The role of thrombin in brain injury after hemorrhagic and ischemic stroke

Abstract

Thrombin is increased in the brain after hemorrhagic and ischemic stroke primarily due to the prothrombin entry from blood either with a hemorrhage or following blood-brain barrier disruption. Increasing evidence indicates that thrombin and its receptors (protease-activated receptors; PARs) play a major role in brain pathology following ischemic and hemorrhagic stroke (including intracerebral, intraventricular and subarachnoid hemorrhage). Thrombin and PARs affect brain injury via multiple mechanisms that can be detrimental or protective. The cleavage of prothrombin into thrombin is the key step of hemostasis and thrombosis which takes place in every stroke and subsequent brain injury. The extravascular effects and direct cellular interactions of thrombin are mediated by PARs (PAR-1, −3 and −4) and their downstream signaling in multiple brain cell types. Such effects include inducing blood-brain-barrier disruption, brain edema, neuroinflammation and neuronal death, although low thrombin concentrations can promote cell survival. Also, thrombin directly links the coagulation system to the immune system by activating interleukin-1α. Such effects of thrombin can result in both short-term brain injury and long-term functional deficits, making extravascular thrombin an understudied therapeutic target for stroke. This review examines the role of thrombin and PARs in brain injury following hemorrhagic and ischemic stroke and the potential treatment strategies which are complicated by their role in both hemostasis and brain.

Introduction

Thrombin, a serine protease, is generated by cleavage during activation of the coagulation cascade to provide hemostasis. Thrombin generation and the coagulation cascade has long been a target for preventing ischemic stroke (e.g. warfarin). More recently, direct thrombin inhibitors are being used clinically (e.g. dabigatran). However, thrombin has many actions outside those directly related to hemostasis that impact cerebrovascular disease, the subject of this review.

Prothrombin is predominantly produced by the liver, and circulates in the bloodstream until it is cleaved into the active form of thrombin by Factor Xa[1]. However, prothrombin can also be produced by neurons and astrocytes[2, 3]. The physiological functions of this brain-derived thrombin are mostly unknown, although there is evidence of its roles in brain development[4], synaptic transmission and plasticity[5, 6] as well as neuronal protection after ischemic brain injury[7]. The amount of thrombin in brain is relatively low in pathological conditions like neurodegenerative diseases[8]. It is, though, increased in both hemorrhagic and ischemic stroke[9–11], mostly due to prothrombin or thrombin entering the brain after a cerebral hemorrhage, blood-brain-barrier (BBB) breakdown and, to a lesser extent, from brain prothrombin synthesis (Figure 1)[12]. Experiments where brain prothrombin is genetically deleted would allow assessment of the precise contributions of vascular and extravascular prothrombin to increases in brain thrombin after stroke. Although thrombin is reported to be neuroprotective at a low concentration[13–15], many of in vivo [16–18] and in vitro[13, 19, 20] studies indicate that high concentration of thrombin plays a role in brain injury especially brain edema formation after stroke.

Figure 1.

Sources and sites of action of thrombin in brain after stroke. Firstly, a large amount of prothrombin can enter the brain through ruptured vessels after intracerebral hemorrhage (ICH), intraventricular hemorrhage (IVH), subarachnoid hemorrhage (SAH), or ischemic hemorrhagic transformation and cleaved to thrombin. Secondly, the BBB is disrupted after both hemorrhagic and ischemic stroke and the plasma (pro)thrombin can penetrate through the damaged BBB into the brain parenchyma. Lastly, there is increased prothrombin synthesis from neurons and astrocytes after stroke. Thrombin can have extracellular actions (particularly the conversion of fibrinogen to fibrin) but also has actions via protease-activated receptors (PARs) expressed on multiple cell types. BBB: blood-brain-barrier; CNS: central nervous system.

Thrombin is a key regulator in the primary hemostatic process, in which it activates platelets, and also in the secondary hemostasis, it mediates the conversion of fibrinogen to fibrin. Thus, thrombin contributes to thrombus formation that stops bleeding and hematoma formation after blood enters the brain parenchyma[21]. Both are critical in the pathology of hemorrhagic and ischemic stroke. Besides a central role in hemostasis, thrombin also has roles in brain cell death and survival as well as neuroinflammation, predominantly via the cellular protease-activated receptors (PARs) activation and downstream signaling pathways[22]. PARs are expressed throughout the brain, and the involvement of PARs in stroke has been confirmed by many studies with both protective and neurotoxic actions[23]. There are also extravascular effects of thrombin not related to PARs. For example, a recent study has indicated that thrombin could activate the immune system by directly cleaving pro- interleukin-1α to the active form (IL-1α)[24]. This novel finding suggests thrombin may be „a gas pedal‟ driving the innate immune system [25].

This review addresses the role of thrombin in brain injury after hemorrhagic and ischemic stroke with an emphasis on the mechanisms and pathways involved. For hemorrhagic stroke, we include studies on intracerebral hemorrhage (ICH), intraventricular hemorrhage (IVH) and subarachnoid hemorrhage (SAH).

To comprehensively review the existing research on the role of thrombin in the brain after hemorrhagic stroke and ischemic stroke, published articles in human and animal studies and reviews or meta-analyses were considered acceptable. The literature search was performed using PubMed, with no restriction on publishing date, and the last search was performed on September 17th, 2020. The search algorithms were combinations of the following terms: “thrombin”, “brain”, “injury”, “hemorrhage”, “intracerebral hemorrhage”, “intraventricular hemorrhage”, “subarachnoid hemorrhage”, “ischemia”, “stroke”, “thrombin receptors”. Articles were selected and reviewed if they addressed issues within the scope of this review. All selected articles reviewed in this article were published in peer-reviewed journals. Excluded were research articles written in a language other than English.

Thrombin and coagulation components after stroke

The cleavage of prothrombin following activation of the coagulation cascade is a key process during hemorrhage stroke and ischemic stroke. In turn, thrombin cleaves fibrinogen into insoluble fibrin to form thrombus. In both hemorrhagic and ischemic stroke, vascular endothelial cells are damaged[26]. In cerebral ischemia, intravascular activation of thrombin inducing microthrombus formation beside the initial injury site[11] and can increase the ischemic area. In ICH, the conversion of fibrinogen to fibrin by thrombin is essential to stop the initial bleed and limit hematoma expansion once the blood enters the parenchyma. The hematoma mass can compress normal brain tissue and increase intracranial pressure (ICP), potentially causing brain herniation and even death. Reducing the mass effect after intracerebral hemorrhage (ICH) has been a goal in many preclinical and clinical studies including the recent MISTIE III trial (NCT01827046), in which the investigators injected a thrombolytic agent, alteplase, through a catheter after minimally invasive surgery. The procedure was safe in ICH patients, however, whether fibrinolytic improve the neurofunction outcome is still uncertain[27].

In hemorrhagic stroke, fibrinogen is another essential part of the coagulation cascade. Fibrinogen is converted by thrombin to fibrin and both fibrinogen and fibrin may play important roles in brain injury. Evidence indicates that extravascular fibrinogen can induce strong inflammatory response including microglial activation via the CD11b/CD18 integrin receptor[28] and major histocompatibility complex (MHC) II-dependent antigen presentation[29]. Fibrin-induced activation increases chemokine secretion such as CXC-chemokine ligand10 (CXCL10), CCL2/monocyte chemoattractant protein 1 (MCP1) and interleukin 12, which further recruits T cells and macrophages[30]. Perivascular microglia cluster at fibrin deposition sites further demonstrating the phenomenon. Hence, fibrinogen may result in subsequent brain injury following thrombin production by potentiating inflammation. While thrombin and downstream components of the coagulation cascade may contribute to brain injury, they also play crucial roles in hemostasis limiting hematoma expansion, making them difficult therapeutic targets in hemorrhagic stroke.

Thrombin receptors in the brain

Besides its vital role in hemostasis and thrombosis, thrombin also plays a role in various non-hemostatic biological and pathophysiological processes, predominantly through the PAR activation [31]. First identified by Carney and Cunningham in 1978[32], PARs are a family of the G protein-coupled receptors that are expressed by a variety of cell types including platelets, endothelial cells, monocytes, neurons and astrocytes[33]. While PARs are activated by different proteases, up until now, three members of the PAR family have been found to be activated by thrombin, PAR-1, PAR-3 and PAR-4[34]. Among those receptors, PAR-1 and PAR-3 have a high-affinity for and can be activated at low thrombin concentrations (<5 nM), whereas PAR-4 can be activated at higher concentrations due to the lack of a hirudin-like sequence in the vicinity of the protease cleavage site[34, 35]. PARs are activated by proteolytic cleavage of the N-terminal extracellular domain rather than by ligand binding. The intracellular domains of PARs bind to Gα and Gβγ subunits, and receptor activation results in phosphorylation and release of either one of the Gα subunit families: Gα12/13, Gαi/o or Gαq[36]. Signaling following phosphorylation of Gαq subunit includes phospholipase C, mitogen-activated protein kinase (MAPK) and phosphokinase C (PKC). In contrast, activation of Gαi results in ERK/MAPK activation and inhibition of adenylyl cyclase which is critical in anti-inflammatory responses. The Gβγ subunit can activate phosphoinositide 3-kinase (PI3K) after Gαi activation. Subunit Gα12/13 may induce the activation of MAPK and subsequently small G proteins such as Rho GTPase for Rho-dependent responses[35]. Activation of PARs can regulate gene expression involved in many aspects of cell function and cell-to-cell interaction, including apoptotic and pro- and anti-inflammatory processes.

PARs are expressed throughout the brain parenchyma in neurons, microglia, astrocytes, and oligodendrocytes. PAR-1 is abundant in the pyramidal cell layers of the hippocampus while PAR-3 and PAR-4 are expressed in all cortical layers and the thalamus[37]. Amongst PARs, PAR-1 activation and associated signaling appear to be most important in cell modification and proliferation for astrocytes and microglia[38, 39]. Microglia, the main brain neuroimmune cells, sense environmental changes, respond to brain injury and have both neuroprotective and neurotoxic capacity [40]. In microglia, PAR activation facilitates the release of cyclooxygenase-2 (COX-2), inducible nitric oxide synthase (iNOS), tumor necrosis factor-α (TNF-α), interleukin 1α /1β (IL-1α /1β), IL-6, IL-12, and up-regulates surface CD40 expression thus associated extensive neuroinflammation[12]. Astrocytes also respond to thrombin through the release of mediators of the growth-regulated oncogene/cytokine-induced neutrophil chemoattractant family and have a neuroprotective effect[41, 42]. However, high doses of thrombin induce apoptosis in both cultured astrocytes and neurons via PAR-1 and protein RhoA[43]. PAR-1 activation also leads to astrogliosis via the MAPK pathway and may cause further neuron loss after brain injury[44]. Under pathological conditions, PARs have the capacity to mediate neuronal death and inflammation-induced neurodegeneration, while in other conditions PARs can have a neuroprotective role, indicating the importance of further studies on the pathophysiological functions of PARs.

Another important site of action for thrombin is the cerebral endothelium. Thrombin via PAR-1 increases intracellular Ca²⁺ in cerebral endothelial cells, induces F-actin stress fiber formation, disrupts the tight junctions linking those cells and increases BBB permeability[45]. Direct oral anticoagulants such as dabigatran can block PAR-1 cleavage by thrombin and limit thrombin-induced BBB disruption[46]. However, there is evidence that PAR-1 may also have protective actions at the cerebral endothelium. An analog of activated protein C (3K3A-APC) cleaves PAR-1 and activates cell survival signaling via Akt[47] That analog also decreases tissue plasminogen activator-induced cerebral hemorrhage preclinically[48] and is in a clinical trial to try and reduce reperfusion-induced cerebral hemorrhage in ischemic stroke (RHAPSODY, NCT02222714)[49]. These results indicate that certain PAR-1 agonists and antagonists may be able to shift the actions of PAR-1 and potentially thrombin in stroke by differentially activating specific signaling pathways. As well as the cerebral endothelium, this might occur in parenchymal cells with PAR-1.

Vorapaxar (Zontivity) is a potent oral inhibitor of PAR-1, it is a U.S. Food and Drug Administration approved novel antiplatelet drug to reduce the risk thrombotic events in patients with a history of myocardial infarction or peripheral artery disease[50]. In contrast to the marked effects in platelets, the function of this antagonist in the brain have not been well studied. Future research into finding potent brain-specific PAR-1 antagonists (and potentially agonists) may be very useful for a wild spectrum of brain disorders.

Concentration-dependent thrombin effects

Thrombin exhibits concentration-dependent effects in brain, with low concentrations being generally neuroprotective, while high concentrations can cause brain damage in hemorrhagic and ischemic brain diseases. In vitro studies indicated that thrombin protects rat primary hippocampal neurons from hypoglycemia, hypoxia, and growth supplement deprivation[13]. A similar neuroprotective effect was found in rat primary astrocyte cultures[19]. Also, thrombin pretreatment can protect cells from damage induced by a large dose of thrombin[51]. Those protective effects were attenuated by thrombin inhibitors, hirudin and protease nexin-1[52], as well as with PAR-1 knockout[15]. In addition, during ischemic stroke, neuronally-generated thrombin can promote astrocyte activation towards a neuroprotective phenotype via PAR-1[7].

In vivo data on protective effects of thrombin in ischemic and hemorrhagic stroke have also been reported. Intracerebral injection of a low dose of thrombin can reduce subsequent brain damaged caused by intracerebral hemorrhage or cerebral ischemia[14, 53]. We termed this phenomena thrombin preconditioning (TPC) or thrombin-induced brain tolerance. In contrast, intracerebral infusion of high concentration of thrombin induces brain edema and neuronal death[9, 17, 23]. There is also evidence that thrombin may mediate brain injury after cerebral ischemia. In the injured brain, such as in cerebral ischemia, lower concentrations of thrombin may cause cell death[23].

Although the mechanisms underlying the concentration-dependent effects of thrombin in hemorrhagic and ischemic stroke are not fully elucidated, PAR activation plays a crucial role in both the neuroprotective and neurotoxic processes. Besides, thrombin activates the hypoxia-inducible factor-1 (HIF-1) signaling pathway in vascular smooth muscle cells[54] and HIF-1α was accumulated in experimental ICH models[55] and TPC model. And the HIF-1α and its target genes may be involved in thrombin-induced brain protection[56] and promote angiogenesis after ICH[57]. An up-regulation of endogenous thrombin inhibitors[58, 59] and heat shock proteins (HSPs) especially HSP27and HSP32[14], along with the astrocyte-neuron interaction[7, 60] may also contribute to neuroprotection. In contrast, high concentration of thrombin causes neuron death by inducing neuronal edema, enhancing excitotoxicity, and neuroinflammatory response. Thus, thrombin has a concentration-dependent effect in the brain by both receptor and non-receptor manner following multiple different signaling pathways. Modulating extravascular thrombin activity could be a novel therapeutic strategy for cerebral hemorrhage and ischemia.

Thrombin can directly activate interleukin-1α

The coagulation system and the immune system have long been considered to be linked. At sites of injury, the coagulation cascade acts immediately with both intrinsic and extrinsic pathways converting prothrombin to thrombin. The inflammatory system was activated by injury or pathogen invasion. The immune system works relatively slowly with the activation of immune response cells and the release of cytokines. Among those, IL-1 is one of the most important proinflammatory cytokines in regulating the subsequent inflammatory response[22]. Currently, the IL-1 family is considered to have 11 members that play pleiotropic roles in inflammation and cancer. IL-1α and IL-1β were the first members to be described, and both of them act via the same receptor, IL-1R. During the last decades, many studies have focused on the function of IL-1β, and some recent studies indicate the importance of IL-1α in disease which has been relatively understudied[61]. A recent study identified a direct link between the immune system and the coagulation cascade by thrombin and IL-1α activation. Burzynski and colleagues identified a thrombin cleavage site in pro-IL-1α (p33 IL-1α)[24], which is highly conserved over a broad range of species, implying this link could be important. Thrombin is able to cleavage p33 IL-1α and generates mature p18 IL-1α and therefore promotes rapid inflammatory cell recruitment, including the major players in the inflammatory response like monocytes/macrophages. The authors also verified this hypothesis in a knock-in mouse model carrying an amino acid mutation (R114Q) in IL-1α. This makes IL-1α resistant to thrombin cleavage while retaining the capacity to be cleaved by calpain. Despite showing the same level of circulation hematological, coagulation, immune and inflammatory markers compared to wild type mice, the gene knock-in mice had much less mature IL-1α generation and much-reduced recruitment of neutrophils and monocytes at the site of skin wounds. Administration of thrombin inhibitor dabigatran in wide-type mice masked the difference, indicating a crucial role of thrombin in the generation of activated IL-1α. The authors also found some support of their hypothesis in clinical data obtained from patients of acute respiratory distress syndrome (ARDS).

While these results show that thrombin can promote inflammation via direct activation IL-1α, evidence also suggests IL-1α drives coagulation to form a circle of interaction between inflammation and coagulation[62]. Back in 1986, it was reported that IL-1 administration in rabbits upregulates tissue factor, induces fibrin formation while suppressing thrombomodulin[63]. Several different models support this finding by showing that IL-1α rapidly activates endothelial cells via NF-κB pathways and promotes thrombosis[62].

Compared to IL-1β, the role of IL-1α in stroke has received much less attention. Boutin et al.[64] found important crosstalk between the two in cerebral ischemia, while Thornton et al.[65]found that that IL-1α was an important driver of cerebrovascular inflammatory changes after ischemic stroke. The role of IL-1α hemorrhagic stroke has been neglected, including whether thrombin-induced IL-1α activation contributes to neuroinflammation and brain injury. It would be interesting to see if the current thrombin inhibitors will inhibit IL-1α activation by thrombin in the brain. Indeed, targeting cleavage of IL-1α might be an efficacious therapeutic target in many diseases, especially hemorrhagic stroke, since it might selectively interfere with the proinflammation effects of thrombin without affecting the coagulation cascade. Further investigation of thrombin-induced activation of IL-α in the setting of stroke should be done to address those issues.

Thrombin in intracerebral hemorrhage

In ICH, large amounts of prothrombin will enter the brain with the hemorrhage. Thus, human plasma prothrombin concentrations are ~1.5 μM[66] compared to the <5 nM thrombin required to activate PAR-1[34]. Gong et al.[67] measured thrombin activity in the ipsilateral basal ganglia after a striatal 100 μl blood injection in rats. They found thrombin activities 3.3 and 2.4 U/g at 1 and 24 hours after ICH compared to 0.1 U/g in saline injected rats (3.3 U/g corresponds to about 30 nM thrombin). It should be noted that the thrombin activity in perihematomal tissue after an ICH will depend not only on the generation of thrombin by the hematoma but also on the presence of members of the serpin family that inhibit thrombin (e.g. antithrombin within the hematoma and protease nexin I in astrocytes[68]. Also, while thrombin generation from prothrombin during hematoma formation is rapid, thrombin binds to fibrin[69] and this might be source of thrombin during hematoma resolution.

Much evidence indicates that thrombin plays an important role in inducing brain injury after ICH[9, 17, 70]. Thus, while thrombin facilitates hemostasis after ICH, it can be neurotoxic, by eliciting DNA fragmentation[71], and can induce BBB disruption[10] and vasogenic brain edema in experimental animal models[16] (Figure 2). Perihematomal edema after ICH was reduced by delayed administration of a specific thrombin inhibitor, argatroban, in rat models confirm the neurotoxicity of thrombin[72, 73]. Rat autologous whole blood and thrombin injection models induce marked behavioral deficits by day 1 with progressive recovery over 4 weeks. Co-injection of a thrombin inhibitor, hirudin, with whole blood can reduce the ICH-induced neurological deficits[74]. Iron is another hematoma derived factor with a role in post-hemorrhagic neuronal injury[75]. Thrombin may facilitate cellular iron uptake with upregulation of transferrin[76] and lipocalin-2[77] thus enhancing neuronal injury. These results indicate that thrombin neurotoxicity could have both short-term brain injury and long-term behavioral impacts after ICH, making thrombin a potential therapeutic target in ICH treatment, which has not advanced in decades except perhaps adequate blood pressure control[78].

Figure 2.

Mechanisms of thrombin-induced brain injury after intracerebral hemorrhage. ICH: intracerebral hemorrhage; BBB: blood-brain-barrier.

Importantly, intracerebral injection of thrombin induces a marked neuroinflammatory response[75, 79]. Intracerebral injection of thrombin in rodents activates microglia, upregulates many proinflammatory cytokines, such as TNF-α[80]and IL-1β [81], and initiates the complement cascade[82]. The source of IL-1 after brain injury is unclear, although it can be produced by many cell types, such as microglia, astrocytes and brain endothelial cells, and hematogenous macrophages and neutrophils[83]. IL-1, and especially IL-1β, is believed to take a part in ICH induced brain injury[84, 85]. Now it has been reported that IL-1α is directly activated by thrombin[24], this could be an important mechanism in thrombin-induced brain injury. We have previously reported that overexpressing interleukin-1 receptor antagonist (IL-1ra) using an adenovirus vector can attenuate the inflammatory reaction and brain edema following ICH and thrombin injection[86]. Thrombin-IL-1 interactions in ICH as a therapeutic target merit further investigation.

Thrombin receptors, particularly PAR-1, play a major role in thrombin-induced brain injury. In PAR-1 knockout (KO) mice, thrombin-induced brain lesions were markedly reduced and IL-1β levels were much lower compared to wild-type mice[87]. Another study from our group also demonstrated that after autologous blood injection into the basal ganglia, PAR-1 was upregulated in neurons and microglia, and ICH induced less brain swelling and neuronal death in PAR-1 KO mice. In addition, the loss of PAR-1 affected microglial polarization with less M1 polarization and reduced proinflammatory cytokines, again indicating a major role of PAR-1 in neuroinflammation after ICH[88]. Moreover, a PAR-1 mediated potentiation of neuronal NMDA receptor function[89] was observed in rodent hippocampus which may exacerbate glutamate-mediated cell death and possibly participate in post-hemorrhagic seizures.

While thrombin has a role in ICH-induced brain injury, whether thrombin inhibitors are potential therapeutics is very controversial because such inhibition might cause undesirable hematoma enlargement or rebleeding. We examined whether delayed (6 hours) systemic administration of argatroban did not increase hematoma size in a rat collagenase ICH model[72]. Lauer et al.[90] also found that the pretreatment with dabigatran etexilate, a new oral specific thrombin inhibitor, did not increase hematoma volume in both collagenase-induced and laser-induced ICH models, in contrast to warfarin. In contrast, another study suggested both warfarin and dabigatran exacerbated secondary hemorrhage in two kinds of mouse ICH model (intrastriatal liquid polymer injection and intrastriatal blood injection)[91]. Further investigation is needed for a better understanding of the pathophysiology of thrombin-induced brain injury after ICH to determine how and when to target it (or downstream pathways) without impacting hematoma growth.

Thrombin can also be beneficial in ICH. First of all, the hemostasis role of thrombin involves in the initial phase of ICH to stop bleeding and hematoma enlargement. In addition, studies have found that precondition with a low dose of thrombin reduces brain injury in ICH models or high dose thrombin injection[23, 92]. The mechanisms of thrombin-induced brain protection in hemorrhagic stroke may be related to the activation of PAR-1, the upregulation of heat shock proteins, and ceruloplasmin[92–94]. Thus, it seems that thrombin PAR activation and downstream signaling play an essential role in both neuroprotective and neurotoxic effects after ICH. How to manipulate the effect of thrombin and PAR-1 to enhance the former and diminish the latter is important for future treatment of ICH.

It is also possible that the effects of thrombin and PAR-1 in ICH may be time dependent. While there is considerable evidence that inflammation potentiates brain injury after ICH it is also involved in hematoma resolution and brain repair[95, 96]. Given the important role of thrombin and PAR-1 in inflammation, their temporal effects after ICH merit investigation.

Thrombin in intraventricular hemorrhage

Intraventricular hemorrhage affects both premature infants and elderly adults with high mortality and morbidity. In preterm neonates, IVH is typically due to germinal matrix hemorrhage (GMH), hemorrhage from a brain region of progenitor cells destined to be neurons and glia which is present until 34-week gestation in humans[97]. Post-hemorrhagic hydrocephalus (PHH) is a common co-morbidity with neonatal IVH patients and an independent predictor of poor prognosis which requires continuous care such as Cerebrospinal fluid (CSF) drainage[98]. In adults, IVH occurs primarily as a secondary consequence of ICH, when the ventricular ependyma is ruptured allowing blood to enter the ventricular system. It can also occur in SAH when blood moves from the subarachnoid space into the ventricles. IVH is associated with worse outcomes in both ICH and SAH patients[99, 100].

Prothrombin enters the ventricle with the IVH and activation of the coagulation cascade generates thrombin which then helps to form blood clots inside the ventricular system. Those clots impair CSF circulation which can lead to non-communication hydrocephalus. Researchers believe that intraventricular administration of thrombolytic agents such as tissue plasminogen might be beneficial[101]. The very recent CLEAR III trial (NCT00784134) tested the effect of intracerebroventricular thrombolytic agent alteplase in patients with IVH. Although it significantly reduced IVH volume and lowered mortality at 180 days, functional outcomes (as assessed by modified Rankin score) were, however, similar[102]. In neonatal IVH-GMH, multiple studies of intracerebroventricular fibrinolytic agents including alteplase, urokinase, streptokinase have been undertaken but, so far, no substantial beneficiary evidence concerning PHH has been found[103]. As preterm born infants are the primary victims of GMH-IVH, clinical studies have examined ways to target them specifically and prevent the development of the disease. Despite prolonged clotting time, which is not associated with IVH, the preterm infant appears to generate thrombin normally[104]. Clinical trials have not found evidence that prophylactic administration of the thrombin inhibitor heparin nor antithrombin is beneficial in high-risk neonates with respect to IVH incidence and severity[105, 106].

As well as hematoma formation, thrombin may play a variety of other crucial roles in IVH. Intraventricular thrombin injection causes significant hydrocephalus (Figure 3), ventricular wall damage, and periventricular BBB disruption. These effects were partially blocked by co-injection of a PAR-1 antagonist SCH79797, indicating that thrombin may contribute to hydrocephalus development after IVH via PAR-1[107]. Other studies have also shown that the inhibition of PAR-1 downstream Src family kinases attenuates brain edema, BBB disruption[108], and reduces hippocampal neuronal cell death and spatial memory deficits[109] after IVH. Thrombin was also found to play a role in neonatal periventricular hemorrhage[110]. In neonatal rat GMH models, forebrain thrombin levels are increased[111], and coadministration of a PAR-1 and PAR-4 inhibitor (P4pal10) could normalize increased mTOR expression, while early inhibition of mTOR improved long-term neurobehavior outcome[112]. Recently, a study using an oral thrombin antagonist, dabigatran, in a rat neonatal GMH model reported improved long-term morphological and neurofunctional outcomes as well as decreased PHH. Those effects tended to be reversed by direct PAR-1 stimulation. However, a PAR-1 inhibitor alone failed to achieve the same outcome as dabigatran indicating a PAR-1 independent mechanism in thrombin-induced brain injury after GMH[113].

Figure 3.

Mechanisms of thrombin-induced brain injury after intraventricular hemorrhage. IVH: intraventricular hemorrhage; CSF: cerebrospinal fluid.

Recently, Karimy and colleagues described a previously unrecognized contribution of inflammation-dependent CSF hypersecretion in post-hemorrhagic hydrocephalus. They found that IVH causes a Toll-like receptor 4 (TLR4)- and NF-κB-dependent inflammatory response at the choroid plexus (CP) epithelium that induced a ~3-fold increase CSF secretion through the activation of Ste20-type stress kinase (SPAK) and NKCC1 co-transport[114]. There have been prior attempts to inhibit ion and fluid transport by the CP in PHH, but the NKCC1 co-transporter is a new target and its apical (CSF-facing) distribution at the CP epithelium may explain the lack of effectiveness of furosemide. In a previous study, CSF secretion inhibitors such as acetazolamide (a carbonic anhydrase inhibitor) attenuated hydrocephalus following intraventricular thrombin injection[115]. Intraventricular injection of thrombin induces epiplexus (macrophage) cell activation at the CP as well as hydrocephalus suggesting an inflammatory mechanism[116]. Thrombin was also reported to disrupt VE-cadherin of CP and lead to hydrocephalus via PAR-1/p-Src/p-PAK1 pathway [117]. Those studies suggest potentially significant effects of thrombin at the CP that may be a future therapeutic target for treating PHH.

Since IVH mainly affects the two very end of the age spectrum, and age-related change is partially responsible for the difference in outcome seen in neonates and elderly patients. We previously observed more severe hydrocephalus and more intense CP inflammation in aged rats compared with young rats[118], while the effect of thrombin between different age groups were not established. Therefore, the investigation concerning age difference in the effect of thrombin may direct an individualized therapeutic strategy for IVH.

Thrombin in subarachnoid hemorrhage

Thrombin levels and fibrinolytic components are increased in the CSF of patients with SAH[119, 120]. As with ICH, prothrombin enters the brain with the initial bleed in SAH and result in neurological deficits (Figure 4). However, one of the primary pathological changes following SAH is increased microvascular permeability between endothelial cells and it has been suggested BBB disruption in both the early stage and late stage after SAH[121, 122] leads to prothrombin entry from plasma into the brain. BBB breakdown was also thought to directly contribute to brain edema, one potentially fatal consequence of SAH[123]. Earlier, Sugawara and et al.[124] reported that BBB disruption, as well as brain edema, cell death, neuronal inflammation, were attenuated by intraperitoneal administration of argatroban in a rat SAH model. They also found improved behavioral tests with the thrombin inhibitor. Their further investigation indicated thrombin might induce the BBB disruption via the PAR-1/c-Src/p21-PAK-1 pathway which could be partially alleviated by PAR-1 antagonist[125].

Figure 4.

Mechanisms of thrombin-induced brain injury after subarachnoid hemorrhage. SAH: subarachnoid hemorrhage; CP: choroid plexus; BBB: blood-brain-barrier.

Another potential lethal pathological change of SAH is generalized cerebral vasospasm leading to delayed cerebral ischemia[126, 127]. The activation of serine protease cascade and/or thrombin played an important part in the development of cerebral vasospasm owing to the upregulation of PAR-1 and impaired receptor desensitization[128] and thrombin-induced increased expression of a principal vascular contractile agent platelet-derived growth factor (PDGF)[129] by endothelial cells, macrophages and platelets[130] which might contribute to the chronic vasoconstriction. Microvascular thrombi formation has been found in autopsy studies in the brain of SAH patients[131, 132] and experimental animal models of SAH[133, 134]. Microthrombi is defined as histological finding of platelet plugs with fibrinogen/fibrin deposition with the arterioles and capillaries of brain[135], which begin to form just 10 min after SAH[136]. By 24h after rats SAH induction, ventricular dilation and white matter injury are found on MRI[134]. Both cerebral vasospasm and microthrombi formation contribute to delayed cerebral ischemia after SAH and neurological deficits [137, 138]. A recent study revealed a correlation of microthrombi formation with brain infarction and delayed neurological deficits after SAH in a mice model[139].

Hydrocephalus is one of the most common complications and reasons for readmission after SAH, according to previous reports, hydrocephalus was diagnosed in 40.4% in aneurysmal SAH patients undergoing surgical clipping or endovascular coiling[140, 141]. About 44–63% of rats develop hydrocephalus after experimental SAH induced by vascular perforation[142, 134]. Obstruction of CSF circulation pathway by blood clots and brain tissue factor may also related to pathologies such as hydrocephalus[143]. Macrophage/Microglia activation also occurs after SAH in the CP[116], CSF[144, 145] and brain tissue. We found that epiplexus macrophage activation and hydrocephalus after SAH could be mimicked by intraventricular thrombin injection[116]. Therefore, thrombin generated during the coagulation process or pass through the disrupted BBB after SAH might participate in immune cell activation at the CP and, thus, induce hydrocephalus.

The promotion of vasospasm, microthrombi formation and neuroinflammation by thrombin make it a potential therapeutic target in SAH and a thrombin inhibitor ameliorated vasoconstriction and suppressed MAPK expression in the vascular wall after SAH[146]. Moreover, there is pathological and radiographical evidence that argatroban reduces SAH-related vasospasm[147, 148], but clinical efficacy is yet to be determined. A rat experiment SAH study reported that administration of low-dose intravenous unfractionated heparin (LDIVH) significantly decreased neuroinflammation and neuronal apoptosis[149]. Retrospective studies of SAH patients using subcutaneous heparin for deep venous thrombosis prophylaxis showed a significant decrease in delayed cerebral ischemic[150] and improved cognitive outcome[151]. Previously, several clinical trials have investigated the effect of anticoagulation agents such as enoxaparin and heparin for aneurysmal SAH patients with mixed results[152, 153]. Currently, there is an on-going phase II randomized clinical trial examining the efficiency of LDIVH treatment in aneurysmal SAH patients (ASTROH trial, NCT02501434).

According to recent epidemiological studies, the risk of SAH is 45% higher in women than in men for unknown reasons and mortality rate is also greater in female patients, especially post-menopause[154, 155]. Animal SAH studies have also exhibited a worse outcome in females than males[134, 139]. These sex differences in SAH have not been extensively studied and results of different animal studies investigating sex steroids as a potential therapeutic target have produced conflicting results[156]. The relative effect of thrombin in males and females has not been well studied nor has it been studied in different age groups after SAH. Thus, future research is needed for a better understanding of sex and age differences in SAH.

Thrombin in brain ischemia

Intravascular thrombin and the coagulation cascade have long been a therapeutic target in ischemic stroke, primarily in relation to stroke prevention with warfarin, platelet inhibitors and now direct thrombin inhibitors (Table 1). However, increased brain thrombin levels have been detected in the infarct area after ischemia due to both BBB breakdown and prothrombin entry and brain prothrombin synthesis[157, 158]. Multiple groups have investigated the role of thrombin in brain ischemia using both in vitro and in vivo models. Thrombin at high concentrations activates microglia and induces astrocyte and neuronal death, while low concentrations are neuroprotective in brain ischemia[23].

Table 1.

Clinically Available Thrombin Inhibitors in Acute Stroke

Mechanism of Action	Agent(s)	Related Clinical Studies	Comments
Bind to Antithrombin Inactivate Thrombin and Factor Xa	Heparin	Whiteley et al. [177]	Limited Use in Acute Ischemic Stroke or TIA for Selected Patients*[179, 180]
	LMW Heparin	Sandercock et al.[178]
		Simard et al.[150]	May Be Safe and Beneficial for Post-Procedure SAH Patients
Direct Thrombin Inhibitor	Argatroban (Parenteral)	Sandercock et al.[178]	Usefulness in Acute Stoke Not Established
		Barreto et al. (ARTSS trial)[181, 172]	May Increase Hemorrhagic Transformation[182]
	Dabigatran (Oral)	Kate et al.[183]
		Ken S et al.[182]
Platelet PAR-1 Antagonist	Vorapaxar	Bonaca et al. [50]	Usefulness in Acute Stoke Not Established
	Atopaxar**		Potentially Reduces Incidence of Stroke [50]

^*Only for select patients with acute cardioembolic ischemic stroke or TIA due to intracardiac thrombus in the left ventricle or thrombus associated with mechanical or native heart valves who are at high short-term risk for recurrent stroke

^**Ongoing Clinical Study, Not FDA Approved

Thrombin may have adverse effects in cerebral ischemia by several pathways: Firstly, it can induce microthrombosis formation in small distal vessels which may exacerbate the ischemia after the initial arterial occlusion. Thrombin also alters brain pericyte function through two independent pathways via PAR-1 activation to release matrix metalloproteinase (MMP)-9, the PKCθ-Akt pathway and the PKCδ-ERK1/2 pathway[159]. There is also remodeling of the endothelial junctional structure and, thus, vascular hyperpermeability causing brain edema and cell death[11, 26, 160]. Moreover, in vitro studies demonstrate thrombin causes synaptic dysfunction by inducing ischemic long-term potentiation (iLTP) of synaptic transmission through the activation of N-Methyl-D-aspartate receptors (NMDAR) via PAR-1[5, 161]. Excessive neuronal stimulation results in neuronal death. Finally, thrombin infiltrates into parenchymal tissues and causes direct cytotoxic effects on both glia cells and neurons[43, 162]. PAR-1 appears to be the most important thrombin receptor during the thrombin-induced neuronal damage caused by ischemia[157, 163]. Indeed, the neurotoxic effects of thrombin have been shown to be attenuated by PAR-1 deletion[164, 165].

Intravenous administration of thrombin inhibitor hirudin increased hippocampal CA1 neuronal survival after global cerebral ischemia in gerbils[13]. Argatroban, a specific thrombin inhibitor, reduced infarct volume, improved neurological outcome and extended therapeutic window for recombinant tissue-type plasminogen activator (r-tPA) without increasing hemorrhage rates in rat ischemia models[166, 167], and reduced neurodegeneration and brain edema following bilateral common carotid artery occlusion and reperfusion in gerbils[168]. Dabigatran etexilate reduced endothelial permeability through inhibition of thrombin-induced cytoskeleton reorganization[169, 170], which may explain its favorable benefit-to-risk profile compared to warfarin for preventing ischemic stroke in patients with atrial fibrillation in a clinical trial (NCT00262600)[171] by lowering the risk of intracerebral hemorrhage. Also, a randomized exploratory clinical trial was conducted to test the safety and the outcome with adjunctive argatroban, administered in tPA-treated ischemic stroke patients (NCT01464788)[172]. Results showed that combination therapy is not associated with an increased risk of ICH. However, the study was terminated due to the beneficial results of thrombectomy clinical trials.

Despite the brain injury inducing effects of thrombin, some studies have demonstrated a neuroprotective role after brain ischemia[23]. Prior intracerebral injection of low-dose thrombin can induce brain tolerance to subsequent ischemic attack[14, 51]. Interestingly, the neuroprotective effect of thrombin also seems to be related to PAR-1[173] [15]. A recent study demonstrated that neuron-generated thrombin, released during ischemia, acts via PAR-1 activating astrocytes and paracrine neuroprotection[7]. The pleiotropic effects of thrombin in ischemic brain disease are likely mediated by distinct signaling pathways dependent on thrombin levels, the cell types involved and, potentially, timing. PAR-1 cleavage with an activated protein C analog, 3K3A-APC, can induce vasculo- and neuroprotection by activating cell Akt survival [47, 174] rather than injurious pathways. Further investigation into the thrombin effects in ischemic stroke can focus on the role of PAR-1 and downstream pathways in different cell types and how they interact with different injury/protection pathways.

Sex differences were also observed in ischemic stroke patients. Women present with higher lifetime risk and higher mortality and disability from ischemic stroke than men, especially post-menopause[175, 176]. However, the mechanisms underline the sex differences are poorly understood other than the involvement of sex steroids. Studies on the effects of thrombin in different sexes and ages are needed given its role in both coagulation and neuroinflammation.

Conclusions and Future Directions

Thrombin is not only an important component in the coagulation system but also has a major impact on the brain after both hemorrhagic and ischemic stroke. While intravascular thrombin generation is extremely important in stroke, prothrombin can also enter the CNS following hemorrhage and BBB disruption or be generated by neural cells and then converted to thrombin. While low thrombin concentrations and thrombin preconditioning potentially protect the brain, high thrombin concentrations result in brain injury. The mechanisms of thrombin-mediated brain injury involve multiple pathways including thrombus formation, activation of PARs and downstream signaling pathways, and non-receptor mediated inflammatory processes, including direct activation of IL-1α. So far, little is known about the precise source and the exact function of thrombin in the brain. Further investigation targeting extravascular thrombin and its receptors as treatment strategies are warranted, especially in hemorrhagic stroke. In ICH, potential inhibition on the thrombin receptors might be a future research target to minimize the proinflammatory property of thrombin but preserve its hemostasis ability. In terms of IVH and SAH, the role of thrombin in CP inflammation and CP epithelial hypersecretion merit more study, particularly in relation to hydrocephalus development. The very early occurrence of microthrombosis after SAH and the prolonged neurological deficits call for more investigation into the role of thrombin. We suspect agents targeting thrombin and its receptors may be beneficial to the SAH patients once aneurysms are repaired with surgical clipping or endovascular coiling.