR3/CXCR4 clones compared with those in the control (Fig. 5c; p < 0.05). Regarding the transplanted primary tumors, there was no significant difference in size among the mice inoculated with their respective clones.

Figure 5.

Inhibition of metastasis to LNs, lungs and liver by CXCR3-, CXCR4- and CXCR3/CXCR4 double-knockdowns. (a) Representative photographs of the para-aortic LNs from mice 6 weeks after inoculations with control, miCXCR3, miCXCR4 and miCXCR3/CXCR4 clones, respectively. Metastatic foci are visualized by GFP fluorescence. Scale in millimeters. (b) Representative photographs of the lung from mice 6 weeks after inoculations with control, miCXCR3, miCXCR4 and miCXCR3/CXCR4 clones, respectively. Metastatic foci are visualized by GFP fluorescence. Insets show higher magnifications of the boxed areas. Scale in millimeters. (c) The number of metastatic foci per mouse within para-aortic LNs, liver and lungs in the control (n = 29), miCXCR3 (n = 28), miCXCR4 (n = 33) and miCXCR3/CXCR4 (n = 35). Each circle represents one mouse sample. Mean; bars (Student's t-test; *p < 0.05 compared to control clone).

To elucidate the formation of micrometastasis, we quantified the metastasized cancer cells within the three organs 2 weeks after inoculations, by quantitative PCR of human β-globin-related sequence (HBB; Fig. 4b). The results indicated that there was no apparent difference in the number of cancer cells disseminated to LNs. For lung metastasis, the numbers of cancer cells of miCXCR3 and miCXCR3/CXCR4 clones were significantly decreased compared to that of the control (parental, 6,200 ± 5,500 cells; control, 4,000 ± 4,200 cells; miCXCR3-1, 180 ± 140 cells; miCXCR4-1, 1,400 ± 1,900 cells and miCXCR3/CXCR4-1, 230 ± 200 cells: mean ± SD of eight transplants/group) (p = 0.038 and 0.038, respectively), although that of miCXCR4 was not decreased. For liver metastasis, miCXCR3 and miCXCR3/CXCR4 clones significantly decreased the numbers of cancer cells compared with the control clones (parental, 2,400 ± 4,400 cells; control, 1,500 ± 2,900 cells; miCXCR3-1, 140 ± 260 cells; miCXCR4-1, 910 ± 940 cells; and miCXCR3/CXCR4-1, 120 ± 180 cells: mean ± SD of eight transplants/group) (p = 0.021 and 0.028, respectively). Taken together, these results indicate that targeting CXCR3 or CXCR4 can suppress CRC metastasis to LNs, liver and lungs, although targeting CXCR3 may have a stronger effect for the CRC patients with severe LN metastases.

Discussion

Our study revealed that both CXCR3 and CXCR4 play pivotal roles in CRC metastasis to LNs and distant organs (liver and lungs). We previously reported that CXCR3 promotes CRC metastasis preferentially to LNs with poorer prognosis and that the patients with CXCR3-positive CRC exhibited significantly shorter survival than those without them as well as CXCR4-positive CRC.6 Although CXCR4 has been reported to be associated with distant metastasis of CRC, the effect of CXCR3 on distant metastasis to liver and lungs has not been elucidated. Here, we have demonstrated that both CXCR3 and CXCR4 were expressed at higher levels in SW620 (high metastatic potential) compared to SW480 cells (low metastatic potential; Fig. 2a) and that both CXCR3 and CXCR4 were expressed at a higher proportion in metastases than in primary tumors using clinical samples (Fig. 1). We also found that both CXCR3 and CXCR4 were expressed at similar levels in both HCT116 cells with mutated KRAS and isogenic clones with wild-type KRAS alleles (data not shown), which may indicate that expressions of CXCR3 and CXCR4 are not affected by KRAS mutation. Although a positive correlation was reported for CXCR3 expression and poor prognosis in melanoma, CRC and breast cancer,6, 28, 29 the opposite relationship was observed in chronic B-cell lymphocytic leukemia and clear cell renal carcinoma.30, 31 In these two diseases, low CXCR3 expression, rather than abundant CXCR3, was associated with shorter survival. Thus, the contribution of CXCR3 to tumor behavior may not be generalized to all malignancies. CXCR3 has two splice variants with completely opposite functions in different cell types; CXCR3-A promotes cell proliferation and migration, whereas CXCR3-B inhibits these processes.15, 32 In our study, we found that the expression of CXCR3-A was markedly increased, whereas that of CXCR3-B was low in human CRC cell lines. In addition, we found that CXCL4, specific ligand for CXCR3-B, did not change the migratory and proliferative responses of HT29 cells (data not shown). Accordingly, it is likely that CXCR3-A, not CXCR3-B, plays an important role in CRC invasion and metastasis.

CXCR3 ligands (CXCL9, CXCL10 and CXCL11) have been demonstrated to exhibit both pro-tumor and anti-tumor functions in tumorigenesis.33, 34 For example, CXCL10 promotes tumor invasion and metastasis in CRC,7, 8, 35, 36 whereas downregulated CXCL10 expression correlates with poorer prognosis in Stages II and III CRC.37 Regarding the role of CXCR4 ligand (CXCL12) in CRC, its expression within colonic carcinoma cells is silenced by DNA hypermethylation to promote their metastatic potential,38 whereas high expression of CXCL12 at tumor budding significantly correlates with shorter survival.39 It was also reported CXCL12 G801A polymorphism predicted LN metastasis in CRC.40 Here, we have demonstrated that expressions of all three CXCR3 ligands (CXCL9, CXCL10 and CXCL11) and CXCR4 ligand (CXCL12) at primary tumors were not associated with the stage and metastasis (Supporting Information Fig. 1). Large-cohort studies in a multicenter setting will be necessary to validate these findings and examine potential mechanisms of ligands for CXCR3 and CXCR4 in CRC metastasis.

This is the first study to show a synergistic effect among chemokine receptors expressed on cancer cells. Here, we have found that activation of CXCR3 pathway strengthens the CXCR4-mediated cell migration of SW620 and SW480 cells in a synergistic manner (Fig. 2b). Interestingly, this synergistic effect between CXCR3 and CXCR4 was observed in other cell types, such as plasmacytoid dendritic cells and IFN-producing cells,25, 26 which suggest that this phenomenon may be generalized to various kinds of mammalian cells. Regarding the molecular mechanism of this synergistic effect, we showed that pretreatment with CXCL10 did not upregulate CXCR4 expression (Supporting Information Fig. 2b), but increased the CXCL12-induced phosphorylation of ERK1/2 and Akt/PKB (Fig. 3b). Collectively, it is conceivable that CXCR3 responds to CXCR3 ligands (CXCL9, CXCL10 and CXCL11), and increases the sensitivity of CXCR4 to CXCL12. Regarding the metastatic activities in vivo, we showed that CXCR3-knockdown clones significantly decreased metastasis to distant organs (liver and lungs) as well as LNs. On the other hand, we previously reported that forced expression of CXCR3 promoted LN metastasis of DLD-1 cells in vivo without affecting distant metastasis to liver and lungs.6 We speculate that one reason for this discrepancy among these results is the absence of the synergistic effect between CXCR3 and CXCR4, because DLD-1 cells do not endogenously express CXCR4. Several lines of clinical evidence are beginning to show that LN metastasis not only provides the prognostic information but also serves as a launch pad for further distant metastasis to liver and lungs, causing eventual lethality.9 Although little is known about the molecular mechanism how tumor cells within LNs promote further distant metastasis, the synergistic effect between CXCR3 and CXCR4 may be involved in this mechanism. We have also demonstrated that CXCR3-knockdown clones disseminated to LNs at a similar frequency to that of control clones 2 weeks after inoculation. However, by 6 weeks, CXCR3-knockdown clones did not colonize metastatic foci in LNs to the same extent as the control clones. Therefore, it is conceivable that CXCR3 stimulates the expansion of the metastatic foci in LNs rather than helping the initial dissemination of CRC cells. On the other hand, CXCR3-knockdown clones significantly decreased the disseminated cancer cells within liver and lungs even 2 weeks after inoculation, suggesting that CXCR3 may be involved in both early and later stages of distant metastasis. It is worth noting that systemic administration of a small molecular inhibitor of CXCR3, AMG487, inhibits lung metastasis of HT29 CRC cells and breast cancer cells in a mouse model.8, 14 In addition, CXCR4-knockdown clones also significantly decreased metastasis to LNs, liver and lungs 6 weeks after inoculation, although its effect on LN metastasis was less. However, 2 weeks after inoculation, CXCR4-knockdown clones disseminated to LNs, liver and lungs at a similar frequency to that of the control clones, which was consistent with a previous report that CXCR4 is required for outgrowth of CRC micrometastases.4 The CCL20/CCR6 axis has been shown to be associated with liver metastasis of CRC.23, 24 We have also found that pretreatment with CXCL10 increased the CCL20-induced phosphorylation of ERK1/2 and Akt/PKB in SW620 cells (Supporting Information Fig. 2d), whereas such an increase in phosphorylation was not observed when pretreated with CXCL12 (Supporting Information Fig. 2e). These results may indicate that activation of CXCR3 pathway strengthens the function of CCR6 as well as CXCR4 in a synergistic manner.

Chemokines are released locally and their effects are usually confined to local tissues, which differs from cytokines that often cause systemic effects. Because of these characteristics, they can be better targets for drug therapy with less off-target side effects. Because most of the drugs targeting chemokine signals were originally developed for autoimmune and inflammatory diseases such as rheumatoid arthritis, psoriasis, multiple sclerosis and asthma,41 the clinical trials targeting chemokine signals in cancer remain limited. Our study provides preclinical evidence that both CXCR3 and CXCR4 are potential therapeutic targets for suppressing CRC metastasis. Therefore, it would be worthwhile to make use of the inhibitors for CXCR3 and CXCR4 to prevent CRC metastasis, because they have been already developed for other disease and tested for safety.

Acknowledgements

This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan, from Sagawa Foundation for promotion of Cancer Research, and from Takeda Science Foundation (to K.Kawada)

Ancillary