Understanding the Functions of Endogenous and Exogenous Tissue-Type Plasminogen Activator During Stroke

Alteplase (Actilyse or Activase) is the gold standard acute treatment of ischemic stroke1 and receives attention for hematoma resolution in hemorrhagic stroke. The active principle of Alteplase is the recombinant form of an endogenous protease, tissue-type plasminogen activator (tPA). Its intravascular thrombolytic activity is well known, but less are its multifaceted functions in the central nervous system (CNS). Endogenous tPA is not only released in the blood by endothelial cells but also expressed by many cells within the brain parenchyma (online-only Data Supplement), and it can act on all cell types of the brain virtually. Endogenous tPA has been involved in an ever-increasing number of brain functions, of which several are highly relevant during and after stroke. Importantly, knowledge on the mechanisms of action of endogenous tPA may hold true for recombinant tPA.

In this review, we provide an up-to-date overview of the current knowledge on the enzymatic or cytokine-like effects of action of tPA in the CNS, its various molecular substrates or receptors, focusing on the processes occurring during and after ischemic or hemorrhagic stroke, including excitotoxicity, apoptosis, blood–brain barrier breakdown, inflammation, axonal damage, and demyelination.

tPA is a mosaic protease of 527 amino acids consisting of 5 distinct modules: a Finger domain, an epidermal growth factor (EGF)–like domain, 2 kringle domains (K1 and K2), and a serine protease proteolytic domain. Through these domains, tPA can interact with a variety of binding proteins and receptors in the brain parenchyma, thus extending its functions above the conversion of plasminogen into plasmin (Figure 1).



Figure 1. 
Main cellular and molecular targets of tissue-type plasminogen activator in the central nervous system. BBB indicates blood–brain barrier, EGFR, epidermal growth factor receptor; LRP, low-density lipoprotein receptor–related protein; NMDAR, N-methyl-d-aspartate receptors; and PDGFR, platelet-derived growth factor receptor.



tPA is not fibrinolytic by itself. …

A lteplase (Actilyse or Activase) is the gold standard acute treatment of ischemic stroke 1 and receives attention for hematoma resolution in hemorrhagic stroke. The active principle of Alteplase is the recombinant form of an endogenous protease, tissue-type plasminogen activator (tPA). Its intravascular thrombolytic activity is well known, but less are its multifaceted functions in the central nervous system (CNS). Endogenous tPA is not only released in the blood by endothelial cells but also expressed by many cells within the brain parenchyma ( online-only Data Supplement), and it can act on all cell types of the brain virtually. Endogenous tPA has been involved in an ever-increasing number of brain functions, of which several are highly relevant during and after stroke. Importantly, knowledge on the mechanisms of action of endogenous tPA may hold true for recombinant tPA.
In this review, we provide an up-to-date overview of the current knowledge on the enzymatic or cytokine-like effects of action of tPA in the CNS, its various molecular substrates or receptors, focusing on the processes occurring during and after ischemic or hemorrhagic stroke, including excitotoxicity, apoptosis, blood-brain barrier breakdown, inflammation, axonal damage, and demyelination.
tPA Is More Than a Fibrinolytic Enzyme tPA is a mosaic protease of 527 amino acids consisting of 5 distinct modules: a Finger domain, an epidermal growth factor (EGF)-like domain, 2 kringle domains (K1 and K2), and a serine protease proteolytic domain. Through these domains, tPA can interact with a variety of binding proteins and receptors in the brain parenchyma, thus extending its functions above the conversion of plasminogen into plasmin ( Figure 1). tPA is not fibrinolytic by itself. To promote fibrinolysis, tPA activates fibrin-bound plasminogen into plasmin. The binding of plasminogen to fibrin is a necessary step to change its closed conformation to an open form, allowing its cleavage by tPA. 2 The Finger domain of tPA is involved in its binding to fibrin and is necessary to promote fibrinolytic activity at low plasminogen activator concentrations. 3 In the brain, the Finger domain, by interacting with low-density lipoprotein receptor-related proteins (LRPs), supports blood-brain barrier (BBB) crossing, 4 astrocytic clearance, 5 or microglial activation. 6,7 The EGF-like domain shows homology with EGF and is accordingly thought to mediate the trophic and mitogenic functions of tPA in the brain. [8][9][10][11] O-glycosylation of the EGFlike domain has been shown to promote the hepatic recapture of tPA. 12 The role of the K1 domain in the brain is poorly investigated, but its high mannose-type glycosylation could be involved in uptake processes as shown on liver endothelial cells via mannose receptors. 13 Because of the presence of loops and of a lysine binding site, the K2 domain binds to various proteins in the blood and in the brain parenchyma, including the platelet-derived growth factor-CC, 14 N-methyl-D-aspartate receptors (NMDARs) 15 or high mobility group box-1 protein. 16 In the K2 domain, the Asn184 glycosylation site segregates 2 tPA variants (type I is fully glycosylated, whereas type II lacks glycosylation), 17 which may exert different functions although this remains to be investigated. Interestingly, desmoteplase, the thrombolytic agent used in Desmoteplase in Acute Ischemic Stroke (DIAS) clinical trials, derived from a bat salivary gland, is closely related to tPA but lacks a K2 domain 18 and is accordingly devoid of toxic actions, likely because of the inability to interact with NMDARs. 19,20 Also, the specific amino acids involved in the backbone structure of the lysine binding site are also key in the ability of tPA to promote excitotoxicity on neurons. 21 The catalytic domain supports the functions of tPA dependent on its protease activity. A specific cleavage at the Arg275-Ile276 peptide bound converts the single-chain form to the 2-chain form, maintained by a disulphide bound. In contrast to the other members of the chymotrypsin family of serine proteases, tPA is active in both its single-and 2-chain forms. However, some functions of tPA are specific to 1 form: for example, only the singlechain form of tPA can promote NMDAR-driven neurotoxicity. 22 deprivation and delay to reperfusion. 25 In endothelial cells, tPA induces the synthesis of metalloproteinases, matrix metalloproteinase-9 and matrix metalloproteinase-3, which in turn contribute to enhanced BBB permeability and intracranial bleeding. 26,27 In perivascular astrocytes, tPA induces the shedding of LRPs, 28 which activates nuclear factor-κB 29 and Akt 30 pathways leading to the expression of matrix metalloproteinase-9, 31 which finally promotes the detachment of astrocyte end-feet processes. 28 tPA can also alter the BBB by activating the platelet-derived growth factor-CC pathway on perivascular astrocytes 14,32 (Figures 1 and 2). tPA also contributes to cleavage of the monocyte chemoattractant protein 1 (officially known as C-C motif chemokine 2), leading to a disruption of tight-junctions and thus to BBB leakage. 33 It is important to note that the role of plasmin generation in the effect of tPA on the BBB remains controversial, as recently deeply reviewed. 25 All these mechanisms have mainly been associated with BBB leakage and which of them is/are involved in hemorrhagic transformation remains unclear. The extent of ischemia is likely critical to determine whether BBB leakage will convert to bleeding. The delay of tPA administration is also suspected to influence the occurrence of bleeding but this lacks clinical evidence.

Effects of tPA on NMDAR-Mediated Signaling and Subsequent Neuronal Outcome
Several studies have shown that inhibitors of tPA in the CNS can protect neurons against toxicity induced by the overactivation of NMDARs. [34][35][36] This suggested that tPA can promote neurotoxicity by acting on NMDARs. The mechanism of action of tPA on NMDARs has been a subject of debate 37,38 ( Figures 1 and 3), turning around 3 main questions: Is it proteolytic or nonproteolytic? Does it require plasmin generation? Is a coreceptor involved?
It is now established that the enhancement of NMDAR signaling by tPA depends on its proteolytic activity. 22,39,40 Both plasmin-dependent and plasmin-independent mechanisms have been reported to sustain the potentiation of NMDAR signaling by tPA, 37,41,42 but the most recent studies agree that it can occur independently of plasminogen activation. 22,39,43,44 LRP can act as a coreceptor for tPA. Indeed, tPA would act on a nonplasminogen substrate, engaging LRP receptors, which in turn would enhance Ca 2+ downstream of NMDARs. 45 Whether tPA requires LRP to enhance NMDAR signals could depend on the type of neurons (hippocampal versus cortical), their state of maturation, the kinetic of tPA application or on Figure 1. Main cellular and molecular targets of tissue-type plasminogen activator in the central nervous system. BBB indicates blood-brain barrier, EGFR, epidermal growth factor receptor; LRP, low-density lipoprotein receptor-related protein; NMDAR, N-methyl-d-aspartate receptors; and PDGFR, platelet-derived growth factor receptor.

Figure 2.
Mechanisms of tissue-type plasminogen activator (tPA) action on blood-brain barrier (BBB) breakdown. In endothelial cells, tPA activates nuclear factor (NF)-κB signaling through low-density lipoprotein receptor-related protein (LRP) to induce the synthesis of metalloproteinases matrix metalloproteinase (MMP)-9 and MMP-3, which in turn contribute to enhanced BBB permeability and intracranial bleeding. In perivascular astrocytes, tPA induces the shedding of LRPs, which activates NF-κB and Akt pathways, induces the expression of MMP-9 by perivascular astrocytes, which finally promotes the detachment of astrocyte end-feet processes. tPA effect on perivascular astrocytes can also be mediated by activating the platelet-derived growth factor (PDGF) pathway.  the action of astrocytes. 39,44,46 Interestingly, in PC12 and N2a neuron-like cells, tPA may signal through a complex containing NMDAR, LPR1, and Trk receptors. 46 tPA interacts with the GluN1 subunit of NMDARs, and as already mentioned, this interaction involves the lysine binding site of tPA K2 domain. 15,21 Accordingly, desmoteplase, a variant of tPA lacking the K2 domain, does not interact with NMDAR and does not promote NMDAR signaling in cortical cultures. [18][19][20] Several groups reported that the cleavage of the amino-terminal domain of GluN1 subunit is necessary for enhancement of NMDAR signaling by tPA (Figure 3), 47,48 whereas others did not detect tPA-dependent cleavage of GluN1, despite enhancement of NMDAR function by exogenous tPA in cortical cultures. 45 Plasmin generated by tPA has also been reported to cleave NMDARs, specifically the GluN2 subunit ( Figure 3) at 2 sites: Lys317 on GluN2A, which relieves Zn 2+ inhibition (Zn 2+ is a negative allosteric modulator of NMDAR) and thereby increases NMDAR function, 49 and Arg67 on GluN2B, which increases sensitivity of the NMDAR to the coagonist of NMDARs, glycine. 50 A recent study showed that only single-chain tPA can promote NMDAR signaling and neurotoxicity in cortical neurons, as well as late phase of long-term potentiation (LTP) in hippocampal neurons. 22 Of note, Actilyse is a mix of single-chain recombinant tPA (90%) and 2-chain recombinant tPA (10%). 22 Studies in transgenic mice overexpressing tPA in neurons (T4 transgenic mice) suggested that tPA can also have neuroprotective effects 9,44 through a mechanism that is also dependent on the activation of NMDARs and independent on plasmin. The fact that tPA induces toxic or protective effects in neurons could depend on the different subtypes of GluN subunits involved, as well as on their location (synaptic versus extrasynaptic). In fact, exogenous tPA promotes neurotoxicity on cortical neurons by activating extrasynaptic GluN2D-containing NMDARs 51,52 but leads to a neuroprotective effect by activating synaptic GluN2A-containing NMDARs. 53 tPA may also have opposite effects depending on its concentration, with low concentrations of tPA being protective, 53 and higher concentrations being deleterious, 22,46 which suggests that there are multiple receptors for tPA with distinct affinities or coreceptors.
Interestingly, tPA can enhance NMDAR activity not only in neurons but also in brain endothelial cells. 54,55 tPA also seems to play role in neurovascular coupling, an effect involving NMDARs, but the actual mechanisms involved need to be investigated further. 56,57

Effects of tPA on Apoptosis
Several in vitro studies reported antiapoptotic effects of tPA on neurons 10,28,58 and oligodendrocytes progenitors. 8 Despite the heterogeneity of the toxic paradigms used in these different studies, they all show that this effect of tPA occurs independently of its proteolytic activity, through the so-called cytokineor growth factor-like effect. A consensus also emerges around the necessary activation of PI3K/Akt pathway for neuroprotection by tPA. 8,10,59 Two candidates have been proposed as the receptors mediating the antiapoptotic effects of tPA: annexin II and EGF receptor (Figure 2). Annexin II mediates these nonproteolytic effects of tPA in neurons (but the participation of EGF receptor was not addressed), 59 whereas EGF receptor does so in oligodendrocytes progenitors (but the participation of annexin II was not addressed). 8 Through this antiapoptotic effect on oligodendrocytes, tPA sustains the protection of the white matter after experimental ischemic stroke. 8 In contrast to its reported antiapoptotic effects, tPA was shown to play proapoptotic effects on neurons, 60 an effect possibly involving NMDARs and counteracted by the anticoagulant serine protease, activated protein C. 61 However, the mechanisms through which this antiapoptotic effect occurs remain rather controversial. [60][61][62] The state of maturation of neurons critically influences whether the effects of tPA are neurotoxic or neuroprotective. Indeed, during cortical development, tPA is toxic to neurons in deep layers (which are mature) through an NMDAR-dependent effect, whereas it protects neurons against apoptosis in superficial layers (which are immature) through an EGF receptor-dependent effect. 63 Thus, it seems that the target of the antiapoptotic effects of tPA is generally immature cells, such as developing neurons or oligodendrocyte progenitors.

tPA and Cerebral Inflammatory Processes
tPA can promote microglial activation (Figure 1), independently of plasminogen activation, and via its finger domain interacting with microglial LRP and annexin II. 6,7,64 A study using cell-specific tPA knockout mice challenged with kainate injection in the hippocampus revealed a regulatory loop in which neuron-derived tPA activates microglia, which in turn produces additional tPA. This microglia-derived tPA has autocrine, self-proliferative effects on microglia, and paracrine, neurotoxic effects. 65 Data also indicate that annexin II may act in concert with the lectin galectin-1, which has previously been described as a receptor for tPA outside the CNS, 66 to induce microglial activation by activating extracellular signalregulated kinases 1/2 and c-Jun N-terminal kinase pathways and inflammatory responses by activating the Akt pathway. 7 Interestingly, in the context of ischemia-reperfusion, plasmin 67 and tPA 68 have been reported to contribute to neutrophil and leukocyte infiltrations of reperfused tissue. Interestingly, tPA was suggested to have a biphasic action: first, via plasmin and matrix metalloproteinase, the activation would promote neutrophil transmigration and BBB injuries, and then would favor tPA extravasation in the brain, amplifying neutrophil recruitment via nonproteolytic activation of mast cells and lipid mediator release. 68 Similarly, in in vitro models of BBB, tPA was reported to contribute to the adhesion and transmigration of monocytes and T lymphocytes, an effect prevented by blockage of the tPA-NMDAR interaction or of the LRP. 54,69 Dare are sparse, and additional studies are warranted to determine the actual influence of tPA on brain inflammatory processes.

Role of tPA in Axonal Damage and Regeneration
Fibrin deposits within the CNS can lead to axonal damage and limited capacities of axonal regeneration. tPA might prevent both side effects of fibrin deposits. 70 In fact, tPA is a key regulator of cellular migration and extension of cellular processes during development, by promoting the degradation of extracellular matrix (ECM) and cell adhesions. Similarly, tPA can promote the formation of new axonal varicosities during regenerative processes after axon degeneration. 71 In general, the activation of plasmin by tPA results in the degradation of the ECM toward a more permissive environment for axonal growth. More specifically, in models of spinal cord injury, tPA action on axonal regeneration was shown to involve the degradation of chondroitin sulfate proteoglycans, 72,73 a set of ECM proteins with inhibitory action on axon regrowth. Accordingly, tPA activates the chondroitin sulfate proteoglycan-degrading protease, a disintegrin and metalloprotease with thrombospondin domains-4, thus promoting axonal growth and functional recovery. 74 These effects of tPA could be amplified by the fact that tPA, through its proteolytic activity, can facilitate the migration of macrophages, which in turn may locally release proteases that can degrade ECM components. 75 Unfortunately, because of a low percentage of white matter in the brain of a rodent, these pathways remain poorly investigated in stroke models.

tPA Is Involved in Neuronal Plasticity
Via its ability to activate plasminogen and to regulate metabolism of the ECM, particularly at the growth cone, 76 tPA is surely a crucial regulator of neuronal growth and motility. Indeed, tPA-dependent plasminogen activation was first demonstrated in neuronal cell lines, suggesting that it may have roles in neuronal growth. [77][78][79] tPA-mediated neuronal plasticity thus, at least in part, occurs via facilitation of structural changes by degradation of the ECM to establish new or reinforce existing synapses. [80][81][82] These effects on neuronal plasticity underline the link between tPA and LTP. Indeed, experimentally induced LTP rapidly induces the transcription of tPA in hippocampal neurons via an NMDAR-dependent mechanism. 83 Accordingly, the treatment of hippocampal slices with tPA enhances the late phase of LTP 84 . Furthermore, transgenic mice overexpressing tPA show an enhanced and prolonged hippocampal LTP, 85 whereas tPA knockout mice show reduced hippocampal 86 and corticostriatal 87 LTP. There are several other mechanisms by which tPA can promote synaptic plasticity. These include binding to and the activation of LRP, leading to protein kinase A activation. 88 Also, recent data show a link between tPA and the neurotrophin brain-derived neurotrophic factor (BDNF) in hippocampal LTP. In hippocampal neurons, tPA, plasminogen, and pro-BDNF are copackaged in dense-core granules that can be efficiently recruited into active spines and can undergo activity-dependent release. 89 Plasmin, generated by tPA, is necessary for late phase of LTP and cleaves pro-BDNF to its active form mBDNF 90 that is known to promote the conversion of early to late hippocampal LTP. 91 In agreement with this, a study 92 demonstrated that high-frequency stimulation of hippocampal neurons leads to neuronal secretion of tPA, concomitant with an increase in extracellular mBDNF. tPA, in particular single-chain tPA, could also promote LTP via its ability to enhance NMDAR signaling. 21 To sum up, tPA (of neuronal and astrocytic origin) 93 might promote poststroke recovery, via both plasmin-dependent effects (ECM degradation and BDNF activation) and plasminindependent effects (activation of LRP-induced pathways and enhancement of NMDAR signaling).

Conclusions and Prospects
Research of the past 20 years has changed our image of tPA from a protease with vascular functions to a protease/cytokine with key roles in the brain parenchyma. Compelling evidence shows that tPA exerts multiple and sometimes opposite effects in the CNS, depending on the target cell type and the surrounding environment. Recent advances have explained this variety of effects by the different mechanisms of action of tPA, which are supported by its 5 functional domains and their interaction with their respective binding partners. In animal models of stroke, tPA is mainly reported to display deleterious effects, by promoting excitotoxicity and by enhancing BBB permeability. By contrast, in addition to fibrinolysis, tPA could also have beneficial effects after stroke, as it was reported to display oligotrophic and neurotrophic functions and to promote axonal regeneration and brain plasticity.
Several adjunctive therapies have been investigated in preclinical models to counteract the possible deleterious effects of tPA after stroke. For instance, thanks to the everincreasing knowledge on how tPA might be noxious, such strategies include among other, coadministration of tPA with an antibody targeting the effect of tPA on NMDAR signaling, 94 with annexin-2 to prevent BBB damages, 95 with activated protein C 96 or progesterone 97 to prevent bleeding. Although efficient in preclinical models, the transfer to the clinic of these candidates remains to be established. However, with our ongoing improvement in the comprehension of the molecular mechanisms of action of tPA, as shown in the present review, one can truly expect an optimization of the clinical use of tPA.
A challenging question is whether these effects can be exerted by exogenous recombinant tPA, injected to patients in an attempt to achieve reperfusion. Overall, Alteplase improves the outcome of patients with stroke, 1 despite variable efficiencies of reperfusion, as a function of the cause, size, and location of the clot, as well as the time-to-treatment delay. [98][99][100] Clinical evidence of the occurrence of tPA effects shown in animal models is challenging, but it is worth mentioning that after ischemic stroke, exogenous tPA was independently associated with seizure occurrence (a mechanism that involves tPA) 101 and a worse outcome at 3 months in this seizure subgroup of patients. 102 In addition, brain imaging studies evidenced that exogenous tPA indeed promotes BBB leakage in patients with stroke. 24 Similar questions can be asked in the context of hemorrhagic stroke, for which hematoma evacuation with tPA seems safe and efficient, 103 while first preclinical evidence supports proedema, proinflammatory, and proneurotoxic effects in addition to hematoma resolution. [104][105][106][107][108] Disclosures None.