Inhibition of VEGF signaling can promote lymph node metastasis in preclinical models, but the mechanism is not fully understood, and successful methods of prevention have not been found. Signaling of hepatocyte growth factor (HGF) and its receptor c-Met can promote the growth of lymphatics and metastasis of some tumors. We sought to explore the contributions of c-Met signaling to lymph node metastasis after inhibition of VEGF signaling. In particular, we examined whether c-Met is upregulated in lymphatics in or near pancreatic neuroendocrine tumors in RIP-Tag2 transgenic mice and whether lymph node metastasis can be reduced by concurrent inhibition of VEGF and c-Met signaling. Inhibition of VEGF signaling by anti-VEGF antibody or sunitinib in mice from the age of 14 to 17 weeks was accompanied by more intratumoral lymphatics, more tumor cells inside lymphatics, and more lymph node metastases. Under these conditions, lymphatic endothelial cells, like tumor cells, had strong immunoreactivity for c-Met and phospho-c-Met. c-Met blockade by the selective inhibitor, PF-04217903, significantly reduced metastasis to local lymph nodes. Together, these results indicate that inhibition of VEGF signaling in RIP-Tag2 mice upregulates c-Met expression in lymphatic endothelial cells, increases the number of intratumoral lymphatics and number of tumor cells within lymphatics, and promotes metastasis to local lymph nodes. Prevention of lymph node metastasis by PF-04217903 in this setting implicates c-Met signaling in tumor cell spread to lymph nodes. Cancer Res; 73(12); 3692–703. ©2013 AACR.
Metastasis to regional lymph nodes is a feature of many solid tumors. The presence of lymph node metastasis is an important prognostic factor and the basis for surgical excision and radiation of local lymph nodes (1). Lymph node metastasis occurs after tumor cells enter lymphatics within or near tumors and drain to sentinel nodes (1). Tumor-associated lymphangiogenesis promotes the process (2–4) by increasing the number of routes to lymph nodes.
Reports of recent preclinical studies indicate that tumor invasiveness and metastasis can increase after inhibition of VEGF signaling (5–8). The mechanism of the increased aggressiveness is unknown, but contributing factors are likely to include increased intratumoral hypoxia as a result of vessel pruning. Hypoxia can increase expression of c-Met, the hepatocyte growth factor receptor (HGFR), the receptor tyrosine kinase (RTK) activated by HGF (9). Activation of c-Met can drive tumor cell motility, proliferation, invasion, and survival (10–12). The HGF/c-Met pathway is activated in a wide variety of solid tumors (12, 13), correlates with poor prognosis (14–16), and is thought to contribute to tumor aggressiveness and resistance (17). c-Met expression in tumors can increase after treatment with inhibitors of VEGF signaling that promote vascular pruning and intratumoral hypoxia (8–10). c-Met activation can drive lymphangiogenesis (18, 19), which could favor lymph node metastasis.
Metastases are more abundant in the liver of RIP-Tag2 transgenic mice after treatment with function-blocking anti-VEGFR2 antibody, sunitinib, or neutralizing anti-VEGF antibody (7, 8). The same has been found in lymph nodes of these mice after treatment with anti-VEGFR2 antibody (7), although the effects of age and duration of treatment have not been examined in detail.
The present study examined the involvement of c-Met signaling in lymph node metastasis after inhibition of VEGF signaling. Specifically, we sought to learn whether the treatment increases c-Met expression and activation in the lymphatic vessels and augments lymph node metastasis, and whether inhibition of c-Met signaling can reduce tumor spread to lymph nodes. We addressed these issues by determining the effects of VEGF signaling blockade on c-Met expression in lymphatic vessels and on number of intratumoral lymphatics, tumor cells inside lymphatics, and amount of lymph node metastasis. We then determined whether inhibition of c-Met signaling reduced tumor cells inside lymphatics and lymph node metastasis.
The approach was to manipulate c-Met signaling in RIP-Tag2 mice, which are known to develop lymph node metastasis after inhibition of VEGF signaling (7). VEGF signaling was blocked by treatment with a neutralizing anti-VEGF antibody or with sunitinib, a multitargeted RTK inhibitor of VEGF receptor (VEGFR), platelet-derived growth factor, c-KIT, and related kinases (20). c-Met signaling was blocked by the selective inhibitor, PF-04217903, to determine the effects on lymph node metastasis (21).
The experiments revealed that inhibition of VEGF signaling increased c-Met expression in lymphatics and tumor cells and also increased the number of intratumoral lymphatics, tumor cells inside lymphatics, and metastases in local lymph nodes. Inhibition of c-Met signaling blocked the exaggerated lymph node metastasis accompanying inhibition of VEGF signaling.
Materials and Methods
Animals and treatment
Lymph node metastasis was studied in RIP-Tag2 transgenic mice (C57BL/6 background), which develop spontaneous multifocal, multistage pancreatic neuroendocrine tumors driven by expression of SV40 T antigen in pancreatic βcells (22). Six groups of mice (5–7 mice/group each experiment) were treated from the age of 14 to 17 weeks with: (i) vehicle (0.5% sodium carboxymethyl cellulose (ICN Biomedicals, Inc), 1.8% (w/v) NaCl, 0.4% (v/v) Tween-80 (Sigma Chemical), 0.9% benzyl alcohol (v/v; Sigma-Aldrich), 5 μL/g given daily by gavage, (ii) affinity purified, function-blocking goat anti-mouse VEGF antibody (AF-493-NA; R&D Systems; 150 μg in 50 μL sterile PBS) injected intraperitoneally 3 times per week (Monday, Wednesday, Friday), (iii) sunitinib (Pfizer; 40 mg/kg in 5 μL vehicle) given daily by gavage, (iv) PF-04217903 (Pfizer; 30 mg/kg in 5 μL vehicle) given daily by gavage, (v) anti-VEGF antibody plus PF-04217903, or (vi) sunitinib plus PF-04217903. In group (vi), the agents were premixed and administered together.
Body weight and survival were recorded during the treatment. Some RIP-Tag2 mice received pimonidazole hydrochloride (60 mg/kg i.v; Hypoxyprobe Plus Kit HP2; Chemicon) 1 hour before perfusion to detect hypoxic regions in tumors. Untreated 14-week-old RIP-Tag2 mice (n = 5) were used as baseline control. Fourteen-week-old wild-type C57BL/6 mice (n = 5) were also examined.
In separate experiments, mice were treated from the age of 14 to 15 weeks with vehicle, anti-mouse VEGF antibody, or sunitinib, and tumors were removed for quantitative real-time PCR (qRT-PCR) analysis.
Mice were housed under barrier conditions in the animal care facility at the University of California, San Francisco (UCSF; San Francisco, CA). The UCSF Institutional Animal Care and Use Committee approved all experimental procedures.
Tissue preparation, immunohistochemistry, and imaging
At the end of treatment, mice were anesthetized with ketamine (100 mg/kg i.p.) plus xylazine (10 mg/kg i.p.), and tissues were fixed by vascular perfusion of 1% paraformaldehyde in PBS (23). Some tissues were prepared for hematoxylin and eosin (H&E) staining. Cryostat sections (80 μm thick) were processed for immunohistochemistry using 2 or 3 primary antibodies (23): guinea pig anti-swine insulin antibody (1:100; Dako), rabbit anti-SV40 T-antigen antibody (1:500; Santa Cruz Biotechnology), rat anti-mouse lymphatic vessel endothelial hyaluronic acid receptor (LYVE)–1 antibody (1:500; R&D Systems), goat anti-mouse HGFR antibody (1:500; R&D Systems), rabbit anti-mouse phosphorylated c-Met antibody (1:250; Invitrogen), goat anti-VEGFR-3 antibody (1:1,000; R&D Systems), rabbit anti-Prox1 antibody (1:500; AngioBio), goat anti-mouse HGF antibody (1:500; R&D Systems), rabbit anti-VEGFR-2 antibody T014 (1:2,000; from Rolf Brekken, University of Texas Southwestern Medical Center, Dallas, TX), rabbit anti-phosphohistone H3 antibody (PHH3 1:1,000; Millipore), or fluorescein isothiocyanate (FITC)-conjugated mouse anti-pimonidazole hydrochloride (1:100; Chemicon).
Secondary antibodies were FITC-, Cy3- or Cy5-labeled donkey anti-goat, anti-rabbit, anti-rat, or anti-guinea pig immunoglobulin G antibody (Jackson ImmunoResearch; all diluted 1:500). Cell nuclei were stained with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (Vector laboratories). Specimens were examined with a Zeiss Axiophot fluorescence microscope and a Zeiss LSM 510 laser scanning confocal microscope (23).
The methods used for measuring tumor size and mRNA expression of Prox1, VEGF-A, VEGF-C, and VEGF-D are presented in the Supplementary Materials and Methods.
Measurements of lymphatics and lymph node metastases
Pancreas and pancreatic lymph nodes from each mouse (5–7 mice/group) were removed and processed together. Sections of pancreas (80 μm thick) from 5 to 7 mice per group were examined. Intratumoral lymphatics, identified as LYVE-1-immunoreactive vessels completely surrounded by insulin-immunoreactive tumor cells, were counted in all tumors present in two 80 μm thick sections of pancreas. As tumors were larger in the vehicle and PF-04217903 groups, an average of 5 tumors were analyzed in those groups, whereas 10 to 15 smaller tumors were analyzed in the other treatment groups. Digital fluorescence microscopic images (×5 objective, ×1 Optovar) of tumors were combined as montages to show the entire tumor. The sectional area of tumors was measured, and the number of intratumoral lymphatics was expressed as lymphatics per mm2 of tumor section.
To distinguish growing lymphatics in tumors from lymphatics co-opted by invading tumor cells, we counted intratumoral lymphatics with evidence of endothelial cell proliferation, defined as one or more phosphohistone H3-positive endothelial cells. Tumors in 5 mice per group were analyzed.
The frequency of lymphatics containing tumor cells was expressed as the proportion of LYVE-1–positive lymphatics with one or more SV40 T-antigen–positive tumor cells within the lumen per section of pancreas. A total of 25 to 30 intratumoral lymphatics were examined in each mouse in all groups.
Lymph node metastases were identified by the presence of SV40 T-antigen–positive tumor cells in pancreatic lymph nodes. The incidence of metastasis was determined in 5 to 7 mice per group and was expressed as the proportion of lymph nodes that contained SV40 T-antigen–positive cells in a survey of at least four 80 μm thick sections cut at multiple levels (average 240 μm increment). Each section contained 3 to 5 lymph nodes.
Assessment of HGF, c-Met, and phospho-c-Met immunoreactivities
The abundance of HGF immunoreactive cells in tumors or lymph nodes was estimated as the area density of HGF staining in digital fluorescence microscopic images (10× objective, 1× Optovar) of RIP-Tag2 tumors or lymph nodes. Tumor cells were marked in the same sections by staining for SV40 T antigen or insulin. Lymph nodes were identified by the distinctive LYVE-1 staining. The area density was measured at a predetermined threshold with ImageJ (23).
The proportion of c-Met or phospho-c-Met–positive lymphatic vessels was expressed as the proportion of LYVE-1–positive intratumoral lymphatics that had c-Met or phospho-c-Met immunoreactivity in 80 μm sections of pancreas. The abundance of tumor cells with c-Met or phospho-c-Met immunoreactivity was expressed as the proportion of SV40 T-antigen or insulin immunoreactive cells that colocalized with c-Met or phospho-c-Met.
Differences among groups were assessed by ANOVA followed by Fisher least significant difference test for multiple comparisons. Error bars show mean ± SEM. Fisher exact test was used to compare the incidence of lymph node metastasis. P < 0.05 was considered significant.
Increased lymph node metastasis after inhibition of VEGF signaling
To build on the published report of exaggeration of lymph node metastasis by inhibition of VEGF signaling in RIP-Tag2 mice (7), we sought to obtain confirmatory data after treatment with anti-VEGF antibody or sunitinib before attempting to block it. As an indicator of lymph node metastasis, we used the presence of SV40 T-antigen–immunoreactive tumor cells in pancreatic lymph nodes. These lymph nodes include gastric, pancreatic-duodenal, and colic lymph nodes (24). The lymph nodes were removed, processed, and examined together with the pancreas.
At least one of the pancreatic lymph nodes contained SV40 T-antigen–positive cells in all RIP-Tag2 mice examined at 17 weeks of age, regardless of the treatment (Fig. 1A), but the proportion of positive nodes varied with the treatment. Only 50% of the pancreatic lymph nodes had metastases after vehicle for 3 weeks, but 78% had metastases after anti-VEGF antibody, and 81% had metastases after sunitinib (Fig. 1B). Most metastatic tumor cells were within the LYVE-1–positive cortical sinus of the lymph node (Fig. 1A).
Increased intratumoral lymphatics after inhibition of VEGF signaling
To explore these findings further, we asked whether the higher incidence of lymph node metastasis after inhibition of VEGF signaling was accompanied by more intratumoral lymphatics, more invasion of tumor cells into peritumoral lymphatics, or both (1, 3, 25, 26). We found that intratumoral lymphatics detected by LYVE-1 immunoreactivity were more numerous after inhibition of VEGF signaling from the age of 14 to 17 weeks (Fig. 2A).
We assessed the effectiveness of LYVE-1 as a marker of lymphatics in RIP-Tag2 tumors by comparing the staining for VEGFR-3, podoplanin, and Prox1 (27). VEGFR-3 immunoreactivity in RIP-Tag2 tumors was found in lymphatics and blood vessels (Fig. 2B, i; ref. 23). Podoplanin immunoreactivity was detected in lymphatics and epithelial cells of pancreatic ducts (data not shown), but most Prox1 immunoreactivity was in lymphatic endothelial cell nuclei (Fig. 2B, ii). In comparison, VEGFR-2 immunoreactivity was in blood vessels and lymphatics of tumors and lymph nodes (Supplementary Fig. S1A).
Intratumoral lymphatics were more abundant in tumors of mice treated with vehicle from the age of 14 to 17 weeks than in untreated RIP-Tag2 mice at the age of 14 weeks (onset control; Fig. 2C, i). After anti-VEGF antibody or sunitinib, intratumoral lymphatics were 93% or 79% more numerous than after vehicle (Fig. 2C, i). Consistent with the greater abundance of intratumoral lymphatics, expression of Prox1 mRNA was 2.5 times greater in tumors after anti-VEGF antibody and 3 times greater after sunitinib, from the age of 14 to 15 weeks (Fig. 2C, ii).
Expression of the lymphangiogenic factors, VEGF-C and VEGF-D, in untreated RIP-Tag2 tumors was much lower than VEGF-A (Supplementary Fig. S1B, i), which is known to be highly expressed in RIP-Tag2 tumor cells (28). However, expression of both VEGF-C and VEGF-D was increased after treatment with sunitinib from the age of 14 to 15 weeks (Supplementary Fig. S1B, ii and S1B, iii).
The contribution of lymphangiogenesis to the increase in intratumoral lymphatics after anti-VEGF therapy was assessed by determining the frequency of proliferating lymphatic endothelial cells detected by phosphohistone H3 immunoreactivity. In vehicle-treated mice, an average of 5 lymphatics had one or more phosphohistone H3-positive endothelial cells per section of pancreas (5–6 tumors analyzed in each of 5 mice; Fig. 2D, i). In comparison, after sunitinib, none of the tumors had phosphohistone H3-positive cells that colocalized with LYVE-1 staining (10–13 tumors analyzed in each of the 5 mice; Fig. 2D, ii).
Because we found no evidence for lymphatic proliferation after sunitinib, we asked whether inhibition of VEGF signaling was accompanied by tumor cell invasion into peritumoral lymphatics. Tumors of mice treated with sunitinib had the appearance of being more invasive, in that their border was more irregular and penetrated the acinar pancreas (Fig. 3A, i and ii). After sunitinib, tumors were also accompanied by more LYVE-1–positive lymphatics (Fig. 3B, i and ii and C, i and ii), and more lymphatics had intraluminal tumor cells (Fig. 3D, i and ii). The proportion of lymphatics with SV40 T-antigen–positive tumor cells inside was 58% higher after sunitinib than after vehicle (Fig. 3D, iii).
These differences in RIP-Tag2 mice treated from the age of 14 to 17 weeks accompanied conspicuous treatment-related differences in tumor size, indicated by mean sectional area: 19.2 ± 6.3 mm2 after vehicle, 4.3 ± 0.8 mm2 after anti-VEGF antibody (78% less than vehicle), and 3.7 ± 0.8 mm2 after sunitinib (81% less than vehicle; ref. 8).
Increased c-Met expression after inhibition of VEGF signaling
The involvement of c-Met signaling in the changes in lymphatics and lymphatic metastases was examined by immunohistochemical assessment of the amount of the receptor and receptor activation reflected by c-Met and phospho-c-Met immunoreactivities. In the normal pancreas of wild-type mice, c-Met immunoreactivity was not detected in pancreatic islets or in lymphatics (Fig. 4A, i). In RIP-Tag2 mice treated with vehicle from the age of 14 to 17 weeks, c-Met immunoreactivity was strong in some blood vessels, but was weak or absent in tumor cells and lymphatics (Fig. 4A, ii and Supplementary Fig. S2A, i). In comparison, after treatment with sunitinib, strong c-Met immunoreactivity was found in tumor cells, lymphatics (Fig. 4A, iii and Supplementary Fig. S2A, ii and S2A, iii), and blood vessels (Supplementary Fig. S2B).
The proportion of lymphatics with c-Met immunoreactivity, which was zero in wild-type islets, was 43% in tumors after vehicle, although weak in intensity, and was significantly greater (73%–75%) after treatment with anti-VEGF antibody or sunitinib (Fig. 4B, i). The intensity of c-Met immunoreactivity in lymphatics was also greater after anti-VEGF therapy (Fig. 4B, ii). The proportion of tumor cells with c-Met immunoreactivity increased from 2% after vehicle to 73% after anti-VEGF antibody and 74% after sunitinib (Fig. 4B, iii).
The increase in c-Met immunoreactivity after sunitinib was accompanied by reduced vascular density (8) and more severe intratumoral hypoxia, assessed by pimonidazole staining (Supplementary Fig. S2C, i). However, staining for c-Met was present throughout tumors, though variable in intensity, and was not restricted to regions of pimonidazole staining (Supplementary Fig. S2C, ii).
Phospho-c-Met immunoreactivity was detected in scattered cells in tumors treated with vehicle (Fig. 4C, i), but was stronger and more widespread after sunitinib (Fig. 4C, ii). Staining for phospho-c-Met was found in lymphatic endothelial cells as well as in tumor cells (Fig. 4D, i and Supplementary Fig. S3A). As evidence of specificity, this staining was absent when the primary antibody was omitted (Supplementary Fig. S3B). The proportion of phospho-c-Met immunoreactive lymphatics and tumor cells was significantly more after anti-VEGF antibody or sunitinib (Fig. 4D, ii and iii). After sunitinib, some tumor cells inside lymphatics had immunoreactivity for c-Met and for phospho-c-Met (Supplementary Fig. S3D).
Decreased intratumoral lymphatics after inhibition of c-Met and VEGF signaling
To determine whether the number of intratumoral lymphatics in RIP-Tag2 mice treated with anti-VEGF antibody or sunitinib could be reduced by inhibiting c-Met, we examined the effects of the selective inhibitor PF-04217903 given from the age of 14 to 17 weeks. Tumors treated with PF-04217903 together with anti-VEGF antibody or sunitinib had significantly fewer lymphatics than after VEGF inhibitor alone and were comparable with the vehicle (Fig. 5A). Intratumoral lymphatics were 53% less abundant after PF-04217903 plus anti-VEGF antibody and 75% less after PF-04217903 plus sunitinib (Fig. 5B). With both drug combinations, c-Met–positive lymphatics were less than half as numerous as after the anti-VEGF antibody or sunitinib alone (Supplementary Fig. S2C, i), and phospho-c-Met–positive lymphatics were reduced essentially to zero (Supplementary Fig. S3C, i). Tumors treated with PF-04217903 alone tended to have less intratumoral lymphatics than after vehicle, but the difference was not statistically significant (Fig. 5A).
Decreased tumor cells in lymphatics after inhibition of c-Met and VEGF signaling
Blockade of c-Met by PF-04217903, together with VEGF signaling inhibition by sunitinib, reduced tumor cell invasion into lymphatics and into the acinar pancreas. Addition of the c-Met inhibitor to sunitinib did not lead to further reduction in tumor size, but did change tumor borders from an irregular, invasive phenotype to a smoother contour (Fig. 6A). Intratumoral and peritumoral lymphatics, which were abundant and contained tumor cells after sunitinib, were less numerous and had few or no tumor cells when PF-04217903 was also given (Fig. 6B and C).
The proportion of lymphatics with tumor cells inside was 58% less after PF-04217903 plus sunitinib than after sunitinib alone (Fig. 6D, i–iii) and was 46% less after PF-04217903 alone than after vehicle (Fig. 6D, iii). Fewer tumor cells stained for c-Met (Supplementary Fig. S2D, ii) or phospho-c-Met (Supplementary Fig. S3C, ii) after treatment with PF-04217903 alone or with anti-VEGF antibody or sunitinib.
HGF immunoreactivity was found in scattered cells in tumors and lymph nodes (HGF area density 4% in tumors and 20% in lymph nodes) of vehicle-treated RIP-Tag2 mice (Supplementary Fig. S4A, i and S4A, ii). The amount of HGF immunoreactivity was less after PF-04217903 alone but was similar in all other treatment groups (Supplementary Fig. S4B, i and S4B, ii).
Decreased lymph node metastasis after inhibition of c-Met and VEGF signaling
The pronounced effect of PF-04217903, on intratumoral lymphatics and lymphatics that had tumor cells inside, led us to determine whether lymph node metastases were also reduced. We found significantly fewer lymph node metastases in mice treated with PF-04217903 as a single agent (80% reduction; Fig. 7A, i and ii and B). Importantly, lymph node metastases were also less numerous when PF-04217903 was given together with anti-VEGF antibody (75% reduction) or sunitinib (75% reduction; Fig. 7A, iii–vi and B).
The present study sought to determine the involvement of c-Met signaling in lymph node metastasis after inhibition of VEGF signaling in RIP-Tag2 mice. We found that the number of intratumoral lymphatics, proportion of lymphatics containing tumor cells, and incidence of lymph node metastasis were all increased after inhibition of VEGF signaling by a function blocking anti-VEGF antibody or sunitinib. However, this was not found when c-Met was blocked by concurrent administration of the selective inhibitor PF-04217903. The addition of the c-Met blocker significantly reduced intratumoral lymphatics, tumor cells inside lymphatics, and lymph node metastases. Together, the results indicate that inhibition of c-Met in this model can reduce lymph node metastasis in the presence of VEGF blockade.
Lymph node metastasis after VEGF inhibition
Most angiogenesis inhibitors used for cancer therapy block VEGF signaling (29). This action stops sprouting angiogenesis, prunes tumor vessels, normalizes surviving tumor vessels, and reduces tumor growth (23, 30, 31).
Vascular pruning in tumors can lead to hypoxia, upregulation of hypoxia inducible factor-1α, increase of c-Met expression, and promotion of cell proliferation, motility, and invasion (9, 10, 32, 33). Increased tumor invasiveness found after inhibition of VEGF signaling could reflect this mechanism (5–8).
Treatment of RIP-Tag2 mice with anti-VEGF antibody or sunitinib from the age of 14 to 17 weeks increased c-Met and phospho-c-Met immunoreactivities in lymphatic endothelial cells. Intratumoral lymphatics and lymph node metastases were also more abundant. The increase in lymph node metastasis fits with the known relationship between the abundance of intratumoral lymphatics and incidence of lymph node metastasis (1, 3, 25, 26, 34, 35).
In the present study, 4 different markers were used to identify lymphatic vessels: LYVE-1, Prox1, VEGFR-3, and podoplanin. All are considered reliable markers of lymphatics, but some also label other cells. Prox1 is present in some colorectal cancer cells (36). VEGFR-3 and podoplanin marked lymphatic endothelial cells in RIP-Tag2 tumors but also marked some blood vessels and other cell types. However, LYVE-1 and Prox1 immunoreactivities in RIP-Tag2 tumors were mainly associated with lymphatic endothelial cells.
Treatment of RIP-Tag2 mice with anti-VEGF antibody or sunitinib from the age of 14 to 17 weeks led to a 75% or more decrease in tumor size but a 4-fold increase in number of intratumoral lymphatics. The increase in intratumoral lymphatics could result from growth of new lymphatics, from tumor cell co-option of peritumoral lymphatics, or both. The anti-VEGF antibody blocks VEGF-A, but not VEGF-C or VEGF-D, and sunitinib blocks the signaling of all VEGFRs. Lymphangiogenesis is generally thought to be driven by activation of VEGFR-2 and/or VEGFR-3 signaling by VEGF-C, VEGF-D, or in some cases VEGF-A (37–39). VEGF-C and VEGF-D are expressed by α cells of the islets of Langerhans in human pancreas (40) but were relatively low in untreated RIP-Tag2 tumors. However, VEGF-C and VEGF-D were significantly increased in tumors after treatment with sunitinib.
Lymphatic endothelial cell growth and lymph node metastasis are reduced in a breast cancer model after inhibition of VEGFR-2 and VEGFR-3 by antibodies or sunitinib (41, 42). Overexpression of VEGF-A in the skin of transgenic mice can promote lymphangiogenesis at sites of wound healing, but systemic blockade of VEGFR-2 prevents the formation of lymphatics (43). Sunitinib inhibits lymphangiogenesis in tumors under some conditions (7). Differences between our findings and some published data could reflect the tumor model, stage of tumor progression, and duration of treatment.
Proliferating endothelial cells marked by phosphohistone H3 immunoreactivity were found in intratumoral lymphatics of mice treated with vehicle but not with sunitinib. This finding fits better with co-option of peritumoral lymphatics than with lymphangiogenesis as the source of more abundant intratumoral lymphatics after sunitinib.
Effects of inhibition of VEGF signaling on lymph node metastasis seem to be context- and time-dependent. Promotion of lymph node metastasis by sunitinib was not evident in RIP-Tag2 mice treated from the age of 10 to 15 weeks (7), unlike the findings in our mice treated from the age of 14 to 17 weeks. This difference could reflect tumor stage or treatment duration. Promotion of tumor invasion and metastasis by VEGF inhibition has been reported in some studies (6–8, 44) but not in others (45, 46). In a model of nonmelanoma skin cancer, inhibition of VEGFR-2 by a blocking antibody stopped angiogenesis and prevented invasion of malignant cells (47). These effects could depend on multiple factors including tumor stage, tumor microenvironment, nature, dose, and schedule of the inhibitor.
Suppression of lymph node metastasis by c-Met inhibition
Although c-Met was not detected by immunohistochemistry in normal pancreatic islets and was limited to blood vessels in vehicle-treated RIP-Tag2 tumors, c-Met immunoreactivity was widespread in tumor cells, blood vessels, and lymphatics after treatment with anti-VEGF antibody or sunitinib. c-Met immunoreactivity was also found in tumor cells inside lymphatics after sunitinib. Treatment with the c-Met-selective RTK inhibitor PF-04217903 reduced tumor invasion, intratumoral lymphatics, and lymph node metastasis in RIP-Tag2 mice treated with vehicle, anti-VEGF antibody, or sunitinib from the age of 14 to 17 weeks. As the amount of HGF did not change under these conditions, c-Met activation could be partially ligand independent (17).
In humans, c-Met is overexpressed, constitutively active, mutated, or otherwise amplified in many aggressive tumors (11, 12), and high c-Met activity is associated with poor prognosis (14–16). Consistent with c-Met activation, inhibitors of c-Met signaling, including anti-HGF and anti-c-Met antibodies, soluble decoy receptors, and small molecule RTK inhibitors, have shown promise, particularly in combination with inhibitors of VEGF or EGF signaling, in preclinical tumor models (8, 12, 48–52) and in some clinical trials (53–59).
Multiple mechanisms are likely to be involved in the antimetastatic action of PF-04217903. Many intratumoral and peritumoral lymphatics had strong c-Met immunoreactivity after treatment with sunitinib or anti-VEGF antibody, but c-Met staining was not detected in lymphatics of the normal pancreas and was found in less than half of the lymphatics in untreated RIP-Tag2 tumors. Similarly, c-Met is strongly expressed by activated or growing lymphatics at sites of inflammation or wound healing but not in normal skin (19).
The absence of proliferating intratumoral lymphatics after sunitinib suggests that factors other than lymphangiogenesis promote the increase in intratumoral lymphatics and lymph node metastasis. Tumor borders were conspicuously smoother after treatment with PF-04217903, consistent with an anti-invasive action (8, 60). PF-04217903 could decrease intratumoral lymphatics by reducing tumor invasion into peritumoral lymphatics and by pruning lymphatics with high c-Met expression.
Effects of inhibiting c-Met and VEGF signaling together
We previously reported that inhibition of c-Met and VEGF signaling together reduced liver metastasis and prolonged survival of RIP-Tag2 mice (8). The present study expands these actions to include the reduction of lymph node metastasis. Inhibition of c-Met by PF-04217903 not only reduced lymph node metastasis when given as a single agent, but also prevented the increase in intratumoral lymphatics, tumor cell trafficking into lymphatics, and lymph node metastasis that occurred during treatment with anti-VEGF antibody or sunitinib. Together, the findings are consistent with the involvement of c-Met signaling in metastasis to liver and lymph nodes in this model and with the efficacy of blocking c-Met signaling in reducing metastasis.
In conclusion, inhibition of VEGF signaling by function-blocking antibody or by sunitinib increases intratumoral lymphatics, invasion of tumor cells into lymphatics, and lymph node metastasis in RIP-Tag2 mice with late-stage tumors. Co-option of existing lymphatics by invading tumor cells is likely to contribute to the population of intratumoral lymphatics and perhaps to promotion of lymph node metastasis. Inhibition of c-Met by PF-04217903 reduces intratumoral lymphatics and lymph node metastasis under these conditions.
Disclosure of Potential Conflicts of Interest
J.G. Christensen is employed (other than primary affiliation; e.g., consulting) as a senior director and has ownership interest in Pfizer Inc. No potential conflicts of interest were disclosed by the other authors.
Conception and design: B. Sennino, T. Ishiguro-Oonuma, J.G. Christensen, D.M. McDonald
Development of methodology: T. Ishiguro-Oonuma
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Sennino, T. Ishiguro-Oonuma, B.J. Schriver
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Sennino, T. Ishiguro-Oonuma, B.J. Schriver, J.G. Christensen
Writing, review, and/or revision of the manuscript: B. Sennino, T. Ishiguro-Oonuma, J.G. Christensen, D.M. McDonald
Study supervision: D.M. McDonald
The authors thank Peter Baluk, Casey Williamson, and Grace Laidlaw for critical review of the manuscript and Debbie Chen for genotyping the RIP-Tag2 mice.
This work was supported in part by an Overseas Research Fellowship from Ehime University (T. Ishiguro-Oonuma) and by the National Heart, Lung, and Blood Institute grants HL24136 and HL59157, the National Cancer Institute grant CA82923, a grant from Pfizer Global Research and Development, and funding from AngelWorks Foundation (D.M. McDonald).
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.