Distinct but critical roles for integrin aIIbb3 in platelet lamellipodia formation on fibrinogen, collagen-related peptide and thrombin Kelly Thornber1, Owen J. T. McCarty2,3, Steve P. Watson2 and Catherine J. Pears1
1 Department of Biochemistry, University of Oxford, UK 2 Centre for Cardiovascular Sciences, Institute of Biomedical Research, University of Birmingham, UK 3 Department of Biomedical Engineering, Oregon Health & Science University, Portland, OR, USA
Keywords Integrins are the major receptor type known to facilitate cell adhesion and aIIbb3; adhesion; integrins; lamellipodia; lamellipodia formation on extracellular matrix proteins. However, collagen- platelets related peptide and thrombin have recently been shown to mediate platelet lamellipodia formation when presented as immobilized surfaces. The aims Correspondence C. Pears, Department of Biochemistry, of this study were to establish if there exists a role for the platelet integrin South Parks Road, University of Oxford, aIIbb3 in this response; and if so, whether signalling from the integrin is Oxford, OX1 3QU, UK required for lamellipodia formation on these surfaces. Real-time analysis Fax: +44 1865 275259 was used to compare platelet morphological changes on surfaces of fibrino- Tel: +44 1865 275737 gen, collagen-related peptide or thrombin in the presence of various E-mail: [email protected] pharmacological inhibitors and platelets from ‘knockout’ mice. We demon- Website: http://www.bioch.ox.ac.uk strate that collagen-related peptide and thrombin stimulate distinct patterns 2+ (Received 11 July 2006, revised 22 August of platelet lamellipodia formation and elevation of intracellular Ca to 2006, accepted 12 September 2006) that induced by the integrin aIIbb3 ligand, fibrinogen. Nevertheless, lamelli- podia formation on collagen-related peptide and thrombin is dependent doi:10.1111/j.1742-4658.2006.05500.x upon engagement of aIIbb3, consistent with release of aIIbb3 ligand(s) from platelet granules. However, the requirement for signalling by the integrin on fibrinogen can be bypassed by the addition of thrombin to the solution.
These observations reveal a critical role for aIIbb3 in forming lamellipodia on collagen-related peptide and thrombin which is dependent on its ability to function as an adhesive receptor but not necessarily on its ability to sig- nal. These results suggest that integrins may play an important role in lamellipodia formation triggered by nonintegrin ligands in platelets and possibly in other cell types.
Platelets play an essential role in the formation of a series of morphological changes that include round- haemostatic plug at the site of vascular injury. This ing, generation of filopodia and lamellipodia, and for- process requires adhesion of the platelet to the mation of actin stress fibres [1]. These events serve to exposed subendothelial matrix, followed by powerful stabilize the thrombus, thereby enabling it to with- intracellular signalling events that lead to platelet– stand the high shear forces found in arteries and arte- platelet interactions and thrombus formation. One rioles. critical feature in this scheme is the dramatic alter- Stable adhesion of platelets to the subendothelial ation in platelet morphology in response to activa- matrix is dependent upon sustained activation of tion. Thus, the resting, discoid platelet undergoes a integrins [2]. Integrins are glycoprotein heterodimers
Abbreviations CRP, collagen-related peptide.
5032 FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS K. Thornber et al. Platelet lamellipodia formation via aIIbb3
composed of a-subunits and b-subunits that exist in an formation but demonstrate that signalling by aIIbb3 is inactive or low-affinity conformation in nonactivated not essential for this response. cells. Intracellular signals within the platelet (known as ‘inside-out’ signalling) promote a conformational Results change in the extracellular domain, leading to an increase in affinity, thereby promoting integrin–ligand Morphological changes of human platelets on interactions [2]. In turn, clustering of integrins gene- fibrinogen, CRP and thrombin rates a series of intracellular signals (‘outside-in’ signal- ling) that serve to reinforce platelet activation [3]. The three ligands fibrinogen, CRP and thrombin sup-
The major platelet integrin aIIbb3 is a receptor for port platelet adhesion and lamellipodia formation when fibrinogen, von Willebrand factor, vitronectin, CD40 presented as a monolayer, even though they bind to ligand and fibronectin. Integrin aIIbb3 plays a vital role distinct classes of surface receptor. This raises the ques- in supporting platelet adhesion to the extracellular tion of the molecular basis of adhesion to these ligands matrix and promoting platelet–platelet interaction and whether they induce distinct patterns of change in
(aggregation). In addition, integrin aIIbb3 generates morphology. To address this, real-time imaging of outside-in signals that mediate platelet activation. It is platelets adhering to each surface was undertaken using now established that engagement of aIIbb3 activates a coating of ligands that induces maximal platelet adhe- Src family kinases, leading to activation of Syk [4], sion. These experiments were performed in the presence SLP-76 [5], Vav1 ⁄ 3 [4] and phospholipase Cc2 (PLCc2) of concentrations of apyrase and indomethacin shown [6], and thereby to activation of several second messen- to fully block the effect of the feedback agonists ADP 2+ ger pathways, including protein kinase C [7] and Ca and thromboxane A2, respectively, in order to directly [8]. These signalling events promote actin assembly, monitor the ability of each ligand to support adhesion leading to formation of filopodia, lamellipodia and and lamellipodia formation. stress fibres [9]. Consistent with previous reports, our results dem- In addition to integrin ligands, formation of filo- onstrate that platelets exposed to immobilized fibrin- podia and lamellipodia has been described on a ogen go through sequential formation of filopodia monolayer of collagen-related peptide (CRP), which and lamellipodia over a period of 30 min (Fig. 1 and selectively activates the immunoglobulin receptor glyco- supplementary Video S1). Both structures were stable protein VI (GPVI) [10,11], and on thrombin that has and did not retract once formed, although discrete been immobilized by fibrin. In platelets, thrombin binds movements could still be seen around the periphery to and signals via GPIba and the G protein-coupled of the cell. Fluorescent labelling of the actin cytoske- protease-activated receptor (PAR) receptors (PAR 1 leton revealed that stress fibres were formed within and 4) [12,13]. Immobilized thrombin that has become platelets that had undergone full lamellipodia forma- trapped by fibrin is able to promote platelet adhesion tion (Fig. 1B). In contrast, a distinct pattern of and aggregate formation at intermediate rates of flow, platelet morphological changes was observed on the leading to the speculation that it may function as an GPVI-specific agonist CRP (Fig. 1A and supplement- adhesion ligand in vivo [13]. In epithelial cells, it has ary Video S2). Limited small filopodia were seen, been suggested that immobilized thrombin can bind to with wave-like lamellipodia appearing before filopo- integrins through an RGD site, raising the possibility dia formation was complete, in contrast to the that is may bind directly to integrins in platelets [14]. spherically synchronized growth on fibrinogen. Full The present study was undertaken to investigate the lamellipodia formation was reached within 5–9 min, mechanism by which CRP and thrombin are able to three times more rapidly than on fibrinogen support platelet lamellipodia formation in comparison (Fig. 1C). Strikingly, even after platelets had reached to that induced by fibrinogen. It was of particular 90% of their final surface area, the lamellipodia were interest to discern whether thrombin and CRP stimu- very dynamic (supplementary Video S2), even though late lamellipodia formation directly, or whether they they were accompanied by formation of stress fibres require aIIbb3. The results demonstrate a critical role (Fig. 1B). A similar pattern of rapid yet unstable for aIIbb3 in promoting platelet lamellipodia formation lamellipodia formation and stress fibre formation on CRP, thrombin and on fibrinogen, but that out- was observed for platelets adhering to immobilized side-in signalling by the integrin is not required for thrombin (Fig. 1A,B and supplementary Video S3). lamellipodia formation on fibrinogen in the presence Significantly, lamellipodia formation on thrombin of thrombin. These results further emphasize the was not altered in the presence of the fibrin poly- importance of integrin engagement in lamellipodia merization inhibitor Gly-Pro-Arg-Pro (GPRP) (data
FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS 5033 Platelet lamellipodia formation via aIIbb3 K. Thornber et al.
A
B
C 120
100
80
60 FG CRP 40 THR % of maximum surface area 20
0 0 200 400 600 800 1000 1200 1400 1600 Time (sec)
Fig. 1. Distinct morphological changes in human washed platelets exposed to fibrinogen, collagen-related peptide (CRP) and thrombin. (A) Human washed platelets were exposed to surfaces of fibrinogen (FG), CRP or thrombin (THR). Representative morphology of a single plate- let on each surface at the time points shown (s). (B) Rhodamine–phalloidin staining to show actin stress fibres of platelets after 30 min of exposure to fibrinogen, CRP and thrombin. (C) Mean surface area of platelets at indicated time points, quantified using IMAGEJ. Values are mean ± SEM of the percentage of maximum surface area of five platelets from three independent experiments. not shown), demonstrating that this response was Different Ca2+ mobilization patterns in platelets not dependent upon fibrin formation. exposed to fibrinogen, CRP and thrombin These results demonstrate that a distinct pattern of morphological change is induced by platelet adhesion to Experiments were undertaken to investigate whether CRP and thrombin, compared to that seen on fibrin- the distinct pattern of lamellipodia formation on ogen. the three ligands is associated with differences in the
5034 FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS K. Thornber et al. Platelet lamellipodia formation via aIIbb3 elevation of intracellular Ca2+, in view of the pivotal adhesion and lamellipodia formation are mediated role that the divalent cation plays in the regulation of through integrin activation, in view of the critical role actin rearrangements in platelets through pathways of integrins in mediating adhesion and cell spreading. such as activation of the capping protein gelsolin [10] To address this, experiments were designed to compare 2+ and myosin light chain kinase. Intracellular Ca was the role of the major platelet integrin aIIbb3, in sup- measured by loading with the calcium reporter dye porting spreading on the three surfaces, using the
Oregon Green-BAPTA 1-AM and monitoring fluore- aIIbb3 antagonist lotrafiban [15,16]. scent fluctuations for a minimum of 5 min after adhe- As expected, lotrafiban abrogated platelet adhesion sion to each surface. The functional role of Ca2+ was to fibrinogen (Fig. 2B), but, importantly, dramatically investigated by pretreatment with the Ca2+ chelator reduced lamellipodia formation on CRP and thrombin 1,2-bis-(o-aminophenoxy)-ethane-N,N,N¢,N¢-tetraacetic surfaces, although it also increased the level of adhe- acid (BAPTA-AM). sion (Table 1). These findings demonstrate that despite Our results demonstrate that platelet adhesion to the distinct morphological changes on CRP and fibrinogen was followed by a characteristic lag phase thrombin in comparison to those on fibrinogen, the of 1–3 min before a series of rhythmic spikes of Ca2+. generation of lamellipodia is dependent upon engage-
These were seen at intervals of approximately 30 s, ment of aIIbb3 in all cases. This is most likely explained each lasting 5–15 s. Generally, the magnitude of spikes by release of fibrinogen, von Willebrand factor and declined slightly over time (Fig. 2A). In contrast, adhe- other aIIbb3 ligands from platelet a-granules. Consistent sion to CRP generated an immediate increase in intra- with this, secretion of platelet a-granules in platelets cellular Ca2+, which subsequently declined over a that had adhered to CRP and thrombin was confirmed period of 3–10 min. Minimal oscillations were by immunofluorescence staining for the a-granule observed during the sustained elevation in Ca2+. marker P-selectin, on the surface of adhered platelets A distinct pattern of intracellular Ca2+ signalling was (Fig. 3). observed in platelets on thrombin. An initial rapid ele- These results demonstrate that formation of lamelli- vation in intracellular Ca2+ was followed by a sus- podia on CRP, thrombin and fibrinogen is dependent 2+ tained phase of increased Ca levels superimposed by on engagement of integrin aIIbb3 as a consequence of 2+ a series of small Ca oscillations (Fig. 2A). release of aIIbb3 ligands from platelet a-granules. Maximal concentrations of the Ca2+ chelator BAPTA- AM inhibited intracellular Ca2+ elevation on all three Thrombin can bypass a b outside-in signalling surfaces by over 90% (Fig. 2A). This was accompanied IIb 3 in mediating lamellipodia formation by a complete inhibition of lamellipodia on fibrinogen, whereas only a partial inhibition of lamellipodia was The above observations demonstrate a critical role for observed on CRP and thrombin (Fig. 2B, Table 1). aIIbb3 in lamellipodia formation on CRP and throm- Platelet adhesion to thrombin was significantly reduced bin. This could be due to a role of aIIbb3 in supporting in the presence of BAPTA-AM, whereas the degree of adhesive events or in generating intracellular signals adhesion to CRP and fibrinogen remained unchanged that drive formation of lamellipodia. It is difficult to (Table 1). Filopodia formation was observed on all distinguish between these two possibilities in the case three surfaces in the presence of BAPTA-AM, indica- of CRP, because of the similarity in the signalling 2+ ting that their formation is independent of Ca pathways used by GPVI and integrin aIIbb3, both of mobilization. which are mediated by sequential activation of Src and Taken together, these results demonstrate a critical Syk family kinases and subsequent activation of role for Ca2+ in lamellipodia formation on all three PLCc2 [6]. In contrast, thrombin signals through a surfaces and suggest that distinct patterns of Ca2+ G protein-dependent pathway and induces a full signalling could contribute to the different patterns of repertoire of platelet responses in the presence of the lamellipodia formation induced by each ligand. Src family kinase inhibitor 4-amino-5-(4-chlorophenyl)- 7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2). Experi- ments were therefore designed to investigate the effect a b is required for platelet lamellipodia IIb 3 of PP2 on the ability of thrombin to promote lamelli- formation on fibrinogen, CRP and thrombin podia formation during platelet adhesion. Although the above results demonstrate marked differ- As previously reported, the Src kinase inhibitor PP2 ences between the three ligands in the pattern of Ca2+ blocked formation of lamellipodia on fibrinogen [17] mobilization, which presumably reflect the differing (Fig. 4), with average platelet surface area being signi- signalling strengths of their receptors, it is possible that ficantly reduced from 26.3 ± 0.9 to 17.6 ± 0.7 lm2.
FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS 5035 Platelet lamellipodia formation via aIIbb3 K. Thornber et al.
A
B
Fig. 2. Different Ca2+ signalling patterns on each surface. Washed human platelets were exposed to surfaces of fibrinogen (FG), collagen- related peptide (CRP) or thrombin (THR). (A) Platelets were loaded with Oregon Green-BAPTA 1-AM calcium dye and exposed to each sur- face for 10 min. The fluorescent Ca2+ fluctuations of three single platelets on each surface are shown in the presence (+) or absence (–) of BAPTA-AM (10 lM, added 10 min prior to surface exposure). Representative traces from three individual platelets are shown. (B) Platelets were exposed to each surface for 45 min in the absence or presence of BAPTA-AM (10 lM, added 10 min prior to surface exposure) or the aIIbb3 antagonist lotrafiban (10 lM,10 min). Images shown are representative of at least three independent experiments. Adhesion and sur- face area data from these experiments are shown in Table 1.
5036 FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS K. Thornber et al. Platelet lamellipodia formation via aIIbb3
Table 1. Human platelet data. See Fig. 2 for experimental details. Values are reported as follows: adherent platelets ¼ mean ± SEM of three experiments; platelet surface area ¼ mean ± SEM of at least 100 cells. On a control surface of BSA, adhesion and surface area data are 1.5 ± 0.1 · 102 mm)2 and 7.3 ± 0.2 lm2, respectively.
Fibrinogen CRP Thrombin
Adherent platelets ⁄ Platelet surface Adherent platelets ⁄ Platelet surface Adherent platelets ⁄ Platelet surface Treatment mm2 (· 102) area (lm2) mm2 (· 102) area (lm2) mm2 (· 102) area (lm2)
None 57.6 ± 0.9 26.3 ± 0.9 65.7 ± 0.9 35.0 ± 1.1 67.9 ± 1.9 32.2 ± 1.1 BAPTA-AM 57.2 ± 0.6 17.7 ± 0.6a 50.5 ± 0.6 22.3 ± 1.2a 36.3 ± 1.0a 25.8 ± 0.9a Lotrafiban 3.3 ± 0.4a 9.3 ± 0.4a 95.1 ± 1.3a 16.7 ± 0.5a 93.0 ± 1.4a 12.8 ± 0.4a aP < 0.05 with respect to untreated samples for each surface.
Fig. 3. P-Selectin exposure on collagen-rela- ted peptide (CRP) and thrombin. Washed human platelets were exposed to surfaces of CRP or thrombin (THR) for 45 min before fixing and staining with fluorescein isothio- cyanate (FITC)-conjugated anti-P-selectin serum. Cells were imaged using differential interference contrast (DIC) or fluorescence (P-selectin). Images shown are representa- tive of at least three independent experi- ments.
In contrast, adhesion was not significantly altered. (37.5 ± 1.5 lm2), most likely reflecting the increased Importantly, the inhibitory effect of PP2 on lamelli- signalling strength of the platelet thrombin receptors podia formation could be completely overcome by PAR1 and PAR4. A similar effect was seen with the addition of thrombin (Fig. 4), which induced a larger PAR1-specific peptide, thrombin receptor activating increase in surface area to that induced by fibrinogen peptide [TRAP (data not shown)].
Fig. 4. Thrombin can overcome Src kinase inhibition of lamellipodia on fibrinogen. Washed human platelets were exposed to a surface of fibrinogen for 45 min in the absence or presence of the Src kinase inhibitor PP2 (20 lM, added 5 min prior to surface exposure) and ⁄ or thrombin (1 UÆmL)1, 1 min) in suspension (+ THR). Images shown are representative of at least three independent experiments.
FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS 5037 Platelet lamellipodia formation via aIIbb3 K. Thornber et al.
These results demonstrate that thrombin is able to These experiments extend the observations on stimulate formation of lamellipodia on fibrinogen in human platelets to mouse platelets, namely that inte- the absence of signalling downstream of the integrin. grin aIIbb3 is necessary for platelet lamellipodia forma- Under these circumstances, therefore, the integrin is tion on CRP, thrombin and fibrinogen, but that functioning solely as an adhesive receptor. thrombin is able to mediate lamellipodia formation in the absence of outside-in signalling from the integrin.
Role of aIIbb3 and PLCc2 in adhesion and lamellipodia formation of murine platelets The roles of aIIbb3 and a2b1 in platelet lamellipodia formation on collagen Experiments were undertaken in mouse platelets to confirm the role of integrin aIIbb3 in mediating lamelli- Experiments were designed to investigate whether the podia formation on fibrinogen, CRP and thrombin, critical role of aIIbb3 in mediating lamellipodia forma- and to establish the importance of PLCc2 activation in tion in human and mouse platelets on fibrinogen, CRP this response. These experiments served to extend the and thrombin also extends to a further platelet ligand, observations made in human to mice platelets and also collagen, which binds directly to a second platelet to provide an alternative line of evidence to support integrin, a2b1 as well as GPVI. This question was the results obtained above through the use of platelets addressed by monitoring lamellipodia formation on deficient in the aIIb integrin subunit and PLCc2. collagen in human platelets in the presence of the As previously documented [9,18,19], and in contrast aIIbb3 antagonist lotrafiban and in mouse platelets defi- to human platelets, mouse platelets generate only filo- cient in the integrin subunit aIIb. podia and limited lamellipodia on fibrinogen in the Human and mouse platelets generate filopodia and presence of apyrase and indomethacin (Fig. 4A). In lamellipodia on fibrillar collagen (Fig. 6), and this is contrast, robust lamellipodia formation in mouse reduced by approximately 55% in human platelets in platelets is observed on CRP or thrombin surfaces. the presence of lotrafiban (28.7 ± 1.1 to 17.8 ± Adhesion and lamellipodia formation on fibrinogen 0.5 lm2, compared to 26.3 ± 0.9 to 9.3 ± 0.4 lm2 on were abrogated in the presence of the aIIbb3 antagonist fibrinogen) and in aIIb-deficient mouse platelets (from 2 lotrafiban (Fig. 5A, Table 2) and in mice lacking the 18.4 ± 0.6 to 11.3 ± 0.4 lm in wild-type and aIIb- aIIb gene (Fig. 5A, Table 2). Similarly, adhesion of deficient platelets, respectively, compared with 11.1 ± mouse platelets to CRP was fully blocked in the pres- 0.3 to 5.5 ± 0.5 lm2 on fibrinogen). The degree of ence of lotrafiban or in the absence of aIIb (Fig. 5A, platelet adhesion, however, was not significantly Table 2), demonstrating that adhesion on CRP is crit- altered by loss of aIIbb3 function in either human or ically dependent on functional aIIbb3, presumably as a mouse platelets (not shown), presumably because it is consequence of secretion of fibrinogen and other aIIbb3 mediated through integrin a2b1. = ligands. In contrast, aIIbb3-blocked and aIIb mouse These results demonstrate that aIIbb3 contributes to platelets retained the ability to adhere to immobilized lamellipodia formation on an integrin-binding ligand, thrombin, although lamellipodia formation was elimin- namely collagen, but that, in its absence, limited ated (Fig. 5A, Table 2). lamellipodia formation is mediated by integrin a2b1. The role of Src kinases and Ca2+ mobilization in lamellipodia formation in mouse platelets on fibrino- Discussion gen and CRP was investigated using platelets deficient in the major isoform of PLCc in platelets, PLCc2 It is well established that integrins play a critical role [19,20]. Blockade of Src kinases with PP2 (data not in lamellipodia formation in a wide variety of cell shown) or the absence of PLCc2 led to inhibition of types, although it is unclear whether lamellipodia for- lamellipodia formation on fibrinogen, but had no mation can also be induced by engagement of noninte- effect on adhesion (Fig. 5B, Table 2), in agreement grin receptors. Nevertheless, it is well established that with previous observations [18,19]. The addition of other cell surface receptors may facilitate cell adhesion thrombin in suspension could overcome this inhibitory and actin remodelling through integrin activation and effect, leading to extensive lamellipodia formation stimulation of actin polymerization. An example of (Fig. 5, Table 2). In contrast, the absence of PLCc2 this is Syndecan-1, which has been shown to mediate abrogated adhesion of mouse platelets to CRP lamellipodia formation in Raji lymphoblastoid cells (Fig. 5B), although adhesion and lamellipodia forma- independently of integrins [21], although other synde- tion could be restored by addition of thrombin (Fig. 5, cans are thought to operate through interaction with Table 2). neighbouring integrins [22]. More recently, two studies
5038 FEBS Journal 273 (2006) 5032–5043 ª 2006 The Authors Journal compilation ª 2006 FEBS K. Thornber et al. Platelet lamellipodia formation via aIIbb3
A
B