Differential effects of tumor–platelet interaction in vitro and in vivo in glioblastoma
- First Online:
- Cite this article as:
- Brockmann, M.A., Bender, B., Plaxina, E. et al. J Neurooncol (2011) 105: 45. doi:10.1007/s11060-011-0560-2
- 166 Views
An elevated platelet count is considered an independent predictor of short survival in glioblastoma and various other tumor entities. Prothrombotic activity of the tumor microcirculation resulting in platelet activation and release of cytokines from activated platelets has been suggested to play a role. This study was designed to analyze the effects of platelet-released cytokines on glioblastoma and endothelial cell proliferation and migration in vitro, and the influence of platelet count on glioblastoma growth and angiogenesis in vivo. In cultured human glioblastoma, umbilical cord and cerebral microvascular endothelial cells platelet-released cytokines significantly stimulated proliferation and migration as well as sprouting and formation of capillary-like structures. In vivo, glioblastoma cells were implanted in mice followed by platelet depletion starting 1 or 8 days later. Tumor volume, proliferative index, and vessel density analyzed 14 days after engraftment did not differ between animals with a normal and a low platelet count. Likewise, no effect of platelet depletion over 20 days upon the volume of intracerebrally growing tumors was observed in mice. Additionally, proliferative activity and vessel density determined in tumor samples from patients operated upon glioblastoma did not show any correlation with the patients’ preoperative platelet count. Thus, we conclude that distinct proliferation- and chemotaxis-stimulating effects of platelet-derived cytokines can be achieved in vitro, while the platelet count does not exert a major influence on tumor growth and tumor angiogenesis in GBM in vivo.
Platelets serve a couple of important functions including the control of vessel wall integrity and regulation of hemostasis. Upon activation, platelets undergo a shape change with local adherence, aggregation, and eventually release of a cocktail of factors which orchestrate the activation of further platelets, blood coagulation, and thrombus formation. Activated platelets may release more than 300 peptides and proteins from different storage sites including alpha-granules, dense granules, and lysosomes . These platelet-derived factors affect the hemostatic processes, as well as vasomotor, inflammatory, proliferative, and angiogenic activity. Cytokines exerting either pro-angiogenic effects, such as vascular endothelial growth factor (VEGF) [2, 3], platelet-derived growth factor (PDGF) , basic fibroblast growth factor (bFGF) , endothelial cell growth factor (ECGF) , transforming growth factor (TGF) , insulin-like growth factor ILGF [8, 9], angiopoietin-1 , sphingosine-1-phosphate [11, 12], and matrix metalloproteinases (MMPs) [13–15], or antiangiogenic activity, such as thrombospondin I , platelet factor 4 (PF4) , plasminogen activator inhibitor I (PAI) , and angiostatin , have been identified to be among the released cytokines. In most instances, the pro-angiogenic effects may be prevailing. Accordingly, preparations of platelet releasate have been used for treating chronic foot ulceration, a widespread complication of diabetic neuropathy .
The pro-angiogenic activity of the cocktail of cytokines released by activated platelets has been suggested to play a role in tumor growth and tumor angiogenesis . Thus, a negative relationship between the platelet count and survival time has been described for a large variety of malignancies [22–31]. Furthermore, it has recently been reported that an increased preoperative platelet count is significantly correlated with shorter survival in patients with newly diagnosed glioblastoma multiforme (GBM)  and that platelet counts in patients with GBM rise within the months and years before diagnosis of GBM . However, the pathophysiological mechanisms underlying this association are not yet clear. The present investigation was designed to investigate (1) the influence of platelet-secreted cytokines on growth and migration of glioblastoma and endothelial cells in vitro, and (2) the relevance of platelet count on glioblastoma growth and angiogenesis in vivo.
Materials and methods
Cell isolation and culture
Human umbilical cord endothelial cells (HUVECs) were obtained from umbilical cords provided by the local Department of Obstetrics. Isolation and culturing was performed as described previously . Cerebral microvascular endothelial cells (CMECs) were isolated and grown as described by Lamszus et al. . Glioblastoma cell lines (U87, U373, and G55T2) were cultured in Dulbecco’s modified Eagle’s medium as described recently [36, 37]. All cell cultures were maintained in a humidified atmosphere of 5% (v/v) CO2 in air at 37°C.
Preparation of washed platelet suspension
Venous blood samples were taken from a healthy volunteer. After discarding the first 10 ml, the blood (seven parts) was anticoagulated with one part of acid citrate dextrose (ACD; composition: 0.8% w/v citric acid, 2.2% w/v sodium citrate, and 2.45% w/v glucose). All samples were properly mixed with the anticoagulant. An additional blood sample was anticoagulated with K-EDTA for complete blood cell counts using an automated cell analyzer (CellDyn 3500; Abbott, IL, USA).
Preparation of platelet-derived cytokine solution
To prepare platelet-rich plasma (PRP), ACD-anticoagulated blood was centrifuged for 15 min at 230g (brakes turned off). The PRP was collected in plastic tubes and centrifuged again (300g, 15 min). The supernatant was discarded, the pellet resuspended using a stabilizing washing solution (120 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 5 mM glucose, 500 μg/ml bovine albumin, 50 μg/ml apyrase, 30 mM Na-citrate) and centrifuged again at 300g (15 min). The pellet was resuspended in PBS and the platelet count measured (CellDyn 3500). The platelets were diluted with PBS to yield a platelet count of 600 or 300/nl for enzyme-linked immunosorbent assay (ELISA) measurements. Finally, platelets were activated by addition of 10 μM thrombin receptor agonist peptide-6 (TRAP-6; Bachem, Weil am Rhein, Germany) for 15 min at room temperature (RT). After centrifugation (2,000g, 10 min) the supernatant containing platelet-derived cytokines [termed cytokine-rich (CR)-PBS] was used for measurements of individual cytokine concentrations and proliferation and migration assays at the described dilutions.
We determined the concentrations of bFGF, hepatocyte growth factor (HGF), PAI-1, PDGF-AA, PDGF-BB, TGF-α, and VEGF secreted from human platelets using ELISA kits (Quantikine Immunoassays; R&D Systems, Wiesbaden, Germany). All tests were performed according to the manufacturer’s instructions. Platelet count before activation was 300/nl. All experiments were done at least in duplicate on appropriately diluted samples.
Cell proliferation assays
Glioblastoma and endothelial cells were seeded into 96-well plates (2,000 cells/well; for endothelial cells wells were collagen-coated) and cultured overnight. On day 1, cells were washed with PBS followed by incubation in medium containing CR-PBS at different dilutions. Incubation media were renewed on day 4. Serum-reduced medium (0.5% Fetal Calf Serum (FCS) for glioblastoma cell lines, 2% for HUVECs, 10% for CMECs) was used as negative control, while medium containing 10% FCS (HUVECs) or 20% (CMECs) was used as positive control. In HUVECs, additional control experiments were performed with medium containing 2% FCS and TRAP-6 at concentrations used for platelet activation.
Wells were fixed with 1% glutaraldehyde in quadruplicates at daily intervals up to 8 days after seeding. The fixed cells were stained with crystal violet, washed with PBS, and solubilized in 10% Sodium Dodecyl Sulfate (SDS). The absorbance of the lysate was determined photometrically at 540 nm wave length (Titertek Multiskan Plus; Labsystems, Helsinki, Finland).
Proliferation of CMECs in the presence of platelet cytokines was assayed using a bromdesoxyuridine (BrdU)-based ELISA kit (Roche, Mannheim, Germany). Briefly, 20,000 CMECs were seeded in 96-well plates and incubated for 24 h. Thereafter, plain medium was replaced by medium containing different concentrations of platelet releasate. Cells were cultured again for 48 h and the BrdU-assay was carried out according to the manufacturer’s instructions. Values were assessed in quadruplicate.
Modified Boyden chamber migration assay
Assay medium (serum-free M199 medium with 0.1% bovine serum albumin) containing CR-PBS in appropriate dilutions was added to the lower wells of a 96-well modified Boyden chamber (Neuroprobe, Cabin John, MD, USA). Wells were covered with a Nucleopore filter (pore size, 8-μm; Neuroprobe) coated with Vitrogen 100 (Cohesion, Palo Alto, CA, USA). 1.5 × 104 cells suspended in 50 μl of assay medium were seeded into the upper wells. After incubation for 6 h, non-migrated cells were scraped off from the upper side of the filter followed by staining with Diff-Quick (Dade, Unterschleissheim, Germany). Nuclei of migrated cells were counted in 4 high-power fields (hpfs, 340 × 270 μm) using a ×40 objective with a calibrated ocular grid. Values were assessed in triplicate or quadruplicate. For checkerboard analyses, CR-PBS was added to either the upper or the lower wells, or both.
In vitro angiogenesis assay
The tube formation assay was carried out according to the manufacturer’s instructions (Chemicon International, Hampshire, UK) with slight modifications. Endothelial cells (1 × 104) suspended in 100 μl endothelial cell growth medium (ECGM; PAA Laboratories) containing 2% FCS were seeded onto the polymerized matrix gel. The cells were incubated for 2 h to let them attach. Thereafter, 100 μl of ECGM containing serum-free medium (control) or different dilutions of CR-PBS were added to the wells. Photographs of all wells were taken using an inverted light microscope after 5, 12, and 18 h to count the numbers of closed polygons (one field at fourfold magnification) and of branching points (5 fields at tenfold magnification).
Female nude mice (NMRI-nu/nu, 6–8 weeks of age) were used. Institutional guidelines for animal welfare and experimental conduct were followed.
Heterotopic glioblastoma model
G55T2 glioblastoma cells (7 × 105) dissolved in 100 μl PBS were injected subcutaneously into each quadrant of the back in anaesthetized mice. In a first series of experiments, platelet depletion was initiated in 5 mice on day 1 after tumor cell engraftment by intraperitoneal injection of a mixture of purified rat monoclonal antibodies directed against mouse GPIb alpha [3 μg/g body weight (bw); anti-GPIb alpha; R300; Emfret Analytics, Wuerzburg, Germany]. To maintain thrombocytopenia, injections (1.5 μg/g bw) were repeated every third day until the end of the experiment. Control mice were either treated with unspecific non-immune rat anti-mouse antibodies (n = 5; C301; Emfret Analytics) or saline only (n = 5). In a second series of experiments, platelet depletion was initiated on day 8 after tumor engraftment and maintained as described above. In both experiments, mice were killed 14 days after cell implantation and the tumors excised for caliper-based measurement of size and subsequent immunohistochemical analysis. Tumor volume (V) was calculated using the formula described by Hudd et al.  with V = (width2 × length)/2.
Orthotopic glioblastoma model
In this series, 3 × 105 U87 glioblastoma cells dissolved in 3 μl PBS were stereotactically injected into the caudate nucleus of nude mice as described recently . One group of mice (n = 7) was platelet depleted over 20 days as described above, while a second group was treated by repeated injection of saline only (n = 8). After 20 days of treatment, contrast-enhanced MRI was performed followed by three-dimensional volumetric analysis as described recently .
Determination of microvessel density and proliferative activity in the heterotopic xenografts
Cryosections (10 μm) were fixed in acetone and blocked with 0.5% casein in PBS. Sections were incubated with a rat monoclonal antibody against PECAM-1 (1:50; BD Pharmingen, Heidelberg, Germany) for 2 h at RT. Bound antibody was detected using the Vectastain kit (Vector Laboratories, Burlingame, CA, USA). The numbers of stained vessels were counted in 5 hpfs that were either selected in the most densely vascularized “hot spot” areas or in randomly chosen regions located approximately in the center of the tumor, avoiding necrotic areas.
To analyze the proliferative activity of the tumor cells, sections were stained with a rat anti Ki-67 monoclonal antibody raised against the Ki-67 antigen (1:50; DAKO, Hamburg, Germany). Visualization was achieved using the Vectastain kit. The percentage of MIB-1-positive nuclei was determined by counting immunoreactive tumor cell nuclei in 5 randomly selected hpfs, as well as in 5 hpfs in the most actively proliferating tumor area (“hot spot”).
Patient data and tumor tissue samples
Neuropathologically confirmed glioblastoma specimens were obtained from 47 patients operated upon de novo GBM at the local Department of Neurosurgery. All preoperative hematological analyses including platelet counts were taken within 14 days prior to tumor resection. None of the patients suffered from myeloproliferative disorders, acute inflammatory disease, or status post splenectomy.
Immunohistochemistry for CD31 and Ki-67 in GBM specimens
Sections (thickness, 3 μm) were cut from formalin-fixed paraffin-embedded tissue blocks. Antigen retrieval was achieved by microwave treatment. The primary antibodies used were rabbit polyclonal antiserum against von Willebrand factor (vWF) and murine monoclonal anti-Ki-67 (both DAKO). Bound antibody was detected using the Vectastain kit. Vessel density and proliferative indices were determined as described above. Proliferative index was determined in 45 of the 47 tumor specimens, as immunohistochemistry failed in specimens from two patients.
Differences between platelet-depleted and control mice, as well as differences in the in vitro experiments, were analyzed using the unpaired t test or Mann–Whitney rank-sum test. P values ≤0.05 were considered statistically significant. Values are given as mean ± SD unless indicated otherwise. The relationship between vessel density, proliferative index, and platelet count was analyzed using the Pearson correlation coefficient.
Measurement of cytokine concentrations (released from activated thrombocytes) revealed the highest values for PDGF-AA (1,954 ± 44 pg/ml) and PDGF-BB (1,148 ± 65 pg/ml) while HGF (322 ± 152 pg/ml) and VEGF (146.4 ± 3.7 pg/ml) displayed moderate levels. In addition, TGF-α was barely detectable (8.3 ± 0.5 pg/ml), while bFGF was even below the detection limit.
Proliferation of tumor and endothelial cells
In U373MG cells, proliferation increased significantly during incubation with CR-PBS at concentrations equivalent to 100 or 1,000 platelets per initially seeded tumor cell (both P < 0.05; Fig. 1b). In additional experiments, TRAP-6 used for platelet activation did not by itself affect proliferation of both glioblastoma cell lines (results not shown).
The HUVECs reacted much more sensitive to platelet-released cytokines. From day 2 onward, proliferation of HUVECs was significantly stimulated at all tested concentrations of platelet-released cytokines (P < 0.05; Fig. 1c). Again, TRAP-6 did not by itself affect proliferation (results not shown).
In CMECs, proliferation was enhanced over controls after 2 days of stimulation with platelet releasate (Fig. 1d). Similar to HUVECs, concentrations of CR-PBS equivalent to or above 100 platelets per initially seeded tumor cell induced a significant increase of proliferation (P < 0.05).
Stimulation of endothelial and glioma cell chemotaxis
A dilution-related stimulation of chemotactic movement also occurred in U373MG cells (Fig. 2b). However, the maximum response did not reach the level observed in the respective positive controls.
Platelet releasate also increased chemotactic migration of HUVECs which appeared to react even stronger than the glioma cell lines (Fig. 2c). In the presence of CR-PBS in a dilution as low as 1:200, migration of HUVECs reached a level comparable to that of the positive controls (containing 10% FCS). At lower dilutions, HUVEC migration was almost doubled compared to the positive control.
In CMECs, stimulation of chemotaxis by CR-PBS appeared to be of a similar magnitude as seen in HUVECs. Again, a dilution of CR-PBS as low as 1:100 resulted in significant stimulation (P < 0.05). However, the maximum level of chemotactic stimulation did not exceed that of the positive controls (Fig. 2d).
Checker-board assays carried out to distinguish between chemotaxis (directed movement) and chemokinesis (random movement) did not reveal significant chemokinetic effects elicited by CR-PBS (results not shown). Furthermore, TRAP-6 used for platelet activation did not by itself affect chemotaxis of glioma or endothelial cells (not shown).
Stimulation of capillary-like structure formation by platelet derived cytokines
Tumor growth in experimental in vivo models in platelet depleted mice
We also employed an orthotopic model of glioblastoma growth. Three weeks after cell implantation, tumor volumes as determined by MRI were 173.9 ± 23.9 mm3 (mean ± SEM; n = 8) in saline-treated animals and 153.6 ± 30.2 mm3 (n = 7) in platelet-depleted mice (P > 0.05). These results underscore the observations made in the heterotopic model suggesting that platelet depletion does not affect glioblastoma growth.
Tumor microvessel density and cell proliferation index in the heterotopic in vivo model
In platelet-depleted mice, the intratumoral mean vessel densities (MVD) was not significantly different from that in the control groups treated with either saline or an unspecific polyclonal antibody (Fig. 4b). Moreover, the vessel density in the “hot spot” areas of the tumors did not significantly differ from the control groups (Fig. 4c).
Histopathological correlations in human GBM specimens
Platelet–endothelium interactions have been suggested to affect tumor growth and angiogenesis in vivo [21, 40, 41]. While for many of the cytokines released by platelets evidence has been presented in favor of either enhancing or inhibiting tumor and endothelial cell proliferation and migration , observations on the effect of individual cytokines do not necessarily predict the overall effect of the cocktail of platelet-released cytokines. We therefore initially tried to address this topic by focusing on cell proliferation and chemotactic movement of endothelial and glioma cells in vitro.
In our initially conducted experiments, we thought to analyze tumor–platelet interaction by incubation of platelet-rich plasma with GBM-cells. This resulted in reduced growth of tumor cells in vitro. We did not find the exact reason for this, but we speculate that the plasma itself or the “detritus” of the activated platelets within the cell culture over several days led to this observation. For this reason, and because the cytokines released from platelets have been suggested to be the major trigger for tumor angiogenesis, we decided to analyze the effects of the cytokines released from washed and subsequently activated platelets. We found the platelet releasate to strongly increase proliferation and migration of two human GBM cell lines, U87 and U373, and of HUVECs and CMECs. The receptor status of these cell lines has been analyzed before [36, 42], and the cells are known to express many of the receptors necessary to interact with cytokines released by platelets. Since endothelial cells displaying increased levels of proliferation and migration may form new capillary structures as outlined previously , it is not surprising that sprouting activity and the formation of tube-like structures were also significantly stimulated by the platelet releasate. These observations relate to HUVECs, a widely used model for the study of endothelial cell function as well as to CMECs, the host vasculature for a naturally developing GBM. Although the results of our in vitro assays point to a marked pro-angiogenic action of the platelet releasate, one cannot rule out that the doses used in our in vitro assays are higher than under physiological conditions. However, it would be extremely difficult to determine the true local concentrations of cytokines released from platelets within the tumor micro-vasculature, and how much of the secreted cytokine cocktail is simply washed out of the tumor microcirculation without having any effect upon endothelial or even tumor cells.
Based upon these considerations, we performed several series of in vivo experiments to test the effect of platelet depletion on tumor growth and angiogenesis. In a heterotopic xenotransplantation model in mice, platelet depletion was elicited and maintained using a monoclonal rat anti-mouse GPIb alpha antibody that has previously been described to result in an almost immediate, profound and irreversible Fc-independent platelet depletion [40, 43, 44]. Interestingly, despite the efficient platelet depletion, tumor growth did not differ from control animals treated with saline or with unspecific antibodies. This result was consistently obtained irrespective of whether the platelet depletion was initiated before or after the tumor and the tumor vasculature had established (i.e., mice were platelet depleted on day 1 or day 8 after injection of tumor cells). Virtually identical results were observed in an orthotopic GBM model using intracerebral injection of U87 cells and MRI analysis after 3 weeks of tumor growth. The lack of an anti-tumor activity by platelet depletion was corroborated in the subcutaneously grown tumors by immunohistochemical analyses. Platelet depletion completely lacked anti-proliferative and anti-angiogenic activity as evidenced from cell and vessel density counts in randomly chosen and in hot-spot areas. The negative findings of the in vivo experiments are supported by findings of other groups: Manegold and coworkers observed only a moderate increase of platelet rolling but not of adherence in the developing Lewis lung carcinoma using the mouse skinfold chamber assay .
Since the lack of an effect of platelet depletion on tumor proliferation and new vessel formation in vivo is in contrast to the results obtained in cultured cells, several explanations might be conceivable to explain this finding. For example, platelet activation within the GBM microvasculature simply may not be strong enough or may result in the release of a platelet cocktail somewhat different from that used in our in vitro experiments. One may also speculate about a wash-out of platelet-released cytokines by the blood stream, thus preventing the generation of a sufficiently high concentration of cytokines in the tumor microcirculation.
In addition to the studies in platelet-depleted mice, we determined cell proliferation and tumor MVD in samples from patients operated on GBM using the same methodology as employed in the mouse experiments. Interestingly, we did not find any significant correlation between the pre-operative number of platelets and tumor proliferation or vessel density, respectively. These observations are in full agreement with the results of our animal experiments. Since the issue of the potential impact of platelets on GBM growth has recently been brought up by the observation that preoperative thrombocytosis is a predictor of poor outcome in GBM patients , our present results suggest much more that thrombocytosis is not the cause but rather the consequence of the growing tumors secreting thrombopoieitic factors like interleukin (IL)-6, IL-1, VEGF, macrophage and granulocyte–macrophage colony stimulating factor (M-CSF, GM-CSF), and tumor necrosis factor α (TNF-α) [46–49]. This explanation would also fit with our recent findings that platelet counts increase from normal to elevated levels in patients with GBM .
To conclude, the present results suggest that, although the cocktail of platelet-released cytokines exerts proliferative and chemotactic effects on glioblastoma and endothelial cells in vitro, tumor growth and tumor angiogenesis appear not to be affected by platelet count in a functionally relevant manner in vivo. Therefore, the frequently observed correlation between increased platelet counts and reduced survival time in patients with malignant tumors is most likely to result from a tumor-induced increase of platelet counts, but not vice versa.
The protocol for the platelet-stabilizing washing solution was kindly provided by Dr. A. Ruf, Zentrum f. Labormedizin, Mikrobiologie und Transfusionsmedizin, Klinikum Karlsruhe, Germany. This work was supported by a fellowship granted to Elena Plaxina by the German Academic Exchange Service (DAAD). The authors have no conflicting financial interests.