The incidence of melanoma has strongly increased over the last 40 years.1 Melanoma metastasis shows a high rate of mortality due to the strong resistance to conventional therapy, including chemotherapy and radiation therapy. Up to now, no therapeutic regimen has yet been proven to significantly impact long-term outcome of metastatic melanoma patients in multicenter randomized studies. The propensity of tumor cells to colonize remote organs is greatly affected by active tumor-host interactions as well as by factors that are inherent both to the tumor cells and to the locally invaded host microenvironment.2, 3 One characteristic feature of cancer is the induction of an inflammatory response in the tumor tissue. The inflammatory infiltrate, composed of a diverse leukocyte population (neutrophils, macrophages, mast cells and lymphocytes), may promote tumor progression by the production of cytokines, growth factors and proteases.4 Secreted cytokines can directly influence tumor cell proliferation, whereas the release of proangiogenic factors like vascular endothelial growth factor (VEGF), IL-8 and matrix metalloproteinases (MMPs) may enable micrometastasis to acquire the ability to eventually become vascularized. In addition, remodeling of extracellular matrix components might further alter the balance towards proangiogenic stimuli, enhancing neovascularization.5, 6
There is accumulating evidence suggesting that bone marrow-derived endothelial progenitor cells (EPCs) contribute to adult neovascularization. Moreover, subsets of hematopoietic progenitor cells (HPCs) may be recruited to sites of cancer cell proliferation.7, 8 Plasma elevation of angiogenic factors, including VEGF and placenta growth factor (PlGF), has been shown to induce mobilization and recruitment of both vascular endothelial growth factor receptor-2 (VEGFR-2)+EPCs and VEGFR-1+HPCs,9, 10 contributing to tumor cell survival and neovessel-dependent tumor growth. However, the extent to which postnatal vasculogenesis as opposed to angiogenesis contributes to neovessel-dependent tumor development is still a matter of controversy.11
VEGF and its receptor tyrosine kinases VEGFR-1 and -2 have received great attention as important therapeutic targets in pathologic angiogenesis.12, 13 Inhibition of VEGFR-2 signaling was demonstrated previously to suppress tumor angiogenesis and growth in different tumor models.14, 15, 16, 17, 18 More recently, anti-VEGFR-1 antibody (Ab) strategies have been shown to block angiogenesis in both malignant and inflammatory diseases.19 While the degree of anti-VEGFR-1-mediated inhibition alone may vary,19 simultaneous inhibition of both VEGFR-1 and -2 tends to augment the antivascular effects of anti-VEGFR-2 in a local tumor model, which was attributed in part to blocking recruitment of VEGFR+ bone marrow-derived progenitors.7
In our study we analyzed the influence of VEGFR signaling on subcutaneous B16 melanoma growth and lung metastasis formation. The growth of solid B16 melanoma was not significantly influenced by the inhibition of VEGFR-1 or -2 alone through the retroviral gene transfer of VEGFR mutants. Blockade of VEGFR-1 by a neutralizing Ab strongly reduced the inflammatory infiltrate of B16 melanoma metastasis without a significant decrease in metastasis formation. Also, the inhibition of endothelial cell (EC) function by the inhibition of VEGFR-2 signaling had no significant impact on the formation of B16 melanoma lung metastasis. In contrast, simultaneous blockade of both, VEGFR-1 and -2 signaling strongly suppressed the growth of subcutaneous melanoma and metastasis formation. However, the enhanced mobilization and proliferation of VEGFR-1+ myeloid cells and VEGFR-2+ endothelial progenitors induced by the formation of lung metastasis was not altered by the blockade of VEGFR signaling. Therefore, the antimetastatic effects achieved by combined inhibition of VEGFR-1 and -2 signaling may likely be mediated via targeting cell populations other than myeloid and endothelial progenitor-like cells alone.
The results of our study suggest that in the B16 tumor models used the blockade of both the inflammatory and the VEGFR-2 dependent angiogenic response is necessary to successfully inhibit solid tumor progression and formation of lung metastasis.
Material and methods
Laboratory reagents were purchased from Sigma-Aldrich (Schnelldorf, Germany) unless otherwise stated. Media and supplements were obtained from Invitrogen Gibco (Karlsruhe, Germany). B16 melanoma cells were grown in DMEM (Invitrogen) supplemented with 10% FCS, 1% penicillin/streptomycin and 2% glutamine. GP+E86 retrovirus producer cells were grown in DMEM supplemented with 10% calf serum, 2% glutamine, 1% pyruvate, 0,1% minimal essential aminoacids and 1% penicillin/streptomycin.
Generation of recombinant retrovirus producer cell lines
The generation of GP+E86 cells producing murine ecotropic retroviruses encoding the truncated VEGF receptors, mutVEGFR-1 or -2, was described earlier.18 Virus titer was determined by infecting NIH3T3 cells and counting of the G418-resistant colonies. For infection, NIH3T3 cells were grown to 50% confluency and infected twice for 2 hr at 37°C with virus-containing conditioned medium of GP+E86 retrovirus producer cell lines20 supplemented with 8 μg/ml polybrene. Virus titers were ∼1 × 106 pfu/ml for GP+E86/mutVEGFR-1 and 2 × 106 pfu/ml for GP+E86/mutVEGFR-2.
Antibodies and immunoconjugates
For flow cytometric analysis, the following antibodies were used: rabbit anti-VEGFR-1 (ImClone Systems, New York, NY), rat anti-CD11b-allophycocyanin (BD PharMingen, Heidelberg, Germany), rat anti-VEGFR-2-phycoerythrin (PE; BD PharMingen), rat anti-CD34-fluorescein isothiocyanate (FITC; BD PharMingen). For immunocytochemistry, rat anti-CD45 (clone M1/9; American Type Culture Collection, Manassas, VA), monoclonal rat anti-PECAM-121 and monoclonal rat anti-VEGFR-222 was used.
CD45 staining was performed on 8-μm frozen sections. Slides were fixed with ice-cold acetone and air-dried. Sections were then incubated with the respective anti-CD45, anti-PECAM-1 or anti-VEGFR-2 Ab, followed by incubation with a biotinylated rabbit anti-rat secondary Ab (Vectastain Elite ABC kit; Vector Laboratories, Bulingame, VT), which was detected with avidin-biotinylated horseradish peroxidase complex (Vectastain Elite ABC kit).
Flow cytometry analysis
Subsets of hematopoietic and endothelial stem/progenitor cells were identified by their pattern of surface markers. ECs with progenitor potential were defined by positive staining for VEGFR-2 and CD34 (26,27), myeloid hematopoietic cells by positive staining for VEGFR-1 and CD11b. Peripheral blood was drawn from the retrobulbar plexus into heparinized capillaries. Cells were stained as reported previously.23, 24 After incubation with the indicated antibodies for 30 min at 4°C, cells were washed in PBS and then fixed in 1.5% formaldehyde/PBS. Data acquisition and analysis was performed by using a BD FACScan™ Cytometer (Becton Dickinson, Mountain View, CA), equipped with CellQuest™ software (BD Biosciences, Heidelberg, Germany). Staining was related to isotype-matched control antibodies, purchased from BD PharMingen. Each analysis included at least 100,000 events.
Human umbilical vein endothelial cells (HUVECs) or B16 cells were trypsinized, washed with PBS containing 0.1% FBS (wash buffer), and aliquots were transferred into tubes for Ab staining. HUVECs were incubated with mouse anti-human VEGFR-1-PE Ab (clone 49560, IgG1, R&D Systems, Wiesbaden–Nordenstadt, Germany), mouse anti-human VEGFR-1-PE Ab (clone 49560, mouse IgG1, R&D Systems), or isotype control mouse IgG1 PE (Caltag, Hamburg, Germany) for 30 min on ice in a total volume of 50 μl. B16 cells were incubated with goat anti-mouse VEGFR-1 Ab (clone 141515, goat IgG, R&D Systems), followed by rabbit anti-goat IgG FITC (DakoCytomation, Hamburg, Germany), rat anti-mouse VEGFR-2-PE Ab (clone Avas 12α1, IgG2A, BD PharMingen, Heidelberg, Germany), or isotype control goat IgG FITC or rat IgG PE (Caltag). Subsequently, cells were washed twice and analyzed by a BD FACScan™ Cytometer (Becton Dickinson, Mountain View, CA), equipped with CellQuest™ software for data acquisition and analysis (BD Biosciences, Heidelberg, Germany). About 10,000 cells were collected for analysis. Nonviable cells were identified and excluded by propidium iodide staining.
For reverse transcriptase-polymerase chain reaction (RT-PCR) analyses, 2 μg of total RNA was prepared from HUVEC or B16 melanoma cells by the RNeasy Mini Kit (Qiagen, Hilden, Germany), and reverse transcribed with the enzyme AMV Reverse Transcriptase from the Titan 1 Tube RT-PCR Kit (Roche Diagnostics, Mannheim, Germany). Specific primers were used for amplification of a 620-bp fragment of both the murine or human VEGFR-1 cDNA (5′-AGCAAGTGGGAGTTTGC-3′, 5′-CTGCCAGGTCCCGATGAATG-3′; nt 2,706–3,324, accession number NM 010228.2; nt 2,452–3,073, AF063657.2), a 659-bp fragment of both the murine or human VEGFR-2 cDNA (5′-TACGGACCGTTAAGCGGGCC-3′, 5′-ATGAGATGCTCCAAGGTC- AG-3′; nt 2,558–3,218, 59397.1; nt 2,660–3,320, M 002253.1) and a 751-bp coding sequence of both the murine or human GAPDH cDNA (5′-CACCATCTTCCAGGAGCG-3′, 5′-CCACCCTGTTGCTGTAGC-3′; nt 252–997; BC083149.1; nt 246–997; BC023632.2). After a hot start at 94°C, primers were annealed for 45 sec (VEGFR-1 at 57°C, VEGFR-2 at 52°C and GAPDH at 45°C) and extended for 45 sec at 72°C. PCR reactions were carried out for 45 cycles (VEGFR-1, VEGFR-2) or 25 cycles (GAPDH) in a T3 Thermocycler (Biometra, Goettingen, Germany). The resultant PCR products were visualized in 1.5% ethidium bromide-stained agarose gels.
C57BL/6J mice, aged 8–12 weeks, were purchased from Charles River (Sulzfeld, Germany) and maintained on a 12-hr light-dark cycle. BLACK-nu immunodeficient mice (3–4 backcrossings of the nu gene to C57BL/6J) were from Taconic M&B A/S (Ry, Denmark). Mice were maintained on a 12-hr light-dark cycle at the animal facility. Animals were fed acidified drinking water and standard chow ad libitum. All protocols were approved by the Governmental administration (Darmstadt, Hessen).
Subcutaneous B16 melanoma model
Confluent cultures of B16 melanoma cells25 and GP+E86 cells were washed 2 times with PBS and trypsinized. Cell suspensions were collected by centrifugation (1,200g for 5 min at room temperature). For coinjection of B16 melanoma cells and GP+E86/LXSN or GP+E86/mutVEGFR-1 or GP+E86/mutVEGFR-2, the cell pellets were resuspended in PBS, and 1.5 × 106 cells/cell line in 50 μl were injected s.c. into adult C57BL/6J mice. Triple-injections were performed by injecting 1.5 × 106 B16 melanoma cells in 50 μl together with 0.75 × 106 GP+E86/mutVEGFR-1 and 0.75 × 106 GP+E86/mutVEGFR-2 in 50 μl. Tumors were harvested 14 days postinjection for determination of tumor weight.
Experimental lung metastasis assay
As described previously,26 B16 cells were prepared for inoculation by brief exposure to 0.25% trypsin/0.02% EDTA solution (from Sigma-Aldrich, Schnelldorf, Germany). Cell viability in single-cell suspensions was determined by trypan blue exclusion. A total of 3 × 105 B16 cells (>90% viability) in 100-μl PBS were injected into the lateral tail vein of 8–12 week old male syngeneic C57BL/6J or BLACK-nu mice, respectively. As indicated, mice received rat IgG (Sigma-Aldrich; IgG from rat serum, reagent grade 95%; 1 mg/mouse at 3-days interval), anti-mVEGFR-1 (ImClone Systems; MF1, 750 μg/mouse at 3-days interval), and/or anti-mVEGFR-2 (ImClone Systems; DC101, 750 μg/mouse at 3-days interval) by intraperitoneal injection, starting 48 hr after tail vein-injection. Animals were killed at Day 11 (C57BL/6J) or Day 20 (BLACK-nu) postinjection. Lungs of C57BL/6J mice were excised and fixed in phosphate-buffered 10% formaldehyde. Metastatic foci at lung surfaces were counted by 2 observers in a blinded fashion. Tumor-bearing BLACK-nu mice were transcardially perfused at Day 20 with 1% paraformaldehyde, and lungs were excised and instillated 0.5 ml Tissue-Tek® O.C.T™ (Sakura Finetek, Zoeterwoude, Netherlands) intratracheally. Images of the lungs were taken in a standardized fashion, and organs were then embedded in Tissue-Tek® O.C.T™, and stored at −80°C. Area calculation of metastatic foci at lung surfaces was performed on standardized photographs, utilizing the public domain Java image processing program ImageJ (v1.30).
Comparisons between groups were analyzed by two-sided t-test or the Wilcoxon test for 2 independent samples. The p values for pairwise multiple comparisons were adjusted according to the Bonferroni procedure (p values multiplied by the number of comparisons); significance was established for p values < 0.05.
Simultaneous retroviral gene transfer of truncated VEGFR-1 and -2 mutants is necessary to block the growth of subcutaneous B16 melanoma
To analyze the therapeutic potential of VEGFR mutants on the growth of highly metastatic cancer cells, we used a model of subcutaneously induced B16 melanoma in a first step to establish a functional therapeutic regimen. To block VEGFR signaling, retrovirally produced VEGFR-1 and -2 mutants that lack the intracellular tyrosine kinase domain (mutVEGFR-1 and -2) were used. Coinjections of B16 melanoma cells and GP+E86 cells that produce recombinant retrovirus encoding either mutVEGFR-1 (GP+E86/mutVEGFR-1) or mutVEGFR-2 (GP+E86/mutVEGFR-2) were performed in syngeneic mice. In addition, tumor cells were coinjected with a combination of both GP+E86/mutVEGFR-1 and GP+E86/mutVEGFR-2. As a negative control, B16 melanoma cells were coinjected with retrovirus producer cells that contained the empty retroviral vector only (GP+E86/LXSN). Coinjection with GP+E86/LXSN did not affect the in vivo-growth behavior of the parent B16 cell line (Suppl. 1). Tumors were harvested 14 days post injection and the tumor weight was determined (Fig. 1). No significant inhibition of tumor growth was observed by the separate retroviral gene transfer of either mutVEGFR-1 or -2. Remarkably, however, the simultaneous injection of tumor cells along with both GP+E86/mutVEGFR-1 and GP+E86/mutVEGFR-2 led to a strong and significant reduction of tumor growth. Immunohistological examinations on tumor vessels do not reveal a significant reduction of vascular density in tumors that were treated separately with either mutVEGFR-1 or mutVEGFR-2 (Figs. 2a, 2c, 2e and 2i). In contrast, a significant 39% decrease in tumor vascularity is observed in tumors, in which simultaneous retroviral gene transfer of both GP+E86/mutVEGFR-1 and GP+E86/mutVEGFR-2 (Figs. 2g and 2i) had been performed compared to B16/LXSN tumors. Therefore, these findings correlate well with the observed treatment response in those tumors. However, the frequency of VEGFR-2 positive capillaries was not influenced by the retroviral gene transfer of the VEGF receptor mutants, since the majority of all tumor vessels stained positive for VEGFR-2 (Figs. 2b, 2d, 2f and 2h), irrespective of the experimental condition. Interestingly, the retroviral gene transfer of the VEGFR-1 or -2 mutants, either alone or in combination, led to a strong increase in dilated tumor vessels (Figs. 2c–2h). Though application of the VEGFR-1 or -2 mutants alone is able to change tumor vessels morphology, simultaneous treatment with both receptor mutants appears to be required to inhibit tumor neovascularization and tumor growth in this model.
Experimental melanoma metastasis is suppressed by simultaneous inhibition of VEGFR-1 and -2
Next we investigated whether the inhibition of VEGFR-1 and/or -2 has equivalent effects on the development of lung melanoma metastasis. B16 melanoma cells were injected systemically in syngeneic mice and VEGFR signaling was blocked by the administration of neutralizing antibodies directed against VEGFR-1 (clone MF1; ImClone Systems) or VEGFR-2 (clone DC101; ImClone Systems). The antibodies were applied either alone or in combination. The number of metastatic foci at lung surfaces at Day 11 (Fig. 3) and Day 20 (Suppl. 2) was taken to monitor formation of B16 melanoma lung metastasis growth. To avoid any immune response to the injected antibodies because of the extended observation time BLACK-nu mice were used for the evaluation at Day 20. Administration of neutralizing Ab directed against VEGFR-1 (starting at Day 3 p.i., 750 μg/day, every third day) led to a modest, yet insignificant decrease in lung melanoma metastatic foci (Fig. 3, Suppl. 2). In a comparable fashion, neutralizing Ab against VEGFR-2 (starting at Day 3 p.i., 750 μg/day, every third day) resulted in only marginal effects on lung colony formation. In contrast, formation and growth of B16 cell metastasis was strongly suppressed by the simultaneous application of neutralizing antibodies against VEGFR-1 and -2.
A direct influence of VEGF on melanoma growth has been suggested by the observed expression of VEGFR-1 and -2 on several human melanoma cell lines and their increased proliferation after VEGF stimulation in vitro.27, 28 Therefore, we analyzed the expression of VEGFR-1 and -2 on the B16 melanoma cells used in our study in vitro (Fig. 4). However, no discernable expression of either VEGF receptor by B16 cells was detected on the mRNA (Fig. 4a) and protein level (Fig. 4b). These results exclude the tumor cells as direct target cells for the inhibitory effect of the simultaneous VEGFR-1 and -2 blockades on the growth of subcutaneous B16 melanoma and the formation of lung melanoma metastasis.
Mobilization of bone marrow-derived myeloid hematopoietic cells and endothelial progenitors is resistant to the blockade of VEGFR signaling
To explore the potential relevance of bone marrow-derived progenitor populations to the process of metastasis formation and successive growth, bone marrow- and peripheral blood-derived cells were evaluated for the numbers of VEGFR-1+ and -2+ cell subsets in metastasis-bearing C57BL/6J mice compared to healthy controls (Fig. 5). A significant increase in peripheral blood-derived VEGFR-1+CD11b+ myeloid hematopoietic cells and VEGFR-2+CD34+ ECs with progenitor potential was detectable at Day 11 post B16 cell inoculationem. To investigate whether the increased mobilization into the circulation was a consequence of an increased proliferation in the bone marrow, the number of VEGFR-1+CD11b+ and VEGFR-2+CD34+ cells were analyzed in the bone marrow of diseased mice and unaffected controls. Moreover, the amount of hematopoietic VEGFR-1+CD117+ and endothelial VEGFR-2+CD117+ progenitor cells were determined. An increase of both VEGFR-1+ hematopoietic cell subsets could be demonstrated (Figs. 5d and 5f). Whereas no significant increase in the numbers of VEGFR-2+CD117+ cells was noticed (Fig. 5c), the count of VEGFR-2+CD34+ cells was found to be significantly higher in the bone marrow from metastatic mice (Fig. 5e). Together, these data indicate that the establishment of experimental melanoma lung metastasis induces the proliferation of both VEGFR-1+ and VEGFR-2+ progenitor cells and their subsequent mobilization to the circulation.
However, the increase in both peripheral blood-derived VEGFR-1+CD11b+ myeloid hematopoietic cells and VEGFR-2+CD34+ cells in metastasis-bearing mice at Day 11 was found to be resistant to treatment with either neutralizing VEGFR-2 or -1 Ab (Fig. 6). In addition, simultaneous Ab-mediated inhibition of VEGFR-1 and -2 signaling, which suppressed lung metastasis formation (Fig. 4), failed to reduce the numbers of the defined cell subsets in metastasis-bearing C57BL/6J mice. Furthermore, no changes in the over-all blood counts were observed by VEGFR blockade (Suppl. 3). Therefore, the induced mobilization of both studied cell populations into the circulation may not constitute an essential target of antimetastatic effects of combined VEGFR-1 and -2 inhibition in experimental lung metastasis. In addition, our data indicate that levels of selected circulating VEGFR+ cell subsets fail to correlate with treatment responses in our metastasis model. In contrast to previous findings (reviewed in Ref.29), mere numbers of the studied hematopoietic and endothelial circulating cells may therefore not serve as appropriate pharmacodynamic markers in B16 lung metastasis to monitor treatment activity.
The over-all reduction of CD45+ inflammatory infiltrates has no effect on experimental melanoma lung metastasis formation
Inhibition of VEGFR-1 signaling was demonstrated to reduce the accumulation of mononuclear or CD45+ cells in malignant or inflammatory disease models.7, 19 To explore whether cellular changes of the inflammatory infiltrate could provide evidences as to the mode of action of anti-VEGFR Ab-mediated effects, metastatic lesions were stained with the leukocyte-specific marker CD45 (Figs. 7a–7d). Whereas only combined inhibition of both VEGFR-1 and -2 was seen to inhibit formation of lung melanoma colonies, our histochemical analysis of Day-20 metastasis for CD45 expression revealed that neutralizing Ab against VEGFR-1 alone already conveyed reduction in the inflammatory infiltrate of melanoma lung metastasis (Fig. 7b). Though histochemical analysis of metastatic colonies from mice that were treated with both anti-VEGFR-1 and -2 Ab is hampered by sheer deficiency in number, no further reduction in CD45+ cells was noted compared to lung colonies from mice that received VEGFR-1 Ab only (Fig. 7d). Hence, VEGFR-1 Ab-mediated reduction in CD45+ inflammatory infiltrates may be necessary but not sufficient for the inhibition of experimental lung metastasis formation.
Malignant melanoma represents one of the most aggressive human tumor types, whose incidence is rapidly increasing.30 The high resistance of advanced stage and metastatic melanoma to standard chemotherapy and radiation therapy is the cause for most of the deaths from malignant melanoma.31 At present, there are no therapeutic regimens available that lead to a substantial increase in life expectancy once melanoma has formed metastasis, which can not be treated by local surgical excision.1
Inhibitors of VEGF and its receptors are being evaluated as viable antiangiogenic agents in several malignancies and inflammatory diseases.13 However, the role of tumor angiogenesis during melanoma progression and metastasis is still under discussion. Whereas several studies demonstrated a reduced disease-free and overall survival of melanoma patients with increasing tumor vascularity,32, 33, 34 others showed no association between melanoma vascularization and prognosis.35, 36
In our study, we analyzed the role of VEGFR signaling in solid B16 tumor growth and the formation of experimental melanoma lung metastasis. Inhibition of VEGFR-2 signaling has already been demonstrated to suppress angiogenesis and growth in different local tumor models.16, 17, 18 Recently, also anti-VEGFR-1 Ab was shown to block tumor angiogenesis and growth of epidermoid A431 tumors in a fashion comparable to that seen with anti-VEGFR-2.19 Dose-dependent effects may explain as to why neutralizing anti-VEGFR-1 Ab alone did not reveal discernible therapeutic effects on the growth of tumors derived from Lewis lung carcinoma (LCC) cells.7 However, addition of anti-VEGFR-1 at doses that alone did not affect tumor growth seemed to augment the antivascular effects of anti-VEGFR-2 in the LCC tumor model.7
We first analyzed the efficiency of VEGF receptor inhibition on the growth of subcutaneously induced B16 melanoma. Retroviral gene transfer of VEGFR-1 or -2 mutants lacking the intracellular tyrosine kinase domain was not sufficient to effectively inhibit tumor growth. In contrast, a significant reduction of solid tumor growth and tumor vascular density was observed by the simultaneous retroviral gene transfer of both VEGFR mutants (Figs. 1 and 2), indicating that at least in our model signaling via both VEGF receptors is necessary for the growth of solid B16 melanoma. In a previous study, the administration of a neutralizing anti-VEGFR-2 Ab led to a growth retardation of subcutaneous B16 melanoma.16 Likely, the apparent differences compared to our findings may result from the lower amount of injected tumor cells, as also untreated control tumors in that study showed a strongly delayed onset of tumor growth. These data suggest that the inhibitory effect of the Ab-mediated VEGFR-2 blockade may, at least in part, be a consequence of slower growth kinetics. Moreover, Prewett et al.16 may have used a different B16 melanoma cell line that varied in terms of growth behavior.
To analyze the dependence of lung melanoma metastasis on VEGFR signaling, mice were treated with neutralizing antibodies against VEGFR-1 or -2, respectively, after systemic injection of B16 melanoma cells. Taking metastatic foci at lung surfaces as end-point in our experimental metastasis assay, sole administration of neutralizing Ab directed against either VEGFR-1 or -2 failed to exert significant effects on the formation of lung melanoma metastasis (Fig. 3). Similar to the growth inhibition of solid B16 melanoma, only the simultaneous application of neutralizing antibodies against both, VEGFR-1 and -2 substantially suppressed the formation and growth of lung metastasis. The failure of the neutralizing VEGFR-2 Ab to inhibit the formation of lung melanoma metastasis may be due to subpopulations of melanoma cells that are able to survive under hypoxic conditions or metabolic stress,37 suggesting that these cells are capable to resist an antiangiogenic therapy. Furthermore, the formation of alternative vessel structures, which are lined by not only ECs, but also by tumor cells, has been described.38 This so called vascular mimicry, which has been also observed in aggressive uveal melanomas,39 might render the developing metastasis less dependent on angiogenic processes mediated by VEGFR-2 alone. Several human melanoma cell lines derived from primary and metastatic tumors express both VEGF receptors and show an increased proliferation after VEGF stimulation in vitro,27, 28 suggesting a direct influence of VEGF on melanoma growth. However, the B16 melanoma cells used in our study were negative for VEGFR-1 and -2 expression on mRNA and protein level (Fig. 4). Therefore, it is unlikely that the inhibitory effects observed by the simultaneous blockade of both VEGF receptors are mediated by direct inhibitory effect on the tumor cells themselves.
While accumulating evidence suggests that bone marrow-derived EPCs contribute to adult neovascularization,7, 40, 41, 42 the validity and significance of the concept that endothelial progenitors incorporate into newly formed vessels and develop into mature ECs is still controversial (reviewed in Refs.29 and43). The level of endothelial progenitor contribution to vessel-dependent tumor growth is viewed to be highly variable, ranging from estimates of minor contributions44, 45, 46 to notions of a crucial role in new vessel formation.9, 47 These diverse assumptions result from the variable degree of endothelial progenitor incorporation, which may best be explained by the diversity of different tumor models used. Pertinent to this conclusion, the rate of incorporation has been shown to be higher in metastatic as opposed to primary tumors, and may also greatly vary as a function of tumor site, tumor type and tumor differentiation (reviewed in Refs.29 and48).
In addition, antiangiogenic therapies have been found to critically affect the level of peripheral blood endothelial VEGFR-2+ progenitor-like cells. Specifically, suppressive effects of different antiangiogenic drugs, including the DC101 mAb, on the level of circulating endothelial progenitors have been demonstrated (reviewed in Ref.29), suggesting that measurements of circulating ECs may serve as useful surrogate markers for monitoring therapeutic efficacy of antiangiogenic treatments. However, our studies reveal that neither administration of VEGFR-1- and -2-directed Ab alone nor the effective combination of both antibodies have a significant impact on selected VEGFR-1+ and VEGFR-2+ cell subsets in the B16 lung metastasis model. Therefore, counts of the selected VEGFR+ myeloid and EC subsets in our metastasis model failed to correlate with preclinical response. While the differences to previous data may possibly be explained by the metastasis tumor model and potentially by different mouse strains used, the methods and definitions of phenotypic progenitor cell detection may also contribute to diverse results, as phenotypical discrimination between mature and progenitor-like ECs in the peripheral blood is still very complex and has yet to be fully resolved (reviewed in Refs.29, 43 and49).
In addition to endothelial progenitors, also subsets of HPCs may be recruited to sites of cancer cell proliferation.7, 8 Plasma elevation of angiogenic factors, including VEGF and PlGF, has been shown to induce mobilization and recruitment of both VEGFR-1+HPCs and VEGFR-2+ECs,9, 10 contributing to tumor cell survival and neovessel-dependent tumor growth. After induction of melanoma lung metastasis, we observed an increase in VEGFR-1+ hematopoietic progenitor-like cells both in the bone marrow and in the peripheral blood (Fig. 5). Comparable to endothelial progenitor-like cells, enhanced mobilization of myeloid hematopoietic cells was also not changed by inhibition of VEGFR signaling (Fig. 6). Though our studies do not address the controversy on the validity of the concept of endothelial progenitors, as the fate of progenitor cells is not actively followed in the tumor compartment, we speculate based on our findings that it is rather unlikely that the analyzed circulating VEGR1+ and VEGFR-2+ cell populations play a pivotal role during B16 experimental metastasis. Mobilization of these cell subsets from the bone marrow and their recruitment to sites of tumorigenesis has been shown previously to be inducible by VEGF.9, 50 As cell count in the marrow and the peripheral blood remained unchanged in the presence of VEGFR-1 and -2 antibodies in our study, additional VEGFR-independent mechanisms of mobilization may compensate for VEGFR-directed inhibition, e.g., by the effects of factors like SDF-1 that are known to mobilize precursor cells from the bone marrow as well.50, 51
Beside the local host microenvironment, inflammation has been recognized as an important contributor to tumor growth.4, 52 Cells of the mononuclear lineage were demonstrated to constitute a major component of the leukocyte infiltrate at sites of tumor growth.53 Macrophages associated with tumor cells originate from monocytic precursors and are both target and source of cytokines and chemokines within the tumor microenvironment.54 Significantly, tumor-derived growth factors (e.g., VEGF, PlGF, macrophage-colony stimulating factor) and different chemokines (e.g., CC chemokine CCL2) both contribute to macrophage recruitment and to induction of monocytes and macrophages to secrete proangiogenic factors and MMPs.54 These observations foster the concept that angiogenesis-dependent tumor growth is closely related to a network of mutual interactions between tumor cells, leukocytes, ECs and components of the host stroma.2, 3 Pertinent to this conclusion, mononuclear cell infiltration has been shown to be strongly associated with tumor angiogenesis in breast cancer and in cutaneous melanoma.55, 56 In addition, the mobilization and recruitment of bone marrow-derived CD45-positive cells to tumor tissues has been shown to contribute to tumor growth7, 8 and neovascularization of VEGF overexpressing organs.50 Here, we observed a markedly reduced accumulation of CD45+ inflammatory cells in Day-20 metastasis in anti-VEGFR-1 treated animals (Fig. 7), comparable to previous findings on subcutaneous epidermoid A431 tumors.19 The findings within our tumor model suggest that homing of CD45+ cells into the metastatic tissue is dependent on VEGFR-1 signaling, whereas the mobilization of bone marrow-derived myeloid hematopoietic cells in the peripheral blood may be mediated by VEGFR-independent mechanisms. However, single inhibition of VEGFR-1 signaling merely resulted in minor effects on lung colony formation (Fig. 3), indicating that the over-all reduction in CD45+ inflammatory cells may be required, but by itself may not be sufficient to inhibit experimental lung metastasis formation. The latter conclusion is supported by the fact that only simultaneous inhibition of both VEGFR-1 and -2 sufficiently suppressed metastasis formation. Tumor-associated leukocytes may likely contribute to tumor progression by secreting MMPs, growth and angiogenic factors. In addition, they promote the establishment of metastasis by the release of chemokines, thereby supporting the recruitment of chemokine receptor-expressing tumor cells with metastatic potential (reviewed in Ref.57).
Whereas the precise mechanisms of action of anti-VEGFR blockade are yet to be defined, our data clearly indicate that simultaneous signaling via VEGFR-1 and -2 is critical for experimental melanoma lung metastasis to occur. Furthermore, our findings provide a rationale for further evaluating combined anti-VEGFR-1 and anti-VEGFR-2 treatment approaches to interfere with inflammation-associated and angiogenesis-dependent metastatic growth.
This work was supported by Deutsche Forschungsgemeinschaft grants Gi 229/5-1 (J.G.) and SFB/TR23 Project C3 (J.G. and R.H.), and by the BMBF and the EU (G.B.).