The CXCL12–CXCR4 Chemokine Pathway: A Novel Axis Regulates Lymphangiogenesis

The major cause of cancer mortality is metastasis that occurs via multiple pathways including lymphatic vasculature. A considerable number of studies has documented that lymphangiogenesis, the growth of lymphatic vessels, contributes to lymphatic metastasis in experimental tumor models and in clinic (1, 2), although it remains controversial in several types of human cancer (3, 4). Tumors can actively induce the growth of lymphatic vessels toward tumor tissues. Studies over the past decade have identified several growth factors and receptors, particularly vascular endothelial growth factor-C (VEGF-C), VEGF-D, and their receptor vascular endothelial growth factor receptor-3 (VEGFR-3) on lymphatic endothelial cells (LEC), contributing to LEC growth, migration, and survival (5, 6). Nevertheless, the precise mechanism underlying lymphangiogenesis is still a focus of intensive investigation.

The chemokine system, which is known to comprise more than 40 chemokines and 18 chemokine receptors to date, correlates with many biologic and pathologic processes, for instance, inflammation, HIV-infection, angiogenesis, and tumor growth (7–9). Chemokines are small monomeric cytokines of 8 to15 kDa. On the basis of a cysteine motif, they have been classified into 4 subgroups: CXC, CC, C, and CX3C (10). Chemokines interact with 7 transmembrane G-protein-coupled cell surface receptors and promote chemotaxis, a process that induces the directional cell migration toward a gradient of chemotactic cytokine (10). Though chemokine receptors were initially found on leukocytes, many nonhematopoietic cell types have been identified to express chemokine receptors. These chemokine/chemokine receptor interactions help coordinate cell trafficking and reorganization within various tissue compartments (11). Previous studies have documented that chemokines and their receptors play important roles in tumor development and angiogenesis (9), among which Augustin et al. (12) reported that blood endothelial cells express chemokine receptors CXCR1-4. Considering the abundant levels of chemokines existing in the local tumor and their advantages in directing cell migration (10, 13), it is reasonable to hypothesize that LECs also use the chemokine-mediated mechanism during the process of lymphangiogenesis, an event that requires directional migration of LECs, such as those inducing angiogenesis or regulating leukocyte trafficking.

In this study, we have analyzed specific chemokine patterns that contribute to the chemoattraction of LECs and the lymphangiogenesis process. We report here that mouse LECs (mLEC) express abundant amount of chemokine receptor CXCR4. CXCL12 executes lymphangiogenic activities via CXCR4. CXCL12, also known as stromal cell-derived factor-1 (SDF-1), is highly conserved with 99% homology between human and mouse, allowing it to act across species barriers (14). The CXCL12–CXCR4 axis is essential for many biologic processes, including development, hematopoiesis, organogenesis, as well as vascularization (15–18). Furthermore, CXCR4 is highly expressed in a variety of cancers (19). The interaction between CXCL12 and CXCR4 plays a prominent role in tumorigenesis (20) and metastasis (19, 21).

We found that among all chemokine receptors expressed in mLECs, VEGF-C stimulation dramatically upregulates the expression level of CXCR4 on lymphangiogenic endothelial cells compared with quiescent mLECs. Subsequent analyses suggest that hypoxia-inducible factor-1α (HIF-1α) is responsible for the upregulation of CXCR4 by VEGF-C. In addition, chemokine CXCL12 induces migration, tubule formation of mLECs, activates intracellular signaling of Akt and Erk, and stimulates lymphangiogenesis in a Matrigel plug assay in vivo, which can be impeded by blocking or knocking down CXCR4. Additionally, blocking the VEGFR-3 pathway has no effect on CXCL12-induced lymphangiogenic activities. However, antibodies targeting both CXCL12 and VEGF-C pathways show a more dramatic inhibitory effect on the tumor lymphangiogenesis and lymphatic metastasis in MDA-MB-231 breast cancer model. Taken together, our study provides direct evidence that the chemokine system is involved in regulating lymphangiogenesis and shows that the CXCL12–CXCR4 chemotactic axis plays a critical role in promoting lymphangiogenesis in a VEGFR-3 independent pattern.

Primary mLECs were isolated, characterized, and cultured in endothelial cell culture medium (ECM) as previously documented (2). Human LECs (hLEC) were purchased from Sciencell. B16/F10 and MDA-MB-231 cell lines were purchased from the American Type Culture Collection. Enhanced green fluorescent protein (eGFP)-labeled MDA-MB-231 cell lines (MDA-MB-231/eGFP) were constructed using the Kit from Genepharma according to the manufacturer's instructions. Hamster antimouse Podoplanin antibody, antibodies against Erk 1/2, phosphorylated Erk 1/2 (p-Erk 1/2), Lamin B, and Actin were from Santa Cruz Biotechnology. Rat antihuman Podoplanin antibody was from Biolegend. Akt antibody was from Bioworld Technology. P-Akt antibody was from Cell Signaling Technology. CXCR4 antibody was from Abcam. TRITC- and FITC-conjuncted secondary antibodies were from Santa Cruz Biotechnology. Anti-CXCR4, anti-VEGFR-3, anti-CXCL12, anti-VEGF-C-blocking antibodies were from Bioss. AMD3100, U0126, and LY294002 were from Sigma-Aldrich. VEGF-C and CXCL12 were from R&D Systems.

Cultured mLECs were starved overnight, and then stimulated with or without 100 ng/mL of VEGF-C for 24 hours. The total RNA from cultured mLEC was isolated using TRIZOL Reagent (Invitrogen), and converted into cDNA using the First Strand cDNA Synthesis Kit (Fermentas). Quantitative RT-PCR (qRT-PCR) was conducted using the Brilliant II SYBR Green qRT-PCR Master Mix Kit (Stratagene) with the standard PCR conditions applied in this study (40 cycles). All primers for chemokine receptors were listed in supplementary Table S1. Relative quantitation was analyzed using the ΔΔC_t method. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. Independent experiments were repeated in triplicates.

For detection of the cell surface expression level of CXCR4, mLECs were incubated with CXCR4 antibody (Abcam), or rabbit IgG as an isotype control, and subjected to the standard protocol of cytometry. The data were analyzed using a FACS Calibur flow cytometry system (Becton Dickinson).

The migration efficiency of mLECs was assessed using 8-μm-pore Transwell filter membrane (Costar) as previously described (2). Migrated cells were quantified by counting in 8 fields under an Olympus IX71 optical microscope (Olympus). Experiments were conducted in triplicate and repeated twice.

The tubule formation assay was conducted as previously described (2). Tubule structures were imaged by the Olympus microscope and quantified by measuring the length of cords in 6 randomly viewed fields. The tubule length was defined as previously described (22). Experiments were conducted in triplicate and repeated twice.

The siRNA for HIF-1α and siRNA (1#) for CXCR4 were purchased from Santa Cruz Biotechnology. The siRNA for Akt isoforms (Akt1, Akt2, Akt3) were purchased from GenePharma. Scrambled siRNA was purchased from GenePharma (Shanghai, China). The sequence of ineffective siRNA (2#) for CXCR4 is 5′-gagacuaugacuccaacaatt-3′. The sequence of siRNA for Akt1 is 5′-gcaccuuuauuggcuacaaggtt-3′, siRNA for Akt2 is 5′-aagaguggaugcgggcuaucctt-3′, and siRNA for Akt3 is 5′-aaggaugaaguggcacacacutt-3′. The siRNAs were transfected with Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. After 72-hour transfection, knock-down efficiency was detected by Western blot analysis.

Cells were harvested, denatured, and subjected to SDS-PAGE. Proteins were transferred to polyvinylidene difluoride (PVDF) membrane, immunolabeled with appropriated primary antibodies overnight at 4°C, incubated with horseradish peroxidase-conjugated secondary antibodies for 1 hour at room temperature, then detected with an enhanced chemiluminescence system (Roche) according to the manufacturer's protocol.

Proliferating mLECs were cultured on coverslips in 12-well plates in serum-containing ECM. After serum-starved overnight, the cells were treated with or without 100 ng/mL of VEGF-C for 24 hours. The cells were further fixed by 4% paraformaldehyde. According to the previous study (23), Podoplanin and CXCR4 were immunostained and detected by Nikon A1 laser scanning confocal microscope using 60X/1.49 Oil DIC objectives.

For the immunofluorescence of tissue samples, fluorescence images were detected by Nikon A1 laser scanning confocal microscope using Plan Apo 20X/0.75 objectives. Images were captured and analyzed with Nikon image software (NIS-Elements AR 3.0).

Lymphatic vessel formation in vivo was evaluated by the Matrigel plug assay. Briefly, 0.5 mL of Matrigel (Becton-Dickinson Labware) containing CXCL12, VEGF-C (500 ng/mL), antibodies (10 μg/mL), or AMD3100 (a CXCR4 antagonist that can block CXCL12 signaling, 50 μg/mL) at indicated concentrations was injected subcutaneously into the abdominal midline of BALB/c mouse (female, 5 weeks old, 5 per group). After 8 days, plugs were dissected and subjected to the immunofluorescent analysis. Evaluation of the lymphatic vessel density was assessed in 5 independent fields imaged by the Nikon A1 laser scanning confocal microscope using Nikon image software (NIS-Elements AR 3.0).

Multitumor tissue microarray was purchased from Xi'an Aomei (Aomei) that contained 54 clinical specimens (median age 55.6 years, range 15–81, male and female were about 31–23, including brain glioma, esophagus, stomach, liver, colon, rectum, lung, bladder, heart, kidney, thyroid, pancreas, cervix, skin, breast, ovarian, prostate, and testis, 3 specimens for each). Tissue microarray was immunostained with antihuman Podoplanin (Biolegend) and CXCL12 (Bioss) antibodies according to the protocol of immunofluorescent staining.

All animal studies were approved by the Institutional Animal Care and Use Committee of Tsinghua University. Constructed MDA-MB-231/eGFP cells (3 × 10⁶ in 100 μL of Matrigel solution) were inoculated into the mammary fat pad of nude mice (female, 6–8 weeks). The mice were divided randomly into 4 groups (6 mice per group). Control rabbit IgG (2 mg/kg), rabbit anti-CXCL12 antibody (2 mg/kg), rabbit anti-VEGF-C antibody (2 mg/kg), or anti-CXCL12 antibody (1 mg/kg) plus anti-VEGF-C antibody (1 mg/kg) were administered intraperitoneally into the mice every other day, respectively. After 3 weeks, primary tumor, axillary lymph nodes were dissected, photographed, and applied to immunofluorescent staining. Metastasized MDA-MB-231/eGFP cells in lymph node were detected by Nikon A1 laser scanning confocal microscope. Area of eGFP-positive signals in lymph nodes was assessed in at least 6 independent fields in different sections using Nikon image software (NIS-Elements AR 3.0).

Data are presented as mean. Statistical analyses were assessed by a 2-tailed Student's t-test.

To determine whether the chemokine attractant/chemokine receptor interaction participates in the lymphangiogenesis process, we systematically analyzed the mRNA levels of well-known chemokine receptors (CCR1–CCR10, CXCR1–CXCR7, and CX3CR1) in isolated primary mLECs. PCR results showed that primary mLECs prominently express chemokine receptors CCR5, CCR9, CXCR4, CXCR6, and CXCR7, however, weakly express chemokine receptors CCR4, CCR6, CCR8, CCR10, CXCR3, and CX3CR1 (Supplementary Fig. S1). VEGF-C, which is secreted by tumor cells and mast cells in tumor microenvironment, has been reported to be a predominant pro-lymphangiogenesis factor (5). We wondered whether VEGF-C stimulation can regulate the expression of chemokine receptors on LECs. Intriguingly, qRT-PCR showed that only CXCR4 mRNA level was dramatically increased upon VEGF-C treatment, among those chemokine receptors (CCR4-6, CCR8-10, CXCR3, CXCR4, CXCR6, CXCR7, and CX3CR1) expressed on mLECs (Fig. 1A), indicating that LECs selectively express certain chemokine receptors, and VEGF-C stimulation can specifically upregulate CXCR4 expression in lymphangiogenic endothelial cells.

To verify this result, cultured mLECs were starved of endothelial cell growth supplement overnight, and then stimulated by VEGF-C, followed by flow cytometry analyses. As expected, the level of cell surface CXCR4 protein was much higher after 24-hour culture with VEGF-C (Fig. 1B). Consistent observation was obtained from immunofluorescence assay showing that CXCR4 expression is associated with the proliferative status of mLECs and hLECs (Supplementary Fig. S2A). The regulation of CXCR4 surface expression was also analyzed by immunoblotting. VEGF-C could dramatically upregulate CXCR4 expression, as shown in Fig. 1C. Because CXCR4 is one of the target genes of HIF-1α (24), we hypothesized that VEGF-C upregulates CXCR4 via HIF-1α. Indeed, VEGF-C treatment increased the expression level of HIF-1α (Fig. 1C). HIF-1α accumulation in mLECs induced by cobalt chloride treatment and hypoxia exposure increased the expression of CXCR4 (Supplementary Fig. S2B). When HIF-1α was knocked down in mLECs by transfection of HIF-1α siRNA, VEGF-C failed to increase the expression of CXCR4 in comparison with scrambled siRNA (Fig. 1D). Collectively, these results provide compelling evidence that VEGF-C is a positive-regulator of CXCR4 in LECs, in which HIF-1α is involved.

We were therefore prompted to explore the expression of CXCR4 on lymphatic vessels in vivo. Stained tissue sections from Matrigel plugs, dissected 8 days after injected into mice subcutaneously, revealed that VEGF-C-induced lymphangiogenic vessels express higher levels of CXCR4, as compared with Matrigels without VEGF-C (Fig. 2A). Then we further investigated the distribution of CXCR4 on lymphatic vessels in both normal tissues and tumor tissues. Investigation on mouse normal intestine, colon, and lymph node confirmed that matured lymphatic vessels do not express detectable level of CXCR4. In comparison, strong expression of CXCR4 was observed in melanoma tissues and in lymph nodes from tumor-bearing mice (Fig. 2B). We also observed high expression of CXCR4 on lymphatic vessels in human rectal cancer, colon cancer, and skin squamous cell carcinoma (Supplementary Fig. S2C). These immunofluorescence studies in vivo confirmed that CXCR4 expression is upregulated on lymphangiogenic vessels.

Because CXCL12 is one of the major chemokines corresponding to CXCR4 (14), we speculated that CXCL12 has the potential to attract CXCR4-positive lymphangiogenic endothelial cells. In a chemotaxis assay, mLECs were seeded in the upper chamber of a Transwell insert, and CXCL12 was added in the lower chamber. The results showed that CXCL12 induced a dose-dependent increase in cell migration (Fig. 3A). We also observed that CXCL12 promotes the migration of hLECs, which could be blocked by AMD3100, a well-known antagonist of CXCR4 (Supplementary Fig. S3). In addition, CXCL12 also had an effect on mLECs tubule formation in a dose-dependent manner, as assessed by measuring the length of cords formed by mLECs (Fig. 3B). To further confirm this result, we conducted a Matrigel plug assay in vivo. Plugs supplemented with different doses of CXCL12 were injected into mice subcutaneously. VEGF-C served as a positive control. In agreement with the findings in vitro, Matrigels with CXCL12 showed significantly enhanced lymphatic vessel formation with a tubule-like structure compared with control group (Fig. 3C). To test the relationship between lymphatic vessel density and CXCL12 level in human cancers, expression levels of CXCL12 and Podoplanin, a specific marker for lymphatic endothelium, were examined in a multitumor tissue microarray containing a total of 54 clinical tissue samples. Immunostaining analyses revealed that the density of tumor-associated lymphatic vessels was positively correlated with the expression level of CXCL12 in 50 samples among 54 specimens detected (Fig. 3D and Supplementary Table S2). Taken together, these results show that CXCL12 is a positive regulator of lymphangiogenesis.

Stimulation of mLECs motility by CXCL12 suggests that this chemokine can activate intracellular signaling pathways. Previous studies have documented a direct involvement of PI3K/Akt and Erk signaling pathway in cell migration mediated by the CXCL12–CXCR4 axis (25, 26). As expected, increased phosphorylation of Erk 1/2 and Akt was detected in mLECs upon CXCL12 stimulation (Fig. 4A). When CXCR4 was blocked by its neutralizing antibody, the phosphorylation of Erk 1/2 and Akt in response to CXCL12 was significantly decreased, as compared with control IgG (Fig. 4B). Similarly, knocking down of CXCR4 using effective siRNA eliminated the effect of CXCL12 on activation of Erk 1/2 comparing to scramble siRNA or ineffective CXCR4 siRNA (Fig. 4C). To evaluate the contributions of the Akt and/or Erk 1/2 pathways to CXCL12-induced chemotaxis, the inhibitors of PI3k/Akt (LY294002) or Erk 1/2 (U0126) were used to pretreat mLECs, which dramatically attenuated the migrated mLEC number induced by CXCL12 (Fig. 4D). Because Akt has 3 different isoforms and mediates distinct functions, we thus used siRNAs to knock down Akt1, Akt2, and Akt3, respectively. Chemotaxis assay revealed that knocking down Akt1 dramatically inhibits the function of CXCL12, whereas either Akt2 siRNA or Akt3 siRNA has slight effect, showing that Akt1, other than Akt2 or Akt3, plays a principal role in the CXCL12-induced migration of mLECs (Supplementary Fig. S4). In summary, these results show that activation of Akt1 and Erk 1/2 signaling components is essential for the chemotactic activity of CXCL12 on lymphangiogenic endothelial cells.

The above-mentioned evidences showed that CXCL12 is a chemoattractant for lymphangiogenic endothelial cells. However, whether CXCL12 stimulates lymphangiogenesis directly via CXCR4 or indirectly via other pathways is still unknown. Among the known prolymphangiogenic factors, VEGF-C and -D are the most specific and potent growth factors which directly bind to VEGFR-3 on LECs (5, 6). In addition, it was also reported that the abilities of basic fibroblast growth factor (bFGF), VEGF-A, and hepatocyte growth factor to induce lymphangiogenesis are mediated at least in part via the VEGF-C/VEGF-D/VEGFR-3 pathway (27–29). To investigate whether CXCL12 may employ the similar pathway to execute its prolymphangiogenic functions, we treated mLECs with CXCL12 in the presence or absence of VEGFR-3-neutralizing antibody. Surprisingly, incubation of mLECs with VEGFR-3-neutralizing antibody did not affect CXCL12-induced mLECs migration, whereas VEGF-C-induced mLECs migration was dramatically inhibited (Fig. 5A). Therefore, we detected the effect of inhibiting CXCL12–CXCR4 pathway on the function of VEGF-C/VEGFR-3 pathway. When we used AMD3100 or CXCR4-neutralizing antibody to treat mLECs in a chemotaxis assay, the CXCL12-mediated migration of mLECs was effectively blocked, but did not exert any significant inhibitory effect against VEGF-C-induced cell migration (Fig. 5B). In agreement with the in vitro results, Matrigel plug assay in vivo further confirmed that blocking CXCR4 by CXCR4-neutralizing antibody or AMD3100 inhibited the function of CXCL12 without affecting the VEGF-C activity. Moreover, blocking VEGFR-3 dramatically inhibited the function of VEGF-C, whereas it did not have any effect on CXCL12-induced lymphangiogenesis (Fig. 5C). Taken together, these results show that the CXCL12–CXCR4 pathway is independent of the VEGF-C/VEGFR-3 pathway in LEC chemoattraction and lymphangiogenesis. This conclusion means that in addition to the VEGF-C/VEGFR-3 pathway, we discovered a novel CXCL12–CXCR4 pathway regulating lymphangiogenesis.

On the basis of our findings, it is naturally to speculate that simultaneous inhibition of CXCL12–CXCR4 and VEGF-C/VEGFR-3 pathways should result in more potent blockade of tumor lymphangiogenesis and lymphatic metastasis. First, chemotaxis assay in vitro (Supplementary Fig. S5) and Matrigel plug assay in vivo (Fig. 6A) showed that CXCL12 and VEGF-C have additive effect on promoting migration of mLECs. Second, in an orthotopic breast cancer model in immunodeficient mice, eGFP-labeled human breast carcinoma cell line MDA-MB-231 (MDA-MB-231/eGFP) was injected orthotopically into the mammary fat pad of nude mice. Intraperitoneally treatment every other day with either CXCL12-neutralizing antibody, VEGF-C-neutralizing antibody, or together for 3 weeks inhibited the primary tumor growth (Supplementary Fig. S6). A significant decrease in lymphatic vessel density in primary tumor tissue was observed in either anti-CXCL12- or anti-VEGF-C-treated mice (Fig. 6B). Furthermore, combination of both anti-CXCL12 and anti-VEGF-C blocking antibodies showed an even more efficient inhibition in tumor lymphangiogenesis than each antibody alone (Fig. 6B). Because tumor lymphangiogenesis actively contributes to cancer dissemination (30), and because it was reported that CXCL12 is associated with lymph node metastasis (20, 21), thus axillary lymph nodes were dissected from tumor-bearing mice. Our results showed that treatment with both anti-CXCL12 and anti-VEGF-C antibodies led to a significant reduction in lymph node metastasis (Fig. 6C). As MDA-MB-231 cell line was labeled with eGFP, disseminated tumor cells were further analyzed by confocal fluorescence microscopy (Fig. 6D). Quantified results showed that blockade of CXCL12 dramatically impedes the tumor cell metastasis to the lymphatic system, and targeting both CXCL12 and VEGF-C pathways shows an additive effect on prevention of lymphatic metastasis (Fig. 6D).

Chemokines and their receptors have been documented in tumor cell growth, angiogenesis and metastasis (9, 31). However, few studies have explored the chemokine system in the tumor-lymphatic microenvironment. Tumor cells, when expressing receptors for lymphatic-derived chemokines, may gain access to lymphatic vessels via chemoattraction. Kim and coworkers (32) reported that LECs promote the migration of CXCR4-positive tumor cells by secretion of CXCL12. Shields and coworkers (33) reported that tumor cells utilize interstitial flow and CCR7 signaling to access lymphatics. They found that VEGF-C expressed by tumor cells induces upregulation of chemokine CCL21, a ligand of CCR7, by lymphatic vessels, which in turn guides the migration of tumor cells toward lymphatics (34). Herein, our study provides strong evidence that VEGF-C stimulation can also induce upregulation of chemokine receptor CXCR4, not only just chemokines such as CCL21, on LECs, and the chemokine/chemokine receptor interaction promotes lymphangiogenesis by directly acting on LEC chemotaxis.

The lymphatic vasculature is a network that transports interstitial fluid from tissues to the blood. Previous studies have shown that lymphangiogenesis, growth of lymphatic vessels in the solid tumors, correlates with lymphatic metastasis. However, the exact mechanism for lymphangiogenesis remains unclear. One of the major directions related to lymphangiogenesis is how LECs being regulated. Both Alitalo group (35) and Detmar group (36) have investigated the differences of gene profile between LECs and blood endothelial cells. Their results indicate that LECs express a large number of distinct genes involved in multiple endothelial cell functions such as inflammatory processes and cell–cell interactions. In addition, tumor-related LECs have also been showed having distinct profile from that of normal LECs (37). Given the critical role of the chemokine system in guiding cells migration, as the foregoing analyses, we focus on the chemokine system and have identified several chemokine receptors expressed on LECs. The present study highlights the variation of gene profile, chemokine receptors in particular, in LECs in response to the stimulation of VEGF-C. It is possible that lymphangiogenic factors can activate the lymphatic endothelium to express increased amounts of chemotactic factors and receptors, which contribute to the migration of LECs toward tumor tissues and LEC-tumor cell interactions. Indeed, we have shown that, among all chemokine receptors expressed in LECs, VEGF-C specifically upregulates CXCR4. Because other chemokine receptors, such as CCR5, CCR9, CXCR6, and CXCR7, are also expressed by LECs, we can not exclude the possibility that other growth factors or inflammatory factors may regulate the expression of those chemokine receptors, which further mediate the chemoattraction of activated LECs. Therefore, exploring the role of the chemokine system on the properties of lymphangiogenic endothelial cells and the lymphangiogenesis process will be of considerable interest in the immediate future.

It has been reported that tumor cells and carcinoma-associated fibroblasts express high levels of CXCL12 (13, 38), one of the major chemokines corresponding to CXCR4. Here, we emphasize that the CXCL12–CXCR4 axis plays a pivotal role in promoting the chemoattraction of LECs. Additionally, we found that LECs also express high level of CXCR7, which has been reported to be a novel receptor for CXCL12 by Bachelerie group (39) and Schall group (40), respectively. However, unlike CXCR4, CXCL12 activation of CXCR7 results in a proliferative effect but not cell migration. Further, CXCR7 expression usually endorses tumor cells with a growth and survival advantage and increased adhesion properties (40, 41). In this study, VEGF-C stimulation exclusively upregulates CXCR4 but not CXCR7 on LECs. Thus, we focus on the role of the CXCL12–CXCR4 axis in the chemoattraction of LECs. Nevertheless, the function of CXCR7 on LECs remains to be further explored. As the expression of CXCR7 can be upregulated on the activated blood endothelial cells after stimulation with the cytokines TNF-α and IL-1β (40), it is possible that lymphangiogenic endothelial cells express high levels of CXCR7 in the tumor microenvironment, which in turn supports lymphangiogenic endothelial cells growth and lymphangiogenesis.

Another major concern is the regulation mechanism of CXCR4 expression in LECs. Many studies have stated that CXCR4 is expressed in a variety of tumors, supporting its role in cell survival, proliferation, adhesion, and migration (19, 20). In addition, CXCR4 also localizes on vascular endothelial cells and mediates the angiogenic activity of CXCL12 (42). It has been described that angiogenic factors VEGF and bFGF can upregulate the expression of CXCR4 on vascular endothelial cells (12, 43). Consistently, our result shows inducible expression of CXCR4 in cultured mLECs by VEGF-C, whereas the expression level of CXCR4 in matured lymphatic vessels is much lower. Though we only choose VEGF-C as the stimulator, it is still possible that activation of the lymphatic system by other lymphangiogenic factors like VEGF-A, platelet-derived growth factor BB (PDGF-BB), hepatocyte growth factor, may also increase the CXCR4 expression on LECs. Our group (44) has reported that PDGF-BB can stimulate the CXCL12–CXCR4 axis during tumor angiogenesis. Moreover, we have also observed that CXCL12 stimulation led to an upregulation of CXCR4 on LECs (Data not shown), indicating the possibility that CXCR4 expression is, at least in part, regulated under an autocrine manner. This autocrine fashion might explain the fact that CXCL12 alone can also induce considerable number of newly formed lymphatic vessels in a Matrigel plug assay. On the other hand, it is known that HIF-1α, a central mediator of hypoxia, can induce a significant increase in the expression of CXCR4 on normal and malignant cells through the von Hippel–Lindau-HIF-1α pathway (24). In our study, both cobalt chloride treatment and hypoxia exposure increase the expression of CXCR4 on LECs. It is possible that the CXCL12–CXCR4 axis is related to lymphangiogenesis under tissue repair of hypoxic–ischemic injury. Therefore, the regulation of CXCR4 expression in LECs is sophisticated in that it might be regulated in both a paracrine and an autocrine manner, involving lymphangiogenic factors and/or hypoxia factors under different pathologic conditions.

Initial studies characterized CXCL12 as a pre-B cell survival factor (45). Considerable clinical-pathologic studies have showed that CXCL12 and CXCR4 expression are significantly associated with lymph node metastasis (20, 21). However, the mechanism that how the CXCL12–CXCR4 axis regulates lymphatic metastasis remains unclear. CXCR4-positive tumor cells may metastasize to areas with high CXCL12 expression (46). Hirakawa and coworkers (47) have reported that tumor associated LECs express CXCL12 and attract invasive Paget cells via the CXCL12–CXCR4 axis. In turn, the present study shows that functionally activated LECs also express increased level of CXCR4. The CXCL12–CXCR4 axis is involved in promoting the chemoattraction of LECs and their subsequent lymphangiogenesis, which contributes to lymph node metastasis. Therefore, in addition to its angiogenic activity (38), the CXCL12–CXCR4 axis should play a more pivotal role in regulating tumor growth and metastasis than previously thought.

Another intriguing aspect of this study is the observation that the CXCL12–CXCR4 axis is independent of VEGFR-3 pathway in promoting lymphangiogenesis, even though VEGF-C can induce the expression of CXCR4. Among the known lymphangiogenic factors, VEGF-C, which exerts its functions via VEGFR-3, is the most potent and specific growth factor acting directly on LECs. Although VEGF-A, bFGF, angiopoient-1/2, insulin-like growth factor-1/2 (IGF-1/2), hepatocyte growth factor, PDGF-BB have been shown to be prolymphangiogenesis factors, the VEGF-C-VEGFR-3 signaling is a common pathway for several lymphangiogenic factors. The bFGF was reported to upregulate VEGF-C expression in endothelial cells, and its lymphangiogenic property is mediated by VEGF-C (27). The VEGF-A-induced lymphangiogenesis may partially mediated by the VEGF-C/-D-VEGFR-3 signaling (28). Tie-2, IGF-1/2 receptor, and PDGF receptor have been detected on LECs, and their respective ligands are shown to promote lymphangiogenesis (48–50). Intriguingly, blockage of VEGF-C/-D/VEGFR-3 inhibits the angiopoient-1-induced lymphatic sprouting but does not affect the lymphangiogenic activities of IGF-1/2 and PDGF-BB, indicating that IGF-1/2 and PDGF-BB activate a signaling pathway independent of that triggered by the VEGF-C/-D/VEGFR-3 axis. Our present study shows that CXCL12 is a direct chemoattractant for LECs. In contrast to the antibody of CXCR4, the VEGFR-3-neutralizing antibody has no effect on the cell chemoattraction of LECs stimulated by CXCL12 as well as CXCL12-induced lymphangiogenesis in vivo. Although we can not exclude the possibility that CXCL12 might also have indirect effects in promoting lymphangiogenesis via inducing other factors, the observed effects of the CXCL12–CXCR4 axis on LECs chemotaxis show that it at least has a direct impact on certain steps of lymphangiogenesis. Importantly, we are more interested in the additive effect of CXCL12 and VEGF-C in promoting cell migration of LECs. It seems that CXCL12 is at least as potent as, if not more than, VEGF-C in chemoattraction of LECs according to our results, whereas VEGF-C has been shown to execute a potent mitogenic effect on LECs. We assume that these 2 factors may act in a synergistic manner by conducting different functions in lymphangiogenesis. Having shown that a combination of anti-CXCL12 and anti-VEGF-C antibodies more efficiently inhibit tumor lymphangiogenesis and lymph node metastasis than individual antibody alone, therefore, the development of antagonists for both CXCR4 and VEGFR-3 pathway should be a very promising approach for the control of tumor metastasis and lymphangiogenesis-related diseases.

In conclusion, our study shows that the CXCL12–CXCR4 axis is a potent positive-regulator of lymphangiogenesis by directly acting on LECs. Of all chemokine receptors analyzed in LECs, only CXCR4 is significantly upregulated by VEGF-C compared with that in quiescent LECs. CXCL12 induces chemotaxis of LECs in vitro and lymphangiogenesis in vivo mediated via CXCR4. Interestingly, lymphangiogenesis triggered by the CXCL12–CXCR4 axis is independent of the VEGF-C-VEGFR-3 pathway. Targeting both CXCL12 and VEGF-C by neutralizing-antibodies results in an additive inhibitory effect on both tumor lymphangiogenesis and lymphatic metastasis. This study, for the first time, points out that the chemokine/chemokine receptor system is directly involved in the process of lymphangiogenesis, and provides a novel route to inhibit lymph node metastasis via targeting both chemokines and lymphangiogenic growth factors, which must attract intensive investigation in this direction in the future.

No potential conflicts of interest were disclosed.

Conception and design: W. Zhuo, Y. Luo

Development of methodology: W. Zhuo, N. Song, X. Song, Y. Fu, Y. Luo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Zhuo, L. Jia, X. Lu, Y. Luo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Zhuo, L. Jia, N. Song, X. Lu, X. Wang, X. Song, Y. Fu, Y. Luo

Writing, review, and/or revision of the manuscript: W. Zhuo, L. Jia, N. Song, Y. Ding, X. Wang, Y. Luo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Zhuo, Y. Fu, Y. Luo

Study supervision: W. Zhuo, Y. Luo

The authors thank the members of the Luo laboratory for their kind suggestions on this work.

This work was supported in part by the General Programs of the National Natural Science Foundation of China (No. 81071742, No. 81171998 and No. 81171999) and the Doctoral Fund of the New Teacher Programe of Ministry of Education of China (No. 20110002120039).

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.