Stroke is one of the leading public health problems and the stroke-related death is expected to be doubled in the next decades due to an increase in aged populations all across the globe. Quite a number of therapeutic agents developed in pre-clinical studies have miserably failed in clinical trials, revitalizing the long-held belief that stroke is an incurable disease (Feuerstein and Chavez 2009). Fortunately, the success obtained with recanalization therapies still keeps the door open for optimism (Donnan et al. 2011). Thrombolysis trials encompassing thousands of patients have unambiguously demonstrated the presence of a salvageable penumbra after arterial occlusions (Donnan et al. 2009; Heiss 2011). Imaging studies have illustrated that the penumbral tissue can be as large as two third of the middle cerebral artery (MCA) territory and may remain viable for a couple of hours although rapidly vanishes over time (Ma et al. 2009; Lansberg et al. 2011). These encouraging developments strongly suggest that neuroprotection could indeed be a attainable option. Currently, the best conceivable way of neuroprotection for acute ischemic stroke appears to be restoration of blood flow to the ischemic area by thrombolysis. Unfortunately, a short therapeutic time window limits the use of recanalization therapies (Lees et al. 2010; Donnan et al. 2011). Only a small proportion of acute ischemic stroke patients receive thrombolysis because most of them arrive at hospital more than 4.5 h after the ictus. One of 4.5 patients benefits from i.v. tissue plasminogen activator (tPA) treatment when given within 90 min of the stroke onset, whereas only one out of 14.9 patients' neurological status improves when tPA is administered between 3 and 4.5 h (Donnan et al. 2011). This decline in tPA's effectiveness is generally attributed to rapid loss of viability of the penumbral tissue. However, a low recanalization rate (Saver 2011) as well as incomplete microcirculatory reperfusion (no-reflow) despite satisfactory recanalization may also contribute as recent experimental and clinical findings suggest (De Silva et al. 2009; Yemisci et al. 2009; Soares et al. 2010; del Zoppo and Hamann 2011). Moreover, thrombolysis is complicated by hemorrhage and edema due to increased vascular permeability in about 6% of patients receiving i.v. tPA (Donnan et al. 2011). These drawbacks of thrombolysis point out that we need to better understand the events associated with reperfusion for improving the effectiveness and decreasing the unwanted effects of recanalization therapies. Here, we review the potential mechanisms leading to reperfusion-induced blood–brain barrier (BBB) permeability increase and to incomplete reperfusion after recanalization. Interestingly, both mechanisms appear to share common toxic pathways and, therefore, similar treatment approaches may significantly improve the outcome of recanalization therapies by improving microcirculatory reflow as well as preventing hemorrhagic conversion and vasogenic edema.
Currently, the best way of neuroprotection for acute ischemic stroke appears to be restoration of blood flow to the ischemic area by thrombolysis. Unfortunately, a short therapeutic time window as well as thrombolysis-induced bleeding and edema limit the use of recanalization therapies. Here, we review the evidence suggesting that ischemia/reperfusion-induced microvascular injury plays a critical role in determining tissue survival after recanalization in focal cerebral ischemia by disrupting the blood–brain barrier integrity and promoting microcirculatory clogging. Among many complex mechanisms of the ischemia–reperfusion injury, overproduction of oxygen and nitrogen radicals on the microvascular wall appears to significantly contribute to these pathological processes. These developments bring about the exciting possibility that effective suppression of oxidative/nitrative stress during pharmacological or interventional re-opening of the occluded artery may significantly improve the outcome of recanalization therapies in stroke patients by improving microcirculatory reflow as well as by preventing hemorrhagic conversion and vasogenic edema. They also point to the critical (but partly neglected) importance of the microcirculation in neuroprotection.
endothelial nitric oxide synthase
middle cerebral artery
neuronal nitric oxide synthase
tissue plasminogen activator
Neurovascular unit and reperfusion injury
The neurovascular unit limits and regulates passage of blood constituents to brain parenchyma and, plays a role in coupling the metabolic demand created by neuronal activity with blood flow (Iadecola 2004; Abbott et al. 2006; Attwell et al. 2010). Majority of the nutrient and waste exchange takes place at microcirculation except the capillary-free peri-arteriollar areas (Krogh cylinders; Hudetz 1997; Harrison et al. 2002; Kasischke et al. 2011). The total length of microvessels present in 1 mm3 of human cortex is about 50 cm and creates a surface of about 10 cm2 (Lauwers et al. 2008). Therefore, the immense microcirculatory network (Fig. 1) has a huge surface area, which consumes considerable amounts of oxygen and produces nitric oxide (NO) although the quantity of both are smaller compared with those used/produced by the brain parenchyma (Wei et al. 1999; Pacher et al. 2007). At microcirculatory level, the neurovascular unit is composed of the vascular endothelium, tight junctions among endothelia, basal membrane surrounding the endothelia, pericytes located at the periphery between layers of the basal lamina and, the astrocyte endfeet encircling the microvessel (Fig. 2) (Iadecola 2004). Endothelia, pericytes and astrocyte endfeet harbor several mitochondria to match the high metabolic demand (Oldendorf et al. 1977; Ruetzler et al. 2001; Gursoy-Ozdemir et al. 2004; Mathiisen et al. 2010). During ischemia but especially at reperfusion when more O2 becomes available, mitochondria produce excess O2− (Adam-Vizi 2005; Fraser 2011). Another important source of oxygen radicals in the vascular wall during ischemia/reperfusion (IR) is enzymes such as NAPDH oxidase that is highly expressed in the cerebrovascular endothelium (Kahles et al. 2007; Chrissobolis and Faraci 2008; Miller et al. 2009). Endothelial nitric oxide synthase (eNOS), which is also abundant in the endothelium, becomes a continues source of NO as long as intracellular Ca+2 levels are high during IR unlike physiological conditions in which eNOS is only briefly activated with transient intracellular Ca+2 rises (Wei et al. 1999; Guix et al. 2005; Pacher et al. 2007; Fraser 2011). However, Ca-independent NO production by inducible NOS becomes noticeable 12 h after reperfusion on induction of inducible nitric oxide synthase in vascular cells and, plays role in post-stroke inflammation (Iadecola et al. 1996). O2− and NO have very high affinity to each other such that NO outcompetes superoxide dismutase and reacts with O2− before it is dismutated to hydrogen peroxide (H2O2) (Pacher et al. 2007). Hence, concomitant generation of NO and O2− leads to high throughput formation of the strong oxidant peroxynitrite (ONOO−). ONOO− is believed to mediate an important part of the radical toxicity during IR because it is a much stronger oxidant than NO and O2− alone and, has a longer tissue half life (Pacher et al. 2007). Supporting this view, we observed that O2− producing foci in the ischemic brain were all 3-nitrotyrosine immunopositive, indicating that O2− was converted to ONOO− by interacting with NO that spreads freely and rapidly across cell membranes when we examined O2− and ONOO− production with fluorescent probes in the intact mouse brain subjected to 2 h of MCA occlusion and 6 h of reperfusion (Gursoy-Ozdemir et al. 2004) (Figs 2 and 3). In these studies, 3-nitrotyrosine immunolabeling, a footprint of ONOO− illustrating the nitrated tyrosine residues on proteins, was most prominent at microvasculature although it was also evident in larger vessels and all across the parenchyma (Gursoy-Ozdemir et al. 2000, 2004) (Fig. 3). We attribute this intense immunoreactivity in microvessels to a larger surface to volume ratio in smaller vessels because the volume determining the radical concentration changes with third power of the radius, whereas the vascular surface producing the radicals varies with its second power.
The parenchymal source of NO leading to ONOO− formation is neuronal NOS (nNOS) (Dalkara and Moskowitz 1994; Huang et al. 1994; Irikura et al. 1995; Gursoy-Ozdemir et al. 2000). nNOS expressing neurons are intermittently spaced to supply an area equaling to NO's diffusion limit determined by its rapid removal with hemoglobulin in circulating erythrocytes (Guix et al. 2005; Pacher et al. 2007). Inhibition of parenchymal NO synthesis with 7-nitroindazole, an in vivo selective inhibitor of nNOS reduces the infarct volume when given prior to ischemia (Dalkara et al. 1994; Gursoy-Ozdemir et al. 2000) (Fig. 4). Interestingly, 7-nitroindazole was not protective when administered at reperfusion following 2 h of ischemia, whereas inhibition of eNOS at reperfusion with Nω-nitro-L-arginine (L-NA), a non-selective NOS inhibitor was neuroprotective (Gursoy-Ozdemir et al. 2000). This pharmacological evidence suggests that NO generated on the vascular wall plays a crucial role in determining fate of the reperfused tissue. Of note, NOS inhibition has a bimodal effect: whereas partial inhibition provides neuroprotection by reducing excess NO and ONOO− formation without suppressing basal NO supply, further inhibition of NOS increases the infarct size (similar to development of larger infarcts in eNOS knockout mice) because a basal NO synthesis is required for maintaining the normal circulatory and vascular physiology (Huang et al. 1996; Margaill et al. 1997). Importantly, partial inhibition of eNOS with a low dose of l-N5-(1-iminoethyl)-ornithine, a NOS inhibitor that does not cross the BBB provided neuroprotection comparable to those obtained with a BBB-permeable NOS inhibitor L-NA or O2− scavenger N-tertbutyL-α-phenylnitrone (PBN) (Fig. 4), suggesting that vascular protection during reperfusion can provide an efficient neuroprotection by itself (Yemisci et al. 2009). Supporting this idea, S-PBN, which is not BBB permeable unlike PBN, was reported to be equally neuroprotective to PBN in a rat embolic stroke model (Yang et al. 2000). Microvessels had an almost normal histology in sections obtained from animals subjected to 2 h MCA occlusion and treated with NOS inhibitors or O2− radical scavengers at reperfusion (Yemisci et al. 2009) (Fig. 4). Suggesting an association between peroxynitrite and microvascular injury, we showed by biochemical analyses that NOS inhibitors [L-NA, l-N5-(1-iminoethyl)-ornithine] as well as O2− scavengers (PBN) at doses used in the above studies all efficiently inhibited ONOO− formation in the brain tissue and albumin extravasation (Gursoy-Ozdemir et al. 2000, 2004; Yemisci et al. 2009).
Experimental studies have consistently shown that oxygen and nitrogen radicals generated in excess amounts during reperfusion play an important role in loss of BBB integrity and consequent vasogenic edema and hemorrhagic transformation (Wei et al. 1999; Gursoy-Ozdemir et al. 2000, 2004; Pacher et al. 2007; Fraser 2011). Radical species activate matrix metalloproteinases (MMPs) or directly modify the structure and localization of tight junction proteins, both of which lead to an increase in BBB permeability (Lochhead et al. 2010; Willis et al. 2010; Lehner et al. 2011). Activated MMPs degrade extracellular matrix of vascular wall as well as tight junction proteins, whereas chemical modification of occludin or claudin-5 causes loss of their function and strategic location (Rosenberg et al. 1996; Sumii and Lo 2002; Haorah et al. 2007; Schreibelt et al. 2007; Candelario-Jalil et al. 2009; Lochhead et al. 2010). Immunohistochemical stainings have shown that MMP-9 expression after IR is especially intense on microvessels, which is reduced by anti-radical agents or in animals over-expressing superoxide dismutase (Gursoy-Ozdemir et al. 2000; Asahi et al. 2001; Aoki et al. 2002; Maier et al. 2006). In addition to the radical-mediated toxicity detected in stroke models based on mechanical occlusion and re-opening of the MCA, reperfusion obtained with tPA in embolic-clot models inflicts further damage on BBB possibly owing to activation of MMP-9 by tPA as demonstrated by Eng Lo and his colleagues (Aoki et al. 2002; Wang et al. 2003; Zhao et al. 2007). In parallel with clinical observations (Thomalla et al. 2007; Donnan et al. 2011), tPA-treated animals exhibit more hemorrhagic transformation and brain edema, which are reduced with anti-radical therapies or MMP inhibitors (Aoki et al. 2002; Sumii and Lo 2002). There are also experimental findings suggesting that chronic hypertension may aggravate hemorrhagic transformation and stroke outcome by altering tight junction proteins (Hom et al. 2007). Finally, recently identified reversible opening of the BBB after embolic occlusions that allows transmigration of emboli across microvessels may also be one of the mechanisms contributing to the BBB leakiness after IR (Lam et al. 2010). The latter phenomenon is also thought to be mediated by MMPs.
Microcirculatory no-reflow after recanalization
An impaired reflow due to loss of microvascular patency (no-reflow phenomenon) was first noted following global cerebral ischemia (Ames et al. 1968) and subsequently after focal ischemia (Crowell and Olsson 1972; Little et al. 1975, 1976). Starting as soon as 1 h after MCA occlusion, capillaries show constrictions and luminal narrowing whereas pre-capillary arterioles remain open (Little et al. 1976; Belayev et al. 2002). Narrowed capillary lumina are filled with entrapped erythrocytes, leukocytes and fibrin-platelet deposits (Little et al. 1976; del Zoppo et al. 1991; Garcia et al. 1994; Garcia 1997; Choudhri et al. 1998; Zhang et al. 1999b; Morris et al. 2000). Clogged capillaries are not adequately perfused as demonstrated by carbon tracers and fluorescent-labeled intravascular markers (Anwar et al. 1988; del Zoppo et al. 1991; Belayev et al. 2002). The increased microcirculatory resistance is further aggravated by obstructions in post-capillary venules, of which lumina are surrounded with adherent leukocytes and may contain aggregates made of fibrin, platelets and leukocytes (Hallenbeck et al. 1986; del Zoppo et al. 1991; Garcia et al. 1994; Ritter et al. 2000; Belayev et al. 2002).
Incomplete restoration of the microcirculatory blood flow may negatively impact tissue recovery if re-opening of the occluded artery is achieved when there is still salvageable penumbral tissue. Improving microcirculatory reperfusion, therefore, appears to be a promising strategy for recanalization therapies. In fact, despite the lack of direct evidence connecting the microvasculature obstructions to the evolution of tissue injury, there is quite a number of studies supporting this view. For example, a reduction in microtubule-associated protein-2 immunoreactivity, a marker of neuronal injury, was reported to be spatially linked to local microvascular obstructions after 1 h of MCA occlusion (Zhang et al. 1999a). Secondary microvascular obstructions has also been shown to contribute to the evolution of selective neuronal necrosis after temporary carotid occlusion in Mongolian gerbils (Ito et al. 2011). Pharmacological agents and genetic manipulations reducing microvascular clogging by inhibiting leukocyte adherence, platelet activation or fibrin-platelet interactions have been shown to restore microcirculation and improve stroke outcome in animal models (Mori et al. 1992; Choudhri et al. 1998; Abumiya et al. 2000; Belayev et al. 2002; Ishikawa et al. 2005), raising the exciting possibility that anti-thrombotic agents could perhaps reduce microcirculatory thrombosis after recanalization in stroke patients. Finally, as noted above, a BBB-impermeable NO synthase inhibitor (l-N5-(1-iminoethyl)-ornithine) decreases microvascular clogging and reduces the infarct volume, reinforcing the idea that restoring microvascular patency may improve stroke outcome without direct parenchymal neuroprotection (Yemisci et al. 2009).
Narrowing of the capillary lumen is generally attributed to compression by swollen astrocyte endfeet encircling microvessels (Little et al. 1976; Garcia et al. 1994). Recently, we reported another mechanism mediated by pericytes that might also be critical in tissue survival after reperfusion (Yemisci et al. 2009). We observed that about half of the pial microvessels was not sufficiently reperfused after recanalization following 2 h of MCA occlusion. Pericytes on microvessels appeared to play an important role in this incomplete microcirculatory reflow because they contracted during ischemia and remained contracted despite reopening of the occluded artery (Yemisci et al. 2009) (Fig. 5). Pericytes are located at periphery of the microvessel wall and wrap it with their processes (Mathiisen et al. 2010; Winkler et al. 2011). They communicate with other cells of the vascular wall and regulate several microcirculatory functions in addition to their role in angiogenesis (Attwell et al. 2010). Recently, pericytes have been shown to contract in response to vasoactive stimuli and change the capillary diameter (Peppiatt et al. 2006; Fernandez-Klett et al. 2010). In fact, ischemia-induced segmental constrictions giving a sausage-like appearance to microvessels are more compatible with an active contraction rather than passive compression by the astrocyte endfeet covering capillaries along their longitudinal course. Supporting this idea, we found that the constricted nodes were co-localized with the contracted pericytes that were intermittently located along microvessels (Little et al. 1976; Garcia et al. 1994; Yemisci et al. 2009; Fig. 5). Although pericyte contractions did not obliterate the lumen, small decreases in capillary radius caused by pericyte contractions led to erythrocyte entrapments because capillary luminal size hardly allows passage of erythrocytes even under normal physiological conditions (Yemisci et al. 2009; Hamilton et al. 2010) (Fig. 6). Erythrocyte accumulations may also hinder passage of other blood cells and promote platelet aggregation together with fibrin, as commonly observed in ischemic microcirculation (Garcia et al. 1994; del Zoppo and Hamann 2011). Pericyte contraction together with erythrocyte entrapments first appears 1 h after MCA occlusion in the mouse cortex and increases over time but is not reversed after recanalization (Yemisci et al. 2009; Dalkara et al. 2011). Considering the fact that the level of local cerebral blood flow within the first hour of ischemia and reperfusion is the most critical factor determining tissue survival (Zhao et al. 1997), the failure of erythrocyte circulation within part of the microvessels is expected to unfavorably effect the survival of reperfused tissue. Interestingly, we found that the unregulated pericyte contraction could be restored by NOS inhibitors and O2− scavengers indicating that oxygen and nitrogen radicals that are intensely generated on the microvascular wall during IR play an important role in pericyte dysfunction (Yemisci et al. 2009; Fig. 6, see also Fig. 2). Restoration of microcirculatory patency with these agents administered during recanalization was correlated with the size of tissue surviving, suggesting that microcirculatory obstructions negatively impact recovery after recanalization (Fig. 6). In addition to changing the capillary diameter with various chemical stimuli originating from neighboring astrocytes and neurons (Peppiatt et al. 2006; Hamilton et al. 2010), pericytes also play an important role in maintenance of the BBB integrity (Armulik et al. 2010; Winkler et al. 2011). The IR-induced pericyte dysfunction may therefore contribute also to BBB leakiness mediated by MMPs. However, the potential role of pericytes in incomplete microcirculatory reperfusion as well as loss of the BBB integrity and, finally, the clinical significance of partial no-reflow after focal ischemia requires direct experimental evidence and evaluation in stroke patients with imaging techniques.
In conclusion, the above reviewed evidence suggests that IR-induced vascular injury plays a critical role in determining tissue survival after recanalization in focal cerebral ischemia by disrupting the BBB integrity and leading to microcirculatory clogging. Among many complex mechanisms of the IR injury, overproduction of oxygen and nitrogen radicals on the microvascular wall appears to significantly contribute to these pathological processes. Accordingly, free radical scavengers administered during recanalization of the occluded artery may improve stroke outcome by decreasing BBB disruption and restoring microcirculatory flow. Restitution of microcirculation along with recanalization positively impacts tissue survival and, hence, may increase the number of patients who benefit from thrombolytic treatment, whereas preservation of the BBB integrity may reduce complications like brain swelling and hemorrhage. These developments also point to critical importance of the microcirculation for a successful neuroprotection. Therefore, without being discouraged by previous negative clinical trials, we should continue working on translating these experimental finding to clinical practice.
Authors are grateful to Kivilcim Kilic for her expert help with preparing Fig. 2.
Conflicts of interest
Turgay Dalkara's work is supported by the Turkish Academy of Sciences. Authors have no conflict of interest.