Cancer and dendritic cells recognize and migrate toward chemokines secreted from lymphatics and use this mechanism to invade the lymphatic system, and cancer cells metastasize through it. The lymphatic-secreted chemokine ligand CCL21 has been identified as a key regulatory molecule in the switch to a metastatic phenotype in melanoma and breast cancer cells. However, it is not known whether CCL21 inhibition is a potential therapeutic strategy for inhibition of metastasis. Here, we describe an engineered CCL21-soluble inhibitor, Chemotrap-1, which inhibits migration of metastatic melanoma cells in vivo. Two-hybrid, pull-down, and coimmunoprecipitation assays allowed us to identify a naturally occurring human zinc finger protein with CCL21 chemokine-binding properties. Further analyses revealed a short peptide (∼70 amino acids), with a predicted coiled-coil structure, which is sufficient for association with CCL21. This CCL21 chemokine-binding peptide was then fused to the Fc region of human IgG1 to generate Chemotrap-1, a human chemokine-binding Fc fusion protein. Surface plasmon resonance and chemotaxis assays showed that Chemotrap-1 binds CCL21 and inhibits CCL21-induced migration of melanoma cells in vitro with subnanomolar affinity. In addition, Chemotrap-1 blocked migration of melanoma cells toward lymphatic endothelial cells in vitro and in vivo. Finally, Chemotrap-1 strongly reduced lymphatic invasion, tracking, and metastasis of CCR7-expressing melanoma cells in vivo. Together, these results show that CCL21 chemokine inhibition by Chemotrap-1 is a potential therapeutic strategy for metastasis and provide further support for the hypothesis that lymphatic-mediated metastasis is a chemokine-dependent process. Cancer Res; 70(20); 8138–48. ©2010 AACR.

Tumor metastasis occurs through either the lymphatic or the vascular systems. Lymphatic metastasis is the most common route for melanoma, breast, prostate, and other cancers. When they metastasize specifically through the lymphatic system, tumor cells locate and recognize lymphatic endothelial cells (LEC). The secreted lymphatic chemokine CCL21 is released from LEC (1) and used by dendritic cells (DC) to guide their invasion into lymphatics by acting on its receptor CCR7, expressed on DCs (2). CCR7 expression is increased in patients with metastatic melanoma (3); prostate (4), head and neck (5), and squamous cell carcinoma (6); and breast cancer (7) compared with nonmetastatic cancers. Thyroid, lung, and oral squamous cell carcinoma; melanoma; and breast cancer cells that upregulate receptors for lymphatic-secreted chemokines such as CCL21 are more likely to metastasize (3, 810). It has therefore been hypothesized that CCR7 is used by metastatic melanoma cells to detect locally secreted lymphatic-specific chemokines, such as CCL21, and migrate toward it to find areas of increased lymphatic density (11). One potential therapeutic strategy to prevent metastasis would be to inhibit this chemokine-mediated entry of cancer cells into lymphatics. Therapeutic strategies based on soluble human receptors prepared as Fc fusion proteins, termed Traps, have been successfully developed for several cytokines, including tumor necrosis factor α (etanercept; refs. 12, 13), interleukin-1 (IL-1-Trap; ref. 14), and vascular endothelial growth factor (Aflibercept; ref. 15). These are generated using the fixed chain (Fc) domain of the human IgG fused to a protein-binding domain to provide a stable, dimeric protein with long plasma half-life. The protein-binding domain has been either the ligand-binding domain of the receptor (e.g., etanercept; ref. 16) or a fusion protein of different ligand-binding domains of multiple receptors (e.g., aflibercept; ref. 15) to improve the affinity of the Trap. To determine whether CCL21 could be a potentially useful antimetastatic target, we identified a naturally occurring human protein with CCL21-binding properties and used its chemokine-binding domain to generate a human IgG1-Fc fusion protein that binds CCL21 with high affinity and inhibits its activity. We then determined whether this protein, Chemotrap-1, could inhibit melanoma growth toward areas of high lymphatic density.

Cell culture and transfection

Melanoma cells, purchased from the American Type Culture Collection, from Bioware, or supplied by I.J. Fidler (University of Texas M.D. Anderson Cancer Center, Houston, TX), were maintained in 10% fetal calf serum (FCS)–DMEM (A375) or 10% FCS–Eagle's MEM (mouse B16-Luc). Human dermal LECs (CC-2814, Cambrex) were maintained in EGM-2 MV medium (CC-3202, Cambrex). For transfection, cells at 80% to 90% confluency were seeded in six-well plates and transfected with 1 or 1.5 µg of DNA and 4 μL of Lipofectamine 2000 (Invitrogen) in Opti-MEM according to the manufacturer's instructions. Antibiotic selection was started on day 2 following transfection.

Plasmid constructions

pGBT9-hCCL21 vector was generated by inserting a cDNA encoding the mature form of human CCL21 (Genbank accession no. NP_002989; amino acids 24–134) into the BamHI site of Matchmaker Yeast Two-Hybrid (Y2H) System 2 vector pGBT9 (Clontech). pGBKT7-hCCL21 vector was generated by subcloning the BamHI CCL21 fragment from pGBT9-hCCL21 into the BamHI site of pGBKT7 (Clontech). pGADT7-hCCL21 and pGADT7-hCCL21ΔCOOH expression vectors were generated by subcloning the BamHI fragment (encoding CCL21 amino acids 24–134) or the BamHI-PstI fragment (encoding CCL21, amino acids 24–102) from pGBKT7-CCL21 into pGADT7 expression vector (Clontech). pGADT7-mCCL21 and pGADT7-CCL27 expression vectors were generated by cloning the cDNAs encoding mature forms of mCCL21 (Genbank accession no. NP_035254; amino acids 24–133) or hCCL27 (Genbank accession no. NP_006655; amino acids 25–112) into pGADT7. Construction of pGBKT7-THAP1 and pEGFP.C2-THAP1 expression vectors has been previously described (17). The glutathione S-transferase (GST)–THAP1 expression vector was generated by cloning the full-length coding region of human THAP1 (amino acids 1–213) in the pGEX-2T prokaryotic expression vector (Amersham Pharmacia Biotech). THAP1 deletion mutants THAP1-C1 (amino acids 90–213), THAP1-C2 (amino acids 120–213), THAP1-C3 (amino acids 143–213), THAP1-N1 (amino acids 1–90), THAP1-N2 (amino acids 1–166), and THAP1-N3 (amino acids 1–192) were amplified by PCR using pEGFP.C2-THAP1 as template. The PCR fragments were digested with EcoRI and BamHI and cloned in-frame downstream of the Gal4 activation domain in the pGADT7 two-hybrid vector (Clontech).

Y2H assays

Two-hybrid screening of a high endothelial venule (HEV) cDNA library (17) was performed using human CCL21 as bait. pGBT9-hCCL21 was cotransformed with the pGAD424-HEV cDNA library in yeast strain Y190 (Clontech). Transformants (1.5 × 107) were screened, and positive protein interactions were selected by His auxotrophy. Two-hybrid interactions were confirmed in strain AH109 using Matchmaker Two-Hybrid System 3 (Clontech). AH109 cells were cotransformed with pGBKT7-THAP1, pGADT7-hCCL21, pGADT7-mCCL21, pGADT7-hCCL27, or pGADT7-hCCL21ΔCOOH expression vectors. Transformants were selected on medium lacking histidine and adenine. Two-hybrid interaction between THAP1 mutants and chemokine CCL21 or wild-type THAP1 was tested by cotransformation of AH109 with pGADT7-THAP1-C1, pGADT7-THAP1-C2, pGADT7-THAP1-C3, pGADT7-THAP1-N1, pGADT7-THAP1-N2, pGADT7-THAP1-N3, pGBKT7-CCL21, or pGBKT7-THAP1.

GST pull-down and coimmunoprecipitation assays

GST and GST-THAP1 fusion proteins were produced and purified as previously described (17). In vitro–translated hCCL21, mCCL21, hCCL27, and hCCL21ΔCOOH were generated with the TNT Coupled Reticulocyte Lysate System (Promega) using pGADT7-hCCL21, pGADT7-mCCL21, pGADT7-hCCL27, or pGADT7-hCCL21ΔCOOH as template. 35S-labeled chemokines (25 μL) were incubated with immobilized GST-THAP1 or GST proteins overnight at 4°C in 10 mmol/L NaPO4 (pH 8.0), 140 mmol/L NaCl, 3 mmol/L MgCl2, 1 mmol/L DTT, 0.05% NP40, 0.2 mmol/L phenylmethylsulfonyl fluoride, 1 mmol/L sodium vanadate, 50 mmol/L β-glycerophosphate, 25 μg/mL chymotrypsin, 5 μg/mL aprotinin, and 10 μg/mL leupeptin. Beads were washed five times in 1-mL binding buffer. Bound proteins were eluted with 2× Laemmli SDS-PAGE sample buffer, fractionated by 10% SDS-PAGE, and visualized by fluorography using Amplify (Amersham Pharmacia Biotech). For coimmunoprecipitation experiments, U2OS cells were transfected with plasmids encoding CCL21-Flag and/or HA-THAP1 using a calcium phosphate precipitation procedure. Two days after transfection, cellular extracts were prepared [15 mmol/L Tris-HCl (pH 7.5), 0.4 mol/L NaCl, 5 mmol/L MgCl2, and 0.1% Tween 20 containing protease and phosphatase inhibitors (Roche)]. After three freeze/thaw cycles, cellular extracts were centrifuged at 12,000 × g for 10 minutes and then precleared at 4°C for 1 hour in a rotating wheel with 20 μL protein G–Sepharose beads (Amersham). Beads were precipitated by centrifugation, and supernatants were incubated overnight at 4°C with 5 μg of Flag M2 antibody (Sigma). Immune complexes were captured with 20 μL of protein G–Sepharose beads at 4°C for 1 hour and washed four times with 0.4 mol/L NaCl. Bound proteins were eluted and subjected to immunoblotting with an anti-HA antibody.

In silico sequence analysis and molecular modeling

Coiled-coil predictions were performed at PAIRCOIL (http://paircoil.lcs.mit.edu/cgi-bin/paircoil) and MULTICOIL (http://multicoil.lcs.mit.edu/cgi-bin/multicoil) Web sites. The structure of the THAP1 dimeric coiled-coil was built by homology with the X-ray structure of the GCN4 parallel coiled-coil (PDB code: 1KD8). The energy of the interaction interface was minimized by using the Affinity module within Insight II (Accelrys).

Generation of Chemotrap-1 expression vector

The sequence encoding residues 140 to 213 of human THAP1 was amplified by PCR with oligonucleotides THAP1-XhoI-5′(5′-ccgctcgaggatacaatgcacc-3′) and THAP1-BamH1-3′(5′-gcgggatccgctggtacttcaactatttcaaag-3′). The resulting XhoI-BamHI fragment was used to replace the XhoI-BamHI fragment encoding L-selectin in the pCDM8–L-selectin–IgG1 plasmid (18, 19). A linker encoding the Igκ chain signal peptide (SP; 21 amino acids) from plasmid pSecTag2 (Invitrogen) was ligated into the XhoI site of pCDM8-THAP1140–213-IgG1 to obtain the expression vector pCDM8-SP-THAP1140–213-IgG1. The whole construct was inserted into the pcDNA3.1 vector (Invitrogen) to generate pcDNA3.1–Chemotrap-1. pcDNA3.1–Chemotrap-189 expression vector was obtained by replacing the XhoI-THAP1140–213-BamHI fragment of pcDNA3.1–Chemotrap-1 by an XhoI-THAP1189–213-BamHI fragment generated by PCR.

Chemotrap-1 production and purification

pcDNA3.1Zeo-Chemotrap or pcDNA3.1–Chemotrap-189 expression vectors were transfected into Chinese hamster ovary (CHO) cells using JetPEI (Ozyme). Cells were grown with zeocin (100 mg/mL; Invivogen) to establish clonal lines. Supernatants were diluted in ImmunoPure-Gentle Ag/Ab Binding Buffer (Pierce) and mixed with protein A–Sepharose CL-4B (Amersham Biosciences) for 1 hour at room temperature under gentle agitation. The packed resin was rinsed using 15 volumes of binding buffer before elution with ImmunoPure-Gentle Elution Buffer (Pierce). The fractions containing Chemotrap were concentrated on an Amicon Ultra-4 (molecular weight cutoff, 10,000) and analyzed by immunoblotting with goat α-human IgG horseradish peroxidase (1:5,000) and an enhanced chemiluminescence kit (Amersham Bioscience).

Surface plasmon resonance experiments

Purified recombinant THAP11–213-Fc and Chemotrap-1 (THAP1140–213-Fc) were covalently bound from their amino groups to the gold sensor chip surface (CM5 sensor chip). Purified recombinant human CCL21 (Chemicon) was used as analyte protein at a fixed concentration in the fluid phase [10 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 3 mmol/L EDTA, 0.005% P20 surfactant, as a running buffer] and at a constant flow rate of 20 μL/min in a BIAcore 3000 system (BIAcore AB). Association and dissociation curves were established for 12.5, 25, 50, 100, and 200 nmol/L CCL21. Chemokine was injected during the association phase for 4 minutes (20 μL/min). The dissociation phase was carried out over 5 minutes, and flow cells were regenerated by injection of 0.05% SDS (30 seconds at a flow of 20 μL/min) between each phase of association-dissociation. BIAevaluation (BIAcore) program was used for calculations of association/dissociation kinetic constants (Kd = kd/ka).

ELISA

The CCL21 DuoSet ELISA was carried out as per the manufacturer's instructions. To determine the effect of Chemotrap-1 on CCL21 release from LEC, cells were cultured for 24 hours in 0.5% EGM-2, and 5-mL medium was transferred to a T25 flask of B16-Luc melanoma cells for a further 24 hours. Medium was then collected, spun to remove cells, and then concentrated through an Amicon 10000 spin column.

In vitro cell migration

Migration assays were performed in a modified Boyden chamber consisting of a cell culture insert with an 8-μm pore polycarbonate membrane (Millipore) seated in each well of a 24-well plate as previously described (11). For dose response, migration assays were performed as above in 150 ng/mL CCL21 and increasing concentrations of Chemotrap. The number of cells that migrated through were normalized to the lowest concentration of Chemotrap (10 pmol/L) and to 0.1% FCS alone (no CCL21), and EC50 was determined by nonlinear curve fitting with fixed top and bottom parameters (100% and 0%, respectively) using a sigmoidal dose-response curve.

In vivo metastatic chemotaxis model

Experiments were carried out as previously described (11). Briefly, nude mice were injected with 1 × 106 tumor cells s.c. with a Monastral blue–coated needle. The mice were then injected with 1 × 105 human LEC ∼10 mm caudal to the melanoma injection site. Mice were left until tumors reached 8 to 10 mm when measured through the skin. The animals were then killed, tumors were excised, and the skin was pronated and photographed. Directional growth of melanoma was measured as the plan area of the tumor below the perpendicular axis (i.e., closer to the endothelial cell injection site) and expressed as a percentage of the total area of the tumor.

In vivo bioluminescent imaging of tumors

CD1 mice were injected s.c. with 1 × 106 B16-Luc cells. To image bioluminescence, mice were injected with 0.15 mg/g luciferin i.p. at 20 mg/mL. Five minutes later, they were anesthetized by isofluorane (5% in 95% O2) and imaged using an IVIS Lumina (Caliper Life Sciences). Lesion maximum width and length were measured, and the ratio was used to calculate directional growth. Once tumors reached 16 mm in size, the mice were imaged as above and then killed, and the flank was dissected to investigate in-transit metastases. Areas of black tracking along lymphatics were photographed using a Nikon Coolpix digital camera.

Identification of a CCL21 chemokine-binding protein

While searching for CCL21-interacting proteins in a two-hybrid screen of a HEV cDNA library, we identified the cDNA encoding a novel nuclear factor, which we designated THAP1 (17). Eight positive clones (of 5 million transformants) were identified with the CCL21 bait, and all corresponded to the COOH-terminal part of THAP1 (amino acids 90–213; Fig. 1A). To confirm the interaction in yeast, we performed in vitro GST pull-down assays. Full-length THAP1-GST protein was incubated with radiolabeled in vitro–translated CCL21. Both human and mouse CCL21 proteins bind to GST-THAP1 but not to GST (Fig. 1A). The interaction was also observed when the basic COOH-terminal extension of CCL21 was deleted, indicating that THAP1 interacts with the core chemokine domain of CCL21 (Fig. 1A). No binding of THAP1 was observed to CCL27, another CC chemokine. We next addressed whether THAP1 is able to interact with CCL21 in cells. We performed immunoprecipitation experiments in cells coexpressing epitope-tagged CCL21 (Flag-CCL21) and THAP1 (THAP1-HA). We observed specific immunoprecipitation of THAP1 with anti-Flag antibodies in cells coexpressing Flag-CCL21, whereas no precipitation of THAP1 with anti-Flag antibodies was observed in control cells (Fig. 1B). These findings showed that THAP1 interacts with CCL21 both in vitro and in cells. To determine the kinetic parameters of the THAP1/CCL21 complex, we performed surface plasmon resonance (SPR). THAP11-213-Fc fusion protein was purified from cell supernatants and immobilized on a sensor chip. Addition of purified recombinant human CCL21 revealed association with immobilized THAP11-213-Fc with a Kd of 87 nmol/L (Fig. 1C). Together, our observations indicated that the human THAP–zinc finger protein THAP1 is a chemokine-binding protein that binds CCL21 with nanomolar affinity.

Figure 1.

Interaction between CCL21 and THAP1. A, two-hybrid and GST pull-down interaction assays show interaction between human CCL21 (hCCL21) and human THAP1. This interaction was also observed with mouse CCL21 (mCCL21), indicating evolutionary conservation. Schematic representation of results obtained in two-hybrid and GST pull-down assays using human THAP1 together with human CCL21, mouse CCL21, human CCL21 deleted from its COOH-terminal extension (hCCL21DCterm), and human CCL27, another CC-chemokine used as a control. B, coimmunoprecipitation of THAP1-HA with Flag-CCL21 in cellular extracts (0.4 mol/L NaCl) followed by immunoblot with anti-HA antibody. Stars represent Ig subunits, and arrows indicate the HA-tagged THAP1. C, SPR analysis of THAP1/CCL21 association. Full-length THAP1 protein (amino acids 1–213) was immobilized on the sensor chip as an IgG1-Fc fusion protein and incubated with increasing concentrations of soluble CCL21 protein. The Kd for the THAP1/CCL21 association had a calculated value of 87 nmol/L.

Figure 1.

Interaction between CCL21 and THAP1. A, two-hybrid and GST pull-down interaction assays show interaction between human CCL21 (hCCL21) and human THAP1. This interaction was also observed with mouse CCL21 (mCCL21), indicating evolutionary conservation. Schematic representation of results obtained in two-hybrid and GST pull-down assays using human THAP1 together with human CCL21, mouse CCL21, human CCL21 deleted from its COOH-terminal extension (hCCL21DCterm), and human CCL27, another CC-chemokine used as a control. B, coimmunoprecipitation of THAP1-HA with Flag-CCL21 in cellular extracts (0.4 mol/L NaCl) followed by immunoblot with anti-HA antibody. Stars represent Ig subunits, and arrows indicate the HA-tagged THAP1. C, SPR analysis of THAP1/CCL21 association. Full-length THAP1 protein (amino acids 1–213) was immobilized on the sensor chip as an IgG1-Fc fusion protein and incubated with increasing concentrations of soluble CCL21 protein. The Kd for the THAP1/CCL21 association had a calculated value of 87 nmol/L.

Close modal

Mapping the CCL21 chemokine-binding domain

To identify sequences mediating CCL21 binding, a series of THAP1 deletion mutants were generated. These truncated versions were used as Gal4 DNA-binding domain fusion proteins (baits) in the Y2H system, together with preys corresponding to mature CCL21 (amino acids 24–134) in fusion with the Gal4-activating domain. Schematic representation of all THAP1 deletion mutants and summary of two-hybrid results are presented in Fig. 2A. We found that neither the THAP1-N1 mutant, corresponding to the DNA-binding THAP–zinc finger (20), nor THAP1-N2 and THAP1-N3, which contain the THAP–zinc finger and the proline-rich linker region, interact with CCL21. In contrast, cells coexpressing CCL21 with NH2-terminal deletion constructs, THAP1-C1, THAP1-C2, or THAP1-C3, exhibited strong growth on selective medium, indicating that the COOH-terminal domain of THAP1 (amino acids 143–213) is necessary and sufficient for CCL21 binding. To confirm the Y2H results, we performed in vitro GST pull-down assays with GST-THAP1 deletion constructs that showed the COOH-terminal part of THAP1 (amino acids 142–213) sufficient for interaction with radiolabeled in vitro–translated CCL21 (Fig. 2B), but that it required amino acids 142 to 166, as deletion of amino acids 142 to 166 reduced binding and deletion abolished it completely. These results show that the CCL21 chemokine-binding domain of THAP1 corresponds to the last 70 amino acids (amino acids 143–213).

Figure 2.

The CCL21 chemokine-binding domain of THAP1 has characteristics of a coiled-coil domain. A, summary of two-hybrid results obtained with THAP1 deletion mutants and CCL21 or wild-type (wt) THAP1 (homodimerization). B, GST pull-down interaction assays show interaction between human CCL21 and human THAP1 residues 126 to 213 and 142 to 213, less so for 166 to 213, and not for 186 to 213. C, THAP1 contains a predicted coiled-coil domain between residues 142 and 190. The 213–amino acid sequence of THAP1 was analyzed using the PAIRCOIL program. The probability of each residue within the THAP1 sequence contributing to a coiled-coil structure is plotted (maximal probability score = 1). D, molecular modeling of the THAP1 coiled-coil domain based on the crystallographic structure of the GCN4 homodimeric parallel coiled-coil. Molecular modeling was performed as described in Materials and Methods.

Figure 2.

The CCL21 chemokine-binding domain of THAP1 has characteristics of a coiled-coil domain. A, summary of two-hybrid results obtained with THAP1 deletion mutants and CCL21 or wild-type (wt) THAP1 (homodimerization). B, GST pull-down interaction assays show interaction between human CCL21 and human THAP1 residues 126 to 213 and 142 to 213, less so for 166 to 213, and not for 186 to 213. C, THAP1 contains a predicted coiled-coil domain between residues 142 and 190. The 213–amino acid sequence of THAP1 was analyzed using the PAIRCOIL program. The probability of each residue within the THAP1 sequence contributing to a coiled-coil structure is plotted (maximal probability score = 1). D, molecular modeling of the THAP1 coiled-coil domain based on the crystallographic structure of the GCN4 homodimeric parallel coiled-coil. Molecular modeling was performed as described in Materials and Methods.

Close modal

The CCL21 chemokine-binding domain has characteristics of a coiled-coil

Two-hybrid assays with THAP1 bait (Fig. 2A) revealed that the CCL21 chemokine-binding domain (amino acids 143–213) also mediates interaction of THAP1 with itself (homodimerization or oligomerization). The PAIRCOIL (21) and MULTICOIL (22) algorithms revealed that the CCL21 chemokine-binding domain of THAP1 has characteristics of a coiled-coil, one of the principal dimerization/oligomerization motifs found in proteins (23). The coiled-coil structure consists of two α-helices wrapped around each other with a slight superhelical twist. Its most characteristic feature is a heptad repeat pattern of primarily apolar residues that constitute the dimer/oligomer interface. Both PAIRCOIL and MULTICOIL programs predicted a parallel left-handed coiled-coil domain with seven heptad repeats (49 residues), extending between amino acids 142 and 190 of human THAP1. The probability for formation of the coiled-coil structure of this region was 1 (maximum score) with both programs, indicating that this region of THAP1 has a very high probability of existing as a coiled-coil structure (Fig. 2C). Molecular modeling of the THAP1 coiled-coil domain was performed using the X-ray crystal structures of known coiled-coil domains as a template. The resulting three-dimensional model of the THAP1 coiled-coil is shown in Fig. 2D. The parallel homodimer is stabilized by a perfect zipper based on hydrophobic interactions between identical residue side chains from each monomer. The CCL21-binding domain also includes the last 23 COOH-terminal amino acids of THAP1 (amino acids 191–213), not required for homodimerization/oligomerization (Fig. 2A) but, with residues 142 to 166, essential for CCL21 binding.

Chemotrap-1—an engineered soluble inhibitor based on the CCL21 chemokine-binding domain

The CCL21 chemokine-binding domain of THAP1 could be useful to engineer a high-affinity soluble inhibitor capable of blocking CCL21 chemokine activity in vivo. We thus generated a fusion protein between the Fc region of human IgG1 and the CCL21 chemokine-binding domain of human THAP1 (Fig. 3A). This fusion protein—Chemotrap-1—was then expressed in CHO cells and purified. As a negative control, we generated a fusion protein of human IgG1 with a 26–amino acid region (amino acids 189–213) of THAP1 that does not bind CCL21(Fig. 2B). Chemotrap-1 bound to CCL21 with the same affinity as the full-length THAP1-IgG1 fusion protein (∼100 nmol/L), as determined by SPR (Fig. 3B). In contrast, Chemotrap-189 did not bind CCL21 (data not shown).

Figure 3.

Chemotrap-1 binds CCL21 and blocks CCL21-induced melanoma cell migration in vitro. A, the chemokine-binding peptide of THAP1 was fused to the Fc region of human IgG1 to generate Chemotrap-1, a fully human chemokine-binding Fc fusion protein. B, SPR analysis of Chemotrap-1/CCL21 association. Chemotrap-1 was immobilized on the sensor chip as an IgG1-Fc fusion protein and incubated with increasing concentrations of soluble CCL21 protein. The Kd for the Chemotrap-1/CCL21 association had a calculated value of 102 nmol/L. C, Chemotrap-1 blocks CCL21-induced melanoma cell chemotaxis in vitro. Cell migration of medium containing cells transfected with 150 ng/mL CCL21 or 150 ng/mL. Columns, mean; bars, SE. ***, P < 0.01, compared with medium and Chemotrap-189. ANOVA, Student-Neuman-Keuls test. D, Chemotrap-1 has a high affinity for inhibition of CCL21-mediated migration. A375P-CCR7 cells were subjected to migration assay toward 150 ng/mL CCL21 or increasing concentrations of Chemotrap-1. The IC50 was 77 pmol/L. E, Chemotrap-1 inhibition of migration is restricted to CCR7 ligands. Migration of A375P-CCR7 melanoma cells across polycarbonate filters toward 150 ng/mL of CCL21, CCL19, CXCL12, or CXCL10 was measured in the presence or absence of 1 μg/mL Chemotrap-1. +, P < 0.05, compared with control (value of 1); *, P < 0.05; ***, P < 0.001, compared with cytokine alone, one-way ANOVA, Dunnet's test.

Figure 3.

Chemotrap-1 binds CCL21 and blocks CCL21-induced melanoma cell migration in vitro. A, the chemokine-binding peptide of THAP1 was fused to the Fc region of human IgG1 to generate Chemotrap-1, a fully human chemokine-binding Fc fusion protein. B, SPR analysis of Chemotrap-1/CCL21 association. Chemotrap-1 was immobilized on the sensor chip as an IgG1-Fc fusion protein and incubated with increasing concentrations of soluble CCL21 protein. The Kd for the Chemotrap-1/CCL21 association had a calculated value of 102 nmol/L. C, Chemotrap-1 blocks CCL21-induced melanoma cell chemotaxis in vitro. Cell migration of medium containing cells transfected with 150 ng/mL CCL21 or 150 ng/mL. Columns, mean; bars, SE. ***, P < 0.01, compared with medium and Chemotrap-189. ANOVA, Student-Neuman-Keuls test. D, Chemotrap-1 has a high affinity for inhibition of CCL21-mediated migration. A375P-CCR7 cells were subjected to migration assay toward 150 ng/mL CCL21 or increasing concentrations of Chemotrap-1. The IC50 was 77 pmol/L. E, Chemotrap-1 inhibition of migration is restricted to CCR7 ligands. Migration of A375P-CCR7 melanoma cells across polycarbonate filters toward 150 ng/mL of CCL21, CCL19, CXCL12, or CXCL10 was measured in the presence or absence of 1 μg/mL Chemotrap-1. +, P < 0.05, compared with control (value of 1); *, P < 0.05; ***, P < 0.001, compared with cytokine alone, one-way ANOVA, Dunnet's test.

Close modal

To determine whether Chemotrap-1 could inhibit CCL21 chemokine-mediated migration of cancer cells, we seeded A375P-CCR7–transfected melanoma cells on polycarbonate inserts and measured migration across those inserts to CCL21, or CCL21 and Chemotrap-1. Chemotrap-1 (1 μg/mL) significantly inhibited migration of these melanoma cells toward 150 ng/mL CCL21 (Fig. 3C). To determine the potency of Chemotrap-1 in vitro, we performed a dose-escalating study using 150 ng/mL CCL21 as a promigration agent. Figure 3D shows that Chemotrap-1 dose dependently inhibited CCL21-mediated migration with an IC50 of 77 pmol/L, significantly more potent than to immobilized CCL21 on a chip in vitro. To determine the specificity of Chemotrap-1 in vitro, we investigated its effect on migration of A375P-CCR7 cells induced by three different chemokines: CCL19 (binds CCR7), CXCL12 (binds CXCR4), and CXCL10 (binds CXCR3). Any of the four chemokines (150 ng/mL) induced migration of melanoma cells, but only the CCR7 ligands CCL21 and CCL19 were inhibited by Chemotrap-1 (Fig. 3E).

Chemotrap-1 blocks migration of metastatic melanoma cells toward lymphatics in vitro and in vivo

We previously showed metastatic melanoma cells migrating toward LEC conditioned medium (CM) in a CCL21-dependent manner (11). To determine whether this could be inhibited by Chemotrap-1, A375 cells were transfected with expression vectors for Chemotrap-1 (pcDNA3.1–Chemotrap-1) or Chemotrap-189 (pcDNA3.1–Chemotrap-189). Expression of the Fc fusion protein was tested by immunoblotting, and the transfected A375 cells were then seeded into the top of Boyden chambers. LEC CM was placed on the bottom part of the chamber, and A375 migration was measured over 24 hours. Figure 4A shows that expression of Chemotrap-1, but not Chemotrap-189, resulted in complete inhibition of migration of A375 cells toward LEC CM. We have also previously shown that when implanted into nude mice 8 to 10 mm from a depot of human LECs, metastatic, but not nonmetastatic, melanomas grow toward the LEC injection depot. To determine whether Chemotrap-1 could inhibit directed metastatic growth, 1 × 106 A375 cells transfected with pcDNA3.1–Chemotrap-1, pcDNA3.1–Chemotrap-189, or pcDNA3 (control vector) were implanted s.c. into mice 8 to 10 mm from a simultaneous implantation of 1 × 105 human LEC. After 14 days, tumors were apparent in all mice. Whereas tumors in mice injected with the pcDNA3– and pcDNA3.1–Chemotrap-189–transfected cells grew significantly back toward the LEC depot, the pcDNA3.1–Chemotrap-1–transfected cells grew in the location where they were injected (Fig. 4B). Tumors were excised and pronated, and the area of tumor on each side of the injection site was quantified. Figure 4C shows that Chemotrap-1 expression resulted in a statistically significant inhibition of tumor growth toward LEC to a level equivalent to nondirected growth (toward the site of a saline injection).

Figure 4.

Chemotrap-1 inhibits directed growth of melanoma toward LECs. A, migration of melanoma cells or melanoma cells expressing empty vector (pcDNA3), negative control Fc (Chemotrap-189), or Chemotrap-1 was measured in vitro to LEC CM. B, melanoma cells as above were injected into nude mice 8 to 10 mm from a depot of LEC, and direction of tumor growth was measured. The red highlighted area is growing toward, and the blue away, from the LEC. C, the % growth toward the LEC is reduced in A375 cells expressing Chemotrap-1 to the same level as in the absence of LECs (PBS). Columns, mean; bars, SE. ***, P < 0.001; **, P < 0.01; *, P < 0.05, compared with pcDNA3 control (ANOVA).

Figure 4.

Chemotrap-1 inhibits directed growth of melanoma toward LECs. A, migration of melanoma cells or melanoma cells expressing empty vector (pcDNA3), negative control Fc (Chemotrap-189), or Chemotrap-1 was measured in vitro to LEC CM. B, melanoma cells as above were injected into nude mice 8 to 10 mm from a depot of LEC, and direction of tumor growth was measured. The red highlighted area is growing toward, and the blue away, from the LEC. C, the % growth toward the LEC is reduced in A375 cells expressing Chemotrap-1 to the same level as in the absence of LECs (PBS). Columns, mean; bars, SE. ***, P < 0.001; **, P < 0.01; *, P < 0.05, compared with pcDNA3 control (ANOVA).

Close modal

Chemotrap-1 prevents in-transit metastasis of melanoma cells in lymphatics in vivo

CCR7 overexpression increases lymph node metastasis of B16 mouse metastatic melanoma cells, resulting from uptake into and tracking along the local lymphatics (24). To determine whether Chemotrap-1 could inhibit this tracking, in a CCR7-dependent manner, we imaged tumors growing in CD1 mice by bioluminescence. B16 melanomas formed tumors that were elongated (Fig. 5A). The maximum major to minor axis ratio was 4.1 ± 1.1 (mean ± SEM, n = 6; Fig. 5A, i). In contrast, tumors expressing Chemotrap-1 had a mean ratio of 2.5 ± 0.2 (Fig. 5A, iii), which was significantly lower than that observed with tumors expressing Chemotrap-189 (3.9 ± 0.5; Fig. 5A, ii). Moreover, on excision, tracking of the black melanomas (Fig. 5A, iv) could be seen along lymphatics in one third of the B16 melanomas and in 60% of the cells expressing Chemotrap-189, whereas this tracking was rarely seen with cells expressing Chemotrap-1 (17%; Fig. 5A, v). CCR7/Chemotrap-189–cotransfected cells all showed evidence of tracking (Fig. 5B, i, ii, and iv). In contrast, CCR7–Chemotrap-1–transfected cells showed a significant inhibition of tracking (Fig. 5A, iii and v) to 33% (P < 0.01, χ2 test; Fig. 5C), indicating that coexpression of Chemotrap-1 completely inhibited the effect of CCR7 overexpression. To determine whether recombinant Chemotrap could prevent metastasis in vivo, we used a CCR7-dependent model of lymph node metastasis in syngeneic mice. B16-Luc cells were stably transfected with CCR7, and 50 μL of 1 × 105 cells were injected into the footpad of 10 C57/BL6 mice. Three animals were treated twice weekly with 50 μg Chemotrap-1 by i.p. injection. After 22 days, the mice were injected with luciferin and tumors were imaged. Mice were killed, and lymph nodes were exposed and imaged. Figure 5D shows luminescence of the primary tumor and of a lymph node metastasis in the popliteal lymph node. Metastases were seen in all seven control mice, but in only one of three treated mice (P < 0.02, χ2 test), suggesting that systemic administration could reduce lymph node metastases in this mouse model.

Figure 5.

Chemotrap-1 inhibits tracking of melanoma cells into and along lymphatics. A, B16-Luc melanoma cells (i), transfected with Chemotrap-189 (ii) or Chemotrap-1 (iii), were injected into CD1 mice, and circularity of tumor growth was measured by bioluminescence. On excision, tracking (arrow) of the tumor was seen in control (iv) but not in Chemotrap-1–transfected (v) tumors. B, effect of overexpression of CCR7 (i) or CCR7 and Chemotrap-189 (ii) increased nonuniform growth but CCR7–Chemotrap-1 expression again results in more circular growth (iii). Tracking was increased in CCR7-transfected tumors (iv) but reduced in CCR7- and Chemotrap-1–expressing tumors (v). C, percentage of tumor tracking. **, P < 0.01; *, P < 0.05, χ2 test, compared with control. D, C57/BL6 mice were injected with B16-Luc–CCR7 melanoma cells in the footpad. Twenty-two days later, the tumors were imaged by bioluminescence (i) and then the mouse was killed and the lymph node was exposed and imaged (ii). E, Chemotrap-1 treatment significantly reduced the incidence of lymph node metastasis (P < 0.02, χ2 test). F, CCL21 DuoSet ELISA kit was used to determine the concentration of CCL21 in the LEC CM and LEC CM incubated with B16 cells expressing Chemotrap-1 or Chemotrap-189. *, P < 0.05, Dunnet's test.

Figure 5.

Chemotrap-1 inhibits tracking of melanoma cells into and along lymphatics. A, B16-Luc melanoma cells (i), transfected with Chemotrap-189 (ii) or Chemotrap-1 (iii), were injected into CD1 mice, and circularity of tumor growth was measured by bioluminescence. On excision, tracking (arrow) of the tumor was seen in control (iv) but not in Chemotrap-1–transfected (v) tumors. B, effect of overexpression of CCR7 (i) or CCR7 and Chemotrap-189 (ii) increased nonuniform growth but CCR7–Chemotrap-1 expression again results in more circular growth (iii). Tracking was increased in CCR7-transfected tumors (iv) but reduced in CCR7- and Chemotrap-1–expressing tumors (v). C, percentage of tumor tracking. **, P < 0.01; *, P < 0.05, χ2 test, compared with control. D, C57/BL6 mice were injected with B16-Luc–CCR7 melanoma cells in the footpad. Twenty-two days later, the tumors were imaged by bioluminescence (i) and then the mouse was killed and the lymph node was exposed and imaged (ii). E, Chemotrap-1 treatment significantly reduced the incidence of lymph node metastasis (P < 0.02, χ2 test). F, CCL21 DuoSet ELISA kit was used to determine the concentration of CCL21 in the LEC CM and LEC CM incubated with B16 cells expressing Chemotrap-1 or Chemotrap-189. *, P < 0.05, Dunnet's test.

Close modal

To confirm that Chemotrap-1 was being secreted from the transfected cells, we used a commercial DuoSet CCL21 ELISA. Figure 5F shows that the melanoma cells do not express CCL21 at levels detectable on the CCL21 DuoSet kit [confirmed by reverse transcription-PCR (RT-PCR); data not shown]. LEC expressed CCL21 protein (1.79 pg/mL per flask per hour) and mRNA as determined by RT-PCR (data not shown). When LEC CM was mixed with medium from Chemotrap-1–expressing cells (B16-CT1), there was a reduction in available CCL21, indicating binding of CCL21 to Chemotrap-1 to below the detection limit of the assay. In contrast, when LEC CM was added to Chemotrap-189, there was no reduction in CCL21.

Metastatic mechanisms are potential therapeutic targets, as cancers can be detected and excised before metastasis is apparent, yet spread months or even years after excision. For melanoma, prognosis is poor once metastasis has occurred and there are no widely accepted therapies. We show here that one mechanism of metastasis, chemokine-mediated LEC targeting, is potentially therapeutically amenable, as inhibition of the CCL21/CCR7 axis can prevent melanoma cells from recognizing and migrating toward areas of high lymphatic density and invading those lymphatics.

Chemokine receptors, including the CCL21 receptor CCR7 and the CXCL12 receptor CXCR4, have been associated with metastatic melanoma in humans and increased metastatic growth in mouse and human tumors in experimental animals (3, 25). Two nonmutually exclusive mechanisms have been proposed for CCR7-CCL21–mediated metastasis: metastatic chemotaxis toward areas of high lymphatic density (11) and autologous chemotaxis (26). Current experiments did not determine the relative contributions of these two mechanisms toward invasion of melanoma cells into the lymphatic system. Although we showed that secreted Chemotrap-1 bound to CCL21 secreted from LEC, and that recombinant Chemotrap-1 could inhibit metastasis, we do not exclude the potential for synthesis of Chemotrap-1 acting intracellularly or autocrinely on the tumor cells. We show here that CCR7-dependent lymphatic recognition by metastatic melanoma is inhibited by the CCL21-binding agent Chemotrap-1. Chemotrap-1 is a human IgG1-Fc fusion protein containing the CCL21 chemokine-binding domain of human THAP1, a physiologic regulator of endothelial cell proliferation (17, 27). CCL21 contains a functional nuclear localization sequence in its COOH-terminal extension and may localize to the nucleus under certain conditions,5

5J-P. Girard, unpublished data.

but the physiologic relevance of the THAP1/CCL21 interaction has not yet been shown. Chemotrap-1 is the first example of a human recombinant chemokine-binding Fc fusion protein. The THAP1 chemokine-binding domain, predicted to adopt a coiled-coil helical structure, presents no sequence or structural homologies with CCR7 and other chemokine receptors. Determination of the three-dimensional structure of the complex will be therefore required to understand the molecular basis for specific recognition of the chemokine core domain of CCL21 by the THAP1 chemokine-binding peptide. Inhibitors of chemokine receptors have been shown to be effective in animal models of melanoma, specifically the CXCR4 receptor inhibitor AMD3100 (25). However, there are no specific small-molecule inhibitors of CCR7, and CCR7-IgG1–soluble receptors cannot be prepared because of the seven-transmembrane nature of chemokine receptors.

Chemotrap-1 blocked CCR7-mediated tracking of metastatic melanoma cells into and along lymphatics, indicating that Chemotrap-1 can inhibit CCL21 activity in vivo. Another chemokine-binding protein, the M3 protein encoded by murine gammaherpesvirus-68, has previously been shown to bind CCL21 with high affinity and to inhibit CCL21-mediated recruitment of lymphocytes in vivo (28). A major advantage of Chemotrap-1 is that it is a fully human recombinant protein. The M3 protein also binds and blocks the activity of other chemokines in vivo (28), and it will be important in future studies to widen the chemokines against which Chemotrap-1 has been tested to understand the full binding potential of this protein.

Most human melanomas metastasize through the local lymphatics and, through the lymph node, gain access to the circulation and reach distant organs such as the lung, brain, and liver. We therefore used a model in experimental animals whereby lymphatic metastasis is generated from s.c. injection of primary tumor cells, in contrast with experimental models using i.v. injection of tumor cells. This is not the usual route for melanoma metastasis. In patients, melanomas can form secondary lesions either locally within the skin (local metastasis), along the lymphatics (in-transit metastasis), or distantly (lymph node, primarily, followed by brain, lung, liver, etc.). We show here that chemokine-binding agents can prevent melanoma migration toward areas of high lymphatic density, invasion into lymphatics, and in-transit metastasis.

The findings here indicate that inhibition of chemokines acting on tumor cells can prevent tumor migration toward and into draining lymphatics. This suggests that chemokine antagonists may be potential therapeutic agents. Further evidence for this would be required before moving to clinical studies, particularly in terms of generating pharmacologically relevant inhibitors, and protein therapeutics, but the results here show for the first time that an endogenous human protein sequence can be used to block lymphatic metastases.

M. Roussigné and J-P. Girard are inventors on a patent associated with this work.

We thank Dr. Thomas Clouaire for help with the generation of THAP1 deletion mutants, Dr. Luc Aguilar for expert advice on Chemotrap-1 studies, and the BIAcore technological platform of the IFR150-CHU Toulouse for SPR experiments.

Grant Support: Ligue Nationale contre le Cancer (“Equipe Labellisée Ligue 2009”; J-P. Girard), Wellcome Trust grant 083583 (D.O. Bates and S. Lanati), Skin Cancer Research Foundation (M.S. Emmett), Michael Brough Healing Foundation (D.B. Dunn), and British Heart Foundation grant BS06/009 (D.O. Bates).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Podgrabinska
S
,
Braun
P
,
Velasco
P
, et al
. 
Molecular characterization of lymphatic endothelial cells
.
Proc Natl Acad Sci U S A
2002
;
22
:
22
.
2
Cyster
JG
. 
Chemokines and the homing of dendritic cells to the T cell areas of lymphoid organs
.
J Exp Med
1999
;
189
:
447
50
.
3
Takeuchi
H
,
Fujimoto
A
,
Tanaka
M
,
Yamano
T
,
Hsueh
E
,
Hoon
DS
. 
CCL21 chemokine regulates chemokine receptor CCR7 bearing malignant melanoma cells
.
Clin Cancer Res
2004
;
10
:
2351
8
.
4
Heresi
GA
,
Wang
J
,
Taichman
R
, et al
. 
Expression of the chemokine receptor CCR7 in prostate cancer presenting with generalized lymphadenopathy: report of a case, review of the literature, and analysis of chemokine receptor expression
.
Urol Oncol
2005
;
23
:
261
7
.
5
Wang
J
,
Xi
L
,
Hunt
JL
, et al
. 
Expression pattern of chemokine receptor 6 (CCR6) and CCR7 in squamous cell carcinoma of the head and neck identifies a novel metastatic phenotype
.
Cancer Res
2004
;
64
:
1861
6
.
6
Ishida
K
,
Iwahashi
M
,
Nakamori
M
, et al
. 
High CCR7 mRNA expression of cancer cells is associated with lymph node involvement in patients with esophageal squamous cell carcinoma
.
Int J Oncol
2009
;
34
:
915
22
.
7
Muller
A
,
Homey
B
,
Soto
H
, et al
. 
Involvement of chemokine receptors in breast cancer metastasis
.
Nature
2001
;
410
:
50
6
.
8
Maekawa
S
,
Iwasaki
A
,
Shirakusa
T
, et al
. 
Association between the expression of chemokine receptors CCR7 and CXCR3, and lymph node metastatic potential in lung adenocarcinoma
.
Oncol Rep
2008
;
19
:
1461
8
.
9
Sancho
M
,
Vieira
JM
,
Casalou
C
, et al
. 
Expression and function of the chemokine receptor CCR7 in thyroid carcinomas
.
J Endocrinol
2006
;
191
:
229
38
.
10
Shang
ZJ
,
Liu
K
,
Shao
Z
. 
Expression of chemokine receptor CCR7 is associated with cervical lymph node metastasis of oral squamous cell carcinoma
.
Oral Oncol
2009
;
45
:
480
5
.
11
Shields
JD
,
Emmett
MS
,
Dunn
DB
, et al
. 
Chemokine-mediated migration of melanoma cells towards lymphatics—a mechanism contributing to metastasis
.
Oncogene
2007
;
26
:
2997
3005
.
12
Mohler
KM
,
Torrance
DS
,
Smith
CA
, et al
. 
Soluble tumor necrosis factor (TNF) receptors are effective therapeutic agents in lethal endotoxemia and function simultaneously as both TNF carriers and TNF antagonists
.
J Immunol
1993
;
151
:
1548
61
.
13
Moreland
LW
,
Baumgartner
SW
,
Schiff
MH
, et al
. 
Treatment of rheumatoid arthritis with a recombinant human tumor necrosis factor receptor (p75)-Fc fusion protein
.
N Engl J Med
1997
;
337
:
141
7
.
14
Economides
AN
,
Carpenter
LR
,
Rudge
JS
, et al
. 
Cytokine traps: multi-component, high-affinity blockers of cytokine action
.
Nat Med
2003
;
9
:
47
52
.
15
Holash
J
,
Davis
S
,
Papadopoulos
N
, et al
. 
VEGF-Trap: a VEGF blocker with potent antitumor effects
.
Proc Natl Acad Sci U S A
2002
;
99
:
11393
8
.
16
Peppel
K
,
Crawford
D
,
Beutler
B
. 
A tumor necrosis factor (TNF) receptor-IgG heavy chain chimeric protein as a bivalent antagonist of TNF activity
.
J Exp Med
1991
;
174
:
1483
9
.
17
Roussigne
M
,
Cayrol
C
,
Clouaire
T
,
Amalric
F
,
Girard
JP
. 
THAP1 is a nuclear proapoptotic factor that links prostate-apoptosis-response-4 (Par-4) to PML nuclear bodies
.
Oncogene
2003
;
22
:
2432
42
.
18
Aruffo
A
,
Kolanus
W
,
Walz
G
,
Fredman
P
,
Seed
B
. 
CD62/P-selectin recognition of myeloid and tumor cell sulfatides
.
Cell
1991
;
67
:
35
44
.
19
Walz
G
,
Aruffo
A
,
Kolanus
W
,
Bevilacqua
M
,
Seed
B
. 
Recognition by ELAM-1 of the sialyl-Lex determinant on myeloid and tumor cells
.
Science
1990
;
250
:
1132
5
.
20
Clouaire
T
,
Roussigne
M
,
Ecochard
V
,
Mathe
C
,
Amalric
F
,
Girard
JP
. 
The THAP domain of THAP1 is a large C2CH module with zinc-dependent sequence-specific DNA-binding activity
.
Proc Natl Acad Sci U S A
2005
;
102
:
6907
12
.
21
Berger
B
,
Wilson
DB
,
Wolf
E
,
Tonchev
T
,
Milla
M
,
Kim
PS
. 
Predicting coiled coils by use of pairwise residue correlations
.
Proc Natl Acad Sci U S A
1995
;
92
:
8259
63
.
22
Wolf
E
,
Kim
PS
,
Berger
B
. 
MultiCoil: a program for predicting two- and three-stranded coiled coils
.
Protein Sci
1997
;
6
:
1179
89
.
23
Burkhard
P
,
Stetefeld
J
,
Strelkov
SV
. 
Coiled coils: a highly versatile protein folding motif
.
Trends Cell Biol
2001
;
11
:
82
8
.
24
Wiley
HE
,
Gonzalez
EB
,
Maki
W
,
Wu
MT
,
Hwang
ST
. 
Expression of CC chemokine receptor-7 and regional lymph node metastasis of B16 murine melanoma
.
J Natl Cancer Inst
2001
;
93
:
1638
43
.
25
Vianello
F
,
Papeta
N
,
Chen
T
, et al
. 
Murine B16 melanomas expressing high levels of the chemokine stromal-derived factor-1/CXCL12 induce tumor-specific T cell chemorepulsion and escape from immune control
.
J Immunol
2006
;
176
:
2902
14
.
26
Shields
JD
,
Fleury
ME
,
Yong
C
,
Tomei
AA
,
Randolph
GJ
,
Swartz
MA
. 
Autologous chemotaxis as a mechanism of tumor cell homing to lymphatics via interstitial flow and autocrine CCR7 signaling
.
Cancer Cell
2007
;
11
:
526
38
.
27
Cayrol
C
,
Lacroix
C
,
Mathe
C
, et al
. 
The THAP-zinc finger protein THAP1 regulates endothelial cell proliferation through modulation of pRB/E2F cell-cycle target genes
.
Blood
2007
;
109
:
584
94
.
28
Jensen
KK
,
Chen
SC
,
Hipkin
RW
, et al
. 
Disruption of CCL21-induced chemotaxis in vitro and in vivo by M3, a chemokine-binding protein encoded by murine gammaherpesvirus 68
.
J Virol
2003
;
77
:
624
30
.