Nitric oxide mediates neurodegeneration and breakdown of the blood-brain barrier in tPA-dependent excitotoxic injury in mice

Tissue plasminogen activator (tPA) is a serine protease well known for its function in fibrinolysis. tPA also functions in the central nervous system (CNS) by mediating excitotoxicity in the hippocampus (Tsirka et al., 1995). Mice deficient in tPA are resistant to the excitotoxic neuronal cell death that normally follows intrahippocampal injection of kainate (KA), and this resistance reflects a role for tPA in cell death acutely, rather than it being the result of altered brain development in the absence of tPA (Tsirka et al., 1995; Tsirka et al., 1996).

Another molecule that has been implicated in neurodegeneration and excitotoxicity is nitric oxide (NO). NO is produced by multiple cell types in the CNS and its production increases in stroke events and neurodegenerative disease (Endoh et al., 1994; Murphy, 2000; Nomura and Kitamura, 1993; Zhang et al., 1995). Despite the increased presence of NO during excitotoxicity, its specific role remains unclear. Depending on the context in which NO is produced, it appears to elicit either a degenerative effect (Dawson et al., 1991; Gabriel et al., 2000; Schulz et al., 1995) or has negligible effects on excitotoxicity (Lerner-Natoli et al., 1992; Pauwels and Leysen, 1992). NO might even exhibit protective properties by reducing oxidative stress (Chiueh, 1999; Kiprianova et al., 1997).

In models implicating NO in neurodegeneration, it is believed that the toxic effects are a consequence of the actions of its downstream metabolite, peroxynitrite (ONOO^–). ONOO^– is a highly reactive oxidant formed when NO reacts with superoxide radicals. ONOO^– can modify proteins through the formation of nTyr adducts, and also regulates excitotoxicity (Brown and Bal-Price, 2003; Kroncke et al., 2001), apoptosis (Estevez et al., 1995; Lin et al., 1995) and induces oxidative DNA damage (Grace et al., 1998; Inoue and Kawanishi, 1995). However, in one study, ONOO^– was reported to have neuroprotective effects, thus complicating the potential role of NO in the brain (Garcia-Nogales et al., 2003).

tPA, NO and ONOO^– have also been implicated in BBB breakdown (Mayhan, 2000; Tan et al., 2004; Yepes et al., 2003), which occurs during excitotoxic injury (Chen et al., 1999b; Lassmann et al., 1984) and correlates with neuronal damage (Ruth, 1986; Ting et al., 1986). The effects of glutamate on BBB are mediated by the NO produced during stroke (Mayhan and Didion, 1996).

It is evident that the roles of NO and ONOO^– are highly contextual. Therefore, it is necessary to evaluate their roles using the tPA-mediated excitotoxicity model to determine whether their path of action converges with that of tPA. We demonstrate here that mice deficient in tPA have attenuated NO synthase (NOS) activity and hence generate less NO and ONOO^–. Intriguingly, injection of an NO donor prior to KA injection restores the toxic effects of KA in tPA-deficient mice. Moreover, whereas both NO and ONOO^– mediate excitotoxicity, only NO and tPA alter BBB permeability. These results provide insights into the role of NO in mediating neurotoxicity through ONOO^– production and BBB breakdown, and identify a downstream effector pathway in tPA-mediated neurodegeneration.

The production of NO is attenuated for tPA^–/– microglia stimulated in culture with bacterial lipopolysaccharide (LPS) in comparison with LPS-stimulated wild-type microglia (Yepes et al., 2003). To determine if this observation in culture is reflective of microglial behavior in vivo, wild-type and tPA^–/– mice were injected unilaterally with KA to induce excitotoxic injury, sacrificed at multiple time points, and extracts were prepared from the excised hippocampi. Six hours after the injection of KA into wild-type mice, there was a sharp and significant increase in NOS activity. Apart from a slight decline at day 1 post injection of KA, NOS activity remained elevated at days 3 and 5 (Fig. 1A). By contrast, NOS activity in tPA^–/– hippocampi did not increase in response to KA. This failure to respond reflected an acute requirement for tPA, since exogenous provision of tPA using a mini-infusion pump restored the activation of NOS to wild-type levels. Interestingly, even basal levels of NOS activity were higher in wild-type mice than in tPA^–/– mice, suggesting that NOS and tPA pathways might mediate normal functional pathways distinct from those related to neurodegeneration; a non-degenerative role has previously been proposed for tPA in the context of memory consolidation (Baranes et al., 1998).

tPA functions through both proteolytic and non-proteolytic pathways (Tsirka et al., 1997; Rogove et al., 1999; Nicole et al., 2001; Melchor et al., 2003). To determine whether tPA activity is crucial for the increase in NOS activity, we pharmacologically inhibited tPA in wild-type mice prior to the injection of KA using the synthetic tPA inhibitor tPA Stop, which successfully prevented the increase in NOS activity (Fig. 1B). Conversely, exogenous provision of a catalytically inactive mutant of tPA, S481A tPA, into tPA^–/– mice did not restore the activation of NOS (Fig. 1B). Taken together, these three approaches demonstrate that the catalytic activity of tPA is required for KA-elicited increases in NOS activity.

The initial upsurge of NOS activity is observed within 6 hours of KA injection (Fig. 1A). This time point is considerably earlier than any morphologically detectable microglial activation (Andersson et al., 1991). This result suggested that the NO generated at this early time point would more plausibly be produced from neurons through the activation of neuronal NOS (nNOS). We explored two possibilities as mechanisms through which tPA might promote nNOS activation. The first was based on reports that tPA can cleave the NR1 subunit of the NMDA receptor (NMDAR), leading to potentiated signaling (Nicole et al., 2001), and that NMDAR signaling activates nNOS (Garthwaite et al., 1989). To examine this potential pathway, we injected 1 mg/kg of the NMDAR antagonist MK-801 30 minutes prior to KA injection and found that inhibition of NMDAR signaling blocked NOS activation (Fig. 1B). The second potential mechanism was based on the well-known function of tPA in converting plasminogen to plasmin, which then degrades laminin and triggers apoptosis (Chen and Strickland, 1997). Plasminogen-deficient mice, whose hippocampal neurons are also resistant to KA injury (Tsirka et al., 1997), were subjected to KA injection and analyzed; the NOS activity in hippocampal extracts was reduced to non-stimulated levels (NOS activity at 6 hours was measured at 0.26±0.04 units/mg protein). Taken together, these results indicate that the proteolytic activity of tPA is important for the upregulation in NOS activity, and suggest that both mechanisms examined might be involved in nNOS activation.

In addition to measuring NOS activity, relative levels of ONOO^– were also determined, as ONOO^– is the most common downstream metabolite of NO generated during pathological events. Since ONOO^– is very unstable, the most accurate method for examining its prior formation is through detection of its downstream product, nitrotyrosine (nTyr), which accumulates on nearby proteins. Western blot analysis to detect nTyr-modified proteins was performed on hippocampal extracts from wild-type and tPA^–/– mice injected with KA (Fig. 1C). In wild-type mice, but not in tPA^–/– mice, the levels of total nTyr-modified proteins on the side injected compared with the non-injected side were increased. In fact, wild-type mice appear to have a higher basal level of nTyr in the non-injected hippocampus compared with the tPA^–/– non-injected hippocampus (Fig. 1C). These results correlate with the higher level of NOS activity found in non-treated wild-type mice in comparison with non-treated tPA^–/– mice (Fig. 1A).

To corroborate the western blot analysis and determine the cellular specificity of nTyr-modified proteins, immunohistochemistry was performed on tissue sections using an anti-nTyr antibody. As reported previously, wild-type but not tPA^–/– mice exhibit neurodegeneration in the hippocampus after intrahippocampal KA injection (Fig. 1D, top row; compare loss of Cresyl-Violet-staining pyramidal neurons in CA1-3 regions on the injected side of the wild-type mouse with the comparable regions on the non-injected side of the wild-type mouse and with the injected side in the tPA^–/– mouse) (Tsirka et al., 1995). We now show that the extent of degeneration observed colocalizes with the region where nTyr formation is detected (Fig. 1D panels c-h, and Fig. S1, supplementary material). Diffuse nTyr-positive scattered cells are observed in all animals regardless of treatment. However, nTyr staining in pyramidal neurons is observed only on the KA-injected side of wild-type mice (arrows), in agreement with the Western blot analysis. These results suggest that NO and ONOO^– could act as downstream effectors in the tPA-mediated neurodegenerative pathway.

The increased levels of nTyr staining in wild-type compared with tPA^–/– mice after KA injection raised the possibility that it is ONOO^– that mediates the ensuing neuronal cell death. To establish causality, wild-type mice were infused with the ONOO^– scavengers FeTMPyP and FeTPPS prior to KA injection. The infusions conferred protection to excitotoxicity (Fig. 2A,C and Table 1) as well as a reduction in nTyr staining (Fig. S1, supplementary material). Protection was not observed with infusion of TMPyP (Table 1), an analog to FeTMPyP that lacks the iron center and therefore cannot scavenge ONOO^– (Crow, 2000; Stern et al., 1996). Injection of 5 μg/μl tPA prior to KA injection restored the toxic effects of KA (Fig. 2A,C). This effect was then reversed by the infusion of FeTMPyP prior to KA injection (Fig. 2A,C). We thus conclude that ONOO^– is a functional contributor to neuronal death after excitotoxic injection.

Since ONOO–is a potent oxidant, we also examined whether treatment with FeTMPyP or FeTPPS would reduce oxidative stress on neurons following KA injection. Oxidative stress was visualized by injection of dihydroethidium (DHE), which is oxidized to ethidium during oxidative stress and can be visualized by emission of bright red fluorescence after it binds to DNA. Three days following injection with KA, high levels of oxidative stress were evident in wild-type mice along the neuronal layer of the CA1 and CA3 regions of the hippocampus (Fig. 2B). Oxidation of DHE was not detectable in tPA–/– mice injected with KA, indicating that tPA–/– neurons do not exhibit significant oxidative stress in response to KA. However, acute administration of tPA prior to the injection of KA restores both the oxidative stress in hippocampal neurons to wild-type levels and neurodegeneration (Fig. 2B-D). Finally, wild-type mice and tPA-injected tPA–/– mice, infused with 2 mM FeTMPyP or 10 mM FeTPPS two days prior to KA injection, exhibited minimal oxidative stress along the CA1 and attenuated oxidation in the CA3 hippocampal subfields (Fig. 2B,D). Moreover, low levels of fluorescence, indicative of oxidative stress observed in some CA3 FeTMPyP-treated neurons, coincided with the few degenerating neurons seen (Fig. 2A,B). These results demonstrate that infusion of 2 mM FeTMPyP or 10 mM FeTPPS (Fig. 2C) reduces the oxidative stress that occurs during KA-induced excitotoxicity and that the oxidative stress plays a key role in the neurodegenerative process.

The finding above suggests that oxidative stress is a functional contributor to neuronal death after excitotoxic injection and that inhibiting the formation of NO results in neuroprotection. To address this possibility, wild-type mice were infused with the general NOS inhibitor N^G-methyl-L-arginine (NMMA). Infusion of NMMA prior to KA injection conferred neuroprotection in a dose-dependent manner (Fig. 3A,C and Table 1). Significant protection was observed with a 5 μM infusion and almost complete protection was evident at 200 μM.

To ensure that the effects seen with NMMA were a result of eliminating NO, and not a result of inhibition of NOS function, the NO scavenger PTIO was infused prior to KA injection. PTIO scavenges NO generated, but does not inhibit NOS activity or alter blood flow (Tozer et al., 1997). Infusion of 300 μM PTIO prior to KA injection was sufficient to protect neurons significantly from degeneration (Table 1), indicating that alterations in blood flow (vasoconstriction or vasodilation) do not affect detectable neuronal cell death in this model.

To investigate whether the neurodegeneration seen in tPA^–/– mice following tPA and KA injection could result from NO formation, tPA^–/– mice were infused with NMMA prior to tPA and KA injection (Fig. 3B). Infusion of 200 μM and 400 μM NMMA prior to tPA and KA co-injection conferred neuroprotection in tPA^–/– mice (Fig. 3B,D).

Injection of high concentrations of KA (0.9 nmols) results in global hippocampal pyramidal neuronal death. By contrast, injection of lower concentrations (0.4-0.2 nmol) results only in limited excitotoxic neurodegeneration localized to specific regions of the hippocampus (Andersson et al., 1991). NOC-18, an NO donor, when injected at 1 mM concentration into wild-type mice, is not sufficient to cause neurodegeneration in the hippocampus (Fig. 4). However, NOC-18 injection prior to the injection of low concentrations (0.2-0.4 nmol) of KA increases neuronal death in a dose-dependent manner. Injection of NOC-18 prior to the injection of 0.9 nmol KA had no further effect on the subsequent neuronal degeneration, which was already encompassing the CA1-CA3 hippocampal subfields (data not shown). These results indicate that, although NO is not toxic by itself in wild-type mice, it is capable of exacerbating KA-induced excitotoxic injury, possibly by meeting a threshold level of NO required for toxicity to occur.

Injection of 0.9 nmol KA causes neurodegeneration in wild-type, but not in tPA-deficient, mice. To determine whether tPA was acting upstream or after the generation of NO, we delivered NO into the hippocampus by injecting the NO donor NOC-18 (Fig. 5). NOC-18 alone did not cause neurodegeneration in tPA^–/– mice. However, injection of NOC-18 prior to KA in tPA^–/– mice resulted in degeneration comparable with that seen in wild-type mice. Quantification using Scion imaging indicated a significant difference (P<0.001) between control and experimental groups of mice (data not shown). Infusion of FeTMPyP in tPA-deficient mice prior to KA and NOC-18 injections protected hippocampal neurons from excitotoxicity. Taken together, these findings demonstrate that ONOO^– mediates the toxic effects of NO in hippocampal neurodegeneration, and NO and ONOO^– are necessary downstream effectors of tPA in KA-induced neurotoxicity.

Since tPA, NO and ONOO^– have each been implicated in BBB breakdown, which in turn might affect neurodegeneration, we evaluated the individual effects of these components on BBB integrity by assessing the diffusion of Evans Blue dye in the brain parenchyma. The presence of Evans Blue after perfusion would indicate breakdown of the BBB. Mice were injected with 2% Evans Blue intravenously immediately after the injection of KA. To evaluate whether the BBB breakdown occurs specifically in the hippocampus after intrahippocampal KA injection, the hippocampus was removed and extravasated Evans Blue was determined solely in this region of the brain. Injection of KA causes breakdown of the BBB in the hippocampus of wild-type mice, but not in tPA-deficient mice (Fig. 6A). The BBB breakdown is mediated by the acute presence of tPA since infusion of tPA into the tPA^–/– mice prior to KA injection restored the extravasation of Evans Blue.

To determine if there is a link between the neuroprotective effect of NMMA, PTIO, FeTPPS and FeTMPyP with the breakdown of the BBB, wild-type mice were infused with these compounds prior to KA injection. Infusion of either 100 μM or 400 μM NMMA not only protected neurons from neurodegeneration, but also prevented BBB breakdown in the hippocampus (Fig. 6B). By contrast, infusion of 2 μM FeTMPyP or 10 μM FeTPPS allowed for BBB breakdown to occur, even while having neuroprotective effects. The results from this experiment indicate that neurodegeneration and BBB breakdown are independently mediated by NO and ONOO^–.

To determine if NO alone can mediate BBB breakdown, wild-type and tPA^–/– mice were injected with NOC-18 and disruption of the BBB was assessed by extravasation of Evans Blue (Fig. 6C). NOC-18 alone was unable to promote BBB breakdown in either wild-type or tPA^–/– mice. However, co-injection of NOC-18 with KA into tPA^–/– mice restored the ability of KA to mediate BBB breakdown, indicating that NO is downstream of tPA in promoting BBB breakdown (Fig. 6C).

Specific breakdown of the BBB in the hippocampus was also investigated by immunofluorescence using anti-occludin antibodies, since occludin is exclusively localized at tight junctions, which are important for maintaining BBB integrity (Hirase et al., 1997). On the uninjected side of KA-treated mice, occludin shows strong and typical staining of the endothelial layer at contact points between cells (Fig. 6D and Fig. S2, supplementary material). Similar immunolocalization can be observed in the KA-injected side of tPA^–/– mice, and in wild-type mice that were infused with 400 μM NMMA prior to KA injection. By contrast, the staining was decreased in wild-type mice injected with KA, and the decrease was not prevented by prior infusion of 2 μM FeTMPyP, confirming the results obtained by quantification of Evans Blue extravasation. We therefore conclude, that whereas NO might contribute to BBB breakdown, BBB breakdown is not crucial for the progression of neurodegeneration, as previously reported (Chen et al., 1999a).

The role of NO in neurotoxicity remains unclear (reviewed by Boje, 2004). The cellular source from which the NO is generated appears to be important in regulating whether the effects are degenerative or protective (Samadani et al., 1997). For instance, NO is produced in response to stimulation of the NMDAR, a subtype of glutamate receptors that leads to activation of nNOS (Garthwaite et al., 1989). nNOS is constitutively expressed and requires a rise in intracellular Ca²⁺ to become activated (Brown and Bal-Price, 2003; Sattler and Tymianski, 2000). In addition to the contribution of nNOS, endothelial cells are also capable of generating NO through endothelial NOS (eNOS). eNOS is also constitutively expressed and regulated by multiple mechanisms, including rises in intracellular Ca²⁺, mRNA stability, phosphorylation and protein-protein interactions (Endres et al., 2004; Fleming and Busse, 2003; Nathan and Xie, 1994). Activation of nNOS and the generation of neuronally derived NO has toxic consequences, whereas eNOS activation was reported to result in neuroprotection in a model of focal ischemia (Samadani et al., 1997). However, during KA-induced excitotoxicity, nNOS proteolysis and subsequent inactivation limits neuronal cell death in the hippocampus (Araujo et al., 2003).

Glia can also produce NO, most notably through the induction of inducible NOS (iNOS) (Brown and Bal-Price, 2003). Unlike nNOS and eNOS, iNOS is transcriptionally regulated and does not require increased intracellular Ca²⁺ levels to produce NO. Generally, iNOS is considered to be responsible for the high output of NO seen in various pathological states of CNS that involve inflammation. Microglia have been implicated in neuronal cell death, including excitotoxicity (Rogove and Tsirka, 1998), ischemia (Yrjanheikki et al., 1998) and multiple sclerosis (Carson, 2002). It has been suggested that microglia are neurotoxic through an NO mechanism (Chao et al., 1992). However, even in the case of glial-produced NO, the effects of NO are highly contextual. In Alzheimer's disease, β-amyloid has been shown to activate microglia (Meda et al., 1995). However, the role of NO produced by the microglia, once they are active, has been reported to be either neuroprotective (Troy et al., 2000) or degenerative (Xie et al., 2002).

Although the majority of toxic effects exhibited by NO have been attributed to ONOO^–, a similar discrepancy is found when looking at its toxic effects in different models. Specifically, whereas ONOO^– has been shown to be toxic to neurons treated with β-amyloid-activated microglia (Xie et al., 2002), it is protective to neurons against NO-mediated apoptosis (Garcia-Nogales et al., 2003). It is apparent that the role of NO and ONOO^– in neuronal injury is dependent on various factors including the cellular source, the method of inducing toxicity and the animal model. In the mouse model of tPA-dependent excitotoxicity after intrahippocampal KA injection, the effects of NO appear to be deleterious, as they result in production of ONOO^–. Our results show a relative increase in nTyr formation after KA injection in wild-type mice. This nTyr increase, which is not observed in tPA^–/– mice (Fig. 1A,B), although usually attributed to ONOO^–, is not completely specific for it, since the nitrite-hemeperoxidase pathway can also result in tyrosine nitration (Chao et al., 1992; Hazen et al., 1999). Scavenging ONOO^– by FeTMPyP attenuates the neurotoxic properties of KA and reduces the levels of oxidation (Fig. 2) and nTyr (Fig. S1, supplementary material). This function of FeTMPyP might be attributed to ONOO^– scavenging; however, it should be noted that FeTMPyP can also scavenge superoxide. Interestingly, infusion of 2 μM FeTMPyP along with KA injection occasionally causes very limited degeneration in the CA3 (Fig. 2). In these cases, the degenerating neurons coincide with those neurons exhibiting oxidative stress, suggesting that ONOO^– increases the levels of oxidative stress.

These results lead us to speculate that tPA has several roles in mediating NO-dependent excitotoxicity (Fig. 7). One possibility is that tPA might interact with the NMDAR on the surface of neurons. These receptors generate excitatory signals following glutamate stimulation. It has been reported that tPA can directly interact with the NMDAR and cleave the NR1 subunit, thus allowing for an increase in the influx of Ca²⁺ (Nicole et al., 2001). Since nNOS activation leads to the production of NO (Bredt and Snyder, 1990), and nNOS is closely linked to the NMDAR, it is possible that the interaction of tPA with the NMDAR can result in the production of NO. The production of NO ultimately leads to neuronal cell death through ONOO^– and induction of BBB breakdown. Our results show a transient increase in NO production 6 hours after KA injection, possibly mediated by modulation of nNOS activity by tPA. After 24 hours, a significant number of neurons in the hippocampus are dead or undergoing degeneration, potentially accounting for the sudden decrease in NOS activity at this time point. Supporting this idea, infusion of a tPA inhibitor into wild-type mice, or catalytically inactive tPA into tPA^–/– mice, abolishes increased NO production at 6 hours post-KA injection (Fig. 1B), suggesting that the catalytic activity of tPA is necessary for increased NO production during excitotoxicity. Not surprisingly, inhibition of the NMDAR by MK-801 also attenuates the increase in NOS activity seen at 6 hours (Fig. 1B). Although these experiments do not show a clear link between tPA and cleavage of the NR1 subunit, they do suggest that both the catalytic activity of tPA and the NMDAR are involved in increasing NOS activity during excitotoxicity. An alternative or parallel possibility could be that tPA might be cleaving plasminogen to plasmin, thus leading to increased NOS activity through unknown mechanisms. This possibility is supported by the finding that plasminogen-deficient mice also lack the increase of NOS activity at 6 hours post KA injection.

Additionally, tPA has been found to activate microglia following excitotoxic insult by interacting with annexin II on the surface of microglia (Rogove et al., 1999; Siao and Tsirka, 2002). Once microglia are activated, they can release various neurotoxic factors including NO (Chao et al., 1992). Attenuation of microglial activation provides neuroprotection from KA-induced excitotoxicity (Rogove and Tsirka, 1998). The iNOS found in microglia is transcriptionally regulated, does not require high levels of Ca²⁺, and is thought to be responsible for the high output of NO seen during inflammation (Nathan and Xie, 1994). In light of this evidence, we propose that KA induces the release of tPA (Baranes et al., 1998; Gualandris et al., 1996), which then interacts with the NMDAR (Nicole et al., 2001) to activate nNOS. Additional tPA produced by neurons can activate microglia through interactions with annexin II. Activated microglia produce large quantities of NO that, along with the NO produced from neurons, leads to the formation of ONOO^– and, ultimately, to additional neuronal cell death (Fig. 7). The increase in NOS activity seen at days 3 and 5 in wild-type mice, and not in tPA-deficient mice, suggests that glia might be acting as the principle source of NO at later time points, since microglial activation becomes maximal at these time points (Chao et al., 1992). Production of NO in glial cells correlates with the presence of tPA (Fig. 1A). Our data suggest that the production of NO occurs downstream of tPA secretion, since KA toxicity can be induced in tPA-deficient mice by the injection of NO (Fig. 5).

Both neurons and microglia can produce superoxide, which is required for ONOO^– formation (Colton and Gilbert, 1987; Heinzel et al., 1992; Kim and Ko, 1998; Liu et al., 2002; Xia et al., 1996). Once formed, ONOO^– is a powerful oxidant that can promote neurodegeneration (Trackey et al., 2001; Xie et al., 2002). This study indicates that ONOO^– plays an important part in mediating neuronal cell death after KA injection, since scavenging ONOO^– affords protection from excitotoxicity (Fig. 3). These results also offer an explanation to why inhibiting the formation of NO provides neuroprotection, since ONOO^– is a downstream metabolite of NO (Fig. 2). Further evidence shows that NO-dependent excitotoxicity is mediated by ONOO^– and is substantiated by the fact that NO alone is not toxic, but requires the additional presence of an injury factor (Fig. 4), as this injury possibly stimulates the generation of superoxide.

In addition to being associated with neurodegeneration, NO and ONOO^– have also been implicated in BBB breakdown (Mayhan, 2000; Mayhan and Didion, 1996; Tan et al., 2004). Our results show that NO contributes to BBB compromise during KA-induced excitotoxicity (Fig. 6). This property of NO might be attributed to its vasodilatory effects, and might be beneficial in certain circumstances, such as when it is necessary to increase blood flow after stroke. Of note is the fact that infusion of tPA alone not only increased NOS activity (Fig. 1A), but also caused an increase in BBB permeability (Fig. 6). These results are consistent with other studies that have indicated that tPA can modulate BBB permeability (Yepes et al., 2003). It is possible therefore that the modulation of BBB permeability by tPA occurs through the regulation of NO production. NO can also limit excitotoxicity by interacting with the NMDAR and directly preventing further activation. Therefore, NO might not itself be neurotoxic, but rather might be beneficial in limiting toxic effects during excitotoxicity.

Our results also support evidence that BBB breakdown does not necessarily result in neuronal cell death. This is evident in Fig. 3, in which scavenging ONOO^– provides neuroprotection, but is unable to prevent BBB disruption (Fig. 6). If BBB disruption had a direct link to neuronal damage, then BBB breakdown should have resulted in toxicity even in the presence of FeTMPyP or FeTPPS. Previous cell culture studies have shown that NO mediates BBB breakdown through ONOO^– (Mayhan, 2000; Tan et al., 2004). In the whole animal, ONOO^– might not have an opportunity to interact with the BBB. ONOO^– has a short half-life (Beckman et al., 1990), reacts quickly with lipids (Radi et al., 1991; Rubbo et al., 1994) and might influence protein function by nitrating tyrosine (Blanchard-Fillion et al., 2001; MacMillan-Crow et al., 1998; Nielsen et al., 2004). If the ONOO^– is formed inside the cells, it might cause DNA damage (Estevez et al., 1995), ultimately leading to apoptosis. For these reasons, it seems more likely that whereas NO contributes to BBB breakdown, ONOO^– contributes to neuronal cell death. However, ONOO^– appears to be important in mediating BBB breakdown in disease models in which a hyperactivated immune system could provide large quantities of NO and ONOO^– over extended periods of time (Hooper et al., 2000; Kean et al., 2000). It is important to note that, although NO contributes to BBB breakdown, blocking NO production with 400 μM NMMA does not completely abolish Evans Blue extravasation. Therefore, other factors in addition to NO, such as tPA, facilitate BBB disruption. In conclusion, our results show that interfering with the production of NO or ONOO^– in vivo might aid in suppressing neurotoxicity mediated by KA injection. Furthermore, we have shown for the first time a link between the modulatory effects of tPA on excitotoxicity and the production of NO and ONOO^–.

All experiments with animals followed National Institutes of Health guidelines and were approved by the Department of Laboratory Animal Research at the State University of New York (SUNY, Stony Brook, NY). tPA-deficient (tPA^–/–) (Carmeliet et al., 1994) and C57BL6 (wild-type, wt) mice had ad libitum access to food and water and were placed on a 12 hour light/dark schedule. The tPA^–/– mice used had been backcrossed for more than 12 generations into the C57BL6 background.

For intrahippocampal injections, adult mice were injected as described earlier (Tsirka et al., 1995; Tsirka et al., 1996) with 0.9 nmol kainate (KA; Sigma), unless otherwise stated. Mice were sacrificed at the indicated times after injection, and brains were removed, frozen and sectioned. For infusion experiments, 7-day microosmotic pumps (Durect) were subcutaneously inserted (Tsirka et al., 1995; Tsirka et al., 1996) after being filled with 1 μg/μl tPA, 1 μg/μl S481A non-enzymatic tPA (Molecular Innovations), 2,7-bis-(4-amidinobenzylidene)-cycloheptanone-1-one dihydrochloride (tPA Stop; American Diagnostica), N^G-methyl-L-arginine (NMMA; Sigma), 5,10,15,20-tetrakis(N-methyl-4′-pyridyl)porphinato iron (III) chloride (FeTMPyP; Calbiochem), 5,10,15,20-tetrakis(N-methyl-4′-pyridyl)porphinato tetra(p-toluenesulfonate) (TMPyP; Sigma,), 5,10,15,20-tetrakis(4-sulfonatophenyl)porphyrinato iron (III) chloride (FeTPPS; Calbiochem), or 300 μM 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (PTIO; Calbiochem). Two days after the onset of infusion, the mice were injected with kainic acid (KA). Mice were sacrificed at day 7 after pump insertion unless otherwise specified. 1 mg/kg of 5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate (dizocilpine hydrogen maleate MK-801; Sigma) was injected intraperitoneally 30 minutes prior to KA injection. tPA at a concentration of 5 μg/μl was also injected directly into the hippocampus prior to KA injection where indicated.

Wild-type or tPA^–/– mice were anesthetized as above and injected with the indicated concentrations of the NO donor 2,2′-(hydroxynitroshydrazino)bis-ethanamine (NOC-18; Calbiochem) in 500 nl at the previously stated coordinates. If the injection was in combination with KA, NOC-18 was injected first and allowed to dissipate for 2 minutes. KA was then injected as previously described.

Fresh-frozen coronal sections (20 μm) cut using a Leica cryostat were mounted onto slides, air dried and postfixed in 4% paraformaldehyde/PBS. Sections were stained with Cresyl Violet, as described (Tsirka et al., 1995; Tsirka et al., 1996). The stained sections were photographed using a SPOT RT camera with a Nikon Eclipse E1600W microscope. Neuronal survival was quantified using the freeware Scion Image beta 4.02 (Scion). In brief, TIF files were loaded into Scion Image and the freehand tool used to define the area of pyramidal neuronal loss. The entire length of the hippocampal pyramidal layer was measured in the same manner using arbitrary units, and the percentage loss of neurons on the injected side were calculated. To ensure accuracy, the process was repeated five times per section and the average percentage taken as the final measurement. No measured percentage had a standard deviation greater than 2. A minimum of five sections was quantified for each experimental point. A minimum of 6-8 different mice was used for each quantification to ensure measurements along the entire A/P axis of the hippocampus.

Wild-type mice, tPA^–/– mice, and wild-type mice infused with 2 μM FeTMPyP were injected with KA as described above. Two hours prior to sacrifice, the mice were injected intraperitoneally with 1 mg/ml dihydroethidium (DHE) (Molecular Probes). In the presence of oxidative stress, DHE becomes oxidized to ethidium (Melendez et al., 1999; Scanlon and Reynolds, 1998), binds to DNA, and emits a red fluorescent signal. Brains were sectioned and immediately visualized using a Nikon Eclipse E1600W microscope. Fluorescent intensity was determined using Scion Image beta 4.02 by quantifying the color density of the injected hippocampal side after subtraction of the color density on the non-injected side (Wang and Tsirka, 2005).

Following treatment, KA-injected wild-type mice, tPA^–/– mice, and tPA^–/– mice infused with 1 μg/μl tPA were sacrificed at the indicated time points and the brains removed. The hippocampus of each animal was immediately dissected out and frozen at –80°C. NOS activity was determined using the NOS Assay Kit (Cayman Chemical) according to the protocol described by the manufacturer. In brief, hippocampal tissue was homogenized and incubated with reaction buffer for 30 minutes in the presence of [³H]L-arginine. NOS activity was determined by loading the reaction mix on a resin that binds to positively charged [³H]L-arginine. Following removal of [³H]L-citrulline, [³H]L-arginine was eluted from the resin using elution buffer. NOS activity was determined as the percentage conversion of [³H]L-arginine to [³H]L-citrulline catalyzed by NOS (percentage of amount of [³H]L-arginine loaded on the column minus the amount of [³H]L-arginine eluted). The specificity of the reaction was confirmed by inhibition by L-NAME. NOS activity was expressed as units (nmole of NO generated per min) per milligram of protein in each homogenate.

Fresh-frozen coronal sections (20 μm) were postfixed in 4% paraformaldehyde/PBS. Endogenous peroxidase activity was quenched with peroxide treatment (3% H₂O₂ in PBS, 30 minutes). After blocking in serum of the host of the secondary antibody [5% serum in PBS-T (0.5% Triton X-100 in PBS)] with or without nTyr to test for antibody specificity, the primary anti-nTyr antibody (Molecular Probes or Cayman Chemical) was added at a 1:100 dilution. Sections were incubated in primary antibody overnight at 4°C. After washing for 10 minutes in PBS, appropriated biotinylated secondary antibodies were added in serum/PBS-T and incubated for 30 minutes. After washing in PBS for 10 minutes, the ABC reagent was added (Vector Laboratories) according to the manufacturer's directions for 30 minutes. The sections were washed for 10 minutes and the signal was visualized using DAB/H₂O₂. Sections were successively dehydrated in ethanol, defatted in xylene and mounted with Permount.

Three days post KA injection, wild-type and tPA^–/– mice were sacrificed, the brains removed, and the hippocampi dissected and divided into ipsilateral (injected) and contralateral (non-injected) sides. The tissues were homogenized in 0.25% Triton X-100 in PBS containing one tablet of complete mini-protease inhibitors (Roche) per 10 ml of buffer. After centrifugation to remove debris, total protein content was measured using the Bio-Rad Bradford detergent-compatible (DC) assay. Equal amounts of protein (25 μg) were loaded onto a 10% SDS-PAGE gel, transferred onto polyvinylidene difluoride membrane, and blocked overnight in 5% non-fat milk in PBS-T. The membrane was then incubated with anti-nTyr antibody (1:1000 dilution) in milk/PBS-T. Horseradish peroxidase (HRP)-conjugated secondary antibody (Jackson Immunoresearch) was used (1:2000 dilution) and detected with LumiGLO Chemiluminescent Substrate System (KPL).

BBB breakdown was assessed as described elsewhere (Yepes et al., 2003). In brief, tPA^–/– and wild-type mice were injected with KA and immediately afterwards injected intravenously with a solution of 2% Evans Blue (Sigma). The mice were then sacrificed by transcardiac perfusion 1 day later. The hippocampi were removed, divided into ipsilateral (injected) and contralateral (non-injected) hemispheres, and weighed. The hemispheres were homogenized in 0.5% Triton X-100 in PBS and centrifuged at 21,000 g for 30 minutes. The amount of Evans Blue in the supernatant was quantified at 620 nm, subtracted from the background, and divided by the wet hemisphere brain weight. Values are presented as a percentage of the total signal and represent average values for a minimum of five mice per experimental group.

tPA^–/– mice, wild-type mice and wild-type mice infused with NMMA or FeTMPyP were injected with KA. Sections (20 μm) were fixed in 95% ethanol for 30 minutes at 4°C followed by an additional 3 minutes incubation with acetone. After blocking in serum (5% serum in PBS-T), sections were incubated with 3 μg/ml mouse anti-occludin (Zymed) overnight at 4°C. After washing for 10 minutes in PBS, FITC-goat anti-mouse secondary antibody was added for 1 hour at room temperature. Sections were washed twice for 10 minutes in PBS and mounted with Vectashield mounting medium (Vector Laboratories).

Statistical analysis was performed using one-way ANOVA followed by a Bonferroni-Dunn test for multiple comparisons within a group, or a two-tailed t test for comparisons between groups, as indicated by the figure legends; P<0.05 was considered significant and is marked by * or ^‡; P<0.01 and P<0.001 were considered very significant and are marked by ** or ***, respectively. All results are represented as average, with error bars indicating the standard error of the mean (s.e.m.). In all experiments, n refers to the number of animals used for each genotype or condition. For each experiment, a minimum of six mice were used.

The authors thank members of the Tsirka laboratories as well as Howard Crawford and Michael Frohman for helpful suggestions. This work was supported by NIH grant to S.E.T. and a Turner Fellowship to S.R.P.