von Willebrand factor fibers promote cancer-associated platelet aggregation in malignant melanoma of mice and humans

Abstract

Tumor-mediated procoagulatory activity leads to venous thromboembolism and supports metastasis in cancer patients. A prerequisite for metastasis formation is the interaction of cancer cells with endothelial cells (ECs) followed by their extravasation. Although it is known that activation of ECs and the release of the procoagulatory protein von Willebrand factor (VWF) is essential for malignancy, the underlying mechanisms remain poorly understood. We hypothesized that VWF fibers in tumor vessels promote tumor-associated thromboembolism and metastasis. Using in vitro settings, mouse models, and human tumor samples, we showed that melanoma cells activate ECs followed by the luminal release of VWF fibers and platelet aggregation in tumor microvessels. Analysis of human blood samples and tumor tissue revealed that a promoted VWF release combined with a local inhibition of proteolytic activity and protein expression of ADAMTS13 (a disintegrin-like and metalloproteinase with thrombospondin type I repeats 13) accounts for this procoagulatory milieu. Blocking endothelial cell activation by the low-molecular-weight heparin tinzaparin was accompanied by a lack of VWF networks and inhibited tumor progression in a transgenic mouse model. Our findings implicate a mechanism wherein tumor-derived vascular endothelial growth factor-A (VEGF-A) promotes tumor progression and angiogenesis. Thus, targeting EC activation envisions new therapeutic strategies attenuating tumor-related angiogenesis and coagulation.

Figure 1

Immunofluorescence analysis of tumor microvessels compared with healthy control skin in ret transgenic mice. Cryosections were stained for VWF and CD31 (A,C). An anti-GPIb antibody was used to identify platelets (B,D). Nuclei were stained with DAPI. Representative images of control skin show VWF localized within the vessel wall, lacking ULVWF fibers within the lumen (A-B, arrowheads) and only few platelets are visible (B, red). By contrast, in tumor microvessels, ULVWF fibers are detectable within the vessel lumen indicating EC activation (C, arrows). These ULVWF fibers bind platelets as shown in the same vessel (D, red, arrows). Insets, A higher magnification of the presented images (n = 4 to 10 animals; scale bars = 20 µm). See also supplemental Figures 1-2.

Figure 2

Local inhibition of ADAMTS13 promotes VWF fiber formation in microvessels obtained from human melanoma patients. Cryosections of human malignant melanoma tissue, healthy control skin, and basal cell carcinoma were analyzed by immunofluorescence stainings for VWF and CD31 (A-C) or thrombospondin (TSP) (D-E). Nuclei were stained with DAPI. Analysis of healthy skin (A) and human basal cell carcinoma (B) as control demonstrate storage of VWF in the vessel wall (arrowheads). ULVWF fibers are detected in the lumen of the microvessels, correlating to reduced VWF within the vessel wall indicative of EC activation (C, arrows). These ULVWF fibers bind platelets (D, arrows) and are associated with microthrombi formation in distinct microvessels (E, asterisk; n = 5 to 6; scale bars = 20 µm). Quantification showed significantly increased numbers of vessels with luminal VWF fibers in tumor vasculature compared with healthy skin (F). Systemic VWF level in blood samples of malignant melanoma patients was increased compared with healthy control (G). By contrast, only a slight reduction of ADAMTS13 activity was observed (H). Tumor-derived cytokines and growth factors were measured in healthy control skin and tumor of human malignant melanoma by bio-plex. Cytokine levels of IFN-γ (I), TNF-α (J), and IL-6 (K) were increased in tumor samples compared with control skin. The concentration of VEGF-A was significantly increased within melanoma compared with control (L). Results of 9 different melanoma patients are shown (*P < .05, **P < .005, ***P < .001). Bars indicate the mean ± SD.

Figure 3

Reduced ADAMTS13 activity in tumor tissue promotes intraluminal VWF fiber formation in tumor microvessels. Immunofluorescence staining of cryosections for VWF (green), platelets (red), and the endothelial cell marker CD31 (red). DNA was stained with DAPI (blue). Mouse melanoma cells (Ret) were injected intradermally and mice were treated with recombinant ADAMTS13 (rADAMTS13; E-F) or 0.9% NaCl (C-D) as control. Reconstitution of ADAMTS13 reduced the formation of intraluminal ULVWF networks and platelet aggregation compared with vehicle treatment. In comparison with wild-type skin (A-B), less VWF in the vessel wall indicates ULVWF degradation after exocytosis (n = 5-6 animals; scale bars = 20 µm). To analyze the impact of ADAMTS13 on VWF fiber formation, vessels with or without luminal fibers were quantified. Infusion with rADAMTS13 significantly decreased the number of vessels with ULVWF (G). The activity and the protein expression of ADAMTS13 in tumor tissue were significantly reduced compared with healthy skin measured by a FRET-based assay and western blot (H; n = 3-5 animals, *P < .05, **P < .005). See also supplemental Table 2.

Figure 4

The melanoma cell line Ret induces endothelial cell stimulation via VEGF-A. HUVECs were stimulated for 15 minutes with the supernatant (sn) of the melanoma cell line Ret alone or supplemented with 0.65 mg/mL bevacizumab and thrombin (0.5 IU/mL) was used as a positive control. The efficiency of the melanoma cell-induced EC stimulation was quantified by measurement of VWF release by immunofluorescence staining (A) and by ELISA for VWF (B). Supplementation of Ret cells with bevacizumab or tinzaparin (100 IU/mL) reduced tumor cell-induced VWF release (B). Measurement of VEGF-A revealed that incubation with tinzaparin for 24 hours, 48 hours, and 72 hours resulted in a significant reduction of VEGF-A release by melanoma cells (C). The VEGF-A mRNA expression measured by real-time PCR was not affected by tinzaparin treatment of 48 hours (D). Bio-plex assays of different cell fractions showed a tinzaparin-induced reduction of VEGF-A in all fractions (E). Adding tinzaparin to Ret cell supernatant immediately before the measurements revealed an interaction of VEGF-A and tinzaparin (F). Binding of tinzaparin to VEGF-A was determined using the intrinsic tryptophan fluorescence (emission: 290 nm, excitation: 340 nm). The fluorescence of VEGF-A increases in a dose-dependent manner (G). In contrast to fondaparinux, tinzaparin exhibits a high binding affinity to VEGF-A (H). The binding of VEGF-A to tinzaparin inhibited VEGF-mediated ATP production of endothelial cells indicative for reduced cell proliferation (I). Data are presented as the mean ± SD of n = 4 of at least 2 independent experiments (*P < .05, **P < .005, ***P < .001; scale bars = 20 µm).

Figure 5

Blocking VEGF-A using tinzaparin reduces VWF fiber formation in tumor microvasculature. Immunofluorescence stainings for VWF (green) and anti-CD31 (red) in cryosections of ret transgenic tumors were performed. Nuclei were stained with DAPI. Tumor microvessels of control mice showed formation of ULVWF fibers in the vessel lumen (A, arrows). By contrast, microvessels of tinzaparin-treated mice showed almost no ULVWF fiber formation and a punctual pattern of VWF within the vessel wall (B, arrowheads), indicative of reduced endothelial cell activation. Representative pictures of tumor microvessels are shown (n = 10 animals of 2 independent experiments; scale bars = 20 µm). Tinzaparin treatment correlated with a significant reduction of vessels with intraluminal VWF fibers (C). Tumor vessels were analyzed for platelet aggregation using VWF (green) and GPIb (red) staining. Quantification showed a significant increase in platelet-covered area in the lumen of tumor vessels compared with control. This effect was abolished by treatment with tinzaparin (D-E). Panel Ei shows a single platelet in the lumen of a tumor blood vessel. In addition, the treatment of ret transgenic mice with tinzaparin (gray) reduced the appearance of middle (ii) and big aggregates (iii) to healthy control skin levels (white) compared with vehicle-treated control tumors (black). Plot shows mean ± SD (n = 5-10; *P < .05; **P < .005; ***P < .001; scale bars = 5 µm).

Figure 6

Tinzaparin attenuates proliferation of melanoma cells in primary tumor and lymph nodes and attenuates tumor progression in ret transgenic mice. Immunofluorescence stainings of tumor cell proliferation with Ki67 (green) and DAPI (nuclei, blue). Cryosections of vehicle-treated tumors (A) show more proliferating melanoma cells compared with tinzaparin-treated tumors (B; scale bars = 50 µm). Quantification showed a significant reduction of Ki67-positive cells after treatment with tinzaparin (C). Analysis of Ki67-positive tumor cells in lymph nodes of vehicle-treated ret mice and tinzaparin-treated animals was assessed by flow cytometry. Results show that tinzaparin induces a significant reduction of proliferating melanoma cells (D). Bars show mean ± SD (n = 7-10). Tumor-derived VEGF was measured in tumor and lymph nodes of ret transgenic mice (E-F) by bio-plex assay. VEGF levels were decreased in tumor samples (E) and lymph nodes (F) after treatment with tinzaparin (n = 5-7 different mice each group). Consequently, a survival benefit (G) and a reduced tumor weight (H) compared with vehicle-treated control was observed after tinzaparin treatment (n = 13 of 2 independent experiments). Plot shows mean ± SD; *P < .05. See also supplemental Video 1.

Figure 7

Tinzaparin inhibits VEGF-A–mediated angiogenesis in primary skin tumors and impedes tumor cell metastasis. Tumor-bearing mice were treated with vehicle (A; control) or tinzaparin (B) and cryosections of primary tumors were analyzed by immunofluorescences for CD31. Morphometric quantification of the vessel density (C) demonstrates a significant difference in vessel density upon tinzaparin treatment compared with control tumors. Quantitative assessment of vessels in tinzaparin-treated tumors (D) shows that tinzaparin treatment results in a significant increase of small vessels (<50 µm) and decrease of big vessels (>150 µm). Plots show mean ± SD (n = 4-6 animals of 2 independent experiments; scale bar = 100 µm; *P < .05). Schematic overview shows the role of tumor cell-mediated EC activation in cancer progression (E-G). Circulating tumor cells secrete VEGF (E) followed by activation of endothelial cells and the release of procoagulatory VWF fibers (F). VWF mediates platelet aggregation and tumor cell binding, promoting extravasation (G). Binding of the LMWH tinzaparin blocks VEGF-mediated angiogenesis and VWF release attenuating tumor progression (H).