Research ArticlePhysiology

The collagen receptor glycoprotein VI promotes platelet-mediated aggregation of β-amyloid

See allHide authors and affiliations

Science Signaling  04 Aug 2020:
Vol. 13, Issue 643, eaba9872
DOI: 10.1126/scisignal.aba9872

Unclogging amyloid from arteries

Cerebral amyloid angiopathy (CAA) is a common feature of Alzheimer’s disease in which the deposition of amyloid in vessels in the brain impairs blood flow. Amyloid can activate platelets, which promotes amyloid aggregation. Using platelets from patients and mice, Donner et al. found that blocking the collagen receptor GPVI may reduce CAA. Amyloid bound to GPVI on platelets and stimulated the release of fibrinogen that both clustered with soluble amyloid and bound a platelet-surface integrin that further promoted amyloid aggregation. Blockade of GPVI or the integrin reduced amyloid aggregation in platelet cultures, and loss of GPVI reduced the adhesion of amyloid-activated platelets in arteries in mice, revealing a potential therapeutic target to ameliorate CAA in patients.

Abstract

Cerebral amyloid angiopathy (CAA) and β-amyloid (Aβ) deposition in the brain parenchyma are hallmarks of Alzheimer’s disease (AD). We previously reported that platelets contribute to Aβ aggregation in cerebral vessels by secreting the factor clusterin upon binding of Aβ40 to the fibrinogen receptor integrin αIIbβ3. Here, we investigated the contribution of the collagen receptor GPVI (glycoprotein VI) in platelet-induced amyloid aggregation. Using platelets isolated from GPVI–wild type and GPVI-deficient human donors and mice, we found that Aβ40 bound to GPVI, which induced the release of ATP and fibrinogen, resulting in platelet aggregation. Binding of Aβ40 to integrin αIIbβ3, fibrinogen, and GPVI collectively contributed to the formation of amyloid clusters at the platelet surface. Consequently, blockade of αIIbβ3 or genetic loss of GPVI reduced amyloid fibril formation in cultured platelets and decreased the adhesion of Aβ-activated platelets to injured carotid arteries in mice. Application of losartan to inhibit collagen binding to GPVI resulted in decreased Aβ40-stimulated platelet activation, factor secretion, and platelet aggregation. Furthermore, the application of GPVI- or integrin-blocking antibodies reduced the formation of platelet-associated amyloid aggregates. Our findings indicate that Aβ40 promotes platelet-mediated amyloid aggregation by binding to both GPVI and integrin αIIbβ3. Blocking these pathways may therapeutically reduce amyloid plaque formation in cerebral vessels and the brain parenchyma of patients.

INTRODUCTION

In 2015, there were more than 47 million people living with dementia worldwide. With increasing age and the lack of effective therapeutic strategies, this number is projected to rapidly increase, reaching 135 million people by 2050 (1, 2). Alzheimer’s disease (AD) is the most frequent cause of dementia, accounting for 60% of dementia cases (3). The pathological hallmarks of AD are elevated misfolding; oligomerization and aggregation of β-amyloid (Aβ) peptides in brain parenchyma and in the cerebral vessels, known as cerebral amyloid angiopathy (CAA); and accumulation of intracellular neurofibrillary tangles in neurons (4, 5). The consequences are neurodegeneration with synaptic and neuronal loss, leading to brain atrophy (6, 7).

Several studies indicate that vascular damage and dysfunction, including reduction of cerebral blood flow (CBF), CAA, and blood-brain barrier (BBB) disturbances, contribute to the onset and progression of AD (8). Vascular risk factors such as atherosclerosis, stroke, hypertension, and diabetes lead to vascular damage and are associated with AD. However, whether the processes in the vasculature initiate the pathologic process of Aβ aggregation is still uncertain. Identifying the mechanisms underlying vascular pathophysiology that contribute to neurodegeneration in AD will help identify novel therapeutic targets.

Besides the role of platelets in thrombus formation during hemostasis, it is becoming clear that platelets play a crucial role in a number of other processes within the vasculature such as angiogenesis, inflammation, and cancer (911). Moreover, alterations in platelet function are also observed in diverse neurological diseases such as Parkinson’s disease, schizophrenia, autism, and AD (1215). A higher baseline expression of platelet activation biomarkers was measured in patients with AD (16), and the analysis of the Alzheimer mouse model APP23 showed that these mice have a pro-thrombotic phenotype (17, 18). Moreover, APP23 mice develop CAA and exhibit platelet accumulation at vascular plaques, leading to the reduction of CBF and probably to occlusion of cerebral vessels (18). The ability of platelets to modulate soluble, synthetic Aβ40 into fibrillar Aβ in vitro indicates a direct impact of platelets in the aggregation property of Aβ40 peptides (1820). Previously, we demonstrated that platelets contribute to Aβ aggregation through the binding of Aβ40 to the fibrinogen receptor integrin αIIbβ3, leading to outside-in signaling in platelets (19, 20), and we found that the inhibition of integrin αIIbβ3 on the surface of platelets prevents the aggregation of Aβ40 in cultured cells (19). An important indication of the involvement of platelets in Aβ aggregation in vivo was evident through the treatment of APP23 mice with the antiplatelet agent clopidogrel, a P2Y12 antagonist. Clopidogrel reduced the incidence of CAA with less adherent platelets at vascular Aβ deposits in transgenic AD mice (19). Here, we investigated the involvement of other receptors on the surface of platelets and uncovered a critical role for the collagen receptor glycoprotein VI (GPVI) in platelet-mediated aggregation of Aβ.

RESULTS

Phosphorylation of tyrosine residues in LAT and other proteins by GPVI and integrin αIIbβ3 in response to Aβ40 stimulation

Binding of collagen to GPVI leads to a series of downstream signals in platelets, resulting in phosphorylation and activation of various signaling proteins, including the adaptor protein LAT (linker of activated T cells) (21). The stimulation of human platelets with soluble Aβ40 induced a similar pattern of tyrosine phosphorylation compared to the stimulation of platelets with collagen-related peptide (CRP) as shown by Western blot analysis (Fig. 1A). Aβ40 also induced the phosphorylation of LAT (Fig. 1B) and in a time-dependent manner, with a maximum abundance detected at 90 s of incubation (fig. S1). Previously, we demonstrated that binding of Aβ40 to integrin αIIbβ3 induces integrin outside-in signaling. To exclude this effect, we performed studies in the presence of the human blocking integrin αIIbβ3 antibody abciximab. Blocking of integrin αIIbβ3 decreased the phosphorylation of tyrosine (Fig. 1A) and LAT (Fig. 1B) induced by Aβ40. In addition, we used platelets from patients who lack the GPVI receptor. Human GPVI-deficient platelets did not show phosphorylation of LAT, neither after stimulation with collagen nor by Aβ40 stimulation (Fig. 1C). However, Aβ40-induced phosphorylation of LAT was higher than upon activation with low concentration of collagen. Moreover, we studied the effect of Aβ40 in mouse platelets deficient in GPVI. Compared to wild-type (WT) platelets, GPVI-deficient platelets failed to induce phosphorylation of LAT both upon CRP and Aβ40 activation (Fig. 1D). These results indicate the ability of Aβ40 to activate GPVI in human and mouse platelets.

Fig. 1 Aβ40 stimulates tyrosine and LAT phosphorylation in a GPVI- and integrin αIIbβ3–dependent manner.

(A) Western blotting for tyrosine phosphorylation (antibody 4G10) in isolated human platelets at rest or upon stimulation with collagen-related peptide (CRP; 5 μg/ml) or Aβ40 (20 μM) for 120 s. In lane 4, as indicated, cultures were pretreated with the integrin αIIbβ3 antibody abciximab (0.5 μg per 1 Mio cell) for 15 min at room temperature. β-Actin served as loading control; n = 5 donors. (B) Western blotting for LAT phosphorylation in human platelets treated as described in (A) for 30 or 120 s. Total LAT served as loading control; n = 5 donors. (C) Western blotting for LAT phosphorylation in platelets isolated from a control donor and a GPVI-deficient patient; cells were unperturbed (resting) or stimulated with collagen (1 μg/ml) or Aβ40 (30 μM). α-Tubulin served as loading control; n = 2 GPVI-deficient patients and n = 2 healthy controls. (D) Western blotting for LAT phosphorylation in isolated Gp6−/− and WT platelets stimulated with CRP (5 μg/ml) or Aβ40 (20 μM) for 30 or 120 s. α-Tubulin served as loading control; n = 6 to 7 mice per group.

Reduced ATP release of GPVI-deficient platelets in response to Aβ40

Activation of GPVI through collagen induces platelet activation, leading to secretion of granules, inside-out signaling of integrin αIIbβ3, and platelet aggregation (22). To study the consequence of GPVI activation through Aβ40, we measured the release of adenosine 5′-triphosphate (ATP) upon Aβ40 stimulation in human platelets. Previous studies showed that losartan inhibits collagen-induced platelet aggregation through GPVI (23, 24). Therefore, we analyzed the effects of losartan treatment on Aβ40-induced platelet stimulation. Aβ40 induced the release of ATP; however, the amount of ATP was lower compared to CRP stimulation of platelets (Fig. 2, A and B). To test whether Aβ40-induced ATP release is altered by losartan, platelets were preincubated with losartan. The release of ATP was reduced by losartan after stimulation of platelets with either CRP or Aβ40 (Fig. 2, A and B). WT and GPVI-knockout mice were analyzed to confirm that Aβ40 induces a release of ATP via GPVI. The release of ATP in response to CRP or Aβ40 was significantly reduced using GPVI-deficient platelets compared to WT controls. Blocking of integrin αIIbβ3 using the Leo.H4 antibody in WT platelets resulted in significantly reduced ATP release as well (Fig. 2C). Thus, Aβ40-induced release of ATP is mediated by GPVI and integrin αIIbβ3. Therefore, blocking of integrin αIIbβ3 in WT platelets reduced ATP release to resting levels (Fig. 2C).

Fig. 2 Reduced ATP release of GPVI-deficient platelets in response to Aβ40.

(A and B) Representative ATP release curves (A) and analysis (B) in human platelets pretreated (gray traces) with losartan (100 μM) for 20 min and then stimulated with CRP (5 μg/ml) or Aβ40 (20 μM). Data are means ± SEM from n = 5 donors; two-way ANOVA with Bonferroni’s multiple comparison post hoc test: **P ≤ 0.01 and ***P ≤ 0.001. (C) ATP release in WT and Gp6−/−platelets at rest, treated with CRP (5 μg/ml), Aβ40 (10 μM), or Aβ40 pretreated with the integrin αIIbβ3-blocking antibody Leo.H4 for 20 min. Data are means ± SEM from six to seven mice per group; CRP-stimulated platelets served as controls; two-way ANOVA with Bonferroni’s multiple comparison post hoc test: *P < 0.05 and **P < 0.01.

Strongly reduced aggregation of GPVI-deficient platelets in response to Aβ40

Next, we analyzed platelet aggregation after Aβ40 stimulation. Aβ40-induced platelet aggregation was comparable to that induced by CRP (Fig. 3A). In addition, we analyzed the effect of losartan on Aβ40-induced platelet aggregation (Fig. 3, A and B). In agreement with reported data, losartan significantly inhibited CRP- and Aβ40-induced platelet aggregation (Fig. 3, A and B). To confirm the role of GPVI in Aβ40-induced platelet aggregation, we used platelets from patients with GPVI deficiency. As expected, these platelets showed no platelet aggregation in response to Aβ40 compared to platelets from healthy controls (Fig. 3, C and D). Moreover, we analyzed platelets from WT and GPVI-deficient mice. The aggregation response of WT mouse platelets with Aβ40 was comparable to CRP-induced platelet aggregation (Fig. 3, E and F). As expected, GPVI-deficient mouse platelets showed no aggregation upon CRP stimulation. Platelet aggregation upon Aβ40 stimulation was reduced in GPVI-deficient platelets compared to WT platelets (Fig. 3F). In contrast to CRP stimulation, we still measured a slight platelet aggregation of GPVI-deficient platelets in response to Aβ40, suggesting that Aβ40 can induce platelet aggregation without GPVI. These results demonstrated that the activation of GPVI by Aβ40 binding induced platelet aggregation.

Fig. 3 Reduced aggregation of GPVI-deficient platelets in response to Aβ40.

(A and B) Representative aggregation curves (A) and the quantified maximum aggregation of platelets treated with CRP (5 μg/ml) or Aβ40 (20 μM) in the presence or absence of losartan (100 μM pretreatment for 20 min). Data are means ± SEM from n = 3 to 4 donors; two-way ANOVA with Bonferroni’s multiple comparison post hoc test: ***P < 0.001. (C and D) As described in (A) and (B) in control donor and GPVI-deficient patient platelets stimulated with Aβ40 (30 μM). Data are means and range from two GPVI-deficient patients and two healthy controls. (E) As described in (A) and (B) in platelets isolated from WT and Gp6−/−mice and treated with CRP (5 μg/ml) or Aβ40 (10 μM). Data are means ± SEM from n = 5 mice per group; two-way ANOVA with Bonferroni’s multiple comparison post hoc test: ***P < 0.001.

Decreased amyloid aggregate formation by GPVI inhibition or genetic deletion in vitro

In our previous study, we showed that platelets are able to modulate soluble Aβ40 to fibrillar Aβ aggregates, whereas blocking of integrin αIIbβ3 on the surface of platelets prevents Aβ aggregate formation (19). To investigate a role of GPVI in platelet-mediated Aβ aggregate formation, we pretreated human platelets with losartan and incubated with soluble, synthetic Aβ40 for 3 days. The formation of fibrillar Aβ aggregates was analyzed by Congo red staining. Although Aβ40-induced platelet aggregation and ATP release were reduced in the presence of losartan, we did not observe alterations in Aβ aggregation in platelet cell culture (Fig. 4A). Neither daily addition nor different concentrations of losartan were able to reduce fibrillar Aβ aggregate formation (fig. S2, A to C).

Fig. 4 Inhibition or genetic deletion of GPVI decreases amyloid aggregate formation.

(A) Isolated human platelets were incubated with Aβ40 (5 μM) at 37°C for 3 days. Afterward, amyloid aggregates were stained by Congo red. Representative pictures of Congo red–stained amyloid aggregates platelet culture in the presence or absence of losartan (100 μM). Scale bar, 50 μm; n = 5 experiments. (B) Samples of murine platelets treated as in (A) were stained with Congo red to visualize amyloid aggregates in the presence of either the GPVI-blocking antibody JAQ1 or the integrin αIIbβ3-blocking antibody Leo.H4 (each at 6 μg per 2 × 106 cells). Scale bar, 50 μm; n = 5 experiments. (C and D) Corresponding Western blotting and quantification of soluble Aβ in supernatants from murine platelets cultured as in (B). Leo.H4-treated platelets served as control. Data are means ± SEM from n = 5 experiments; Student’s t test, *P ≤ 0.05. (E) Congo red staining of amyloid aggregates in cultures of platelets from WT and Gp6−/−mice without Aβ40 and with Aβ40 (5 μM) in the presence or absence of Leo.H4 (6 μg per 2 × 106cells). Scale bar, 50 μm; n = 5 mice per group. (F and G) Representative Western blots (F) and quantification (G) of soluble Aβ in supernatants from WT and GPVI−/− murine platelet cultures of remaining soluble Aβ. Controls lacking Aβ40 (lane 1 in the blot) were not regarded in the analysis. Data are means ± SEM from n = 5 mice per group; two-way ANOVA with Bonferroni’s post hoc test, *P < 0.05.

Because losartan is not a specific GPVI inhibitor, mouse platelet experiments were performed where GPVI was blocked by antibody treatment with JAQ1. In addition, the antibody Leo.H4 was used to block integrin αIIbβ3 to confirm the essential role of integrin αIIbβ3 in Aβ aggregate formation. The formation of Aβ aggregates was completely inhibited by blocking of integrin αIIbβ3 and strongly reduced by blocking of GPVI (Fig. 4B). The inhibitory effect of GPVI blockage in platelet cell culture was dose dependent (fig. S3A). The quantification of remaining soluble Aβ40 in the supernatants of platelet cell culture by Western blot analysis showed significantly increased amounts of Aβ40 when GPVI was blocked compared to untreated platelets, consistent with reduced Aβ aggregate formation (Fig. 4, C and D). To confirm these results, we used platelets from GPVI-deficient mice for cell culture experiments. Cultures of platelets from GPVI-knockout mice displayed markedly reduced Aβ aggregate formation (Fig. 4E). In the supernatants of GPVI-deficient platelets, we measured significantly increased amounts of soluble Aβ40 compared to those from WT platelets (Fig. 4, F and G). The additional blocking of integrin αIIbβ3 by the antibody Leo.H4 led to increasing amounts of soluble Aβ40 in the supernatant compared to GPVI deficiency alone and to complete inhibition of Aβ aggregates in cell culture (Fig. 4, E to G).

Direct binding of Aβ40 to GPVI

To elucidate the mechanisms by which Aβ40 peptides induce GPVI activation, we investigated the interaction between GPVI and Aβ40. First, we used the microarray AVEXIS (avidity-based extracellular interaction screen) screening assay (25). No binding with a control protein (CD200R-BLH) but direct binding of pentameric GPVI to Aβ40 peptides was observed (Fig. 5, A and B). In a second approach, we confirmed the interaction of both proteins by the use of immobilized magnetic beads coated with recombinant GPVI and incubated with soluble Aβ40. After pulldown, the association was visualized by Western blotting using antibodies to GPVI and Aβ (Fig. 4C). When Aβ40 was passed through GPVI-bound beads, a large amount of Aβ was detected along with GPVI. To verify the interaction between GPVI on platelets and Aβ40 in vitro, we incubated murine platelets with Aβ40 peptides and immunoprecipitated GPVI with the antibody JAQ1. Western blot analysis demonstrated that Aβ peptides were coimmunoprecipitated with GPVI (Fig. 4D). To show the relevance of GPVI for Aβ binding to platelets, platelets from GPVI-deficient and WT mice were incubated with Aβ40 peptides. Using flow cytometry, binding of Aβ40 to platelets was detected by fluorescein isothiocyanate (FITC)–labeled Aβ antibody (Fig. 4E). Binding of Aβ to GPVI-deficient platelets was significantly reduced compared to WT platelets. In addition, binding of Aβ to platelets was increased upon stimulation with both CRP and soluble Aβ40 and significantly reduced by integrin αIIbβ3 blocking in WT platelets. This might be due to an increased number of integrins at the platelet surface after CRP stimulation that allows augmented Aβ40 binding to integrin αIIbβ3.

Fig. 5 Aβ40 binds to GPVI.

(A) Interaction screening using AVEXIS (avidity-based extracellular interaction screen). Biotinylated bait peptides Aβ40 (CD200R-BLH is used as control) are arrayed on the surface of a streptavidin-coated plate and incubated with pentameric prey protein s5-GPVI (s5-CD200 is used as control). Interaction produces a color change to red. (B) Corresponding quantification of the colorimetric change after prey binding at 485 nm as represented in (A). n = 5 experiments; two-way ANOVA with Bonferroni’s post hoc test, ***P < 0.001 (C) Pulldown was accomplished using immobilized GPVI magnetic beads and incubated without and with Aβ40 (20 μM). Uncoated beads served as control. Immunoprecipitates were blotted against Aβ (6E10) and GPVI. Input = cell lysate. n = 3 experiments. (D) Isolated platelets were stimulated with Aβ40 (20 μM) and immunoprecipitated with GPVI antibody. Immunoprecipitates were analyzed via Western blotting against Aβ and GPVI. Representative of n = 3. (E) WT and Gp6−/− platelets were preincubated with Aβ40 (5 μM), followed by an incubation with anti–Aβ-FITC antibody. When indicated, platelets were pretreated with integrin αIIbβ3-blocking antibody Leo.H4. Binding of Aβ to platelet surface was measured by flow cytometry (n = 9 to 13 mice per group; mean ± SEM; one-way ANOVA with Dunnett’s post hoc test within each group: *P < 0.05 and **P < 0.01).

Release of fibrinogen through Aβ-induced GPVI activation and colocalization of fibrinogen with Aβ aggregates

The most abundant of platelet secretory granules are α-granules, which contain about 300 proteins, including von Willebrand factor, integrin αIIbβ3, and fibrinogen (26). The release of the α-granule content is important for all platelet functions, including hemostasis, inflammation, and angiogenesis (27). In a previously reported study, we showed that monomeric and oligomeric Aβ40 bound to fibrinogen and concluded that fibrinogen bridges Aβ/integrin αIIbβ3 complexes of platelets and contributes to the occlusion of cerebral vessels in APP23 mice, an AD model (19). Thus, we analyzed the release of fibrinogen from platelet α-granules upon Aβ40 stimulation. The release of fibrinogen from platelets in response to Aβ40 was increased and inhibited by losartan, comparable to blocking of integrin αIIbβ3 (Fig. 6A). In the presence of GPVI-blocking (JAQ1) or αIIbβ3 integrin–blocking (Leo.H4) antibodies, the release of fibrinogen was strongly reduced in response to Aβ40 (Fig. 6B). To characterize the impact of released fibrinogen on the formation of Aβ aggregates, we incubated murine platelets with Aβ40 for 3 days and analyzed fibrinogen and Aβ localization by immunofluorescence staining, which revealed that fibrinogen and Aβ aggregates colocalized (Fig. 6C). We also observed colocalization of Aβ aggregates and fibrinogen in cultures of human platelets (fig. S4A). Blocking of active factor X with the selective inhibitor arixtra did not alter binding of fibrinogen or amyloid fibril aggregate formation, suggesting that the conversion of fibrinogen to fibrin did not play a role in platelet-mediated amyloid fibril aggregate formation (fig. S5). The inhibition of GPVI on platelets not only led to reduced aggregation of Aβ but also to less accumulation of fibrinogen in cell culture (Fig. 6C, middle). In addition, the inhibition of integrin αIIbβ3 by blocking antibodies prevented the formation of Aβ aggregates and the accumulation of fibrinogen in cell culture (Fig. 6C, bottom). To confirm the impact of GPVI on the release of fibrinogen upon Aβ40 stimulation of platelets, we used platelets from WT and GPVI-deficient mice. Western blot analysis revealed that GPVI-deficient platelets did not release fibrinogen neither upon stimulation with Aβ40 nor in response to the GPVI agonist CRP (Fig. 6D). Reduced formation of Aβ aggregates was accompanied by reduced fibrinogen in cell culture using GPVI-deficient platelets compared to WT controls (Fig. 6E). Together, these results suggested that Aβ40 induced the release of fibrinogen from platelets via GPVI and the released fibrinogen colocalized with Aβ aggregates in cell culture.

Fig. 6 Fibrinogen release in response to Aβ and colocalization with amyloid aggregates in cell culture.

(A) Western blotting of fibrinogen release in murine platelets upon stimulation with CRP (5 μg/ml) or Aβ40 (20 μM) [pretreated as indicated with the integrin αIIbβ3-blocking antibody Leo.H4 (0.5 μg per 1 Mio cell) or with losartan (100 μM)]. β-Actin served as loading control; representative images of n = 3 experiments. (B) Western blotting of fibrinogen release in murine platelets as described in (A) pretreated with the GPVI-blocking antibody JAQ1 and with the integrin αIIbβ3-blocking antibody Leo.H4 (each 0.5 μg per 1 Mio cell). Representative Western blot of n = 3 experiments. (C) Murine platelets were incubated with Aβ40 (5 μM) at 37°C for 3 days. When indicated, platelets were pretreated with the GPVI-blocking antibody JAQ1 or the integrin αIIbβ3 antibody Leo.H4. Immunostaining against Aβ aggregates (green) and fibrinogen (red) visualizes colocalization. Scale bar, 20 μm. Representative images of n = 3 experiments. (D) Western blot analysis of fibrinogen release in platelets from WT and Gp6−/− mice upon stimulation with CRP (5 μg/ml) or Aβ40 (20 μM). When indicated, platelets were pretreated with integrin αIIbβ3 antibody Leo.H4. Representative images of n = 3 mice per group. (E) Platelets from WT and Gp6−/−mice were incubated with Aβ40 (5 μM) at 37°C for 3 days in the presence and absence of the integrin αIIbβ3-blocking antibody Leo.H4. Immunofluorescence staining of fibrinogen (red) and Aβ (green). Scale bar, 20 μm. Representative images of n = 3 mice per group.

Reduced Aβ-induced platelet adhesion in vivo by blocking or genetically deleting GPVI

Platelet adhesion to vascular Aβ plaques in cerebral vessels of transgenic AD model mice and enhanced Aβ40–triggered platelet adhesion at the injured vessels of WT mice in vivo were shown in a previous study (18). To explore the inhibitory effects of losartan on Aβ40-enhanced platelet adhesion at the vessel in vivo, we analyzed platelet adhesion at the injured carotid artery by in vivo fluorescence microscopy. Platelets from donor mice were stained with CellTracker Red and activated with Aβ40 in the absence or presence of losartan (Fig. 7A and movies S1 and S2). As expected, Aβ40-induced tethering and stable adhesion of platelets at sites of injury in recipient mice were observed (Fig. 7, A to C). In contrast, treatment of donor platelets with Aβ and losartan led to a statistically significant reduction of tethered (Fig. 7B) and stable adherent (Fig. 7C) platelets at the injured vessel in recipient mice.

Fig. 7 Blocking or deletion of GPVI reduces Aβ-induced platelet adhesion in vivo.

(A) Images of stable adherent and tethering Aβ40-activated platelets 20 min after carotid artery ligation in vivo. WT platelets were either incubated only with Aβ40 (top row) or were pretreated with losartan then incubated with Aβ40 (middle row), and Gp6−/− platelets were incubated with only Aβ40. Carotid artery vessel wall is outlined using dotted lines. Arrows indicate adherent platelets. n = 4 to 5 mice per group. Scale bar, 100 μm. (B and C) Quantification of tethering (B) and stable adherent (C) Aβ40-activated platelets. Data are means ± SEM; n = 4 to 5 mice per group. One-way ANOVA with Dunnett’s post hoc test: *P ≤ 0.05 and **P ≤ 0.01. (D) Tentative schematic illustration. Direct binding of Aβ40 to the collagen receptor GPVI (“1”) initiates phosphorylation of LAT (“2”), leading to secretion of granules and thus to the release of ATP, ADP, and fibrinogen (“3”). Activation of GPVI and binding of ADP to the P2Y12 receptor induces a shifting of integrin αIIbβ3 from a closed (inactive) to open (active) form leading to enhanced binding of Aβ to integrin αIIbβ3. Released fibrinogen bridges binding of soluble Aβ to GPVI and integrin αIIbβ3 to induce the formation of amyloid aggregates at the platelet surface (“4”).

To confirm an important role of GPVI on Aβ40-triggered platelet adhesion at the injured vessel in vivo, we used platelets from donor mice lacking GPVI. In vivo fluorescence imaging of platelet adhesion at sites of injury in WT recipient mice showed reduced adhesion of Aβ40-stimulated GPVI-deficient platelets compared to WT controls (Fig. 7, A to C, and movie S3).

DISCUSSION

GPVI is one of the key receptors involved in hemostasis and the prothrombotic state of acute coronary syndrome; thus, targeting GPVI may be therapeutic for thrombosis. Recombinant GPVI-Fc improves left ventricular function after experimental myocardial infarction in mice (28). Injection of GPVI-specific antibodies into mice leads to the depletion of the receptor and provides strong protection against arterial thrombosis (29, 30). GPVI is also implicated in vascular integrity during development and inflammation (31). Here, our study, using platelets from patients and mice, revealed that GPVI may also contribute to AD through direct interaction with Aβ40 and the consequent release of fibrinogen that amplifies platelet-mediated formation of amyloid fibrils. GPVI-blocking antibodies reduced platelet-associated amyloid aggregate formation. Aβ40 induced tyrosine phosphorylation, including the phosphorylation of LAT, in a GPVI-dependent manner. Platelet aggregation, Aβ40-induced ATP release, and LAT phosphorylation were reduced in GPVI-deficient murine and human platelets. The GPVI-induced release of fibrinogen accounted for amyloid aggregate formation in vitro. In vivo, enhanced platelet accumulation at injured vessels after stimulation of platelets with Aβ40 was markedly reduced when we injected GPVI-deficient platelets into the mice or pretreated the platelets with losartan, a small molecule that has been described to inhibit collagen-induced platelet aggregation in mice (23, 24). Losartan inhibited Aβ40-induced platelet aggregation and ATP and fibrinogen release but had no effect on platelet-mediated amyloid aggregate formation. These results are in line with studies, showing that the use of angiotensin receptor blockers, such as losartan, restore cerebrovascular dysfunction but have no effects on memory decline or AD pathology (as in, specifically, amyloidosis) (32, 33). The selective blocking of the angiotensin IV and its receptor (AngIV/AT4R)–mediated cascade is suggested to represent the underlying mechanism in losartan’s benefits. However, our data suggest that the beneficial effect on cerebrovascular function is not restricted to the AngIV/AT4R cascade but rather also includes reduction of GPVI-induced platelet activation and aggregation, demonstrating broader implications of losartan. These results are in line with a study by Elaskalani and colleagues (34) who showed reduced platelet aggregation and phospholipase Cγ2 phosphorylation in response to Aβ42 when they block GPVI by losartan.

Besides collagen, several GPVI ligands have been identified; these include diesel exhaust particles and large polysaccharides, such as fucoidan and dextran sulfate (35), as well as fibrin (36). Here, we provide evidence for Aβ40 binding to GPVI and acting as a regulator of GPVI signaling, including tyrosine phosphorylation, ATP and fibrinogen release, and platelet aggregation. Activation of GPVI was induced by direct binding of Aβ40 to the receptor and most likely not as secondary effect of, say, fibrinogen release, conversion of fibrinogen to fibrin, and fibrin-mediated GPVI activation.

To date, there is only one study that has investigated GPVI in AD. Those authors showed that, compared to healthy controls, patients with AD have decreased plasma levels of soluble GPVI (sGPVI) (37). This finding is of notable interest in terms of an antithrombotic strategy, given that sGPVI could bind collagen exposed upon vessel injury and thus reduces its binding to platelet GPVI. Reduced sGPVI plasma levels imply increased GPVI exposure at the surface of AD platelets, suggesting an increased number of Aβ40-sensitive receptors at the platelet surface and thus potentially enhanced Aβ40 binding to platelets in patients with AD.

Our data suggest that the binding of Aβ40 to GPVI induces the release of fibrinogen that is then incorporated into amyloid aggregates (Fig. 6). The formation of fibrin might not play a role since treatment of platelets with factor X inhibitor arixtra did not alter platelet-induced amyloid aggregate formation in culture. Studies have shown that fibrinogen only binds to human but not to mouse GPVI (38, 39). Because we did not observe differences in the integration of fibrinogen into Aβ fibrils using either human or mouse platelets, we do not believe that fibrinogen binding to GPVI plays a role in platelet-induced amyloid aggregate formation. Fibrinogen has been identified as possible contributor to the pathology of AD, and reducing fibrinogen decreases neurovascular damage, BBB permeability, and neuroinflammation in AD (40). Fibrinogen is a cerebrovascular risk factor that is able to bind to Aβ, thereby altering fibrin clot structure and degradation (41, 42). The interaction of Aβ and fibrinogen induces fibrinogen oligomerization (42). Targeting the interaction of Aβ and fibrinogen is a promising new therapeutic approach in AD (43). However, the authors had not taken into consideration that platelets might play a role by binding to fibrinogen and/or Aβ. Here, we provide evidence for platelets playing an important role in Aβ40-induced release of fibrinogen via GPVI and integrin αIIbβ3 and for fibrinogen being involved in platelet-induced amyloid aggregate formation.

We propose that engagement of GPVI and integrin αIIbβ3 by Aβ40 at the platelet surface induces the formation of an Aβ fibril network that included binding of Aβ40 to GPVI and integrin αIIbβ3 and fibrinogen binding to Aβ40 and integrin αIIbβ3 (Fig. 7D). These different binding possibilities might induce the formation of a specific type of “clustering” of GPVI and integrin αIIbβ3. Therefore, it is feasible that the failure of losartan to prevent platelet-mediated Aβ aggregate formation is due to its inability to block GPVI clustering, as already shown in the presence of collagen (24). However, because we previously did not observe integrin activation in the presence of Aβ40 alone (19), binding of Aβ40 to GPVI and integrin αIIbβ3 probably did not induce integrin inside-out signaling. According to our data, both previously published (19) and extended here, Aβ40 binds to nonactivated integrin αIIbβ3 on the surface of platelets, and this binding is enhanced in the presence of adenosine 5′-diphosphate (ADP) and CRP, probably due to activation-induced up-regulation of αIIbβ3 at the platelet surface.

Together, our findings reveal that GPVI mediates platelet-induced amyloid aggregate formation through the release of ATP and fibrinogen in response to direct binding of Aβ40 at the platelet surface. Further analysis is needed to validate whether blocking GPVI is beneficial to reduce amyloid plaque formation in cerebral vessels (as in CAA) and in brain parenchyma.

MATERIALS AND METHODS

Chemicals and antibodies

Platelets were activated with CRP (CambCol Laboratories Limited) or soluble Aβ40 (1-40; Bachem Peptide, catalog no. 4014442.1000) sequence single-letter code (DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV). Aβ1-40 stock solutions with a concentration of 1 mg/ml were solved in sterile H2O and stored at −20°C. Apyrase (grade II, from potato) and prostacyclin from Calbiochem were used for isolation. Antibodies against phosphotyrosine (Millipore, clone 4G10; catalog no. 05-321), phospho-LAT (p-LAT) (Tyr200; Abcam, catalog no. ab68139), Aβ1-16 (BioLegend, 6E10, catalog no. SIG-39320), and fibrinogen (Dako, catalog no. A0080) were used for immunoblotting. The antibodies to LAT (catalog no. 9166), β-actin (catalog no. 4967), α-tubulin (catalog no. 2144), and horseradish peroxidase (HRP)–linked secondary antibodies (catalog nos. 7074 and 7076) were from Cell Signaling Technology.

Animals

Mice with targeted deletion of GPVI were provided by J. Ware (University of Arkansas for Medical Sciences) and backcrossed to C57BL/6 mice. For the generation of homozygous WT and Gp6−/−mice, heterozygous breeding partners were mated. The animals were maintained in an environmentally controlled room at 22° ± 1°C with a 12-hour day-night cycle. Two to five mice were housed in Makrolon cages type III with ad libitum access to food (standard chow diet) and water. All animal experiments were conducted according the Declaration of Helsinki and approved by the Ethics Committee of the State Ministry of Agriculture, Nutrition and Forestry State of North Rhine-Westphalia, Germany (reference numbers AZ 84-02.05.40.16.073 and AZ 81-02.4.2019.A232).

Human platelet preparation

Platelets were prepared, as previously described (19). Fresh ACD-anticoagulated blood was obtained from healthy volunteers (age of 18 to 50 years, from the blood bank, not AD or GPVI deficient) and GPVI-deficient patients, as indicated. Participants provided their written informed consent to participate in this study according to the Ethics Committee and the Declaration of Helsinki (study number 2018-140-KFogU). Collected blood was centrifuged at 200g for 10 min at room temperature. The supernatant (platelet-rich plasma; PRP) was added to phosphate-buffered saline [PBS; pH 6.5, apyrase (2.5 U/ml) and 1 μM PGI2 in 1:1 volumetric ratio and centrifuged at 1000g for 6 min]. Platelets were resuspended in Tyrode’s buffer solution (140 mM NaCl, 2.8 mM KCl, 12 mM NaHCO3, 0.5 mM Na2HPO4, and 5.5 mM glucose, pH 7.4).

Murine platelet preparation

Murine blood was acquired by retro bulbar puncture and centrifuged at 250g for 5 min. The samples were centrifuged at 50g for 6 min to obtain PRP. PRP was washed two times (650g for 5 min at room temperature) before the pellet was resuspended in Tyrode’s buffer [136 mM NaCl, 0.4 mM Na2HPO4, 2.7 mM KCl, 12 mM NaHCO3, 0.1% glucose, and 0.35% bovine serum albumin (BSA; pH 7.4)] supplemented with prostacyclin (0.5 μM) and apyrase (0.02 U/ml). Before use, platelets were resuspended in the same Tyrode’s buffer supplemented with 1 mM CaCl2.

Platelet culture, supernatant blotting, and Congo red and immunofluorescent staining

Isolated human or murine platelets were preincubated for 15 min with 6 μg per 2 × 106 platelets anti-mouse integrin αIIbβ3 antibody [Leo.H4/rat immunoglobulin G2b (IgG2b); emfret ANALYTICS, catalog no. M021-0] or 6 μg per 2 × 106 platelets anti-mouse GPVI antibody (JAQ1 Rat IgG2a; emfret ANALYTICS, catalog no. M011-0) or 100 μM losartan (Tocris, catalog no. 3798). The final concentration of 2 × 106 platelets per well were added to 150 μl of Dulbecco’s modified Eagle’s medium. Platelets were stimulated with 5 μM Aβ40 or CRP (5 μg/ml) for 3 days at 37°C and 5% CO2. After 3 days of incubation, unbound platelets were removed by rinsing with PBS. Adherent platelets were fixed with 2% paraformaldehyde and stained against fibrillary Aβ aggregates with Congo red according to the manufacturer’s protocol (Millipore catalog no. 101641). For immunofluorescence staining, slides were washed with 100 μl of PBS before being fixed with 2% paraformaldehyde and blocked for 1 hour with 5% goat serum in PBS. Afterward, slides were incubated overnight at 4°C with the primary antibodies against Aβ (mouse anti-human; 6E10), fibrin-[ogen] (rabbit anti-mouse; DAKO) and the IgG controls in a 1:100 dilution containing 1% BSA and 5% goat serum in PBS. The next day, the chamber slide was washed three times with PBS and afterward incubated for 1 hour at room temperature with the secondary antibodies labeled with Alexa Fluor 488 and Alexa Fluor 555 (Life Technologies catalog nos. A32727 and A32790) in a 1:250 dilution containing 1% BSA and 5% goat serum in PBS. For immunoblotting analysis of supernatants, the cell culture supernatants were removed and centrifuged at 10,000g for 10 min at 4°C. The supernatant was collected, prepared with reducing sample buffer (Laemmli buffer), and denatured at 95°C for 5 min.

Cell lysis and immunoblotting

Platelets (60 × 106) were stimulated with 20 μM soluble Aβ40 or CRP (5 μg/ml) in Tyrode’s buffer (pH 7.4) at 37°C for the indicated time. Pretreatment, when indicated, with anti-mouse integrin αIIbβ3 antibody (Leo.H4; emfret ANALYTICS), anti-mouse GPVI antibody (JAQ1; emfret ANALYTICS), abciximab (Janssen-Cilag GmbH), or losartan (Cayman Chemical company) occurred at 37°C for 15 to 30 min. For separation into supernatant and pellet, platelets were centrifuged at 650g. For platelet lysis, human platelets were incubated for 15 min on ice with lysis buffer containing 145 mM NaCl, 20 mM tris-HCl, 5 mM EDTA, 0.5% sodium deoxycholate, 1% Triton X-100, and complete protease inhibitor (PI) cocktail (Roche, catalog no. 5892970001). Murine platelets were incubated for 15 min on ice with lysis buffer containing 15 mM tris-HCl, 155 mM NaCl, 1 mM EDTA (pH 8.05), 0.005% NaN3, 1% IGPAL, and PI. Platelet lysates (30 μl) and supernatants (30 μl) were subjected to SDS–polyacrylamide gel under reducing conditions and transferred onto nitrocellulose blotting membrane (GE Healthcare Life Sciences). Membrane was blocked using 5% BSA or 5% nonfat dry milk in TBST (tris-buffered saline with 0.1% Tween 20) and probed with the appropriate primary antibody (dilution 1:1000 in 5% BSA in TBST) and secondary (dilution 1:2500 in 5% nonfat dry milk in TBST) HRP-conjugated antibody. Band intensities were quantified in relation to untreated platelets using the FUSION FX7 software (Vilber).

p-LAT: Under nonaggregating conditions (apyrase, 0.5 U/ml; lotrafiban, 10 μM; and indomethacin, 10 μM), human and mouse platelets (1.5 × 106) were stimulated with 30 μM soluble Aβ40, CRP (5 μg/ml), or collagen (1 μg/ml) in Tyrode’s buffer (pH 7.4) for the indicated time at 37°C. Cells were immediately lysed on ice with NP-40 lysis buffer (300 mM NaCl, 20 mM tris, 2 mM EGTA, 2 mM EDTA, and 2% NP-40 detergent) in addition to the protease and phosphatase inhibitors [5 mM sodium orthovanadate, 1 mM AEBSF [4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride], leupeptin (10 μg/ml), aprotinin (10 μg/ml), and pepstatin (1 μg/ml)]. Platelet lysates were loaded in a gradient gel (NuPAGE 4 to 12%; Invitrogen) under reducing conditions and transferred onto polyvinylidene difluoride blotting membrane (TransBlot Turbo, Bio-Rad). Membrane was blocked using 5% BSA in TBST and probed with the appropriate primary antibody p-LAT (dilution 1:500; Abcam) or α-tubulin (dilution 1:1000; Sigma-Aldrich) and secondary HRP-conjugated antibody anti-mouse IgG (dilution 1:5000; GE Healthcare) or anti-rabbit IgG (dilution 1:5000; GE Healthcare). Band signals were detected using Odyssey Fc imaging system (LI-COR).

Immunoprecipitation

Platelets (1 × 109) were stimulated with 20 μM Aβ40 for 30 min at 37°C while being shaken. Platelets without stimulation with Aβ40 were used as a control (resting). Murine resting and 20 μM Aβ40-stimulated platelets were lysed with 5× lysis buffer (as described in the “Cell lysis and immunoblotting” section) for 10 min on ice. Afterward, the lysate was centrifuged at 10,000g for 10 min at 4°C to clear the lysate from remaining cell fragments. The cleared lysate was transferred to a new reaction tube and incubated with GPVI antibody or corresponding IgG control (JAQ1, emfret ANALYTICS; mouse IgG2b, Cell Signaling) for 1 hour at 4°C. Samples were transferred to a new reaction tube and incubated with washed G-sepharose protein overnight at 4°C. Samples were washed three times: first time with immunoprecipitation buffer (15 mM tris-HCl, 155 mM NaCl, 1 mM EDTA, and 0.005% NaN3) additionally containing 1% IGPAL, and second and third times with only immunoprecipitation buffer before adding 2× Laemmli, containing 5% mercaptoethanol, and incubated at 95°C for 5 min. After centrifugation at 10,000g for 2 min, supernatants were removed and analyzed via immunoblotting against Aβ (BioLegend, 6E10) and GPVI (R&D Systems, catalog no. AF6758).

Pulldown

Recombinant GPVI (R&D Systems, catalog no. 6758-GP-050) was covalently immobilized to Pierce NHS (N-hydroxysuccinimide)–Activated Magnetic Beads according to the manufacturer’s information (Thermo Fisher Scientific, catalog no. 88802). Protein solution with and without 20 μM Aβ40 was added to the GPVI-coupled beads and incubated at room temperature on a rotator for 1 to 2 hours. Beads were collected with the magnetic stand and washed for three times with wash buffer (TBS with 0.05% Tween 20 detergent) and afterward washed with ultrapure water. For protein elution, beads were washed with 100 μl of elution buffer (0.1 M glycine, pH 2.0), and pH was neutralized by adding 10 μl of neutralization buffer (1 M tris, pH 9). Laemmli buffer was added, and samples were analyzed via immunoblotting under reducing conditions against Aβ (BioLegend, 6E10) and GPVI (R&D Systems, catalog no. AF6758).

AVEXIS screening

Aβ40 peptides or CD200R bait proteins were incubated in MaxiSorp 96-well microtiter plates (Nunc) for 1 hour and then blocked with 1% BSA for 30 min. The peptide-coated plate was incubated with full-length recombinant soluble pentameric (s5) GPVI and s5CD200 for 1 hour. Three wash steps were performed between each incubation using PBS with 0.1% Tween 20. After addition of nitrocefin (125 μg/ml; #N005, Toku-e) and incubation for 1 hour, absorbance was measured at 485 nm on a VersaMax microplate reader (Molecular Devices).

Platelet aggregation and ATP release

Aggregation was measured as percentage light transmission compared to Tyrode’s buffer (as = 100%) using a Chrono-Log dual-channel lumi-aggregometer (model 700) at 37°C stirring at 1000 rpm. Human ATP release was assessed by applying a luciferin/luciferase bioluminescent assay and calculated using a provided ATP standard protocol (all from Chrono-Log). Murine ATP release was measured using ATP Bioluminescence Assay Kit HS II (Roche, catalog no. 11699709001) according to the manufacturer’s information and normalized to resting.

Flow cytometry

Flow cytometry was performed, as described (18, 44). Analysis of Aβ40 binding to platelets surface was carried out using fluorophore-labeled antibodies for Aβ (anti–Aβ-FITC; Santa Cruz Biotechnology, catalog no. sc-28365). Twenty-five microliters of washed blood samples was diluted in Tyrode’s buffer with 1 mM CaCl2 and stimulated with indicated agonist [5 μM Aβ40 and CRP (5 μg/ml)] and antibody at room temperature for 15 min. Reaction was stopped using 300 μl of PBS. Samples were analyzed on a FACSCalibur flow cytometer (BD Biosciences).

Carotid artery ligation model

Carotid ligation in mice was performed, as described elsewhere (18). Platelets from WT and Gp6−/− donor mice were stained with a CellTracker Red CMTPX (Invitrogen) according to the manufacturer’s guidelines and incubated with losartan (100 μM) or vehicle and Aβ (50 μg/ml) for 30 min. WT littermates mice were anaesthetized using ketamine (Zoetis) and xylazine (WDT) and put on a heating pad. The right common carotid artery was prepared, and after intravenous injection of fluorescently labeled and treated platelets, a film of 30 s was taken using a DM6FS microscope (Leica Microsystems, Wetzlar, Germany). Afterward, the carotid artery was ligated vigorously for 5 min, thus inducing vascular injury. The interaction of the fluorescent platelets with the injured vessel wall was visualized 20 min after ligation by in vivo video microscopy. Tethering and adherent cells were counted as means from 10 different pictures throughout the film with the same time span between these pictures, but always in the same phase of vessel pulsation.

Statistical analysis

Data are provided as arithmetic means ± SEM. Significant differences were calculated using the two-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison post hoc test, one-way ANOVA with Dunnett’s post hoc test, or Student’s t test as indicated in the figure legends. Outliers were excluded using Grubb’s test.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/13/643/eaba9872/DC1

Fig. S1. Time-dependent LAT phosphorylation of human platelets stimulated with Aβ40.

Fig. S2. Different concentrations of losartan on platelet cell cultures.

Fig. S3. Reduced amyloid aggregate formation through GPVI inhibition in a concentration-dependent manner.

Fig. S4. Immunofluorescence staining of fibrinogen and Aβ in human and murine platelet cell cultures.

Fig. S5. No alteration in amyloid fibril formation upon inhibition of active factor Xa.

Movie S1. Adhesion of WT platelets at the injured carotid artery in vivo.

Movie S2. Adhesion of WT platelets at the injured carotid artery in vivo after losartan treatment.

Movie S3. Adhesion of Gp6−/− platelets at the injured carotid artery in vivo.

REFERENCES AND NOTES

Acknowledgments: We thank M. Spelleken for providing outstanding technical assistance. Funding: The study was supported by the Deutsche Forschungsgemeinschaft (DFG) grant number EL651/5-1 and the Collaborative Research Center (CRC) 1116 (Project A05) to M.E. and to M.K. (project B06) and N.G. (project B09). S.P.W. is a British Heart Foundation Chair (CH03/003). S.P. and Y.S. acknowledge funding from the COMPARE and the British Heart Foundation (PG/16/53/32242). We acknowledge the support of the Susanne-Bunnenberg-Stiftung at the Düsseldorf Heart Center. Author contributions: L.D., L.M.T., I.K., S.G., R.B., A.B., D.M., S.P., M.K., N.G., and Y.S. performed experiments and analyzed the data. D.M. arranged for the acquisition of GPVI-deficient patient samples. L.M.T. discussed the data and helped draft the manuscript. S.P.W. and Y.S. designed and performed experiments and read and edited the manuscript. L.D. and M.E. designed experiments, discussed the data, and wrote the manuscript. A.B. is currently affiliated with the Infection, Inflammation and Rheumatology at UCL Great Ormond Street, Institute of Child Health, London, WC1N 1EH, UK. Competing interests: The authors declare they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Stay Connected to Science Signaling

Navigate This Article