Antibodies Preventing the Interaction of Tissue-Type Plasminogen Activator With N-Methyl-D-Aspartate Receptors Reduce Stroke Damages and Extend the Therapeutic Window of Thrombolysis

Background and Purpose— Tissue-type plasminogen activator (tPA) is the only drug approved for the acute treatment of ischemic stroke but with two faces in the disease: beneficial fibrinolysis in the vasculature and damaging effects on the neurovascular unit and brain parenchyma. To improve this profile, we developed a novel strategy, relying on antibodies targeting the proneurotoxic effects of tPA. Methods— After production and characterization of antibodies (&agr;ATD-NR1) that specifically prevent the interaction of tPA with the ATD-NR1 of N-methyl-D-aspartate receptors, we have evaluated their efficacy in a model of murine thromboembolic stroke with or without recombinant tPA-induced reperfusion, coupled to MRI, near-infrared fluorescence imaging, and behavior assessments. Results— In vitro, &agr;ATD-NR1 prevented the proexcitotoxic effect of tPA without altering N-methyl-D-aspartate-induced neurotransmission. In vivo, after a single administration alone or with late recombinant tPA-induced thrombolysis, antibodies dramatically reduced brain injuries and blood–brain barrier leakage, thus improving long-term neurological outcome. Conclusions— Our strategy limits ischemic damages and extends the therapeutic window of tPA-driven thrombolysis. Thus, the prospect of this immunotherapy is an extension of the range of treatable patients.

D espite tremendous advances in understanding the pathophysiology of stroke and the significant number of clinical trials, the thrombolytic agent recombinant tissue-type plasminogen activator (rtPA) remains the only approved acute pharmacological treatment for ischemic stroke. 1 Nevertheless, its use is limited to a short therapeutic window (4.5 hours post-onset, excluding 90% of patients), 2,3 and associated with a threat of hemorrhage 1 and neurotoxicity. 4 -8 Thus, there is a critical need for a larger therapeutic window of rtPA and for the development of alternative treatments for patients who are not eligible for thrombolysis.
Vascular tissue-type plasminogen activator (tPA) promotes fibrinolysis by converting plasminogen into active plasmin but can cross the even uninjured blood-brain barrier (BBB). 9 Together with tPA released by neurons and glia, it can interact with several substrates within and beyond the neurovascular unit. For instance, tPA might promote BBB leakage through the conversion of pro-platelet-derived growth factor (PDGF-C) into PDGF-CC, the shedding of low-density lipoprotein receptor-related protein, and/or matrix metalloproteinase (MMP-3 and MMP-9) activation. 10 -12 In the brain parenchyma, the interaction of tPA with N-methyl-Daspartate receptors (NMDAR), lipoprotein receptor-related protein, annexin II in glial cells and/or neurons activates signaling processes that result in adverse outcomes, including cerebral edema, hemorrhage, and cell death. [12][13][14] Based on our knowledge of the molecular events through which tPA aggravates NMDAR-mediated neurotoxicity, 7 we generated a polyclonal antibody against the interaction site of tPA on the NR1 subunit of NMDAR, which could prevent the proexcitotoxic effect of rtPA in vitro. In a clinically relevant model of thromboembolic stroke in mice, 15 a single early or late intravenous administration of the antibody strongly and durably reduced brain lesions and neurological deficits. Moreover, this strategy increased the therapeutic window of rtPA-driven thrombolysis.

Methods
Unless specified, chemicals were from Sigma Aldrich. Experimentations complied with the European Directives and the French Legislation on Animal Experimentation and were approved by the local ethical committee. For more detailed protocols, refer to the supplementary methods (http://stroke.ahajournals.org).

Production and Purification of Polyclonal Antibodies
The region of the aminoterminal domain of the NR1-1a subunit (amino acids 19 to 371, a sequence absent from other glutamate receptor subunits) corresponding to the domain of interaction with tPA (designed rATD-NR1), was produced as previously. 7 Then, slightly modified from the previous procedure, 16 mice were immunized by intraperitoneal injection of immunogenic mixtures: complete Freund's (first injection) and incomplete Freund's (once a week during 3 weeks) adjuvant alone (control) or containing the rATD-NR1 (30 g; nϭ60 per group). Sera collected 2 weeks after the last inoculation were purified on hydroxyapatite columns (Proteogenix) to obtain ␣ATD-NR1 or control polyclonal antibodies. Two different batches of ␣ATD-NR1 were used to conduct experiments.

Thromboembolic Focal Cerebral Ischemia
Male Swiss mice (28 to 30 g; CERJ, Paris, France) or tPA knockout and their wild-type C57BL/6J littermate mice were subjected to focal cerebral ischemia by injection of murine ␣-thrombin (0.75UI, 1 L, Gentaur) into the middle cerebral artery (MCA) 15 under deep anesthesia (isoflurane; induction 5%, maintenance 2% in O 2 /N 2 O). Thrombolysis was initiated by rtPA (10 mg/kg; Actilyse; Boehringer Ingelheim; tail vein injection, 10% bolus, 90% perfusion during 40 minutes). Controls received saline under identical conditions. Rectal temperature was maintained at 37Ϯ0.5°C throughout the surgical procedure using a feedback-regulated heating system. Cerebral blood flow was continuously measured by laser Doppler flowmetry using an optic fiber probe (Oxford Optronix) affixed on the skull above the MCA downstream of the thrombin injection site. The post-ischemic cerebral blood flow was expressed as the percentage of cerebral blood flow during the last 5 minutes of rtPA or vehicle infusion over the baseline cerebral blood flow before the injection of ␣-thrombin.

Passive Immunization
A single injection in the tail vein with saline, purified ␣ATD-NR1, or control antibodies (160 g in 200 L each) was performed at the time indicated after ischemic challenges. Animals were randomly assigned to the experimental groups. Mortality was null for all procedures/treatments. Also, no evidence of major intracerebral bleeding was observed.

Histological Analysis of Brain Lesions
After 24 hours, brains were collected, frozen in isopentane, cryostatcut (20-m sections), stained with thionine, and analyzed (blindly to the group) as described earlier. 15

MRI Analyses
MRI analyses were performed on ischemic mice that had been intravenously injected with saline or the ␣ATD-NR1 antibodies 4 hours post-middle cerebral artery occlusion (MCAO) (see also supplementary methods).

Evans Blue Extravasation
Evans blue (200 L, 2%) was intravenously injected in ischemic mice 3 hours before euthanasia and quantified in perfused cortices as previously described. 17 Briefly, ipsilateral and contralateral hemispheres were weighed, homogenized in N,N-dimethylformamide, and centrifuged (21 000 g, 30 minutes). Evans blue was quantified in supernatants at 620 nm minus the background calculated from the baseline absorbance between 500 and 740 nm and divided by the wet weight of each hemisphere.

Post-ischemic Neurological Deficit
A global evaluation of locomotor activity was performed 24 hours, 1 month, and 3 month(s) post-surgery by placing mice in a chamber (detailed in supplementary methods) and recording their movements over 5 minutes. The neurological outcome is expressed as the percentage of freezing time during a 5-minute period.

Cell Culture
Primary cultures of cortical neurons were prepared from fetal mice (E15 to E16) as described previously. 5

In Vitro Excitotoxicity
Rapidly triggered excitotoxicity was induced, alone or in combination with rtPA (20 g/mL) and/or ␣ATD-NR1 or control Igs (0.01 mg/mL), by exposing neurons to 50 mol/L NMDA for 1 hour and transferring cells back to serum-free medium. 18 Neuronal death was quantified 24 hours later by lactate dehydrogenase release (Roche).

Near-Infrared Fluorescence Imaging
Four hours after thromboembolic ischemia, saline or Alexa 680labeled ␣ATD-NR1 (160 g per mouse) were intravenously injected. After 2 hours, mice (nϭ3 per group) were perfused with heparinized saline and brains were harvested and placed in the planar near-infrared fluorescence imaging system (Photon Imager; Biospace) for a first global picture acquisition. Then coronal brain slices of 1-mm thickness were cut in a matrix. Excitation wavelength was set to 650 nm with a 700-nm high-pass emission filter. Autofluorescence images were acquired using an excitation wavelength of 590 nm with the same emission filter. Fluorescence was quantified on the whole hemispheres using M3Vision software (Biospace).

Statistical Analyses
Results are the meanϮSEM. Statistical analyses were performed by the Kruskall-Wallis test followed by post hoc comparison with the Mann-Whitney U test.

Antibodies Against the ATD-NR1 Prevent tPA-Promoted NMDAR-Mediated Neurotoxicity In Vitro
The aminoterminal domain of the NMDAR NR1 subunit (ATD-NR1) is the site of interaction with tPA, mediating increased NMDAR signaling and the attendant neurotoxicity. 7 Accordingly, recombinant ATD-NR1 was used as an immunogenic peptide in mice. The resultant purified polyclonal immunoglobulins (termed ␣ATD-NR1) were then characterized and shown to recognize the NR1 subunit of murine and human neurons (Supplemental Figure I). The next step was to validate the ability of ␣ATD-NR1 to prevent the aggravating effect of rtPA on NMDA toxicity. 5 In cultured murine cortical neurons exposed to rapidly triggered excitotoxicity, the coapplication of the ␣ATD-NR1 prevented rtPA (0.3 mol/L)-promoted neuronal loss ( Figure 1A

␣ATD-NR1 Improves Neurological Outcome, Protects the Brain Against Stroke, and Increases the Therapeutic Window of rtPA-Induced Thrombolysis
The therapeutic value of the ␣ATD-NR1 was then investigated in vivo. The ability of antibodies to cross the BBB was investigated. Immunohistology on brain slices harvested 2 hours after an intravenous injection of Alexa 555 -labeled ␣ATD-NR1 (160 g) showed positive stainings in both the endothelial vascular wall and brain parenchyma (Figure 2A), arguing for a trans-BBB passage even under normal conditions. Accordingly, in a model of excitotoxicity with no BBB leakage and involving endogenous tPA, 9 a single intravenous injection ␣ATD-NR1 (160 g) reduced by 47% the lesion size induced by the striatal administration of NMDA compared with mice injected with control antibodies (nϭ8, PϽ0.01; data not shown). In sham-operated animals, the proportion of Alexa 555 -labeled ␣ATD-NR1 (injected after 4 hours) transferred to the brain after 2 hours was 10.4%. Mice Black and white bars represent the 2 consequent stimulations with NMDA, respectively, before and after treatments. *PϽ0.05 before vs after treatment. NMDA indicates N-methyl-D-aspartate; rtPA, recombinant tissue-type plasminogen activator. Because endogenously produced tPA is known to participate in the evolution of ischemic damages, 6,16 the use of the ␣ATD-NR1 as a standalone therapy was then tested in a model of thromboembolic stroke in mice. Both early ( Figure  3A) and late ( Figure 3B) deliveries of the ␣ATD-NR1 alone (160 g, intravenous, single bolus) conferred brain protection (44% and 41% smaller lesions compared with saline-injected controls when injected 20 minutes and 4 hours postocclusion, respectively; PϽ0.001).
Regarding exogenous tPA, although early thrombolysis reduced ischemic lesions by 31.5%, this effect was not improved when rtPA was combined with immunotherapy ( Figure 3A). When thrombolysis was performed 4 hours post-clot formation, rtPA still restored blood flow as effi-ciently as on early administration ( Figure 3C), but this time with a significant 33% increase in brain lesion volumes (24.9Ϯ1.12 mm 3 for saline compared with 33.03Ϯ1.47 mm 3 for rtPA, PϽ0.002; Figure 3B). Interestingly, the deleterious effect of late rtPA administration was completely suppressed when combined with a single coinjection of ␣ATD-NR1. Importantly, the beneficial effect of the late administration of the antibodies was comparable to that seen without rtPA or on early thrombolysis. This was evident from the resulting lesion volumes, which fell to 55.5% and 37.3% compared with those seen in animals receiving no treatment and only late rtPA, respectively ( Figure 3B; PϽ0.005). Importantly, the coadministration of antibodies did not alter the ability of early or late rtPA injection to induce reperfusion ( Figure 3C), consistent with their lack of influence on rtPA-promoted clot lysis in vitro (Supplemental Figure II).
Two further experiments demonstrated that the ␣ATD-NR1 specifically targets the tPA/NMDAR interaction. Control Igs were ineffective against stroke damage ( Figure  3A-B) and ␣ATD-NR1 antibodies conferred protection in wild-type but not in tPA-deficient mice ( Figure 3D). Confirming previous studies, 6 absence of endogenous tPA en- Figure 2. ␣ATD-NR1 cross the BBB and protect the brain against the neurotoxic effect of endogenous tPA. A, Fluorescent ␣ATD-NR1-Alexa 555 was injected intravenously (160 g). After 2 hours, mice were perfused, brains harvested, and cut for immunohistochemistry. Vessels were stained with an antibody against Type IV collagen (green) nuclei with DAPI. The picture illustrates the passage of ␣ATD-NR1-Alexa 555 (red) across the intact BBB (nϭ3). B, Quantification of the passage of ␣ATD-NR1-Alexa 555 2 hours after intravenous injection in perfused brain extracts harvested from sham or ischemic animals. Antibodies were injected 20 minutes or 4 hours post-MCAO (c indicates contralateral; i, ipsilateral hemisphere; nϭ3 per group: *different from the corresponding sham control or the contralateral side, PϽ0.05). C1, Representative DWI-MRI analysis of a lesion 3 hours post-MCAO (nϭ3). C2-3, Representative ex vivo NIRF analyses of the parenchymal distribution (C2, whole brain; C3, NIRF images obtained on 1 mm-thick slices transposed on the corresponding MRI-acquired slices) of ␣ATD-NR1-Alexa 680 injected intravenously 4 hours after initiating MCAO; brains were harvested 2 hours after antibody administration. There was no signal in control animals. D, Fluorescence quantification of C showing an increased permeation of the ␣ATD-NR1-Alexa 680 in the ipsilateral (i) vs the contralateral (c) side of MCAO (nϭ3, *PϽ0.05). BBB indicates bloodbrain barrier; tPA, tissue-type plasminogen activator; DAPI, 4Ј,6-diamidino-2-phenylindole; MCAO, middle cerebral artery occlusion; DWI, diffusion-weighted imaging; NIRF, near-infrared fluorescence imaging. tailed a significant protection from thromboembolic brain injury. Altogether these data demonstrate two important features of the ␣ATD-NR1: a single injection is sufficient to confer brain protection irrespective of its timing (20 minutes or 4 hours post-MCAO) due to the blockade of the deleterious effects of endogenous tPA and can suppress the injurious effects of rtPA when injected late after stroke onset and thus extends the therapeutic window of rtPA-induced thrombolysis.

Late Immunotherapy Targeting the ATD-NR1 Is Sufficient to Improve Neurological Outcome and Protects the Brain Against Stroke
The clinical relevance of targeting the potentiating effect of endogenous tPA on NMDAR neurotoxicity, even by a delayed immunotherapy, was further confirmed by an MRIbased longitudinal follow-up of animals treated 4 hours post-clot onset. ADC sequences, recorded 24 hours postischemia, as well as T2 MRI analyses ( Figure 4A) confirmed  At the indicated times, they were placed in a 7-Tesla MRI for T2 brain imaging and determination of the apparent diffusion coefficient (ADC) at 24 hours (representative images, nϭ3 per group). B, Lesion volumes (illustrated in A) were measured by MRI 24 hours post-clot onset (nϭ3 per group; *PϽ0.05 vs corresponding control). C, Mice were treated with saline or ␣ATD-NR1 (160 g) 4 hours after stroke and their neurological deficits were evaluated 24 hours, 30 days, and 3 months post-ischemia (nϭ10 per group: *PϽ0.05 vs corresponding treated control). rtPA indicates recombinant tissue-type plasminogen activator; MRI, magnetic resonance imaging. the patterns of thionine stainings (45% reduction of the lesion volume 24 hours post-ischemia; PϽ0.05; Figure 4B), brain protection being maintained for up to 1 week. Like in the clinical setting, we then investigated whether brain protection translated into improvement of neurological deficits. Strokeinduced neurological deficits evaluated at 24 hours were potently reduced by the intravenous administration of the ␣ATD-NR1 20 minutes (Supplemental Figure III) or 4 hours ( Figure 4C; PϽ0.05) post-MCAO. Providing further support for the clinical relevance of this treatment modality, this beneficial effect was maintained for at least 3 months ( Figure  4C).
We further investigated whether the global beneficial effect of the ␣ATD-NR1 could also affect the pro-PDGF-C/ PDGF-CC/PDGF-R␣ cascade. 10 Immunoblots from ischemic cortices confirmed that rtPA treatment leads to huge increases of PDGF-CC levels (almost 8-fold when injected 4 hours post-MCAO). This stimulation was significantly antagonized (Ϫ40%) by the ␣ATD-NR1 ( Figure 5C). Similarly, rtPA increased the activity of PDGF-R␣ after ischemia (ϩ74% when injected 4 hours post-MCAO; Figure 5D). This increase was abolished by the ␣ATD-NR1 and even reverted into a significant decrease compared with animals exposed to MCAO with no treatment (approximately Ϫ75% versus MCAO alone and Ϫ90% versus MCAO with tPA; PϽ0.05).

Discussion
Despite the beneficial results of the National Institute of Neurological Disorders and Stroke trial, 1 stroke has remained a challenging clinical problem. Indeed, thrombolysis can be applied to few patients only, and improved treatment modalities for stroke are mandatory. The limitations of the use of rtPA are mainly due to nonfibrinolytic effects. 12 Controlling the prohemorrhagic and proneurotoxic effects of tPA (endogenous and exogenous) is one of the most urgent issues in the field of stroke. We demonstrate that an antibody preventing the interaction of tPA with the NMDAR could be an ideal therapeutic candidate in ischemic stroke, alone or in combination with tPA.
Our aim was to assess whether targeting the deleterious effects initiated by the tPA/NMDAR interaction can afford brain protection. Central to our strategy was a dedicated thromboembolic model of stroke, in which we previously demonstrated that early rtPA-induced thrombolysis reduces ischemic lesions and cognitive deficits. 15 We have further validated this model by showing that like in the clinical setting, brain protection is better the sooner rtPA-based thrombolysis is performed. The noxious effects of tPA, whether released from parenchymal cells and/or administered late after stroke, can largely be attributed to its potentiating effect on NMDAR activity. 5,7 Although the mechanism has been debated, tPA is undoubtedly a positive modulator of the NMDAR-mediated glutamatergic signaling pathways in neurons, [12][13][14] and possibly in endothelial cells. 19 Based on our proposed mechanism of action, in which tPA directly (ie, with no requirement of plasmin[ogen]) interacts with the NMDAR NR1 subunit, 7 we produced polyclonal antibodies recognizing the N-terminal end of NR1 in both rodents and humans. Importantly, these antibodies do not alter NMDAinduced neurotransmission (toxicity, calcium influx) but specifically target the effect of tPA on these receptors. We show in vitro that these antibodies limit rtPA-induced NMDAevoked neuronal calcium overload and, by this, are highly effective against rtPA-dependent aggravation of excitotoxicity. This ability finds its expression in our in vivo investigations, which demonstrate that a single intravenous injection of the ␣ATD-NR1 alone leads to massive neuroprotection, irrespective of the timing (20 minutes or 4 hours post-clot onset), whereas coadministration averts the deleterious effects of late rtPA administration, making late thrombolysis as safe as early reperfusion. Perhaps most important, clinically relevant criteria (MRI analysis and neurological scores) support the contention that ␣ATD-NR1 cannot only confer brain protection but also improve long-term neurological outcome.
A growing body of evidence indicates that tPA is a central mediator of ischemia-induced increase in neurovascular unit permeability (and consequently edema and hemorrhagic transformation) in experimental models 12 and possibly in thrombolyzed stroke patients. 20 We demonstrate that the ␣ATD-NR1 cause beneficial changes in events considered as hallmarks of BBB disruption, that is, BBB permeabilization, MMP-3 and MMP-9 activation, and PDGF-R␣ activation by PDGF-CC, irrespective of being caused by MCAO alone (ie, mediated by endogenous tPA) or by late administration of rtPA. These observations strengthen the therapeutic interest of the ␣ATD-NR1, because brain edema and bleeding are associated with a poor clinical outcome in patients with stroke. 1 They also support the view that parenchymal damages are at the origin of (or at least contribute to) BBB leakage, tPA, and other mediators being released from stressed neurons to promote harmful events at the BBB. 17 Targeting the brain with antibodies has been the subject of question of feasibility and major safety concerns. However, consistent with previous observations, [21][22][23][24][25] we show that low levels of antibodies can enter the brain parenchyma in the absence of injury and that the passage of the BBB is increased under ischemic conditions. Moreover, our antibodies do not induce encephalopathy or cognitive, behavioral, or mnesic deficits (not shown). This last observation rules out the possibility that the protective neurological effect induced by the ␣ATD-NR1 after stroke could be due to a downregulation of the inhibitory neurotransmission. Moreover, ␣ATD-NR1 is highly specific of their target, because it does not mediate protection in tPA-deficient mice and does not alter the antiapoptotic effect of tPA (data not shown), which is independent on the interaction with NMDAR. 26 The necessity for a more specific NMDAR blockage 27 lends greater weight to our antibody-based strategy, because it only targets a neuromodulatory rather than a whole neurotransmission system. 28 In line with ongoing experimental strategies targeting undesirable rtPA functions to optimize stroke treatment, 8,29,30 we demonstrate that this immunotherapy targeting the ability of endogenous tPA to potentiate NMDAR signaling reduces excitotoxic damages, BBB leakage, and the attendant brain injury and neurological deficits. We also provide proof of concept for the use of these antibodies as an adjunct to rtPA-based thrombolysis with the potential to antagonize brain and neurovascular unit damage and, by this, to extend the therapeutic window.

Sources of Funding
This work was supported by the Institut National de la Santé Et de la Recherche Médicale, the French Ministry of Research and Technology, and the Eurostroke-Arise Program (FP7/2007(FP7/ -2013.