Conceived and designed the experiments: DMP RD. Performed the experiments: DMP GC SB. Analyzed the data: DMP KI BS. Contributed reagents/materials/analysis tools: RD LS GC PS MT. Wrote the paper: DMP RD HM MT.
The authors have declared that no competing interests exist.
Recent molecular studies provide evidence for a significant transcriptional plasticity of tumor cell subpopulations that facilitate an active contribution to tumor vasculature. This feature is accompanied by morphological changes both
Melanoma is a highly aggressive neoplasm. Even very thin primary tumors (Breslow tumor thickness <1 mm) may seed metastases and precipitate rapid death
The introduction of high through put assays such as gene expression arrays have offered new tools for cancer classifications including melanoma.
In the present study we investigated melanoma metastases for neovascularization in both human biopsies and mouse xenografts of human melanoma cell lines. We found VMs and a tremendous blood micro-neovascularization in human biopsies as well as in xenografts. In the mouse xenografts of human melanoma cell lines we showed melanoma cell participation in capillary formation.
Immunohistochemistry was performed on three different tissue microarrays (TMAs) representing a total of 49 primary melanomas and 127 melanoma metastasis. The metastasis array included 21 lymph node and 106 organ metastases. Construction of TMAs was performed as previously described
Immunohistochemical stainings on TMAs and melanoma metastases were performed using an automated immunostainer (Ventana Medical Systems, Tucson, AZ, USA), utilizing the antibodies anti-D2–40 (Dako A/S, clone D2–40, dilution 1∶50, pretreatment: CC1-buffer for 90 min./98°C), anti-CD34 (Novocastra, NCL-END, dilution 1∶60, w/o pre-treatment), detection was performed with Ultra View-HRP-Kit on Ventana Benchmark.
PanMelanoma Cocktail (Biocare Medical, clone HMB-45+M2-7C10+M2-9E3, dilution 1∶50, pre-treatment: CC1-buffer for 20 min/98°C), detection was performed with Ultra View-AP Kit (substrate: New-fuchsin) on Ventana Benchmark. Anti-Glut-1 (Chemicon, AB1341, dilution 1∶1000, pre-treatment: CC1-buffer for 20 min/98°C) and anti-MECA32 (PharMingen, #553849, dilution 1∶100, pre-treatment: Leica buffer H2 for 40 min/98°C) detection were performed with Refine-HRP-Kit on BondMax from Leica. Collagen IV (BMA Biomedicals AG, dilution 1∶30, pre-treatment: Ventana protease 1/12 min.) Detection was performed with iVIEW-HRP-Kit on Ventana Discovery, using secondary antibody Donkey anti-Rabbit biotinylated, Jackson 711-065-152, diluted 1∶80). Substrate for all HRP-Kits was Diaminobenzidin (DAB). Counterstain was done with Hematoxylin from the Immunostaining-Kits
Studies described in this publication were performed according to Swiss Animal Welfare laws and specifically as described in animal licenses No. 77/2008 (Int. 3816) issued and approved by the Cantonal Veterinary Office (official state office), of the Canton of Zurich (permission submitted in the supplement). The approval was signed by Mrs. Dr. Simone Gilg, scientific board deputy and responsible for the review board of the Cantonal Veterinary Office, of the Canton of Zurich. The approval ID/permit number is 2008077.
The melanoma cell line (M010308) and the melanoma cell lines used for chromosome 17/HER-2 probe were obtained from consenting patients, according to the guidelines of the University Hospital of Zurich. The permission to produce cell cultures from fresh tumor tissue was given from the cantonal ethic commission in Zürich. The consent statement was written (EK-No. 647, see attached permission in the supplement). Melanoma cell lines used for the chromosome 17/HER-2 probe are additionally described in more detail in a recent publication
In order to demonstrate mosaic microvessels 2 xenotransplantations with male/female constellation were established as follows: a male melanoma cell culture (M010308) established 4 years ago which went through at least 20 passages, which was frozen and re thawed several times and finally was used for xenotransplantations in female athymic nude mice. The cell line (M010308) was proven to be negative for CD31 and CD34 by immunohistochemistry (data not shown). From the melanoma culture, a total of 3×10E6 cells were injected into both flanks of 8-week-old female athymic nude mice. Mice were kept in individually ventilated cages for a maximum of 75 days postinjection. Tumour volume was measured using vernier calipers once every 1 or 2 days during the linear growth phase. If at least one xenograft tumor reached 1 cm3, the mouse was sacrificed and the tumors were investigated with the y-chromosome probe.
In addition using melanoma patients cell line
FICTION simultaneously enables to perform cytogenetic investigation by FISH and immunophenotyping by fluorescence marked immunohistochemistry (FICITION =
Immunohistochemistry on BondMax (Leica), detection of primary Rat anti-Mouse panendothelial MECA32 (Pharmingen Becton/Dickinson, dilution 1∶100) with Refine HRP-Kit (Leica) and Link-antibody Rabbit anti-Rat (Jackson, dilution: 1∶150). Antigen-Retrieval: HIER buffer2 (Leica) for 40 min, Chromogen DAB, without counterstain. Followed by FISH with HER-2- and centromeric probe for chromosome 17 (ABBOTT Molecular), according to the manufacturers guidelines. Counterstain with DAPI.
FISH analysis for CEP X (DXZ1) Spectrum Green/CEP Y (DYZ3) Spectrum Orange Probe (ABBOTT Molecular) was performed according to the manufacturers' protocol. FISH procedure was followed by the detection of human endothelial cells: Mouse anti-human CD31 (DAKO, dilution 1∶10) was coupled with a fluorescence labeled secondary antibody: Goat anti-mouse Alexa488 (Molecular Probes, dilution 1∶100), counterstain with DAPI (Incubation at room-temperature for 30 min each).
A cell was defined as human-derived as follows: on confocal microscopy, the nucleus had to present typical endothelial morphology, being positive for the Y-probe, and the CD31 positive cytoplasm had to be fully discernible.
All procedures were done on BondMax (Leica) with Kits and solutions from Leica according to the manufacturer's guidelines.
Epitope-retrieval was performed with Bond heat-pre-treatment-buffer 2 for 20 min. at 100°C. Immunohistochemistry was done with rabbit anti-CD31 (Abcam Ltd. Ab28364 dil. 1∶100) and Bond™ Polymer Refine Red Detection (Leica) without nuclear staining. Endogenous biotin was subsequently blocked with Avidin-Biotin-Blocking Kit (CellMarque, Cat. 928B-02) and in-situ-hybridization was performed using enzyme pretreatment (Leica, Enzyme) followed by the DNA-hybridization with a biotinylated human ALU-repeat DNA_positive-control-probe (Leica, Cat. PB0682) according to the manufacturer's protocol. Bound probes were detected with a rabbit anti-biotin antibody (Bethyl Cat. A150-109A, dilution 1∶1500) and with donkey anti-rabbit Dylight488 (Jackson, Cat. 711-486-152, dilution 1∶1000). Slides were mounted with Vectashield Hard-Set Mounting Medium with DAPI (Vector Laboratories, H-1500). The CD31-specific fast red color precipitate was detected in the red and the human DNA-specific ALU-repeat in situ hybridization in the green fluorescence channel, respectively.
Immunohistochemistry on BondMax (Leica) with Refine-DAB-Kit including counterstain. Epitop-retrieval with Bond Heat-pre-treatment-buffer 2 for 30 min. Primary antibody Chicken anti-GFP (Abcam ab13970, 1∶2000), secondary antibody Rabbit anti-Chicken Y (Abcam ab6753, 1∶300).
Confocal image acquisition was done using sequential mode on a Leica TCS SP5 confocal laser scanning microscope (Mannheim, Germany). Co-localization studies were conducted using IMARIS software (Bitplane, Zurich, Switzerland).
Freshly isolated melanoma cells were injected into nude athymic mice. Patient tumour xenografts were harvested and cells dissociated as described above. Some of those cells were cultured in neuro-sphere condition and passed 8 times. At the 8th passage the cells were infected with a GFP driving virus.
GFP-expressing lentivirus was produced by transient four-plasmids cotransfection into HEK 293
CD34 and D2–40 expressions are presented as median with range and compared between different patient groups using the Mann-Whitney and the Krustal-Wallis test. Glut-1 expression was compared using Fisher's exact test. Differences between CD34 and D2–40 expressions were compared using the Wilcoxon signed rank test. P-values below 0.05 were considered as significant. SPSS 16.0 for Windows (SPSS Inc., Chicago, IL) was used for statistical analyses.
We investigated 127 melanoma metastases (21 lymph nodes and 106 other sites) and 49 primary melanomas for vascularisation. The number of vessels was counted per core. There were significantly more D2–40 positive lymphatic vessels in primary melanoma than in melanoma metastases (p<0.0001;
(A) Box plot comparison of lymphatic vessels (number of D2–40 positive vessels per core) in primary versus metastatic melanoma. * One case, * two cases. (B) Absence of D2–40 staining within metastatic tissue showing no lymphatic vessels (×30). (C) D2–40 positivity shows dilated lymphatic vessels present at the tumor/lymphatic parenchyma interface (arrow, ×150). (D) CD34 staining shows important blood micro-neovascularisation within metastases. (×50). (E) CD34-positive blood microvessels (arrowhead) associated with CD34–negative cells showing vascular mimicry (arrow, ×400). (F) and (G) Cells of vascular mimicries embedded in a PAS (F) and collagen IV (G) positive laminina (arrow, ×320)).
Primary melanomas | Metastases of melanoma | P-value | |
Sample number | 49 | 127 | |
Lymphatic vessels (D2–40) | 4±4 | 0.6±2 | <0.0001 |
Blood vessels (CD34) | 20±14 | 15±14 | 0.01 |
± Mean ± SD.
There were significantly more blood than lymphatic microvessels in primary melanomas (p<0.0001) as well as in metastases (p<0.0001). In addition primary melanomas had significantly more blood vessels than metastases (p = 0.012). Primary melanomas had an average of 20 blood vessels per core (median 19; range 1–60) and metastases had an average of 15 blood vessels (median 11; range 0–69) per core (
We investigated six xenografts of human melanoma cell lines labelled with p75 and sorted by FACS as described in material and methods for MECA32, a marker specific for mouse vessel endothelia. There was significant neovascularization within the tumor (
(A) Mouse vessel-specific MECA32 staining shows micro-neovascularisation within melanoma xenografts (arrowhead, ×100) and positive vessels of the adjacent soft tissue (arrow). (B) Vascular mimicries negative for MECA32 (arrow) associated with and connecting MECA32-positive vessels (arrowhead, ×175). (C) Interphase FISH for human chromosome 17 and HER-2 combined with immunofluorescence staining for the mouse-specific vessel marker MECA32: Tumor capillaries with endothelial cells positive for MECA32 (yellow) and negative for HER-2 (arrow) as well as endothelial-like cell with negativity for MECA32 and positivity for human chromosome 17 and HER-2 (arrowhead, green; ×500). (D) and (G) Melanoma tumor cell-derived endothelial-like cells: (D) Capillary endothelial-like cell positive for human chromosome 17 and HER-2 (arrow) and positive internal control of tumor cells (arrowhead, ×500. (E) and (F) Negative internal controls of slide D and G. Melanoma xenograft adjacent soft tissue with mouse capillaries and fibroblasts negative for human chromosome 17_HER-2 probe (E, ×500) and negative for FISH y probe and CD31 (F, ×700). (G) Interphase FISH for human y-probe combined with immunofluorescence staining for the human-specific vessel marker CD31: Vessel endothelial-like cell positive for the Y chromosome (red) and CD31 (arrowhead, green; ×1000). (H) Vascular mimicries (arrow): Endothelial-like cells positive for human chromosome 17 and HER-2 (arrowhead, ×300). (K) and (L) Green fluorescent protein labelled melanoma cell lines: Melanoma cells (arrowhead, green) and endothelial cells (arrow) contributing to complex branching neo-vascularisation. (K ×200; L ×500).
Melanoma derived endothelial-like cells of tumor capillaries with nuclear positivity for the human-DNA specific ALU-repeat sequences (A, D arrow, green, ×700) and cytoplasmic positivity for CD31 (A, B, arrow, red, ×700). Mouse derived endothelial cells with negativity for the human-DNA specific ALU-repeat sequences (D arrow head) and cytoplasmic positivity for CD31 (B arrow head, red). Internal control with nuclear positivity for the human-DNA specific ALU-repeat sequences A, D) and negativity for CD31 in melanoma cells (A, B, asterix).
Method | Result | Shown in figure |
Interphase FISH for human chromosome 17 and HER-2 combined with immunofluorescence staining for the mouse-specific vessel marker MECA32: | Capillary lining cells (endothelial-like cells) negative for MECA32 and positive for human chromosome 17 and HER-2 | 2C |
Chromogenic |
6% of all capillary lining cells (endothelial-like cells) are positive | 2D |
Interphase FISH for human y-probe combined with immunofluorescence staining for the human-specific vessel marker CD31 | Endothelial-like cells positive for the Y chromosome and CD31 | 2G |
Interphase FISH for the human DNA specific ALU-repeat combined with immunofluorescence staining for CD31 | Endothelial-like cells positive for the human-DNA specific ALU-repeat sequences and CD31 | 3A–D |
Vascular mimicries (VM) were observed in both human metastases (
(A) and (B) Vascular mimicries (arrow): (A) Melanoma cells with endothelial-like morphology, condensed nuclei, minimal Masson Fontana positive (inset) pigmentation (H&E, ×500), (B) mostly negative for PanMelanoma (red) (×150).
In contrast to their endothelial-like morphology, they were negative for CD31, CD34 and D2–40. In xenografts tumor cell VM were associated with MECA32-positive mouse derived microvessels (
The establishment of metastases is a complex process, which includes angiogenic development. Several studies have shown that increased lymphatic density in primary melanomas is associated with a higher incidence of regional lymph node metastases
In contrast to the lymphangiogenesis seen in primary lesions, there were few lymphatic vessels observed within lymph node and organ metastases. The few lymphatic vessels identified were mostly found in foci at the metastatic lymph node interphase, or in the lymph node parenchyma close to metastases, and these were often angiectatic (
However, we observed a significant anatomical difference in lymphatic density with a mean of 8 lymphatic vessels in normal skin 0,5 cm adjacent to the primary melanoma and absence in lymph node tissue 0,5 cm adjacent to the metastases. However, despite those anatomical differences, there still remained a significant difference in lymphatic vascularization between primary melanoma and adjacent normal tissue indicating the lymphatic induction potential of primary melanoma, which was almost absent in metastases. Lymphatic induction in primary melanoma and its prognostic relevance is well documented. In addition lymphatic induction is also a part of regression in primary melanoma
Melanoma metastases presented a highly complex neovascularization with immunoreactivity for CD34 and negativity for D2–40. Furthermore, the highly significant inverse correlation between expression of Glut-1 (implying local hypoxia) and the presence of micro vessels suggests that hypoxia is a major driver (p<0.0001;
Similar blood neovascularization was found in xenografts of human melanoma cell cultures. These vessels were mostly positive for the mouse-specific vessel marker MECA32. However, about 6% of endothelial-like cells were negative for this marker. We propose that these MECA32-negative cells are tumor-derived because they are positive for the human chromosome 17 and HER-2 (
Cells involved in VM express genes associated with embryonic stem cells including those associated with primordial vascular development
Approximately one million tumor cells are shed per g of tumor per day
In summary, we found a complex neovascularization consisting of VMs and leaky blood micro vessels in human melanoma biopsies and melanoma xenografts. In addition, melanoma cells contributed to tumor capillaries.
We are grateful to M. Storz for TMA construction, N. Wey for photographic and computer-assisted reproductions, Dieter Zimmermann for supporting the method part and Nadine Mihic for reviewing the manuscript.