The 1027^th target candidate in stroke: Will NADPH oxidase hold up?

NOX1, 2, 4 and 5 are expressed in blood vessels. However, NOX5 is not present in rats and mice. NOX3 is expressed in the inner ear and DUOX1 and DUOX2 are expressed in thyroid, with DUOX2 also being present in lung and gastrointestinal tract epithelium. [47, 48]. As NOX3, as well as DUOX1 and DUOX2, are neither expressed in the vasculature nor in the brain, they are not relevant for stroke and will thus not be further discussed in this review. A summary of the tissue and cellular distribution of NOX1, 2, 4 and 5 is given in Table 2.

NOX1 is predominantly found in colon epithelia [49], but also at lower levels in cerebral cells (neurons, astrocytes and microglia) [50] and in vessels. In the vasculature, NOX1 is usually restricted to vascular smooth muscle cells (VSMCs) [51]. Indeed, NOX1 knockout (KO) mice have slightly reduced blood pressure and a reduced pressor response to angiotensin II [52, 53].

NOX2 was first discovered in phagocytes and was previously called gp91^phox. The role of NOX2 in neutrophils is host defence [41]. Besides circulating neutrophils, it was also detected in the brain (mainly in microglia) and in vascular cells [41]. In vessels, NOX2 is present in endothelial cells (ECs) and fibroblasts [54], and also in infiltrating monocytes, macrophages and T-cells when underlying pathology is present.

NOX4 (previously called renox) is widely distributed with high physiological levels in the kidneys [55], lung [56] and vasculature [57]. Therefore, knocking out the NOX4 gene was expected to induce arteriolar hypotension, altered pulmonary and renal function. However, surprisingly, deletion of NOX4 did not cause any respective abnormal basal phenotype [12, 58]. NOX4 is the most dominant isoform in vessels and its levels are even higher in cerebral vessels [59]. NOX4 mRNA is expressed in ECs, VSMCs and fibroblasts [40]. Expressed and functionally active NOX4 has also been found in neurons [60] and astrocytes [11].

As NOX5 is not expressed in rodent cells, it has not been studied in classical NOX KO studies [61]. In other species, NOX5 was detected in testis, spleen, kidney, lymphocytes, ECs and VSMCs [62].

In summary, different NOX homologues have different functions in different cell types, and subcellular localization is also an important determinant of NOX function. For more detailed information we refer to other reviews [41, 47].

All NOX isoforms contain six trans-membrane domains with two heme-binding sites and on the cytosolic tails binding-sites for FAD and NADPH [63]. A number of regulatory subunits have been identified for the NADPH oxidases. NOX1-4 form heterodimers with the membrane-bound p22^phox subunit [64].

NOX2 was the first isoform identified and thus is the best-studied isoform. Subunits that are needed for NOX2 activation are divided into two groups: activating molecules (p67^phox) and organizing molecules (p47^phox). Upon phosphorylation of p47^phox, cytosolic subunits (p40^phox, p47^phox and p67^phox) translocate to the cell membrane and bind to intracellular loops of NOX2. Activators bind to Rac in the cytosol and then migrate towards NOX2 with help of the organizers [41] (Figure 1). For ROS production by NOX2, Rac and p47^phox have to be activated simultaneously.

Similar to NOX2, NOX1 is also regulated by cytosolic units. Its organizer molecule NOXO1 constitutively interacts with the membrane and does not need phosphorylation like p47^phox. Therefore, NOX1 seems to be active under basal conditions [65]. The activating molecule of NOX1 is called NOXA1.

NOX4 seems to only need interaction with p22^phox, but does not necessarily require any of the known cytosolic subunits to be active. NOX4 was therefore suggested to be constitutively active, and thus its activity may directly be related to its protein levels [66]. Recently, Rac implication in NOX4 activation has also been excluded [67]. However, stimulus-induced ROS production by NOX4 has been observed; and maybe other, yet unknown factors play a role in NOX4 activity regulation. It has been shown that polymerase (DNA-directed) delta interacting protein 2 (PolDip2) interacts with p22^phox, which promotes NOX4 linking to the cytoskeleton [68] (Figure 1). Furthermore, protein disulphide isomerase (PDI) may also bind to NOX4, as well as to NOX1 [69]. However, the role of PolDip2 and PDI in regulating NADPH oxidases has to be further investigated.

NOX5 does not require heterodimerization with p22^phox. Interestingly, NOX5 is directly activated by calcium, which can bind on the EF-hand motives of the N-terminal tail of NOX5 [41, 46]. Although it is generally agreed that there is no need for other NOX5 binding proteins, a recent study showed that Hsp90 binds to NOX1, 2 and 5 and that this binding is needed to confer enzyme stability [70]. Five splice variants of NOX5 have been discovered, with the shortest isoform 5s lacking the calcium-binding domain [62].

Moreover, it was postulated that NADPH oxidases are involved in the ischemic pathway by functioning as oxygen sensors [71], an important aspect in the field of I/R injury. In hypoxic cells, mRNA and protein levels of NOX increased rapidly. The subsequent increase in ROS production led to an upregulated and activated hypoxia inducible factor (HIF) [31], whereas in NOX KO cells, HIF was not induced by hypoxia. Hypoxia seems to regulate all NOX isoforms [72‐74].

Besides regulation through binding proteins, various stimuli, e.g. angiotensin II, thrombin or growth factors, have also been shown to alter the expression or the activity of NADPH oxidases [47]. NOX homologues can cause different physiological responses when coupled to different agonists.

Reduction of oxygen by transmitting one electron from NADPH results in O₂ ^- that can be converted into H₂O₂ by superoxide dismutases (SOD). All NOX isoforms appear to produce O₂ ^-, except NOX4 that mainly produces H₂O₂[55, 75] (Figure 1). Because of this H₂O₂ production, initially several problems were encountered to detect ROS production from the NOX4 isoform [67, 76]. Mutations of NOX4 can switch ROS release from H₂O₂ to O₂ ^-[75].

In cell cultures of the murine brain, NOX1 protein was found in neurons, astrocytes, microglia and endothelial cells [11]. Results on the role of NOX1 in stroke using NOX1 KO mice are conflicting. Jackman et al. [10] did not observe an effect of NOX1 deletion on total infarct volume, edema or neurological outcome after tMCAO when compared to WT mice. However, they did find an increased cortical infarct in NOX1 KOs, suggesting that NOX1 might play a role in limiting cortical infarct size. Basal ROS levels in the brains were similar in WT and NOX1 KO mice. A second group [11] observed the opposite effects of NOX1 KO on stroke. They reported a 55% smaller stroke lesion that was paralleled by better neurological outcome in NOX1 KO versus littermate WT mice after 1 h of ischemia and 23 h of reperfusion. However, no difference in lesion size was observed when the ischemic period lasted 2 h or when permanent ischemia was investigated. Apoptosis levels were similar in both strains and the response to antioxidants and NOS inhibitors was similar in both genotypes. Thus, the observed infarct size reduction in NOX1 KO mice was neither related to reduced ROS production nor to reduced NO bioavailability. A third study reported no significant difference in stroke outcome between NOX1 deficient and control mice 1 day after cerebral I/R injury [12].

In summary, it is unlikely that NOX1 plays a major role in stroke in mice. One should also keep in mind that NOX1- and p47phox deficient mice had reduced basal blood pressures and blunted pressor responses to angiotensin II [52, 53, 106], which may interfere with stroke outcome.

The most studied isoform in stroke experiments is NOX2. After transient focal cerebral ischemia NOX2 protein levels were elevated, p47^phox translocated to the membrane and ROS production increased [107]. These results show a potential role of NOX2 in stroke pathology (Figure 2). The spatiotemporal profile of NOX2 expression was studied in endothelin-1-induced stroke in conscious rats [108]. NOX2 mRNA was upregulated from 6 hours until 7 days post-stroke in the cortex and striatum. At 6 h post-stroke ROS production was found in neurons, and after 7 days in macrophages and microglia.

Several papers reported a protective effect of deleting NOX2 in mice: NOX2 deficient mice had a 40% smaller infarct volume than WT mice [13‐21]. In contrast to these studies, Kleinschnitz and colleagues could not reproduce the findings of the previous studies concerning the neuroprotective effect of NOX2 deletion despite using a high number of animals (n = 19) [12]. VAS2870 did not display an additional protective effect in NOX4 knockout mice, whereas it did so in WT mice, further suggesting that NOX1 and 2 have no major implication in stroke pathology [12]. Reasons for the discrepancies are unclear, but several factors may play a role, including different occlusion times and different experimental protocols, as discussed in more detail in the review of Radermacher et al. (Antioxid Redox Signal, in revision).

In the brain, NOX4 is most abundant in endothelial cells, but also present in neurons and astrocytes (Figure 2) [11]. As for NOX2, the spatiotemporal profile of NOX4 mRNA expression after stroke in rats has been measured [108]. NOX4 mRNA was upregulated 6 h post-stroke and then returned to control levels. In contrast, others reported elevated NOX4 mRNA levels in the rat cortex 24 h after pMCAO lasting up to 15 days. The early elevation of NOX4 mRNA is probably due to neuronal expression and the later peak expression to neo-angiogenesis [60]. In another study, NOX4 mRNA was elevated at 12 and 24 h after tMCAO [12]. Here, NOX4 expression was also validated at the protein level (by immunohistochemistry) in brain samples from stroke patients and in mouse brain slices. Co-staining clearly showed colocalization of NOX4 with endothelial cells and neurons. To our knowledge only one study performed stroke experiments in NOX4 KO mice [12] (Figure 3). After exclusion of systemic vascular effects in NOX4 deficient mice, 75% smaller infarct volumes were measured in NOX4 KO compared to WT, NOX1 KO and NOX2 KO mice (Figure 3). In addition, functional outcomes were improved, as was survival. Similar results were obtained in female, in older mice (18–20 weeks) and when using a permanent stroke model. This study shows that NOX4 seems to play a major role in brain damage and thus is a promising target in stroke therapy. In addition to this NOX4 KO study, another group recently showed first results confirming the detrimental role of NOX4 post stroke. For their study, this group used transgenic mice that overexpress NOX4 in endothelial cells. One day after MCAO, the infarcted region of these transgenic animals was bigger than in wildtype mice and suppression of eNOS by NOX4 was proposed as being responsible for infarct enlargement [109].

As NOX5 is not expressed in rats and mice, studies investigating this isoform are scarce, and thus to date no data on the role of NOX5 in stroke are published. To elucidate the possible role of this isoform in animals, transgenic expression of NOX5 in mice is needed, or other animal species, such as rabbits, have to be used.