The Glycoprotein VI-Phospholipase Cγ2 Signaling Pathway Controls Thrombus Formation Induced by Collagen and Tissue Factor In Vitro and In Vivo
Objective— Both collagen and tissue factor can be initiating factors in thrombus formation. We investigated the signaling pathway of collagen-induced platelet activation in interaction with tissue factor–triggered coagulation during the thrombus-forming process.
Methods and Results— In murine blood flowing over collagen, platelet exposure of phosphatidylserine and procoagulant activity, but not adhesion, completely relied on each of the following signaling modules: glycoprotein VI (GPVI), FcR γ-chain, Src kinases, adaptor protein LAT, and phospholipase Cγ2 (PLCγ2). On flow in the presence of tissue factor, these signaling components were essential for platelet aggregation and greatly enhanced fibrin clot formation. Collagen-stimulated thrombin generation relied on the presence and activity of GPVI, FcR γ-chain, Src kinase, LAT, and PLCγ2. The physiological importance of this GPVI pathway was shown in a FeCl3-induced in vivo murine thrombosis model. In both venules and arterioles, signaling through GPVI, FcR γ-chain, and Src kinases enhanced the formation of phosphatidylserine-exposing and fibrin-rich thrombi.
Conclusions— The GPVI-PLCγ2 activation pathway regulates collagen-dependent coagulation in venous and arterial thrombus formation.
Thrombus formation can be initiated by both platelet- and coagulation-activating factors. Collagens in the extracellular matrix and other vascular layers are thought to act as principal platelet-activating components of the damaged vessel wall; they also provide a surface for von Willebrand factor adhesion.1 Tissue factor, also exposed in damaged vessels, is a key trigger of the coagulation process.2 Because of the proposed major role of platelets in arterial thrombosis and the importance of coagulation in venous thrombosis, the current understanding is that collagen/von Willebrand factor–mediated events are more important in arteries, whereas tissue factor plays a more prominent role in venous thrombus formation.
Flow studies with human and mouse blood have established that the signaling receptor glycoprotein VI (GPVI) exclusively mediates collagen-induced platelet procoagulant activity, thus linking the processes of platelet activation and coagulation.3–5 This procoagulant platelet response is mediated by a prolonged and potent rise in cytosolic [Ca2+]i, which results in exposure of procoagulant phosphatidylserine (PS) at the platelet outer surface.6 PS exposure is a key regulating factor in the coagulation process. For instance, PS-containing membrane surfaces dramatically increase the formation of factor Xa and thrombin.7 However, several authors have argued that this platelet response has only an assistant role in coagulation and that, in vivo, other platelet reactions may play important roles as well.2,8 Thus, although there is no doubt that platelets enhance thrombin generation (coagulation) in plasma or whole blood, the precise mechanism is still a matter of debate.
The platelet immunoreceptor GPVI is coexpressed with the Fc receptor (FcR) γ-chain, although the latter also interacts with other platelet receptors, eg, GPIb.9 In human and mouse platelets, activation of GPVI by collagen or other ligands results in a complex cascade of signaling events.10–12 The initial step is tyrosine phosphorylation of the FcR γ-chain by the Src family kinases Fyn and Lyn. Subsequent activation of the tyrosine kinase Syk leads to phosphorylation of multiple signaling proteins, including the adaptor proteins LAT, SLP-76, and SLAP-130, and, additionally, the G protein regulator Vav and phospholipase C-γ2 (PLCγ2).13 GPVI stimulation activates various other protein kinases, including Btk, Tec, phosphoinositide 3-kinase, and, further downstream, protein kinase C, extracellular signal regulated kinase 1/2, and focal adhesion kinase, as a result of which platelets respond by integrin activation, Ca2+ increase, aggregation, shape change, secretion, and procoagulant activity. Although earlier work has shown that GPVI plays a key role in the collagen-induced platelet procoagulant activity in both stasis and flow,14 it is still unclear which signaling elements downstream of GPVI contribute to this platelet reaction.
Evidence that GPVI plays a key role in arterial thrombosis comes from recent in vivo studies with mice, where the platelet aggregation in vivo was followed after ligation of arteries, causing exposure of vascular collagen.1,15 Absence of GPVI (as in FcR γ-chain null mice) or downregulation of GPVI (as in mice treated with JAQ1 antibody) appeared to suppress intravascular formation of platelet aggregates. Thus, knowing the in vitro evidence for a procoagulant effect of GPVI, we hypothesized that GPVI can also drive the coagulation process during thrombus formation.
In the present study, we used mice deficient in GPVI or in 1 of the signaling proteins downstream of GPVI to investigate the signal transduction route leading to GPVI-induced PS exposure, coagulation stimulation, and fibrin formation both in vitro and in vivo. We performed whole-blood flow studies partly in the presence of tissue factor–triggered coagulation to determine the functional effects of PS exposure. Furthermore, in vivo experiments where thrombus formation was induced with free radical–forming FeCl3 allowed us to study the importance of this process in a thrombosis model known to rely on thrombin generation and coagulation.16 The data showed that also in vivo the signaling cascade from GPVI to PLCγ2 led to coagulant activity and enhanced thrombus formation.
For a detailed Methods section, please see http://atvb.ahajournals.org.
Components of the GPVI–PLCγ2 Signaling Pathway Required for Collagen-Induced Procoagulant Activity of Platelets in Flowing Blood
In mouse and human platelets, functional activity of GPVI is required for collagen-induced aggregation and surface exposure of procoagulant PS.4,5 To investigate which signaling proteins downstream of GPVI are involved in the procoagulant platelet response, we used genetically modified mice and specific blockers of key proteins in the PLCγ2 activation pathway. We used Fab fragments of JAQ1 monoclonal antibody (mAb) to block GPVI receptors and, in comparison, GPVI-null mice lacking the FcR γ-chain. Src kinase activity was abolished with the compound PD173952, which has been identified recently as a selective and specific inhibitor of the Src-family kinases that is still active in blood. It essentially abolishes all collagen-induced but not thrombin-induced tyrosine phosphorylation events in platelets.17 Furthermore, mice deficient in LAT18 or PLCγ219 were used to study the contribution of these downstream signaling entities.
In the first experiments, PPACK-anticoagulated whole blood from wild-type or genetically modified mice was perfused over a collagen surface at a shear rate of 1000 s−1, which is representative of that found in murine arterioles. Using wild-type blood, platelets rapidly adhered to the collagen surface and partly assembled into aggregates. As shown before,4,20 platelet deposition and aggregate growth on collagen increased approximately linearly with time. Blocking of GPVI with JAQ1 Fab fragments abolished platelet adhesion and aggregation (Figure 1A), also confirming earlier results.4,15 The absence of FcRγ or treatment with Src kinase inhibitor resulted in a moderately reduced deposition of platelets on collagen; these platelets remained single and showed little tendency to aggregate. Similar results were obtained with blood from LAT-null or PLCγ2-null mice (Figure 1A).
Poststaining of wild-type/control thrombi with OG488-annexin A5 revealed many single PS-exposing platelets (Figure 1B). Strikingly, virtually no PS-exposing platelets were detected when GPVI was blocked (with JAQ1 Fab), FcR γ-chain was absent, Src kinases were blocked (with PD173952), or LAT or PLCγ2 was absent (Figure 1B). In sharp contrast, platelet deposition was only slightly affected by blocking of GPIb–von Willebrand factor interactions (with Xia.B2 mAb), secondary platelet responses because of release of ADP (with receptor antagonists MRS2179 and AR-C69931MX), thromboxane A2 (aspirin), or by blocking integrin αIIbβ3 (with lotrafiban; Figure 1A). Also, these blocking conditions did not significantly influence PS exposure compared with the control situation (Figure 1B).
A typical observation under control conditions was that wild-type platelets on the collagen surface only showed PS exposure when not assembled into aggregates. Conversely, aggregated platelets did not expose PS (Figure I, available online at http://atvb.ahajournals.org). We used the high resolution of two-photon laser scanning microscopy, allowing simultaneous detection of 2 fluorescent probes, to better characterize the populations of PS-exposing and aggregated platelets. With OG488-fibrinogen and AF647-annexin A5 added to wild-type blood, the fibrinogen label exclusively incorporated into platelet aggregates, whereas labeled annexin A5 was differently localized, only staining single platelets around the aggregates (Figure I). After perfusion, platelets from FcR γ-chain null mice and also wild-type platelets treated with PD173952 failed to bind fibrinogen (Figure I), pointing to the requirement of the FcR γ-chain and Src kinases to the aggregate-forming process.17 Together, these results indicate that the signaling components of the GPVI–PLCγ2 pathway (involving Src kinases, FcRγ, Syk, and LAT) are indispensable for PS exposure on collagen under flow. The PS-exposing platelets typically represent a different population than the platelets assembled into aggregates and binding fibrinogen.
Components of the GPVI Signaling Pathway Determining Platelet Procoagulant Activity and Thrombus Formation in Flowing Blood Triggered With Tissue Factor
To induce coagulation under flow conditions, citrated blood was mixed with tissue factor and CaCl2 just before entering the flow chamber. The perfusion protocol was such that coagulation occurred at physiological, millimolar Mg2+, and Ca2+ concentrations.21 With wild-type blood, perfusion in the presence of tissue factor resulted in PS exposure and the formation of platelet aggregates, which gradually transformed to clots trapping erythrocytes (Figure II, available online at http://atvb.ahajournals.org). TPLSM analysis indicated that the thrombi contained extensive networks of fibrin and aggregated platelets, which were surrounded by single PS-exposing platelets. Perfusion of JAQ1-treated blood (which blocked GPVI) or blood from FcRγ-null mice in the presence of tissue factor gave only limited platelet adhesion without aggregate formation. Fibrin formation was greatly suppressed, and only a few PS-exposing platelets were observed (Figure II). PD173952 was somewhat less inhibitory in the presence of tissue factor, with small fibrin(ogen)-binding aggregates still being formed (Figure II). This reflects the presence of Src kinase–independent pathways of platelet aggregation. Pretreatment of blood from FcR γ-chain null mice with PD173952 did not additionally reduce platelet adhesion.
Quantitative image analysis indicated that platelet deposition and PS exposure were significantly reduced by the various treatments (Figure 2). Pretreatment of wild-type blood with the anti-GPVI Fab JAQ1 reduced the total surface area coverage of platelets from 25.8% (control) to 6.3% and PS exposure from 7.9% (control) to 0.1% (Figure 2). This was also the case when using FcR γ-chain null blood or PD173952-treated blood, where platelet deposition was decreased to 3.8% and 4.9%, and annexin A5 binding was reduced even more to 0.5% and 0.7%, respectively (Figure 2). Reduced thrombus formation and coagulant activity under all of the above conditions was also seen at a lower shear rate of 150 s−1 (data not shown), which is representative of that in venules.
Together, these results stress the importance of the GPVI pathway in aggregate formation on collagen under flow. They also significantly extend earlier work by showing that, independent of the shear rate, GPVI signaling plays a key role in the coagulation process, that is, in thrombin and fibrin formation triggered by tissue factor. Apparently, collagen-induced PS exposure via the FcR γ-chain is a key regulatory factor in coagulation.
Components of the GPVI–PLCγ2 Signaling Pathway Involved in Collagen-Enhanced Thrombin Generation
To investigate more directly how coagulation is controlled by collagen-GPVI interaction and subsequent signaling, we examined the effect of collagen on thrombin generation in mouse platelet-rich plasma (PRP) triggered with tissue factor under static conditions. The addition of tissue factor/CaCl2 to wild-type PRP resulted in high-thrombin generation, a process that was greatly enhanced with a submaximal dose of collagen (peak level increased by 97%; Figure III, available online at http://atvb.ahajournals.org). Control experiments demonstrated that essentially no thrombin was formed during 60 minutes, when platelets or tissue factor were absent (data not shown). Collagen did not alter the thrombin generation curve with PRP from FcR γ-chain null mice (Figure 3; Figure III). In agreement with this, in PRP from wild-type mice, the blocking of GPVI (with JAQ1 Fab) or Src kinases (with PD173952) completely abolished the enhancing effect of collagen on thrombin generation. These interventions did not affect thrombin generation in PRP from FcR γ-chain null mice. In the absence of collagen, PRP from mice deficient in LAT or PLCγ2 showed normal thrombin generation curves compared with wild-types. Collagen addition did not increase the thrombin-generating activity in this PRP (Figure 3). Accordingly, the enhancing effect of collagen on thrombin generation was lost in case 1 of these components of the GPVI–PLCγ2 signaling pathway was missing.
Absence of GPVI Signaling Diminishes Thrombus Formation in Venules and Arterioles on Exposed Extracellular Matrix
To determine the physiological relevance of the collagen-induced activation pathway, we used an in vivo mouse model of microvascular thrombus formation that relies on exposure of the extracellular matrix and ensuing thrombin generation.22 Injury of the mesenteric vessels of anesthetized mice was induced by topical application of FeCl3, which caused local but complete denudation of the endothelium in venules and nearby arterioles. Intravital microscopy was used for real-time imaging of thrombus formation in the damaged vessels (Figure IV, available online at http://atvb.ahajournals.org).
To study the involvement of GPVI, wild-type mice were injected with JAQ1 antibody, which causes biphasic, long-term in vivo depletion of GPVI on platelets without affecting other platelet glycoproteins.15,23 After 10 minutes, this injection resulted in a greatly reduced platelet concentration (20% of normal platelet count), and a specific disappearance of GPVI on the platelet surface with other platelet glycoproteins remaining unchanged (data not shown).15 At 5 days after injection, however, the platelet count was normalized, whereas GPVI expression on platelets was still completely absent.
In venules from wild-type mice, thrombus formation started after seconds, whereas in arterioles there was a lag time of several minutes. Thrombus size, quantified as thrombus height perpendicular to the vessel wall, increased with time in the vessels from wild-type mice (Figure V, available online at http://atvb.ahajournals.org). After 10 minutes, 39% and 31% of the venules and arterioles were occluded, respectively. When FeCl3 was applied to vessels from mice that were injected with JAQ1 antibody (5 days before), thrombus formation in both venules and arterioles was greatly delayed and reduced (Figure 4). Similarly, in FcR γ-chain null mice, thrombus formation induced by FeCl3 was markedly delayed in both venules and arterioles. Thrombi in this case remained small in size in venules and were not formed at all in arterioles (P≤0.05). No occlusion was observed within 10 minutes. Furthermore, preinjection of mice with PD173952 at 10 minutes before FeCl3 application (estimated final plasma concentration of 50 μmol/L) resulted in a significant decrease in arterial thrombus size (P≤0.05) but not in venular thrombus size (Figure 4). PD173952 injection caused full blockage of collagen-induced aggregation in vitro. We conclude that, in this microvascular model of extracellular matrix exposure, the presence of GPVI, FcRγ, and Src kinase signaling are critical for venous and arterial thrombosis.
Reduced Fibrin Formation and PS Exposure in Arterial Thrombi After GPVI Depletion
TPLSM was used for better visualization of platelet activation and coagulation in arterioles after FeCl3 application. Wild-type mice were preinjected with OG488-labeled fibrinogen or annexin A5, and fluorescent thrombi were subsequently scanned. In FeCl3-treated arterioles from control mice, extensive deposition of OG488-fibrin(ogen) was detected and smaller spots of labeled annexin A5 (Figure 5). In contrast, in arterioles from mice pretreated with JAQ1 antibody (5 days) to downregulate GPVI, FeCl3 application led to formation of only small domains of fibrin labeling and no annexin A5 labeling. Thus, both fibrin formation and PS exposure were reduced after downregulation of platelet GPVI.
The present results indicate that the GPVI receptor, acting via a straightforward signaling pathway, acts as principal mediator of PS exposure, platelet-dependent thrombin generation, and fibrin formation/coagulation in the presence of tissue factor. The signaling modules involved in the procoagulant platelet response are the FcRγ, Src kinases (likely Fyn/Lyn), the adaptor protein LAT, and PLCγ2, the latter of which is responsible for Ca2+ mobilization. In addition, we find that under flow this signaling pathway mediates the build-up of a platelet-fibrin thrombus. Evidence comes from a number of approaches to eliminate GPVI or subsequent activation steps. Blocking anti-GPVI JAQ1 Fab fragments were used. Additionally, the Src kinase inhibitor PD173952 completely inhibits collagen-induced aggregation and tyrosine phosphorylation of mouse platelets in plasma but leaves platelet responses to G protein–coupled receptor agonists unchanged.17 Abolished PS exposure and thrombus formation were also seen in blood from mice lacking the FcR γ-chain, the adaptor protein LAT, or the effector protein PLCγ2. The recognition that LAT is required for the procoagulant response is especially important, because earlier work suggested that LAT is dispensable for collagen- and convulxin-induced platelet aggregation.24
Interestingly, the addition of JAQ1 antibody fragments resulted in lower platelet adhesion than with FcR γ-chain–deficient blood. This can be explained by the recent observation that GPVI and GPIb-IX-V associate on platelets25 and the possibility that the JAQ1 antibody can thereby interfere with GPIb-mediated effects, while GPIb remains functionally active under flow in the absence of FcRγ.26 We have shown before that GPIb plays a more prominent role in platelet-collagen interaction at reduced levels of GPVI and FcRγ, which argues against a significant role of the FcR γ-chain in thrombus formation independently of GPVI, such as others have proposed.27
Earlier perfusion studies have indicated that platelet-collagen interaction via GPVI is the principal trigger of platelet activation and aggregation in mouse and human blood.5,15 Here, we significantly extend this observation by showing that GPVI signaling also controls thrombus formation on collagen in the presence of coagulation. We found that not only platelet aggregate formation, but also PS exposure and fibrin formation were greatly suppressed in blood from FcR γ-chain null mice, both at high (arteriolar) and low (venular) shear rates. This was also the case when GPVI or Src kinases were blocked with JAQ1 Fab or PD173952, respectively. During flow in the absence of GPVI activity, tissue factor was still active in triggering thrombin formation, as deduced from the traces of fibrin that still were formed, but apparently propagation of the coagulation process and fibrin clot formation were prevented. Accordingly, platelet activation by GPVI, most likely via PS exposure, is required for full-coagulation activity.
In a macrovascular thrombosis model triggered by mechanical damage of the carotid artery, it has been demonstrated that GPVI is required for platelet adhesion to the vessel wall.1 Whether or not coagulation contributes to the thrombotic process in that model is unknown. Here, we used a microvascular model, relying on free radical formation with FeCl3, known to be driven by thrombin formation and coagulation.22 The application of FeCl3 resulted in an almost complete disappearance of the endothelium in both arterioles and venules, causing exposure of the collagen-containing extracellular matrix (M Kuijpers, unpublished data, 2004). In either vessel type, GPVI activity appeared to control the thrombus-forming process. For instance, in venules and arterioles from FcR γ-chain null mice, thrombi remained small in size (venules) or were not formed at all (arterioles). Long-term injection with an anti-GPVI JAQ1 mAb, which completely downregulates GPVI on the platelet surface,15 delayed and reduced the thrombus formation in venules and arterioles to a similar extent. Furthermore, observation with TPLSM indicated that in vessels from mice preinjected with JAQ1 mAb, only small spots of labeled fibrin were present, whereas labeled annexin A5 was not detectable at all. Finally, injection of PD173952 into mice to block (GPVI-activated) Src kinases suppressed the thrombotic process in arterioles.
Together, these data provide the first in vivo evidence that GPVI, acting via a relatively simple signaling pathway (Src kinases, FcRγ, Syk, LAT, or PLCγ2), has a key role in platelet procoagulant activity and subsequent thrombin and fibrin formation. The procoagulant function of GPVI appears to contribute to arterial thrombus formation and, interestingly, also to venous thrombus formation. These findings, thus, point to a dual role for GPVI, both procoagulant and aggregatory, in collagen-induced thrombus formation in vivo.
We thank Pfizer Global Research and Development for supplying PD173952.
This work was supported by the Netherlands Heart Foundation (2002B014) and the Netherlands Organization for Scientific Research (902-16-276). A.S. and J.M.A. were supported by a Marie Curie Fellowship from the European Community (QLK5-CT-2000-60007). J.M.A. holds a BHF Studentship.
- Received July 8, 2005.
- Accepted October 12, 2005.
Massberg S, Gawaz M, Grüner S, Schulte V, Konrad I, Zohlnhofer D, Heinzmann U, Nieswandt B. A crucial role of glycoprotein VI for platelet recruitment to the injured arterial wall in vivo. J Exp Med. 2003; 197: 41–49.
Butenas S, Cawthern KM, van ’t Veer C, Dilorenzo ME, Lock JB, Mann KG. Antiplatelet agents in tissue factor-induced blood coagulation. Blood. 2001; 97: 2314–2322.
Nieswandt B, Brakebusch C, Bergmeier W, Schulte V, Bouvard D, Mohtari-Nejad R, Lindhout T, Heemskerk JWM, Zirngibl H, Fässler R. Glycoprotein VI but not α2β1 integrin is essential for platelet interaction with collagen. EMBO J. 2001; 20: 2120–2130.
Kuijpers MJE, Schulte V, Bergmeier W, Lindhout T, Brakebusch C, Offermanns S, Fässler R, Heemskerk JWM, Nieswandt B. Complementary roles of glycoprotein VI and α2β1 integrin in collagen-induced thrombus formation in flowing whole blood ex vivo. FASEB J. 2003; 17: 685–687.
Siljander PRM, Munnix ICA, Smethurst PA, Deckmyn H, Lindhout T, Ouwehand WH, Farndale RW, Heemskerk JWM. Platelet receptor interplay regulates collagen-induced thrombus formation in flowing human blood. Blood. 2004; 103: 1333–1341.
Zwaal RFA, Schroit AJ. Pathophysiological implications of membrane phospholipid asymmetry in blood cells. Blood. 1997; 89: 1121–1132.
Monroe DM, Hoffman M, Roberts HR. Platelets and thrombin generation. Arterioscler Thromb Vasc Biol. 2002; 22: 1381–1389.
Nieswandt B, Bergmeier W, Schulte V, Rackebrandt K, Gessner JE, Zirngibl H. Expression and function of the mouse collagen receptor glycoprotein VI is strictly dependent on its association with the FcRγ chain. J Biol Chem. 2000; 275: 23998–234002.
Ezumi Y, Shindoh K, Tsuji M, Takayama H. 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. 1998; 188: 267–276.
Nieswandt B, Watson SP. Platelet-collagen interaction: is GPVI the central receptor? Blood. 2003; 102: 449–461.
Wonerow P, Pearce AC, Vaux DJ, Watson SP. A critical role for phospholipase Cγ2 in αIIbβ3-mediated platelet spreading. J Biol Chem. 2003; 278: 37520–37529.
Nieswandt B, Schulte V, Bergmeier W, Mokhtari-Nejad R, Rackebrandt K, Cazenave JP, Ohlmann P, Gachet C, Zirngibl H. Long-term antithrombotic protection by in vivo depletion of platelet glycoprotein VI in mice. J Exp Med. 2001; 193: 459–469.
Ni H, Papalia JM, Degen JL, Wagner DD. Control of thrombus embolization and fibronectin internalization by integrin αIIbβ3 engagement of the fibrinogen-γ chain. Blood. 2003; 102: 3609–3614.
Auger JM, Kuijpers MJE, Senis YA, Watson SP, Heemskerk JWM. Adhesion of human and mouse platelets to collagen under shear. FASEB J. 2005; 19: 825–827.
Zhang W, Irvin BJ, Trible RP, Abraham RT, Samelson LE. Functional analysis of LAT in TCR-mediated signaling pathways using a LAT-deficient Jurkat cell line. Int Immunol. 1999; 11: 943–950.
Lecut C, Schoolmeester A, Kuijpers MJE, Broers JLV, van Zandvoort MAMJ, Vanhoorelbeke K, Deckmyn H, Jandrot-Perrus M, Heemskerk JWM. Principal role of glycoprotein VI in α2β1 and αIIbβ3 activation during collagen-induced thrombus formation. Arterioscler Thromb Vasc Biol. 2004; 24: 1727–1733.
Schulte V, Rabie T, Prostredna M, Aktas B, Grüner S, Nieswandt B. Targeting of the collagen-binding site on glycoprotein VI is not essential for in vivo depletion of the receptor. Blood. 2003; 101: 3948–3952.
Judd BA, Myung PS, Obergfell A, Myers EE, Cheng AM, Watson SP, Pear WS, Allman D, Shattil SJ, Koretzky GA. Differential requirement for LAT and SLP-76 in GPVI versus T cell receptor signaling. J Exp Med. 2002; 195: 705–717.
Liu J, Pestina TI, Berndt MC, Jackson CW, Gartner TK Botrocetin/vWf-induced signaling through GPIb-IX-V produces TxA2 in an αIIbβ3 and aggregation-independent manner. Blood. 2005; 106: 2750–2756.