Suppression of the inflammatory response by diphenyleneiodonium after transient focal cerebral ischemia


Address correspondence and reprint requests to Dr Michalis Papadakis, PhD, Laboratory of Cerebral Ischemia, Acute Stroke Programme, Nuffield Department of Clinical Medicine, University of Oxford, Level 7, John Radcliffe Hospital, Oxford OX3 9DU, UK. E-mail:


Diphenyleneiodonium (DPI), a NADPH oxidase inhibitor, reduces production of reactive oxygen species (ROS) and confers neuroprotection to focal cerebral ischemia. Our objective was to investigate whether the neuroprotective action of DPI extends to averting the immune response. DPI-induced gene changes were analyzed by microarray analysis from rat brains subjected to 90 min of middle cerebral artery occlusion, treated with NaCl (ischemia), dimethylsulfoxide (DMSO), or DMSO and DPI (DPI), and reperfused for 48 h. The genomic expression profile was compared between groups using ingenuity pathway analysis at the pathway and network level. DPI selectively up-regulated 23 genes and down-regulated 75 genes more than twofold compared with both DMSO and ischemia. It significantly suppressed inducible nitric oxide synthase signaling and increased the expression of methionine adenosyltransferasesynthetase 2A and adenosylmethionine decarboxylase 1 genes, which are involved in increasing the production of the antioxidant glutathione. The most significantly affected gene network comprised genes implicated in the inflammatory response with an expression change indicating an overall suppression. Both integrin- and interleukin-17A-signaling pathways were also significantly associated and suppressed. In conclusion, the neuroprotective effects of DPI are mediated not only by suppressing ischemia-triggered oxidative stress but also by limiting leukocyte migration and infiltration.

Abbreviations used

adenosylmethionine decarboxylase 1


methionine adenosyltransferasesynthetase 2A


actin-related protein 2/3


Chemokine C-C motif ligand 20






flavin adenine dinucleotide




FYN oncogene


intracellular adhesion molecule 1


inhibitor of DNA binding 1






inducible nitric oxide synthase


ingenuity pathway analysis


Janus kinase


Kinesin family member


mitogen-activated protein kinase-14


Middle cerebral artery occlusion


matrix metalloproteinase 9


magnetic resonance imaging


P38 mitogen-activated protein kinase


proteolipid protein 1


perfusion-weighted imaging




v-ral simian leukemia viral oncogene homolog B


reactive oxygen species




Sialyltransferase 8 A


T2-weighted imaging


helper T lymphocytes


tumor necrosis factor-alpha


vascular cell adhesion molecule 1

Ischemic stroke results from the obstruction of blood flow of an artery within the brain, leading to cell death and infarction. The lack of oxygen and glucose during ischemia activates an array of pathways, including glutamate excitotoxicity, inflammation, oxidative stress, and ultimately induction of apoptosis and necrosis (Sutherland et al. 2012). The central role of oxidative stress and free radicals both in ischemic and reperfusion injury has attracted attempts to pharmacologically inhibit their production or scavenge their formation (Chan 2001; Kontos 2001). The neuroprotective effect of compounds such as the free radical scavenger NXY-059, although promising in preclinical animal studies (Papadakis et al. 2008), failed to translate to an improvement of neurological outcome in acute stroke clinical studies (SAINT II trial) (Shuaib et al. 2007). Delineating the molecular cascades induced by such compounds, downstream or related to oxidative stress suppression, would improve the understanding of their mechanism of action and allow the optimization of future neuroprotective therapies.

We previously demonstrated that the NADPH oxidase inhibitor diphenyleneiodonium (DPI) was neuroprotective, following focal cerebral ischemia (Nagel et al. 2007). NADPH oxidase is a major mediator of oxidative stress by catalyzing the formation of superoxide anion (Segal and Abo 1993) and is activated and up-regulated by focal cerebral ischemia, persisting for at least 30 days of reperfusion (Vallet et al. 2005). Oxidative stress is an early factor in the pathophysiology of stroke resulting to the formation of reactive oxygen species (ROS), which are generated mainly by infiltrating leukocytes and activated microglia, following an immune response triggered by ischemia reperfusion (Iadecola and Anrather 2011). In vitro, superoxide anion generation in isolated blood leukocytes stimulated with the chemotactic peptide N-formyl-methionine-leucine-phenylalanine was inhibited with DPI (Nagel et al. 2007). DPI also suppressed the inflammatory microglial response after lipopolysaccharide stimulation (Wang et al. 2004). Our objective was to gain insights into the neuroprotective molecular mechanisms exerted by DPI, by carrying out a microarray study. The in vivo effect of DPI on the genomic response following cerebral ischemia and reperfusion was investigated aiming at the dissection of any putative effects on the inflammatory response and on oxidative stress-related pathways.


Middle cerebral artery occlusion protocol

The experimental protocols used in this study were approved by the ethics committee for animal research, Karlsruhe, Germany, and all measures were taken to minimize pain or discomfort. Male Wistar rats (250–300 g) were randomized into different groups: a sham ischemia group (Sham), a control ischemia group (Ischemia), an ischemia group treated with the vehicle dimethylsulfoxide (DMSO), and an ischemia group treated with DMSO and DPI (DPI). The Sham and Ischemia groups were carried out contemporaneously with the DMSO and DPI groups and have previously been reported in detail (Nagel et al. 2012).

Anesthesia was performed with 1–2% halothane, 75% N2O, and 23% O2. During surgery, body core temperature was maintained at 37°C by a rectal thermometer connected to a feedback controlled heating pad in all animals. Middle cerebral artery occlusion (MCAO) for 90 min with subsequent reperfusion for 48 h was induced according to a modification of the filament method of Longa et al. (1989) and has been described in detail elsewhere (Garcia et al. 1993, 1995a). The left femoral artery was cannulated for continuous monitoring of arterial blood pressure and for measurement of arterial blood gases before, both during occlusion and reperfusion. In sham animals, the filament was introduced and immediately withdrawn. In all other rats, after induction of reperfusion by withdrawal of the filament, a bolus of either 0.25 mL of isotonic NaCl (Ischemia), 0.25 mL of DMSO, or 0.25 mL of a 50-mM solution of DPI in DMSO (DPI) was intravenously injected. Dose finding of DPI is previously described (Nagel et al. 2007). After wound closure, the rats were allowed to wake up and recover with free access to food and water.

Neurological assessment

Rats were evaluated with a functional neuroscore for neurological outcome on a scale from 3 to 18 (no deficit) before decapitation at 48 h after reperfusion (Garcia et al. 1995b). Data are presented as median, range: min–max.

MRI protocol

The animals were examined in a 2.35 T scanner (Biospec 24/40, BRUKER Medizintechnik, Ettlingen, Germany), and during all scans, the body temperature was maintained at 37°C by a heating fan. The full magnetic resonance imaging (MRI) protocol is described elsewhere (Nagel et al. 2007). Briefly, to prove occlusion of the MCA, a perfusion-weighted imaging (PWI) sequence was obtained directly after placement of the filament. MCAO was considered successful if ≥ 50% of the MCA territory showed a perfusion deficit on the cerebral blood flow map of one slice through the center of the striatum. At 48 h, T2-weighted imaging (T2WI) was performed. Image data were transferred to a SUN Sparcstation 10 (SUN Microsystems, Santa Clara, CA, USA). Measurement of infarct volumes in T2WI was done by a blinded investigator. Infarct volumes were corrected for edema using the formula: (volume of contralateral hemisphere/volume of ipsilateral hemisphere) * volume of infarct. Results are presented as means ± SEM in mm3.

Tissue processing, RNA purification, and preparation for microarray analysis

Rats were killed by transcardiac perfusion with chilled (4°C) isotonic NaCl solution, and brains were removed from the skull, frozen in isopentane, and stored at −80°C until processing. The complete tissue processing protocol has been described previously (Nagel et al. 2012). Briefly, the ipsilateral hemisphere was homogenized in Trizol (Invitrogen, Carlsbad, CA, USA) and the RNA was purified on an RNeasy MinElute Cleanup column (Qiagen, Valencia, CA, USA) after chloroform/isopropanol precipitation. Thereafter, the RNA concentration of the eluate was determined and its integrity was analyzed with an Agilent 2100 Caliper LabChipBio-analyzer (Agilent Technologies, Palo Alto, CA, USA). For Affymetrix GeneChip U34A microarray preparation and analysis, the protocols from Affymetrix Inc. (Santa Clara, CA, USA) were followed, as previously described in detail (Nagel et al. 2012). The raw expression data derived from Affymetrix microarray Suite 4.0.1 software gave each transcript an absolute expression level (signal intensity) and a ‘present’ or ‘absent’ call based on the signal/noise ratio.

Real-time reverse-transcription polymerase chain reaction

The transcripts from four different genes of individual RNA specimens from all animals were quantified by real-time (RT)-PCR. The primers used and the method for cDNA generation and RT-PCR are described in Nagel et al. (2012). All reactions were carried out at least in duplicates for every sample, and melting curve analysis was performed after each run. Data were quantified using the 2-delta CT method and relative fold changes were normalized to sham values. As leptin showed no expression changes on the microarray dataset and no relative changes in the RT-PCR experiments between the groups, it was used as housekeeping gene.

Ingenuity pathway analysis networks

Ingenuity pathway analysis (IPA) composes gene networks using the Ingenuity Knowledge Base, a database based on molecular interactions and gene to phenotype studies from over 200 000 peer-reviewed scientific articles. The networks are composed of genes having direct and/or indirect interactions. Direct gene relationships are those that their protein products have direct physical interaction with one another, such as enzymatic interactions. Indirect gene relationships are those that their protein products do not physically interact, but, for example, their expression levels can be affected by each other. IPA can also identify biological functions statistically associated with certain genes that are part of a particular network.

Statistical analysis

For all intergroup comparisons of physiological parameters, outcome, and MRI, data were presented as mean ± SEM or median (min–max) and anova as well as anova for repeated measurements with Bonferroni post hoc test for parametric and the Kruskal–Wallis test with Dunn's multiple comparison test for non-parametric data was applied. The mortality rates between the treatment groups were tested with the Fisher's exact test. Identified genes from the microarray analysis together with the quantitative data were analyzed using the IPA software and database (Ingenuity Systems®, Only changes equal or greater than ± 2-fold were included for further analysis. IPA calculates the significance value of a given pathway or network as the probability that the pathway or network is associated with the data set by random chance. The p-value was calculated using Fisher's exact test, with values of < 0.05 considered statistically significant.


We carried out a microarray analysis of rat brains subjected to sham ischemia, ischemia, ischemia treated with DMSO, and ischemia treated with DMSO and DPI, 48 h following reperfusion. Three animals had to be excluded because of unsuccessful MCAO, while in all other animals, PWI MRI confirmed qualitatively the occlusion of the MCA territory. Two animals died in the Ischemia group and one each in the DMSO and DPI groups, while one animal of the Ischemia group had to be excluded because of subarachnoid hemorrhage. Finally, each group consisted of five animals.

Neurological outcome and infarct volumetry following focal ischemia and DPI treatment

Animals in the DPI group compared with the Ischemia group had significantly smaller infarct volumes, determined by T2WI (82.8 ± 12 vs. 252 ± 32 mm3, mean diff. 169 mm3, 95% CI 74–264, p < 0.05, Fig. 1a and b) and significantly better neuroscores (14, 13–16 vs. 9, 8–10, diff. in rank sum 9.4, p < 0.05, Fig. 1c), after 48 h of reperfusion. To establish any effect of DMSO on neurological and histological outcome, comparisons between the DMSO and both the Ischemia and DPI groups were carried out. Neuroscores and infarct volumes did not differ significantly between the DMSO and either the Ischemia or DPI groups (Fig. 1c). Physiological variables, including arterial blood gases, pH, and mean arterial blood pressure were similar between time points and groups (Table 1).

Figure 1.

Diphenyleneiodonium (DPI) reduces infarct volumes and improves neurological outcome following ischemia. (a) Four consecutive and representative T2-weighted MR images at 48 h after reperfusion following middle cerebral artery occlusion (MCAO) from one animal of each group. The circumscribed area of hyperintensity represents the cerebral infarct. (b) Quantitative analysis of the corrected infarct volumes after 48 h of reperfusion after MCAO derived from the T2-weighted imaging (T2WI) data. Ischemia animals showed the biggest infarct volumes, whereas DPI-treated animals had significantly smaller infarcts (*p < 0.05, n = 5 per group). (c) Bar graph showing the behavioral scores after 48 h of reperfusion following MCAO. DPI/DMSO-treated animals performed significantly better than Ischemia-only animals (*p < 0.05, n = 5 per group).

Table 1. Summary of the physiological data before, during, and after the MCAO experiments
ParameterTime point in relation to MCAOGroups
  1. Physiological data were assessed by arterial blood gas analysis and continuous invasive blood pressure measurements. No statistical significant differences were observed in the measured variables.

MAP (mmHg)Before85.3 ± 5.293.33 ± 6.698 ± 4.1
During89.6 ± 6.599.04 ± 5.5103 ± 2.7
2 h post89.5 ± 4.299.04 ± 3.587 ± 4
pHBefore7.41 ± 0.017.4 ± 0.017.41 ± 0.03
During7.39 ± 0.017.38 ± 0.027.39 ± 0.01
2 h post7.37 ± 0.027.38 ± 0.017.38 ± 0.02
pCO2 (mmHg)Before45.12 ± 1.2246.37 ± 0.946.82 ± 1.7
During46.97 ± 1.4448.7 ± 1.545.52 ± 2.3
2 h post48.01 ± 1.4347.75 ± 1.449.87 ± 3.5
pO2 (mmHg)Before144 ± 8.7143 ± 5.6129 ± 12.9
During135 ± 8.4124 ± 4.1124 ± 3
2 h post137 ± 14.6122 ± 9.9118 ± 9.6

Gene expression changes by DPI

In all conditions, that is, Sham, Ischemia, DPI, and DMSO, 4067 genes were detected on the Affymetrix array chip. A threshold of twofold was used as a minimum expression change for comparisons between groups (Fig. 2). DPI compared with Ischemia led to an up-regulation of 103 genes (2.53% of all detected genes) and a down-regulation of 194 genes (4.77%), while compared with DMSO, DPI led to an up-regulation of 180 genes (4.43% of all detected genes) and a down-regulation of 117 genes (2.88%). Compared with Sham, DPI resulted to an up-regulation of 306 (7.52%) genes and a down-regulation of 211 genes (5.19%).

Figure 2.

Gene changes induced by diphenyleneiodonium (DPI) compared with Ischemia. Scatterplot of gene fold changes according to the Affymetrix data of the DPI expression against the ischemia group. The data points outside the dotted lines represent genes that were differentially regulated equal or greater than ± 2-fold.

Our hypothesis relied on dissecting DPI-specific effects, which we assumed were correlated with neuroprotection. To identify genes selectively altered by DPI, genes that were either up-regulated or down-regulated more than twofold compared with both DMSO and Ischemia were selected. In total, 98 genes were shortlisted, with 23 and 75 genes up-regulated and down-regulated, respectively, by DPI compared with both DMSO and Ischemia (Table S1). This list of genes was subjected to IPA analysis to identify biological functions, pathways, and gene networks significantly affected by DPI, relevant to oxidative stress and to inflammatory response.

Oxidative stress-related effects by DPI

IPA identified effects of DPI on oxidative stress-related pathways. Inducible nitric oxide synthase (iNOS) signaling was significantly affected by DPI treatment (p = 0.0487), with the expression of two genes participating in this pathway, P38 mitogen-activated protein kinase (P38 MAPK) and Janus kinase (JAK) down-regulated by DPI compared with both Ischemia and DMSO. An additional pathway significantly affected by DPI was methionine metabolism (p = 0.0142). From this pathway, methionine adensyltransferasesynthetase 2A and adenosylmethionine decarboxylase 1 (AMD1) genes, which are implicated in oxidative stress response, were up-regulated by DPI compared with both Ischemia and DMSO (Table 2).

Table 2. Oxidative stress-related effects of DPI compared with Ischemia and DMSO
PathwayGene symbolDPI/Ischemia fold changeDPI/DMSO fold changep-value for pathway
  1. Ingenuity analysis revealed statistically significant changes in two pathways related to oxidative stress. Presented are the p-values for the pathways in line with the according expression changes on detected individual genes within the comparisons.

iNOS signalingMAPK 14−2.35−2.48 0.0487
Methionine metabolismAMT2A2.182.11 0.0142

IPA also identified networks, which are composed of genes with protein products that may participate in different pathways, but directly or indirectly physically interact with each other. The gene network with the highest significant selective alteration by DPI included mitogen-activated protein kinase 14 (MAPK14) which was down-regulated compared with both Ischemia and DMSO and significantly affects oxidative stress response (p = 0.0212; Fig. 3).

Figure 3.

Gene network altered by diphenyleneiodonium. Gene network, with the most significant association with the expression changes induced by diphenyleneiodonium compared with both Ischemia and DMSO. Red-colored genes were up-regulated, whereas green-colored proteins were down-regulated. White-colored genes participate in the pathway, but were not detected in the microarray screen. Significantly affected relevant functions (Fx) are displayed on the right with yellow arrows indicating the genes involved. Image was created using the Ingenuity Pathway Analysis software.

Inflammatory response-related effects by DPI

The aforementioned network also comprised genes that are involved in inflammatory response (Fig. 3). In detail, 21 of the 33 genes composing this network significantly affect the following biological functions: activation of classical (p = 0.00642) and alternative (p = 0.00173) complement pathway, chemotaxis of B (p = 0.00263), γδT (p = 0.00409) and natural killer T (p = 0.00642) lymphocytes, mobilization of dendritic precursor cells (p = 0.000467), quantity of dendritic cells (p = 0.0146) and monocytes (p = 0.0229), infiltration by neutrophils (p = 0.00317) and leukocytes (p = 0.00869), stimulation of monocytes (p = 0.0162), immune response of helper T lymphocytes (p = 0.0323), adhesion of Th17 cells (p = 0.00263) and memory T lymphocytes (p = 0.00758), detachment of neutrophils (p = 0.00438), migration of Th17 cells (p = 0.0105) and monocytes (p = 0.0168), cell movement of leukocytes (p = 0.024) and phagocytes (p = 0.026). Importantly, 20 of these 21 genes were down-regulated by DPI compared with both Ischemia and DMSO.

Additional effects of DPI relevant to inflammation were detected at the pathway level with integrin signaling (p = 0.022) and interleukin-17A (IL-17A) signaling (p = 0.0194) significantly associated with the gene changes selectively induced with DPI. All genes affected by DPI, participating in these pathways, that is, v-ral simian leukemia viral oncogene homolog B (RAL), FYN oncogene, paxillin (PXN), myosin light chain kinase, actin-related protein 2/3 (integrin-signaling pathway) and p38 MAPK, JAK and Chemokine C-C motif ligand 20 (CCL20) (IL-17A-signaling pathway), were down-regulated compared with both Ischemia and DMSO (Table 3).

Table 3. Effects of DPI at the pathway level relevant to inflammation compared with Ischemia and DMSO
PathwayGene symbolDPI/Ischemia fold changeDPI/DMSO fold changep-value for pathway
  1. Ingenuity analysis revealed statistically significant changes in two pathways related to inflammation. Presented are the p-values for the pathways in line with the according expression changes on detected individual genes within the comparisons.

Integrin signalingFYN−2.93−2.49 0.022
IL-17A signalingMAPK14−2.35−2.48 0.0194

Real-time RT-PCR of selected genes

To validate the quantitative findings of the Affymetrix data set, the expression levels of four genes were confirmed with real-time RT-PCR (Fig. 4a). Those included two genes that were highly up-regulated [Kinesin family member (KIFC1)] or down-regulated [Sialyltransferase 8 A (ST8SIA1)] by Ischemia, DMSO, and DPI all compared with Sham and two genes (β-actin and leptin) that were not significantly affected by Ischemia, DMSO, and DPI. There was a significant correlation between the real-time RT-PCR and Affymetrix microarray data (r2=0.5953, < 0.0033, Fig. 4b).

Figure 4.

Correlation of microarray and real-time RT-PCR data. (a) Bar graphs showing the comparison of the real-time RT-PCR quantitative data (n = 5 per group) with the Affymetrix data for individual genes: Kinesin family member (KIFC1) – AF035951_at, Sialyltransferase 8 A (ST8SIA1) – D45255_s_at, beta-actin (β-actin) – AFFX-BioB-3_at, leptin – D49653_s_at. The fold changes for each group [Ischemia, DMSO, or diphenyleneiodonium (DPI)] all compared with Sham are displayed. (b) For statistical comparison between real-time RT-PCR and Affymetrix microarray data, a linear regression with r2 = 0.5953 and p < 0.0033 was calculated.


Oxidative stress and inflammation are closely linked pathophysiological consequences following cerebral ischemia. Because ROS participate in both the acute and the delayed phase infarct growth, we selected the 48 h time point to determine the accumulated effect of DPI. The genomic response was dissected by identifying selective changes induced by DPI compared with both DMSO- and Ischemia-treated groups. IPA was employed to target our analysis to oxidative stress and inflammation-related events. In this study, we confirmed that DPI-treated rats had reduced infarct volumes and improved neuroscores at 48 h after focal cerebral ischemia and reperfusion. We revealed for the first time that the significantly regulated gene pathways and networks accounting for the neuroprotection conferred by DPI extend beyond its properties as an NADPH oxidase inhibitor and are also associated with an overall suppression of the immune-mediated sequelae of cerebral infarction.

The contribution of ROS production to the pathophysiology following cerebral ischemia is well established (Chan 2001). NAPDH oxidase is a major source of ROS by catalyzing the formation of superoxide anion and its activity is inhibited with DPI (Dorsam et al. 2000). The formation of ROS are detrimental to cellular events activated by ischemia as they result in protein degeneration, lipid and ribonucleic acid peroxidation, causing cell dysfunction and tissue damage. DPI has been implicated in suppressing ROS production in several experimental settings including NRK52E rat renal epithelial cells (Umekawa et al. 2005) and cultures of NOD.H2(h4) mouse thyrocytes (Burek and Rose 2008).

DPI can possibly suppress oxidative stress by also inhibiting xanthine oxidase, because in addition to NADPH oxidase, DPI can inhibit other flavin adenine dinucleotide (FAD) oxidases (O'Donnell et al. 1993; Sanders et al. 1997). Such action could account for some of its neuroprotective effects, as xanthine oxidase contributes to ischemic damage by producing oxygen-free radicals such as the superoxide anion, responsible for the breakdown of membranes and structures within the nucleus and elements of the cytoskeleton (Moro et al. 2004).

Another FAD oxidase that can be inhibited by DPI is NOS. Inhibition of endothelial NOS in macrophages could be an important neuroprotective mechanism (Stuehr et al. 1991). Importantly, our microarray and IPA analysis indicated that DPI overall suppressed iNOS signaling as shown by the down-regulation of p38 MAPK and JAK, which are both implicated in the transcriptional regulation that leads to increased iNOS production (Do et al. 2010; Da Silva et al. 1997; Bhat et al. 1999). Expression of iNOS following cerebral ischemia and reperfusion takes place in both neutrophils infiltrating the ischemic tissue and in reactive astrocytes, resulting to the production of NO (Hu et al. 1995; Samdani et al. 1997; Forster et al. 1999). NO and its by-product peroxynitrite contribute to oxidative stress by protein oxidation and importantly inhibition of iNOS confers neuroprotection to the ischemic brain (Zhang and Iadecola 1998). Potentially, direct action of DPI on its enzymatic activity might further contribute to suppression of iNOS signaling.

In addition, DPI up-regulated MAT2A and AMD1, which metabolically mediate the production of glutathione. MAT2A is the enzyme responsible for the catalysis of S-adenosylmethionine production, which is a glutathione precursor (Cederbaum 2010). Importantly, AMD suppression in HL-60 cells results to an over production of ROS attributed to a decrease in the total amount of glutathione and a decrease in the ratio of reduced to oxidized glutathione (Kim et al. 2006). The reduced form of glutathione is a major antioxidant (Warner et al. 2004; Slemmer et al. 2008), which is depleted by cerebral ischemia resulting to oxidative stress (Park et al. 2000; Namba et al. 2001). In models of global ischemia, administration of a glutathione mimetic improved outcome (Gotoh et al. 1994).

Ingenuity pathway analysis identified a network, containing genes related to oxidative stress, which was selectively and significantly altered by DPI following ischemia and reperfusion. MAPK14 occupied a central hub position within this network and was down-regulated by DPI compared with both Ischemia and DMSO. Suppression of this gene in mouse proximal tubular cells of the kidney can prevent hydrogen peroxide-induced oxidative stress (Kaimori et al. 2003).The reduction of MAPK14 expression to levels lower than sham ischemia suggests an active suppression by DPI. All these findings demonstrate that DPI not only inhibits NADPH oxidase but also induces genes that strengthen the oxidative stress response and down-regulates genes that facilitate ROS production.

NADPH oxidase is expressed in all inflammatory cell types. Therefore, it was of great interest to investigate whether the genomic response triggered by DPI, in addition to inhibiting oxidative stress, prevented or even actively suppressed the inflammatory response following ischemia and reperfusion.

The inflammatory response triggered by ischemia includes an innate response which characterizes acute ischemia, while during the delayed phase of ischemia, 1–3 days after the original insult, the adaptive immune response is considered to potentially play a role (as reviewed by Iadecola and Anrather 2011). During the acute phase, complement is activated, neutrophils and macrophages are recruited to the ischemic site, followed by myeloid dendritic cells which can present antigen, and thus activate T cells. Neutrophils adhere to the cerebral endothelium and transmigrate into the tissue releasing their cytoplasmic granules containing proinflammatory molecules such as iNOS, NADPH oxidase, and matrix metalloproteinase-9, which also contribute to the induction of oxidative stress. Microglia are activated resulting in chemokine and cytokine release and up-regulation of the tumor necrosis factor (TNF)-alpha (TNF-α) receptor (Downes and Crack 2010). Finally, during the delayed phase of ischemia, T helper cells (Th) release the pro-inflammatory mediators such as interleukin 2, IL-12, TNF-α (Downes and Crack 2010).

The gene network mostly affected by DPI contained genes with significant effects on the inflammatory response. Interestingly, the expression change of all of these genes by DPI compared with both Ischemia and DMSO indicated an overall suppression of ischemia-stimulated inflammatory events. Different stages of the inflammatory response were suppressed, from the activation of complement, chemotaxis of B, γδT and natural killer T lymphocytes, migration and mobilization of leukocytes, phagocytes, monocytes and dendritic precursor cells, to infiltration of neutrophils and leukocytes.

Activation of both classical and alternative complement pathways contributes to the activation of the inflammatory response following ischemia and reperfusion (Yanamadala and Friedlander 2010). Complement factor B is a mediator of both pathways acting as a chemotaxis-inducing stimulus for neutrophils (Garcia et al. 1995b; Pedersen et al. 2009) and was down-regulated by DPI.

Further support for the suppression of inflammatory cell migration by DPI was indicated by the down-regulation of C-C motif ligand 20 (CCL20), a member of the CC chemokine family (Campbell et al. 1998). CCL20 is involved in a number of biological functions associated with the immune response, predominantly closely aligned with the orchestration and logistics of the immune response and in particular with migration of leukocytes (Dieu-Nosjean et al. 2001; Nistala et al. 2008; Nawarskas and Clark 2011). CCL20 has been shown to increase chemotaxis of naïve B lymphocytes and γδ and natural killer T lymphocytes (Tanaka et al. 1999; Kim et al. 2002). In addition, CCL20 increases the recruitment of dendritic cells (Kodama et al. 2011). Importantly, CCL20 is up-regulated after ischemia in rat astrocytes and antibody neutralization of its activity reduces infarct volumes (Terao et al. 2009). Another gene that was also down-regulated by DPI and was part of this network, inhibitor of DNA binding 1, has also been implicated in the recruitment of phagocytes and monocytes (Chan et al. 2009).

Additional evidence for the prevention of leukocyte migration and infiltration by DPI was indicated by the up-regulation of proteolipid protein 1 (PLP1). Critically, this was the only gene of this network that was up-regulated by DPI compared with both Ischemia and DMSO with a function related to the inflammatory response. PLP1 encodes a transmembrane proteolipid protein that prevents infiltration of leukocytes and macrophages (Reddy et al. 2005; Kroner et al. 2009). Moreover, PLP1 inhibits the function T helper cells (Karpus et al. 1998). Therefore, the up-regulation of PLP1 by DPI denotes an active suppressive response of inflammation.

At the pathway level, additional effects of DPI relevant to leukocyte adhesion and infiltration were revealed by the finding that integrin- and IL-17A-signaling pathways were significantly affected. All genes detected in our microarray screen participating in these pathways were down-regulated by DPI, suggesting a suppressive effect. The effect of DPI on integrin signaling further suggests ablation of leukocyte migration to ischemic tissue. The recruitment of leukocytes in the cerebrovasculature exposed to ischemia and reperfusion involves different adhesion molecules expressed on vascular endothelium and circulating inflammatory cells (Yilmaz and Granger 2008). Integrins and in particular β1- and β2-integrins are expressed on leukocytes and mediate their firm interaction on endothelial cells with the intracellular adhesion molecule 1 and vascular cell adhesion molecule 1, respectively (Haqqani and Stanimirovic 2011). PXN which is part of the integrin-signaling pathway regulates focal adhesion assemblies (Mazaki et al. 1997) and following cerebral ischemia its expression is increased, denoting the activation and recruitment of leukocytes to the site of injury (Erdo et al. 2004). DPI down-regulated PXN expression levels to lower than sham, indicating an active suppression rather than an epiphenomenon of an overall reduced inflammatory response.

Interleukin-17A is a cytokine produced by γδT lymphocytes, but also from astrocytes following ischemia (Li et al. 2005). In rats, IL-17A levels are increased in the ischemic hemisphere 6 h after permanent MCAO. During the delayed phase of ischemia, astrocytes also produce IL-17A further exacerbating its levels (Li et al. 2005). Similarly, IL-17A levels are also elevated in the brain of ischemic stroke patients 2 days after stroke onset (Li et al. 2005; Yin and Li 2011). IL-17A knockout mice exhibit reduced infarct volumes, indicating infiltration of γδT lymphocytes plays a central role in exacerbating brain infarction (Shichita et al. 2009). DPI down-regulated all genes detected in the microarray participating in the IL-17A pathway, supporting the correlation of DPI action with suppression of leukocyte infiltration. Overall, the observed dampening of the inflammatory response by DPI after cerebral ischemia further strengthens our previous finding of reduced matrix metallopeptidase activity and blood–brain barrier breakdown (Nagel et al. 2007).

In conclusion, for the first time we demonstrated that the neuroprotective effects of DPI are directly linked to the suppression of the cerebral immunological response. Whether this effect can be attributed to the direct action of DPI on NADPH oxidase or it is manifested by acting on other targets remains to be established. It is of importance to note that the non-specific inhibition of FAD containing oxidases by DPI can warrant unexpected toxic side effects, which should be investigated in long-term studies. As we analyzed brain homogenates of the whole ipsilateral hemisphere, we cannot account for cell type or region-specific differences of the gene expressions observed in this model. Because it is also not possible to conclusively ascribe the observed neuroprotective effects to active changes induced by DPI or to a decrease in pathophysiological change, elucidating the temporal profile of this response could prove pivotal to deciphering the underlying mechanism of this novel phenomenon. Future neuroprotective agents should undergo a thorough analysis of related gene expression changes before translation to bedside, to better understand their mechanism of action and minimize the risk of futile clinical trials.


SW conceived and designed the experiments. SN carried out the animal studies and acquired the data. SN and MP carried out the real-time RT-PCR experiments. KP performed and analyzed the microarrays. MP, GH, SW, and SN analyzed the data and drafted the manuscript. MP, CGG, GH, and SN interpreted the data. AB and CGG reviewed and critically edited the manuscript. All authors approved the final version of the manuscript.

We thank Dr Marinos Kallikourdis for his critical feedback on the manuscript.

Conflicts of interest

All authors declare no conflict of interest. This study was supported by a grant (F.203673) to SW of the University of Heidelberg and by the Dunhill Medical Trust. GH received funding by the NIHR Integrated Academic Training Programme and Oxford University Clinical Academic Graduate School.