Plasmatic and cell-based enhancement of in vitro induced coagulopathy by microparticles originated from platelets and endothelial cells

Background: Aggressive trauma management and other external factors lead to hypothermia, acidosis and hemodilution (dened as Lethal Triad, LT) contributing to coagulopathy after trauma (Trauma-induced coagulopathy, TIC) that worsens patients’ outcomes. Procoagulative microparticles (MP) are crucial players at the interface of cellular and plasmatic coagulation. However, their functions remain largely unexplored. This study aimed to characterize effects of MP subtypes and concentrations on functional coagulation under in vitro simulated conditions. Methods: Blood from eleven volunteers were collected to simulate in vitro conditions of haemodilution (HD) and LT, respectively. HD was induced by replacing a blood volume of 33% by crystalloids and for LT, samples were further processed by reducing the temperature to 32 °C and lowering the pH to 6.8. MP were obtained either from platelet concentrates (platelet-derived MP, PDMP) or from cell culture (ECV304 cells for endothelial-derived MP, EDMP) by targeted stimulation. After introducing MP to in vitro conditions, their concentration-dependent effects (1.000, 10.000 and 15.000 MP/µl blood) on coagulation compared to whole blood (WB) were characterized by ow cytometric platelet activation and by quantication of brin clot propagation and spontaneous clotting using Thrombodynamics ® technology. Results: MP originated from platelets and endothelial cells affected blood coagulation in a concentration-dependent manner. Particularly, high PDMP quantities signicantly induced platelet activation and brin clot growth and size in HD conditions. In LT conditions, the highest PDMP concentration enhanced platelet activation, clot growth and size. In contrast, EDMP supplementation did not affect platelet activation, but resulted in enhanced formation of spontaneous clots, irrespective of simulated condition. With increasing EDMP concentration, the time until the onset of spontaneous clotting decreased in both HD and LT conditions. Discussion: study PDMP EDMP clotting factors of the plasmatic coagulation resulting in an increased formation of spontaneous brin clots. Conclusion: The diverse effects of in vitro generated MP from different cellular origin indicate a divergent mechanism of action exhibiting distinct functions within the coagulation process. LPS: Lipopolysaccharide LT: Lethal triad, MP: Microparticles; PDMP: Platelet-derived microparticles, PPP: platelet-poor plasma; Phosphatidylserine, PSGL-1: gylcoprotein


Introduction
Despite continuous improvements in trauma management, traumatic injuries are still the leading cause of death and disability in adults under 40 years [1][2][3]. Particularly, uncontrolled bleeding contributes to more than 50% of all traumarelated deaths. The bleeding phenotype is signi cantly aggravated by a coagulopathy (referred to as trauma-induced coagulopathy, TIC) occurring within hours after injury [4,5]. Approximately, one of four trauma patients arrive at the emergency department with laboratory signs of a compromised coagulation resulting in a four-fold higher mortality [6][7][8].
Efforts to elucidate the underlying pathomechanism led to improved resuscitation strategies in trauma management over the last decade, but still remain unknown in decisive parts [9,10]. Looking at etiology, the described mechanisms are currently divided into either a trauma and/or traumatic shock-induced endogenous coagulopathy (including endotheliopathy) also described as acute traumatic coagulopathy (ATC) or iatrogenic coagulopathy (IC) [3,11,12]. In particular, IC is triggered by an aggressive trauma management leading to hypothermia, acidosis and hemodilution, which worsens the outcomes of severely injured patients signi cantly [12]. Due to the high impact on coagulation of these three external conditions, they are most recently referred as "lethal triad" (LT) although hemodilution was not part of the historical de nition. A progressive course of LT is associated with a deterioration of coagulation accompanied by disturbed clot formation and strength [13][14][15]. For primary haemostasis and particularly platelet function, it has been shown a signi cant decrease in platelet activation and aggregation under clinical conditions of ATC leading to deterioration of clot formation in severely injured trauma patients [16,17].
In contrast, small cell-derived subcellular vesicles de ned by size of 0.1 -0.9 µm (known as microparticles, MP) are produced and released in large quantities from different cell types functioning at the boundary of cellular-and plasmaticdriven coagulation [18][19][20][21]. MP therefore are most likely mediators of the cellular and plasmatic coagulation and interact with various systemic in ammatory pathways initiated after trauma and other acute-phase conditions (e.g. ARDS, [22] and multiple organ failure [21,23]). There is evidence that different MP species trace back their cellular origin depending on their membrane composition and show a characteristic distribution pattern after major trauma [24,25]. In relation to their potential to facilitate coagulation complexes and initiate the coagulation process via TF-/FVII-dependent and independent pathways, we aimed to understand and to differentiate the role of platelet-and endothelial-derived MP (PDMP, EDMP) on functional coagulation [26,27]. In the present study, increasing MP concentrations were used to investigate their impact on primary and plasmatic haemostasis under simulated standardized in vitro IC-conditions.

Methods
Targeted in vitro synthesis of microparticles Cell line and cultivation for EDMP production Human ECV304 cells (Sigma-Aldrich, Steinheim, Germany) were used to investigate endothelial MP generation properties.
For generating particles, ECV304 cells were incubated with 1 mM hydrogen peroxide (Sigma-Aldrich, Steinheim, Germany); platelets with 1.5 µg/ml bacterial lipopolysaccharide (LPS) (Sigma-Aldrich, Steinheim, Germany) -both for 22 hours at 37 °C maintaining established cultivation conditions. Subsequently, a rst centrifugation step was performed to remove cells (10 min at 1.000 x g, 4 °C) followed by a second step to sediment particles from transferred supernatant (45 min at 10.000 x g, 4 °C). After removing residual supernatant, pellets were resuspended in 1x phosphate buffered saline (PBS), pooled and stored at -80 °C until analysis.

Platelet concentrates and PDMP production
Human platelets were obtained from platelet apheresis concentrates of the Institute of Transfusion Medicine (ITM) of the Cologne-Merheim Medical Centre that were not being authorised for transfusion.
For in vitro synthesis of PDMP, 40 mls of platelet concentrates were stimulated with 1.5 µg/ml LPS for 22 h at 37 °C in a humidi ed incubator with 5 % CO 2 (ThermoFisher, Marietta, USA). Subsequently, the platelet suspension was centrifuged at 1.000xg for 10 min at 4 °C (Heraeus, Hanau, Germany). Cell-free supernatant was centrifuged twice at 10.000xg for 45 min at 4 °C (Heraeus, Hanau, Germany) for PDMP pellet sedimentation, which was resuspended in the following in PBS obtaining pure PDMP. PDMP pellets were pooled, aliquoted and stored at -80 °C until analysis.
Blood donation and processing for inducing coagulopathic conditions Study approval was given by the ethical committee of Witten/Herdecke University (#182/2016). Eleven healthy volunteers, ful lling the ITM criteria for blood donation (age ≥18 years; no preexisting history of coagulation disorders, anticoagulant and/or platelet-inhibiting medication or viral infection), consented to participate and donated 60 mls blood, which was collected in citrated monovettes (Sarstedt, Nümbrecht, Germany) Coagulopathic conditions such as haemodilution (HD) and lethal triad (LT) were simulated in vitro as previously described [28,29]. In brief, conditions were either induced by diluting whole blood (WB) by replacing by crystalloids (Sterofundin ® ISO Infusionslösung, B. Braun Melsungen AG, Melsungen, Germany) only (hereafter be referred to as haemodilution, HD) or in combination with lowering the pH value to 6.8 using 2 M HCl and decreasing the temperature from 37 to 32 °C (hereafter be referred to as lethal triad, LT). Constant temperature was realized by continuous storage of samples in an appropriate tempered water bath (Thermolab1070, GFL, Burgwedel, Germany).

Application of PDMP and EDMP to the experimental HD and LT approaches
After introducing the conditions of HD and LT, either no MP (untreated controls) or EDMP/PDMP were supplemented in distinct concentrations of 1000 (1k), 10.000 (10k) or 15.000 (15k) MP/µl ( Figure 1). Exact microparticle quantities required for the experimental approaches had been determined shortly before ow cytometric analysis by using the BD Accuri TM C6 (BD, Heidelberg, Germany). For this purpose, particles were de ned by size (0.5 to 0.9 µm) [21] and by typical surface marker originated from the parental cell such as CD42b for platelets and CD144 for endothelial cells, respectively (BD, Heidelberg, Germany). In addition, Annexin V dye was used as marker for the externalisation of phosphatidylserine (PS). After a microparticle incubation of 5 min, samples were processed for subsequent coagulation analysis.

Extended coagulation analysis
Detection of activated platelets For measuring platelet activation, cells were xed according to Cy x III protocol [30] and P-selectin expression was ow cytometric measured by using CD42b and CD62p antibodies (BD. Heidelberg, Germany) and the BD Accuri TM C6 device (BD, Heidelberg, Germany). Results are presented as relative changes of CD42b + /CD62p + stained platelets referring to the respective unstimulated approach of each experimental group (set as 100%).

Thrombodynamics ® (TD)
Differential centrifugation was applied for TD analysis (1.600 x g for 15 minutes followed by a spin of plasma for 5 minutes at 10.000 x g) and platelet-poor plasma (PPP) was immediately shock-frozen in liquid nitrogen and stored at -80 °C until analysis.
For TD analysis, PPP samples of WB and HD were thawed at 37°C and those of LT were thawed accordingly at 32°C in a water bath. The corresponding temperatures for WB, HD and LT were maintained during the TD measurements.
For the determination of spatial clot growth, the TDX kit consisting of reagent I (lyophilized protein of FXIIa inhibitor) and reagent II (CaCl 2 ) was used according to the manufacturer's recommendation (HemaCore, Moscow, Russia). Brie y, 120 µl of PPP were supplemented with reagent I and incubated in the device thermostat for 15 min (HemaCore, Moscow, Russia). Subsequently, the PPP samples were treated with reagent II and directly placed into the micro chamber. By inserting a with immobilized tissue factor coated insert, the reaction had been initiated and clot growth and spontaneous clot formation was recorded over 45 min. The following parameters were measured: lag-time (Tlag, min), rate of clot growth (V, µm/min), initial rate of clot growth (Vi, µm/min), clot density (D, a.u.) and clot size (Cs, µm). In addition to these parameters, Cs was measured in ve minutes intervals (0-20 min) to estimate the in uence of MP on the dynamics of clot growth. In order to statistically record the changed Cs over time, the individual areas under the curve (AUC) of each donor was calculated.

Statistical analysis
Statistics and graphical data analyses were performed using GraphPad Prism version 7.00 for Windows (GraphPad Software, La Jolla, USA). Non-parametric Friedman test and Dunn's post-hoc test to correct for multiple comparisons were applied to determine signi cant differences across the collected parameters in the groups of HD and LT.
In a rst step, the MP-untreated samples in the groups of WB, HD and LT were compared in pairs as follows: WB vs. HD, WB vs. LT and HD vs. LT. Then, the effects of dose-speci c application of PDMP or EDMP were tested in the HD and LT groups by comparing those two. Finally, untreated WB samples a non-parametric Wilcoxon test revealed potential effects between untreated and MP-supplemented samples (C vs. 1k). P-values with a signi cance level <0.05 were considered statistically signi cant.
Values with respect to platelet count and TD temporal course of clot growth (Cs) are presented as arithmetical mean and standard deviation.

Demographics of healthy volunteers
Overall, 11 healthy donors were enrolled for analysis. 54.5% of donors were males with a median age of 42 years.

Effect of experimental conditions and MP supplementation on platelet count and activation
While the condition of HD led to a marginally decreased platelet quanti cation than WB (WB: 2.7 x10 5 /µl vs. HD: 2.2 x10 5 / µl), the LT caused a signi cant reduction of cell numbers compared to WB samples (Figure 2 A).
The application of MP resulted in a varying platelet activation depending on the supplemented dose. Under HD conditions, high PDMP quantities (10k and 15k MP/µl blood) induced an activation of platelets, which was demonstrated by a signi cant increase of the cell surface marker CD42b + /CD62p + expression compared to controls (Figure 2 B). Particularly the administration of 15k PDMP/µl blood further increased platelet activity in this group. Similarly, the highest PDMP concentration showed also the greatest effect on platelet activation within the LT group compared to the untreated LT controls (Figure 2 B). In contrast to PDMP, EDMP supplementation did not affect platelet activation signi cantly either group. However, a trend of slight enhancement of activated platelets was observed after supplementation of 10k and 15k EDMP/µl blood (Figure 2 C). Irrespective of the supplemented MP subtype, low concentration (1k) of any MP had no effect on platelet activation (Figure 2 B, C).

Kinetics of brin clot formation and density after PDMP application
The kinetics of clot formation measured by Tlag remained unaffected in the experimental controls of WB, HD and LT as well as after supplementation of PDMP (Table 1). Similarly, no kinetic changes were observed for the parameters rate of clot growth (V) and initial rate of clot growth (Vi) in the control groups of WB and HD, but a signi cant enhancement was recorded in LT conditions compared to WB (Table 1). However, low quantities of PDMP (1k MP/µl blood) signi cantly improved V and Vi in WB; in hemodiluted blood the high PDMP concentrations (10k and 15k PDMP/µl blood) resulted in an enhancement of both growth parameters (Table 1). Under LT conditions, the application of 10k and 15k PDMP/µl blood resulted in an even stronger increase of brin growth rate (V) compared to untreated LT controls (Table 1). Furthermore, a tendential but not signi cant increase of initial clot growth could be observed after PDMP administration in LT samples (Table 1).
In contrast to WB, brin polymerization and the associated clot density (D) deteriorated under the traumatic in vitro HD and LT conditions (Table 1) while a signi cant reduction compared with untreated WB was detected for LT (Table 1). An improved clot density of brin polymerization was achieved by PDMP quantities of 1k MP/µl blood in WB and by high PDMP quantities (10k and 15k MP/µl blood) in HD samples (Table 1). A PDMP stimulating effect on clot density could not be detected for LT (Table 1). The following kinetic parameters were determined: Tlag: initial growth rate, V: average clot growth rate and Vi: initial clot growth rate.
Clot density was measured as dynamic parameter. Values are presented as median with the corresponding IQR. Statistical significances were marked with plus symbols for differences between unstimulated groups (WB vs. HD and WB vs. LT) or with asterisks for concentration-dependent differences within one experimental group (* ,+ p ≤0.05, ** ,++ p ≤0.01, *** ,+++ p ≤0.001), respectively. The abbreviation k signifies a thousand.
Due to the formation of spontaneous brin clots following EDMP administration in all experimental groups, no kinetic parameters describing clot kinetic and development (Tlag, V, Vi, and D) could be recorded.

Fibrin clot formation following PDMP supplementation -growth and size
The concentration dependent PDMP application resulted in a signi cantly increased brin clot growth (Cs) over the measured time of 20 min in the WB and HD groups (Figure 3). While the administration of low PDMP concentration 1k PDMP/µl blood caused an enhanced clot growth in WB (Figure 3 A), similar effects had been achieved in the HD group after supplementing high level of PDMP (10k and 15k PDMP/µl blood, Figure 3 B). The improved brin formation in WB and HD was also mirrored in the respective AUCs over a time course of 15 min (Figure 3 D, E).
Noteworthy, the condition of LT alone led to an increased brin growth 15 min after measurement initiation compared to WB controls (Fig. 3 D, E). However, the supply of PDMP had no further impact except for the addition of 15k PDMP/µl blood at the time point of 10 min (Figure 3 C).

Spontaneous clotting after EDMP supplementation
Data collection for the parameters de ning brin clot growth could not be collected in the EDMP groups due to the early formation of spontaneous clotting originating from EDMP that occurred irrespective of the experimental setting ( Figures 5,   6). While no spontaneous clot formation was observed in the control groups of WB and HD, it did occurre in LT control conditions ( Figure 6). The event of spontaneous clot formation only occurred following EDMP but not PDMP supplementation.

Discussion
The present study aimed to elucidate the role of PDMP and EDMP on functional coagulation under in vitro simulated traumatic conditions. In line with previous ndings, the applied experimental settings of HD and LT in this study were appropriate models for standardized in vitro simulations of coagulation disturbances that may occur after trauma [28,29].
Special emphasis was given to the analysis of selected MP on blood coagulation with a particular focus on PDMP and EDMP. Both MP subtypes had been reported to be released in high quantities in patients who sustained major traumatic injuries indicating that these particles might be involved in regulating blood coagulation after trauma [19,25,31,32]. With this study we demonstrated for the rst time a divergent mechanism of action on coagulation originating from MP with different parental cells. Supplemented PDMP mainly affected circulating platelets by inducing their activation, which was measured by enhanced P-selectin expression on platelet membranes (CD42b + /CD62p + ), particularly, but not exclusively, in HD conditions. It is conceivable that an increased platelet activation likely compensates for reduced platelet counts in HD and LT conditions potentially via the ligand-receptor interaction between the P-selectin and P-selectin glycoprotein ligand-1 (PSGL-1) forming a PSGL-1/P-selectin complex that leads to exponential cell activation [33][34][35][36]. On the one hand an increased cell activation is associated with a promoted phosphatidylserine (PS) exposure resulting in procoagulant platelet formation and with an increased PDMP secretion and release of coagulation-promoting components from the platelet storage granules on the other [37]. Furthermore, not only platelets expose PS on their membrane, but also in vitro generated PDMP as determined in this study by ow cytometric measurements. It is well known that PS on platelet surfaces cause an assembly of coagulation factors promoting functional clotting [38,39]. Related to our ndings Lipets et al. also revealed that in vitro generated PDMP contributing to an enhanced coagulation propagation by expressing PS on their cell membrane [40].
PDMP together with activated platelets could result in a surface expansion for clotting factors, which then would explain the locally increased formation of thrombin and brin polymerization under high PDMP treatment mainly in HD samples.
Unexpectedly in contrast to PDMP, in vitro generated EDMP affected the impaired coagulation of HD and LT unexpectedly not by activating platelets, but by inducing spontaneous clotting originating from the particles itself. Therefore, the expected EDMP-mediated platelet activation via membrane P-selectin seems to play a secondary role in coagulation stimulation, which is likely due to a lower P-selectin exposure present on in vitro generated EDMP, which seem to be insu cient to activate platelets in a way PDMP do.
Furthermore, the PDMP and EDMP supplementation only showed effects when particles were available in high concentrations (10k or 15k MP/µl blood) indicating a concentration-dependent mode of action. Regarding spontaneous clotting, the used EDMP quantity has been irrelevant for triggering this process as already low EDMP concentrations (1k MP/µl blood) led to spontaneous brin formation, irrespective of the experimental group. However, higher MP level (10k and 15k MP/µl blood) further accelerated this process in HD (<5 min) and LT (<3 min) conditions resulting in unrestricted clotting. An underlying mechanism by which EDMP cause spontaneous clotting could be based on the presence of tissue factor (TF) on EDMP, which has been shown to be released after targeted stimulation from human endothelial cells in vitro [41][42][43]. TF in turn mediates a coagulation initiation via the coagulation factor VII/TF-dependent extrinsic pathway and likely results in spontaneous brin formation [44,45]. The earlier formation of spontaneous brin clots following EDMP administration in LT in comparison to HD potentially resulted from the combination of both, the TF-bearing EDMP and the acidosis-related increased basal platelet activation under LT conditions.
As recently published by our group, there is a positive correlation between injury severity and MP concentration after trauma likely reaching high quantities inducing procoagulative effects [25]. Our in vitro experimental setup indicates that under LT conditions MP promote procoagulative function by different pathways based on their cellular origin. Therefore, at rst sight, our data seems to be contrary to recent clinical observations of platelet dysfunction in TIC-related patients showing reduced platelet activity levels after trauma [16,17]. These clinical data re ect the nal common path of the multicausal pathophysiology of TIC resulting in the hypocoagulative state after trauma. The results from our experimental model should only be carefully extrapolated to a clinical picture but describes the procoagulative potential. It remains to be conclusively clari ed and requires further investigation whether and how MP are involved in the pathophysiology of TIC after major trauma in vivo and whether those high quantities as used in this study can, in fact, be physiologically achieved in a clinical setting.

Conclusion
The present study revealed a divergent mechanism of action originating from PDMP and EDMP indicating that MP subtypes have diverse function in stimulating coagulation under the investigated traumatic conditions of HD and LT.
Particularly, high MP concentrations of both MP types showed effects on the coagulation process. Likely due to the platelet-mediated activation (increased P-selectin expression) and the presence of PS lipids on PDMP membranes, clotting factor assembly and thus clot formation could be promoted on PDMP surfaces. In association with high EDMP quantities, an improved coagulation with formation of spontaneous brin clots was determined, which could presumably be induced via the TF/FVIIa-dependent extrinsic pathway of coagulation.

Limitations
It would have been desirable to perform the analysis with isolated MP subtypes from real trauma patients. But since high MP quantities would have been required for this study and the extraction capacity from patient's blood is limited, we relied on in vitro synthesized MP. There is a limitation in interpreting the results as in vitro LPS-stimulated MP might have a higher procoagulant activity compared to in vivo MP [46]. However, during the validation process of our assays, we compared the activity of in vitro generated MP to that of in vivo MP and did not see any difference.
Additionally, we are aware of the fact that the current in vitro study re ects the pathophysiological mechanism present in trauma patients only in part. For this reason, ndings should cautiously be transferred to a clinical setting.

Consent for publication
All volunteers (blood donors) gave their written and informed consent for participating in this study.
Availability of data and materials All data that are relevant for the study are included in this published article. Further datasets analyzed during the current study are available from the corresponding author on reasonable request.
Authors' contributions MC, NS and MM designed the study. JB, NS collected, analyzed and interpreted the data, wrote the manuscript, which has been critically reviewed by MM, MC, BS and UB. All authors read and approved the nal manuscript.