Purpose: One of the key factors that promotes angiogenesis is vascular endothelial growth factor (VEGF). Platelets are the main source of VEGF in blood and contribute to angiogenesis by release of growth factors, including VEGF, from their α-granules on activation. The monoclonal antibody bevacizumab blocks VEGF in the blood of patients within hours after administration. Platelets are known to endocytose plasma proteins including immunoglobulins. We tested the hypothesis that platelets take up bevacizumab.

Experimental Design: Fluorescence-activated cell sorting analysis, immunofluorescence imaging, and Western blotting were used to study uptake and release of bevacizumab by platelets in vitro and in vivo. The angiogenic activity of platelets preincubated with bevacizumab was studied in endothelial proliferation assays. Finally, we determined whether treatment with bevacizumab neutralizes VEGF in platelets from cancer patients.

Results: We found that platelets are able to take up bevacizumab. Activation of platelets preincubated with bevacizumab resulted in release of the antibody and release of VEGF neutralized by bevacizumab. Immunofluorescence microscopy revealed that FITC-labeled bevacizumab and P-selectin colocalize, indicating α-granule localization. In addition, bevacizumab uptake inhibited platelet-induced human endothelial cell proliferation. In in vivo rabbit experiments, FITC-labeled bevacizumab was present in platelets after 2 h and up to 2 weeks following i.v. administration. Finally, we found that platelets take up bevacizumab in patients receiving bevacizumab treatment. Within 8 h after bevacizumab administration, platelet VEGF was almost completely neutralized due to this uptake.

Conclusion: These studies show that bevacizumab is taken up by platelets and may explain its clinical effect on wound healing and tumor growth.

Vascular endothelial growth factor (VEGF) is a key factor to promote angiogenesis (1). Platelets contribute to angiogenesis by release of VEGF and other factors (2–5). Platelets, which are important for hemostasis and wound healing (6), also play a role in metastasis formation and tumor-induced angiogenesis (4, 7–10). Megakaryocytes and presumably platelets contain and endocytose plasma proteins in their α-granules, including VEGF and immunoglobulins (11–13). Immunoglobulins, in particular humanized monoclonal antibodies, are used to treat a variety of diseases including hematologic and solid malignancies (14, 15). Bevacizumab is a monoclonal antibody that blocks VEGF and has clinical activity in a variety of cancer types (16). Ligand-antibody interaction for bevacizumab has been closely followed in patients (17, 18). Within hours after administration of bevacizumab, >97% of free serum VEGF is bound by the antibody. We previously reported that serum VEGF is predominantly derived from platelets and only small amounts of VEGF are present in plasma (19). Based on these findings, we hypothesized that platelets take up bevacizumab from the circulation. To address this hypothesis, we studied platelet uptake of labeled bevacizumab. We found that bevacizumab is taken up by platelets and blocks activity of platelet VEGF on human umbilical vein endothelial cells (HUVEC). In platelets from patients treated with bevacizumab, our preclinical results were supported. Within hours after start of treatment, bevacizumab neutralizes platelet VEGF. Taken together, our findings provide a possible explanation for the clinical observed effects of bevacizumab.

Labeling of bevacizumab and immunoglobulin G. Bevacizumab is an immunoglobulin G (IgG) with the backbone of the IgG1 isotype. As a proper control IgG, the IgG1 fraction from human immunoglobulin for i.v. use (Gamunex) was enriched by a previously described method with an agarose protein A column that was eluted at different pH (20). Subsequently, bevacizumab and IgG (the IgG1 fraction) were labeled in parallel using exactly the same method. According to standard procedures provided by the manufacturer, bevacizumab and IgG were labeled with either biotin (0.2 mg/mL) or FITC (80 μg/mL; both from Sigma) and checked for comparable labeling efficiency.

Platelet isolation and incubation with labeled antibodies. Platelets and platelet-rich plasma were obtained as previously described (21). In brief, blood was taken from healthy donors or patients and was anticoagulated with 1/10 volume of 0.13 mol/L sodium citrate. Platelet-rich plasma was prepared by centrifugation at 156 × g for 20 min at 22°C. Platelet-rich plasma was isolated and recentrifuged at 330 × g for 15 min. Platelets were washed and, at the second step, resuspended with glucose HEPES-Tyrodes buffer (0.145 mol/L NaCl, 5 mmol/L KCl, 0.5 mmol/L Na2HPO4, 1 mmol/L MgSO4, 10 mmol/L HEPES, 5 mmol/L glucose, pH 6.5 to prevent activation, final pH 7.3 as physiologic pH) after careful resuspension of the platelet pellet. Washing was done in the presence of prostaglandin I2 (Sigma; end concentration, 10 ng/mL) as a means for transient protection against platelet activation. Subsequently, 200,000 platelets/μL were incubated with biotin-labeled or FITC-labeled bevacizumab or IgG. Flow cytometric [fluorescence-activated cell sorting (FACS)] analysis was done using Beckman Coulter FACS. For immunofluorescence microscopy, platelets were permeabilized after fixation with 0.1% Triton X-100, and α-granule staining of resting platelets was done with a fluorescence-labeled Alexa 647 monoclonal antihuman P-selectin antibody (Biolegend). Confocal microscopy was done using Ultraview LCI (Perkin-Elmer) equipped with Spinning Nipkow disc with microlenses and Kr/Ar laser with lines at 488 and 647 nm. Images were grabbed using a LSI-cooled 12-bit charge-coupled device camera and processed using the NIH ImageJ software.5

5

http://rsb.info.nih.gov/ij/download.html

Platelet release of bevacizumab. Platelets were incubated as described above. Platelets were isolated by centrifugation after incubation and addition of prostaglandin I2 to prevent activation. After centrifugation with one washing step, platelets were resuspended and activated for 20 min at 37°C with thrombin (5 IU/mL). Subsequently, these samples were centrifuged or directly centrifuged to obtain resting platelets. Fluorescence of supernatant and platelet pellets was measured with an automatic fluorescence meter. The average of three independent experiments is shown.

Western blot of biotinylated bevacizumab and fibronectin. For Western blotting, platelets were lysed in Triton X-100 lysis buffer [20 mmol/L Tris-HCl (pH 8.0), 1% Triton-X-100, 140 mmol/L NaCl, 10% glycerol, 0.005% bromophenol blue, 1% DTT]. Triton X-100–insoluble cell remnants containing the cytoskeletal fractions were removed by centrifugation (5 min at 16,100 × g) and cleared lysates were stored at −20°C. The Triton X-100–soluble samples contained the α-granule content and were analyzed by SDS-PAGE followed by Western blotting. Antibodies were diluted in PBS containing 5% nonfat dried milk and 0.1% Tween 20. Streptavidin-horseradish peroxidase (1:1,000) was used to detect the biotinylated bevacizumab and IgG. For fibronectin, a mouse monoclonal antibody against fibronectin was used (1:500; FN30-8, TaKaRa Biomedicals) and detected with rabbit anti-mouse horseradish peroxidase (1:10,000; Pierce) as a secondary antibody. Enhanced chemiluminescence (Perkin-Elmer Life Sciences) was used for detection.

Endothelial cell proliferation. HUVECs were used in a standard proliferation assay (22). In brief, 1 × 104 HUVECs per well on a 24-well plate were plated and incubated overnight in 0.5% fetal bovine serum–containing endothelial cell basal medium (starvation). Cells were then incubated with endothelial cell basal medium containing the released contents of activated (by thrombin 0.5 IU/mL, 20 min at 37°C) platelets preincubated with bevacizumab, IgG, or no immunoglobulin. After 72 h, cells were trypsinized and counted in quadruplicates on a Coulter Z1 cell counter (Coulter Electronics).

VEGF ELISA. VEGF measurements were done with a standard VEGF-ELISA-kit (R&D Systems).

In vivo experiments. Platelet-rich plasma was obtained from New Zealand White female rabbits (3.5-4 kg; Myrtle Rabbitry SC) before bevacizumab infusion and at multiple time points thereafter. Platelet-rich plasma was analyzed by FACS and 1 mL of platelet-rich plasma was centrifuged at 330 × g for 15 min to obtain platelet-poor plasma by sampling supernatant. Fluorescence of 100-μL platelet-poor plasma was determined.

Platelet-rich plasma and platelet-poor plasma of patients. After informed consent was obtained from seven patients with advanced cancer (five colorectal, two kidney cancer), blood was drawn before and after bevacizumab administration in citrate buffer containing tubes. These patients received every 2 weeks bevacizumab (10 mg/kg) for metastatic renal cell carcinoma or in combination with standard chemotherapy for colorectal cancer (5-fluorouracil, leucovorin, and oxaliplatin). Platelets and platelet-poor plasma were isolated as described above and both platelet isolates as well as platelet-poor plasma were stored at −80°C. Platelet counts varied from 150,000 to 400,000 platelets/μL. According to the manufacturer's guidelines (R&D Systems), ELISA for VEGF was done in these samples after lysing the platelet isolates and platelet-poor plasma samples with standard lysate buffer (Pierce). In parallel, protein concentrations of these samples were determined to be able to accurately express the amount of VEGF per milligram of total protein according to standard methods. Subsequently, Western blotting on VEGF and actin was done to determine the quantity of total (unbound and bound) VEGF in platelet samples. Mouse monoclonal antibodies for β-actin (Sigma; 1:10,000) and VEGF (R&D Systems; 1 μg/mL) were used according to standard procedures as described above. A standard curve of VEGF, added to human pooled plasma in a concentration range of 0 to 400 pg/mL, revealed an accurate semiquantitative dose response in the same concentration range of the patients samples (50-400 pg/mL) by Western blotting (data not shown).

Bevacizumab uptake and release by platelets. To determine whether platelets take up bevacizumab, we labeled bevacizumab and IgG with FITC or biotin according to standard procedures. Subsequently, we confirmed that biotinylated or FITC-labeled bevacizumab had biological activity similar to native bevacizumab in an endothelial cell proliferation assay (data not shown).

Platelets from healthy volunteers were incubated with increasing concentrations of FITC-labeled bevacizumab or IgG. FACS analysis revealed uptake of bevacizumab in a dose-dependent manner similar to control IgG (Fig. 1A). In addition, Western blot analysis of platelet lysates that were derived from biotinylated bevacizumab- and IgG-incubated platelets confirmed a concentration-dependent uptake (Fig. 1B). To determine whether immunoglobulins are taken up, and not just bound by platelets, we determined release of bevacizumab and IgG by platelet activation. FITC-labeled bevacizumab was secreted in a concentration-dependent manner on platelet activation (Fig. 1C). These results were also confirmed by release of biotin-labeled bevacizumab (Fig. 1B).

Fig. 1.
Fig. 1. A to C, bevacizumab is taken up by platelets and released on activation. A, FACS analysis of platelets incubated with FITC-labeled bevacizumab or IgG. Freshly isolated and washed platelets were incubated with increasing concentrations of FITC-labeled bevacizumab. Nonactivated platelets take up bevacizumab as shown by the increase in mean fluorescent intensity (MFI; y-axis), after incubation with increasing concentrations of FITC-labeled bevacizumab (x-axis). Representative of three independent experiments; bars, SE. B, Western blotting shows increased uptake of biotinylated bevacizumab detected by streptavidin peroxidase (SA-PO) by platelets at increasing concentrations of bevacizumab (0, 5, 10, and 20 μg/mL respectively). Thrombin receptor activator protein (TRAP; 10 μmol) activation of these platelets causes an almost complete disappearance of the bevacizumab that was taken up. Actin was used as protein loading control. C, quantitative measurement of fluorescence of platelets that were incubated with FITC-labeled bevacizumab (0, 5, 10, and 20 μg/mL). Solid black line, supernatant of thrombin-activated platelets; solid gray line, platelet pellets of the thrombin-activated platelets; interrupted black line, platelet pellets of resting platelets. N = 3 independent experiments; bars, SE. P = 0.0006, significant difference between supernatant and platelet pellets (ANOVA). Fluorescence of the supernatant of activated platelets was corrected for the fluorescence of the supernatant of resting platelets.
View largeDownload slide

A to C, bevacizumab is taken up by platelets and released on activation. A, FACS analysis of platelets incubated with FITC-labeled bevacizumab or IgG. Freshly isolated and washed platelets were incubated with increasing concentrations of FITC-labeled bevacizumab. Nonactivated platelets take up bevacizumab as shown by the increase in mean fluorescent intensity (MFI; y-axis), after incubation with increasing concentrations of FITC-labeled bevacizumab (x-axis). Representative of three independent experiments; bars, SE. B, Western blotting shows increased uptake of biotinylated bevacizumab detected by streptavidin peroxidase (SA-PO) by platelets at increasing concentrations of bevacizumab (0, 5, 10, and 20 μg/mL respectively). Thrombin receptor activator protein (TRAP; 10 μmol) activation of these platelets causes an almost complete disappearance of the bevacizumab that was taken up. Actin was used as protein loading control. C, quantitative measurement of fluorescence of platelets that were incubated with FITC-labeled bevacizumab (0, 5, 10, and 20 μg/mL). Solid black line, supernatant of thrombin-activated platelets; solid gray line, platelet pellets of the thrombin-activated platelets; interrupted black line, platelet pellets of resting platelets. N = 3 independent experiments; bars, SE. P = 0.0006, significant difference between supernatant and platelet pellets (ANOVA). Fluorescence of the supernatant of activated platelets was corrected for the fluorescence of the supernatant of resting platelets.

Fig. 1.
Fig. 1. A to C, bevacizumab is taken up by platelets and released on activation. A, FACS analysis of platelets incubated with FITC-labeled bevacizumab or IgG. Freshly isolated and washed platelets were incubated with increasing concentrations of FITC-labeled bevacizumab. Nonactivated platelets take up bevacizumab as shown by the increase in mean fluorescent intensity (MFI; y-axis), after incubation with increasing concentrations of FITC-labeled bevacizumab (x-axis). Representative of three independent experiments; bars, SE. B, Western blotting shows increased uptake of biotinylated bevacizumab detected by streptavidin peroxidase (SA-PO) by platelets at increasing concentrations of bevacizumab (0, 5, 10, and 20 μg/mL respectively). Thrombin receptor activator protein (TRAP; 10 μmol) activation of these platelets causes an almost complete disappearance of the bevacizumab that was taken up. Actin was used as protein loading control. C, quantitative measurement of fluorescence of platelets that were incubated with FITC-labeled bevacizumab (0, 5, 10, and 20 μg/mL). Solid black line, supernatant of thrombin-activated platelets; solid gray line, platelet pellets of the thrombin-activated platelets; interrupted black line, platelet pellets of resting platelets. N = 3 independent experiments; bars, SE. P = 0.0006, significant difference between supernatant and platelet pellets (ANOVA). Fluorescence of the supernatant of activated platelets was corrected for the fluorescence of the supernatant of resting platelets.
View largeDownload slide

A to C, bevacizumab is taken up by platelets and released on activation. A, FACS analysis of platelets incubated with FITC-labeled bevacizumab or IgG. Freshly isolated and washed platelets were incubated with increasing concentrations of FITC-labeled bevacizumab. Nonactivated platelets take up bevacizumab as shown by the increase in mean fluorescent intensity (MFI; y-axis), after incubation with increasing concentrations of FITC-labeled bevacizumab (x-axis). Representative of three independent experiments; bars, SE. B, Western blotting shows increased uptake of biotinylated bevacizumab detected by streptavidin peroxidase (SA-PO) by platelets at increasing concentrations of bevacizumab (0, 5, 10, and 20 μg/mL respectively). Thrombin receptor activator protein (TRAP; 10 μmol) activation of these platelets causes an almost complete disappearance of the bevacizumab that was taken up. Actin was used as protein loading control. C, quantitative measurement of fluorescence of platelets that were incubated with FITC-labeled bevacizumab (0, 5, 10, and 20 μg/mL). Solid black line, supernatant of thrombin-activated platelets; solid gray line, platelet pellets of the thrombin-activated platelets; interrupted black line, platelet pellets of resting platelets. N = 3 independent experiments; bars, SE. P = 0.0006, significant difference between supernatant and platelet pellets (ANOVA). Fluorescence of the supernatant of activated platelets was corrected for the fluorescence of the supernatant of resting platelets.

Close modal

Localization of bevacizumab in platelets. P-selectin is a marker of α-granules and the intermediate microvesicular bodies (pre-stage of α-granules) of resting platelets, and almost no P-selectin is present on the plasma membrane (23). After incubation with FITC-labeled bevacizumab, platelets were subsequently stained with a fluorescently labeled P-selectin antibody. Immunofluorescence microscopy of FITC-labeled bevacizumab (in green) and P-selectin (in red) showed a double granular staining pattern, indicative for bevacizumab localization in α-granules (Fig. 2A). By Western blotting, we found that biotinylated IgG and fibronectin, another α-granule protein, were present in the same secreted protein fractions of activated platelets that were preincubated with biotinylated IgG (Fig. 2B). To determine whether uptake of bevacizumab blocks platelet VEGF, we measured release of free VEGF in supernatant of activated platelets that were preincubated with bevacizumab by ELISA. This ELISA method of R&D Systems recognizes free, unbound VEGF but is unable to detect VEGF bound to bevacizumab (ref. 18 and Fig. 4A). Platelet (α-granule) derived VEGF is neutralized by platelet uptake of bevacizumab in a dose-dependent manner, as determined in the supernatant of activated platelets (Fig. 2C).

Fig. 2.
Fig. 2. Localization of bevacizumab in platelets. A, confocal microscopy of platelets (depicted is one on each row) incubated with FITC-labeled bevacizumab (green, middle) and stained for P-selectin (red, left); right, the overlay of left and middle images. B, Western blotting of lysates derived from platelets incubated with biotinylated IgG (5 μg/mL; +) or no IgG (−). Top row, Western blot for fibronectin (FN). Middle row, IgG detected by streptavidin peroxidase. Bottom row, actin protein loading control. Left, nonactivated platelets; right, activated platelets. C, decrease of free platelet-derived VEGF. VEGF concentrations were determined in supernatant of activated platelets by thrombin (5 IU/mL), preincubated with increasing concentrations of either bevacizumab (black line) or IgG control (gray line).
View largeDownload slide

Localization of bevacizumab in platelets. A, confocal microscopy of platelets (depicted is one on each row) incubated with FITC-labeled bevacizumab (green, middle) and stained for P-selectin (red, left); right, the overlay of left and middle images. B, Western blotting of lysates derived from platelets incubated with biotinylated IgG (5 μg/mL; +) or no IgG (−). Top row, Western blot for fibronectin (FN). Middle row, IgG detected by streptavidin peroxidase. Bottom row, actin protein loading control. Left, nonactivated platelets; right, activated platelets. C, decrease of free platelet-derived VEGF. VEGF concentrations were determined in supernatant of activated platelets by thrombin (5 IU/mL), preincubated with increasing concentrations of either bevacizumab (black line) or IgG control (gray line).

Fig. 2.
Fig. 2. Localization of bevacizumab in platelets. A, confocal microscopy of platelets (depicted is one on each row) incubated with FITC-labeled bevacizumab (green, middle) and stained for P-selectin (red, left); right, the overlay of left and middle images. B, Western blotting of lysates derived from platelets incubated with biotinylated IgG (5 μg/mL; +) or no IgG (−). Top row, Western blot for fibronectin (FN). Middle row, IgG detected by streptavidin peroxidase. Bottom row, actin protein loading control. Left, nonactivated platelets; right, activated platelets. C, decrease of free platelet-derived VEGF. VEGF concentrations were determined in supernatant of activated platelets by thrombin (5 IU/mL), preincubated with increasing concentrations of either bevacizumab (black line) or IgG control (gray line).
View largeDownload slide

Localization of bevacizumab in platelets. A, confocal microscopy of platelets (depicted is one on each row) incubated with FITC-labeled bevacizumab (green, middle) and stained for P-selectin (red, left); right, the overlay of left and middle images. B, Western blotting of lysates derived from platelets incubated with biotinylated IgG (5 μg/mL; +) or no IgG (−). Top row, Western blot for fibronectin (FN). Middle row, IgG detected by streptavidin peroxidase. Bottom row, actin protein loading control. Left, nonactivated platelets; right, activated platelets. C, decrease of free platelet-derived VEGF. VEGF concentrations were determined in supernatant of activated platelets by thrombin (5 IU/mL), preincubated with increasing concentrations of either bevacizumab (black line) or IgG control (gray line).

Close modal

Angiogenic activity of platelets that were incubated with bevacizumab. To determine whether blockade of platelet-derived VEGF impairs platelet function, we conducted an endothelial cell proliferation assay with HUVECs. HUVECs are responsive to VEGF and are widely used to study aspects of angiogenesis in vitro (2, 24). Platelets stimulate endothelial cell proliferation, indicative of their angiogenic activity (2). Bevacizumab uptake by platelets inhibited platelet-induced endothelial cell proliferation compared with IgG-loaded platelets (maximal inhibition, 74 ± 8%; P = 0.003; Fig. 3A and B).

Fig. 3.
Fig. 3. Uptake of bevacizumab by platelets inhibits platelet-induced endothelial cell proliferation. A and B, endothelial cell proliferation stimulated by supernatant of activated platelets. Proliferation of endothelial cells was stimulated by platelet supernatants after platelet incubation with IgG (bullets and lines in gray) or bevacizumab (bullets and lines in black) and subsequent activation with thrombin (0.5 IU/mL). Representative of three independent experiments. Results are normalized to maximal proliferation of HUVECs by 400,000 platelets/μL; bars, SE. Significant differences at P = 0.0003 (A) and P = 0.014 (B), ANOVA. A, effect of a fixed platelet number of 200,000 platelets/μL that were preincubated with increasing concentrations of bevacizumab or IgG. B, effect of increasing number of platelets preincubated with a fixed concentration of bevacizumab or IgG (20 μg/mL).
View largeDownload slide

Uptake of bevacizumab by platelets inhibits platelet-induced endothelial cell proliferation. A and B, endothelial cell proliferation stimulated by supernatant of activated platelets. Proliferation of endothelial cells was stimulated by platelet supernatants after platelet incubation with IgG (bullets and lines in gray) or bevacizumab (bullets and lines in black) and subsequent activation with thrombin (0.5 IU/mL). Representative of three independent experiments. Results are normalized to maximal proliferation of HUVECs by 400,000 platelets/μL; bars, SE. Significant differences at P = 0.0003 (A) and P = 0.014 (B), ANOVA. A, effect of a fixed platelet number of 200,000 platelets/μL that were preincubated with increasing concentrations of bevacizumab or IgG. B, effect of increasing number of platelets preincubated with a fixed concentration of bevacizumab or IgG (20 μg/mL).

Fig. 3.
Fig. 3. Uptake of bevacizumab by platelets inhibits platelet-induced endothelial cell proliferation. A and B, endothelial cell proliferation stimulated by supernatant of activated platelets. Proliferation of endothelial cells was stimulated by platelet supernatants after platelet incubation with IgG (bullets and lines in gray) or bevacizumab (bullets and lines in black) and subsequent activation with thrombin (0.5 IU/mL). Representative of three independent experiments. Results are normalized to maximal proliferation of HUVECs by 400,000 platelets/μL; bars, SE. Significant differences at P = 0.0003 (A) and P = 0.014 (B), ANOVA. A, effect of a fixed platelet number of 200,000 platelets/μL that were preincubated with increasing concentrations of bevacizumab or IgG. B, effect of increasing number of platelets preincubated with a fixed concentration of bevacizumab or IgG (20 μg/mL).
View largeDownload slide

Uptake of bevacizumab by platelets inhibits platelet-induced endothelial cell proliferation. A and B, endothelial cell proliferation stimulated by supernatant of activated platelets. Proliferation of endothelial cells was stimulated by platelet supernatants after platelet incubation with IgG (bullets and lines in gray) or bevacizumab (bullets and lines in black) and subsequent activation with thrombin (0.5 IU/mL). Representative of three independent experiments. Results are normalized to maximal proliferation of HUVECs by 400,000 platelets/μL; bars, SE. Significant differences at P = 0.0003 (A) and P = 0.014 (B), ANOVA. A, effect of a fixed platelet number of 200,000 platelets/μL that were preincubated with increasing concentrations of bevacizumab or IgG. B, effect of increasing number of platelets preincubated with a fixed concentration of bevacizumab or IgG (20 μg/mL).

Close modal

Bevacizumab uptake by platelets in vivo. To investigate whether platelets take up bevacizumab in vivo, we treated rabbits with FITC-labeled bevacizumab or IgG (10 mg/kg i.v.) and measured fluorescence of isolated platelets at multiple time points. Fluorescent platelets were present 2 h after bevacizumab administration and disappeared almost completely 2 weeks later (Fig. 4). Plasma measurements of FITC-labeled bevacizumab and IgG revealed faster disappearance with a half-life of <24 h. FITC-bevacizumab or IgG in platelets was cleared from the circulation with a half-life of ∼48 h.

Fig. 4.
Fig. 4. Uptake of bevacizumab by platelets in vivo. Uptake of bevacizumab by platelets in rabbits. Before and after i.v. injection of FITC-labeled bevacizumab or IgG control (pretreatment, 2 h, 24 h, 48 h, 7 d, and 14 d), FACS measurement of isolated platelets was done (black) or direct fluorescence of plasma was determined as described (gray). FACS analysis of platelets and plasma shows that plasma half-life of bevacizumab in rabbits is ∼17 h whereas in platelets the half-life fluorescence is 48 h. Bars, SE. In each group, three individual rabbits were analyzed. The experiments were repeated twice with similar results. A representative experiment is shown.
View largeDownload slide

Uptake of bevacizumab by platelets in vivo. Uptake of bevacizumab by platelets in rabbits. Before and after i.v. injection of FITC-labeled bevacizumab or IgG control (pretreatment, 2 h, 24 h, 48 h, 7 d, and 14 d), FACS measurement of isolated platelets was done (black) or direct fluorescence of plasma was determined as described (gray). FACS analysis of platelets and plasma shows that plasma half-life of bevacizumab in rabbits is ∼17 h whereas in platelets the half-life fluorescence is 48 h. Bars, SE. In each group, three individual rabbits were analyzed. The experiments were repeated twice with similar results. A representative experiment is shown.

Fig. 4.
Fig. 4. Uptake of bevacizumab by platelets in vivo. Uptake of bevacizumab by platelets in rabbits. Before and after i.v. injection of FITC-labeled bevacizumab or IgG control (pretreatment, 2 h, 24 h, 48 h, 7 d, and 14 d), FACS measurement of isolated platelets was done (black) or direct fluorescence of plasma was determined as described (gray). FACS analysis of platelets and plasma shows that plasma half-life of bevacizumab in rabbits is ∼17 h whereas in platelets the half-life fluorescence is 48 h. Bars, SE. In each group, three individual rabbits were analyzed. The experiments were repeated twice with similar results. A representative experiment is shown.
View largeDownload slide

Uptake of bevacizumab by platelets in vivo. Uptake of bevacizumab by platelets in rabbits. Before and after i.v. injection of FITC-labeled bevacizumab or IgG control (pretreatment, 2 h, 24 h, 48 h, 7 d, and 14 d), FACS measurement of isolated platelets was done (black) or direct fluorescence of plasma was determined as described (gray). FACS analysis of platelets and plasma shows that plasma half-life of bevacizumab in rabbits is ∼17 h whereas in platelets the half-life fluorescence is 48 h. Bars, SE. In each group, three individual rabbits were analyzed. The experiments were repeated twice with similar results. A representative experiment is shown.

Close modal

Uptake of bevacizumab by platelets in patients. To confirm that antibodies are present in human platelets and can be released after activation, we measured the amount of IgG in platelet-poor plasma and serum of healthy volunteers. We found a 15% higher concentration of IgG in serum compared with platelet poor plasma. IgG concentration was 13 ± 1.3 mg/mL in serum versus 11 ± 1.2 mg/mL in plasma (n = 5; P < 0.001). To study whether bevacizumab is taken up by platelets in patients, we sampled platelet-rich and platelet-poor plasma from seven cancer patients before and after administration of bevacizumab in a 2-week treatment regimen. After isolation of platelet-rich plasma, we isolated the platelets by subsequent centrifugation. Subsequently, we lysed these platelet samples and determined total protein and VEGF content in these samples.

As a surrogate marker for the uptake of bevacizumab by platelets from patients, we hypothesized that the R&D ELISA assay cannot detect VEGF when neutralized by bevacizumab, but we could determine total VEGF content by Western blotting. To test this hypothesis, we measured the concentration of VEGF (12 ng/mL = 5-10× serum concentration) in samples after the addition of increasing concentrations of bevacizumab. The amount of VEGF that could be detected in this assay was already markedly decreased in the presence of only 1 μg/mL bevacizumab (Fig. 5A). In subsequent measurements, we used this finding to create a surrogate marker, by comparing unbound platelet VEGF as measured by ELISA with the total amount of platelet VEGF as measured by Western blotting, for the presence of bevacizumab in platelets.

Fig. 5.
Fig. 5. Evidence for bevacizumab uptake by platelets in bevacizumab treated patients. A, ELISA of samples containing 12 ng/mL VEGF in the presence of IgG control or bevacizumab. The ELISA was done after 30-min incubation with IgG control or bevacizumab. X-axis, concentration of antibody. Y-axis, results of VEGF ELISA in nanograms per milliliter. B, ELISA of free VEGF in isolated platelets and platelet-poor plasma samples of patients. The average of VEGF concentrations in the platelet samples of multiple time points from seven different patients is shown before (7 samples) and during treatment for 8 h or more (15 samples) at multiple time points (up to 154 d). Y-axis, VEGF concentration per milligram of total protein concentration. C, ELISA of free VEGF in platelets of two patients. In two patients, blood was collected pretreatment, 2 h, 8 h, 48 h, and every 2 wk after start of treatment. D, Western blot of total VEGF in platelet samples of patients. Western blotting shows unchanged VEGF levels in the platelet samples, indicating that total VEGF in platelets is not changed but is undetectable by ELISA because of bevacizumab blockade. Actin was used as a protein loading control.
View largeDownload slide

Evidence for bevacizumab uptake by platelets in bevacizumab treated patients. A, ELISA of samples containing 12 ng/mL VEGF in the presence of IgG control or bevacizumab. The ELISA was done after 30-min incubation with IgG control or bevacizumab. X-axis, concentration of antibody. Y-axis, results of VEGF ELISA in nanograms per milliliter. B, ELISA of free VEGF in isolated platelets and platelet-poor plasma samples of patients. The average of VEGF concentrations in the platelet samples of multiple time points from seven different patients is shown before (7 samples) and during treatment for 8 h or more (15 samples) at multiple time points (up to 154 d). Y-axis, VEGF concentration per milligram of total protein concentration. C, ELISA of free VEGF in platelets of two patients. In two patients, blood was collected pretreatment, 2 h, 8 h, 48 h, and every 2 wk after start of treatment. D, Western blot of total VEGF in platelet samples of patients. Western blotting shows unchanged VEGF levels in the platelet samples, indicating that total VEGF in platelets is not changed but is undetectable by ELISA because of bevacizumab blockade. Actin was used as a protein loading control.

Fig. 5.
Fig. 5. Evidence for bevacizumab uptake by platelets in bevacizumab treated patients. A, ELISA of samples containing 12 ng/mL VEGF in the presence of IgG control or bevacizumab. The ELISA was done after 30-min incubation with IgG control or bevacizumab. X-axis, concentration of antibody. Y-axis, results of VEGF ELISA in nanograms per milliliter. B, ELISA of free VEGF in isolated platelets and platelet-poor plasma samples of patients. The average of VEGF concentrations in the platelet samples of multiple time points from seven different patients is shown before (7 samples) and during treatment for 8 h or more (15 samples) at multiple time points (up to 154 d). Y-axis, VEGF concentration per milligram of total protein concentration. C, ELISA of free VEGF in platelets of two patients. In two patients, blood was collected pretreatment, 2 h, 8 h, 48 h, and every 2 wk after start of treatment. D, Western blot of total VEGF in platelet samples of patients. Western blotting shows unchanged VEGF levels in the platelet samples, indicating that total VEGF in platelets is not changed but is undetectable by ELISA because of bevacizumab blockade. Actin was used as a protein loading control.
View largeDownload slide

Evidence for bevacizumab uptake by platelets in bevacizumab treated patients. A, ELISA of samples containing 12 ng/mL VEGF in the presence of IgG control or bevacizumab. The ELISA was done after 30-min incubation with IgG control or bevacizumab. X-axis, concentration of antibody. Y-axis, results of VEGF ELISA in nanograms per milliliter. B, ELISA of free VEGF in isolated platelets and platelet-poor plasma samples of patients. The average of VEGF concentrations in the platelet samples of multiple time points from seven different patients is shown before (7 samples) and during treatment for 8 h or more (15 samples) at multiple time points (up to 154 d). Y-axis, VEGF concentration per milligram of total protein concentration. C, ELISA of free VEGF in platelets of two patients. In two patients, blood was collected pretreatment, 2 h, 8 h, 48 h, and every 2 wk after start of treatment. D, Western blot of total VEGF in platelet samples of patients. Western blotting shows unchanged VEGF levels in the platelet samples, indicating that total VEGF in platelets is not changed but is undetectable by ELISA because of bevacizumab blockade. Actin was used as a protein loading control.

Close modal

Measurements of the isolated platelet samples before treatment with bevacizumab revealed significantly higher VEGF content (123 ± 35.4 pg/mg of total protein) compared with VEGF content of platelet-poor plasma samples (6.5 ± 5.1 pg/mg of total protein; P = 0.01) in these seven patients. After 8 h and at all later time points of blood sampling during two weekly bevacizumab administrations, the free, unbound VEGF content was significantly decreased in the platelet samples compared with pretreatment concentrations, 10.8 ± 0.9 (n = 15) versus 123 ± 35.4 (n = 7) pg/mg of total protein, respectively (P = 0.02; Fig. 5B). In two patients, we were able to collect blood at 2, 8, and 48 h after start of treatment. Bevacizumab was not taken up by platelets to allow neutralization of platelet VEGF within 2 h after administration, whereas after 8 h or at any later time point, bevacizumab uptake by platelets resulted in a significant decrease of platelet-free VEGF (Fig. 5C). Total VEGF content (unbound and bound VEGF) in these platelet samples remained unchanged as assessed by Western blotting (Fig. 5D), indicating that the decrease of free VEGF is due to neutralization by bevacizumab and not by absolute decrease of total VEGF. The mean pretreatment VEGF plasma concentration was not significantly different from the mean treatment VEGF plasma concentration during treatment, 6.5 ± 5.1 pg/mg (n = 7) and 4.0 ± 0.9 pg/mg (n = 17), respectively (P = 0.06). These results confirmed our preclinical findings that bevacizumab blocks platelet VEGF in patients.

To the best of our knowledge, we here report for the first time that platelets take up bevacizumab in their α-granules, resulting in neutralization of platelet VEGF. Uptake of bevacizumab and subsequent neutralization of VEGF inhibited platelet angiogenic activity. We confirmed these preclinical findings in patients by showing that bevacizumab blocks platelet VEGF. Platelet uptake of nonspecific IgG and the fact that trastuzumab and cetuximab (data not shown) were also taken up by platelets suggest that this phenomenon is not dependent on the specificity of the antibody. These results may have important implications for treatment of patients with therapeutic (antiangiogenic) antibodies.

After the initial in vitro studies, we confirmed platelet uptake of bevacizumab in vivo. We used rabbits because bevacizumab has a weak affinity for VEGF in these animals but it has no affinity for VEGF in other rodents (25). The half-life of bevacizumab in rabbit plasma (<24 h) is significantly shorter than the one in human and rodent serum studies (17, 26). To our knowledge, the plasma half-life of bevacizumab in rabbits or other species has never been reported before. FACS analysis revealed that platelets containing bevacizumab and IgG were cleared from the circulation with a half-life of ∼48 h. If the difference between plasma and platelet clearance of bevacizumab holds true in humans and can be extrapolated to other immunoglobulins, this observation may have important biological consequences. These differences in platelet and plasma bevacizumab pharmacokinetics may also be of clinical relevance when taking into account the number of circulating platelets in patients.

It is known that platelets play a pivotal role in primary hemostasis at sites where new blood vessel formation is required. Platelets secrete high quantities of VEGF from their α-granules at sites of wound healing to stimulate angiogenesis (27) and thereby promote cutaneous wound healing (28). It has been shown that platelets also play a significant role in gastric ulcer healing by secretion of their α-granule contents (29). Side effects of bevacizumab treatment include hypertension, impaired wound healing, bleeding (i.e., hemoptysis), and gastrointestinal perforations (30, 31). These side effects are commonly linked to inhibition of free circulating VEGF activity by bevacizumab. However, we hypothesized that the underlying mechanism of these side effects may be related to the uptake of bevacizumab and specific blockade of platelet VEGF, which is important for wound healing. We focused our studies on the biological role of bevacizumab uptake in platelets. Platelet-induced endothelial cell proliferation is inhibited by platelet uptake of bevacizumab, which indicates that this uptake impairs the biological activity of platelets. It is possible that the effects of bevacizumab seen in patients, such as the antitumor effect, impaired wound healing, bleeding, and gastrointestinal perforations, are linked to an impaired platelet function caused by uptake of bevacizumab.

In our opinion, the finding that platelet VEGF is also neutralized in patients treated with bevacizumab is of major importance. The recent clinical development of new antiangiogenic therapeutic antibodies and small-molecule receptor tyrosine kinase inhibitors has shown differences in toxicity. These differences may be in part due to our findings that antibodies are taken up by platelets. Concentrations as low as 1 μg/mL bevacizumab blocked most of the VEGF stored in the α-granules of platelets (Fig. 2C). Based on these and additional measurements (data not shown), we were able to calculate that after incubation with 20 μg/mL of the antibody, ∼100 times more molecules of bevacizumab than VEGF are present in the platelets. This finding implicates that VEGF is not only neutralized by bevacizumab inside the platelets but also that platelets can act as a vehicle. At the sites of endothelial damage or activation, platelets may release functional bevacizumab. We hypothesize that on activation of platelets, bevacizumab may be locally delivered at procoagulatory angiogenic tumor sites at relatively high concentrations and subsequently target tumor cell or other host cell–derived VEGF. Such a local delivery to sites of platelet activation may represent a critical difference in the pharmacodynamics of therapeutic antibodies as compared with receptor tyrosine kinase inhibitors. Platelet-transported antibodies may act at different sites and at different concentrations than small-molecule inhibitors. These differences may be relevant for the interpretation of clinical trial data. Further studies should be done to explore the importance of platelet delivery of therapeutic antibodies to tumor vasculature.

Another possible mechanism associated with the uptake of bevacizumab by platelets could be the clearance of VEGF from the plasma through the uptake of VEGF-bevacizumab complexes. Our data do not support a role for this mechanism at this point, but a recent publication revealed that by using a different ELISA system, VEGF in complex with bevacizumab can be measured as well. Using both ELISA systems in future studies may provide more insight into the possible uptake of VEGF-bevacizumab complexes in platelets (32).

In conclusion, we propose that platelets play a role in the mechanism of action of bevacizumab due to direct uptake and subsequent release on activation. This observation has an effect on the pharmacodynamics and clinical side effects of bevacizumab and potentially other therapeutic antibodies.

Grant support: The Adriana Van Coevorden Stichting (H.M.W. Verheul) and The Commonwealth Foundation (H.M.W. Verheul and R. Pili). H.M.W. Verheul is a recipient of the American Society of Clinical Oncology Young Investigator's Award 2006 and a fellow in the Drug Development Fellowship Program of the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins Medical Institutions.

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.

Note: H.M.W. Verheul, M.P.J. Lolkema, R. Pili, and E.E. Voest contributed equally to this work.

We thank S. Kachhap for his assistance with the confocal microscopy studies, E. Sugar for her help with the statistical analysis, and J. Geschwind for his support with the rabbit studies.

1
Ferrara N, Kerbel RS. Angiogenesis as a therapeutic target.
Nature
 
2005
;
438
:
967
–74.
2
Verheul HM, Jorna AS, Hoekman K, Broxterman HJ, Gebbink MF, Pinedo HM. Vascular endothelial growth factor-stimulated endothelial cells promote adhesion and activation of platelets.
Blood
 
2000
;
96
:
4216
–21.
3
Brill A, Elinav H, Varon D. Differential role of platelet granular mediators in angiogenesis.
Cardiovasc Res
 
2004
;
63
:
226
–35.
4
Caine GJ, Lip GY, Blann AD. Platelet-derived VEGF, Flt-1, angiopoietin-1 and P-selectin in breast and prostate cancer: further evidence for a role of platelets in tumour angiogenesis.
Ann Med
 
2004
;
36
:
273
–7.
5
Kisucka J, Butterfield CE, Duda DG, et al. Platelets and platelet adhesion support angiogenesis while preventing excessive hemorrhage.
Proc Natl Acad Sci U S A
 
2006
;
103
:
855
–60.
6
Plow EF, Ginsberg MH. The molecular basis for platelet function. In: Hoffman DR, Benz EJ, Jr., Shattil SJ, et al, editors. Hematology: basic principles and practice. Philadelphia: Churchill Livingstone; 2000. p. 1741–52.
7
Gasic GJ, Gasic TB, Stewart CC. Antimetastatic effects associated with platelet reduction.
Proc Natl Acad Sci U S A
 
1968
;
61
:
46
–52.
8
Honn KV, Tang DG, Crissman JD. Platelets and cancer metastasis: a causal relationship?
Cancer Metastasis Rev
 
1992
;
11
:
325
–51.
9
Pinedo HM, Verheul HM, D'Amato RJ, Folkman J. Involvement of platelets in tumour angiogenesis?
Lancet
 
1998
;
352
:
1775
–7.
10
Verheul HM, Hoekman K, Lupu F, et al. Platelet and coagulation activation with vascular endothelial growth factor generation in soft tissue sarcomas.
Clin Cancer Res
 
2000
;
6
:
166
–71.
11
Klement G, Kikuchi L, Kieran M, Almog N, Yip T, Folkman J. Early tumor detection using platelet uptake of angiogenesis regulators. Blood (ASH Annual Meeting 2006);2004:104:abstract 833.
12
Handagama PJ, Shuman MA, Bainton DF. Incorporation of intravenously injected albumin, immunoglobulin G, and fibrinogen in guinea pig megakaryocyte granules.
J Clin Invest
 
1989
;
84
:
73
–82.
13
George JN. Platelet immunoglobulin G: its significance for the evaluation of thrombocytopenia and for understanding the origin of α-granule proteins.
Blood
 
1990
;
76
:
859
–70.
14
Cartron G, Watier H, Golay J, Solal-Celigny P. From the bench to the bedside: ways to improve rituximab efficacy.
Blood
 
2004
;
104
:
2635
–42.
15
Adams GP, Weiner LM. Monoclonal antibody therapy of cancer.
Nat Biotechnol
 
2005
;
23
:
1147
–57.
16
Jain RK, Duda DG, Clark JW, Loeffler JS. Lessons from phase III clinical trials on anti-VEGF therapy for cancer.
Nat Clin Pract Oncol
 
2006
;
3
:
24
–40.
17
Gordon MS, Margolin K, Talpaz M, et al. Phase I safety and pharmacokinetic study of recombinant human anti-vascular endothelial growth factor in patients with advanced cancer.
J Clin Oncol
 
2001
;
19
:
843
–50.
18
Karp JE, Gojo I, Pili R, et al. Targeting vascular endothelial growth factor for relapsed and refractory adult acute myelogenous leukemias: therapy with sequential 1-β-d-arabinofuranosylcytosine, mitoxantrone, and bevacizumab.
Clin Cancer Res
 
2004
;
10
:
3577
–85.
19
Verheul HM, Hoekman K, Luykx-de Bakker S, et al. Platelet: transporter of vascular endothelial growth factor.
Clin Cancer Res
 
1997
;
3
:
2187
–90.
20
Duhamel RC, Schur PH, Brendel K, Meezan E. pH gradient elution of human IgG1, IgG2 and IgG4 from protein A-sepharose.
J Immunol Methods
 
1979
;
31
:
211
–7.
21
Ferreira IA, Mocking AI, Urbanus RT, Varlack S, Wnuk M, Akkerman JW. Glucose uptake via glucose transporter 3 by human platelets is regulated by protein kinase B.
J Biol Chem
 
2005
;
280
:
32625
–33.
22
Qian DZ, Wang X, Kachhap SK, et al. The histone deacetylase inhibitor NVP-LAQ824 inhibits angiogenesis and has a greater antitumor effect in combination with the vascular endothelial growth factor receptor tyrosine kinase inhibitor PTK787/ZK222584.
Cancer Res
 
2004
;
64
:
6626
–34.
23
Heijnen FG, Debili N, Vainchencker W, Breton-Gorius J, Geuze HJ, Sixma JJ. Multivesicular bodies are and intermediate stage in the formation of platelet α-granules.
Blood
 
1998
;
91
:
2313
–25.
24
Soker S, Gollamudi-Payne S, Fidder H, Charmahelli H, Klagsbrun M. Inhibition of vascular endothelial growth factor (VEGF)-induced endothelial cell proliferation by a peptide corresponding to the exon 7-encoded domain of VEGF165.
J Biol Chem
 
1997
;
272
:
31582
–8.
25
Ferrara N, Hillan KJ, Gerber HP, Novotny W. Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer.
Nat Rev Drug Discov
 
2004
;
3
:
391
–400.
26
Lin YS, Nguyen C, Mendoza JL, et al. Preclinical pharmacokinetics, interspecies scaling, and tissue distribution of a humanized monoclonal antibody against vascular endothelial growth factor.
J Pharmacol Exp Ther
 
1999
;
288
:
371
–8.
27
Weltermann A, Wolzt M, Petersmann K, et al. Large amounts of vascular endothelial growth factor at the site of hemostatic plug formation in vivo.
Arterioscler Thromb Vasc Biol
 
1999
;
19
:
1757
–60.
28
Knighton DR, Ciresi K, Fiegel VD, Schumerth S, Butler E, Cerra F. Stimulation of repair in chronic, nonhealing, cutaneous ulcers using platelet-derived wound healing formula.
Surg Gynecol Obstet
 
1990
;
170
:
56
–60.
29
Ma L, Elliott SN, Cirino G, Buret A, Ignarro LJ, Wallace JL. Platelets modulate gastric ulcer healing: role of endostatin and vascular endothelial growth factor release.
Proc Natl Acad Sci U S A
 
2001
;
98
:
6470
–5.
30
Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer.
N Engl J Med
 
2004
;
350
:
2335
–42.
31
Ellis LM. Bevacizumab. Nat Rev Drug Discov 2005;(Suppl):S8–9.
32
Loupakis F, Falcone A, Masi G, et al. Vascular endothelial growth factor levels in immunodepleted plasma of cancer patients as a possible pharmacodynamic marker for bevacizumab activity.
J Clin Oncol
 
2007
;
25
:
1816
–8.
American Association for Cancer Research
2007