NADPH oxidase inhibition rescues keratinocytes from elevated oxidative stress in a 2D atopic dermatitis and psoriasis model

Emerging evidence suggests oxidative stress plays a role in the pathophysiology of both atopic dermatitis (AD) and psoriasis (PSO). We established in vitro models of AD and PSO skin, and characterized these models in regard to their oxidative stress state. Both AD and PSO model keratinocytes exhibited elevated reactive oxygen species (ROS) levels and accumulated more DNA damage than control cells after oxidative stress induced by 250 µmol/L H2O2. Elevated ROS levels and DNA damage accumulation could be inhibited by the NADPH oxidase (NOX) inhibitor diphenyleneiodonium (DPI). Further, immunofluorescence analysis revealed the presence of both NOX1 and NOX4 in keratinocytes. By inhibiting NOX1, stress‐related signalling cascades and elevated ROS levels could be abrogated, and survival of AD and PSO cells improved. Taken together, this study reveals that inhibition of NOX inhibition could abrogate elevated oxidative stress in a 2D model of AD and PSO.

to initialize either pro-survival gene expression programmes for continuous ROS detoxification (eg Nrf2, pHsp27, pERK) and DNA damage repair (eg ATM and p53) or cell-death-inducing programs (eg NF-κB, p53, p38). [12,13] However, ROS can also lead to the activation of NADPH oxidase (NOX), which themselves generate ROS and therefore reinforce the initial ROS production. The family of NOX comprises of several members (NOX1, NOX2, NOX3, NOX4, NOX5, DUOX1, DUOX2). NOX catalyses the production of superoxide through reduction in NADPH. Via the generation of ROS, NOX mediates diverse functions such as host defense and inflammation, posttranslational processing of proteins, cellular signalling and regulation of gene expression. However, NOX also contributes to a wide range of diseases as a deficiency can lead to immunosuppression and increased activity will lead to excess ROS which is associated with cardiovascular changes, neurodegeneration and inflammatory skin diseases. [29] More evidence for an association between oxidative stress and AD and PSO has been provided by studies which have analysed oxidative stress markers and which have shown that in both AD and PSO malondialdehyde levels are increased, [14][15][16][17][18][19][20][21] and antioxidant enzymes such as superoxide dismutase, catalase and GSH peroxidase are decreased. [16,[21][22][23][24][25] Although extensive evidence exists suggesting an important role of oxidative stress in the pathology of inflammatory skin diseases, no comprehensive study into the molecular mechanisms has been conducted and data on the oxidative state of the skin remain sparse. Furthermore, the mechanism by which oxidative stress might influence pathophysiology has not been discovered yet. The association of oxidative stress with both AD and PSO leads to the question whether ROS play a role in the pathophysiology of both diseases and if yes, whether there are disease-specific differences in redox regulation.

| Induction of a 2-D PSO and AD cell culture model
2D models of AD and PSO were induced by adding respective cytokine mixes to the culture dishes for 24 hours prior to experiments.
We used neonatal human keratinocytes and adapted a protocol to stimulate with either an AD cytokine or a PSO cytokine mix as described previously. [26] For PSO induction, 40 ng/mL IL-22, 40 ng/mL IL-17a and 10 ng/mL TNF-α were added. For AD induction, 40 ng/ mL IL-22, 40 ng/mL IL-4, 40 ng/mL IL-13 and 10 ng/mL TNF-α were added. Control cells were treated accordingly with 10 µmol/L DMSO.

| DCFDA assay
Intracellular ROS levels were measured using the DCFDA assay (Abcam, AB113851) according to the manufacturer's instructions.
Cells were incubated with 10 μmol/L 2′,7′-dichlorofluorescein (DCF) and then washed twice. Oxidative stress was induced by treatment of cells with 250 μmol/L H 2 O 2 . Cells were washed twice with PBS, and ROS levels were quantified by measuring fluorescence at excitation/emission of 495 nm/520 nm.

| Immunofluorescence
Cells grown on coverslips were fixed with 3. For quantification of pH2AX, cell nuclei were manually selected via the DAPI staining in ImageJ to create a mask. The mask was then used to measure fluorescence intensity of both the DAPI and the 488 channels. pH2AX intensity was normalized to the DAPI signal for each image. For fluorescence quantification of NOX1, NOX4 and p22phox, the fluorescence intensity of the total 488 channel signal per image normalized to a blank ROI was measured and then normalized to the total DAPI channel signal normalized to a blank ROI.

| Treatment of a 2D atopic dermatitis and psoriasis skin model with pro-inflammatory cytokines leads to increased oxidative stress
In order to determine the role of oxidative stress in atopic dermatitis (AD) and psoriasis (PSO), we established an adapted 2D in vitro model of AD and PSO keratinocytes. [26] AD, PSO and unstimulated control (Ctrl) keratinocytes were treated with a disease-specific cytokine mix for 24 hours to stimulate the Th1 or Th2 environment, respectively. Treatment with the cytokine mixes had no impact on proliferation or viability (data not shown). To find out whether oxidative stress plays a role in the pathogenesis of inflammatory skin diseases, we first investigated whether stimulation of cells with inflammatory cytokines leads to increased oxidative stress. We induced oxidative stress by treating the cells with 250 µmol/L H 2 O 2 and measured intracellular reactive oxygen species (ROS) using a H2DCFDA assay. Treatment with H 2 O 2 led to an increase in ROS production in all three models, which, however, were significantly greater in the AD and PSO models compared with the control ( Figure 1A; P Value (Ctrl with H 2 O 2 vs AD with H 2 O 2 ) <.0001; P Value (Ctrl with H 2 O 2 vs PSO with H 2 O 2 ) = .0001). The data show that both the AD and PSO models are characterized by an increased oxidative state. To unravel whether the differences in oxidative capacity translate into downstream detrimental biological processes, we analysed DNA damage, as ROS are well-recognized mediators of DNA damage. [27,28] The results showed that H 2 O 2 induced increased DNA damage in the control, as well as in the cytokine-stimulated cells ( Figure 1B

| Inhibition of NADPH oxidase activity leads to abrogation of elevated oxidative stress levels in the AD and PSO models
In order to find out whether the elevated ROS levels in the AD and PSO models are due to increased ROS production or impaired ROS clearance, we measured intracellular ROS after inhibition of NADPH oxidase activity. DPI addition successfully reduced ROS levels after stress induction in all three models ( Analysis of DNA damage accumulation showed that the inhibition of NOX activity significantly reduced DNA damage accumulation after stress in AD and PSO, but reduction in DNA damage accumulation was not significant in control cells.

| Increased oxidative stress in the AD and PSO models is mainly due to NADPH oxidase 1 activity
To further characterize the molecular mechanisms involved in increased oxidative stress sensitivity in the AD and PSO models, we analysed NOX family member expression and localization in keratinocytes. Western blot analysis revealed that only NOX1 and NOX4 can be detected in keratinocytes ( Figure 3A). There was no difference in the expression of NOX1 and NOX4 between AD-and PSO-stimulated cells. DPI treatment reduced NOX1 but not NOX4 expression after DPI and H 2 O 2 treatment in control, and AD and PSO model keratinocytes. Detailed immunofluorescence analysis showed that as expected, there is no difference in expression or localization of either NOX1 or NOX4 in the AD and PSO models ( Figures S2 and   S3), and that neither expression nor localization of the regulatory p22phox subunit of NOX family members is altered, indicating that DPI reduced NOX activity, but not its abundance ( Figure S4). Next, we used the selective NOX1 and NOX4 inhibitors, respectively, to find out which NOX family member is responsible for induction of elevated ROS levels in the AD and PSO models. DCFDA analysis revealed that the inhibition of NOX1 and NOX4 by NOX1/4 inhibi-

| Inhibition of NADPH oxidase activity rescues elevated stress signalling in keratinocytes in the AD and PSO models and promotes survival
For the identification of signalling cascades involved in redox balance in the AD and PSO models, we analysed the stress-related signalling cascades via Western blot. Treatment with H 2 O 2 led to an activation of well-known stress-related signalling cascades such pERK, pP38 and pHsp27 in all three cell models (Figure 4). While P38 signalling is considered to be a pro-apoptosis signal, [10,29] activation of ERK signalling and of the heat shock response via phosphorylation of Hsp27 are considered to be survival-related signalling pathways. [9,10,30,31] Our Western blot analysis revealed activation of both pro-apoptotic and pro-survival signalling pathways. This is not surprising, as we used a non-lethal H 2 O 2 concentration. [32,33] Analysis of cell viability and caspase cleavage showed no significant differences between Ctrl, AD, PSO or H 2 O 2 treatment (data not shown).
We investigated whether stress induction impacts long-term survival by conducting a colony formation assay (CFA). For the CFA, cells were plated very sparsely on a 10-cm cell culture dish, stimulated F I G U R E 1 Atopic dermatitis (AD) and psoriasis (PSO) model keratinocytes react more sensitive to oxidative stress. A, Cells were treated with H 2 O 2 to induce oxidative stress, and reactive oxygen species (ROS) were measured by using a DCFDA assay. AD model keratinocytes exhibit significantly more ROS than untreated control cells. B, DNA damage in the AD and PSO models. Cells were stressed with H 2 O 2 , and DNA damage was measured by analysing levels of pH2AX via immunofluorescence staining (IF). B, shows representative images as well as the quantification of nuclear pH2AX fluorescence intensity normalized to nuclear DAPI. Results represent the means and s.e.m. from more than five independent experiments. *P < .05: statistically significant difference from control value after paired Student t test with the AD or PSO mix for 24 hours, respectively, and then stressed by addition of 250 µmol/L H 2 O 2 for 24 hours. Colony formation was measured after 14 days, thus indicating long-term survival of colonies. Figure 5 shows that treatment of Ctrl, AD and PSO keratinocytes with H 2 O 2 led to a reduction in colony-forming units (CFUs).
Inhibition of NOX activity only led to a better survival rate after oxidative stress in AD and PSO cells models ( Figure 5

| D ISCUSS I ON
Both AD and PSO are common inflammatory skin diseases characterized by epidermal barrier dysfunction and excessive T-cell activation of, however, differing polarity. While AD is driven by the type 2 cytokines, PSO is a Th17-mediated disease. [34][35][36] In both diseases, the increased release of other pro-inflammatory cytokines such as TNF-a and IL-1 in both AD and PSO may cause chronic low-grade systemic inflammation. [34] Our results indicate that the different cytokine milieus associated with AD and PSO have different effects on NOX activity, which might explain why our study revealed differences in the oxidative stress sensitivity of the AD and PSO models in vitro. It is well known that oxidative stress promotes tissue inflammation through upregulation of genes that code pro-inflammatory cytokines, and thus that inflammatory signalling is closely linked to oxidative stress. [37,38] Inflammatory cells in turn release free radicals when activated. Given its prominent inflammatory component, it is conceivable that oxidative stress may play a role in the pathogenesis of inflammatory skin diseases.
As both AD and PSO are inflammatory diseases, inflammatory cells such as T cells and dendritic cells could be the source of elevated oxidative stress marker, but pro-inflammatory signalling in keratinocytes has also been discussed as a possible cause. [8] Our study reveals for the first time that in fact keratinocytes are characterized by high intracellular levels of ROS after treatment with previously described AD and PSO signature cytokines, [26] with IL-4 and IL-13 having a greater effect than IL-17. These differences in oxidative burden translate into downstream detrimental biological processes, such as accumulation of DNA damage and reduced survival rate in the AD and PSO models.
Another important question when regarding impaired redox homeostasis mechanisms that might contribute to disease pathology is whether elevated oxidative stress is due to increased ROS production or due to reduced antioxidative capacity. By utilizing the pan-NADPH oxidase inhibitor DPI, which inhibits the intracellular production of ROS due to environmental stress induction, we were able to show that elevated oxidative stress in the AD and PSO models arises from increased intrinsic ROS production and not impaired antioxidant capacity as hypothesized previously elsewhere. [20,[22][23][24][25] The results indicate a beneficial effect of NOX inhibition on reducing ROS in an AD and PSO models, however, whether these findings translate to in vivo remain to be investigated. Given the multifactorial nature of both AD and PSO, and the high level of complexity of Interestingly, AD model keratinocytes reacted more sensitive to mild H 2 O 2 stimulation than PSO model keratinocytes and had better responsiveness to treatment with NOX inhibitors. This is in line with the observation that while AD skin lesions are characterized by apoptosis of keratinocytes, [39] PSO lesions are characterized by hyperproliferation of keratinocytes. [40,41] It is tempting to speculate that the disease-specific cytokine profile leads to elevated ROS levels in both AD and PSO, but the unique concentration of the key cytokine mediators leads to alternate stress-induced outcomes of cell faith.
As oxidative stress resulting from excess ROS plays an important role in several diseases including AD, PSO, urticaria and allergic diseases, [7] there is a strong need to find strategies lowering ROS in keratinocytes. Possible strategies could include the use of antioxidants or the inhibition of ROS production. Despite many antioxidant therapies that have been evaluated in clinical trials involving tens of thousands of patients, most clinical trial results failed to show effectiveness. [42,43] One of the possible reasons is that in many clinical trials, antioxidants were not chosen because they proved to be the most effective antioxidants, but rather because of their easy availability. This led to the use of antioxidants that were unspecific, ineffective at the doses given or simply also had pro-oxidative effects (eg vitamin E). [42] Due to the clinical fail of many trials dealing with antioxidants, NOX inhibitors are more promising for diseases associated with excess oxidative stress, as NADPH oxidases are the only enzymes solely dedicated to ROS generation and thus prove to be a selective target. [44] Diphenyleneiodonium is a non-specific flavin binder abstracting an electron from an electron transporter forming a radical which F I G U R E 3 Inhibition of NADPH oxidase (NOX) 1 leads to abrogation of elevated oxidative stress levels in the atopic dermatitis (AD) and psoriasis (PSO) models. AD and PSO model keratinocytes were treated with H 2 O 2 and with either DPI, NOX1i (ML171) or Nox1/4i (GKT136901). A, NOX1 and NOX4 protein levels as determined by Western blotting. Figure A shows a representative Western blot as well as quantification of protein levels normalized to actin and to the untreated control keratinocytes. NOX4 levels remain unchanged, while NOX1 levels were reduced after H 2 O 2 stress and treatment with DPI. Results represent the means and s.e.m. from three independent experiments. B, Reactive oxygen species (ROS) production was measured via DCFDA assay. Increased ROS levels in H 2 O 2 -treated control and AD model keratinocytes were abrogated when adding NOX1i and in AD model keratinocytes when adding Nox1/4i, respectively. Results represent the means and s.e.m. from more than five independent experiments. *P < .05; **P < .01; ***P < .001; n.s. P > .05: statistically significant difference after paired Student t test then acts as an inhibitor of the respective electron transporter through a covalent binding step. [45] It is commonly used as a NOX inhibitor not only in chronic inflammatory skin diseases as excess ROS production is associated with many different diseases. A major advantage of DPI is the incomplete suppression of ROS production maintaining a basal ROS level which is needed for physiological processes. [44] DPI has been shown to be a potent and reliable NOX inhibitor; however, DPI has several drawbacks. As it is a flavoprotein inhibitor, it also inhibits CYP450, NO synthase and the mitochondrial electron chain, thus representing an unspecific and toxic inhibitor. [46] In this study, the possible toxic effect of DPI on survival has been addressed via a colony formation assay. We observed reduced colony formation in cells treated with DPI, however, and most importantly, we showed that the combined treatment with H 2 O 2 and DPI leads to increased colony formation as compared to H 2 O 2 alone and thus rescues cells from oxidative damage. Due to the lack of specificity and its toxicity, DPI remains a problematic candidate for drug usage, but is useful as a reference compound for NOX inhibition in vitro. [46] Studies have showed NOX1 and NOX4 KO mice have no significant spontaneous pathologies making the respective selective inhibition of NOX1 and NOX4 more feasible. In addition to the unspecific NOX inhibitor DPI, novel selective NOX1 and NOX4 inhibitors are currently tested in clinical trials. The specific inhibition of respective NOX would lower off-target effects and would make the treatment more effective. The novel inhibitors ML171 and GKT136901 selectively inhibit NOX1 and NOX1/NOX4, respectively. [46][47][48][49][50] Clinical trials with 170 patients have demonstrated that the novel inhibitors are well-tolerated and non-toxic, and thus provide a good alternative for DPI. [50] In this study, we controlled for the possible unspecific effects of DPI by additionally analysing intracellular ROS levels after inhibition of NOX1 and NOX1/4 by the selective inhibitors. Both ML171 and GKT136901 significantly reduced ROS production in AD F I G U R E 4 Inhibition of NADPH oxidase (NOX) activity alters stress signalling. Atopic dermatitis (AD) and psoriasis (PSO) model keratinocytes were treated with H 2 O 2 and with diphenyleneiodonium chloride (DPI). Oxidative stress signalling was characterized by analysing phosphorylation of extracellular signalregulated kinase (ERK), p38 and heat shock protein (Hsp) 27 via Western blotting. A, shows quantification of pHsp27 levels normalized to actin, pERK levels normalized to total ERK and pP38 levels normalized to total p38 from three independent experiments. B, Representative Western blots keratinocytes, but of note the effect was stronger in ML171-treated cells. This indicates that NOX1 is possibly a more important NOX family member in regulating oxidative stress in AD than NOX4, and that a combined inhibition of NOX1 and NOX4 is not as effective as inhibition of NOX1 alone, but further experiments are necessary to elucidate these findings.
Other studies, supporting the important role of NOX inhibitors in the treatment of chronic inflammatory skin diseases, have demonstrated the reduction in AD severity by using the antimicrobial agent gentian violet. Gentian violet is a NOX inhibitor originally used for the treatment of bacterial infections, but studies have proven that the topical administration of gentian violet leads to a reduction in AD severity. [51] Gentian violet is an approved drug and thus might represent a good candidate for further clinical studies.
To our knowledge, there are no in vivo data available regarding NOX inhibition in an AD mouse model. However, there is a study showing a benefit of the antioxidative compound rosmarinic acid in the treatment of NC/Nga mice. [52] Although the study does not mention NADPH oxidases, rosmarinic acid has been described to inhibit NOX2/4 and the effects observed in the above-mentioned study could be due to NOX inhibitory effects. [53] This is the first study to show that keratinocytes treated with AD and PSO signature cytokines elicit an oxidative response, which leads to reduced keratinocyte survival and may contribute to the formation of lesions especially in AD. ROS generation and downstream deleterious effects can be rescued by inhibition of NOX. In the case of AD, where a hallmark of the formation of lesions is the increased apoptosis of keratinocytes, [39,40] the results of this study directly support a link between increased oxidative stress and reduced survival of keratinocytes and lesion formation in patients. In PSO, lesions are characterized by hyperproliferation of keratinocytes instead of apoptosis, but apoptotic features as determined by TUNEL method are still prominent in lesions despite PSO being a hyperproliferative disorder. [54,55] Taken together, the data of this study demonstrate the possible use of NOX inhibitors to counteract detrimental effects of high levels of AD and PSO signature cytokines, but further in vivo experiments are warranted to further elucidate whether NOX inhibition may represent a promising treatment strategy for inflammatory skin diseases. In conclusion, it is important to focus further research on this field in order to characterize the mechanisms of NOX enzymes as major producer of ROS and in order to develop a safe and effective approach to attenuate the ROS production in AD and PSO.

HE designed the experiments; HE, MF, Anke Rose and Steffen
Bachmann acquired the data. HE and MF analysed the data, and HE and SW interpreted the data. HE, ER and SW drafted the article and revised it critically. All authors approved the final version.

CO N FLI C T O F I NTE R E S T
The authors have declared no conflicting interests.