Heterogeneity of Single-platelet Calcium Responses to Activation

Abstract

Cytosolic calcium oscillations play a central role in platelet activation. However, signaling heterogeneity between platelets of the same individual, or between individuals, is poorly characterized. We utilized total internal reflection fluorescence microscopy of calcium fluorophore-loaded, surface-attached human platelets to monitor single-platelet calcium responses to collagen, ADP, and thrombin. For all activation types in healthy adult donors, four types of platelet calcium dynamics (“activation groups) were distinguished: (I) isolated spikes; (II) oscillations with a period of 3–10 s; (III) clusters of spikes following each other with calcium levels never returning to baseline; and (IV) a sustained high calcium level. The activation Groups I and II were predominant in the immobilized platelets of healthy adults (46 ± 22% and 33 ± 10%, respectively), with 18 ± 13% of platelets in Group III. Stimulation with ADP shifted the activation pattern, with Group I fraction falling to 15 ± 9% and Group III fraction rising to 43 ± 13% instead. For stimulation with ADP plus thrombin or collagen, Group III was predominant (71 ± 11% for thrombin, 46 ± 18% for collagen). A combination of all three agonists mainly produced Level III (69 ± 15%) and Level IV (18 ± 12%) platelets. Confocal microscopy revealed a gradual increase in all activation markers (including integrin activation, granule secretion, and spreading type) when one goes from Group I to Group IV. In the triple-negative breast cancer patients samples before therapy, the responses were shifted toward Group I indicating that their platelets were refractory. These results indicate the importance of platelet heterogeneity analysis and suggest a novel technique to investigate minor populations of refractory or hyperactive platelets.

Keywords

Platelet activation - Calcium signaling - TIRF microscopy - Flow chambers - Breast cancer

Introduction

Cytosolic calcium ions are a critically important second messenger during platelet activation by classic physiological agonists such as collagen, ADP, thromboxane A2, and thrombin. The basic type of platelet calcium response is oscillatory or spiking.[1] Upon strong activation, sustained high calcium concentration can be observed, which is associated with necrosis and procoagulant platelet formation.[2] Depending on the degree of stimulation, platelets produce a hierarchy of responses to activation, including changes in their adhesive properties,[3] the expansion of pseudopodia and lamellopodia,[4] [5] the secretion of granules,[6] and the exposure of phosphatidylserine.[7] The degree of platelet activation phenotype is associated with different cytosolic calcium concentration responses.[8]

The main steps in calcium signaling upon platelet activation are considered to be well-established. The agonists bind to platelet receptors, causing the activation of either G protein-associated or tyrosine kinase-associated signaling pathways, which leads to the stimulation of phospholipase C (PLC) isoforms. PLCs catalyze the formation of inositol-1,4,5-triphosphate (IP₃), leading to calcium release from intracellular stores. An interplay between receptor channels to IP₃ (IP₃R) and inner membrane calcium ATPases (Sarcoplasmic/Endoplasmic Reticulum Calcium ATPase [SERCA]) causes oscillations of cytosolic calcium concentration. Due to the small volume of platelets and thus to the small number of molecules per cell in the signaling cascade,[9] these oscillations appear as stochastic spikes.[10] [11] Higher amounts of IP₃ cause an increase in the amplitude and number of calcium spikes, ultimately leading to a sustained high concentration of calcium in platelet cytosol.[11] [12]

Platelets are well known to vary in volume, plasma membrane surface area, and number of proteins per cell.[13] [14] This leads to heterogeneity in platelets’ responses, probably due to variable surface-to-volume ratios.[9] The heterogeneity of platelet calcium response may have significant clinical value, because a minor fraction of superactivated platelets may have major physiological roles, which could hardly be detected by measuring averaged calcium responses. Although there are presently several research and clinical laboratory methods for the study of platelet calcium signaling,[15] [16] [17] they do not focus on the heterogeneity of single-platelet calcium responses in healthy individuals and patients; therefore, the understanding of this variability is limited.

Fluorescence microscopy has been actively employed for the analysis of platelet calcium responses from the beginning of 1990s.[18] [19] Over the years, this technology provided important insights into the understanding of platelet signal transduction,[1] [20] and recent years witnessed numerous major developments and elegant applications of this method.[21] [22] [23] However, due to its low throughput and difficulty of quantitative signal analysis, this method has almost never been employed in a clinical setting.

Here, we propose a method for registering single-platelet calcium responses in calcium-sensitive dye-loaded surface-immobilized platelets in a parallel-plate flow chamber using total internal reflection fluorescence (TIRF) microscopy and an algorithm for classifying them. Using this method, we were able to characterize the heterogeneity of platelet calcium responses in healthy donors and to investigate the effects of pharmacological stimulation or inhibition of their platelets.

Materials and Methods

Materials

The following materials were obtained from the sources shown in parentheses: human thrombin (Hematologic Technologies, Essex Junction, Vermont, United States); nonfibrillar human collagen type I (IMTEK, Moscow, Russia); CalBryte-590 AM (AAT Bioquest, Sunnyvale, California, United States); VM64 antibodies against PECAM-1/CD31 (RRID:AB_782149) was a kind gift from Prof. A. V. Mazurov.[24] Annexin V-Alexa647 and antibodies against P-selectin (CD62p-Alexa647), integrin αIIβ3 activation marker (PAC1-FITC) were from Sony Biotechnology (San Jose, California, United States). All other reagents were from Sigma-Aldrich (San Diego, California, United States), unless otherwise indicated.

Buffer Solutions

Three buffer solutions were used in the experiments. Modified Tyrode’s buffer (137 mM NaCl, 2.7 mM KCl, 12 mM NaHCO₃, 0.36 mM NaH₂PO₄, 1 mM MgCl₂, 2 mM CaCl₂, 5 mM HEPES (pH 7.5), 0.36% bovine serum albumin [BSA], 1 g/l D-glucose, pH 7.35) allows manipulations with cells in flow chambers, and BRB-80 (80 mM PIPES/KOH, pH 6.9, 1 mM ethylene glycol tetraacetic acid, 1 mM MgCl₂) and 1% Pluronic F127 solution in BRB-80 create the conditions for antibody and collagen adhesion. Usage of the BRB-80 buffer was adopted from our fellow team, which studies microtubule dynamics.[25] Pluronic F127 was diluted in BRB-80 on the rocker plate overnight.

Blood Collection and Dye Loading

The study was approved by the Local Ethics Committee of the Loginov Moscow Clinical Scientific Center, Moscow Department of Healthcare (Protocol No. 3/2023 dated February 28, 2023). The study was conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all patients and healthy donors for participation in the study and for the publication of the results without disclosure of personal information. The group of healthy volunteers (24, age 22–50) and five patients with triple-negative breast cancer (Supplementary Table S1, available in online version only) were included in the study. Both patients and healthy donors did not take aspirin or other nonsteroidal anti-inflammatory drugs in the week before the study; all the patients had normal platelet counts. Whole blood was collected into 1.6 mL 525 ATU/mL hirudin-containing tubes (S-Monovette, SARSTEDT AG & Co. KG, Nümbrecht, Germany). The samples were stored in a water bath set to 37 °C before experimentation. A total of 300 μL of blood was loaded with 0.3 μL of 2 mM CalBryte-590 AM in the presence of 0.3 μL 100 U/mL of apyrase for 30 min in a water bath set to 37 °C and then used immediately. The conditions were optimized as described.[26]

Flow Chamber Assembly

Coverslips (24 × 50 mm glass, DeltaLab, Barcelona, Spain) were subjected to cleaning using a PlasmaFlo plasma cleaner (Harrick Plasma, Ithaca, New York, United States) and silanization with 0.05% dimethyldichlorosilane in trichloroethylene, as described.[25] The single-use parallel-plate flow chambers were assembled from a coverslip, a dual-side 0.2-mm-thick adhesive tape (3M #9088-200, Maplewood, Minnesota, United States) and a 25 × 75 × 1.2 mm glass slide (ApexLab, Moscow, Russia) was assembled ([Fig. 1A]). Silicone tubes with an outer diameter of 1.5 mm and an inner diameter of 0.5 mm (MedSil, Mytischi, Russia) were placed to produce flow and sealed with a hot melt adhesive. The anti-CD31 (VM64) coating was performed by placing 40 μg/mL of VM64 in BRB-80 buffer for 15 min in the flow chamber with silanized coverslip ([Fig. 1B]). The ligand (PECAM-1 or CD31) for this antibody is an inhibitory receptor previously shown to suppress ITAM-mediated signaling pathways,[27] yet not to affect platelet activation significantly.[28] Coating with human type I collagen (HC11, Imtek, Moscow, Russia) was performed by placing 1 mg/mL collagen diluted in 0.01 M acetic acid for 15 min in the flow chamber with nonsilanized glass ([Fig. 1C]). A detailed protocols for the glass coating and flow chamber assembly is available at protocols.io[29] and in Supporting Methods (available in online version only).

Fig. 1 Experimental design. (A) Graphical model of the microfluidic chamber used in experiments. (B) Platelet immobilization using VM64. (C) Platelet immobilization using collagen. (D) Differential interference contrast and TIRF microscopy images for several critical timepoints of a typical experiment. TIRF, total internal reflection fluorescence.

Imaging Platelet Calcium Responses

The flow chamber was mounted on a Nikon Eclipse Ti-E fluorescence microscope (Nikon, Japan), equipped with a CFI Apochromat TIRF 100XC Oil objective, Nikon LU-N4 laser excitation source, Nikon TI-TIRF-E TIRF module, and Andor DU-897 digital EMCCD camera. A syringe pump set at 33 μL/min was used to gently infuse the whole blood into the chamber. After the pumping was stopped, differential interference contrast imaging was used to control platelet attachment ([Fig. 1D]). The chamber was washed with Tyrode’s buffer (150 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 0.4 mM NaH₂PO₄, 0.4 mM Na₂CO₃, 5 mM HEPES, 5 mM glucose, 0.5% BSA, pH 7.35) to remove nonattached cells ([Fig. 1D]). A region with 10 to 40 platelets was selected. Their fluorescence was recorded on the first video using low-angle TIRF mode with an angle of 48° ([Fig. 1D]). The power of the 561 nm laser was set to 5% of its maximum, and frames were collected at 12.25 frames per second with an 80-ms exposure and no delay. After that, another region in the same flow chamber was selected (to avoid bleaching), the buffer inside was replaced by Tyrode’s buffer containing 10 μM ADP, and the recording procedure was repeated ([Fig. 1D]). Finally, the third video was recorded in yet another area with Tyrode’s buffer containing both 10 μM ADP and 5 nM thrombin applied ([Fig. 1D]). Background fluorescence was subtracted, platelets as regions of interest were selected using ImageJ software (National Institutes of Health, United States), and mean fluorescence intensities were recorded. Fractions of platelet population with a particular activation strength type were calculated using an algorithm described in the Results section and implemented in Python (Python Software Foundation, Beaverton, Oregon, United States). A detailed protocol for the sample preparation and imaging is available at protocols.io[30] and in Supporting Methods (available in online version only).

Confocal microscopy was used for the assessment of platelet activation markers using Nikon Eclipse Ti2 microscope with an AX confocal scanning module equipped with CFI Plan Apochromat Lambda D 100× Oil objective. FAST resonant was used to ensure no bleaching. Imaging was performed at 1 AU (488 nm) pinhole with Nyquist sampling. Sample preparation and cell adhesion was performed same as previously, then a Tyrode’s buffer with 1% PAC-1-FITC and 1% CD62p-Alexa647 antihuman antibodies were introduced into the chamber for 10 min. During this time, only CalBryte 590 fluorescence was recorded to assess platelet activation group; however, only one shot per 6–10 s was performed in order to collect enough statistics. Then, markers fluorescence was collected in Z-stack mode (step 0.25 μm). The resulting videos were processed as follows: each cell was outlined using ImageJ software. Within the 3D contour, the average value of the fluorescence intensity of FITC or Alexa647 was calculated, then the background correction was made. Using brightfield, the degree of each cell’s spreading on the coating was determined.

Statistical Analysis

Microscopy video analysis was performed using Fiji ImageJ.[31] Statistical analysis was performed using GraphPad Prism 8 software (GraphPad Software, La Jolla, California, United States) using Mann–Whitney U test; the significance level was set as 95%. For the spectral analysis of calcium oscillations, the Fast Fourier Transform (FFT) method was applied in accordance with classical algorithms.[32] Prior to the transformation, the signal was subjected to wavelet filtering to suppress noise components.[33] After filtering, the data were interpolated to increase the sampling density and thereby improve the frequency resolution of the spectral analysis. Before performing the FFT, the mean value was subtracted from the signal to prevent the appearance of a harmonic at zero frequency, which is associated with the constant component of the signal.

Results

Platelet Calcium Responses can be Classified into Four Types

Platelet responses to different types of stimulation (including no stimulation) exhibited significant heterogeneity of calcium dynamics patterns ([Fig. 2], Supplementary Fig. S1 [available in online version only], for collagen). To systematically quantitate them, we focused on the temporal properties of calcium spike sequences rather than absolute concentrations, because it is frequency of cytosolic calcium spikes rather than amplitudes that is believed to represent the degree of platelet stimulation.[34] [35] [36] Noteworthy, there was significant heterogeneity in the amplitude of dye fluorescence in single platelets from the same experiment, despite similar calcium dynamics (e.g., more than 1000 a.u. for platelets 6 and 7 vs. 50 a.u. for platelets 10 and 13 in [Fig. 2]; see also platelets 2 and 4 vs. 12). Even for platelets with steady calcium plateau (platelets 1, 3, 5 in [Fig. 2]) corresponding to a necrotic state with saturated signal because calcium concentration should have reached equilibrium with the outside buffer, the levels of plateau were different. This is probably caused by single-platelet differences in the dye loading and provides additional important reason for the analysis to focus on temporal characteristics rather than absolute signals, which would have required individual calibration for each platelet.

Fig. 2 Typical platelet calcium curves. A typical field of view with adherent platelets and representative fluorescence intensity curves for indicated platelets. Platelets were immobilized on VM64.

Conventional spectral analysis[37] did not allow reliable differentiation between activators or donors (Supplementary Fig. S2, available in online version only). For each cell, the Fourier spectrum time series was calculated; however, individual spectra were too noisy for analysis (Supplementary Fig. S2B–E, available in online version only). Averaging of the individual spectra allowed to obtain a characteristic frequency profile for the entire cell population under the influence of a specific agonist and avoid distortions associated with the individual variability of individual cells. With increasing platelet activation, the spectrum of calcium oscillations broadens due to an increase in amplitude at higher frequencies. This is clearly evident upon thrombin activation, where a hump with a central frequency close to 0.3 Hz could be distinguished (Supplementary Fig. S2A, available in online version only).

The cells that produced no more than five solitary spikes per 60 s were ranked as the least activated Group I ([Fig. 3] A, B, D). The weakly activated Group II was defined as cells that produced multiple spikes, but with the fraction of clusters occupying less than 50% of the observation time ([Fig. 3] C, E). The cells with the fraction of clusters beyond 50% but without sustained high calcium and exponential decay were ranked as a strongly activated Group III ([Fig. 3] F–H). The ultimate state of platelet activation is the procoagulant or necrotic state associated with a sustained high concentration of calcium. When the fluorescence level of a platelet stayed high for at least 30% of the observation time and then demonstrated exponential decay, it was ranked as superactivated, procoagulant Group IV ([Fig. 3] I, J). Other representative calcium profiles for different groups are given in Supplementary Fig. S3 (available in online version only).

Fig. 3 Types of cytosolic calcium responses in CalBryte 590-loaded single platelets from a healthy donor. (A–J) Representative cytosolic calcium responses, which could be segregated into four groups: (A, B, D) Group I, when the cell exhibits less than five random solitary spikes per 60 s of observation; (C, E) Group II, with multiple stochastic spikes that generally do not form clusters; (F–I) Group III, calcium spiking is so frequent that single spikes merge into clusters; (J) Group IV cells exhibit a sustained high concentration of calcium at some timepoint. The decrease in fluorescence intensity is caused by fluorescent dye bleaching and leaking. Blue areas highlight clusters of spikes. Red areas highlight sustained high calcium levels. (K–M) Distribution of platelet activation across four activation groups for platelets on a VM64 surface: (K) VM64 baseline activation Groups I–IV (n = 14); (L) stimulated with ADP (n = 14); (M) stimulated by thrombin, which was added on top of ADP (n = 15).

Distribution of Platelets between Response Types upon Activation Shifts to Higher Average Calcium Levels

[Fig. 3] K–M shows the distribution of platelet response for a typical healthy donor depending on the activation status. When no activators were applied to platelets adherent to neutral coating (VM64), Groups I and II were dominant ([Fig. 3K]) at 46 ± 22% and 33 ± 10%, respectively, followed by Group III at 18 ± 13%, and almost no Group IV. Upon addition of 10 μM ADP, fewer platelets remained in Group I, and more than half of platelets shifted into Group III ([Fig. 3L]). Further platelet stimulation with 10 μM ADP and 5 nM IIa made Group III completely dominant (71 ± 11%; [Fig. 3M]), with Group IV finally becoming significantly increased compared with the case with no stimulation. Interestingly, 10–20% of platelets still remained in Group II, indicating significant heterogeneity and the presence of slowly or low-responding platelets. These data suggest that the use of a four-group distribution adequately describes changes in platelet statuses upon activation, while retaining the ability to represent minor populations with different responses.

In order to identify the functional status of these groups, we performed confocal microscopy of these samples (Supplementary Fig. S4A, B, available in online version only). The mean fluorescence intensity of the calcium dye did not statistically significantly differ between Groups I and II (indicating that the above-described grouping algorithm has a better sensitivity), whereas Group III was an order of magnitude brighter, and Group IV similarly exceeded it (Supplementary Fig. S4C, available in online version only). Group I had a low relative PAC1 and P-selectin signals at about 0.1, whereas a combination of Groups II and III produced a signal several-fold greater, and Group IV similarly had greater expression than Groups II and III (Supplementary Fig. S4D, E, available in online version only). The spreading type of platelets gradually shifted from mostly pseudopodia for Group I up to mostly lamellipodia for Group IV. Only some platelets of the IV group started to bind Annexin V by the end of the 5-minute experiment (data not shown).

Sensitivity of Platelet Activation Profiles to Pharmacological or Pathological Intervention

To test the robustness and the sensitivity of the method, we used a wider range of activations by adding a collagen surface and performed control experiments ([Fig. 4], Supplementary Fig. S1, available in online version only). Platelets attached to nonfibrillar collagen type I coating was originally mostly in the strongly activated Group III at the level of 46 ± 18%, with the rest split between Groups I and II. Further addition of ADP to collagen-bound platelets did not produce a significant change, although Group I became smaller and Group III became slightly larger. The following stimulation, with 10 μM of ADP and 5 nM of IIa, decisively increased Group III up to 69 ± 15%, and Group IV finally became pronounced at 18 ± 12%, whereas Group I became negligibly small.

Fig. 4 Sensitivity of platelet activation profiles to pharmacological and pathological interventions. This figure shows the fraction of activated platelets under six experimental conditions: VBA, platelets immobilized on VM64-covered surface for 5 min; VAA—with the addition of 10 μM ADP; VAT, with the addition of 10 μM ADP and 5 nM IIa; CBA, platelets immobilized on collagen-covered surface for 5 min; CAA, with the addition of 10 μM ADP; CAT, with the addition of 10 μM ADP and 5 nM IIa. Distributions for healthy donors (m = 18) are given as black boxes (25–75% percentiles. Distributions for different activation groups are given separately: (A, E) Group I; (B, F) Group II; (C, G) Group III; (D, H) Group IV. (A–D) Pharmacological analysis. Tyrode’s buffer containing either 0.5 μM PGI₂ (blue), or 400 ng/mL Adrenalin (red) was used for washing away unattached cells (see Materials and Methods). (E–H) Patients with breast cancer (Supplementary Table S1, available in online version only) prior to the application of antitumor therapy.

In order to evaluate the sensitivity of the method, we either inhibited the platelets with PGI₂ (0.5 μM) or promoted their activation with adrenaline (400 ng/mL) prior to the experiments ([Fig. 4] A–D). Inhibition with PGI₂ dramatically increased Group I in all samples at the expense of Groups II and III ([Fig. 4] A–D). In contrast, the addition of adrenaline produced relatively mild effects on top of the stronger activators.

Next, we performed the proposed analysis of platelet response heterogeneity for five patients with triple-negative breast cancer prior to antitumor therapy (Supplementary Table S1 [available in online version only], [Fig. 4] E–H). Our data demonstrate that, for four out of five patients, platelets appear to be more refractory to activation, similar to responses observed on PGI₂ ([Fig. 4] A–D), residing in the first group ([Fig. 4E]). The patient BC-2, responding similar to healthy donors, was the only relatively young patient without any comorbidities (Supplementary Table S1, available in online version only). This observation aligns with previous studies on cancer-associated changes in platelet function, suggesting a baseline platelet activation in the bloodstream, leading to their lower counts and desensitization to further activation.[38] [39] [40]

Discussion

This study proposes a method for assessing whether platelets of an individual have abnormalities in platelet calcium dynamics at the single-cell level. The method employs flow chambers with two types of coating employed side-by-side, weakly inhibitory anti-CD31 antibody coating and strongly activating collagen coating. A distinctive mark of this approach is its focus on platelet heterogeneity, which allows the identification of minor subsets of platelets with unusual properties. Our preliminary evaluation of the cancer patients suggests that the method is sufficiently sensitive to quantitatively detect the refractory response in calcium signaling.

The patterns of calcium dynamics observed were consistent with those described earlier,[1] [18] but the overall distribution of response types for healthy donors is novel from both methodological and research perspective. In the absence of activators, platelets immobilized on an anti-CD31 Ab-coated coverslip produced spontaneous calcium spikes. Upon the addition of ADP, the frequency of the spikes increased significantly. In response to strong stimulation with thrombin and collagen, most platelets responded with a high calcium level, as was observed earlier in multiple studies.[1] [20] [41] [42] In agreement with prior reports, we observed that upon activation, the spikes merge in clusters, and in all types of activation, single spikes have an asymmetric shape with rapid growth and slow decline.[36]

While the general picture of platelet calcium signaling described here is in agreement with previous studies, the overall distribution of response types for healthy donors is novel from both methodological and research perspectives. For larger cells with a more regular calcium dynamics, there are several numerical approaches for quantitative analysis of calcium signaling, including Fourier spectroscopy, wavelet analysis, and others.[43] [44] [45] There are presently no such tools for platelets, and the heterogeneous stochastic nature of platelet calcium signaling makes their development a separate challenging task, which is beyond the scope of the present study. Here, we focused on the semiqualitative classification of platelet calcium response into four groups, which proved to be sensitive to the activation degree and inhibition stimuli. Confocal microscopy revealed a gradual increase in all activation markers (including integrin activation, granule secretion, and spreading type) when one goes from Group I to Group IV. This could be used for physiological interpretation of the findings: for example, one could speculate that weakly activated Group I and II platelets could be found in the thrombus shell.[3] On the other hand, this might be helpful for the clinical interpretation of these groups.

Quite interestingly, the range of calcium responses in platelets was continuous, with tremendous heterogeneity. There were isolated spikes (Group I) in the strongly triple-activated platelets, and there were occasional procoagulant platelets (Group IV) in the samples on VM64 coating with no stimulation, depending on the donor. However, a reliable overall shift between the groups, with an increase in activation, was observed. The results suggest that an individual can have platelet subpopulations with greatly different responses, which could have important physiological and pathological consequences. Additional research is needed to experimentally establish the contribution of specific calcium efflux proteins (such as MCU or SERCA) and signal transduction enzyme copy numbers (PLC, etc.) on this heterogeneity and their roles in shaping the specific features of the four described groups.

Platelet heterogeneity is a widely accepted phenomenon that is the subject of ongoing research.[46] [47] A number of previous studies investigated platelet composition in healthy donors and patients with flow cytometry and mass cytometry and actually revealed several possible platelet subtypes.[47] [48] [49] [50] [51] [52] [53] [54] [55] From earlier studies, we can relate some of the subtypes to the response: for example, procoagulant platelets (Group IV) were found to be more likely to form in platelets with fewer mitochondria.[9] Furthermore, both in vitro and in vivo studies suggested that different platelet subtypes can play different roles in thrombus formation,[56] [57] differently respond to drugs,[58] [59] and different calcium signaling in different parts of the thrombus was also observed.[60] However, a lot of additional research with patients and experimental in vitro thrombosis models is needed to relate the activation groups observed here with the various subtypes and parameters, such as platelet age known from other studies, and to elucidate their significance. Further methodological studies are needed to identify the effects of platelet age, donor heterogeneity and calcium fluorophore loading heterogeneity used on the outcome of the assay. The latter might be especially important for future versions of the method employing more in-depth techniques of spectral analysis, as the present version of the method is not greatly sensitive to the absolute values of the signal. Another major shortcoming of the present study is the limited analysis of the calcium response curves, which could be essentially improved using modern tools of chaos theory and nonlinear systems analysis. Further research into these issues could enable us to better understand the roles of platelets with different responses in thrombus formation and to use single-platelet calcium signaling for diagnostics and therapy personalization.

Contributors’ Statement

F.A.B. and A.N.S. designed research. F.A.B., I.E.Z., and S.V.G. performed experiments and analyzed data. L.V. performed spectral analysis. R.D. and L.G.Z. recruited the patients and analyzed the data. E.V.S. recruited resources and analyzed the data. F.A.B. and A.N.S. prepared figures. F.A.B. and M.A.P. wrote the manuscript with contributions and editing from all authors. A.N.S. planned research, recruited resources, performed experiments, analyzed the data, supervised the project, and wrote the paper.

Acknowledgment

The authors are grateful to Ms. Aleksandra Kolbasnikova and Mr. Ivan Yastrebov (Lomonosov MSU) for technical assistance during experiments.

Data Availability Statement

All data generated or analyzed during this study are included in this article. Raw microscopy data could be found in FigShare.com (https://figshare.com/projects/Single_platelet_calcium_test_Healthy_adults_and_children_/192752). Further enquiries can be directed to the corresponding author.

• References

• 1 Heemskerk JW, Willems GM, Rook MB, Sage SO. J Physiol 2001; 535 Pt [phone]
• 2 Obydennyy SI, Sveshnikova AN, Ataullakhanov FI, Panteleev MA. J Thromb Haemost 2016; 14 (09) 1867-1881
• 3 Stalker TJ, Traxler EA, Wu J. et al. Blood 2013; 121 (10) 1875-1885
• 4 Schurr Y, Sperr A, Volz J. et al. Blood 2019; 134 (25) 2318-2329
• 5 Bender M, Palankar R. Hamostaseologie 2021; 41 (01) 14-21
• 6 Golebiewska EM, Poole AW. Blood Rev 2015; 29 (03) 153-162
• 7 Podoplelova NA, Nechipurenko DY, Ignatova AA, Sveshnikova AN, Panteleev MA. Hamostaseologie 2021; 41 (02) 146-153
• 8 Heemskerk JWM, Mattheij NJA, Cosemans JMEM. J Thromb Haemost 2013; 11 (01) 2-16
• 9 Obydennyi SI, Artemenko EO, Sveshnikova AN. et al. Haematologica 2020; 105 (04) 1095-1106
• 10 Purvis JE, Chatterjee MS, Brass LF, Diamond SL. Blood 2008; 112 (10) 4069-4079
• 11 Sveshnikova AN, Ataullakhanov FI, Panteleev MA. Mol BioSyst 2015; 11 (04) 1052-1060
• 12 Jackson SP, Schoenwaelder SM. Blood 2010; 116 (12) 2011-2018
• 13 Baaten CCFMJ, Swieringa F, Misztal T. et al. Blood Adv 2018; 2 (18) 2320-2331
• 14 Leone G, Agostini A, DeCrescenzo A, Bizzi B. Platelet heterogeneity. Scand J Haematol 1979; 23 (03) 204-210
• 15 Aliotta A, Krüsi M, Bertaggia Calderara D. et al. Int J Mol Sci 2020; 21 (24) 9515
• 16 Hearn JI, Gardiner EE. Arterioscler Thromb Vasc Biol 2023; 43 (10) 1775-1783
• 17 Fernández DI, Provenzale I, Cheung HYF. et al. iScience 2021; 25 (01) 103718
• 18 Ozaki Y, Yatomi Y, Wakasugi S, Shirasawa Y, Saito H, Kume S. Biochem Biophys Res Commun 1992; 183 (02) 864-871
• 19 Dunster JL, Mazet F, Fry MJ, Gibbins JM, Tindall MJ. PLoS Comput Biol 2015; 11 (11) e1004589
• 20 Hussain JF, Mahaut-Smith MP. J Physiol 1999; 514 Pt [phone]
• 21 Starodubtseva ES, Karogodina TY, Moskalensky AE. PLoS One 2024; 19 (10) e0308679
• 22 Fernández DI, Kuijpers MJE, Heemskerk JWM. Platelets 2021; 32 (07) 863-871
• 23 Joubert C, Grichine A, Dolega M. et al. Platelets 2024; 35 (01) 2422437
• 24 Mazurov AV, Vinogradov DV, Kabaeva NV. et al. Thromb Haemost 1991; 66 (04) 494-499
• 25 Gudimchuk N, Vitre B, Kim Y. et al. Nat Cell Biol 2013; 15 (09) 1079-1088
• 26 Galkina S, Balabin F. Syst Biol Physiol Rep 2021; 1: 10-17
• 27 Crockett J, Newman DK, Newman PJ. J Thromb Haemost 2010; 8 (07) 1584-1593
• 28 Jones CI, Moraes LA, Gibbins JM. Platelets 2012; 23 (05) 331-335
• 29 Galkina SV, Balabin FA. Sveshnikova AN. A cover glass silanization and preparation of flow chambers. protocols.io 2025;
• 30 Galkina SV, Balabin FA, Sveshnikova AN. Calcium test: preparations, imaging. protocols.io 2025;
• 31 Schindelin J, Arganda-Carreras I, Frise E. et al. Nat Methods 2012; 9 (07) 676-682
• 32 Cooley JW, Tukey JW. Math Comput 1965; 19: 297-301
• 33 Mallat SG. IEEE Trans Pattern Anal Mach Intell 1989; 11: 674-693
• 34 Skupin A, Falcke M. Chaos 2009; 19 (03) 037111
• 35 Sveshnikova AN, Panteleev MA. Hamostaseologie 2025; 45 (01) 49-62
• 36 Balabin FA, Korobkina JDD, Galkina SV, Panteleev MA, Sveshnikova AN. Biomed Khim 2024; 70 (06) 394-402
• 37 Moein M, Grzyb K, Gonçalves Martins T. et al. Bioinformatics 2018; 34 (17) 3052-3054
• 38 Kawano T, Hisada Y, Grover SP. et al. Thromb Haemost 2023; 123 (05) 501-509
• 39 Riedl J, Kaider A, Marosi C. et al. Thromb Haemost 2017; 117 (01) 90-98
• 40 Boneu B, Bugat R, Boneu A, Eche N, Sie P, Combes P-F. Eur J Cancer Clin Oncol 1984; 20 (07) 899-903
• 41 Heemskerk JWM, Feijge MAH, Henneman L, Rosing J, Hemker HC. Eur J Biochem 1997; 249 (02) 547-555
• 42 Poole AW, Watson SP. Br J Pharmacol 1995; 115 (01) 101-106
• 43 Janicek R, Hotka M, Zahradníková Jr A, Zahradníková A, Zahradník I. PLoS One 2013; 8 (05) e64394
• 44 Uhlén P. Sci STKE 2004; 2004 (258) l15
• 45 Ruffinatti FA, Lovisolo D, Distasi C, Ariano P, Erriquez J, Ferraro M. J Neurosci Methods 2011; 199 (02) 310-320
• 46 Thomas S, Kelliher S, Krishnan A. Thromb Haemost 2024; 8 (02) 102356
• 47 Donati A, Gupta S, Reviakine I. Biointerphases 2016; 11 (02) 029811
• 48 Milovanovic M, Lysen J, Ramström S, Lindahl TL, Richter A, Järemo P. Thromb Res 2003; 111 (01/02) 75-80
• 49 Milovanovic M, Lotfi K, Lindahl T, Hallert C, Järemo P. Pathophysiol Haemost Thromb 2010; 37 (01) 35-42
• 50 Järemo P, Milovanovic M, Buller C, Nilsson S, Winblad B. Platelets 2012; 23 (02) 116-120
• 51 Wang W, Liu D, Wang L. et al. Front Immunol 2025; 16: 1538652
• 52 Hindle MS, Spurgeon BEJ, Cheah LT, Webb BA, Naseem KM. J Thromb Haemost 2021; 19 (07) 1800-1812
• 53 Hamad MA, Krauel K, Schanze N. et al. Front Cardiovasc Med 2022; 9: 823549
• 54 Blair TA, Frelinger III AL. Platelets 2020; 31 (05) 633-640
• 55 Blair TA, Michelson AD, Frelinger III AL. Sci Rep 2018; 8 (01) 10300
• 56 Armstrong PC, Allan HE, Kirkby NS. et al. Blood Adv 2022; 6 (23) 6028-6038
• 57 Thattaliyath B, Cykowski M, Jagadeeswaran P. Blood 2005; 106 (01) 118-124
• 58 Stratz C, Nührenberg T, Amann M. et al. Thromb Haemost 2016; 116 (05) 941-948
• 59 Radziwon-Balicka A, Lesyk G, Back V. et al. Cardiovasc Res 2017; 113 (14) 1719-1731
• 60 Munnix ICA, Kuijpers MJE, Auger J. et al. Arterioscler Thromb Vasc Biol 2007; 27 (11) 2484-2490

Publication History

Received: 28 June 2025

Accepted after revision: 15 December 2025

Article published online:
30 January 2026