A novel antiplatelet antibody therapy that induces cAMP-dependent endocytosis of the GPVI/Fc receptor γ-chain complex

Platelet adhesion to vascular subendothelium, mediated in part by interactions between collagen and glycoprotein VI (GPVI) complexed with Fc receptor γ-chain, is crucial for thrombus formation. Antiplatelet therapy benefits patients with various thrombotic and ischemic diseases, but the safety and efficacy of existing treatments are limited. Recent data suggest GPVI as a promising target for a novel antiplatelet therapy, for example, GPVI-specific Abs that deplete GPVI from the surface of platelets. Here, we characterized GPVI-specific auto-Abs (YA-Abs) from the first reported patient with ongoing platelet GPVI deficiency caused by the YA-Abs. To obtain experimentally useful human GPVI–specific mAbs with characteristics similar to YA-Abs, we generated human GPVI–specific mouse mAbs and selected 2 representative mAbs, mF1201 and mF1232, whose binding to GPVI was inhibited by YA-Abs. In vitro, mF1201, but not mF1232, induced human platelet activation and GPVI shedding, and mF1232 inhibited collagen-induced human platelet aggregation. Administration of mF1201 and mF1232 to monkeys caused GPVI immunodepletion with and without both significant thrombocytopenia and GPVI shedding, respectively. When a human/mouse chimeric form of mF1232 (cF1232) was labeled with a fluorescent endocytosis probe and administered to monkeys, fluorescence increased in circulating platelets and surface GPVI was lost. Loss of platelet surface GPVI mediated by cF1232 was successfully reproduced in vitro in the presence of a cAMP-elevating agent. Thus, we have characterized cAMP-dependent endocytosis of GPVI mediated by a human GPVI–specific mAb as what we believe to be a novel antiplatelet therapy.

Antiplatelet therapy provides significant clinical benefits in patients with a variety of thrombotic or ischemic diseases such as myocardial infarction, unstable angina, and stroke (1). However, the safety and efficacy of existing treatments are limited, and there remains a clinical need for novel antiplatelet interventions to reduce pathological thrombus formation without promoting bleeding (2).

Over the last decade, considerable progress has been made in identifying and describing the molecular events regulating the initiation of thrombus formation. At sites of vascular damage, subendothelial matrix proteins, primarily collagen, are exposed, promoting platelet adhesion and activation through interactions with specific cell-surface receptors, leading to thrombus formation.

Several cellular collagen receptors mediate this response, and, among these, glycoprotein VI (GPVI) plays a critical role in the platelet response to collagen. GPVI is a member of the Ig superfamily (3–5), and it is constitutively associated and expressed with the Fc receptor γ-chain (FcRγ), an immunoreceptor tyrosine-based activating motif–bearing (ITAM-bearing) receptor (6). Upon GPVI engagement, phosphorylation of the ITAM motif of FcRγ initiates a signaling cascade, leading to platelet activation (6–10). The extracellular domain of GPVI contains 2 Ig-like loops of the C2 type (3–5). The extracellular domain 1 (D1) N-terminal loop is of the IgC2-1 type, and the extracellular D2 loop is of the IgC2-2 type (11). Deficiency or blockade of GPVI slightly prolongs bleeding time in both humans and mice (12–14), and this produces antithrombotic effects in several different experimental conditions (14–17). These data suggest that GPVI is a promising target for the development of novel antiplatelet therapies, but such therapies have yet to be successfully developed and/or clinically deployed.

Platelet GPVI deficiency and unresponsiveness to collagen have been first reported in a patient (YA) who developed anti-GPVI Abs in the plasma (12). Monovalent Fab fragments prepared from YA’s serum IgG blocked collagen-induced platelet aggregation, but divalent F(ab′)₂ fragments caused GPVI cross-linking and platelet activation. This report also suggested the possibility of specific Ab-induced GPVI immunodepletion. This assumption was confirmed by later studies in mice (14, 18), and an additional patient with Ab-induced GPVI deficiency was identified (19). Short-term antiplatelet effects may be safely generated by the infusion of anti-GPVI Fab fragments to block the GPVI/collagen interaction. The F(ab′)₂ fragments and intact Ab have a longer half-life in the serum than the Fab fragments, but their divalent nature would likely activate platelets by cross-linking GPVI. Alternatively, specific Ab-induced GPVI immunodepletion could be clinically useful. However, metalloproteinase-dependent proteolytic shedding of GPVI has been demonstrated in vitro to cause considerable loss of GPVI following activation of platelets exposed to GPVI-specific Abs or ligands (20, 21). On the other hand, GPVI internalization and endocytosis may not be denied as another mechanism for Ab-induced GPVI immunodepletion. However, no data exist about a GPVI-specific Ab that does induce GPVI internalization and endocytosis but not GPVI shedding. Most importantly, specific Ab-induced GPVI internalization and endocytosis remains to be completely demonstrated.

With these backgrounds in mind, we aimed to obtain clinically useful mAbs against human GPVI (hGPVI) with characteristics similar to YA’s auto-Abs (YA-Abs) and investigated the mechanisms underlying GPVI immunodepletion. In addition to our first presentation of nonactivating hGPVI-specific mAb-induced GPVI deficiency without significant thrombocytopenia (22), the present work reports the establishment of clinically applicable induction of the Ig-D2–specific mAb-mediated GPVI deficiency and also what we believe to be the first clarification of the mechanisms underlying such induction.

Our initial approach was based upon 2 assumptions: (a) the plasma of patient YA contains auto-Abs that downregulate GPVI expression, and (b) mouse mAbs that bind the same epitope(s) on hGPVI will function similarly to YA-Abs. We first analyzed frozen plasma isolated from patient YA for the presence of anti-GPVI auto-Abs using recombinant hGPVI fused to human Fc (hGPVI-Fc) as a target. As seen in Figure 1A, the concentration of anti-GPVI Abs in the plasma of YA was highest in 1988 and decreased more than 100-fold over the course of 15 years, plateauing at 1 to 5 μg/ml after 2002. However, during this time, the patient’s platelets remained unresponsive to collagen stimulation due to complete GPVI deficiency (23), and the number of platelets was maintained at normal levels (data not shown).

Figure 1

Characterization of YA-Abs over 15 years. (A) YA’s serially diluted plasma stored over 15 years or control solutions containing known amounts of a human/mouse chimera anti-GPVI mAb were incubated with biotin-labeled hGPVI-Fc on streptavidin-coated plates. After washing, the bound Abs were detected using peroxidase-labeled anti-kappa and -lambda chain Abs and subsequent colorimetric reaction. The content of anti-GPVI auto-Abs in the plasma was expressed as an equivalent content of a human/mouse chimera anti-GPVI mAb bound to hGPVI-Fc. Each bar represents the mean of duplicate determinations. (B) Both YA-Abs-88 and -03 were purified from the patient’s plasma stored in 1988 and 2003, respectively, and quantified as described in Methods. Serially diluted YA-Abs-88 (triangles), YA-Abs-03 (circles), mF1201 (diamonds), mF1232 (squares), or mouse IgG (asterisks) as the competing Ab was applied to hGPVI-Fc–immobilized plates followed by the incubation with the peroxidase-labeled mF1201 (left panel) or mF1232 (right panel). After washing, the bound labeled Abs were detected by colorimetric reaction. Data are expressed as the percent of total binding of each peroxidase-labeled mAb without the competitor and are the mean of triplicate determinations.

In order to characterize YA-Abs more in detail, we purified YA-Abs from the patient’s plasma stored in 1988 (YA-Abs-88) and in 2003 (YA-Abs-03) using sequential affinity chromatography on protein A–column and recombinant hGPVI-Fc–column and examined cross-reactivities of those YA-Abs with mouse anti-GPVI mAbs for binding to hGPVI-Fc as described in Methods. YA-Abs efficiently competed with several mAbs for binding to hGPVI-Fc, and we selected 2 representative mAbs, mF1201 and mF1232, for further study. These 2 mAbs did not cross-inhibit one another, indicating that they bind different epitopes of hGPVI. As shown in Figure 1B, mF1201 binding to hGPVI-Fc was inhibited by both YA-Abs-88 and YA-Abs-03 (IC₅₀, 58 ng/ml versus 89 ng/ml). In contrast, mF1232 binding to hGPVI-Fc was inhibited by YA-Abs-03 more strongly than YA-Abs-88 (IC₅₀, 45 ng/ml versus 136 ng/ml). These data suggest that YA-Abs-03 likely shares more epitopes with mF1232 than mF1201, despite the relatively low titer.

We next examined the ability of the isolated mAbs to modulate platelet function either through cross-linking–induced activation or by inhibiting collagen-induced platelet aggregation. Treatment of platelets with up to 10 μg/ml of mF1201 (in excess of reported plasma concentrations of YA-Abs isolated after 1997) induced both platelet aggregation as well as expression of the activation marker CD62P in a dose-dependent manner. In contrast to this finding, mF1232 neither induced aggregation nor significant CD62P expression but dose dependently inhibited collagen-induced platelet aggregation (Figure 2, A–C). However, 10 μg/ml mF1232 did not inhibit platelet aggregation in response to collagen-related peptide (CRP), a synthetic peptide that acts as a GPVI-specific agonist (24), adenosine diphosphate (ADP), ristocetin and vWF, or thrombin receptor-agonist peptide (TRAP) (Figure 2C and data not shown).

Figure 2

mF1201, but not mF1232, induces both platelet aggregation and shedding of surface GPVI, while mF1232 inhibits collagen-induced, but not CRP- or ADP-induced, platelet aggregation. (A) Platelet aggregation was monitored after the addition of the indicated concentrations of mF1201 or mF1232 into human PRP. (B) After human PRP was incubated without (none) or with the indicated concentrations of CRP, mF1232, mF1201, or mIgG for 30 minutes at 37°C, the platelets were fixed and stained with PE-labeled anti-human CD62P Ab for flow cytometry analysis. Data represent mean ± SEM of triplicate determinations. MFI, mean fluorescence intensity. (C) Human PRP was preincubated with the indicated concentrations of mF1232 for 5 minutes at 37°C and stimulated with 1 μg/ml collagen, 4 μg/ml CRP, or 5 μM ADP as indicated by arrows. Traces are representatives of 2 or 3 separate experiments from different donors. (D) Washed human platelets were treated without (control) or with 0.15 μg/ml convulxin (Cvx), 10 μg/ml mF1232, or 10 μg/ml mF1201 for 60 minutes at 37°C, and centrifuged to isolate platelet pellets from supernatants. Platelet pellets (cell lysate) and supernatants were analyzed by western blotting using rabbit anti-GPVI polyclonal Abs. Data are representatives of 3 separate experiments from different donors.

Metalloproteinase-dependent proteolytic shedding is thought to mediate the loss of GPVI expression from platelets exposed to GPVI-specific mAb or ligands (20, 21). Therefore, we examined whether either mF1201 or mF1232 induced GPVI-shedding from the platelet surface. Incubation of washed human platelets with 10 μg/ml mF1201 resulted in the loss of surface GPVI, accompanied by the appearance of its soluble cleavage product, but similar changes were not seen in platelets exposed to mF1232 (Figure 2D).

YA-Abs have been used to probe only hGPVI (6–8, 10, 12, 13, 17, 23, 25), and we tested both mF1201 and mF1232 for their ability to react with GPVI-Fc obtained from different species. We cloned cynomolgus GPVI (cGPVI), and based on amino acid sequence alignment, it is 90% identical to hGPVI. In contrast, mouse GPVI (mGPVI) is only 63% identical to hGPVI over the same region (Figure 3A). Using a competition assay, in which a signal is only obtained if there is no cross reactivity, binding of mF1201 or mF1232 to hGPVI-Fc was not inhibited by mGPVI-Fc (Figure 3, B and C). In contrast, cGPVI-Fc efficiently competed with hGPVI-Fc for binding to either mAb. These data indicate that both generated anti-GPVI mAbs did not recognize mGPVI but cGPVI was bound. To further characterize the epitopes recognized by mF1201 and mF1232, we mutated different regions of hGPVI-Fc to contain the corresponding sequence as indicated in Figure 3A. Using the same competition assay, mutation of loop 9 of the Ig-D2 and mutation of either loop 4 or 5 of the Ig-D1 were found to lose their inhibitory effects on mF1232 and mF1201 binding to hGPVI-Fc, respectively (Figure 3, B and C). These results were confirmed by a direct binding assay using WT and mutant hGPVI-Fc coated on a plate (Figure 3, D and E). Previous studies identified 2 distinct collagen-binding sites on hGPVI. One binding site includes Lys59 and is recognized by the 10B12 single-chain phage Ab (26). The other collagen-binding site is located in Ig-D1 and includes residues Gly30, Val34, and Leu36, and this site is bound by the 9O12 mAb (27). In our study, either mF1201 or mF1232 binding to hGPVI-Fc was inhibited by the mutant at loop 3, including Gly30, Val34, and Leu36 (Figure 3, B and C). However, since Lys59 is included within loop 5, we also generated a hGPVI-Fc point mutant containing the corresponding mouse residue (K59E) at this location. There was no binding of mF1201 to K59E mutant hGPVI-Fc (data not shown), suggesting that mF1201 recognizes the same collagen-binding site as 10B12 (26).

Figure 3

Binding of mF1232 and mF1201 to WT GPVI and to hGPVI mutants. (A) Amino acid alignment of the extracellular domains of GPVI from 3 species are shown. Conserved amino acids among the 3 species are indicated by stars. The Ig-D1 and -D2 regions of GPVI are underlined with solid lines and dotted lines, respectively. To construct of the extracellular portion of GPVI, a comparative molecular modeling was carried out with “Homology” (Accelrys Software Inc.) using PDB 1UGN_A and PDB 1P6F_A as a template. The regions boxed with solid lines were defined as the loop structures. The abbreviation over the box denotes each loop number. (B and C) Fifty ng/ml of peroxidase-labeled mF1232 (B) or mF1201 (C) was incubated in an hGPVI-Fc–coated immunoplate for 60 minutes at 37°C with 1 μg/ml each of the absorption antigens hGPVI, mGPVI, cGPVI, or human/mouse chimera GPVI mutants. Human/mouse chimera GPVI mutant protein is represented as “mL plus number” meaning that the amino acid sequence of loop number on hGPVI is substituted for the counter loop region of mGPVI. (D and E) Peroxidase-labeled mF1232 (D) or mF1201 (E) were incubated with hGPVI-Fc– (open circles), mL9- (filled circles), mL4- (triangles), or mL5-coated (squares) plates. After incubation, the bound peroxidase-labeled mAb was measured. Data are presented as mean ± SD of triplicate determinations.

We next wished to investigate the ability of both mAbs to immunodeplete GPVI in vivo using cynomolgus monkeys as a model. We first confirmed that both mAbs affected cynomolgus platelets in a similar manner to human platelets (data not shown). When either mAb was i.v. injected into monkeys at a dose of 0.3 mg/kg, an amount estimated to be sufficient to saturate the total surface GPVI according to previously reported GPVI expression data (28), expression of both GPVI and FcRγ was decreased within 1 to 2 days, and this was accompanied by platelet unresponsiveness to collagen but not to ADP, ristocetin and vWF, and TRAP (Figure 4, A–C, and data not shown). Additionally, significant thrombocytopenia occurred following injection of mF1201 but not mF1232 (Figure 4D).

Figure 4

Effects of mF1232 or mF1201 injected into cynomolgus monkeys on GPVI immunodepletion, collagen-induced platelet aggregation, and platelet counts. mF1232 (white bars) or mF1201 (black bars) was i.v. injected at a dose of 0.3 mg/kg into monkeys. Blood was collected on day 0 (0d), day 1 (1d), and day 2 (2d) after injection. The level of GPVI, FcRγ, and GPIIIa in platelet lysates (A) and the expression of surface GPVI (B) were analyzed by western blotting and by flow cytometry, respectively, using each specific Ab. (A) Representative data of western blotting from 3 separate experiments are shown at the top of the panel. The band density of GPVI (middle panel) or FcRγ (bottom panel) was quantified by a Typhoon Scanner 9410 and shown as percentage of that on day 0. (B) Mean fluorescence intensity was normalized to 100 for that on day 0. Maximal platelet aggregation was monitored using PRP stimulated with 2 μg/ml collagen (C), and platelet (PLT) counts in whole blood (D) were determined. Data are presented as mean ± SEM of 3 separate experiments. *P < 0.05, **P < 0.01 versus day 0.

Taken together, our data suggest that mF1232 is a more promising therapeutic agent than mF1201, an activating mAb, and we constructed a human/mouse chimera of mF1232 (cF1232) for further study. When we compared cF1232 with mF1232 in assays described in Figures 2, 3, and 4, there were no differences between these 2 reagents (data not shown). Importantly, cF1232 did not induce substantial activation of human platelets such as aggregation and CD62P expression, even at 3 mg/ml, the dose maximally tested here (data not shown).

In order to monitor the behavior of cF1232 after injection into monkeys and binding to platelet surface GPVI, we used CypHer5E, a fluorescent endocytosis probe. This dye is nonfluorescent at neutral pH, but its fluorescence increases exponentially as it passes through the acidic environment of endosomes and lysosomes, with maximum fluorescence at pH 5, the luminal lysosomal pH (29, 30). After CypHer5E-labeled cF1232 was i.v. injected into monkeys, we monitored the fluorescence of CypHer5E in circulating platelets and surface GPVI expression by flow cytometry as well as the expression of GPVI/FcRγ in whole cell lysates by immunoblotting. As shown in Figure 5A, CypHer5E fluorescence in circulating platelets rapidly increased within 1 hour after injection, and this was maintained at a maximum level thereafter. Both surface GPVI (Figure 5B) and total cellular GPVI/FcRγ complex expression (Figure 5C) began to concomitantly decrease, and surface GPVI was reduced by approximately 80% at 6 hours after injection. We wished to confirm that Ab binding and uptake was dependent upon the surface expression of GPVI, and we injected monkeys with unlabeled cF1232, at a dose of 1 mg/kg, to immunodeplete surface GPVI. Five days after this initial injection, monkeys were then treated with CypHer5E-labeled cF1232, and very low levels of fluorescence were seen at 3 hours after injection. This was much lower than the fluorescence seen in nonimmunodepleted monkeys (Figure 5, A and B).

Figure 5

CypHer5E-labeled cF1232 injected into cynomolgus monkeys is endocytosed into circulating platelets correlating with GPVI depletion. CypHer5E-labeled cF1232 or human IgG was i.v. injected at a dose of 1 mg/kg into cynomolgus monkeys. The fluorescence intensity (A) in circulating platelets and platelet surface GPVI expression (B) at the indicated time points after injection of CypHer-5E–labeled cF1232 (open circles) or human IgG (open triangles) were analyzed by flow cytometry. The levels of surface GPVI expression are normalized to 100 for day 0. The filled circles with dotted lines represent monkeys pretreated with 1 mg/kg unlabeled cF1232. GPVI and FcRγ expression in circulating platelets at the indicated time points after injection were analyzed by western blotting (C). All data are presented as mean ± SEM of 3 separate experiments. *P < 0.05, **P < 0.01 versus control groups at the same time points.

We next used monkey platelet-rich plasma (PRP) to study the behavior of cF1232 and GPVI in vitro. Incubation of PRP with CypHer5E-labeled cF1232 at 37°C led to a time-dependent increase in fluorescence corresponding to Ab endocytosis, and little increase in fluorescence was seen when samples were incubated at 4°C (Figure 6A). When CypHer5E-labeled cF1232 was incubated at 37°C with GPVI-deficient PRP that was obtained from monkeys treated with unlabeled cF1232 for 5 days as described above, CypHer5E fluorescence was not detected in platelets (data not shown), confirming that the in vitro cF1232 internalization and endocytosis was strictly dependent on surface GPVI. However, in contrast to what was seen in vivo, there was no change in surface GPVI expression during incubation at 37°C (Figure 6A).

Figure 6

In vitro internalization and endocytosis of GPVI-specific Abs with or without GPVI immunodepletion under various experimental conditions. Cynomolgus PRP was incubated with 10 μg/ml CypHer5E-labeled (Cy5E-labeled) Ab in vitro under various experimental conditions as follows: 60 minutes incubation with cF1232 (black bars) or human IgG (white bars) at 37°C or 4°C (A), incubation with cF1232 (filled circles) or human IgG (open circles) for up to 120 minutes at 37°C in the presence of 3 μM PGE₁ (B), 60 minutes incubation with mF1232, mF1201, or mouse IgG (mIgG) at 37°C in the presence of 3 μM PGE₁ (C). Washed cynomolgus platelets were treated with the indicated Abs in the absence or presence of 3 μM PGE₁ for 60 minutes at 37°C (D). (A–D) The fluorescence in platelets (left panels) and surface GPVI expression (right panels) were analyzed by flow cytometry. Surface GPVI expression before incubation was normalized to 100. (E) Washed cynomolgus platelets were treated with 0.15 μg/ml convulxin, 10 μg/ml mF1232, 10 μg/ml mF1232, or 10 μg/ml mouse IgG for 60 minutes at 37°C in the absence or presence of PGE₁ and sedimented to separate supernatants. The supernatants were analyzed by western blotting with rabbit anti-human GPVI polyclonal Ab. The band density corresponding to soluble GPVI was quantified by a Typhoon Scanner 9410, and the value for convulxin treatment in the absence of PGE₁ was normalized to 100. All results represent mean ± SEM of 3 (A, D, and E) or 4 (B and C) separate experiments. *P < 0.05, **P < 0.01 versus control groups.

We next examined whether cF1232 could be internalized and endocytosed in a platelet activation-independent manner, and PRP was pretreated with 3 μM PGE₁, a well-known potent inhibitor of platelet activation that acts by elevating intracellular cAMP. In PGE₁-treated platelets, as cF1232 was endocytosed into platelets, surface GPVI expression was lost in parallel and it decreased to approximately 30% untreated levels after 2 hours of cF1232 treatment (Figure 6B). The same experiment was also performed by replacing cF1232 with mF1201. As shown in Figure 6C, mF1201 was endocytosed and significant decrease in surface GPVI expression was observed as with mF1232. Since we found that mF1201 induced shedding of surface GPVI in the absence of PGE₁ as described in Figure 2D, we performed experiments to compare Ab endocytosis and the loss of surface GPVI between PGE₁-treated and -untreated platelets, and washed monkey platelets were incubated with CypHer5E-labeled or -unlabeled mAbs in the presence or absence of PGE₁ at 37°C (Figure 6, D and E). As shown in Figure 6D, both CypHer5E-labeled mF1232 and mF1201 were endocytosed whether or not PGE₁ was included in the reaction mixtures. As was seen with PRP, there was a significant decrease in surface GPVI expression in washed platelets incubated with CypHer5E-labeled mF1232 in the presence, but not absence, of PGE₁. In contrast, regardless of PGE₁ treatment, surface GPVI expression of washed platelets incubated with CypHer5E-labeled mF1201 was significantly decreased. We also examined the supernatants of washed platelet suspensions incubated with either mAb in the presence or absence of PGE₁. Once again, cleaved GPVI was detected by immunoblotting in the supernatants of the samples incubated with mF1201, but not with mF1232, and PGE₁ did not significantly affect this (Figure 6E).

We wished to measure an increased amount of soluble GPVI in the plasma from monkeys after mF1232 or mF1201 treatment. For this purpose, we developed a sandwich ELISA assay that could quantitatively detect injected mAb-bound and -unbound soluble GPVI. The increased amounts of soluble GPVI in the plasma from monkeys after mF1232 or mF1201 treatment at a dose of 0.3 mg/kg were assessed as described in the Methods and compared between 2 groups. As shown in Figure 7, most of plasma samples from monkeys after mF1201 treatment revealed detectable increase in soluble GPVI content above the limit of ELISA sensitivity until 4 hours after treatment but not thereafter, and the highest levels of increase in soluble GPVI content were observed 1 hour after treatment. In contrast, little increase in soluble GPVI content over the limit of sensitivity was observed in the plasma from monkeys over 24 hours after mF1232 treatment. Statistical analyses revealed that soluble GPVI content in the plasma significantly increased 1 hour after mF1201, but not mF1232, treatment and that the difference of increased amounts of soluble GPVI between those 2 groups was significant. Interestingly, surface GPVI expression levels on platelets from mF1201-treated monkeys appeared to decrease a little more rapidly than those from mF1232-treated monkeys, but those levels did not differ significantly from each other 24 hours after treatment, confirming the finding shown in Figure 4B (Figure 7, top left panel). Furthermore, although the maximal plasma concentration of either injected mAb was less than 10 μg/ml, plasma concentrations of mF1201 were found to be lower than those of mF1232 at the indicated time points after injection, suggesting more rapid clearance of mF1201-forming immune complexes with activated platelets/microaggregates or soluble GPVI (Figure 7, top right panel).

Figure 7

Comparison of increased amounts of soluble GPVI in the plasma from monkeys after F1232 or F1201 treatment. mF1232 (open symbols) or mF1201 (filled symbols) was i.v. injected at a dose of 0.3 mg/kg into monkeys. Blood was collected before injection and at the indicated time points after injection. The increased amounts of soluble GPVI in the plasma at the indicated time points after each treatment (bottom panel), the expression of surface GPVI (top left panel), and plasma concentrations of each mAb (top right panel) were assessed as described in the Methods and compared between 2 groups. The number of tested monkeys ranged from 2 to 6 at each time point. Data are expressed as the amount of cGPVI-Fc, a standard, or as mean ± SEM. The dotted bar represents the limit of sensitivity of the assay (4.9 ng/ml). *P < 0.05 versus 0 ng/ml or mF1232 versus mF1201.

In patient YA, sustained anti-GPVI Ab concentrations around 1 μg/ml were sufficient to maintain decreased GPVI surface expression, and we wished to maintain the same concentration of cF1232. Monkeys were injected with a single dose of cF1232 of 1 mg/kg either i.v. or s.c., and we followed the plasma Ab concentration over time (Figure 8A). Subcutaneous injection of cF1232 led to a sustained plasma concentration over 1 μg/ml for more than 3 days, but i.v. injection achieved this concentration for less than 2 days. Platelets isolated from injected monkeys lost surface GPVI expression within the first day, and this decrease was maintained for 7 to 10 days without significant thrombocytopenia and accompanied unresponsiveness to collagen but not ADP stimulation (Figure 8B and data not shown).

Figure 8

Pharmacokinetics and long-term antiplatelet effects after cF1232-treatment. (A) cF1232 was i.v. (filled circles) or s.c. (open circles) injected at a dose of 1 mg/kg into cynomolgus monkeys, and plasma concentrations of cF1232 were measured by ELISA. (B) After s.c. injection of cF1232, maximal platelet aggregation in response to 2 μg/ml collagen (filled circles) or 20 μM ADP (filled squares) was periodically monitored. The aggregation rate was normalized to 100% for those on day 0. The experiments were carried out with 3 monkeys and the data represent mean ± SEM.

The first patient with platelet GPVI deficiency has been followed up by our department for more than 20 years and has platelets still lacking GPVI, without thrombocytopenia, or any signs of bleeding. Although the patient has been suffering from chronic atrial fibrillation for the last 7 years, she has not received any antithrombotic medications for fear of inducing bleeding events. Additionally, she denies any signs or symptoms of thrombosis in the brain and heart (unpublished observations). The patient’s serum IgG, isolated and described in our original report, has been an invaluable tool for both the immunological detection of GPVI and studies of GPVI function. However, serum isolated more recently is much less reactive and of limited utility. These clinical and laboratory observations hinted to us that the low titer YA-Abs have generated a long-term state of GPVI immunodepletion without serious episodes of bleeding or thrombosis. Then, we characterized YA-Abs over 15 years to identify conditions compatible with long-term GPVI depletion. The concentration of anti-GPVI YA-Abs-03 was less than 10 μg/ml and 100-fold lower than comparable YA-Abs-88, suggesting that YA-Abs-03 might contain anti-GPVI Ab immunodepleting the target in low titer without causing thrombocytopenia.

We generated mouse anti-hGPVI mAbs with the hope of replicating the performance characteristics of YA-Abs. Indeed, 2 representative mAbs, mF1201 and mF1232, were effectively prevented from binding to hGPVI-Fc by YA-Abs. These mAbs did not cross-inhibit one another, and both recognized distinct epitopes on hGPVI. These data also confirm the oligoclonal nature of YA-Abs.

Upon further characterization of mF1201 and mF1232, we found striking functional differences between these mAbs tested at the concentrations similar to those of anti-GPVI Abs in the recent plasma of YA, namely at less than 10 μg/ml. Treatment of platelets with mF1201 alone resulted in platelet aggregation, activation, and the shedding of surface GPVI. GPVI shedding is thought to be mediated by metalloproteinase-dependent proteolysis, and this may occur in an activation-dependent manner (20, 21). In contrast, mF1232 induced neither significant platelet activation such as aggregation and CD62P expression nor the shedding of surface GPVI.

When we mapped the binding sites of mF1201 and of mF1232 using a systematic mutagenesis approach, we demonstrated that mF1201 recognized a previously reported collagen-binding site on hGPVI (26). In contrast, mF1232 recognized what we believe to be a novel epitope on loop 9 (Thr116-Gln122) of the Ig-D2 of hGPVI. It is interesting to consider the contrasting functional data for these 2 mAbs in the context of their binding sites. Treatment of platelets with collagen induces aggregation and activation, and it is likely that the agonistic nature of mF1201 leads to cross-linking of the collagen-binding sites of GPVI. In contrast, mF1232 inhibits collagen-induced aggregation but does not affect CRP-induced aggregation. Interestingly, Dumont et al. reported that Ig-D2 plays a specific and significant role in GPVI binding to collagen but CRP does not. Additionally, Horii et al. determined the crystal structure of the collagen-binding domain of hGPVI and suggested that Ig-D2 of the 2 GPVI molecules formed a parallel, back-to-back dimer on the platelet membrane, matching the dimensions of a collagen fiber (32). Taken together with these reports, it is tempting to speculate the following mechanisms by which mF1232 binding to loop 9 of Ig-D2 interferes with collagen-mediated, but not CRP-mediated, platelet interaction. Loop 9 of Ig-D2 might contain a heretofore undefined collagen-binding site but not a CRP-binding site. If the site of mF1232 binding in loop 9 does not participate in the back-to-back dimer formation described by Horii et al. (32), mF1232 binding to GPVI might be able to disrupt the adoption of the appropriate dimer structure. Because CRP binds monomeric GPVI, its ability to induce platelet activation might be unaffected by mF1232. These reports and our present findings suggest a novel series of experiments to better clarify the mechanism of collagen and CRP binding to GPVI that should be undertaken in the future.

The generation of 2 mAbs that bind GPVI with distinct effects on platelet activation allowed us to further investigate the mechanisms underlying Ab-induced GPVI immunodepletion using cynomolgus monkeys as a model system. First, we showed that both mAbs bound cGPVI, a protein highly homologous to hGPVI, and they affected platelet function in monkeys in a manner similar to that seen for humans. Notably, both mAbs induced GPVI immunodepletion when i.v. injected into monkeys at a low dose of 0.3 mg/kg. However, injection of mF1201 caused significant thrombocytopenia that might reflect shortened platelet life span. Most of previous studies provided evidence for the metalloproteinase-dependent cleavage of surface GPVI both in vitro and in vivo (20, 21, 33, 34). We clearly showed that 10 μg/ml mF1201, but not mF1232, induced shedding of surface GPVI in vitro, and it is likely that mF1201-induced shedding is at least partly responsible for the reductions in GPVI surface expression seen in vivo. However, there was no shedding seen in vitro following mF1232 treatment, but GPVI was rapidly lost from the platelet surface following mF1232 injection. This suggested that intracellular degradation might be responsible for reductions in GPVI expression following mF1232 treatment.

We wished to obtain direct evidence for the internalization and endocytosis of GPVI using a fluorescent endocytosis probe, CypHer5E, which becomes fluorescent in the acidic environment of intracellular organelles such as endosomes and lysosomes (29, 30). We also used a chimeric Ab based on mF1232 (cF1232) for these studies. The behavior of cF1232 toward both human and cynomolgus platelets was the same as mF1232. Importantly, cF1232 did not activate human platelets in vitro, even at a maximal test concentration of 3 mg/ml. Therefore, it is very likely that cF1232 binding to loop 9 of Ig-D2 on GPVI does not result in GPVI-mediated activation of human platelets, not only in vitro but also when injecting cF1232 into humans. When CypHer5E-labeled cF1232 was injected into monkeys, there was a dramatic GPVI-dependent increase in fluorescence in circulating platelets and there was a corresponding decrease in surface GPVI expression. These data indicate that injected cF1232 bound to surface GPVI is internalized and endocytosed into circulating platelets. A similar surface GPVI-dependent endocytosis of cF1232 was also observed when platelets were incubated with CypHer5E-labeled cF1232 at 37°C but not 4°C in vitro. However, in contrast to the results seen in vivo, there was sustained GPVI expression at the platelet surface. These results suggest that in vitro GPVI was internalized and rapidly recycled back to the platelet surface without degradation in order to maintain surface GPVI level, but an unknown mechanism was operating to downregulate surface GPVI level in vivo.

The solution came from the experiment to see whether in vitro internalization and endocytosis of CypHer5E-labeled cF1232 into platelets could occur when a cAMP-elevating agent, PGE₁, was included in the reaction mixture. In the presence of PGE₁, cF1232 was endocytosed, and there was a corresponding reduction in surface GPVI that was not shed in vitro. In vivo, the vasculature produces prostacyclin, a potent endogenous inhibitor of platelet activation that increases intracellular cAMP levels analogous to PGE₁. It is very likely that prostacyclin constitutively released from the vasculature is essential for the observed downregulation of GPVI seen in vivo, and we must recreate this in vivo effect to observe a similar phenomena in vitro. Importantly, we found a statistically significant increase in plasma concentrations of soluble GPVI in mF1201-treated, but not mF1232-treated, monkeys and also a significant difference of increase in those between 2 groups. Taken together, this is what we believe to be the first direct evidence demonstrating specific Ab-induced GPVI internalization and endocytosis both in vitro and in vivo, and it resolves the questions regarding acquired GPVI deficiency caused by specific auto-Abs (12, 19). Interestingly, mF1201-induced proteolytic shedding occurred even in the presence of PGE₁, indicating that the elevation of intracellular cAMP cannot completely inhibit the proteolytic shedding. Nevertheless, mF1201 was endocytosed to the same extent as cF1232 or mF1232 during incubation at 37°C in vitro, suggesting that mF1201 could deplete surface GPVI through both proteolytic shedding and endocytosis. With regard to receptor internalization, Nieswandt et al. showed that decreased surface GPVI expression preceded reductions in GPVI in whole platelet lysates within 24 hours after injection of the anti-mGPVI mAb JAQ-1 into WT mice (14). However, the presence of a large, internal pool of GPVI has been proven (23) and could explain their findings. Indeed, they confirmed that JAQ-1 caused proteolytic shedding of surface GPVI that was associated with the occurrence of thrombocytopenia (34). In contrast to what we directly demonstrated here using CypHer5E-labeled mAbs and WT platelets or monkeys both in vivo and in vitro, they used biotin-labeled JAQ-1 and mutant mice lacking GPVI downstream signaling molecules to show JAQ-1–induced GPVI internalization in vivo but not in vitro (34). Without using such mutant animals, we definitely showed that the Ig-D2–specific mAbs were endocytosed into platelets irrespective of GPVI downregulation in a surface GPVI-dependent manner but caused GPVI internalization and endocytosis without significant GPVI shedding both in vivo and in vitro when platelets were also exposed to a cAMP-elevating agent. Taken together with our previous report that GPVI downstream signals are cAMP-insensitive (7), the mechanisms underlying specific Ab-induced depletion of surface GPVI through internalization and endocytosis could not be simply explained by GPVI downstream signals themselves. It was of great interest that GPVI immunodepletion could be safely induced without thrombocytopenia following injection of mF1232 into WT monkeys (22). In view of the recent report that JAQ-1 treatment caused significant thrombocytopenia in WT mice but did not in the mutant mice, revealing no shedding of surface GPVI (34), it is noteworthy that injection of mAbs recognizing what we believe to be a novel epitope on Ig-D2 of hGPVI did not cause significant thrombocytopenia in WT monkeys as shown here. Clearly, anti-GPVI mAbs such as mF1201 that strongly cause thrombocytopenia are unsuitable for clinical application.

In a clinical application of cF1232, its injection via s.c. route could be more practical and useful than that via i.v. route to maintain a longer effect of GPVI immunodepletion. In fact, we found that a single s.c. injection of cF1232 into monkeys led to sustained low titer plasma concentrations and long-term antiplatelet effects due to GPVI immunodepletion without significant thrombocytopenia. Present work warrants further studies to emphasize the therapeutic promise of the cF1232 initially characterized here.

Authors thank Yumi Hotta, Kotarou Hoshida, Keiko Hoshida, and Shin Kumadaki for the preparation of chimera Ab and Masayuki Ichioka for the technical assistance with flowcytometric analysis, western blot analysis, and epitope analysis. We also thank Tarou Seo for helping with animal study and maintenance of the animals.

Address correspondence to: Hiroshi Takayama, Department of Health and Nutrition, School of Human Cultures, The University of Shiga Prefecture, 2500 Hassakacho, Hikone, Shiga 522-8533, Japan. Phone: 81-749-28-8441; Fax: 81-749-28-8519; E-mail: htakayam@shc.usp.ac.jp.

Nonstandard abbreviations used: ADP, adenosine diphosphate; cGPVI, cynomolgus GPVI; CRP, collagen-related peptide; D1, domain 1; hGPVI-Fc, hGPVI fused to human Fc; FcRγ, FcR γ-chain; GPVI, glycoprotein VI; hGPVI, human GPVI; mGPVI, mouse GPVI; PRP, platelet-rich plasma; TRAP, thrombin receptor-agonist peptide; YA-Abs, YA’s auto-Abs.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J. Clin. Invest.118:1785–1795 (2008). doi:10.1172/JCI32513

1. Bhatt, D.L., Topol, E.J. 2003. Scientific and therapeutic advances in antiplatelet therapy. Nat. Rev. Drug Discov. 2:15-28.
   View this article via: CrossRef PubMed Google Scholar
2. Jackson, S.P., Schoenwaelder, S.M. 2003. Antiplatelet therapy: in search of the ‘magic bullet’. Nat. Rev. Drug Discov. 2:775-789.
   View this article via: CrossRef PubMed Google Scholar
3. Clemetson, J.M., Polgar, J., Magnenat, E., Wells, T.N., Clemetson, K.J. 1999. The platelet collagen receptor glycoprotein VI is a member of the immunoglobulin superfamily closely related to FcαR and the natural killer receptors. J. Biol. Chem. 274:29019-29024.
   View this article via: CrossRef PubMed Google Scholar
4. Jandrot-Perrus, M., et al. 2000. Cloning, characterization, and functional studies of human and mouse glycoprotein VI: a platelet-specific collagen receptor from the immunoglobulin superfamily. Blood. 96:1798-1807.
   View this article via: PubMed Google Scholar
5. Ezumi, Y., Uchiyama, T., Takayama, H. 2000. Molecular cloning, genomic structure, chromosomal localization, and alternative splice forms of the platelet collagen receptor glycoprotein VI. Biochem. Biophys. Res. Commun. 277:27-36.
   View this article via: CrossRef PubMed Google Scholar
6. Tsuji, M., Ezumi, Y., Arai, M., Takayama, H. 1997. A novel association of Fc receptor γ-chain with glycoprotein VI and their co-expression as a collagen receptor in human platelets. J. Biol. Chem. 272:23528-23531.
   View this article via: CrossRef PubMed Google Scholar
7. Ichinohe, T., et al. 1995. Cyclic AMP-insensitive activation of c-Src and Syk protein-tyrosine kinases through platelet membrane glycoprotein VI. J. Biol. Chem. 270:28029-28036.
   View this article via: CrossRef PubMed Google Scholar
8. Ichinohe, T., et al. 1997. Collagen-stimulated activation of Syk but not c-Src is severely compromised in human platelets lacking membrane glycoprotein VI. J. Biol. Chem. 272:63-68.
   View this article via: CrossRef PubMed Google Scholar
9. Poole, A., et al. 1997. The Fc receptor γ-chain and the tyrosine kinase Syk are essential for activation of mouse platelets by collagen. EMBO J. 16:2333-2341.
   View this article via: CrossRef PubMed Google Scholar
10. Ezumi, Y., Shindoh, K., Tsuji, M., Takayama, H. 1998. Physical and functional association of the Src family kinases Fyn and Lyn with the collagen receptor glycoprotein VI-Fc receptor γ-chain complex on human platelets. J. Exp. Med. 188:267-276.
    View this article via: CrossRef PubMed Google Scholar
11. Martin, A.M., Kulski, J.K., Witt, C., Pontarotti, P., Christiansen, F.T. 2002. Leukocyte Ig-like receptor complex (LRC) in mice and men. Trends Immunol. 23:81-88.
    View this article via: CrossRef PubMed Google Scholar
12. Sugiyama, T., et al. 1987. A novel platelet aggregating factor found in a patient with defective collagen-induced platelet aggregation and autoimmune thrombocytopenia. Blood. 69:1712-1720.
    View this article via: PubMed Google Scholar
13. Moroi, M., Jung, S.M., Okuma, M., Shinmyozu, K. 1989. A patient with platelets deficient in glycoprotein VI that lack both collagen-induced aggregation and adhesion. J. Clin. Invest. 84:1440-1445.
    View this article via: JCI CrossRef PubMed Google Scholar
14. Nieswandt, B., et al. 2001. Long-term antithrombotic protection by in vivo depletion of platelet glycoprotein VI in mice. J. Exp. Med. 193:459-470.
    View this article via: CrossRef PubMed Google Scholar
15. Zheng, Y.M., et al. 2001. Expression of the platelet receptor GPVI confers signaling via the Fc receptor γ-chain in response to the snake venom convulxin but not to collagen. J. Biol. Chem. 276:12999-13006.
    View this article via: CrossRef PubMed Google Scholar
16. Massberg, S., et al. 2003. A crucial role of glycoprotein VI for platelet recruitment to the injured arterial wall in vivo. J. Exp. Med. 197:41-49.
    View this article via: CrossRef PubMed Google Scholar
17. Goto, S., et al. 2002. Involvement of glycoprotein VI in platelet thrombus formation on both collagen and von Willebrand factor surfaces under flow conditions. Circulation. 106:266-272.
    View this article via: CrossRef PubMed Google Scholar
18. Schulte, V., et al. 2003. Targeting of the collagen-binding site on glycoprotein VI is not essential for in vivo depletion of the receptor. Blood. 101:3948-3952.
    View this article via: CrossRef PubMed Google Scholar
19. Boylan, B., et al. 2004. Anti-GPVI-associated ITP: an acquired platelet disorder caused by autoantibody-mediated clearance of the GPVI/FcRγ-chain complex from the human platelet surface. Blood. 104:1350-1355.
    View this article via: CrossRef PubMed Google Scholar
20. Gardiner, E.E., Arthur, J.F., Kahn, M.L., Berndt, M.C., Andrews, R.K. 2004. Regulation of platelet membrane levels of glycoprotein VI by a platelet-derived metalloproteinase. Blood. 104:3611-3617.
    View this article via: CrossRef PubMed Google Scholar
21. Stephens, G., et al. 2005. Platelet activation induces metalloproteinase-dependent GP VI cleavage to down-regulate platelet reactivity to collagen. Blood. 105:186-191.
    View this article via: CrossRef PubMed Google Scholar
22. Takayama, H., et al. 2005. Characterization of anti-human GPVI autoantibodies mediating persistent target immnunodepletion over a 15-year period [abstract]. Blood. 106:612a.
23. Suzuki, H., Murasaki, K., Kodama, K., Takayama, H. 2003. Intracellular localization of glycoprotein VI in human platelets and its surface expression upon activation. Br. J. Haematol. 121:904-912.
    View this article via: CrossRef PubMed Google Scholar
24. Kehrel, B., et al. 1998. Glycoprotein VI is a major collagen receptor for platelet activation: it recognizes the platelet-activating quaternary structure of collagen, whereas CD36, glycoprotein IIb/IIIa, and von Willebrand factor do not. Blood. 91:491-499.
    View this article via: PubMed Google Scholar
25. Ezumi, Y., Kodama, K., Uchiyama, T., Takayama, H. 2002. Constitutive and functional association of the platelet collagen receptor glycoprotein VI-Fc receptor γ-chain complex with membrane rafts. Blood. 99:3250-3255.
    View this article via: CrossRef PubMed Google Scholar
26. Smethurst, P.A., et al. 2004. Identification of the primary collagen-binding surface on human glycoprotein VI by site-directed mutagenesis and by a blocking phage antibody. Blood. 103:903-911.
    View this article via: CrossRef PubMed Google Scholar
27. Lecut, C., et al. 2004. Identification of residues within human glycoprotein VI involved in the binding to collagen: evidence for the existence of distinct binding sites. J. Biol. Chem. 279:52293-52299.
    View this article via: CrossRef PubMed Google Scholar
28. Jandrot-Perrus, M., Lagrue, A.H., Okuma, M., Bon, C. 1997. Adhesion and activation of human platelets induced by convulxin involve glycoprotein VI and integrin α2β1. J. Biol. Chem. 272:27035-27041.
    View this article via: CrossRef PubMed Google Scholar
29. Adie, E.J., et al. 2002. A pH-sensitive fluor, CypHer 5, used to monitor agonist-induced G protein-coupled receptor internalization in live cells. Biotechniques. 33:1152-1157.
    View this article via: PubMed Google Scholar
30. Adie, E.J., et al. 2003. CypHer 5: a generic approach for measuring the activation and trafficking of G protein-coupled receptors in live cells. Assay Drug Dev. Technol. 1:251-259.
    View this article via: CrossRef PubMed Google Scholar
31. Dumont, B., et al. 2006. Chimeric Fc receptors identify ligand binding regions in human glycoprotein VI. J. Mol. Biol. 361:877-887.
    View this article via: CrossRef PubMed Google Scholar
32. Horii, K., Kahn, M.L., Herr, A.B. 2006. Structural basis for platelet collagen responses by the immune-type receptor glycoprotein VI. Blood. 108:936-942.
    View this article via: CrossRef PubMed Google Scholar
33. Boylan, B., Berndt, M.C., Kahn, M.L., Newman, P.J. 2006. Activation-independent, antibody-mediated removal of GPVI from circulating human platelets: development of a novel NOD/SCID mouse model to evaluate the in vivo effectiveness of anti-human platelet agents. Blood. 108:908-914.
    View this article via: CrossRef PubMed Google Scholar
34. Rabie, T., et al. 2007. Diverging signaling events control the pathway of GPVI downregulation in vivo. Blood. 110:529-535.
    View this article via: CrossRef PubMed Google Scholar
35. Morton, L.F., Hargreaves, P.G., Farndale, R.W., Young, R.D., Barnes, M.J. 1995. Integrin α2β1-independent activation of platelets by simple collagen-like peptides: collagen tertiary (triple-helical) and quaternary (polymeric) structures are sufficient alone for a2β1-independent platelet reactivity. Biochem. J. 306:337-344.
    View this article via: PubMed Google Scholar
36. Mizushima, S., Nagata, S. 1990. pEF-BOS, a powerful mammalian expression vector. Nucleic Acids Res. 18:5322.
    View this article via: PubMed Google Scholar
37. Sambrook, J., and Russel, D.W. 2001. Site-specific mutagenesis by overlap extension. In Molecular cloning: a laboratory manual. 3rd edition. Cold Spring Harbor Laboratory Press. Cold Spring Harbor, New York, USA. 13.36–13.39.
38. Nakane, P.K. 1975. Recent progress in the peroxidase-labeled antibody method. Ann. N. Y. Acad. Sci. 254:203-211.
    View this article via: CrossRef PubMed Google Scholar
39. Takayama, H., et al. 1991. Ionophore A23187-induced protein-tyrosine phosphorylation of human platelets: possible synergism between Ca2+ mobilization and protein kinase C activation. Biochem. Biophys. Res. Commun. 174:922-927.