How platelets safeguard vascular integrity

Summary

The haemostatic role of platelets was established in the 1880s by Bizzozero who observed their ability to adhere and aggregate at sites of vascular injury. It was only some 80 years later that the function of platelets in maintaining the structural integrity of intact blood vessels was reported by Danielli. Danielli noted that platelets help preserve the barrier function of endothelium during organ perfusion. Subsequent studies have demonstrated further that platelets are continuously needed to support intact mature blood vessels. More recently, platelets were shown to safeguard developing vessels, lymphatics, as well as the microvasculature at sites of leukocyte infiltration, including inflamed organs and tumours. Interestingly, from a mechanistic point of view, the supporting role of platelets in these various vessels does not necessarily involve the well-understood process of platelet plug formation but, rather, may rely on secretion of the various platelet granules and their many active components. The present review focuses on these nonconventional aspects of platelet biology and function by presenting situations in which platelets intervene to maintain vascular integrity and discusses possible mechanisms of their actions. We propose that modulating these newly described platelet functions may help treat haemorrhage as well as treat cancer by increasing the efficacy of drug delivery to tumours.

Platelets support the semi-permeable barrier function of the resting endothelium

Platelets have long been recognised as the cellular orchestrators of primary haemostasis. As early as 1881, Bizzozero described their ability to promote thrombus formation and coagulation at sites of vascular injury [1–4]. In addition to their crucial role in primary haemostasis, platelets have also long been known to support the semi-permeable function of the endothelium. In 1940, Danielli showed that oedema in perfused frog hind legs was reduced when platelets were added to a perfusate medium [5]. Danielli also noted that this anti-permeability effect of platelets exceeded what could be accounted for by their impact on the colloid osmotic pressure of the perfusate [5]. The enhancing effect of platelets on vascular preservation during organ perfusion was further confirmed by studies showing that, in contrast to platelet-poor plasma (PPP), platelet-rich plasma (PRP) provided optimum perfusion conditions to maintain the barrier function of microvessels in various isolated perfused organs including dog thyroid [6], rat heart [7] and rabbit lungs [8]. In fact, extravasation of both water and plasma proteins in these organs was reduced during perfusion with PRP compared to PPP.

In intact organisms, experimental induction of thrombocytopenia by platelet depleting antiserum or irradiation has been shown to result over time in the disruption of the microvascular endothelium, as evidenced by the excessive leakage of radiolabelled albumin in the lungs and ear of thrombocytopenic sheep [9] and rabbits [10]. The causal relationship between thrombocytopenia and abnormalities in vascular permeability to plasma proteins was demonstrated by the reversal of these defects following transfusion with PRP [9, 10]. While these studies clearly established that circulating platelets support the barrier function of the resting endothelium in intact organisms, they did not determine their mode of action. However, mechanistic hints on this beneficial action of platelets were previously suggested by studies in perfused organs. Here, Danielli hypothesised that the antipermeability effect of platelets may be purely mechanical, with platelets filling gaps within the endothelial lining [5]. Supporting this “adhesion theory,” Gimbrone et al. reported platelets occupying gaps in the endothelium lining of thyroid perfused with PRP [6]. Moreover, these investigators noted signs of vascular degeneration such as endothelial necrosis in organs perfused with PPP in contrast to PRP, suggesting that platelets may also “nurture” the endothelium [6]. In 1975, an electron microscopy study by Kitchens and Weiss provided further explanations on how platelets support the integrity of resting endothelium [11]. In this study, ultrastructural endothelial changes associated with experimental thrombocytopenia were observed in lingual capillaries and post-capillary venules of rabbits. Thinning and fenestration of the endothelium, as well as an increase in microvessel permeability to the radiographic contrast agent Thorotrast injected intravenously, occurred within 6 hours of platelet depletion (Fig. 1A). These structural endothelial abnormalities reversed after spontaneous platelet repletion, demonstrating the long-term need for platelets to maintain the structure and integrity of resting microvessels. In human, ultrastructural abnormalities of the capillary endothelium similar to that observed in thrombocytopenic animals were later described by Kitchens and Pendergast in patients with severe thrombocytopenia (< 15,000 platelets/μL) [12].

Figure 1. Platelets support the integrity and barrier function of resting and developing blood vessels.

A. Ultrastructural changes of the resting endothelium associated with severe thrombocytopenia. Upper panels: the endothelium of a normal capillary (Control) is of a fairly uniform thickness but is attenuated by more than 50% in animals rendered thrombocytopenic by platelet antiserum (Thrombocytopenic), thereby nearly allowing a vesicle (arrow) to bridge its thickness. Arrow: Basement membrane. Lower panels: Localisation of the intravascular contrast agent Thorotrast (T), 5 minutes after intravenous injection in either control or thrombocytopenic rabbits. While Thorotrast was observed only in the vascular lumen of control animals, it was seen also in the extracellular space of thrombocytopenic animals, showing the reduced barrier function of the endothelium during thrombocytopenia. Arrow: Basement membrane. Adapted from Ref. [11], with permission of the American Society of Hematology; permission conveyed through Copyright Clearance Center, Inc.

B. Effect of platelet depletion on angiogenesis. Photographs of eyes from control and thrombocytopenic mice, 72 and 96 h after intracorneal implantation of hydron pellets containing 80 ng of basic FGF. Note the undefined borders of the vessels due to haemorrhage in the eyes of platelet depleted mice. Reproduced from Ref. [41], with permission; copyright (2006) National Academy of Sciences, USA.

In vitro studies have brought further insight regarding the mechanisms underlying the maintenance of vascular integrity and barrier function by platelets. Early on, it was shown that platelets and platelet components, including serotonin and ADP, promote the proliferation of endothelial cells in culture [13–16], suggesting that the supportive role of platelets may include a mitogenic activity. Experiments on cultured endothelial cells have also demonstrated that platelet lysates and conditioned medium from unstimulated platelets can replicate the permeability-decreasing effect of whole platelets [17–21], indicating that platelets release soluble factors and/or microparticles that modulate the permeability of the vascular lining cells. Conditioned medium from platelets was indeed shown to decrease the basal albumin flux and/or to increase the electrical resistance across human [20] and bovine [18, 19, 21, 22] pulmonary artery endothelial cell monolayers. Many types of molecules contained by platelets were shown to reduce vascular permeability in vitro and/or in vivo: adenine nucleotides [19], serotonin (5-HT) [23] and angiopoietin-1 (ang-1) [24, 25], and lipids including sphingosin-1-phosphate (S1P) [20] and lysophosphatidic acid (LPA) [18, 26]. However, the specific contribution of these factor(s) to the enhancement of the vascular barrier by platelets remains unknown and studies have suggested that the active molecule is either a protein [17, 21, 22, 27] or a lipid mediator [18, 26]. These conclusions are not mutually exclusive and the respective contribution of these factors to the maintenance of the vascular barrier in vivo may vary with the experimental model or vascular type and with the particular inducer of platelet secretion. In fact, the expression of junctional molecules, for example, is not uniform along the vascular tree [28] and platelets may differentially secrete the content of their various granules depending on the agonist and platelet receptor stimulated [29, 30].

Taken together, these studies have demonstrated that platelets can support the resting vascular endothelium by at least four mechanisms: 1) platelets physically block potential gaps in the vascular lining [5, 6], 2) platelets and platelet components promote the growth of endothelial cells [13–16], 3) platelets help maintain the endothelium ultrastructure [13–16], and 4) platelets release soluble factors that enhance the barrier function of the endothelium [17–21]. Importantly, it has been implied that the loss of these endothelium-supporting functions of platelets could be the cause of bleeding events, presumed to be spontaneous, that occur during thrombocytopenia [31, 32]. It must be noted that in the absence of injury some thrombocytopenic patients show spontaneous bleeding while others do not, despite equally low platelet counts [33], thus indicating that contributing factors other than thrombocytopenia are required to induce bleeding in these patients. Furthermore, while defects in endothelial junctions would definitely lead to leakage of fluid and proteins into tissues, they are unlikely sufficient to cause bleeding. Supporting this hypothesis, application of the pro-permeability factor VEGF onto the subcutaneous connective tissue of thrombocytopenic mice through a dorsal skinfold chamber, increased Evans blue leakage in the skin but did not cause any skin bleeding [34]. Similarly, although mutant mice selectively lacking S1P in plasma exhibit basal vascular leak to proteins and fluid in the lungs and increased local response to permeability-inducing agents, associated with increased interendothelial cell gaps in venules, they do not show any bleeding either at the basal state or upon stimulation with pro-permeability factors [35]. Extravasation of red blood cells does not occur only through the endothelium but also through the basement membrane, which implies degradation or rupture of the basal lamina surrounding vasculature. Thus, although the absence of platelets may weaken endothelial junctions and favours breakdown of the endothelial permeability barrier, additional injurious factors capable of damaging basement membranes and/or abnormalities of the basement membrane are likely also required for bleeding events to occur during thrombocytopenia. As we will discuss later, the presence of inflammatory cells is one of these factors.

Platelets support vascular development and remodelling

While early studies focused on the role of platelets in the maintenance of the mature vasculature, more recent studies have indicated that platelets also intervene in supporting developing blood and lymphatic vessels, as well as regeneration of blood vessels after vascular injury. Platelets [36, 37], platelet releasate [36, 37], and platelet microparticles [38] have been shown to have proangiogenic effects in various models of angiogenesis. For instance, platelets and/or platelet microparticles stimulated endothelial cell proliferation and differentiation into capillary-like structures in vitro [36, 38, 39], and increased neovascularisation of mouse Matrigel implants in vivo [36, 38]. Rhee et al. also reported that hypoxia–induced retinal neovascularisation in mice was reduced by platelet-depletion or anti-platelet therapy [37]. The pro-angiogenic effect of platelets is likely due in part to the fact that they are an important source of pro- and anti-angiogenic factors that are packaged into distinct granule populations which are released upon stimulation [30, 40]. Our group has shown both the ability of platelets to stimulate angiogenesis but, in addition, that platelets are essential to prevent haemorrhage from the sprouting blood vessels in vivo. Using the cornea micropocket assay and the Matrigel model in mice, we found that platelet-depleted mice experienced a significant reduction in corneal neovascularisation and developed haemorrhage at sites of angiogenesis [41] (Fig. 1B). A similar haemorrhagic phenotype was observed in mice lacking GPIb, one of the major platelet adhesion receptors to the vessel wall, indicating that adhesive interactions between platelets and angiogenic vessels stabilise developing blood vessels [41]. Recent experimental evidence has also indicated that platelets could promote re-endothelialisation following vascular injury. In fact, platelets were shown to recruit circulating progenitor cells at sites of vascular injury or tissue ischaemia [42–45], and to stimulate the differentiation of progenitor cells into endothelial cells in vitro [44]. Blocking of the chemokine stromal-derived factor-1 (SDF-1) inhibited both recruitment and differentiation of progenitor cells, indicating that platelet-derived SDF-1 was responsible for the regenerating actions of platelets [44].

The involvement of platelets in lymphangiogenesis has just been unravelled by a series of 2010 publications that demonstrate a crucial role for platelets in regulating lymphatic vascular development [46–48]. Viewed as a whole, these studies show that activation of the “Src homology 2 (SH2) domain-containing leukocyte protein of 76 kDa” (SLP-76) signalling pathway in platelets, following the interaction between the C-type lectin-like receptor 2 (CLEC-2) expressed by platelets and podoplanin (a transmembrane protein expressed on the surface of lymphatic endothelial cells), is essential for separating the lymphatics from blood vessels during embryonic development. Failure in that platelet activation pathway due to a genetic deficiency in podoplanin [46, 48], CLEC-2 [46, 48] or platelet SLP-76 [46] or to the absence of platelets by targeted genetic ablation of the megakaryocyte/platelet lineage or antibody-mediated platelet depletion [47], systematically resulted in embryonic/neonatal lethality associated with severe lymphatic vascular defects. These vascular defects mostly consisted of abnormal connections between blood and lymphatic vessels resulting in the development of blood-filled lymph sacs and lymphatic vessels, and in impaired lymphatic drainage evidenced by the presence of cutaneous and intestinal oedema in neonatal animals. Interestingly, early studies have shown abnormal red blood cell accumulation within the lymph of animals rendered thrombocytopenic by irradiation [10, 49–51]. Although entry of red blood cells in the lymph of irradiated animals occurred even in the absence of evidence of a haemorrhagic tendency [50], it was interpreted as an indicator of the degree of irradiation-induced capillary damage and erythrodiapedesis. The newly discovered role for platelets in preventing blood-lymphatic vascular mixing may provide an alternative explanation for the changes in cellular composition of the lymph following irradiation-induced sustained thrombocytopenia.

Platelets promote inflammation

Platelets are known to play an important role in inflammatory processes and diseases. The best understood mechanism by which platelets contribute to inflammatory situations is the platelet-dependent enhancement of leukocyte infiltration [52]. The ability of platelets to promote recruitment of leukocytes has been documented experimentally by the reduced number of leukocytes in inflamed organs of thrombocytopenic mice [53–56] or in mice in which one of the major platelet adhesion receptors was neutralised either by a blocking antibody or by genetic deficiency [53–58]. Intravital microscopy experiments provided evidence that the enhancement of leukocyte recruitment in inflamed organs by platelets occurs through direct interactions between platelets, leukocytes and activated endothelial cells. These are mediated by platelet adhesion molecules including P-selectin [56], β3 integrin [57, 58], GPVI [53] and GPIb [57]. Activated platelets have also been shown to promote leukocyte recruitment in the inflamed vasculature by secreting chemokines [59, 60], up-regulating the expression of adhesion molecules on the endothelium [59, 61], and by loosening endothelial cell junctions [62, 63]. The impact on tissue integrity of this leukocyte-supportive effect of platelets can either be beneficial or deleterious, depending on the pathophysiological situation. For example, platelet-mediated leukocyte recruitment has been shown to promote viral clearance in mouse models of acute viral hepatitis [54] and in mice with lymphocytic choriomeningitis [55], but also promotes the development of atherosclerotic lesions in apoE−/− mice [57, 58, 64], cerebral injury after experimental ischaemia-reperfusion [58, 65], and neutrophil-dependent glomerular and lung injury in mouse models of glomerulonephritis [53, 56, 66] and of acute lung injury [62, 67]. From a therapeutic point of view, reducing leukocyte infiltration by targeting platelet/leukocyte interactions may be beneficial in situations in which excessive leukocyte infiltration and activation cause severe tissue damage [67]. The dual consequences of the leukocyte-supportive actions of platelets is further illustrated by their ability to stimulate the formation of neutrophil extracellular traps (NETs) that entrap microbes but also cause significant collateral damage to surrounding tissue [68–70].

In addition to their ability to promote leukocyte infiltration, platelets can also modulate systemic inflammatory responses. The number of circulating blood monocytes and neutrophils, as well as plasma levels of tumour necrosis factor-alpha (TNF-α), interleukin-6, monocyte chemotactic protein-1 (MCP-1), and mortality after experimental thermal injury, have been shown to be increased in platelet-depleted mice compared with control mice [71]. In contrast, levels of the immune-regulatory cytokine transforming growth factor beta-1 (TGFβ₁) were reduced in platelet-depleted mice, leading the authors to propose that modulation of the systemic inflammatory response by platelet-derived TGFβ₁ could contribute to the protective effect of platelets after injury. Another platelet-associated orphan receptor called triggering receptor expressed on myeloid cells–like transcript-1 (TLT-1), a molecule specific to platelet and megakaryocyte α-granules, has been recently shown to limit the systemic inflammatory response to intraperitoneal injection of lipopolysaccharide (LPS) [72]. Mice deficient for TLT-1 (Treml1−/−) developed higher plasma levels of TNF-α than wild-type mice and were more likely to succumb during LPS challenge, indicating that platelet TLT-1 plays a protective role during endotoxemia by dampening the inflammatory response. CD40L, a member of the tumour necrosis factor family that is expressed by platelets and that plays a major role in immune responses, may also contribute to the immunomodulatory properties of platelets. In fact, when infected with lymphocytic choriomeningitis virus (LCMV), platelet-depleted mice have been shown to have a defective viral clearance and a reduced virus-specific cytotoxic T lymphocyte (CTL) response, leading to lethal haemorrhagic anaemia that can be fully rescued by transfusion of platelets from wild-type mice, but only partially by platelets from CD40L deficient mice [55], indicating a role for platelet CD40L in prevention of haemorrhage and viral clearance during LCMV infection [55].

Platelets secure inflamed vessels

Platelets also play a highly positive vascular protective role during the inflammatory response. Recent experimental evidence documents that inflammation is a potent trigger of bleeding during thrombocytopenia, unveiling a crucial role for platelets in maintaining vascular integrity in inflammation. Using a mouse model of acute severe thrombocytopenia induced by platelet depletion (< 2.5% of control platelet counts), our group has shown that while thrombocytopenic mice did not show spontaneous bleeding if unchallenged, these animals were highly susceptible to haemorrhage when subjected to a localised inflammatory stimulus, in either the skin or internal organs such as the brain and lungs [73]. Recently, a case of petechial bleeding restricted to the area of acute inflammation secondary to sunburn, has been reported in a person with mild thrombocytopenia [74], indicating that inflammation may also induce haemorrhaging in thrombocytopenic patients. Intravital microscopic observation in mice of vessels at sites of immune complex-induced acute skin inflammation has revealed that, in platelet-depleted animals, bleeding originated mainly from capillaries (unpublished observation by Ho-Tin-Noé) and post-capillary venules [73]. Haemorrhage starts as early as 20 minutes after the onset of inflammation and spreads extensively and continuously thereafter (Fig. 2). In comparison, only a few petechiae appear over the same period of time in mice with normal platelet count. These observations underline the continuous need for platelets to secure the integrity of the microcirculation from the very onset of inflammatory reactions.

Figure 2. Platelets secure inflamed vessels.

Photographs of progressing immune complex-induced skin inflammation in dorsal skinfold chambers of control and thrombocytopenic mice. In the absence of platelets, petechial bleeding (arrowheads) was detected as early as 20 minutes after the onset of inflammation while there were virtually no petechial spots in nondepleted control animals.

Although the exact mechanisms of vessel protection by platelets during inflammatory reactions remain unclear, there is evidence suggesting that this protective activity of platelets could involve dampening of inflammatory responses and countering of leukocyte activities. In fact, renal bleeding in thrombocytopenic mice subjected to glomerulonephritis was markedly attenuated in mice lacking the leukocyte-specific integrin Mac-1, in which glomerular neutrophil infiltration was prevented [66]. Similarly, depletion of neutrophils prevented TNF-α-induced skin bleeding in thrombocytopenic mice [34]. Thus, in these inflammatory models, platelets prevent and/or counter neutrophil-mediated vessel injury. Activated leukocytes secrete injurious products such as matrix metalloproteases (MMPs), serine proteases and reactive oxygen species (ROS) that can inflict vascular damage. Therefore, prevention of inflammatory bleeding by platelets could simply rely on their well-known ability to form plugs at sites of vessel injury. Surprisingly however, experiments with mice carrying mutations present in major bleeding disorders (human type III von Willebrand disease, Bernard-Soulier syndrome, Glanzmann thrombasthenia), have shown no haemorrhaging at sites of inflammation [73]. Indeed, mice with deficiency in von Willebrand factor, GPIb, or β3, did not show any bleeding when subjected to immune complex-induced acute skin inflammation [73]. Thus, none of these platelet receptors appears to be absolutely required to prevent inflammatory haemorrhage and the role of platelets in this model is likely different from that during platelet plug formation. Supporting this hypothesis, observation of fluorescently-labelled platelets in the inflamed microcirculation by intravital microscopy through a dorsal skinfold chamber revealed platelet rolling, but no thrombi formation at time points for which bleeding could nevertheless already be seen in platelet-depleted mice [73]. Furthermore, Hirahashi et al. showed that platelet accumulation occurred in inflamed glomerular capillaries before any evidence of thrombosis, and that prevention of this initial platelet deposition, by immunodepletion of platelets, resulted in accelerated and exacerbated renal haemorrhage, glomerular injury, and glomerular endothelial damage [66]. Therefore, it appears that early platelet presence exerts cytoprotective effects on inflamed vessels independently of thrombus formation. This may occur through dampening of inflammatory responses [71, 72] but also through platelet-dependent neutralisation of injurious products released by activated leukocytes. Indeed, platelet granules contain a variety of factors like serpins [75–77], tissue inhibitors of metalloproteinases (TIMPs) [78] and ROS scavengers that could be released upon platelet activation [79, 80] and counteract the injurious effects of leukocytes.

Importantly, it should be noted that the vascular protective action of platelets during inflammation has been experimentally shown in mouse models of severe thrombocytopenia (< 5 % normal platelet count, approximately < 30,000–40,000/μL mouse platelets) [55, 66, 73], and that it can be restored by transfusing platelets to replenish as little as 10 % of circulating platelets (approximately <60,000–80,000/μL mouse platelets) [55, 73]. Therefore, platelets are extremely potent at preventing inflammatory bleeding. The fact that a very low threshold of platelet is sufficient for this vasculoprotective action of platelets may account for the absence of inflammatory bleeding in studies using models of thrombocytopenia in which at least 10 % of residual circulating platelets were left [62, 81]. However, in a recent study by Petri et al., no bleeding was observed into thioglycollate-inflamed peritoneum of thrombocytopenic mice, in spite of severe thrombocytopenia (< 5 % normal platelet count) [63]. This suggests that the onset of inflammatory bleeding during thrombocytopenia does not only depend on the number of circulating platelets but probably also on the organ concerned and on the severity and cause of the inflammatory reaction. Also, one should keep in mind that due to the different platelet-leukocyte ratios in peripheral blood between human and mice (e.g. 2500–7500/μL neutrophils in human against 500–1000/μL in mouse and 100,000–400,000/μL platelets in human against 600,000–1000,000/μL in mice), it is difficult to directly transpose the protective threshold of 10 % circulating platelets in mice to humans. Thus, determination of the threshold of platelets needed to restore their vasculoprotective action in humans potentially represents an important issue open to investigation.

Platelets maintain tumour vessel homeostasis

Solid tumours are critically dependent on blood vessels because of their increasing need for nutrients and oxygen supply as they grow [82]. Although tumour blood vessels are formed through angiogenesis from the host vasculature, they display a markedly abnormal phenotype. Tumour vessels have excessive branching, uneven diameters, chaotic flow patterns and thin walls; their endothelium, similar to inflamed vasculature, is hyperpermeable to macromolecules [83, 84]. Tumour vessels are also continually exposed to injuries due to massive leukocyte infiltration [85–87]. Therefore, when considering the supporting role of platelets in angiogenesis and inflammation, solid tumours represent a site in which these platelet functions are likely to come into full effect. Indeed, recently we have demonstrated that platelets protect the tumour vasculature and continuously prevent the tumour from haemorrhaging. The role of platelets in tumour vascular homeostasis was uncovered by inducing acute thrombocytopenia with a platelet-depleting antibody in tumour-bearing mice. Induction of thrombocytopenia consistently resulted in massive bleeding in and surrounding the tumour, without affecting vascular integrity elsewhere [34, 88, 89]. Moreover, thrombocytopenia-induced tumour haemorrhage was independent of tumour age, type, and location. Thus it affected subcutaneous melanoma and mammary carcinoma models, as well as established melanoma lung metastasis [34, 88, 89]. This suggests that the protective effect of platelets on tumour vessels may be generalised to all vascularised tumours.

Interestingly, and similarly to protection of inflamed vessels by platelets during early stages of inflammatory reactions, prevention of tumour haemorrhage by platelets seems to be independent of their capacity to form thrombi. In fact, despite presenting severe haemostatic problems, mice treated with an inhibitor of GPIbα, mice lacking VWF, β3 integrins, or mice defective in activation of platelet integrins, showed no severe spontaneous tumour bleeding [34, 88]. Observation of fluorescently labelled platelets by intravital microscopy and immunostaining of platelets in tumour sections did not reveal the presence of thrombi in the tumour microcirculation [88, 90], further supporting the concept that protection of tumour vessels by platelets occurs independently of platelet plug formation. Importantly, in contrast to resting platelets, thrombin-degranulated platelets failed to prevent tumour bleeding when transfused into thrombocytopenic mice, suggesting positive effects of molecules released upon platelet activation [88]. Activated platelets secrete soluble and membrane-bound granular components, but also shed plasma membrane receptors and pro-coagulant microparticles, which could all contribute to the biological activities activated platelets exert on tumour vessels and tumour infiltrating leukocytes.

The exact identity of the platelet-derived factors responsible for tumour vessel protection remains unknown. We have shown that these factors neutralise damages produced by innate immune cells within the tumour stroma. Immunostaining of haemorrhaging tumour sections revealed that tumour bleeding co-localised with neutrophils and macrophages and occurred in close proximity to areas containing large numbers of dead leukocytes [34]. In mice genetically deficient in leukocyte integrins, we found that thrombocytopenia-induced tumour haemorrhage was reduced as these mice had decreased numbers of tumour-infiltrating neutrophils and macrophages [34]. Thus, platelet-derived factors protect tumour vessels against injuries inflicted by tumour-infiltrating innate immune cells. The fact that protection of tumour vessels by platelets shares common mechanisms with that of inflamed vessels suggests that similar platelet factors may be responsible for protecting vascular integrity in both cases (Fig. 3).

Figure 3. Schematic of the maintenance of tumour vascular integrity by platelets.

Left: The tumour microenvironment provides proinflammatory and pro-coagulant signals that result in endothelial activation, in leukocyte infiltration and local degranulation of platelets. The factors released from activated platelets stabilise the tumour vasculature by inhibiting the activity and/or the release of injurious products from tumour infiltrating-leukocytes. Right: During thrombocytopenia, the absence of the leukocyte-countering activities of platelets leads to cytotoxic intratumour haemorrhage through leukocyte-induced vascular breaches. Induction of tumour haemorrhage by targeting platelets enables improvement of chemotherapy efficacy by increasing the accumulation of the circulating drugs in the tumour. Insets, representative photographs of subcutaneous Lewis lung carcinoma tumours from control (left) and thrombocytopenic (right) mice.

Targeting platelets to improve therapy

An important research goal of cancer therapy is to increase the delivery of anticancer agents to tumour sites for maximum treatment efficacy while minimising side effects on healthy tissues. Inflammation has been found to be present in almost all tumours [91] and given that thrombocytopenia-induced haemorrhage is seen specifically at the tumour site, this phenomenon has great potential as a new drug delivery system to increase the efficacy of chemotherapy. In fact, we have recently demonstrated that the intratumoural vascular breaches induced by thrombocytopenia allow increased accumulation of fluorescently-labelled 1 μm microspheres in the tumours compared to tumours of mice with normal platelet count [89]. This result showed that the openings in the tumour vasculature could be used as delivery ports for circulating particles to the tumour site. Moreover, thrombocytopenia allowed the accumulation of paclitaxel (taxol), a chemotherapeutic agent, in the tumour and the combination of platelet depletion with a single paclitaxel treatment resulted in a significant reduction in tumour growth of murine mammary or lung carcinoma [89]. This greater accumulation of drug at the tumour site after platelet depletion increased the efficacy of the drug leading to an increase in tumour cell apoptosis and a decrease in their proliferation (Fig. 3). No signs of higher toxicity were observed on blood cells and organs of the thrombocytopenic mice. This study suggests that low platelet count or inhibition of the “stabilising” activity of platelets could be used to improve the delivery of chemotherapeutic agents to the tumour. Thrombocytopenia-induced haemorrhage does not require knowledge of specific tumour antigens but rather targets all tumours. This new way of enhancing the delivery of chemotherapy by targeting platelets could therefore be widely applicable. In this context, the identification of the factors released by platelets that prevent the inflammation-induced endothelial injuries, and the associated platelet activation pathways, represent an important area of future research. Targeting these platelet signalling pathways and/or the subsequently released factors could enable the induction of specific breaches in the tumour vasculature without the need to lower platelet count and compromise haemostasis. Interestingly, dual aspirin/clopidogrel antiplatelet therapy administered to a patient following intracoronary stenting, unexpectedly unmasked a colon cancer by precipitating occult gastrointestinal bleeding from the so far silent tumour [92]. Although by blocking global platelet activation combined treatment with clopidogrel and aspirin unselectively blocks both platelet aggregation and the release of soluble factors by platelets, this case illustrates the fact that targeting platelet function represents a feasible strategy to destabilise the tumour vasculature in humans.

Conclusion

The studies described in this review have established that, besides being the cellular orchestrators of primary haemostasis, platelets should also be considered as critical guardians of vascular integrity. Platelets continuously support the barrier function of the resting endothelium and, in inflammation, they prevent or heal vascular injuries caused by the infiltrating leukocytes. Importantly, these vascular protective actions of platelets appear to be mechanistically distinct from their ability to stop bleeding from injured vessels via platelet plug formation. Taking advantage of this mechanistic dichotomy, one might speculate that selective blocking of platelet adhesion and aggregation receptors involved in thrombus formation could efficiently prevent occlusive thrombus formation without affecting the vascular integrity supportive functions of platelets or increasing the risk of spontaneous and inflammatory bleeding. Supporting this hypothesis, it has been shown that selective inhibition of the early steps of platelet adhesion profoundly protects mice from ischaemic stroke without increasing the risk of intracranial haemorrhage [65, 93]. Conversely, inhibition of the leukocyte-countering actions of platelets may enable selective destabilisation of the tumour vasculature without interfering with primary haemostasis. Understanding the molecular mechanisms by which platelets stabilise inflamed blood vessels could lead to novel therapeutics preventing bleeding in thrombocytopenia and reducing the amount of platelets needed for transfusion. Reciprocally, inhibition of this protective pathway would improve drug delivery to inflamed sites such as tumours.