Red Blood Cells Actively Contribute to Blood Coagulation and Thrombus Formation

The chapter describes the likely molecular mechanisms leading to the aggregation of human red blood cells (RBCs) under conditions of physiological coagulation when prostaglandin E 2 (PGE 2 ) or lysophosphatidic acid (LPA) is released from activated platelets and under pathophysiological conditions, in particular thrombi formation in sickle cell disease when patients are in a vaso-occlusive crisis. In both scenarios cation channels are activated. This leads to an increase of the free intracellular Ca 2+ concentration resulting in an activation of the lipid scramblase, which in turn mediates a movement of phosphatidylserine (PS) from the inner to the outer membrane leaflet. In addition, the increased Ca 2+ concentration leads to the activation of the Gardos channel. Experiments suggesting this mechanism have been performed with fluorescence microscopy, flow cytometry as well as single-cell force spectroscopy. The Ca 2+ -triggered RBC aggregation force has been identified to be close to 100 pN, a value large enough to play a significant role during thrombus formation or in pathological situations.


Introduction
It is well known that phospholipids are asymmetrically distributed in the cell membrane of most, if not all, living cells. Sphingomyelin (SM) and phosphatidylcholine (PC) are located predominantly in the outer leaflet of the membrane bilayer, while phosphatidylserine (PS) and phosphatidylethanolamine (PE) are located mostly in the inner leaflet [1] as depicted in Figure 1A. Figure 1B shows the possible movements of phospholipids from one leaflet of the membrane to the other one (for explanation see text below).
The distribution of the membrane phospholipids is realised by three main proteins: flippase [4], floppase and scramblase [2, 5, 6]. The flippase (also named aminophospholipid translocase (APLT)) transports relatively quickly (in some minutes) PS, and a bit slower PE, from the outer to the inner membrane leaflet. The floppase transports PC and SM in the opposite direction, i.e. from the inner to the outer membrane leaflet [2]. The flippase and floppase are ATP-dependent and transport the phospholipids against their respective concentration gradients. The structure of the flippase has been published recently [7]. The scramblase is ATP-independent and activated by an increasing intracellular Ca 2+ concentration in human red blood cells (RBCs) [8]. The scramblase has been identified recently as a member of the TMEM16 protein family, and the crystal structure was published [9]. The activity of the three proteins is shown in Figure 1B. PS exposure on the outer leaflet of the cell membrane has been described as a marker for apoptosis in nucleated cells [5]. Although the programmed cell death of RBCs is still under discussion, these cells show some signs of apoptosis such as PS exposure, membrane blebbing and vesicle formation [10]. This process was denominated eryptosis by Lang et al. [11]. However, the use of this term is very controversial and finally not recommended [12]. Unidirectional phospholipid transport by flippase is directed inwards, whereas floppase promotes outward directed transport. Both transporters are ATP-dependent and frequently move phospholipids against their respective concentration gradients. For example, aminophospholipid translocase ( flippase) rapidly shuttles PS and PE from outer to inner leaflet, while ABCC1 ( floppase) moves both choline phospholipids and aminophospholipids more slowly towards the outer leaflet. The concerted action of both transporters is thought to create a dynamic asymmetric steady state, in which the outer monolayer is rich in choline phospholipids, whereas aminophospholipids predominantly occupy the inner leaflet. Bidirectional phospholipid transport is catalysed by a scramblase, the activation of which may occur following Ca 2+ increase. Since scramblase activity moves all major phospholipid classes back and forth between the two leaflets, it promotes the collapse of membrane phospholipid asymmetry with appearance of PS at the cells ' [18,19], including the participation of the mechanosensitive channel Piezo1 [20,21].
Although it is known that LPA induces PS exposure on the outer membrane leaflet in RBCs, there are conflicting reports about the mechanism. While Chung et al. [22] claim that it is a totally Ca 2+ -independent process, we showed that Ca 2+ alone is sufficient to induce PS exposure in human RBCs [23]. Woon et al. found that an increase of the intracellular Ca 2+ level in RBCs results in the exposure of PS to the outer membrane leaflet due to the activation of the scramblase and inhibition of the flippase [24]. Protein kinase Cα (PKCα) has been also described to be involved in the PS exposure on RBCs [22,25,26]. A further discussion follows below.
An interesting model for lipid studies we consider below is sheep RBCs since it is known that these cells have a completely different phospholipid distribution in the membrane. Like in human RBCs, PS and PE are distributed in the inner membrane leaflet. In contrast, the sheep RBC (like all bovine RBC) membrane does not contain any PC, and the outer layer consists exclusively of SM [27,28].

Stimulation of intracellular Ca 2+ increase in human RBCs
PGE 2 and LPA are local mediators released by platelets after their activation within the coagulation cascade. PGE 2 can be even released by the RBCs themselves under conditions of mechanical stress [29]. We were able to show that the addition of both mediators to suspensions of human RBCs leads to an increase of the intracellular free Ca 2+ concentration in these cells. In the case of PGE 2 (0.1 nM), 45% of the RBCs responded with increased Ca 2+ content; in the case of LPA (5 μM), nearly all cells reacted [17,30] but still showed a strong heterogeneity in the time course and the intensity of the response [31].
The identity of the Ca 2+ entry was proposed to be channel-mediated; on the one hand, the Ca V 2.1-channel [32] and on the other hand the non-selective voltage dependent cation channel [17] were suggested. While the activation of Ca V 2.1 was linked to protein kinase C (PKC) activity [33], we could show that Ca V 2.1 and PKC belong to different signalling branches [18].
However, in a series of papers utilising fluorescence microscopy and flow cytometry, we could link PKCα activity and Ca 2+ entry [30,34].
The above described effect is probably mediated by a specific pathway rather than a non-specific leak since the Ca 2+ entry shows a clear dose-response relationship towards the LPA concentration [35]. For a recent review about Ca 2+ channels in human RBCs, see [36].
Another known port for Ca 2+ entry that is likely to have some pathophysiological relevance is the NMDA-receptor. Among the RBCs, NMDA-receptors were first identified in rat RBCs [37] but later also in human precursor cells as well as in circulating RBCs [38]. NMDA-receptors are cation channels permeable to Ca 2+ and can be activated by homocysteine and homocysteic acid [39].
To extend on this, Figure 2 provides an overview of the current knowledge about Ca 2+ entry and the major cellular consequences in different pathophysiological conditions.

Increased intracellular Ca 2+ content of human RBCs results in an activation of the scramblase
The PS exposure on the outer membrane leaflet of RBCs was studied for LPAtreated cells using fluorescence microscopy. It has been shown that the number of annexin V-positive cells (i.e. cells showing PS exposure) after LPA treatment in the presence of Ca 2+ is about 35% [30].
Experiments have also been carried out after treating the RBCs with LPA or the PKC-activator PMA but in the absence of extracellular Ca 2+ (presence of 1 mM EGTA). In the case of LPA, we did not find significant differences between treatment and control conditions (without activation) regarding PS exposure on human RBCs. Only in the case of PMA treatment, about 50% of RBCs showed PS exposure in the absence of Ca 2+ (compared to about 80% in the presence of Ca 2+ ). The data obtained for control conditions in the presence of Ca 2+ were not significantly different from the data of the control conditions in the absence of Ca 2+ [30].
It is evident that for stimulation with the phorbol ester PMA, there is no correlation between the number of cells showing an elevated Ca 2+ content and the induced PS exposure, because it is known that PMA activates PKCα also in the absence of Ca 2+ .
Furthermore we like to state that the threshold for the forward scatter in the flow cytometry measurements was set in a way to make sure that only the events based on the size of intact cells were counted. This was of importance since we realised the formation of micro-vesicles after treatment with LPA as well as with PMA, which is in agreement with findings reported before [41,42]. Calpain, a Ca 2+ -dependent  proteolytic enzyme, cleaves spectrin and actin, leading to cytoskeleton breakdown [43][44][45] and therefore could be involved in the vesiculation process.
The findings that Ca 2+ content and PS exposure do not correlate in all cells were further supported by confocal fluorescence microscopy, where double-labelling experiments (for intracellular Ca 2+ and PS in the outer membrane leaflet) have been carried out. The results of the LPA treatment showed that most of the cells have an elevated Ca 2+ content and depict a PS exposure. However, some cells are displaying an increased Ca 2+ level, but no PS exposure, whereas other cells externalise PS without having a higher Ca 2+ content.
In the case of PMA treatment, PS exposure without high Ca 2+ content is more pronounced, which is not surprising since PMA is an artificial compound, which activates conventional PKCs without the otherwise necessary presence of Ca 2+ .
The percentage of RBCs showing PS exposure after activation with LPA or PMA is significantly reduced after inhibition of the scramblase using the specific inhibitor R5421 [34]. It is also significantly reduced after the inhibition of the PKCα using chelerythrine chloride or calphostin C. The inhibitory effect is more pronounced when the scramblase and the PKCα are inhibited simultaneously [34].
Based on our experiments, it seems reasonable to assume that at least three different mechanisms could be responsible for the PS exposure on RBCs: • The first mechanism is related to the Ca 2+ -stimulated scramblase activation (and flippase inhibition) [4,8,10,11,47]. This effect involves the activation of the ω-agatoxin-TK-sensitive, Ca V 2.1-like (P/Q-type) Ca 2+ channel [33].
• The second mechanism involves the PKCα, which can be directly activated by PMA. It has been reported that PKCα is involved in the PS exposure in human RBCs [25,26]. In addition, it has been shown that LPA activates in human RBCs both the PKCα (Ca 2+ -dependent) and the PKCζ (Ca 2+ -independent) [22]. Whether there is a direct activation of the scramblase in human RBCs by PKCα and/or PKCζ remains to be proofed. The more pronounced effect of PS exposure observed after treatment of RBCs with PMA as compared to LPA can be explained by assuming that PMA activates all available PKCα, whereas LPA stimulation triggers a signalling cascade [46] resulting only in partial activation of the PKCα pool.
• The third mechanism is the enhanced lipid flop caused by LPA [30]. This mechanism is the only one in sheep RBCs suggesting the absence of scramblase activity in these cells [30].
To consider a possible effect of cell volume alteration on PS exposure, experiments have been carried out where the physiological solution was replaced by a solution containing 150 mM KCl plus 2.5 mM NaCl instead of 145 mM NaCl plus 7.5 mM KCl. Under these conditions of a high extracellular KCl concentration, an opening of the Ca 2+ -dependent K + channel (Gardos channel) does not result in a KCl loss and accompanying cell shrinkage. Based on this strategy, it has been found that a substantial part of the PS exposure in the presence of LPA is caused by the volume decrease.
It remains to be elucidated why, in the experiments reported above, some cells are displaying an increased Ca 2+ level, but no PS exposure, whereas other cells externalise PS without having a higher Ca 2+ content. In that context, we carried out single-labelling experiments (fluo-4 for intracellular Ca 2+ and annexin V-FITC for exposed PS) and focussed on the shape of RBCs using also a fluorescence microscope. The obtained results are presented in Figure 3.
In Figure 3Aa one can see different cell shapes in the transmitted light image for LPA stimulation. However, the corresponding fluorescence image (Figure 3Ab) does not show differences of the increased Ca 2+ content comparing RBCs with different shapes. The situation changes when analysing the PS exposure of the RBCs with different shapes ( Figure 3B). Again, RBCs with different shapes can be seen (Figure 3Ba), but in the case of PS exposure, the cell shape seems to play a significant role. Some echinocytes can be seen in the transmitted light images (red circles in Figure 3Ba). Almost all of them lack PS exposure (corresponding fluorescence image in Figure 3Bb). Only one echinocyte showing PS exposure can be observed. The explanation of such an effect could be based on the bilayer couple hypothesis. If an echinocyte is formed, the outer membrane leaflet is already extended compared with the inner membrane leaflet. It means that any process of externalisation of a membrane lipid located mainly in the inner membrane leaflet is hindered. In addition we carried out experiments where we transferred the RBCs to different solutions as shown in Figure 3C and D. The RBC shapes, which predominantly occur in these solutions, are also indicated in Figure 3. Again, the PS exposure is lowest in solution where one can expect mostly echinocytes although the Ca 2+ content is increased (red curves in Figure 3). The PS exposure of RBCs can be small also in solution where cells predominantly occur with the shape of discocytes or stomatocytes, however, in this case only in solutions where the increase of the intracellular Ca 2+ content is prevented, e.g. physiological solution of low pH (5.6, see violet curves in Figure 3).
Furthermore, we carried out double-labelling experiments (X-Rhod-1 for intracellular Ca 2+ and annexin V-FITC for exposed PS) and investigated the RBCs using a fluorescence microscope. The obtained results are presented in Figure 4. One RBC (indicated with white arrow in all panels of Figure 4) shows an increased intracellular Ca 2+ content, but the cell suddenly disappeared from the transmitted light image   (Figure 4Ab). However, PS exposure of this cell still can be seen (Figure 4Bb). It can be assumed that this RBC lysed, i.e. we see the remaining membrane structure (a ghost) and the Ca 2+ -sensitive fluorescent dye diffused out of the cell.
A different behaviour of RBCs depending on the age of the cell has been controversially discussed. Therefore, we performed additional short-time incubation experiments, comparable to the experiments carried out by Nguyen et al. [30], with RBCs separated in five fractions with different cell age according to the method of Lutz et al. [47]. The intracellular Ca 2+ content and PS exposure at the outer membrane leaflet of human RBCs with different age have been investigated using, e.g. LPA stimulation. Here we present a reanalysis of these already published data [48]. Interestingly, the percentage of RBCs showing increased Ca 2+ content and PS

RBCs were stimulated with lysophosphatidic acid, and panels (A) depict images 1 min after the start of the stimulation and panels (B) 1.5 min after the beginning of the stimulation. Panel (A) shows wide-field images and panel (B) confocal sections of the fluorescent biomarkers X-Rhod-1 (a Ca 2+ indicator) and FITC-labelled annexin V ( for PS detection). Panel (C) is an overlay of panels (A) and (B). The white arrow points to a cell that lysed after capturing images (a), i.e. in images (b) it's a ghost.
exposure depict a linear behaviour in dependence of cell age with a very good regression, R 2 of 0.94 and 0.92, respectively, as outlined in Figure 5A. However, the slope of this linear regression failed to be significantly different from zero.
After stimulation of the RBCs with LPA, the situation is even more complex. Figure 5B depicts the situation after 15 min of LPA stimulation. While the Ca 2+ concentration seems to relate inversely proportional to RBC age, PS-positive cells show a rather quadratic dependence on cell age. This is in contradiction to earlier investigations we performed on mouse RBCs [31]. Although we cannot completely resolve the situation part of the explanation might be caused in the detection technique: While microscopy is a rather gentle approach, in flow cytometry the cells under investigation experience high pressure and significant shear forces [60]. Therefore a significant number of high Ca 2+ cells that are more fragile may lyse in the flow cytometer and hence are excluded from the detection.
Furthermore it is worthwhile to mention that under in vivo conditions, cells with a permanent high Ca 2+ content and/or PS exposure are removed from the circulation, mostly in the spleen, while after long in vitro incubation time (48 h), old RBCs responded with higher increase of intracellular Ca 2+ content as well as higher PS exposure compared to young RBCs [48,49].
The intracellular Ca 2+ content and the PS exposure at the outer membrane leaflet have been investigated for human RBCs also in physiological solutions prepared with oxygen-enriched water (in comparison to normal physiological solution). This was a study for a company (futomat ® ) producing equipment for the production of oxygen-enriched water for drinking. It has been found that Ca 2+ content and the PS exposure are not changed significantly in oxygen-enriched water. However, one interesting effect was found when the RBCs were treated with LPA. Under such conditions the PS exposure was significantly reduced in futomat ® water compared to normal physiological solution. It remains open and requires clinical studies to see whether there is a relevant effect in human beings, meaning a positive effect preventing thrombus formation.

Active role of RBCs in blood coagulation
When the endothelium of blood vessels is damaged, platelets become activated and transport PS to their external membrane surface [50]. After activation, the exposed PS provides a catalytic surface for the formation of active enzyme-substrate  complexes of the coagulation cascade, especially for the tenase and prothrombinase complexes [51]. Under these circumstances exposed PS provides a procoagulant surface and is, in general, needed as a response to injury. Therefore, the mechanism of PS exposure has to occur with a relative high transport rate of the lipids. Platelets treated with a Ca 2+ ionophore show a phospholipid scrambling rate of 78 × 10 3 per second [51]. Human RBCs also show the mechanism of PS exposure after increased intracellular Ca 2+ content (see above) and are able to adhere to endothelial cells under pathophysiological conditions [22,[52][53][54]. In addition, exposed PS is sought to serve as a signalling component for macrophages to eliminate old or damaged RBCs from the circulation [55][56][57][58]. Since PS-exposing RBCs can adhere to the vascular wall, which may lead to disturbance of the microcirculation [59], the elimination of these cells is a very important mechanism. However, compared to platelets, RBCs have a lower phospholipid scrambling rate (0.45 × 10 3 per second) [51].  [23], panels (B) and (D) are reprints from [35] and panel (C) is a reprint from [60].
Even more important was the demonstration that human RBCs play an active role in clot formation [23]. This is lacking in medical textbooks, where one can find statements claiming that RBCs only become part of clots because they are so abundant in the circulation. First demonstration of increased interacting forces between two RBCs when the intracellular free Ca 2+ concentration is increased was performed using non-invasive holographic optical tweezers [23]. In addition, using single-cell force spectroscopy, it has been shown that the upper force limit for Ca 2+triggered adhesion of the RBCs was approximately 100 pN, a value large enough to be of significance during clot formation or in pathological situations [23]. Figure 6 summarises the in vitro force measurements performed.

Active participation of RBCs in thrombotic events
There are numerous indications for the active participation of RBCs in the induction of thrombotic events. The first example we like to mention is the occurrence of thrombotic complications in anaemic patients that experienced a splenectomy. Numerous hereditary anaemic disorders such as spherocytosis, stomatocytosis or elliptocytosis are associated with distorted RBCs, which are preferentially removed in the spleen. Therefore, splenectomy is believed to improve the anaemic symptoms because cells cannot be removed in the spleen. In principle, this concept works out but with the severe side effect that some patients suffer from thrombotic events. Since the 'maintenance' of the RBCs in the spleen is missing, it is likely that the RBCs are the major cause for the thrombotic events. In patients diagnosed with hereditary xerocytosis, mostly related to mutations in the Piezo1 channel, thrombotic complications were regularly reported after these patients underwent splenectomy [61], whereas patients diagnosed for 'Gardos Channelopathy' , even after splenectomy, thrombotic events were not observed [62].
An even more prominent example is sickle cell disease associated with vasoocclusive pain crisis as the major and most severe symptom of the patients. Since the mutation associated with sickle cell disease is in the haemoglobin, it seems obvious that also the symptoms of the disease are associated with RBCs. The common belief is that vaso-occlusive pain crises in sickle cell disease patients are caused by the crystallisation of the mutated haemoglobin under deoxygenation conditions. While the sickle formation under deoxygenation at stasis is undoubted, it's not clear if the same shape change happens in vivo. However, although it is clear that deoxygenated RBCs of sickle cell disease patients have an impaired deformability, the link to the vaso-occlusive crises must be a bit more complicated because deoxygenation happens continuously as deoxygenated RBCs are continuously passing the circulation and vaso-occlusive pain crises happen only sporadically and are so far unpredictable. A possible explanation is the activity of the NMDA-receptors (see above) that are activated by homocysteine and homocysteic acid, which are markers for inflammation in the blood plasma [63].
Such the above described mechanisms triggered by intracellular Ca 2+ increase are likely to happen also during vaso-occlusive crises in sickle cell disease patients. A first clinical pilot study on sickle cell disease patients using memantine, a drug blocking the NMDA-receptor (and approved to treat Alzheimer disease), showed very promising results both in the support of the mechanism we sketch and in the patients showing a lower number and less severe vaso-occlusive pain crises [64,65].
Furthermore, it is well known that in RBCs of sickle cell disease patients, the Gardos channel activity is increased [66], which is an indicator for an increased Ca 2+ since the Gardos channel is a Ca 2+ -activated K + channel. However, a clinical trial testing senicapoc, a Gardos channel inhibitor, failed because vaso-occlusive crises were not improved [66]. Since senicapoc addresses the Gardos channel and © 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. not the upstream increase in Ca 2+ that causes all the effects described above, the failure of senicapoc is explainable.
Additionally, we like to mention and discuss another aspect: The process we describe, in particular the Ca 2+ -triggered aggregation to initiate thrombus formation, takes some time [23], and an argument is that the time required is too long that aggregation happens between the fast-moving RBCs in the circulation. In this context, we like to mention the hydrodynamic clustering [67], which is perfectly reversible but provides the cellular interaction time since the lifetime of the hydrodynamic clusters can be in the range of several seconds.

Conclusions and outlook
It seems obvious that RBC participation in blood coagulation and thrombus formation is more than an accidental trapping in the process. In this chapter we summarised indications, evidences and proofs for active participation of RBCs in blood clotting and thrombus formation. However, this concept so far did not make it into haematological text books and standard medical education. With this book chapter, we like to make a little contribution to better explain and propagate this concept. Although we face severe experimental and clinical evidence for the active participation of RBCs in blood coagulation and thrombus formation, there is a demand for further research on the regulation and manipulation of this aspect in the coagulation sometimes also referred to as RBC hypercoagulation. We are looking forward to the next years of investigations in coagulation and thrombosis research.

Conflict of interest
The authors don't declare a conflict of interests.