Preclinical and clinical studies positively correlate the expression of vascular endothelial growth factor (VEGF)-C in tumors and the incidence of lymph node metastases. However, how VEGF-C regulates individual steps in the transport of tumor cells from the primary tumor to the draining lymph nodes is poorly understood. Here, we image and quantify these steps in tumors growing in the tip of the mouse ear using intravital microscopy of the draining lymphatic vessels and lymph node, which receives spontaneously shed tumor cells. We show that VEGF-C overexpression in cancer cells induces hyperplasia in peritumor lymphatic vessels and increases the volumetric flow rate in lymphatics at the base of the ear by 40%. The increases in lymph flow rate and peritumor lymphatic surface area enhance the rate of tumor cell delivery to lymph nodes, leading to a 200-fold increase in cancer cell accumulation in the lymph node and a 4-fold increase in lymph node metastasis. In our model, VEGF-C overexpression does not confer any survival or growth advantage on cancer cells. We also show that an anti-VEGF receptor (VEGFR)-3 antibody reduces both lymphatic hyperplasia and the delivery of tumor cells to the draining lymph node, leading to a reduction in lymph node metastasis. However, this treatment is unable to prevent the growth of tumor cells already seeded in lymph nodes. Collectively, our results indicate that VEGF-C facilitates lymphatic metastasis by increasing the delivery of cancer cells to lymph nodes and therapies directed against VEGF-C/VEGFR-3 signaling target the initial steps of lymphatic metastasis. (Cancer Res 2006; 66(16): 8065-75)
- intravital imaging
- lymphatic therapy
Multiple steps are required for tumor cells to metastasize from their primary site to regional lymph nodes. These steps include detachment from the primary tumor mass, invasion into lymphatic vessels, transport through draining lymphatic vessels, arrest in lymph nodes, and survival and growth in lymph nodes ( 1, 2). Although the steps leading to hematogenous metastasis have been extensively investigated ( 3– 10), there is a paucity of similar data for lymphatic metastasis. Several experimental ( 11– 15) and clinical (reviewed in refs. 16, 17) studies have shown a positive correlation among vascular endothelial growth factor (VEGF)-C expression, tumor margin lymphangiogenesis, and lymphatic metastasis. However, the individual steps in the lymphatic metastasis cascade that are controlled by VEGF-C are still unexplored. VEGF-C binds to VEGF receptor (VEGFR)-2, found on both blood and lymphatic vessels, and VEGFR-3, found on lymphatic vessels and some angiogenic tumor blood vessels ( 18). VEGFR-3 signaling is primarily responsible for the lymphangiogenic response to VEGF-C stimulation ( 19) and leads to lymphangiogenesis and lymphatic hyperplasia in mouse tumor models ( 11, 14, 15, 20, 21). VEGF-C can also induce angiogenesis ( 22, 23), which can be inhibited by blocking VEGFR-2 signaling ( 24). Determining the step(s) in the lymphatic metastasis process regulated by VEGF-C ( 25) is critical for developing optimal therapeutic strategies to control lymphatic metastasis ( 26– 29).
Here, we adapt and further develop a tumor model in the mouse ear ( 30) to allow intravital microscopy (IVM) of each of the steps leading to lymphatic metastasis. By imaging and quantifying these steps, we show that VEGF-C overexpression leads to hyperplasia of peritumor and draining lymphatics as well as to an increased delivery of cancer cells to the draining cervical lymph node. We then show that an anti-VEGFR-3 antibody can suppress this process. We also show that the anti-VEGFR-3 antibody has differential effects on the prevention versus treatment of lymphatic metastasis in our model. These findings should help identify appropriate clinical situations for the use of anti-VEGF-C/anti-VEGFR-3 therapy.
Materials and Methods
Tumor Cell Lines: Generation and Characterization
VEGF-C-overexpressing (VEGF-C) and mock-transduced (MT) B16F10 melanoma and T241 fibrosarcoma cell lines were established previously and cultured as reported ( 24). To create green fluorescent protein (GFP)–expressing cells, peak12 EF1α GFP vector (a gift from Dr. Brian Seed, Massachusetts General Hospital, Boston, MA) was transduced into T241-VEGF-C and T241-MT cells by lipofection. All cell lines were maintained in DMEM with 10% fetal bovine serum.
Stable expression of VEGF-C was verified by Northern blot analysis of total RNA extracted from cultured cells. A 392-bp mouse VEGF-C probe was generated by PCR using mouse embryo cDNA as a template (forward 5′-CAAGGCTTTTGAAGGCAAAG-3′ and reverse 5′-TGCTGAGGTAACCTGTGCTG-3′). Hybridization probes were labeled with Alkphos direct labeling and detection kit (Amersham Biosciences, Piscataway, NJ). Blots were hybridized overnight at 60°C, washed with high stringency, and exposed on a film. Blots were stripped and rehybridized with a β-actin probe as a loading control. Two bands (3.4 and 2.4 kb) corresponding to the full-length transcript and a splice product lacking ∼1 kb of viral non-long terminal regions were identified.
Western blot analysis was also carried out to confirm the secretion of VEGF-C protein. Conditioned media was obtained from subconfluent cells grown for 24 hours in serum-free medium and concentrated. Protein samples were electrophoresed in SDS-polyacrylamide gel and transferred onto a polyvinylidene difluoride membrane. The membrane was incubated with a polyclonal anti-VEGF-C antibody (dilution, 1:1,000; Santa Cruz Biotechnology, Santa Cruz, CA) followed by incubation with secondary antibody, development with CDP-star reagent, and exposure to a film. Analysis of conditioned medium confirmed secretion of the partially processed 31-kDa form of VEGF-C in the VEGF-C-transduced cell lines but not in MT cell lines.
Reverse transcription-PCR (RT-PCR) was carried out to show that VEGFR-2 and VEGFR-3 were not expressed by the VEGF-C-overexpressing or MT tumor cells. Total RNA was extracted from cultured cells using RNeasy kit (Qiagen, Valencia, CA) and reverse transcribed using Taqman kit (Applied Biosystems, Foster City, CA). PCR was conducted with primer pairs for VEGFR-2 and VEGFR-3 (VEGFR-2 forward 5′-GCTTTCGGTAGTGGGATGAA-3′ and reverse 5′-GGAATCCATAGGCGAGATCA-3′ and VEGFR-3 forward 5′-TTGGCATCAATAAAGGCAG-3′ and reverse 5′-CTGCGTGGTGTACACCTTA-3′). PCR products were electrophoresed in an agarose gel and receptor gene expression was analyzed. Mouse embryo mRNA was used as positive controls.
In vitro proliferation of T241-VEGF-C, T241-MT, T241-VEGF-C-GFP, and T241-MT-GFP cells was determined by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Cells (3 × 103) were plated on 96-well plate in triplicate. After 24, 48, or 72 hours of incubation, the MTT assay was done according to the manufacturer's recommendations. All assays were done in duplicate.
Experiments were done in nude (T241 fibrosarcoma model) and C57BL/6 (B16F10 melanoma model) mice and were approved by the Institutional Animal Care and Use Committee of Massachusetts General Hospital. Animals were anesthetized with ketamine/xylazine (10/100 mg/kg i.m.) for all experiments. Lymphangiography was done by slow injection in the interstitial tissue of the peripheral ear and angiography was done by i.v. injection in the tail vein. In preliminary experiments, lymphangiography with Evan's blue dye (Sigma Chemical Co., St. Louis, MO) and FITC-dextran (2.5%, MW = 2 million; Sigma) revealed a dense auricular network of lymphatic capillaries, draining to a larger vessel at the ear base and subsequently to the exposed superficial cervical lymph node ( Fig. 1A and B ). To establish tumors, we injected 50 μL tumor cell suspension (containing 5 × 106 cells) in the peripheral ear. Tumor cell suspension was created from source tumors grown in the flank of four to six mice. Excised tumors were minced with scissors, treated with trypsin (0.0125% trypsin in HBSS), and centrifuged. A dense cell pellet was collected and injected in the tip of the ear with a 30-gauge needle.
Lymphangiography Using IVM
Lymphatic diameters were measured using lymphangiography in mice with or without tumors in their ears. When tumor volumes reached 150 mm3, mice were anesthetized and placed on a small plate allowing immobilization of the ear. FITC-dextran (2 μL; Sigma) was injected in the surface of the tumors. Ear lymphatics were observed with epifluorescence IVM (E-IVM) and/or multiphoton laser scanning IVM (MP-IVM). Lymphangiography images were captured and lymphatic diameters were measured using ImageJ software (http://rsb.info.nih.gov/ij/; ref. 14). The maximum diameter of each lymphangion (a segment of lymphatics between two valves) was measured. Lymphatics within 700 μm from the edge of the tumor were defined as peritumor lymphatics and the ones farther from the tumor were defined as ear base lymphatics. The afferent lymphatic to the cervical lymph node was observed in the exposed lateral neck area ( Fig. 1B).
Lymph Flow Measurements Using Fluorescence Photobleaching
Lymph fluid velocity in peritumor lymphatics was measured with fluorescence photobleaching as described previously ( 31). Briefly, lymphangiography with FITC-dextran was performed at a constant pressure of 10 cm H2O and the measurements were initiated when sufficient fluorescent material appeared in the lymphatics. Lymph fluid velocity was calculated by tracking the convective movement of a photobleached spot. Flow volume was calculated from the velocity and the lymphatic diameter using the equation: Qlymphflow = πd2vlymph, where Qlymphflow is the flow volume of lymph, d is the lymphatic diameter, and vlymph is the velocity of lymph measured with fluorescence photobleaching.
Tumor Cell Delivery to the Lymph Node
Tumor cell arrival into the exposed cervical lymph node was observed with MP-IVM ( 32, 33) on days 3, 5, 7, 9 10, 14, and 20 after tumor implantation in nude mice. No tumor cells were observed in the lymph nodes within 1 hour after tumor implantation. Collagen I in the lymph node was imaged using second harmonic generation to visualize the lymph node capsule ( 34). T241-MT-GFP or T241-VEGF-C-GFP cells were implanted in the same manner as described above. On the day of imaging, mice were anesthetized and fixed on the microscope stage. Following TRITC-dextran (1%, MW = 2 million; Invitrogen, Carlsbad, CA) injection into the tip of the ear, the cervical lymph node was exposed and imaged with MP-IVM. Images of all the GFP-positive cells detectable in each lymph node were captured as image stacks with a 10 μm step. One to 5 fields per lymph node and 10 to 51 slices per field were acquired. The number of cells was counted using ImageJ software in a blinded fashion by three investigators.
Direct Tumor Cell Injection into Lymph Nodes
Anesthetized mice were immobilized under a stereoscope and the cervical lymph node was exposed. Either T241-MT-GFP or T241-VEGF-C-GFP cells (1 × 103, 5 × 103, or 1 × 104) in 0.5 μL PBS were directly injected into the lymph node using a 30-gauge needle connected to a 2.5 μL microsyringe (Hamilton, Reno, NV) and the surgical site was closed [1 × 103 and 5 × 103 cells, n = 8; 1 × 104 cells, n = 10 (T241-MT-GFP) and n = 12 (T241-VEGFC-GFP)]. After 28 days, lymph node tumor formation was examined macroscopically and microscopically using multiple frozen sections.
For the metastasis assay, formalin-fixed and paraffin-embedded cervical lymph nodes were stained with H&E. Multiple sections spaced 200 μm apart spanning the entire lymph node were examined. For blood and lymphatic vessel evaluation in primary tumors, ears were excised at day 14, fixed in 4% paraformaldehyde by immersion, embedded in paraffin, and immunostained with rabbit anti-mouse LYVE-1-antibody (1:2,000, Upstate, Charlottesville, VA), hamster anti-mouse podoplanin antibody (1:50,000, AngioBio, Del Mar, CA), or rat anti-mouse MECA-32 antibody (1:50, BD Biosciences, San Jose, CA). For terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) and Ki-67 staining, lymph nodes containing 100 to 500 GFP-expressing tumor cells (determined with MP-IVM as described above) were excised, fixed in 4% paraformaldehyde, and frozen in OCT compound. At least three sections were analyzed from each lymph node. Fluorescent TUNEL staining was carried out according to the manufacturer's instructions for the ApopTag Red In situ Apoptosis Detection kit (Chemicon, Temecula, CA). T241-GFP tumor cells (n = 1,045) and T241-VEGF-C-GFP tumor cells (n = 975) were counted in the TUNEL analysis. For Ki-67 staining, sections were exposed to a rat anti-mouse Ki-67 antibody (DAKO, Glostrup, Denmark) and stained with Alexa Fluor 546 (Invitrogen). T241-GFP tumor cells (n = 1,204) and T241-VEGF-C-GFP tumor cells (n = 2,301) were counted in the Ki-67 analysis. Images were captured by confocal microscopy (Fluoview 500, Olympus, Center Valley, PA). The number of GFP-positive/TUNEL-positive cells and GFP-positive/Ki-67-positive cells were counted by two investigators in a blinded manner using ImageJ software. LYVE-1 staining was done on lymph nodes containing 100 to 500 metastatic tumor cells prepared as described above. Sections were stained with rabbit anti-mouse LYVE-1 antibody (Upstate) and developed with diaminobenzidine. Lymphatics visualized in bright-field images were outlined and the ratio of stained lymphatic area per tissue area was analyzed with the use of a NIH Image macro (n = 6-9). VEGFR-3 staining was carried out on T241-VEGF-C-GFP ear tumors. Sections were stained with rat anti-mouse VEGFR-3 antibody (1:100, eBioscience, San Diego, CA) and developed with diaminobenzidine.
Anti-VEGFR-3 and Anti-VEGFR2 Antibody Treatment in the Ear Model
Prevention protocol. A neutralizing rat monoclonal antibody to murine VEGFR-3, mF4-31C1 (ImClone Systems, Inc., New York, NY; ref. 28), was given to mice bearing T241-VEGF-C-GFP ear tumors using different treatment schedules. To study the effects of VEGFR-3 blockade on tumor cell delivery and metastasis prevention, mF4-31C1 (40 mg/kg) was injected i.p. every 2 days for the first 14 days after tumor implantation with a preload of 80 mg/kg on day 0. Rat IgG (Jackson ImmunoResearch, West Grove, PA) was injected in control mice at the same dose and schedule. Lymphatic diameter was measured with E-IVM lymphangiography on day 12 and cell arrival into lymph nodes was quantified with MP-IVM on day 14 as described above. To study the ability of mF4-31C1 treatment to prevent lymph node metastasis, tumors were removed after 14 days of treatment and the treatment was discontinued. Forty-two days after original tumor implantation, macroscopic and microscopic metastases were assessed.
Intervention protocol. To study the ability of mF4-31C1 treatment to control metastasis formation after tumor cell seeding, T241-VEGF-C-GFP tumors were left untreated for 14 days after implantation. On day 14, the tumors were resected and mF4-31C1 (40 mg/kg) was injected i.p. every 2 days for the next 28 days with a preload of 80 mg/kg for the first dose. Rat IgG was injected in control mice at the same dose and schedule. Forty-two days after original tumor implantation, macroscopic and microscopic metastases were assessed.
A neutralizing rat monoclonal antibody to murine VEGFR-2, DC101 (ImClone Systems), was also given using the intervention protocol. After the tumors were resected on day 14 after implantation, DC101 (40 mg/kg) was injected i.p. every 3 days for 28 days. Rat IgG was injected in control mice at the same dose and schedule. Forty-two days after original tumor implantation, macroscopic and microscopic metastases were assessed.
Anti-VEGFR-3 Antibody Treatment in the Dorsal Skin Window Model
The effect of VEGFR-3 blockade on the function of peritumor lymphatics was examined using the dorsal skin fold chamber (DSC) model as described previously ( 21). Briefly, C57/BL6 mice were anesthetized, dorsal skin was shaved and depilated, and two mirror-image titanium frames were mounted to fix the extended double layer of dorsal skin between the frames. One 15-mm-diameter layer of skin was excised, leaving the striated muscle, s.c. tissue, and epidermis of the opposite side intact. The tissue was covered with a glass coverslip mounted into the frame. After the mice recovered from surgery, a small piece (1 mm in diameter) of B16F10-VEGF-C tumor (T241-VEGFC tumors do not display reproducible abnormal lymphatic patterns in the DSC) was implanted into the center of the DSC. mF4-31C1 or rat IgG (40 mg/kg) was injected i.p. on days 3 and 5 and lymphangiography was done on day 6. Evan's blue dye (1-2 μL) was injected i.d. in the back side of the tumor using a 30-gauge needle. Within 10 minutes after the injection, mice were fixed under a stereoscope and images were acquired by a digital camera. The ratio of the length of the lymphatics that fill in the caudal direction (retrograde lymphatics) over the length of total lymphatics was measured (n = 6).
Quantitative data are mean ± SE. Student's t test and Fisher's exact test were used for statistical analysis. For the MTT proliferation assay results, regression analysis was used to compare slopes of graphs. P ≤ 0.05 was considered statistically significant.
Visualization of the steps of lymphatic metastasis. Using Evan's blue or fluorescence lymphangiography with E-IVM and MP-IVM ( 32, 33), we visualized the auricular lymphatic network draining the tip of a normal mouse ear into larger vessels at the base of the ear [as described previously by Nagy et al. ( 30)] and subsequently into the afferent lymphatic vessel of the superficial cervical lymph node (called afferent lymphatic throughout text; Fig. 1A-C). When tumors were implanted in the tip of the ear, each of these sets of vessels could be monitored during the metastatic process ( Fig. 1D-F). We also imaged the blood vasculature of both normal and tumor bearing ears by simultaneous angiography and lymphangiography ( Fig. 1G and H).
We next used these techniques to determine the step(s) in the metastatic process where VEGF-C increases the formation of lymph node metastasis. We created VEGF-C-overexpressing cell lines from murine T241 fibrosarcoma and murine B16F10 melanoma ( Fig. 2A and B ; ref. 14). When implanted in the ear, VEGF-C-overexpressing tumors exhibited an increased incidence of lymphatic metastasis in the cervical lymph node ( Fig. 2C-E) without affecting the growth of the primary tumor (data not shown). The increase in lymph node metastasis resulting from the overexpression of VEGF-C in our tumor models is similar to data from a variety of human cancers ( 16, 17).
VEGF-C overexpression induces hyperplasia in peritumor and draining lymphatics but not in the afferent lymphatics. Using E-IVM and MP-IVM, we showed that peritumor lymphatics had an increased vessel diameter in VEGF-C-overexpressing T241 tumors (T241-VEGF-C) when compared with that in MT controls (T241-MT) ( Fig. 3A and B ). These data are consistent with our previous report using two different models ( 14). Additionally, we found that the lymphatic hyperplasia extended down to the base of the ear but was not observed in the afferent lymphatic. In accordance with this, there was no increase in density of LYVE-1 staining in the draining lymph nodes of VEGF-C-overexpressing tumors compared with control tumors ( Fig. 3C), suggesting that VEGF-C overexpression by the primary tumor did not induce lymphangiogenesis in the draining lymph node in this model ( 35). Occasionally, malformed intraluminal valves were observed in the peritumor lymphatics ( Fig. 3D and E), which may contribute to abnormal flow patterns that were observed previously ( 21). LYVE-1 staining also showed enlarged lymphatic vessels in the margin of T241-VEGF-C tumors when compared with T241-MT tumors ( Fig. 3F-H). Podoplanin staining also showed enlarged lymphatic vessels in the margin of T241-VEGF-C-GFP tumors when compared with T241-MT-GFP tumors ( Fig. 3H). Thus, VEGF-C-induced enlargement of peritumor lymphatic vessels likely provides the opportunity for increased entry of tumor cells into the lymphatic system.
VEGF-C overexpression increases lymph flow. Measurements of lymph flow through large lymphatic vessels connecting the drainage from primary tumors to sentinel lymph nodes have shown a rapid rate of transport in both mice and humans ( 36). However, the intralymphatic flow rates in the small lymphatic vessels associated with a tumor have not yet been measured. Intralymphatic flow in both initial and collecting lymphatic vessels carries cancer cells to the lymph nodes and may be modulated by VEGF-C. To address this possibility, we measured for the first time the lymph fluid velocity in the tumor margin, ear base, and afferent lymphatic vessels with a fluorescence photobleaching technique ( 31). Although the actual velocity of lymph movement was somewhat slower in the peritumor lymphatics and in the lymphatics at the base of the ear in T241-VEGF-C tumors (data not shown), total volumetric flow rate in the lymphatics in the base of the ear was increased by 40% due to lymphatic vessel enlargement ( Fig. 3I). Thus, the total lymph flow rate is significantly increased by VEGF-C overexpression.
VEGF-C overexpression increases tumor cell delivery to lymph nodes. To image the transport of tumor cells through lymphatic vessels and into lymph nodes, we stably expressed GFP in T241-MT cells (T241-MT-GFP) and T241-VEGF-C cells (T241-VEGF-C-GFP). Using MP-IVM, we imaged, for the first time, GFP-expressing cells shed from primary tumors as they traveled in the peritumor and afferent lymphatics ( Fig. 4A ). These cells subsequently localized in the subcapsular sinus and the cortex of the lymph node near the afferent lymphatic ( Fig. 4B-F). The collagen fibers of the lymph node capsule were imaged using second harmonic generation microscopy ( 34), allowing us to visualize the lymph node capsule and thus localize the tumor cells within the node ( Fig. 4D).
Cells in lymphatic vessels are carried to their destination by lymph fluid, making their transport dependent on lymph flow. Because VEGF-C overexpression increases peritumor lymphatic diameters as well as lymph flow rate, we hypothesized that VEGF-C overexpression would increase the delivery of cancer cells. To test this hypothesis, we quantified the number of cells delivered to the lymph node using MP-IVM. Indeed, the number of tumor cells delivered to the lymph node increased with time after tumor implantation ( Fig. 4G and H). We found a 200-fold increase in GFP-positive tumor cells in the lymph node from tumors overexpressing VEGF-C when compared with mock-transfected tumors 10 days after tumor implantation ( Fig. 4H). Furthermore, 20 days after implantation, 6 of 7 lymph nodes from mice bearing T241-VEGF-C-GFP tumors had large metastatic nodules, whereas 8 of 9 lymph nodes in mice bearing T241-MT-GFP tumors had only a few tumor cells ( Fig. 4E and F). H&E staining was used to confirm the formation of metastatic nodules on day 20 (data not shown).
VEGF-C overexpression does not increase tumor cell survival in lymph nodes. An alternative explanation for the increased number of T241-VEGF-C-GFP cells in the cervical lymph nodes would be that VEGF-C promotes proliferation and survival of tumor cells within the lymph nodes. Using RT-PCR, we found that VEGFR-2 and VEGFR-3 are not expressed in the VEGF-C-overexpressing and mock-transfected (control) tumor cells and in vitro proliferation is not different between the cell lines, thus eliminating the possibility of autocrine signaling ( Fig. 5A-C ). We showed previously that these VEGF-C-overexpressing cells do not exhibit increased cell migration ( 14). In addition, VEGFR-2 and VEGFR-3 were not detected on tumor cells in vivo by immunohistochemistry ( Fig. 5D). VEGF-C overexpression did not change the rate of apoptosis or proliferation in intranodal tumor cells when compared with MT cells ( Fig. 5E and F). Furthermore, by injecting equal numbers of either T241-MT-GFP or T241-VEGF-C-GFP cells directly into the lymph node, we showed no differences in the incidence or size of tumor masses in the lymph nodes ( Fig. 5G). Thus, VEGF-C overexpression did not enhance the growth of tumor cells in lymph nodes.
To further investigate any survival benefit of VEGF-C overexpression, we analyzed the formation of lymph node metastasis after equal numbers of spontaneously shed T241-MT-GFP and T241-VEGF-C-GFP tumor cells accumulated in the lymph node. To this end, we implanted tumor cells in the ear on day 0 and removed the primary tumors on day 14 for T241-MT-GFP cells (number of tumor cells in the lymph node, 27 ± 7; Fig. 4H) and on day 7 for T241-VEGF-C-GFP cells (number of tumor cells in the lymph node, 44 ± 23; Fig. 4G). Twenty-eight days after primary tumor removal, cervical lymph nodes were examined by both MP-IVM and histology. No lymph node metastasis was formed in either T241-MT-GFP or T241-VEGF-C-GFP tumors (0 of 13 and 0 of 15, respectively). However, when T241-VEGF-C-GFP ear tumors were resected on day 14 (cell count; 536 ± 144), 5 of 11 mice developed gross lymph node metastases. Thus, VEGF-C-overexpressing tumor cells did not have a survival or growth advantage when compared with MT tumor cells. Taken together, our data show that the delivery of VEGF-C-overexpressing tumor cells to the lymph node is increased >200-fold, without a migratory, proliferative, or survival advantage.
VEGFR-3 blockade reduces peritumor lymphatic hyperplasia and tumor cell delivery to lymph node. We next used our model to test how molecular interventions would alter the individual steps in lymphatic metastasis. To this end, we administered a neutralizing rat monoclonal antibody to murine VEGFR-3, mF4-31C1 ( 28), to mice starting on the day of implantation of T241-VEGF-C-GFP tumor cells into the ear. Fluorescence lymphangiography showed that VEGF-C-induced lymphatic hyperplasia was significantly suppressed by mF4-31C1 treatment ( Fig. 6A-C ). On the other hand, this antibody did not affect intratumor blood vessel density (control, 58 ± 9 vessels/mm2; mF4-31C1, 51 ± 7 vessels/mm2; P > 0.05). Furthermore, a significant decrease in the number of tumor cells delivered to the cervical lymph node of mF4-31C1-treated mice was measured on day 14 ( Fig. 6D-F). These data indicate that VEGFR-3 blockade can abrogate VEGF-C-induced lymphatic hyperplasia and tumor cell delivery in the lymph node.
Anti-VEGFR-3 treatment does not restore normal lymph flow patterns. We also tested the ability of VEGFR-3 blockade to reverse abnormal lymph flow patterns. In the normal skin of the mouse DSC, lymph flow is unidirectional ( Fig. 7A and D ; ref. 21). In agreement with our previous study, peritumor lymphatics of VEGF-C-overexpressing B16F10 melanomas exhibited abnormal multidirectional flow pattern presumably due to malformation of intraluminal valves ( Fig. 7B and E). Due to rapid growth of B16F10 tumors and size limitation of the DSC, we began anti-VEGFR-3 antibody therapy 3 days after tumor implantation and observed the effect 3 days after the start of treatment. Two doses of anti-VEGFR-3 antibody were not sufficient to restore the normal peritumor lymph flow pattern in this model ( Fig. 7C and F).
VEGFR-3 blockade has differential effects on prevention versus treatment. We then examined the ability of VEGFR-3 blockade to interfere with lymphatic metastasis in different clinical scenarios. To this end, we compared the development of lymph node metastasis in two treatment protocols (prevention and intervention) using mF4-31C1. In the prevention protocol, mF4-31C1 was given from the day of T241-VEGF-C-GFP tumor implantation until tumor resection on day 14. Consistent with the cell delivery results, fewer lymph node metastases were identified 28 days after tumor resection in the group treated with mF4-31C1 ( Fig. 6G). In the intervention protocol, T241-VEGF-C-GFP tumors were left untreated for 14 days and then resected. Treatment with mF4-31C1 was started after resection and continued for 28 days. No statistical difference in lymph node metastasis was found between control and mF4-31C1-treated animals in the intervention protocol ( Fig. 6H). Although VEGFR-3 blockade successfully blocked lymphatic hyperplasia and limited the delivery of tumor cells into the lymph node, it was unable to prevent the growth of tumor cells already seeded in the lymph node.
VEGFR-2 blockade reduces tumor growth in both primary and secondary sites. To confirm that our findings were related to VEGF-C/VEGFR-3 signaling, we examined the effect of the anti-VEGFR-2 antibody DC101 in the same models. In agreement with previously published studies ( 37), DC101 greatly inhibited primary tumor growth when treatment was started at the time of tumor implantation (data not shown). In the intervention protocol, unlike mF4-31C1, the DC101 treatment group developed fewer macroscopic lymph node metastases than control-treated animals ( Fig. 6I). In contrast to VEGFR-3 blockade, VEGFR-2 blockade could inhibit the growth of seeded tumor cells. These findings support the use of agents that block both VEGFR-2 and VEGFR-3 signaling—a subject of our ongoing research.
Using a novel tumor model, we showed that lymph node metastasis formation requires a large number of viable tumor cells to be delivered to the lymph node. Using intravital imaging, we also showed that VEGF-C increases the number of cells trafficking from the tumor to the lymph node without affecting their survival. Fibrosarcomas, including T241 used in this study, only rarely metastasize to lymph nodes. By overexpressing the lymphatic growth factor VEGF-C in T241 cells, this fibrosarcoma gained the ability to metastasize to lymph nodes with much greater frequency without changing the proliferation, migration, or tumor growth characteristics of these cells ( Fig. 5; ref. 14). Thus, the increase in lymph node metastasis is most likely a result of changes in tumor-associated lymphatic vessels in response to VEGF-C overexpression, allowing us to specifically dissect the contribution of lymphatic vessel biology to lymphatic metastasis.
Because the growth and survival of VEGF-C-overexpressing cells in the lymph node is the same as that for the control cells, the increased incidence of metastases from T241-VEGF-C-GFP tumors can only be explained by the increased number of cells arriving in the lymph node. This may explain the correlation between VEGF-C expression and the incidence of lymph node metastasis in the patients ( 16, 17). These observations highlight the stochastic nature of the metastatic process. The greater number of cells delivered to the lymph node increases the probability that a cell capable of forming a metastatic nodule will be present and form a metastasis ( Fig. 8 ).
The exact mechanism by which VEGF-C increases tumor cell entry into lymphatic vessels is not known. One hypothesis is that VEGF-C increases the surface area of functional lymphatics in the tumor margin, thus providing more opportunity for a tumor cell to enter the lymphatics and disseminate ( 14). The data presented here support this hypothesis. An alternate hypothesis is that VEGF-C stimulates tumor-associated lymphatics or the draining lymph nodes to release chemotactic factors that recruit tumor cells to enter lymphatics ( 38– 41). Further direct studies on the migration and entry of cancer cells into lymphatic vessels are needed to unambiguously reveal whether the lymphatics act as passive recipients of invasive tumor cells or whether they play an active role in tumor cell invasion.
Interfering with the VEGF-C/VEGFR-3 signaling pathway has been suggested as a useful clinical strategy in the treatment of lymphatic metastasis ( 13, 27, 28, 41, 42). The expression of a soluble VEGFR-3 receptor, which competitively inhibits the binding of VEGF-C and VEGF-D to both VEGFR-2 and VEGFR-3, was able to prevent lymph node metastasis when the soluble receptor is available from the time of tumor implantation ( 27, 41, 43). Similarly, the knockdown of VEGF-C expression using siRNA also led to a reduction in lymph node metastasis ( 44). VEGF-C also stimulates angiogenesis through VEGFR-2 stimulation ( 22– 24). To isolate the contribution of VEGFR-3 in promoting lymphatic metastasis, we directly blocked VEGFR-3 signaling by using the neutralizing rat monoclonal antibody mF4-31C1 and showed that lymphatic hyperplasia, and consequently the number of tumor cells entering into lymphatic vessels, could be reduced. The reduction of the number of tumor cells entering into lymphatic vessels significantly reduced the number of lymph node metastasis in the absence of any direct effect of the treatment on the cancer cells. These data offer novel mechanistic insight into the causal relationship between VEGF-C and lymph node metastasis and indicate that VEGFR-3 blockade may be a valid strategy to prevent lymph node metastasis.
On the other hand, VEGFR-3 blockade was not effective in blocking metastasis formation in lymph nodes already seeded with cancer cells that do not express VEGFR-3. Although VEGFR-3 is present on some tumor blood vessels and can play a role in tumor angiogenesis ( 45, 46), VEGFR-3 blockade did not decrease growth of the primary tumor, blood vessel density, or metastatic growth in our model. On the other hand, VEGFR-2 blockade using DC101 did reduce the growth of primary tumors in the ear as well as the formation of macroscopic lymph node metastasis in lymph nodes seeded with tumor cells. It is likely that both of these effects are due to the antiangiogenic effects of DC101. Thus, a potentially promising strategy to control both hematogenous and lymphatic metastasis would be to block both VEGFR-2 and VEGFR-3 signaling using multitargeted tyrosine kinase inhibitors already approved or in clinical trials ( 47).
Dissecting the steps in lymphatic metastasis has shown that VEGF-C/VEGFR-3 signaling influences tumor cell entry into lymphatic vessels. This information may guide the selection of clinical situations in which anti-VEGF-C/VEGFR-3 therapy will benefit appropriate patients. Patients with disease confined to the primary site who can be successfully treated with local therapy (e.g., surgery) are less likely to benefit from anti-VEGF-C/VEGFR-3 therapy. Similarly, patients with successful treatment of their primary tumor, but tumor cells remaining in their lymph nodes, may derive little benefit from anti-VEGF-C/VEGFR-3 therapy. In contrast, in settings where the prevention of tumor cell seeding into lymphatic vessels is indicated, anti-VEGF-C/VEGFR-3 therapy may be a promising therapeutic strategy. Examples of such settings include patients with inoperable tumors, patients with residual cancer cells after surgical resection, or those at risk of local failure after initial treatment ( 17). In the specific circumstances in which the VEGFR-3 receptor is present on tumor cells or the tumor angiogenesis is dependent on VEGFR-3 signaling, VEGFR-3 blockade may have additional benefit in treating metastatic disease. The efficacy of anti-VEGF-C/VEGFR-3 therapy in these settings now needs to be validated in clinical trials.
Grant support: National Cancer Institute grants R01CA85140 (R.K. Jain) and R01CA96915 (D. Fukumura).
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.
We thank Sergey Kozin, Lance Munn, and Gregory Nelson for scientific and technical help and Sylvie Roberge, Julia Kahn, Jessica Tooredman, Nyall London, and Carolyn Smith for outstanding technical support.
Note: T. Hoshida and N. Isaka contributed equally to this work.
Current affiliation for N. Isaka: Department of Surgery, Seirei-Sakura Citizen Hospital, Chiba, Japan. Current affiliation for J. Hagendoorn: Department of Surgery, University Medical Center, Utrecht, the Netherlands.
- Received April 14, 2006.
- Revision received May 30, 2006.
- Accepted June 15, 2006.
- ©2006 American Association for Cancer Research.