Research ArticleBiochemistry

Ca2+ waves coordinate purinergic receptor–evoked integrin activation and polarization

See allHide authors and affiliations

Science Signaling  21 Jan 2020:
Vol. 13, Issue 615, eaav7354
DOI: 10.1126/scisignal.aav7354

Ca2+ waves and integrin activation

The integrin GPIIb/IIIa is highly abundant on the surface of platelets and can be activated by intracellular Ca2+ signaling in an “inside-out” manner to bind to the adhesive ligand fibrinogen. Bye et al. imaged intracellular Ca2+ signaling and fibrinogen binding events in primary rat megakaryocytes activated through the ADP-stimulated receptors P2Y1 and P2Y12. The authors found that signaling by both receptors was required for full integrin activation, which depended on P2Y1-stimulated Ca2+ signaling and P2Y12-stimulated activation of the kinase PI3K. In addition, fibrinogen binding became polarized in these cells in a manner dependent on the direction of ADP-stimulated Ca2+ waves.


Cells sense extracellular nucleotides through the P2Y class of purinergic G protein–coupled receptors (GPCRs), which stimulate integrin activation through signaling events, including intracellular Ca2+ mobilization. We investigated the relationship between P2Y-stimulated repetitive Ca2+ waves and fibrinogen binding to the platelet integrin αIIbβ3 (GPIIb/IIIa) through confocal fluorescence imaging of primary rat megakaryocytes. Costimulation of the receptors P2Y1 and P2Y12 generated a series of Ca2+ transients that each induced a rapid, discrete increase in fibrinogen binding. The peak and net increase of individual fibrinogen binding events correlated with the Ca2+ transient amplitude and frequency, respectively. Using BAPTA loading and selective receptor antagonists, we found that Ca2+ mobilization downstream of P2Y1 was essential for ADP-evoked fibrinogen binding, whereas P2Y12 and the kinase PI3K were also required for αIIbβ3 activation and enhanced the number of Ca2+ transients. ADP-evoked fibrinogen binding was initially uniform over the cell periphery but subsequently redistributed with a polarity that correlated with the direction of the Ca2+ waves. Polarization of αIIbβ3 may be mediated by the actin cytoskeleton, because surface-bound fibrinogen is highly immobile, and its motility was enhanced by cytoskeletal disruption. In conclusion, spatial and temporal patterns of Ca2+ increase enable fine control of αIIbβ3 activation after cellular stimulation. P2Y1-stimulated Ca2+ transients coupled to αIIbβ3 activation only in the context of P2Y12 coactivation, thereby providing an additional temporal mechanism of synergy between these Gq- and Gi-coupled GPCRs.


The αIIbβ3 integrin (GPIIb/IIIa) is one of the most highly expressed platelet surface proteins and has a key role in hemostasis through its ability to bind to several adhesive ligands, particularly fibrinogen and von Willebrand factor (1). Blood cell integrins exhibit a low affinity for their ligands under resting conditions; however, “inside-out” activation after adequate cellular signaling induces a conformational change that promotes the high-affinity state and, thus, ligand binding (2). Among the signaling events that couple platelet receptors to αIIbβ3 integrin activation, cytosolic Ca2+ flux plays a crucial role through its ability to stimulate the Ca2+-sensor CalDAG-GEFI and, thus, also the small G protein Rap1 (3, 4). Sufficient activation of Rap1 promotes the binding of talin to αIIbβ3 to shift the integrin into its high-affinity state (58). However, guanosine 5′-triphosphate (GTP) loading of Rap1 by CalDAG-GEFI is opposed by the constitutive GTPase-activating protein (GAP) activity of RASA3; thus, αIIbβ3 integrin activation downstream of phospholipase C (PLC)–coupled receptors is limited, unless accompanied by phosphoinositide 3-kinase (PI3K)–induced inhibition of RASA3, for example, after stimulation of Gαi-coupled P2Y12 receptors (5).

Intracellular Ca2+ responses do not occur as uniform increases throughout the cytoplasm, but instead display complex spatiotemporal patterns in the form of restricted punctate events, oscillations, and waves (9, 10). Several studies have demonstrated the functional relevance of the spatial and temporal information within these cytosolic Ca2+ increases. For example, the frequency of oscillations determines both the efficiency and specificity of gene transcription in lymphocytes (11), and unitary Ca2+ events can play different roles compared to those of global Ca2+ increases during smooth muscle contraction (12). Ca2+ waves promote intracellular signaling over a long distance in large cells (1315) and coordinate certain types of cardiac myocyte contraction (16). In the platelet, Ca2+ transients triggered by environmental stimuli induce procoagulant activity (17, 18) in a manner that is dependent on the Ca2+ spiking pattern of the individual cell (19, 20). However, the relationship between the spatiotemporal patterns of cytosolic Ca2+ signals and inside-out integrin activation is unclear.

Previous reports have raised concerns that cell culture systems do not fully reproduce the events underlying agonist-evoked integrin activation in highly specialized cells such as platelets (21). Here, we used freshly isolated megakaryocytes (MKs) as a primary cell type that displays repetitive Ca2+ waves (22, 23) and robust inside-out activation of the αIIbβ3 integrin (24, 25) to investigate the interaction between these two events. We showed that transient Ca2+ waves downstream of P2Y1 receptors coupled to αIIbβ3 activation within seconds, but only when P2Y12 receptors were costimulated. The extent of fibrinogen binding was dependent on both the amplitude and interval of the Ca2+ transient. Although ionophore-generated, sustained Ca2+ increases alone activated the integrin, this occurred with a substantial delay. Thus, the acceleration of Ca2+-dependent inside-out αIIbβ3 activation by P2Y12 receptors represents a key component of the synergy between Gq- and Gi-coupled P2Y receptors. We also provide evidence that the net direction of the adenosine 5′-diphosphate (ADP)–evoked Ca2+ waves may explain the polarization of fibrinogen binding observed during the later stages of this functional response.


P2Y-evoked Ca2+ transients mediate precise temporal activation of αIIbβ3

To investigate the relationship between intracellular Ca2+ and αIIbβ3 activation, primary rat MKs were exposed to a hydrolysis-resistant ligand of both P2Y1 and P2Y12 receptors, ADPβS, which avoids the loss of agonist due to ectonucleotidase activity in Ca2+-containing external saline (26). ADPβS evoked a series of transient increases in whole-cell intracellular Ca2+ and stimulated fibrinogen binding at the plasma membrane, which served as a marker for activation of the integrin αIIbβ3 (Fig. 1A and movie S1). Peripheral fibrinogen binding increased in a stepwise manner that coincided with the occurrence of [Ca2+]i transients (Fig. 1B). Upon closer analysis, a discrete increase in fibrinogen binding (Fig. 1C) was found to follow each [Ca2+]i transient (Fig. 1D). It was therefore possible to quantify the properties of each αIIbβ3 activation event in terms of the peak increase in fibrinogen binding (peak ΔF) and the net change in fibrinogen binding between successive [Ca2+]i transients (net ΔF). We also quantified the peak of each [Ca2+]i transient (peak ΔF/F0) and the interval between transients. Correlation was observed between the maximum increase in fibrinogen binding (peak ΔF) and the peak increase in Ca2+ (peak ΔF/F0) in 91% of MKs (Fig. 1E). This indicates that the amplitude of the Ca2+ transient determines the magnitude of αIIbβ3 activation. A statistically significant correlation was also observed between the net change in bound fibrinogen at the end of the period between successive Ca2+ transients (net ΔF) and the interval of the Ca2+ transient in 75% of MKs (Fig. 1F). This further suggests that the frequency of Ca2+ transients determines whether a net accumulation or reduction in the amount of activated, fibrinogen-bound integrin αIIbβ3 occurs between successive Ca2+ transients. The gradient describing this relationship was not statistically significantly different between individual MKs (mean value, −1.28 ΔF s−1 ± 0.518). The x intercept, which reflects the intertransient interval below which the net fibrinogen binding switches from an increase to a decrease, varied between different cells (range, 12 to 44 s) as expected from the intercellular heterogeneity in the magnitude of the Ca2+ responses and, thus, also the fibrinogen binding events. Together, these data suggest that Ca2+ transients promote both the immediate and the cumulative activation of αIIbβ3 through both their amplitude and frequency of occurrence.

Fig. 1 Integrin αIIbβ3 activation is dependent on the timing and magnitude of ADPβS-evoked, repetitive Ca2+ transients.

(A to F) Characteristics of the [Ca2+]i transients and fibrinogen accumulation as measured by confocal fluorescence microscopy analysis of primary rat MKs after stimulation with ADPβS at the indicated times. Whole-cell images (A) and fluorescence traces (B to D) of the cytoplasmic Ca2+ concentration (blue) and peripheral fibrinogen binding (red) are representative of six independent experiments performed in duplicate. Fibrinogen binding events were quantified in terms of the peak increase (peak ΔF) and net change (net ΔF). (E and F) Linear regression analysis of quantified maximal fibrinogen binding events (peak ΔF) versus peak Ca2+ F/F0 (E) or the net change in fibrinogen binding (net ΔF) versus Ca2+ transient interval (F). Relationships for individual experiments are shown by the gray lines, and the average of all experiments is indicated by the red line.

P2Y1-dependent repetitive Ca2+ increases stimulate αIIbβ3 activation only in combination with P2Y12-dependent signaling

Individual Ca2+ transients evoked by P2Y receptors stimulated a finite increase in integrin αIIbβ3 activation after a mean delay of 2.9 ± 0.75 s (measured from initiation of the [Ca2+]i transient to the onset of fibrinogen binding; Fig. 2, A and B). Loading of MKs with the Ca2+ chelator BAPTA to an extent sufficient to prevent a detectable ADPβS-evoked Ca2+ increase abolished the fibrinogen binding event (Fig. 2C). This confirms previous evidence that increased [Ca2+]i is an essential component of the P2Y receptor signaling events, leading to integrin αIIbβ3 activation (27, 28). Overlay of the integral of the [Ca2+]i increase further highlighted the close association of this cytosolic signal with subsequent integrin activation and the consistent lag between the two events across multiple transients (Fig. 2D).

Fig. 2 Rapid activation of integrin αIIbβ3 is dependent on synergy between P2Y1 and P2Y12 receptors.

(A to I) Confocal fluorescence microscopy analysis of [Ca2+]i transients and fibrinogen accumulation in primary rat MKs after stimulation with ADPβS (A to F) or ionomycin (G to I) over the indicated times. Responses are representative of at least three independent experiments. The blue traces represent the whole-cell intracellular Ca2+ responses, and the red traces represent peripheral fibrinogen binding. (D) Cumulative integrated Ca2+ increase (green) alongside Ca2+ and fibrinogen responses during the initial four Ca2+ transients of the experiment shown in (A). (A to D) Responses to ADPβS were compared after cytoplasmic loading with the Ca2+ chelator BAPTA (C), incubation with the P2Y1 antagonist MRS2179 (E), incubation with the P2Y12 antagonist cangrelor (F), or exposure to vehicle as a control (B). (H and I) Responses to ionomycin after incubation with vehicle (H) compared with responses to ionomycin and ADPβS when MKs had been treated with MRS2179 (I). (J) Quantification of the mean lag period between the onset of [Ca2+]i flux and fibrinogen binding in response to ADPβS or ionomycin with or without MRS2179 or cangrelor. Data are means ± SD of at least three independent experiments. ***P < 0.001 by one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test; n.s., not significant; n/a, not applicable due to complete inhibition.

ADP stimulates two receptors in platelets, P2Y1 and P2Y12, which couple to Gq and Gi, respectively (2935). The selective P2Y1 antagonist MRS2179 ablated both ADPβS-evoked Ca2+ signaling and integrin αIIbβ3 activation (Fig. 2E). In contrast, ADPβS still stimulated a Ca2+ response in the presence of a maximal concentration of the P2Y12 antagonist, cangrelor, although fibrinogen binding was similarly abolished (Fig. 2F). Thus, signaling downstream of both P2Y1 and P2Y12 is required for activation of integrin αIIbβ3 by ADPβS in the rat MK, as was previously observed in platelets and murine MKs (25, 26, 3033, 35, 36). The number of Ca2+ transients was reduced from 22 per 9-min period in control settings to only 1 to 3 transients per recording after blockade of P2Y12 receptors, indicating the ability for this Gi-coupled receptor to prolong the Ca2+ response. A direct enhancement of a sustained [Ca2+]i increase in response to the ionophore ionomycin also induced peripheral fibrinogen binding (Fig. 2G and movie S2). This indicates that a prolonged increased amount of intracellular Ca2+ stimulates inside-out activation of αIIbβ3, consistent with a previous study of platelet aggregation (37). However, despite stimulating a larger peak Ca2+ increase than that observed in response to ADPβS (Fig. 2A), the ionomycin-evoked fibrinogen binding occurred with a statistically significantly greater delay compared with P2Y receptor stimulation (23.6 ± 12.3 s versus 2.9 ± 0.75 s after the initial Ca2+ increase, respectively; Fig. 2H). In addition, stimulation of P2Y12 with ADPβS together with ionomycin under conditions of P2Y1 receptor blockade with MRS 2179 (Fig. 2I) had no statistically significant effect on the rate of onset of αIIbβ3 activation compared with that in response to ionomycin alone (Fig. 2J). This suggests that an additional signal downstream of P2Y1 or a difference between the spatiotemporal pattern of the ADP-evoked versus the ionomycin-stimulated Ca2+ increase contributes to the ability of this Gq-coupled receptor to synergize with P2Y12 in the acceleration of αIIbβ3 activation.

Synergy between Ca2+ transients and PI3K signaling is required for the activation of αIIbβ3

We next explored the mechanism responsible for temporal synergy between P2Y1 and P2Y12 during ADP-dependent αIIbβ3 activation in experiments with pharmacological inhibitors of signaling events downstream of these receptors. In MKs and platelets, the Gq-coupled P2Y1 receptor is required for the Ca2+ response through its coupling to PLC and, thus, stimulation of IP3 (inositol 1,4,5-trisphosphate)–dependent Ca2+ mobilization (25, 29, 33). We found that both the [Ca2+]i increase and the fibrinogen binding evoked by ADPβS (Fig. 3A) were ablated by incubation with the PLC inhibitor U73122, whereas the structurally related, non–PLC-inhibiting analog U73343 had no effect (Fig. 3, B and C). P2Y1 receptors also release diacylglycerol (DAG) and, thus, stimulate protein kinase C (PKC); however, the pan-PKC inhibitor GF109203 had no effect on ADPβS-stimulated fibrinogen binding (Fig. 3D). GF109203 modified the Ca2+ response by preventing its return to baseline between [Ca2+]I transients, which is likely a consequence of the role of PKC in mediating P2Y1 receptor desensitization and internalization (38, 39). The ability of P2Y12 receptors to modulate the P2Y1-dependent Ca2+ response has been suggested previously to be mediated through activation of PI3K and also inhibition of adenylyl cyclase (AC), although amplification of [Ca2+]i by AC inhibition requires costimulation of Gs-coupled receptors (40, 41). Roles for both PI3Kβ and PI3Kγ in P2Y12-mediated integrin activation have been demonstrated using both pharmacological tools and isoform-specific knockout mice (30, 42). Consistent with these previous observations, we found that the pan-PI3K inhibitor LY294002 mimicked the action of cangrelor by abolishing the fibrinogen response (Fig. 3E), confirming the crucial role for PI3K in P2Y12-evoked integrin αIIbβ3 activation (4345). LY294002 also limited the Ca2+ response to only one to two transients. In contrast, inhibition of AC with SQ22536 did not enhance the ADP-evoked Ca2+ response (Fig. 3F) or rescue it in the presence of cangrelor (Fig. 3, G and H). Because PI3K proved critical for ADPβS-evoked activation of αIIbβ3, we treated MKs with LY294002 before stimulation with ionomycin and found that fibrinogen binding still occurred, but only after a statistically significantly increased delay (235 ± 96 s) relative to that in vehicle-treated MKs (Fig. 3I). This effect can be explained by an ability of ionophore-induced [Ca2+]i increases to directly activate PI3K (46). Ionomycin-induced fibrinogen binding was totally abolished by the PLC inhibitor U73122 (Fig. 3J), but not by the control analog U73343 (Fig. 3K), which is consistent with the reported ability of a large sustained [Ca2+]i increase to activate this Gq-coupled enzyme in platelets (47). Together, these results indicate a nonredundant role for PLC in the pathways leading to αIIbβ3 activation, consistent with effects observed in patients with a platelet-specific deficiency in PLCβ (4850).

Fig. 3 Enhanced [Ca2+]i and activation of PI3K are required for rapid αIIbβ3 activation.

(A to K) Confocal fluorescence microscopy analysis of [Ca2+]i transients and fibrinogen accumulation in primary rat MKs after stimulation with ADPβS (A to G) or ionomycin (I to K). Intracellular Ca2+ (blue) and fibrinogen binding (red) responses are representative of at least three independent experiments for ADPβS (A to G) and ionomycin (I to K). Cells were treated with vehicle (A) or the specified inhibitors U73122 (B and J), U73343 (C and K), GF109203 (D), LY294002 (E and I) ,SQ22536 (F), or a combination of SQ22536 and cangrelor (G). For the control ionomycin-evoked response, see Fig. 2G. (H) Peak increase in fibrinogen in response to ADPβS for all cells in each of the conditions shown in (A) to (G); n ≥ 3 for each condition; bar shows the mean ± SD. *P < 0.05 by one-way ANOVA with Dunn’s multiple comparisons test.

P2Y-evoked Ca2+ waves promote spatial reorganization of activated αIIbβ3

In primary MKs, the transient Ca2+ increases occurred as waves that traveled across the cytosol (22, 23) and had a clear point of origin and direction of travel (Fig. 4A). Because [Ca2+]i transients are critical in determining the temporal pattern of αIIbβ3 activation, we investigated whether the waves also modulated the spatial characteristics of fibrinogen binding. The net direction of the Ca2+ waves in each cell was analyzed by calculating a Ca2+ polarity score. A value of 1 represents an MK in which all of the Ca2+ waves traveled in the same direction, whereas a score of 0 represents an MK in which there was no net dominant directionality. We generated images of the first 10 Ca2+ waves and the radar plots to illustrate the percentage of waves traveling in each direction throughout a recording for a cell with a low Ca2+ polarity score (Fig. 4, B and C) and for a cell with a high score (Fig. 4, D and E). Heterogeneity in different MKs was observed in the Ca2+ polarity score, which ranged from 0.73 to 0.08 (Fig. 4F).

Fig. 4 Directional polarity of P2Y receptor–evoked Ca2+ waves in MKs.

(A to F) Spatial analysis of ADPβS-evoked transient [Ca2+]i waves in primary rat MKs. (A to C) Confocal fluorescence images showing the progression of a single [Ca2+]i wave (A) or 10 consecutive waves (B and D). (A) Left: The octants used to calculate radar plots and Ca2+ wave score are displayed in the pre-stimulus image. The progression of a Ca2+ wave is shown by the dotted red lines in subsequent images at the stated time after stimulation. (B and D) The red arrows depict the direction of travel of individual waves, whereas a dotted circle represents Ca2+ increases that lacked a clear direction of travel across the cell. The extent to which all waves within a single cell displayed a net polarity is summarized within radar plots based on the percentage of [Ca2+]i waves traveling toward each octant (C and E). (B and C) Images and analysis of an MK with a low polarity score. (D and E) Images and analysis of an MK with a high Ca2+ polarity score. (F) Quantification of the Ca2+ polarity score calculated for 12 MKs measured in seven independent experiments.

During the initial 1 to 2 min of a recording, P2Y receptor–evoked fibrinogen binding was distributed evenly around the plasma membrane of the MK. However, in many recordings, the pattern of fibrinogen binding then progressively shifted toward one side of the cell (Fig. 5A). This behavior was not observed after stimulation with ionomycin, which induced a nonwaveform, sustained [Ca2+]i increase and an even distribution of bound fibrinogen around the periphery of the MK throughout the duration of the experiment (Fig. 5B). Development of fibrinogen polarity was analyzed by calculating a fibrinogen polarity score for each MK. A score of 1 represents an MK in which 100% of the fibrinogen bound to one 120° sector of the cell surface, whereas a value of 0 indicates that the fibrinogen is evenly distributed over the membrane surface. Development of polarity peaked about 6 min after stimulation with ADPβS (Fig. 5C). As observed for the net direction of the Ca2+ waves, ADPβS-evoked fibrinogen binding polarity scores displayed marked heterogeneity, ranging from 0.08 to 0.73 (Fig. 5D). Despite the heterogeneity, the average fibrinogen polarity score for ADPβS (mean, 0.34 ± 0.24) was statistically significantly higher than that observed in response to ionomycin, which induced a consistently low value (mean, 0.15 ± 0.07).

Fig. 5 Development of a polarized integrin αIIbβ3 distribution after stimulation by ADPβS, but not ionomycin.

(A to D) Confocal fluorescence microscopy analysis of fibrinogen binding to primary rat MKs after stimulation with ADPβS or ionomycin. Fluorescence intensity radar plots (top) and confocal images (bottom) of surface-bound fibrinogen in a single MK before and at 2-min intervals after stimulation with ADPβS (A) or ionomycin (B). (C) Quantification of the fibrinogen polarity score against time after stimulation with ADPβS [red line corresponding to the cell shown in (A)] or ionomycin [blue line corresponding to the cell shown in (B)]. (D) Scatter plot comparing fibrinogen polarity scores calculated 6 min after stimulation with ADPβS (n = 12) or ionomycin (n = 10). Data are means ± SD; *P < 0.05 by Students t test. The cells selected in (A) and (C) show representative data from an MK with a high polarity score after stimulation with ADPβS (n = 6) and all MKs after treatment with ionomycin (n = 10).

After stimulation with ADPβS, MKs with low Ca2+ polarity scores retained a uniform distribution of fibrinogen binding at their periphery (Fig. 6, A to C). In contrast, MKs with a high Ca2+ polarity score displayed a shift in their fibrinogen binding toward one hemisphere (Fig. 6, D to F). Linear regression analysis of the Ca2+ and fibrinogen polarity scores confirmed a statistically significant positive correlation (gradient. 0.6; P < 0.001) between these two parameters (Fig. 6G). For strongly polarized MKs (fibrinogen polarity scores, >0.5), the average point of origin of Ca2+ waves coincided with the side of the cell displaying the center of fibrinogen polarity (mean difference of 18° ± 5.6° between the origin of the [Ca2+]i waves and the center of fibrinogen polarity) (Fig. 6H). This indicates that the side of the MK from which most of the [Ca2+]i waves originated was also the location at which fibrinogen accumulated.

Fig. 6 Ca2+ wave direction coordinates the polarization of activated integrin αIIbβ3.

(A to G) Confocal fluorescence microscopy and analysis of the relationship between [Ca2+]i waves and fibrinogen binding in primary rat MKs after stimulation with ADPβS. (A to C) Representative cell showing low polarity scores for both Ca2+ wave direction and fibrinogen binding. (D to F) Representative of a cell with high polarity scores for both parameters. Radar plot analysis (A, B, D, and E) of Ca2+ wave direction (A and D) and fibrinogen binding (B and D) together with fluorescence images of fibrinogen binding (C and F) before and every 2 min after stimulation. (G) Linear regression analysis of fibrinogen polarity score against Ca2+ polarity score for 12 MKs, which gave a statistically significant nonzero fit (P < 0.05) with a positive gradient (0.59). (H) Schematic diagram representing the difference in angle between the average point of origin of Ca2+ waves and the center of fibrinogen polarity in five MKs that had a fibrinogen polarity score >0.5.

Mobility and reorganization of activated αIIbβ3 depend on its association with the actin cytoskeleton

Inside-out activation of integrins such as αIIbβ3 causes association with the actin cytoskeleton through talin (5154), which limits the mobility of these adhesive receptors within the plasma membrane (55). If αIIbβ3 is unable to diffuse freely, then cytoskeletal reorganization might form part of the mechanism underpinning the development of its polarized distribution. We therefore investigated the diffusive properties of activated αIIbβ3 using fluorescence recovery after photobleaching (FRAP) of Alexa Fluor 647–labeled fibrinogen bound to the plasma membrane 4 min after stimulation with ADPβS (Fig. 7A). The fluorescence recovered to only 24 ± 7.5% of prebleached levels by the end of the 220-s recording (Fig. 7, B and C), suggesting that most of the αIIbβ3 receptors were immobile within the plasma membrane. Thus, diffusion of the integrin is unlikely to be an important factor in the development of polarity. To test whether the association of αIIbβ3 with the cytoskeleton limited its mobility, we repeated the FRAP measurements after treatment with cytochalasin D, an inhibitor of actin polymerization (Fig. 7, B and C). Diffusion of bound fibrinogen was statistically significantly increased by cytochalasin D treatment (mean recovery, 36 ± 13%), suggesting that the actin cytoskeleton has a role in limiting the plasma membrane mobility of activated αIIbβ3 and that major changes in αIIbβ3 localization are likely to occur only through reorganization of the actin cytoskeleton.

Fig. 7 The mobility and distribution of activated αIIbβ3 in the plasma membrane are dependent on its association with the actin cytoskeleton.

(A to C) FRAP analysis of fibrinogen bound to ADPβS-stimulated MKs as measured by confocal fluorescence microscopy. (A) Images of fibrinogen bound to an ADPβS-stimulated MK before and immediately after bleaching of a defined region (dotted circle) at the cell periphery. (B) Normalized fluorescence recovery over time within the bleached region after treatment with vehicle (black line) or cytochalasin D (Cyt D; red line). The responses are representative of six independent experiments. (C) Average maximum recovery of fluorescence within the bleached region after 4 min. Data are means ± SEM of six independent experiments. (D to G) Confocal fluorescence analysis of fibrinogen binding (red) and actin (green) in MKs fixed 4 min after stimulation with ADPβS. Data are representative of an MK with either a low (D and E) or a high (F and G) fibrinogen polarity score, taken from three independent experiments. (D and F) Seven consecutive fluorescence images at 1-μm intervals from a z-stack and with a three-dimensional reconstruction of the whole MK (right panel). (E and G) Radar plots of normalized fluorescence intensity for fibrinogen and actin around the circumference of the MKs for each confocal slice. *P < 0.05 by Students t test.

To further investigate the relationship between activated αIIbβ3 and the actin cytoskeleton, MKs stimulated with ADPβS for 4 min in the presence of Alexa Fluor 647–labeled fibrinogen were fixed and stained with Alexa Fluor 488–conjugated phalloidin. Confocal z-stack images showed that MKs without polarized fibrinogen binding also lacked polarization of actin structure in three dimensions (Fig. 7, D and E). In contrast, MKs in which the bound fibrinogen shifted to one side of the cell also displayed an increase in actin abundance in the same cellular hemisphere (Fig. 7, F and G). These data suggest that repetitive [Ca2+]i waves can drive directional reorganization of actin, leading to spatial redistribution of activated integrin αIIbβ3 at the cell surface.


Regulation of integrin affinity, referred to as inside-out signaling, enables cells to control their interactions with the extracellular environment. Inside-out signaling is a process common to all integrins and involves a conformational change triggered by intracellular events that results in exposure of high-affinity sites for ligand binding. αIIbβ3 has long served as a model for the cellular regulation of integrins and is highly abundant in platelets and their precursor cell, the MK. Although the critical events that underpin receptor-evoked αIIbβ3 activation are well established, the temporal regulation of inside-out signaling has not been explored in detail. Rapid activation of integrins is likely to be important in a number of contexts, not least under the dynamic conditions present in the circulation where the platelet must activate αIIbβ3 quickly enough to enable adhesion at sites of vascular injury. Primary MKs express functional αIIbβ3 that undergoes inside-out activation upon stimulation with ADP (24, 25), and, unlike in the diminutive platelet, it is possible to discern the quantal increase in integrin activation triggered by individual Ca2+ transients (Fig. 1). This has enabled us to dissect the temporal relationship between Ca2+ signaling events and integrin activation in detail.

We found that the activation of αIIbβ3 was efficiently induced by individual cytosolic Ca2+ transients after costimulation of the ADP receptors P2Y1 and P2Y12. Two characteristics of repetitive Ca2+ transients were found to control the overall rate of integrin activation. First, the amplitude of the whole-cell Ca2+ transient determined the magnitude of the increase in integrin αIIbβ3 activation after each transient. Second, because of constitutive deactivation of the integrin, the net change in integrin activation over time was controlled by the frequency of the Ca2+ transients. Another example of a downstream functional event that is controlled by the amplitude and frequency of Ca2+ transients is lymphocyte gene transcription (11, 56). Lymphocyte transcription factors are regulated differentially by the precise characteristics of Ca2+ oscillations and provide a basis for the selective stimulation of individual gene regulation pathways downstream of a single second messenger. Several studies showed that the pattern of Ca2+ increase in platelets can be specific to individual agonists (5759), and the relationship between Ca2+ transients and integrin activation described in the present study may therefore represent a mechanism whereby different agonist species activate integrins with varying levels of efficiency. In addition, because the frequency of Ca2+ oscillations in rat MKs increases with agonist concentration (60), this provides a mechanism whereby the extent of inside-out activation can be controlled by stimulus strength.

Although we found evidence that integrin activation and fibrinogen binding were reversible, the rate of release was much slower than that of binding, resulting in the accumulation of bound fibrinogen even at low Ca2+ transient frequencies. As a result, the pattern of fibrinogen binding bore a strong resemblance to the cumulative Ca2+ signal, suggesting that the accumulation of Ca2+-activated intracellular signaling molecules supports inside-out activation. In other cell types, transient Ca2+ signaling events are “memorized” by Ca2+-sensing signaling molecules that are rapidly activated but slowly deactivated (61). For example, in T cells, the Ca2+ calmodulin–dependent phosphatase calcineurin is activated by Ca2+ transients and subsequently dephosphorylates the transcriptional regulator NFAT (nuclear factor of activated T cells), enabling its translocation to the nucleus (11). Deactivation of NFAT by phosphorylation occurs slowly, such that the interval between Ca2+ transients controls the accumulation or loss of nuclear NFAT, resulting in a pattern that is highly reminiscent of integrin activation by Ca2+ transients in the MK. In platelets and MKs, the Ca2+-sensing molecule CalDAGGEFI mediates the activation of Rap1, which, in turn, mediates integrin activation (3). Rap1 is inactivated by the GAP RASA3 (62), and it is possible that the interplay between CalDAGGEFI, Rap1, and RASA3 regulates integrin activation in response to Ca2+ transients in a manner analogous to NFAT regulation in T cells.

The two ADP receptors expressed in MKs and platelets, P2Y1 and P2Y12, regulate different halves of the CalDAGGEFI-RASA3 axis that controls Rap1 activity. We exploited this by using receptor antagonists to explore the regulatory roles of P2Y1 and P2Y12 alone and in combination. The Gq-coupled P2Y1 receptor stimulates activation of PLCβ and Ca2+ transients that switch on the GTP-exchange activity of CalDAGGEFI, thereby promoting activation of the Ras-like small GTPase Rap1 (3, 4). Inside-out αIIbβ3 activation by ADP also requires coactivation of P2Y12-induced signals (Fig. 2) (34, 63). Although this Gi-coupled receptor is linked to the inhibition of AC and the activation of Akt, the crucial P2Y12 signaling event is the inactivation of RASA3 GAP activity that tonically limits GTP loading of Rap1 (5). RASA3 inactivation is induced by P2Y12-evoked PI3K activation, which is why LY294002 mimics the inhibitory effect of P2Y12 antagonists on ADP-evoked fibrinogen binding and Ca2+ responses (Figs. 2F and 3E) (44). Whereas Rap1 and RASA3 are already predominantly located at the plasma membrane (7, 64, 65), CalDAGGEFI and talin require translocation to the cell membrane during inside-out activation of αIIbβ3 (6668), which could therefore account for a substantial portion of the delay between the Ca2+ transients and integrin activation that we observed (Fig. 2). Although P2Y12 antagonists completely blocked the ADP-induced activation of αIIbβ3 even in the continued presence of Ca2+ transients, substantial fibrinogen binding was observed after the induction of a prolonged Ca2+ increase with ionomycin. This is consistent with the ability of the Ca2+ ionophore to induce Rap1 activation in platelets (69) and indicates that sustained CalDAGGEFI activation can overcome tonic RASA3 activity, although there remains a contribution from PI3K activation (Fig. 3I). However, the effect of an [Ca2+]i increase alone on fibrinogen binding occurred with a marked delay compared with that observed after ADP-evoked Ca2+ transients during P2Y1 and P2Y12 costimulation. The delay after ionomycin-induced Ca2+ increases (23.9 s) was longer than the duration of individual ADP-evoked Ca2+ transients (2.5 to 3 s). Thus, the synergy between CalDAGGEFI and RASA3 in promoting Rap1 activity enables brief Ca2+ transients to stimulate a discrete increase in integrin activation.

Preventing excessive and toxic activation of Ca2+-dependent proteins (70, 71) is envisaged to be one advantage of repetitive transients over sustained increases in this ubiquitous second messenger and, thus, is a further advantage of the P2Y1-P2Y12 synergy occurring at the level of Rap1 activation. We noted that rapid fibrinogen binding was not observed after costimulation of cells through P2Y12 and ionomycin, which suggests that either a difference between the Ca2+ increases evoked by P2Y1 and ionomycin or an additional signal downstream of the Gq-coupled receptor accelerates Rap1 activation alongside PI3K activation. However, PKC activation was not involved (Fig. 3D), consistent with the lack of a role for DAG in CalDAGGEFI activation (72, 73). Furthermore, inhibition of PLC mimicked the effect of BAPTA loading; thus, possibilities for the missing signal include IP3 formation or a decrease in PIP2 abundance, which requires further investigation.

As previously observed in the MK (23), each P2Y-evoked Ca2+ transient was initiated at one edge of the cell and spreads as a wave across the remainder of the cytoplasm before declining to resting levels. A higher density or greater sensitivity of IP3 receptors at the site of origin of the wave has been suggested to account for the point of initiation, which then evokes Ca2+-induced Ca2+ release through nearby IP3 receptors, resulting in a spread of the cytosolic Ca2+ increase across the cell (23, 74). In the MK, the wave origin has a predominant juxtanuclear location (23) and, together with previous observations that the polyploidic nucleus is predominantly located in one hemisphere of the cell (23), provides the basis for the polarity of the Ca2+ signal relative to the cell structure. Thus, the dependence of the fibrinogen binding on the direction of the Ca2+ wave could serve as a signal for directional migration of the MK to the sinusoidal blood vessels where proplatelet extension is projected across the vascular endothelium, enabling release of new platelets into the circulation, for which polarization is critical (75). Polarized regulation of the small GTPases Cdc42 and RhoA stimulates the GPIb-dependent transendothelial migration of MKs into the circulation (76). Furthermore, deletion of the two main platelet isoforms of Rap1 causes reduced proplatelet formation and, thus, macrothrombocytopenia (77).

In conclusion, here, we examined the relationship between spatiotemporal increases in intracellular Ca2+ and activation of high-affinity fibrinogen binding sites of the most prevalent platelet integrin, αIIbβ3. The work demonstrates that synergy between P2Y1 and P2Y12 receptors accelerates Ca2+-dependent inside-out activation of αIIbβ3 to enable individual P2Y1-evoked transient Ca2+ waves to stimulate discrete increases in fibrinogen binding. Spatiotemporal information encoded within the repetitive Ca2+ waves determined the extent of fibrinogen binding and also redistribution of the active integrin within the cell membrane.


Preparation of rat bone marrow MKs

Rat bone marrow was extracted from the tibia and femurs of adult male Wistar rats weighing between 300 and 400 g. Rats were killed by cervical dislocation in accordance with Home Office license regulations and the University of Leicester Home Office certificate of designation. The femurs and tibia were isolated, the epiphyses removed, and the marrow was then flushed out with 1 ml of Hanks’ balanced salt solution (HBSS).

Simultaneous fluorescence imaging of [Ca2+]i and fibrinogen binding in primary rat MKs

A bone marrow suspension was diluted 10-fold in HBSS (with 1.26 mM CaCl2) containing 2 μM Fluo-3 AM and 0.02% Pluronic F127, incubated for 1 hour at room temperature with gentle agitation, pelleted by centrifugation at 200g, and resupended in HBSS. Alexa Fluor 647–labeled fibrinogen (centrifuged at 13,600g for 10 min to remove precipitates) was added at a final concentration of 60 μg/ml. MKs were identified by their uniquely large size (diameter, >25 μm) and multilobular nucleus (78) and were imaged with an Olympus IX81 inverted confocal laser scanning microscope (Olympus, UK) and a 60× oil immersion 1.35–numerical aperture UPLSAPO objective. Cells were mounted in a 200-μl volume chamber (Digitimer Ltd., Welwyn Garden City, UK) in a stable “Z” temperature–regulated holder (Bioptechs Inc., Butler, PA) set to 25°C. Images were captured at 5.4 Hz with a 6× digital zoom at a resolution of 128 × 128 pixels for a total of 9 min. MKs were preincubated for 10 min with MRS2179 (100 μM), cangrelor (1 μM), LY294002 (50 μM), SQ 22536 (30 μM), U73343 (10 μM), U73122 (10 μM), or GF109203 (10 μM) as described in the figure legends. MKs were incubated with 10 μM BAPTA-AM for 60 min to chelate intracellular Ca2+ in some experiments as described. Stimulation with 100 μM ADPβS or 1 μM ionomycin was achieved by direct addition of these compounds to the chamber with an air displacement pipette held in position by a retort stand.

Solutions and chemicals

HBSS contained 1.26 mM CaCl2, 0.493 mM MgCl2, 0.407 mM MgSO4, 5.33 mM KCl, 0.441 mM KH2PO4, 4.17 mM NaHCO3, 137.93 mM NaCl, 0.338 mM Na2HPO4, 5.56 mM d-glucose, and 5 mM Hepes (pH 7.35) with NaOH. Pluronic F127, MRS2179, cytochalasin D, LY294002, formyl saline, and ADPβS were obtained from Sigma-Aldrich (Gillingham, UK). Fluo-3 AM, Alex Fluor 488–conjugated phalloidin, and Alexa Fluor 647–conjugated fibrinogen were from Invitrogen (Paisley, UK). SQ 22536, U73343, and U73122 were from Tocris Bioscience (Bristol, UK). Cangrelor was a gift from AstraZeneca (Moindal, Sweden).

Temporal analysis of whole-cell Ca2+ increases and fibrinogen binding

Temporal changes in intracellular Ca2+ and fibrinogen binding across the entire cell were analyzed with ImageJ software (Rasband, W.S., ImageJ, U.S. National Institutes of Health, Bethesda, Maryland, USA,, 1997–2016). Mean fluorescence intensity values were measured within a circular region of interest (ROI) containing the entire MK. Background-subtracted Fluo-3 fluorescence values (F) were converted to F/F0 ratios to normalize for the amount of fluorescence before the addition of agonist (F0). Increases in Alexa Fluor 647–conjugated fibrinogen fluorescence are expressed as the change in signal above that under prestimulated conditions (F-F0).

Calculation of Ca2+ wave and fibrinogen binding polarity scores

Ca2+ wave direction was assigned to one of eight directions (representing 45° sectors of the MK) and used to calculate the average wave direction and magnitude (Ca2+ polarity score) using vector analysis similar to that used to calculate average wind direction and speed (79). The following equations were used for the vector analysis:u=ui×sin[2π×θi360](1)v=ui×cos[2π×θi360](2)Ca2+ polarization score=(u2+v2)12(3)Ca2+ wave direction=arctan(uv)±180(4)

Equations 1 and 2 calculate vector components (u and v) using vector magnitude (ui, assigned an arbitrary value of 1) and wave direction in degrees (θi). Equation 3 calculates the average vector magnitude (Ca2+ polarity score) using the mean u and v for all of the Ca2+ waves measured in a single MK. Equation 4 calculates the mean Ca2+ wave direction using mean u and v, where 180° is added if arctan(uv) is <180 and 180° is subtracted if arctan(uv) >180.

Fibrinogen polarity was quantified by first identifying the 120° sector of each MK that had developed the greatest fluorescence intensity 6 min after stimulation. The sum fluorescence intensity of this sector (Fsectmax) and the total fluorescence intensity of the entire MK (FWC) were used to calculate the degree of fibrinogen polarity with Eq. 5.Fibrinogen polarity=Fsectmax×3Fwc1(5)

FRAP measurements

MKs were preincubated with 10 μM cytochalasin D or vehicle and then stimulated for 2 min with 100 μM ADPβS in the presence of Alexa Fluor 647–labeled fibrinogen (60 μg/ml). The labeled fibrinogen was then bleached by exposure to a 635-nm laser at maximum intensity for 300 ms within a 16-μm2 circular region at the cell periphery. Recovery of fluorescence was monitored by confocal imaging (Olympus FV1000, see above) at a rate of 0.2 Hz for 220 s. Longer recordings could not be performed because this assay was particularly sensitive to the small movements of MKs that naturally occur over time. Data were analyzed with the following equation to correct for nonspecific bleaching and normalized to the prebleach and postbleach fluorescence intensity values:Normalized Bx=(BxBgRxBg)(BbBgRbBg) (BiBgRiBg)(BbBgRbBg)(6)where Bx is the fluorescence intensity of the bleached ROI at x s, Bb is the fluorescence intensity of the bleached ROI immediately after photobleaching, Bi is the initial fluorescence intensity of the bleached ROI at, Rx is the fluorescence intensity of the reference ROI (nonbleached region of the MK), Rb is the fluorescence intensity of the reference ROI immediately after photobleaching, Ri is the initial fluorescence intensity of the reference ROI, and Bg is the average fluorescence intensity of the background ROI.


MKs suspended in HBSS with Alexa Fluor 647–labeled fibrinogen (60 μg/ml) were stimulated with 100 μM ADPβS for 6 min and then fixed in 8% formyl saline for 10 min. Cells were washed in HBSS, permeablized with 0.2% Triton X-100 for 10 min, and then washed again in HBSS. Cells were blocked for 1 hour in 1% (w/v) bovine serum albumin (BSA) and stained for 1 hour with Alexa Fluor 488–conjugated phalloidin (1:1000) in 1% BSA. Cells were then washed three times in HBSS and imaged with a 60× objective lens on a Nikon F1000 confocal microscope.

Statistical analysis

Statistical testing was performed as described in the individual figure legends with GraphPad Prism software (La Jolla, CA).


Movie S1. Intracellular Ca2+ signaling and fibrinogen binding after stimulation with ADPβS.

Movie S2. Intracellular Ca2+ signaling and fibrinogen binding after stimulation with ionomycin.


Acknowledgments: We thank C. Jones for the helpful comments and suggestions during manuscript preparation. We also thank A. Inayat for participation in the preliminary imaging experiments. Funding: This work was funded by the British Heart Foundation (PG/05/014 to M.P.M.-S. and RG/15/2/31224 to J.M.G.) and a BBSRC Studentship (to A.P.B.). Author contributions: A.P.B. designed the study, performed the experiments, analyzed the data, and wrote the manuscript. M.P.M.S. secured the funding, designed the study, and wrote the manuscript. J.M.G. wrote the manuscript. Competing interests: The authors declare that 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.

Stay Connected to Science Signaling

Navigate This Article