Platelets at the crossroads of thrombosis, inflammation and haemolysis

Summary

Platelets play a critical role at the interphase of thrombosis and inflammation, key features in haemolysis-associated disorders. Exercising this role requires expression of pattern recognition receptors by platelets, including toll-like receptor 4 (TLR4) and nucleotide-binding domain leucine rich repeat containing protein 3 (NLRP3), the latter forming intraplatelet multiprotein inflammasome complexes. Platelets are a potential target of various damage-associated molecular pattern (DAMP) molecules, such as free haem, a degradation by-product of haemoglobin oxidation during haemolysis, and high-mobility group box 1 (HMGB1), a DNA-binding protein released by dying or stressed cells and activated platelets. We have recently identified platelet TLR4, NLRP3, and Bruton tyrosine kinase (BTK) as critical regulators of platelet aggregation and thrombus formation, suggesting that the BTK inhibitor ibrutinib is a potential therapeutic target. Increasing evidence suggests that these and other DAMP-driven signalling mechanisms employed by platelets might be key in mediating inflammation and thrombosis encountered in haemolytic disorders. However, the precise regulatory triggers and their clinical relevance are poorly understood. We provide new insights into these less-well characterised platelet mechanisms, which are potentially targetable in haemolytic disorders.

An accelerated destruction of red blood cells (haemolysis) is a critical feature of various inherited and acquired disorders that are intrinsically linked with vascular thrombosis and inflammation, including sepsis (Levi et al, 2013), valveinduced haemolysis (Edmunds et al, 1996), thrombotic thrombocytopenic purpura/haemolytic uraemic syndrome (HUS) (George & Nester, 2014), autoimmune haemolytic anaemias (Hendrick, 2003), thalassaemia (Eldor & Rachmilewitz, 2002) and sickle cell disease (SCD) (Ataga et al, 2007). Typical thrombotic complications in patients with haemolytic disorders are venous thromboembolism, pulmonary thrombosis and stroke (Hendrick, 2003; Stein et al, 2006; Ataga et al, 2007; Au et al, 2009; Levi et al, 2013). Several mechanisms may underlie the pathophysiology of the hypercoagulable and inflammatory state in haemolysis.

In SCD and thalassaemia, red blood cells express phosphatidylserine in significant amounts on the cell surface (Eldor & Rachmilewitz, 2002; Ataga et al, 2007), which promotes endothelial cell adhesion and triggers activation of thromboinflammatory pathways (Zwaal & Schroit, 1997; Setty et al, 2002). Tissue factor is a key initiator of coagulation and is upregulated on circulating mononuclear cells in SCD patients in steady state and pain crisis (Key et al, 1998). Moreover, plasma levels of procoagulant soluble CD40 ligand (sCD40L) derived from platelets are increased in SCD patients at steady state and further elevated in patients with painful crises (Lee et al, 2006).

Haemoglobin released from the ruptured red blood cells is oxidised to free haem, which regulates transcription of antioxidant enzymes, such as haem-oxygenase 1 (Ogawa et al, 2001). Free haem and other molecules, such as adenosine diphosphate (ADP) and uric acid, act as erythrocytederived damage-associated molecular pattern molecules (DAMPs) that promote thrombosis and inflammation (Gladwin & Ofori-Acquah, 2014). The DNA-binding protein high-mobility group box 1 (HMGB1) is a non-erythrocyte DAMP associated with haemolysis and abnormal coagulation, including sepsis (Wang et al, 1999), disseminated intravascular coagulation (DIC) (Hatada et al, 2005), trauma (Levy et al, 2007), HUS (Lee et al, 2013) and SCD (Xu et al, 2014). HMGB1 plasma levels are elevated in SCD patients at baseline and further increased during acute sickling events (Xu et al, 2014). HMGB1 induces thrombosis and inflammation when exported to the cell surface or released into the extracellular space by stressed or dying cells and activated platelets (Rouhiainen et al, 2000; Andersson & Tracey, 2011; Maugeri et al, 2012; Gawaz & Vogel, 2013; Vogel et al, 2014, 2015a,b, 2016). The effects of HMGB1 upregulation in the context of haemolysis are poorly understood.

HMGB1 and haem signal through pattern recognition receptors, including toll-like receptor 4 (TLR4) (Park et al, 2004, 2006; Figueiredo et al, 2007; Belcher et al, 2014) and nucleotide-binding domain leucine rich repeat containing protein 3 (NLRP3) (Dutra et al, 2014; Li et al, 2014; Chi et al, 2015). Platelets express TLRs and nucleotide-binding oligomerization domain-like receptors (NLRs) (Shiraki et al, 2004; Hottz et al, 2013), implying that platelets are a putative target of these haemolysis-associated DAMPs. In a trauma/haemorrhagic shock model in mice with platelet-specific ablation of HMGB1, we have shown that platelets are the major source of HMGB1 within thrombi and that HMGB1 derived from platelets is a critical mediator of thrombosis and inflammation, via activation of a non-canonical TLR4/MyD88/soluble guanylate cyclase (sGC) platelet signalling pathway that was dependent on expression of platelet P-selectin (SELP, CD62P) (Vogel et al, 2015a). In another recent study, we have identified NLRP3 and Bruton tyrosine kinase (BTK) in platelets as critical co-regulators of platelet aggregation and thrombus formation (Murthy et al, 2017).

Here, we highlight the current evidence on these less-well characterised platelet danger signalling mechanisms in the context of haemolysis-associated thrombosis and inflammation, with an emphasis on SCD.

Platelets in sickle cell disease

SCD is an inherited blood disorder, caused by the presence of an abnormal haemoglobin (HbS). Polymerization of the deoxygenated HbS leads to “sickling” of red blood cells, followed by adhesion of the sickled erythrocytes to vascular endothelium, neutrophils and platelets. These cell interactions set off a cascade of downstream pathophysiological mechanisms that underlie the two hallmarks of the disease – chronic haemolysis and recurrent vaso-occlusions (Zhang et al, 2016; Thein et al, 2017). Patients with SCD experience a wide range of symptoms, including acute pain, chronic pain and multiorgan failure, due to a combination of pathological factors arising from haemolysis, inflammation and thrombosis (Thein et al, 2017). The high rate of red blood cell breakdown and release of cell-free haemoglobin and haem deplete endogenous haptoglobin and haemopexin, leading to free circulating haem, which triggers vaso-occlusions and inflammation (Belcher et al, 2014). Treatment options of SCD are still limited. Established disease management options include the myelosuppressive agent hydroxycarbamide, blood transfusions, penicillin prophylaxis, vaccinations against Streptococcus penumoniae and other pathogens, and patient education.

Even under SCD steady state conditions, circulating platelets are significantly activated in patients, and are further activated in vaso-occlusive crisis and acute clinical events, such as acute chest syndrome (Westwick et al, 1983; Wun et al, 1997; Villagra et al, 2007; Garrido et al, 2017). Beyond their role in haemostasis, platelets provide a link between thrombosis and inflammation (Semple et al, 2011; Rondina et al, 2013; Morrell et al, 2014; Vogel et al, 2015a, 2016; Murthy et al, 2017). More than two decades ago, Brittain et al (1993) proposed that activation of platelets has a contributory role in sickle vaso-occlusions. Platelets are capable of forming aggregates with monocytes and neutrophils; the aggregates were significantly elevated in SCD patients (Frelinger et al, 2014) and caused arteriolar platelet-neutrophil microemboli in the lungs of transgenic humanised SCD mice (Bennewitz et al, 2017). Platelets also facilitated the formation of circulating neutrophil-red blood cell aggregates in SCD patients in steady state, potentially contributing to the vaso-occlusion pathophysiology (Dominical et al, 2014).

The platelet activation marker P-selectin plays a critical role in the formation of platelet-neutrophil aggregates. Lung vascular permeability in sickle mice was significantly improved in the presence of a neutralising P-selectin antibody or the platelet inhibitor clopidogrel (Polanowska-Grabowska et al, 2010). Transgenic sickle mice deficient in P-selectin and E-selectin showed defective leucocyte recruitment to the vessel wall and were protected from vaso-occlusions (Turhan et al, 2002). In a double-blind, randomised phase 2 trial, SCD patients that received crizanlizumab, a P-selectin neutralising antibody, had significantly lower annual rates of sickle cell-related vaso-occlusive pain crises, suggesting that P-selectin is a therapeutic target in SCD (Ataga et al, 2017). On the other hand, the platelet inhibitor prasugrel did not significantly affect the rate of vaso-occlusive crises in children and adolescents with SCD in phase 3 trial (Heeney et al, 2016), indicating that further studies investigating the effectiveness of various platelet-specific targets in SCD are warranted.

A hypercoagulable state is known to contribute to the occurrence of vascular events in the microcirculation in SCD, which is at least partly mediated by upregulated activation and aggregation of circulating platelets (Ataga & Key, 2007; De Franceschi et al, 2011; Sparkenbaugh & Pawlinski, 2013). The specific mechanisms that control these activated platelet responses, and their pathophysiological significance for thromboinflammatory processes in SCD and other haemolytic conditions, however, are still poorly understood.

Pattern recognition receptor signaling in platelets

The NLRP3 inflammasome is a multiprotein complex expressed in immune cells and platelets that consists of the pattern recognition receptor NLRP3, apoptosis-associated speck-like protein containing a C-terminal caspase recruitment domain (CARD) (ASC), caspase-1 and BTK (Agostini et al, 2004; Willingham et al, 2009; Hottz et al, 2013; Ito et al, 2015; Liu et al, 2017; Murthy et al, 2017). Activation of the NLRP3 inflammasome upregulates caspase-1 activity and triggers the production of bioactive interleukin-1b and other proinflammatory cytokines, such as interleukin-18 (Fig 1).

Fig 1.

Schematic overview of how platelets may orchestrate thrombosis and inflammation in haemolysis. Extracellular HMGB1 and haem are haemolysis-associated DAMPs that trigger pattern recognition receptor signalling pathways in platelets. HMGB1 signals via platelet TLR4, inducing complex formation between MyD88 and sGC leading to upregulation of enzymatic activity of sGC and upregulated platelet CD62P expression. HMGB1 sensing and secretion by platelets may also be controlled by intraplatelet NLRP3, which cooperates with ASC and caspase-1 within inflammasome complexes that activate IL-1b and IL-18 signalling. BTK and SIRT are critical regulators of the NLRP3 inflammasome. Free haem may activate both plateletspecific TLR4 and NLRP3. Activation of NLRP3 may or may not depend on purinergic P2X receptors. TLR4 and NLRP3 signalling in platelets upregulates expression of CD62P on the platelet surface. Expression of CD62P and secretion/release of IL-1b, IL-18 and HMGB1 by activated platelets are key events in regulating thrombosis and inflammation. Therapeutics that potentially target these platelet mechanisms include: Crizanlizumab, a P-selectin neutralizing antibody; glycyrrhizin, an HMGB1 inhibitor with potent inhibitory effects on platelet-specific HMGB1 signalling; and ibrutinib, an orally administered selective and covalent BTK inhibitor. (+, blue arrows) stimulatory and (-, red arrows) inhibitory signalling effects. ASC, apoptosis-associated speck-like protein containing a C-terminal caspase recruitment domain; BTK, Bruton tyrosine kinase; CD62P, P-selectin; cGMP, cyclic guanosine monophosphate; DAMP, damage-associated molecular pattern molecule; GTP, guanosine-5⁰-triphosphate; HMGB1, high-mobility group box 1; IL-18, interleukin-18; IL-1b, interleukin-1b; MyD88, myeloid differentiation primary response gene 88; NLRP3, nucleotide-binding domain leucine rich repeat containing protein 3; sGC, soluble guanylate cyclase; SIRT, sirtuin; TLR4, toll-like receptor 4.

The NLRP3 inflammasome in platelets was initially found to be upregulated by the dengue virus (Hottz et al, 2013). We have demonstrated recently that activation of platelets by collagen or thrombin also upregulates platelet NLRP3 inflammasome activity (Murthy et al, 2017). Platelets constitutively express NLRP3, ASC and BTK. Activation of the NLRP3 inflammasome in platelets does not require an initial priming signal, such as the exposure to cytokines or a specific pattern recognition ligand as in immune cells (Hottz et al, 2015). HMGB1 (Chi et al, 2015) and haem (Dutra et al, 2014; Li et al, 2014) have been identified as regulatory triggers of NLRP3 signalling, and haem may (Li et al, 2014) or may not (Dutra et al, 2014) require purinergic P2X receptors. Cellfree haemoglobin may even synergistically increase HMGB1-mediated inflammatory responses, as shown in bone marrow-derived mouse macrophages (Lin et al, 2012).

Secretion of HMGB1 is controlled by the NLRP3 inflammasome in monocytes/macrophages (Willingham et al, 2009). Plasma derived from SCD patients at baseline and plasma from SCD mice induced TLR4 activity in a TLR4 reporter cell line in an HMGB1-dependent manner, further increased when plasma from SCD crisis patients was used (Xu et al, 2014). Other studies have shown that HMGB1 promotes inflammation in human microvascular endothelial cells (Fiuza et al, 2003) and promotes thickening of the pulmonary artery wall in a rat model of pulmonary arterial hypertension (SadamuraTakenaka et al, 2014). In a trauma/haemorrhagic shock mouse model, HMGB1-mediated activation of the TLR4/MyD88/sGC pathway in platelets induced thrombosis and the formation of neutrophil extracellular traps (NETs) (Vogel et al, 2015a). NETs formation and thrombosis are intrinsically linked with haemolysis (Ataga, 2009; Chen et al, 2014). The clinical relevance of HMGB1-mediated mechanisms in the context of haemolysis remains unknown. A schematic overview of the discussed pattern recognition receptor signalling mechanisms in platelets is presented in Fig 1.

Targeting HMGB1 and NLRP3 signalling in platelets

Hydroxycarbamide is still the main therapeutic option for SCD patients. A key component of its therapeutic effect is fetal haemoglobin induction in red blood cells and reduction of haemolysis (Bunn, 1997). Therapeutic efficacy of hydroxycarbamide also involves nitric oxide release from activation of sGC in erythroid progenitor cells (Cokic et al, 2008), reduction of neutrophil counts and activity (Saleh et al, 1999) and downregulation of proinflammatory cytokines (Lanaro et al, 2009). Targeting the adhesion molecule P-selectin, expressed by platelets and endothelial cells, with a blocking antibody has emerged as a new, promising strategy to significantly decrease vascular complications in SCD, as shown in SCD mice (Polanowska-Grabowska et al, 2010) and in a recent phase 2 trial with the humanised monoclonal P-selectin antibody crizanlizumab (Ataga et al, 2017). Crizanlizumab had an added effect in patients who were receiving concomitant hydroxycarbamide. The annual crisis rate was 32·1% lower with high-dose crizanlizumab than with placebo among patients receiving hydroxycarbamide.

P-selectin expression by platelets and platelet aggregation are critically regulated by the NLRP3 inflammasome (Murthy et al, 2017) and the HMGB1/TLR4/sGC (Vogel et al, 2015a) signalling pathway in platelets. HMGB1 and TLR4 are key players in promoting inflammation in SCD (Xu et al, 2014). Targeted ablation of HMGB1 and TLR4 in circulating platelets in transgenic mice promoted anti-thrombotic and antiinflammatory effects (Vogel et al, 2015a, 2016). The HMGB1 inhibitor Glycyrrhizin (Mollica et al, 2007), a triterpene glycoside extracted from liquorice root, exerted potent inhibitory effects on platelet HMGB1 signalling (Vogel et al, 2014, 2015a). Interfering with these pattern recognition receptor pathways in platelets might have significant effects on thrombosis and vascular inflammation in SCD via P-selectindependent and -independent mechanisms, and are currently under investigation.

The mitochondrial-enriched sirtuin deacetylases SIRT1 (Hwang et al, 2015) and SIRT3 (Zhao et al, 2016) have been identified as regulatory triggers that downregulate the proinflammatory effects of HMGB1 and NLRP3 signalling, respectively. Depletion or deletion of SIRT3 increased NLRP3 inflammasome activation in human macrophages and in murine kidney tissues during sepsis-induced acute kidney injury (Traba et al, 2015; Zhao et al, 2016). Platelets express functionally active SIRT1 and SIRT3 (Kumari et al, 2015), and interfering with sirtuins in platelets might affect platelet HMGB1/NLRP3 signalling.

BTK is another recently identified regulator of the NLRP3 inflammasome in immune cells (Liu et al, 2017) and platelets (Murthy et al, 2017), whose activation can be targeted with the US Food and Drug Administration-approved BTK inhibitor ibrutinib (PCI-32765) (Ito et al, 2015). Genetic ablation of BTK in platelets suppressed platelet activation, aggregation and thrombus formation, which was restored by the NLRP3 activator Nigericin, indicating a functionally relevant link between BTK and NLRP3 in platelets (Murthy et al, 2017). Thus, ibrutinib-mediated NLRP3 inhibition in platelets might be a novel therapeutic approach for targeting plateletmediated thromboinflammatory events during haemolysis. Interestingly, two SCD patients who developed chronic myeloid leukaemia (CML) reported marked reduction in the occurrence of SCD pain crises on treatment with the BCRABL1 tyrosine kinase inhibitor imatinib (Stankovic Stojanovic et al, 2011; Murphy et al, 2014). However, it is unknown if the potential effect of imatinib on vaso-occlusive crisis is due to inhibition of tyrosine kinases.

The clinical significance of platelet danger signalling mechanisms in haemolysis-driven inflammation and thrombosis is still poorly understood. Existing studies have reported promising data thus far, and further investigations in this field might identify valuable platelet-based therapeutic options for haemolytic diseases, such as SCD.