Plasmin Activation of Glial Cells through Protease-Activated Receptor 1

The objective of this study was to determine whether plasmin could induce morphological changes in human glial cells via PAR1. Human glioblastoma A172 cells were cultured in the presence of plasmin or the PAR1 specific activating hexapeptide, SFLLRN. Cells were monitored by flow cytometry to detect proteolytic activation of PAR1 receptor. Morphological changes were recorded by photomicroscopy and apoptosis was measured by annexinV staining. Plasmin cleaved the PAR1 receptor on glial cells at 5 minutes (P = 0.02). After 30 minutes, cellular processes had begun to retract from the basal substratum and by 4 hours glial cells had become detached. Similar results were obtained by generating plasmin de novo from plasminogen. Morphological transformation was blocked by plasmin inhibitors aprotinin or epsilon-aminocaproic acid (P = 0.03). Cell viability was unimpaired during early morphological changes, but by 24 hours following plasmin treatment 22% of glial cells were apoptotic. PAR1 activating peptide SFLLRN (but not inactive isomer FSLLRN) promoted analogous glial cell detachment (P = 0.03), proving the role for PAR1 in this process. This study has identified a plasmin/PAR1 axis of glial cell activation, linked to changes in glial cell morophology. This adds to our understanding of pathophysiological disease mechanisms of plasmin and the plasminogen system in neuroinjury.


Introduction
Plasmin is a serine protease best known for its thrombolytic properties in the coagulation system. However, it can also act on cells that bear receptors belonging to the protease-activated receptor (PAR) family to cause secretion of in�ammatory cytokines, oxidative radicals, matrix metalloproteinases, proliferation, cell migration, and platelet aggregation [1][2][3][4][5][6]. PARs are widely expressed in the central nervous system [7].
Plasmin is generated from plasminogen, by proteolytic cleavage with either tissue-type plasminogen activator (tPA), urinary plasminogen activator (uPA), or bacterial streptokinase. It catalyzes the breakdown of �brin into D-dimers, hence acting as a brake on coagulation. Anti�brinolytics are in clinical use to limit bleeding in cardiac surgery and intracranial bleeding in traumatic brain injury [8,9].
A pathophysiological role has been recognized for the plasminogen-activating system in exacerbating intracranial bleeding, excitotoxicity and cell death in neurons or white matter, ischemia reperfusion injury, and increased permeability of the blood-brain barrier (BBB) [11][12][13][14][15][16][17]. Alterations in BBB permeability were accompanied by profound changes in cell shape to astroglial cell lines or primary glial cells, via an incompletely understood receptor mechanism [15]. is phenomenon has been con�rmed in cardiac surgery patients with clinical evidence for glial cell injury [18].
PAR receptors share a common activation mechanism, whereby proteolytic cleavage unmasks a hexapeptide ligand sequence in the exodomain of the receptor, which can then dock intramolecularly and transmit G-protein-coupled signals into the cell [19]. PAR1 is activated by any serine protease that cuts speci�cally at arginine 41 -serine 42 of the receptor generating the hexapeptide serine 42 -phenylalanineleucine-leucine-arginine-asparagine 47 (SFLLRN) at the new N-terminus. PAR1 activating proteases include thrombin, plasmin, and trypsin [20][21][22]. Although plasmin and PAR1 have independently been implicated in pathways of cerebral injury, a plasmin/PAR1 axis remains to be identi�ed in cells of the central nervous system. In mice, genetic deletion of the PAR1 homolog or deletion of tPA confers neuroprotection in a model of transient cerebral ischemia [16,22,23]. Although plasmin has not been studied in this context, other PAR1 agonists trigger excitotoxicity, cell invasion, and apoptosis in neurons, astrocytes, and microglia [13,16,22,[24][25][26][27][28]. Hence, we investigated the ability of plasmin to activate central nervous system glial cells via PAR1 in an effort to identify key cellular events and highlight potential therapeutic targets.

Cell
Culture. e human glioma cell line A172 was purchased from the American Type Culture Collection (ATCC; Manassas, VA). Cells were maintained in Dulbecco's Modi�ed Eagle's Medium (DMEM) enriched with 10% fetal calf serum, 2 mmoles/L L-glutamine, 100 U/mL penicillin, and 100 g/mL streptomycin (all Sigma-Aldrich) at 37 ∘ C in a humidi�ed environment with 5% CO 2 . At con�uence cells were detached using cell dissociation medium (Sigma Aldrich), a trypsin-free detachment step that avoids possible PAR1 activation by trypsin [20,29]. For plasmin activation, cells were cultured to con�uence in Corning Costar 24well tissue culture plates (Sigma-Aldrich). Puri�ed plasmin (Sigma-Aldrich) was used at 5 U/mL. Plasmin was also generated in situ from plasminogen and streptokinase (Sigma-Aldrich), monitoring plasmin generation spectrophotometrically with Spectrozyme colour reagent (American Diagnostica, Stanford, CT) at 405 nm using a Multimode Detector (Dynex Technologies, Chantilly, VA). Where indicated, aprotinin was used at 200 KIU/mL and -aminocaproic acid at 10 mmoles/L. SFLLRN or FSLLRN peptides were used at a concentration of 25 moles/L.

PAR1 Expression and
Cleavage. PAR1 receptor expression and cleavage was carried out by �ow cytometry as previously described [29]. Brie�y, total receptor expression was assessed using the pan-reactive anti-PAR1 antibody WEDE15, whereas receptor activation due to plasmin cleavage was monitored using antibody SPAN12. SPAN12 detects only the intact (i.e., unactivated) PAR1 receptor; therefore, loss of SPAN12 staining provides a linear measure of receptor activation [30]. Staining was carried out on freshly passaged cells in suspension following stimulation with plasmin for 5 minutes at 37 ∘ C. Staining with WEDE15, SPAN12, or classmatched (IgG 1 ) control antibody was carried out at 20 g/mL for 15 minutes on ice, followed by three washes in icecold PBS. Flow cytometric analysis was performed using an EPICS �L �ow cytometer (Beckman Coulter). WEDE-15 or SPAN12 staining was expressed in units of relative �uorescent intensity (RFI), a ratio of the mean �uorescent staining intensity obtained with detection antibody versus the staining intensity of a class-matched control antibody (i.e., an RFI of 1.00 is equivalent to no detectable expression). Results were expressed as median ± IQR (RFI units).

Morphological Changes.
Cells were observed and photographed at 0 min, 30 min, 4 h, and 24 h using a Leica DM IL inverted microscope (Leica Microsystems, Wetzlar, Germany) at ×40 or ×100 magni�cation. Morphological changes were scored on a 6-point ordinal scale according to the scheme: 1 = fully con�uent lawn of cells; 2 = cellular pseudopodia have started to retract from one area; 3 = cell processes has started to retract from multiple areas; 4 = cell lawns show visible detachment from well substrata and �apping; 5 = cells have completely detached from well substrata to form a �oating island; 6 = �oating island of cells with shrivelled appearance.   by determining the area under the histogram greater than background staining in resting cells. Serum starvation was used as a positive control for apoptosis.

Statistics.
Relative �ourescent intensity and the morphological change scale were summarised using robust measures: median and interquartile range. Group differences were plotted using boxplots and assessed formally using the Wilcoxon-Mann-Whitney test for two-group comparisons. Statistical analyses used exact algorithms performed in Stata 10 (Stata Corp., College Station, TX).

PAR1 Receptor Cleavage Induced by Plasmin.
Initial experiments examined whether PAR1 was expressed on resting A172 glioma cells. A representative �ow cytometric histogram illustrates expression of PAR1 detected with a panreceptor antibody WEDE15 (Figure 1). WEDE15 staining from pooled samples ( ) demonstrated expression at an intensity of 2.31 ± 0.33 (median ± interquartile range (IQR)) relative �uorescent intensity units (RFI units) at the cell surface. Proteolytic activation of PAR1 due to plasmin was monitored using a different antibody, SPAN12, speci�c to only the intact (i.e., unactivated) receptor (Figures 1 and  2). is antibody was expressed at a �uorescent intensity of 1.69 ± 0.05 RFI units (median ± IQR.) on resting cells ( ). Activation with a puri�ed plasmin preparation caused a statistically signi�cant loss of SPAN12 expression at 5 minutes (1.69 ± 0.07 resting cells versus 1. 3 ± 0.20 plasmin activated; 0.02; Figure 2(a)). To demonstrate whether this was due to proteolytic cleavage of the receptor by plasmin, experiments were repeated in the presence of the serine protease inhibitor aprotinin: this showed complete restoration of SPAN12 expression (1.65 ± 0.0 ; 0.02 versus plasmin alone). Analogous results were obtained when plasmin was generated in situ from plasminogen ( Figure  2(b)).

Morphological Changes Induced by Plasmin.
A timecourse of photomicrographs taken at 30 minutes, 4 hours and 24 hours aer plasmin activation illustrates remarkable morphological transformation of A172 glioma cells ( Figures  3 and 4(a)). At 30 minutes cell processes could be observed retracting from the basal substratum, and �ap formation was seen at the edges of cell lawns (Figure 3(b)). By 4 hours the cell lawn had completely detached from the well into a �oating island, apparently preserving cell-to-cell contacts (Figure 3(c)). Video microscopy showing cell detachment in real time illustrated the release of individual tethers from the end of cell processes (http://youtube/FkSUdWKfoxE). e morphological changes were abrogated by 200 KIU/mL aprotinin, indicating they were dependent on serine proteolytic activity of plasmin (Figure 3(e)). To demonstrate a speci�c role for PAR1 in this process, a PAR1 speci�c activating peptide SFLLRN (25 moles/L) was examined. is revealed exactly analagous morphological changes triggered through SFLLRN, but not through an inert scramble peptide FSLLRN even when used at high concentrations up to 250 moles/L (Figures 3(f) and 4(b)). Morphological changes up to 4 hours following plasmin treatment were not accompanied by any detectable apoptosis as assessed with Annexin V staining by �ow cytometry (Figure 5). �owever, at 24 hours the �oating mass of glial cells had a shrivelled appearance (Figure 3(d)) and �ow cytometric analysis indicated 22% of cells were apoptotic ( Figure 5).

��ect o�
Anti�brinolytics. e effect of anti�brinolytics on glial cell morphology was determined in cultures in which plasmin was generated in situ from plasminogen. e antiplasmin(ogen) agents aprotinin (200 KIU/mL) and -aminocaproic acid (10 mmoles/L), used at concentrations e�uivalent to their clinical usage, statistically signi�cantly inhibited glial cell detachment due to plasmin (each 0.03; Figure 6). Biochemical assays for plasmin activity in cell cultures con�rmed that each of the anti�brinolytics had abrogated plasmin activity at the concentrations employed (data not shown).

Discussion
e present study has proven the principle that plasmin can activate human glial cells via proteolytic cleavage of PAR1. e type of cell-lawn detachment observed was similar to that previously described in A172 cells treated with an integrin antagonist, that also caused separation from basal substratum and accumulation of cells into �oating spheroids [31]. Although immediate cytotoxicity due to plasmin was ruled out in apoptosis assays, gradual expression of annexin V occurred in our culture system aer 24 hours following plasmin activation. e fact that SFLLRN (but not the inactive scramble peptide FSLLRN) could recapitulate the morphological changes in A172 cells proved the role of PAR1 in this process. ese �ndings add to a growing literature that morphological transformation of astroglial cells can take place in a pathway involving the plasminogen activating system and components of the focal cell adhesion machinery [15,31,32]. An important recent study showed that tPA was capable of eliciting morphological changes and increased BBB permeability in mouse astroglial cells, through alterations to the Rho kinase (ROCK) pathway regulating focal adhesion contacts with extracellular matrix proteins [15]. A two-receptor mechanism was postulated in that study, one of which remained unknown. e unknown receptor was a receptor for plasminogen, as shown by antagonism with the anti�brinolytic agents aprotinin or tranexamic acid. e plasmin/PAR1 axis described by us was similarly antagonized by aprotinin or a lysine analog and may therefore present a plausible alternative receptor pathway.
Most studies investigating the effect of serine proteases on neuronal PAR1 have focused on thrombin, which can either be neuroprotective or, conversely, induce apoptosis in neurons, depending on concentration and length of exposure to thrombin [7,24,25,28,33]. Initial studies using our culture system indicated that alpha-thrombin did not activate PAR1 on A172 glial cells within a physiological concentration range, but that plasmin activated PAR1 efficiently [34]. e plasmin/PAR1 axis of glial activation identi�ed here may synergise to cause neurodegeneration with pathways of excitotoxic potentiation of N-methyl D-aspartate responses in neurons, also mediated via PAR1 activation with plasmin or thrombin [27,35].
Cell lawn detachment may represent a manifestation of natural morphological processes, such as cell migration or astrocyte stellation. Reversal of the stellate phenotype by thrombin or changes in astrocyte morphology described for tPA may utilize the same cellular pathway, since both are regulated by ROCK [15,36]. Local restructuring of focal adhesion contacts and cell detachment may emulate important physiological functions linked to cell migration, wound healing, proliferation, glial scarring, and neurite formation [17,[37][38][39][40]. Video microscopy of the cell detachment process appeared to indicate the release of individual pseudopod tethers, similar to what had been described for release or thinning of astrocyte cell processes induced by thrombin agonist [38,41].
e morphological changes described here may be most relevant in the setting of neurotrauma or neuroinjury secondary to BBB breakdown, when zymogens such as plasmin that are normally con�ned to the systemic circulation can enter the brain [42]. A common surgical insult that compromises BBB function is cardiopulmonary bypass: this is linked to changes in glial cell morphology, neurocognitive de�cits and stroke [18,[43][44][45][46]. e demonstration that glial cells are activated via plasmin/PAR1 expands our understanding of neurodegenerative mechanisms and focuses attention on PAR1 as a possible drug target for neuroprotection. While PAR1 de�cient mice are protected in various models of ischemia, it is simplistic to consider PAR1 antagonism as a treatment option to block astroglial pathways of neuroinjury, since a clinical trial with the oral PAR1 antagonist vorapaxar for treatment of acute coronary syndromes was associated with signi�cantly higher rates of intracerebral hemorrhage [22,47,48]. A desirable outcome would be to combine thrombolytic properties without hemorrhage or compromise of BBB permeability; however, this outcome remains elusive. However, plasmin inhibitors are neuroprotective, and the plasmin/PAR1 axis identi�ed here may be explored further as a therapeutic target [17,[49][50][51].
ere are some limitations to this study. e cell culture system may fail adequately to model the homeostatic environment of the brain, which is endowed not only with serine proteases and their zymogens but also naturally occurring serine protease inhibitors (serpins) [52,53]. Furthermore, the in vitro cell culture model may not faithfully replicate in vivo pathways of glial injury caused by plasmin nor the three-dimensional cell-cell and cell-substrate interaction that arises from the precise cellular organization that exists within the brain. However, a reductionist cell culture approach as employed here is well suited to prove unequivocally that glial cells can be activated through proteolytic cleavage of PAR1 due to plasmin.
In conclusion, we have identi�ed a PAR1 axis of glial cell activation triggered by plasmin or the plasminogen system. is may be especially relevant under conditions of BBB breakdown or intracranial hemorrhage when serine proteases gain access to the neurovascular unit.