Ragged spiking of free calcium in ADP-stimulated human platelets: regulation of puff-like calcium signals in vitro and ex vivo

Abstract

1. Human platelets respond to agonists of G protein (G_q)-coupled receptors by generating an irregular pattern of spiking changes in cytosolic Ca²⁺ ([Ca²⁺]_i). We have investigated the ADP-induced Ca²⁺ responses of single, Fluo-3-loaded platelets in the presence or absence of autologous plasma or whole blood under flow conditions.

2. In plasma-free platelets, incubated in buffer medium, baseline separated [Ca²⁺]_i peaks always consisted of a rapid rising phase (median time 0.8 s) which was abruptly followed by a slower, mono-exponential decay phase. The decay constant differed from platelet to platelet, ranging from 0.23 ± 0.02 to 0.63 ± 0.03 s⁻¹ (mean ±s.e.m., n = 3–5), and was used to identify individual Ca²⁺ release events and to determine the Ca²⁺ fluxes of the events.

3. Confocal, high-frequency measurements of adherent, spread platelets (diameter 3-5 μm) indicated that different optical regions had simultaneous patterns of both low- and high-amplitude Ca²⁺ release events.

4. With or without plasma or flowing blood, the ADP-induced Ca²⁺ signals in platelets had the characteristics of irregular Ca²⁺ puffs as well as more regular Ca²⁺ oscillations. Individual [Ca²⁺]_i peaks varied in amplitude and peak-to-peak interval, as observed for separated Ca²⁺ puffs within larger cells. On the other hand, the peaks appeared to group into periods of ragged, shorter-interval Ca²⁺ release events with little integration, which were alternated with longer-interval events.

5. We conclude that the spiking Ca²⁺ signal generated in these small cells has the characteristics of a ‘poor’ oscillator with an irregular frequency being reactivated from period to period. This platelet signal appears to be similar in an environment of non-physiological buffer medium and in flowing, whole blood.

Putatively, because their function is to acutely interact with each other and with a damaged vessel wall, platelets are abundantly equipped with signalling receptors and a highly reactive signalling machinery. One of the earliest responses of platelets upon activation of G_q protein-coupled receptors is an increase in [Ca²⁺]_i. Thus, stopped-flow experiments, in which platelets were rapidly mixed with agonists such as thrombin or platelet-activating factor, have indicated that rises in [Ca²⁺]_i start within 200 ms of agonist-receptor contact (Sage & Rink, 1987). When mixed with ADP, a compound that is secreted by platelets themselves, the [Ca²⁺]_i rise begins even faster, i.e. after a lag time shorter than 10 ms (Sage et al. 1990). This Ca²⁺ signal involves stimulation of at least two P₂ purinergic receptor types. The initial component is caused by Ca²⁺ influx via the ionotropic P_2X1 receptor (MacKenzie et al. 1996). This is followed by a second, more prolonged component that is dependent on stimulation of the G_q-coupled metabotropic P_2Y1 receptor, resulting in the generation of inositol 1,4,5-trisphosphate (InsP₃) and subsequent release of Ca²⁺ from intracellular stores (Heemskerk et al. 1993; Offermanns et al. 1997; Jin et al. 1998; Hechler et al. 1998). Calcium store depletion in turn activates the process of store-mediated Ca²⁺ entry (Sargeant et al. 1992). We have recently established that the P_2Y1 pathway comprises the major part (about 90%) of the Ca²⁺ signal attributed to ADP receptor activation (Sage et al. 2000).

When monitored in individual platelets, moderate stimulation by agonists of G_q protein-coupled receptors usually causes an irregular Ca²⁺ response, consisting of multiple spikes in [Ca²⁺]_i (Heemskerk et al. 1992, 1997a; Hussain & Mahaut-Smith, 1999). Higher concentrations of a strong agonist such as thrombin evoke a more continuous [Ca²⁺]_i elevation, but the Ca²⁺ response evoked by ADP remains spiking in character even at high doses. This irregular spiking persists for at least 5 min and is abolished upon removal of the ADP, suggesting that in platelets P₂ purinergic receptor occupation leads to multiple Ca²⁺ release events of variable magnitude. This is in contrast to the situation in megakaryocytes, i.e. the platelet progenitor cells, in which P₂ purinergic receptor stimulation elicits a train of regular oscillations in [Ca²⁺]_i with fixed amplitude and frequency (Tertyshnikova & Fein, 1997, 1998). This suggests that during the separation of platelets from their megakaryocyte precursors some of the fine regulation of Ca²⁺ peak generation is lost, perhaps because of the decrease in cellular volume upon shedding the platelets (platelets have a volume of about 6 fl and are around 2 μm diameter).

In both electrically excitable and non-excitable cell types, global Ca²⁺ signals are assumed to result from the regenerative recruitment of Ca²⁺ from discrete subcellular Ca²⁺ release units. These units, representing the elementary building blocks of Ca²⁺ signalling, usually give rise to short-lived, highly localised signals (Bootman et al. 1997a). Such local events are indicated as Ca²⁺ sparks in cardiac (Lipp & Niggli, 1998), skeletal (Stern et al. 2000) and smooth (Jagger et al. 2000) muscle cells, and as Ca²⁺ puffs or blips in other (non-excitable) cells such as Xenopus oocytes (Parker et al. 1996), neuronal PC12 cells (Reber & Schindleholz, 1996) and HeLa cells (Bootman et al. 1997b). In the hierarchical scheme of Ca²⁺ release events, blips may represent the activity of individual InsP₃ receptor channels, whereas puffs depict the release from clusters of InsP₃ receptors, which are usually found at intervals of several micrometers apart. In both oocytes and HeLa cells, individual blips and puffs are assumed to act as pacemaker release events. They tend to integrate in a spatio-temporal way and give rise to global oscillations and waves when reaching the critical [Ca²⁺]_i threshold for whole-cell regenerative Ca²⁺ release (Allbritton & Meyer, 1993; Parker et al. 1996; Bootman et al. 1997a). The triggering of individual puffs is likely to be a random process, possibly due to the stochastic binding of InsP₃ to single molecules or clusters of receptors, whereas nearby puffs potentiate each other, facilitating the next release event by local positive feedback loops.

For the experiments described in this paper, we hypothesised that the irregular Ca²⁺ peaks detected in ADP-stimulated platelets resemble the Ca²⁺ puffs that are locally observed in larger cells. We designed a method to determine individual Ca²⁺ release events in the complex signals in platelets. The spatial and temporal heterogeneity of the ADP-induced Ca²⁺ signal was then evaluated in platelets in artificial buffer medium, and also in platelets in a physiological environment, i.e. autologous plasma or flowing whole blood.

METHODS

Materials

H-Phe-Pro-Arg chloromethyl ketone (PPACK) was obtained from Calbiochem (La Jolla, CA, USA). Bovine fibrinogen (type IV), apyrase (325 units ATPase and 110 units ADPase (mg protein)⁻¹) and ADP, sodium salt (98.2% purity with 1.4% AMP), were purchased from Sigma (St Louis, MO, USA). Glass coverslips were obtained from Menzel-Glaser, Braunschweig, Germany). Fluo-3 pentaacetoxy methyl ester and Fluo-3 free acid were purchased from Molecular Probes (Leiden, The Netherlands). Other materials were obtained from sources as described previously (Heemskerk et al. 1997a).

Platelet labelling and preparation of plasma and blood

Blood was drawn from adult, healthy volunteers, from whom informed consent was obtained and the protocol was approved by the local ethical committee. The blood was collected into 1/6 volume of acid citrate-glucose (80 mm trisodium citrate, 52 mm citric acid and 180 mm glucose). Platelet-rich plasma, supplemented with apyrase (0.2 units ADPase ml⁻¹), was collected after centrifugation at 190 g for 15 min. Platelets in plasma were loaded with the Ca²⁺ probe Fluo-3, because of its bright fluorescence signal and the long excitation wavelength of 485 nm, at which fluorescence from plasma proteins is low. This allowed Ca²⁺ measurements from platelets in plasma and blood. Samples of 1 ml platelet-rich plasma were incubated with Fluo-3 pentaacetoxy methyl ester (5 μm) in the presence of lysine acetylsalicylate (aspirin, 100 μm) at 37 °C during 45 min. After the addition of 1/25 volume of acid citrate-glucose, the platelets were centrifuged in an Eppendorf microfuge at 1700 g for 2 min. The pelleted platelets were suspended in Hepes buffer pH 6.6 (136 mm NaCl, 10 mm glucose, 5 mm Hepes, 2.7 mm KCl, 2 mm MgCl₂, apyrase at 0.2 units ADPase ml⁻¹, and 0.1% (w/v) bovine serum albumin). After another centrifugation step, the platelets were resuspended in Hepes buffer pH 7.45 (composition as above, pH adjusted with Tris) at a concentration of 5 × 10⁷ platelets ml⁻¹ (Felige et al. 1998). Platelets were counted with a thrombocounter (Coulter Electronics, Luton, UK), and used within 3 h of loading.

A number of experiments were carried out with platelets in plasma or whole blood anti-coagulated with PPACK. The blood was collected into 1/10 volume of a mixture of 153 mm NaCl and 40 μm PPACK. Aspirin (100 μm) and apyrase (0.8 units ADPase ml⁻¹) were added. Additional PPACK (10 μm) was added 2 h later. Immediately before measuring, Fluo-3-loaded platelets were added to the anticoagulated blood (autologous) so that 5% of the platelets were labelled. When required, plasma was prepared by centrifuging the PPACK-anticoagulated blood at 2800 g for 10 min.

Set-up of steady-state and flow experiments

Round glass coverslips were degreased with a mixture of ethanol and 2 m HCl (1:1, v/v), and rinsed with water and 153 mm NaCl. The coverslips were then exposed to a fibrinogen solution (10 mg ml⁻¹ in 153 mm NaCl), rinsed twice with 153 mm NaCl, and incubated with Hepes buffer pH 7.45 containing 2% (w/v) bovine serum albumin, to shield uncovered patches of glass. The fibrinogen-coated coverslips were mounted into an open 1 ml incubation chamber and placed on the stage of an inverted microscope (Nikon Diaphot 200, Tokyo, Japan). Suspensions of platelets were directly added to these coverslips (Heemskerk et al. 1993; Briedéet al. 1999).

For flow experiments, rectangular (50 mm × 24 mm) coverslips were coated with fibrinogen and mounted on the bottom of a parallel-plate flow chamber, modified after Sakariassen et al. (1988). The dimensions of the chamber, punched in a block of poly(methyl methacrylate), were 42.5 mm (length) × 5.0 mm (width) × 0.2 mm (depth). Including tapered ends at the inlet and outlet of the flow cell, the chamber volume was 60 μl; the coverslip surface in contact with the perfusion fluid was 2.1 cm². Flow chambers were mounted on the stage of an inverted microscope and connected, via siliconised plastic tubes and a mixing device, to two gas-tight syringes that were placed on pulse-free perfusion pumps (Harvard Apparatus, South Natick, MA, USA). The first syringe (5 ml plastic, Terumo, Leuven, Belgium) was filled with PPACK-anticoagulated blood containing Fluo-3-loaded platelets. The pump rate was set at 25 ml h⁻¹. The second syringe (2.5 ml glass, Hamilton, Bonaduz, Switzerland) was filled with 330 μm ADP and 2 mm CaCl₂ in Hepes buffer pH 7.45. The second pump was operating at 2.5 ml h⁻¹. The total perfusion rate of 27.5 ml h⁻¹ caused, given the dimensions of the perfusion chamber, a wall shear rate of 225 s⁻¹ (Billy et al. 1997). Surgical clamps were used to prevent flow in the tubes when required. Syringes and tubes were extensively rinsed with 153 mm NaCl before use.

Fluorescence video imaging and confocal laser scanning microscopy

Non-confocal microscopic images of Fluo-3-loaded platelets were collected with a Nikon × 40 oil objective (Ph4/DL, numerical aperture 1.3) upon epi-illumination of the sample with a 100 W xenon lamp, which was connected to the microscope through a liquid light guide. A 0.2 neutral density grey filter was inserted in the excitation light beam. Excitation and emission wavelengths were 485 and 530 nm, respectively (half-bandwidths 22 and 30 nm). The fluorescent light was separated with a 505 nm dichroic long-pass filter. Images were captured with a low-light level, charge-coupled device camera (Photonic Sciences, Robertsbridge, UK) working at standard video rate. Image recording was at a rate of 1-5 Hz using a Unix/Quanticell 700-driven computer (Visitech, Sunderland, UK), as described elsewhere (Heemskerk et al. 1997b). Optical resolution in the x-y plane was 1.0 pixel μm⁻¹. Fluorescence measurements were sometimes alternated with recordings of (infrared) transmission light. Phase-contrast transmission images were recorded with a second VPM 6132 monochrome high-resolution charge-coupled device camera (Vista, Norbain, UK) and a Unix/Quanticell 900-driven computer system.

High-resolution, confocal images of fluorescent platelets were recorded with a Nikon RCM 8000 real-time confocal laser scanning system (Tokyo, Japan), operating at 30 Hz. The light source was an argon laser working at an excitation power of 91 μW. The excitation wavelength was 488 nm, and emission light was measured in the range 500-550 nm. Light was collected with a Nikon × 60 oil objective (APO, numerical aperture 1.4). Using a small pinhole, confocality in the x-y plane was 0.2 μm, matching the optical resolution of 6.0 pixels μm⁻¹. Confocality along the z-axis was 0.5 μm. Because of the limited fluorescence of the platelets, image frames were × 4 averaged to give a final temporal resolution of 8 Hz.

Activation of platelets

Fluo-3-loaded platelets in suspension were allowed to adhere to a fibrinogen-coated coverslip for 10-20 min, after which unbound platelets were removed by flushing. The platelets having spread on the coverslip (3-4 μm in diameter) were incubated with 500 μl of either Hepes buffer pH 7.45 containing 2 mm CaCl₂ or PPACK-anticoagulated platelet-poor plasma. Fluorescence and phase-contrast images were collected from the platelets, while ADP (final concentration 20 μm) was added from a concentrated stock solution.

In flow experiments, PPACK-anticoagulated blood with 5% Fluo-3-loaded platelets was perfused through a flow chamber using a pulse-free pump. The blood was mixed with ADP (final concentration 20 μm) immediately before reaching the inlet of the chamber. Fluorescence images were recorded from the plane of the coverslip.

Calibration of fluorescence signal

Digitised images were analysed as changes in fluorescence per platelet (region). Background fluorescence was obtained from a region of pixels adjacent to the platelet. Background-subtracted, fluorescence signals of each cell were corrected for the bleaching of the Fluo-3 fluorescence signal. Under non-confocal conditions, bleaching was typically 10% after 400 illuminations of 45 ms duration. The corrected fluorescence values (F) for each cell were converted into pseudo-ratio values by normalising for the fluorescence under resting conditions (F₀), as described elsewhere (Parker et al. 1996; Bootman et al. 1997b). Using Fura-2 as an indicator, we have established that the resting [Ca²⁺]_i in platelets attached to fibrinogen is around 40 nm (Heemskerk et al. 1997a). Assuming that [Ca²⁺]_i is at rest at this level, the pseudo-ratio value F’=F/F₀ can be used for calibration purposes. Using such F‘values, [Ca²⁺]_i in the platelets was estimated from the equation [Ca²⁺]_i=K_dF‘/(F‘_max - F‘). The maximal pseudo-ratio fluorescence level, F‘_max, was determined for each optical condition with calibration solutions of free Fluo-3. A K_d value of 316 nm was used (room temperature).

Statistics

Where indicated, data were compared by Student's two-tailed t test for unpaired observations.

RESULTS

Composition of calcium signals in platelets in artificial buffered medium

We used a sensitive microscopic fluorescence imaging system to monitor the morphological and fluorescence changes in human Fluo-3-loaded platelets adhering to immobilised fibrinogen. A suspension of platelets was added to an incubation chamber containing a fibrinogen-coated glass coverslip. Phase-contrast images showed that platelets made contact with the surface and gradually spread over the fibrinogen layer. Platelet binding and spreading was completely inhibited by H-Arg- Gly-Asp-OH peptides, indicating that the adhesion was mediated by the integrin α_IIbβ₃ fibrinogen receptors. After about 15 min of adhesion, the platelets had spread over a surface area of 3-5 μm in diameter. Apyrase, which degrades autocrine-released ADP, was added to the spreading platelets to prevent ‘spontaneous’ increases in [Ca²⁺]_i, as described before (Heemskerk et al. 1997a,b).

Microscopic fluorescence images were continuously taken from these adherent platelets. The low intensity of fluorescence, due to of the small volume of the platelet cytosol (about 6 fl), limited the rate of image capturing to ≥ 5 Hz. Upon bathing the fibrinogen-bound platelets in Hepes buffer containing 2 mm CaCl₂, levels of Fluo-3 fluorescence (F) remained low in most of the cells, indicating low levels of [Ca²⁺]_i. Addition of 20 μm ADP caused a series of fluorescence peaks in most of the platelets. The F values were first normalised to pseudo-ratio values relative to the basal fluorescence (F₀), in order to reduce signal heterogeneity caused by different amounts of probe (variations in platelet size). In the F/F₀ plots, fluorescence peaks were either baseline separated or, more often, seemed to consist of multiple spikes (Fig. 1A). The peak composition of complex spiking patterns was determined by extensive analysis.

Figure 1. ADP-evoked calcium responses in single platelets in buffered medium.

Human Fluo-3-loaded platelets, immobilised on a fibrinogen-coated surface, were stimulated with 20 μm ADP in the presence of 2 mm CaCl₂. A, traces represent F/F₀ values (upper line) and rates of estimated [Ca²⁺] changes of a representative platelet, calculated with a fixed decay value D_Ca= d/dt (F/F₀) +k_decay(F/F₀) (lower line). B, determination of decay constant (k_decay) of a single Ca²⁺ peak. C and D, distribution of values of rise time and decay constant of individual Ca²⁺ peaks in traces for 10 platelets.

It is known that individual [Ca²⁺]_i peaks in single ADP-stimulated platelets originate from both the mobilisation of Ca²⁺ from internal stores and the influx of extracellular Ca²⁺ (Heemskerk et al. 1997a). As indicated in Fig. 1B, the rising phases of individual F/F₀ peaks appears to be relatively short. The median time to peak maximum was 0.8 s, while 89% of the peaks reached a maximum in 0.4-1.2 s (n = 64) (Fig. 1C). When examining baseline separated peaks, it was apparent that the rising phase was abruptly followed by a slower decay phase. This decay could be described empirically as a first-order process, i.e. be fitted with a mono-exponential decay function (Fig. 1B). Values of the decay constants (k_decay) determined for individual peaks varied from 0.25 to 0.75 s⁻¹, corresponding to a 50% fluorescence decay in 1.3 to 4.0 s (Fig. 1D).

The peak decay constant seemed to be a cell-specific property. For instance, for platelets in a single experiment (i.e. originating from one subject), the k_decay varied from 0.23 ± 0.02 to 0.63 ± 0.03 s⁻¹ (mean ±s.e.m., n = 3-5 peaks per cell). Although being an empirical value, this constant can be considered to reflect the rate of Ca²⁺-ATPase-mediated back-pumping of Ca²⁺ from the cytosol. Cell-to-cell variability may then be due to variations in either Ca²⁺-ATPase activity or cytosolic Ca²⁺ buffering.

This analysis suggested that individual Ca²⁺ release events in platelets have the characteristics of a rapid increase in F/F₀ followed by a cell-specific, mono-exponential decay phase. This implies that the k_decay values can be used to identify individual signal events within the complex pattern of fluorescence changes of ADP-stimulated platelets. Reasoning that all [Ca²⁺]_i peaks can be resolved into periods of rapid Ca²⁺ release followed by slower Ca²⁺ removal, the rate of [Ca²⁺] changes (as an indication of the ‘net’ Ca²⁺ flux) during individual release events can be estimated using the equation D_Ca= d/dt (F/F₀) +k_decay(F/F₀) (s⁻¹). Because the F/F₀ signal is relatively noisy in its derivative form, the flux traces needed to be smoothed by a moving average (5 point) filter. Accordingly, derivative curves of D_Ca were obtained which showed sharpened and easily distinguishable peaks (Fig. 1A, lower trace). In such derived curves, we tentatively considered the peaks with amplitudes larger than 1.5 times the noise level as significant events. This deconvolution procedure thus represents a simple, empirical way to separate individual Ca²⁺ peaks from complex fluorescence traces, and to identify putative single Ca²⁺ release events.

Long-term activation studies were needed to determine peak frequencies and amplitudes. To limit the bleaching of Fluo-3, these experiments were carried out at a lower imaging frequency of 1 Hz. Conventional calibration procedures were used to convert the pseudo-ratio F/F₀ values into nanomolar increases in [Ca²⁺]_i (see Methods). The reduced imaging frequency caused a systematic underestimation of peak levels in [Ca²⁺]_i of no more than 20-30%. Under these measurement conditions, the ADP-evoked Ca²⁺ response (in the presence of CaCl₂) was characterised by a spiking pattern that remained irregular over a period of at least 10 min (Fig. 2A). Calculation of the derivative D_Ca revealed that k_decay values ranged in individual platelets from 0.27 ± 0.03 to 0.59 ± 0.06 s⁻¹ (mean ±s.e.m., n = 5-6 peaks per cell). Time curves of D_Ca were then generated to determine individual Ca²⁺ peaks. As shown for the curve from a representative platelet (Fig. 2A), both the peak interval times (ranging from 4 to 40 s; Fig. 2B) and the peak amplitudes (ranging from 20 to 300 nm; Fig. 2C) were found to be quite variable.

Figure 2. Long-term recording of ADP-evoked calcium responses in single platelets.

A, changes in [Ca²⁺]_i of a representative cell after calibration from F/F₀ values (upper line) and derived [Ca²⁺] change curve (D_Ca, lower line). The [Ca²⁺]_i trace of this cell was used to make histograms of peak interval times (B) and peak amplitudes (C).

This variability was also apparent from histograms of the peak interval times and amplitudes that were constructed from the [Ca²⁺]_i patterns of 10 stimulated platelets. Peak intervals of 4-11 s occurred most frequently (67%), though about 15% of the interval times were longer than 20 s (Fig. 3A). Peaks of all amplitude sizes were observed, although lower amplitude peaks (< 30% of maximum) were relatively abundant (40%) (Fig. 3B). We questioned whether, like in other cell types, small peaks or long intervals might prime for larger peaks, i.e. whether the peak level is determined by the amplitude of the previous peak or by the previous peak-to-peak interval. This appeared not to be the case: plots of the previous peak interval versus the peak level failed to give a significant relation (Fig. 3C). At first glance this suggests that there is little relation between the peak interval times and amplitudes. However, when plotting the data from representative cells as peak number versus peak time, it was apparent that often relatively long peak intervals were alternated with periods of shorter intervals (note nodes in Fig. 3D). This is an indication that the Ca²⁺ signal is composed of clusters of short-interval peaks (see below).

Figure 3. Analysis of peak patterns of ADP-induced calcium responses in platelets in buffer medium.

Fluo-3-loaded platelets on fibrinogen-coated coverslips were stimulated with ADP, as indicated for Fig. 1. Amplitudes of individual peaks in nanomolar [Ca²⁺]_i were normalised as percentages of the maximal amplitude of that cell. A and B, cumulative histograms of peak interval times and peak magnitudes. C, plot of peak amplitude as function of the previous peak interval (R² of correlation = 0.06, n = 375).D, plot of peak number versus peak time (four platelets from one experiment). Summarised data are given from 10 platelets (except for panel D), representative of > 50 cells.

Puff-like character of calcium release events in platelets

In larger cells, such as HeLa cells, local Ca²⁺ puffs appear to co-operate spatially or temporally in the generation of global Ca²⁺ oscillations. This means that series of puffs in the same region and/or within a short time interval can synergise to a high-amplitude Ca²⁺ signal (Bootman et al. 1997a). We investigated whether, in platelets, small or subcellular Ca²⁺ signals might also prime for larger peaks. By high-speed confocal laser scanning microscopy, confocal images from Fluo-3-loaded platelets were recorded at a final rate of 8 Hz, which allowed the analysis of the regional heterogeneity of the Ca²⁺ signal. The x-y confocality of the scanning system approached the final pixel resolution (see Methods). Signal analysis from the spread platelets, stimulated as before with 20 μm ADP in buffer medium containing 2 mm CaCl₂, was performed for three non-overlapping regions (about 100 pixels) per cell. We detected synchronous patterns with high as well as low Ca²⁺ peaks in every platelet region (Fig. 4A), although closer inspection of the traces indicated that, in some of the platelets, individual peaks seemed to start slightly earlier in one of the subcellular regions (Fig. 4B). The derived D_Ca curves (using a fixed k_decay value per cell), gave similar flux patterns for each of the three subcellular regions. However, occasionally an additional flux peak was observed in one of the regions (see asterisk in Fig. 4C). Together, this points to a proportional extension of the Ca²⁺ signal through the various regions of a spread platelet, although some of the platelets may have specific sites where the Ca²⁺ mobilisation starts or extra release events can be generated. Assuming that the Ca²⁺ signal in platelets is a summation of elementary release events, it thus appears that these events are variable in size and propagate almost instantaneously through the whole spread cell.

Figure 4. Confocal analysis of ADP-induced calcium responses in platelets in buffer medium.

Fluo-3-loaded platelets on fibrinogen-coated coverslips were stimulated with ADP, as indicated for Fig. 1. Fluorescence changes from single platelets were observed by high-speed confocal laser scanning microscopy, and analysed for three non-overlapping regions of 0.8 × 1.0 μm (100-150 pixels per region). A, changes in [Ca²⁺]_i after calibration from F/F₀ values; traces are from the three regions of interest. B, enlarged part of the traces under A; arrow indicates possible priming event. C, derived [Ca²⁺] change curves (D_Ca) from the same three regions using a single k_decay value of 0.64 s⁻¹; asterisk indicates putative local release event. Data are representative of > 35 platelets.

Calcium release events in platelets in plasma or flowing whole blood

Blood plasma contains many proteins modulating the platelet activation process. It is thus of importance to determine how this physiological environment influences the platelet Ca²⁺ signal. The adhering Fluo-3-loaded platelets were therefore incubated with plasma (of autologous origin) before additions of ADP. The plasma was anticoagulated with the thrombin scavenger PPACK to preserve physiological divalent cation concentrations (e.g. about 2.4 mm free Ca²⁺). Under these conditions, 20 μm ADP induced a complex pattern of [Ca²⁺]_i spiking (Fig. 5A). Peak analysis using derived D_Ca curves again pointed to a large variation in peak interval times (Fig. 5B) and peak amplitudes (Fig. 5C). Cumulative histograms, using the derived tracings from 10 platelets, showed a surplus of smaller peaks with short interval times, though we again noted a significant fraction of peaks with long intervals (Fig. 5D and E). These longer intervals seemed often to separate groups of short-interval peaks (see Fig. 5A).

Figure 5. ADP-induced calcium responses in platelets in autologous plasma.

Fibrinogen-bound platelets in PPACK-anticoagulated plasma, containing 2.4 mm free Ca²⁺, were stimulated with 20 μm ADP. A, curve of changes in [Ca²⁺]_i in a representative platelet after calibration from F/F₀ values (upper line), and first-derivative Ca²⁺ curve (D_Ca, lower line). B and C, histograms of peak intervals and peak amplitudes of the same platelet. D and E, cumulative histograms of peak intervals and peak amplitudes from 10 platelets.

To further approach physiological conditions, Ca²⁺ signals were determined in Fluo-3-loaded platelets that were returned to autologous (PPACK-anticoagulated) whole blood. The blood was mixed with ADP, perfused at a relatively low shear rate of 225 s⁻¹ over a fibrinogen-coated coverslip, and fluorescence was collected from platelets assembling on the fibrinogen surface as single cells or small aggregates. Bright-field trans-illumination images indicated that platelets started to adhere to the fibrinogen coating once infusion of ADP was started. Video imaging of the Fluo-3 signal showed that the adhesion was usually following by the appearance of spiking changes in [Ca²⁺]_i. Again, the recorded traces from individual cells showed a large variation in [Ca²⁺]_i amplitudes and interval times, often with recognisable groups of Ca²⁺ peaks (Fig. 6).

Figure 6. ADP-induced calcium responses in platelets in flowing whole blood.

Fluo-3-loaded platelets in autologous PPACK-anticoagulated blood were perfused over a fibrinogen-coated surface in the presence of 20 μm ADP. Fluorescence was recorded from single, adhering platelets. A and B, traces of changes in [Ca²⁺]_i (upper lines) and derived Ca²⁺ curves (D_Ca, lower lines) of two representative cells.

Regulation of ragged calcium signal in platelets

The observations thus far suggest that, for various experimental conditions (washed platelets, platelets in plasma or platelets in flowing blood), the ADP-evoked Ca²⁺ signal consists of peaks appearing after shorter- or longer-term intervals. Inspection of the plots of peak number versus peak time (e.g.Figs 3C, D and 7) reinforces the impression that the short peak intervals tend to cluster. Frequency distribution patterns of peak intervals (Figs 2B, 3A and 5B) show that the intervals peak around 9 s and that beyond 15-18 s a long tail with nearly constant frequency extends up to > 34 s. This tail contains 15-25% of the peak intervals. Therefore we arbitrarily choose the fifth quintile as a cut-off point, where the lower first to fourth quintile peak intervals were considered as short intervals. Analysis of run lengths of subsequent short and long peak intervals revealed average series of 8.3 and 2.0 peaks, respectively (washed platelets). This is nearly twice the expected run length (5.0 and 1.2, respectively) for a random distribution of peak intervals. In addition, the plots of peak number versus peak time gave a significantly better fit (P < 0.03) when constructed for only the lower four quintiles of intervals. Subsequent cluster analysis was thus performed by omitting the upper quintile of interval times and analysing the remaining four quintiles.

Figure 7. Pattern analysis of peak appearance.

Spikes of [Ca²⁺]_i in ADP-stimulated platelets in buffer medium (A and B), in plasma (C and D) and in whole blood (E and F). Upper panels (A, C and E): graphs of peak number versus peak amplitude (bars) and peak interval time (○). Lines represent fitted curves for fixed interval times, omitting the upper quintile of intervals. Lower panels (B, D and F): residuals representing deviation from the fit. Data are derived from the traces presented in Figs 2A, 5A and 6A, respectively.

The peak groups (four lower quintiles) were separately analysed by constructing plots of peak number versus peak time and fitting for a fixed peak interval per cluster of peaks. Examples are shown of such fits for platelets in buffer medium (Fig. 7A and B), and platelets in plasma (Fig. 7C and D) and whole blood (Fig. 7E and F). The fit characteristics are summarised in Table 1. It is apparent that, with respect to mean peak intervals, there are no differences between the various experimental conditions (presence of plasma or whole blood), and that the deviation of the fit (residual peak time) of 2.3-2.8 s is still large in comparison to the slope of 8.1-9.4 s. Further analysis indicated that there was no correlation between the duration of previous and next intervals (R²= 0.003, 0.06 and 0.008 for washed platelets, platelets in plasma and platelets in whole blood, respectively, n = 181-330). In addition, there was no evidence for the shortening of peak intervals along a sequence. Together, this indicates that - for these shorter-term intervals - the variation of peak interval is sizeable with Ca²⁺ release events that are not recruited in a temporal way.

Table 1.

In contrast to this frequency analysis, analysis of the peak amplitudes provided evidence for co-ordination. In the presence of buffer medium, the first peaks in a cluster were 5.7 ± 2.2 times higher in amplitude than second peaks, while first peaks were 3.3 ± 0.6 times higher in magnitude than third peaks (mean ±s.e.m., n = 36 clusters, P < 0.001 in either case). For platelets in plasma, these factors were 1.5 ± 0.1 and 2.5 ± 0.4, respectively (P < 0.001, n = 29). This points to a desensitisation rather than sensitisation of the initial Ca²⁺ release events in a peak cluster.

DISCUSSION

In this paper, we evaluated the ADP-evoked spiking Ca²⁺ responses of Fluo-3-loaded human platelets under artificial and more physiological conditions. Analysis of the complex Ca²⁺ peak patterns was performed using simple analytical procedures. High-frequency measurements showed that individual Ca²⁺ peaks consist of a rapid increase in [Ca²⁺]_i immediately followed by a slower decay phase, most likely representing back-pumping of Ca²⁺ out of the cytosol. The decay could be fitted mono-exponentially by a single constant, k_decay, which was applicable to all Ca²⁺ peaks from the trace of a given platelet. Second, by using this decay constant, it was possible to construct derived traces, as indications of net Ca²⁺ mobilisation, D_Ca, from the spiking fluorescence traces. With this approach, individual peaks, and thus individual Ca²⁺ release events, could be recognised and quantified.

From the pattern analyses, we conclude that the ADP-evoked response of most platelets consists of (longer or shorter) ‘ruffles’ of Ca²⁺ peaks with a ragged appearance. It was estimated that these ruffles make up about 80% of the peak interval times, and are separated by intervals of > 15 s. Peaks in a ruffle are generated at narrow, though variable, intervals, and the peak clusters are separated by periods of rest. Within a cluster, the averaged peak interval time is 8 s with a mean deviation of 2.3-2.8 s, pointing to an absence of periodicity. Within clusters there is large variation in peak amplitudes, but with a tendency for second and third peaks to decrease in amplitude, pointing to desensitisation after a first Ca²⁺ release event. Such a pattern of clustered peaks was found not only for washed platelets, but also for platelets bathed in plasma or whole blood.

Previously, we have shown that ADP-evoked [Ca²⁺]_i spiking in platelets requires InsP₃ formation (Heemskerk et al. 1993). Under conditions as described in this paper (i.e. resting levels of cAMP), the Ca²⁺ signal with ADP was found to be largely due to intracellular Ca²⁺ mobilisation followed by store-regulated entry of extracellular Ca²⁺ (Heemskerk et al. 1997a; Sage et al. 2000). Together with the recent evidence that signalling through the purinergic P_2Y1 receptors is responsible for the G_q- and thus InsP₃-dependent Ca²⁺ signal in ADP-stimulated platelets (Hechler et al. 1998; Jin et al. 1998), it appears that InsP₃-mediated mobilisation of Ca²⁺ is the principal driving force for the generation of Ca²⁺ peaks. The derived D_Ca curves described in this paper thus basically visualise the repetitive gate opening of the platelet InsP₃ receptors in response to ADP, although amplified by the Ca²⁺ signal due to store-regulated entry.

In non-electrically excitable cells, it is thought that local concentrations of InsP₃ determine the phasic, elementary Ca²⁺ release through InsP₃ receptor channels and the subsequent effect of (local, slowly diffusing) Ca²⁺ on InsP₃ receptor functioning (Jafri & Keizer, 1994; Yao et al. 1995; Callamaras & Parker, 2000). In the hierarchical scheme of Ca²⁺ signalling, elementary puffs thus represent the often short Ca²⁺ transients that arise stochastically from local release sites by a concerted action of changing numbers of InsP₃ receptors (Parker et al. 1996; Bootman et al. 1997a). These puffs typically synergise in either frequency or amplitude mode to reach a threshold for recruitment of successive sites and, thus, for regular and global Ca²⁺ signals. The Ca²⁺ signals in small platelets, even when spread over a fibrinogen surface and with plasma being present or absent, clearly show a different organisation. On the one hand, the Ca²⁺ peaks in a platelet resemble the puffs observed in larger cells, because they extend through various parts of the platelet and arise at random intervals and amplitudes. Thus, in platelets, almost all ‘local’ events appear to extend to quantitatively similar ‘global’ events, suggesting no diffusion limitations of InsP₃ or Ca²⁺. On the other hand, the clustering of spikes into groups of random Ca²⁺ peaks with decreasing amplitudes points to a certain degree of signal organisation as seen in larger cells. We conclude that the spiking Ca²⁺ signal generated in platelets has the characteristics of a ‘poor’ oscillator with a rather unstable frequency that is reactivated from period to period. The occurrence of such clustered, puff-like events in platelets in whole blood suggests that these play a role in the platelet activation process in vivo.