Introduction Ischemic stroke has outstanding medical relevance as it is the second leading cause of death in industrialized countries [1]. Due to the aging of the population, the incidence of stroke is projected to rise even further in the future [2]. Despite tremendous research activity, with more than 100 clinical trials in human stroke patients [3], only one therapy approved by the United States Food and Drug Administration is available, i.e., thrombolysis using recombinant tissue plasminogen activator (rt-PA). However, the efficacy of rt-PA on functional outcomes is moderate at best, and more than 90% of all stroke patients must be excluded from rt-PA treatment because of over 25 labeled contraindications. Therefore, there is an unmet need for more effective therapies in acute stroke. Although a plethora of drugs for the treatment of acute stroke are effective in animal models, their translation into clinical practice has completely failed [3],[4]. As a result, many pharmaceutical companies have withdrawn from drug discovery in this area. To overcome this lack of clinically effective neuroprotective drugs, innovative strategies are urgently needed to identify pathways that can be targeted with innovative therapies [5]. Higher quality study designs are also required [6],[7]. One such high-potential pathway in ischemic stroke may be the occurrence of oxidative stress, i.e., the increased occurrence of reactive oxygen species (ROS) above physiological levels. Oxidative stress has been suggested for many years to cause tissue damage and neuronal death. The toxicity of ROS can be further increased by nitric oxide to produce reactive nitrogen species such as peroxynitrite (ONOO−), a molecule that causes oxidation and nitration of tyrosine residues on proteins [8]. Disappointingly, there is no conclusive evidence of a causal link between oxidative stress and the development of disease, and there is no successful therapeutic application targeting oxidative stress. To date, clinical attempts to scavenge ROS by applying antioxidants did not result in clinical benefit [9] or even caused harm [10],[11]. However, the characterization of the relevant enzymatic sources of oxidative stress may allow therapeutic targeting of oxidative stress by preventing the formation of ROS in the first place, instead of scavenging ROS after they have been formed. A potential source of ROS are NADPH oxidases, the only known enzyme family that is only dedicated to ROS production [12]. Four rodent genes of the catalytic subunit NOX, Nox1, Nox2, Nox3, and Nox4, have been identified, of which Nox1, Nox2, and Nox4 are expressed in the vasculature. NOX4 is the most abundant vascular isoform; its expression is even higher in cerebral than in peripheral blood vessels [13] and, further, induced in stroke [14]. Therefore, we hypothesized that NOX4 is the most relevant source of ROS in stroke. To test this hypothesis, we generated constitutively NOX4-deficient (Nox4 −/−) mice and directly compared them to NOX1-deficient (Nox1y/ −) and NOX2-deficient (Nox2y/ −) mice. NOX4 has been implicated in the regulation of systemic and hypoxic vascular responses. Therefore, we had to exclude systemic vascular effects of NOX4 deletion on blood pressure, which may affect stroke outcome independent of a specific neuronal or neurovascular mechanism. Finally, to examine the therapeutic potential of NOX4 as a drug target, we infused the specific NADPH oxidase inhibitor VAS2870 [15] after ischemia, thus mirroring the clinical scenario. Results NOX4 Is Induced during Ischemic Stroke in Mice and Humans Because NOX4 mRNA is expressed at higher levels in cerebral than in peripheral blood vessels [13] and is induced in stroke [14], we first sought to validate these data not only at the mRNA but also at the protein level. In all experiments, we followed current guidelines defining methodological standards for experimental stroke studies [4],[6],[7],[16],[17]. Here we chose a model of acute ischemic stroke in which mice are subjected to transient middle cerebral artery occlusion (tMCAO). This disease model is thought to involve oxidative stress and an induction of Nox4 expression [18]. Indeed, expression of NOX4 mRNA was significantly higher 12 h and 24 h after tMCAO in the basal ganglia and neocortex of wild-type mice than in sham-operated controls, in which basal NOX4 expression was low (Figure 1A). This result was validated by immunohistochemistry using a specific NOX4 antibody. We detected a stronger staining in neurons and cerebral blood vessels in wild-type mice subjected to tMCAO than in sham-operated controls. Although immunohistochemistry is not quantitative, this finding suggests higher levels of NOX4 protein (Figure 1B). Importantly, NOX4 staining was also stronger in brain samples from stroke patients. Although NOX4 was barely detectable in healthy brain regions, clear positive labeling of NOX4 was seen in neurons and vascular endothelial cells from the forebrain cortex of stroke patients. This finding was confirmed by double labeling for NeuN (a neuronal marker) or von Willebrand factor (an endothelial marker) and NOX4 in brain tissue (Figure 1B). These data indicate that NOX4 protein is induced during brain ischemia in mice, and this observation would be in agreement with a major functional role for NOX4 in ischemic stroke. Our limited observations in a small number of human cases provide some support to the hypothesis that these processes are also important in human stroke. 10.1371/journal.pbio.1000479.g001 Figure 1 Induction of NOX4 expression after ischemic stroke in mice and humans. (A) Relative gene expression of Nox4 in the ischemic basal ganglia (left) and cortex (right) of wild-type mice after sham operation and 4 h, 12 h, and 24 h after tMCAO (n = 5). *, p 0.05; Figure 2A). Ischemic stroke is usually a disease of the elderly and, consequently, one should verify any stroke-protective effects observed in young adult laboratory animals also in an older cohort [4],[16],[17]. Indeed, 18- to 20-wk-old Nox4 −/− mice also developed significantly smaller brain infarctions (27.8±15.1 mm3 versus 81.8±19.0 mm3, respectively) and less severe neurological deficits than age-matched controls, thereby confirming our results in young animals (Figure 2B). We also determined the functional outcome and mortality of 6- to 8-wk-old male Nox4 −/− mice and matched wild-type controls over a longer time period after ischemic stroke (Figure 2D). Five days after 60 min of tMCAO, 15 of 15 wild-type mice (100%) had died, which is in line with previous reports [19]. In contrast, seven of ten Nox4 −/− mice (70%) survived until day 5, and five of these were still alive after 1 wk (p = 0.0039) (Figure 2D). In line with these findings, Nox4-deficient mice showed significantly better Bederson scores than controls over the whole observation period, and neurological deficits remained low until day 7 (Figures 2D and S4). According to the current experimental stroke guidelines [4],[16],[17], any protective effect also requires evaluation in models of both transient and permanent ischemia. We therefore subjected Nox4 −/− mice to filament-induced permanent middle cerebral artery occlusion (pMCAO), a procedure in which no tissue reperfusion occurs (Figure 2E). In the absence of Nox4, infarct volumes (66.7±28.6 mm3 versus 120.1±15.6 mm3, p 0.05). ROS formation from neurons after 24 h was also significantly reduced in Nox4 −/− mice subjected to pMCAO (Figure S6). Because the dihydroethidium stain may also indicate oxidative chemistry events, including formation of ONOO− and nitration of protein tyrosine residues [8], we analyzed the extent of protein nitration in Nox4 −/− and wild-type mice subjected to tMCAO. In agreement with our findings on the generation of ROS, tissue nitration occurred to a lesser extent in ischemic brains from Nox4 −/− mice than in those from wild-type controls (Figure 3B). Oxidative chemistry events such as the formation of ROS and peroxynitrite, as detected by dihydroethidium staining and nitrotyrosine immunolabeling, can induce neuronal apoptosis, which is a well-established mechanism of tissue damage in ischemic stroke [29],[30]. Indeed, superimposed TUNEL and NeuN immunolabeling revealed widespread apoptosis of neurons in wild-type mice 24 h after stroke onset (Figure 3C). In contrast, the number of apoptotic neurons in Nox4 −/− mice subjected to tMCAO was significantly lower, and the basal apoptotic turnover rate in Nox4 −/− mice fell within the range found in sham-operated mice (p>0.05) (Figure 3C). 10.1371/journal.pbio.1000479.g003 Figure 3 Nox4 deficiency confers neuroprotection by reducing oxidative stress, neuronal apoptosis, and disruption of the blood–brain barrier. (A and B) Left panels show representative brain sections from sham-operated wild-type (WT) mice and wild-type and Nox4 −/− mice 24 h after tMCAO. Sections were stained for ROS and oxidative chemistry using dihydroethidium (DHE) (A), or stained for reactive nitrogen species by using nitrotyrosine (B). Right panels show the number of cells per square millimeter that are positive for ROS or oxidative stress (A) or reactive nitrogen species (B) in the ischemic hemispheres of sham-operated wild-type mice and wild-type and Nox4 −/− mice 12 h and 24 h after tMCAO (n = 4 per group). (C) Left panels show representative brain sections from sham-operated wild-type mice and wild-type and Nox4 −/− mice 24 h after tMCAO, immunolabeled for the neuronal marker NeuN and subjected to TUNEL to show apoptosis. Right panel shows the number of TUNEL-positive neurons per square millimeter in the ischemic hemispheres of sham-operated wild-type mice and wild-type and Nox4 −/− mice 24 h after tMCAO (n = 4 per group). (D) Left panels show corresponding coronal brain sections of wild-type and Nox4 −/− mice on day 1 after tMCAO and injection of Evans blue. Extravasation of Evans blue was reduced in areas where infarcts were regularly present in Nox4 −/− mice (basal ganglia, broken white line). The right panel shows the extent of extravasation (i.e., edema volume) as determined by planimetry in the wild-type and Nox4 −/− mice 24 h after tMCAO (n = 6 per group). For (A–C), ###, p 0.05, two-way ANOVA, Bonferroni post-hoc test, compared with baseline rCBF. (B) Assessment of the cerebral vasculature in wild-type and Nox4 −/− mice. A complete circle of Willis (white arrows) was identified in all animals studied, and the distribution of the trunk and branch of the middle cerebral artery appeared to be anatomically identical among the genotypes. (C) Normal brain structure in Nox4 −/− mice. Representative Nissl-stained 5-µm coronal paraffin-wax-embedded brain sections of 3-mo-old wild-type and Nox4 −/− mice (n = 3 each), showing a macroscopic view (uppermost panel), formation of the hippocampus formation (center panel), and somatomotor areas of the neocortex (lowermost panel). (1.64 MB TIF) Click here for additional data file. Figure S3 Generation of Nox4 knockout mice and counter-regulation of NOX1 and NOX2. (A) Construct development for Nox4 knockout mice. Exons 14 and 15 are flanked by loxP sites and followed by a floxed neomycin resistance gene (neo) and a negative-selection cassette coding for diphtheria toxin A (dta) as described in the Text S1. Embryonic stem cell clones were generated by homologous recombination with the targeting vector. Transient expression of Cre recombinase results in three different recombination events. Type 1 results in deletion of the neo cassette and thus floxed exons 14 and 15. These cells can be used to generate conditional Nox4 knockout. Type 2 results in deletion of the floxed exons, and type 3 results in the deletion of exons 14 and 15 and the neo cassette. These cells were used to generate the Nox4 knockout mice. (B) Western blot demonstrating the absence of the 64-kDa NOX4 band in the aorta, lung, and kidney of Nox4 −/− mice. (C) Expression of NOX1 and NOX2 is not upregulated in Nox4 −/− mice. The uppermost left panel shows results of densitometric analysis of the NOX1 134-kDa band in brain samples of the cortex and basal ganglia from Nox4 −/− (pale bar) and wild-type mice (black bar). Data are presented as the relative amount of the NOX1 band normalized to GAPDH and represent the mean ± standard error of three samples. The right panel shows a Western blot comparison of brain and aorta samples from wild-type mice demonstrating the presence of the 134-kDa band in both samples. The center and lowest panels show results of densitometric analysis of the 91- and 53-kDa NOX2 bands seen in brain samples from the cortex and basal ganglia of Nox4 −/− (pale bar) and wild-type mice (black bar). Data are presented as the relative amount of either the 91-kDa band or 53 k-Da band normalized to GAPDH and represent the mean ± standard error of three samples. The bottom right panel shows a Western blot comparison of NOX2 expression in the brain and aorta of wild-type mice, demonstrating the presence of the 91-kDa and 53-kDa bands in both tissues. (24.75 MB TIF) Click here for additional data file. Figure S4 Long-term outcomes are improved in Nox4 −/− mice after tMCAO. Long-term outcome of motor function (grip test) in 6- to 8-wk-old male Nox4 −/− mice (n = 10) and wild-type (WT) controls (n = 15) after tMCAO. Nox4 −/− mice performed better over the whole observation period. **, p<0.001 and *, p<0.05, one-way ANOVA, Bonferroni post-hoc test compared with wild-type mice. (0.27 MB TIF) Click here for additional data file. Figure S5 Motor function after pMCAO. Motor function was assessed by the grip test in 6- to 8-wk-old male Nox4 −/− mice (n = 7) and wild-type (WT) controls (n = 11) 24 h after pMCAO. Two-tailed Student's t-test compared with wild-type mice. ns, not significant. (0.18 MB TIF) Click here for additional data file. Figure S6 Oxidative stress is reduced in brains from Nox4 −/− mice after pMACO. Left panels show representative brain sections from wild-type (WT) and Nox4 −/− mice 24 h after sham operation of pMCAO. Sections were stained for ROS and oxidative chemistry using dihydroethidium. Right panel shows the number of cells per square millimeter that are positive for ROS or oxidative stress in the ischemic hemisphere of wild-type and Nox4 −/− mice 24 h after sham operation or pMCAO (n = 3–5 per group). ##, p<0.001 compared with sham-treated mice; **, p<0.001 compared with wild-type mice by one-way ANOVA, Bonferroni post-hoc test. (1.76 MB TIF) Click here for additional data file. Table S1 Results of blood gas analysis and posterior communicating artery (PComA) score in wild-type and Nox4 −/− mice. (0.04 MB PDF) Click here for additional data file. Table S2 Stroke study population. (0.09 MB PDF) Click here for additional data file. Table S3 Power and type-II (beta) error calculations on infarct volumes depicted in Figure 2A . (0.06 MB PDF) Click here for additional data file. Table S4 Power and type-II (beta) error calculations on infarct volumes depicted in Figure 2E . (0.05 MB PDF) Click here for additional data file. Table S5 Power and type-II (beta) error calculations on infarct volumes depicted in Figure 4B . (0.06 MB PDF) Click here for additional data file. Text S1 Supplementary results, supplementary methods, and supplementary references. (0.32 MB DOC) Click here for additional data file.
