Keratinocyte‐derived IL‐1β induces PPARG downregulation and PPARD upregulation in human reconstructed epidermis following barrier impairment

Abstract Peroxisome proliferator‐activated receptors (PPARs) are a family of nuclear hormone receptors. In skin, PPARs modulate inflammation, lipid synthesis, keratinocyte differentiation and proliferation and thus are important for skin barrier homeostasis. Accordingly, PPAR expression is altered in various skin conditions that entail epidermal barrier impairment, that is atopic dermatitis (AD) and psoriasis. Using human epidermal equivalents (HEEs), we established models of acute epidermal barrier impairment devoid of immune cells. We assessed PPAR and cytokine expression after barrier perturbation and examined effects of keratinocyte‐derived cytokines on PPAR expression. We show that acetone or SDS treatment causes graded impairment of epidermal barrier function. Furthermore, we demonstrate that besides IL‐1β and TNFα, IL‐33 and TSLP are highly relevant markers for acute epidermal barrier impairment. Both SDS‐ and acetone‐mediated epidermal barrier impairment reduce PPARG expression levels, whereas only SDS enhances PPARD expression. In line with findings in IL‐1β and TNFα‐treated HEEs, abrogation of IL‐1 signalling restores PPARG expression and limits the increase of PPARD expression in SDS‐induced epidermal barrier impairment. Thus, following epidermal barrier perturbation, keratinocyte‐derived IL‐1β and partly TNFα modulate PPARG and PPARD expression. These results emphasize a role for PPARγ and PPARβ/δ in acute epidermal barrier impairment with possible implications for diseases such as AD and psoriasis.


| INTRODUC TI ON
Peroxisome proliferator-activated receptors (PPARs) are a family of ligand-dependent nuclear hormone receptors that comprise PPARα, PPARβ/δ and PPARγ. After ligand activation, PPARs heterodimerize with the retinoid-X-receptor (RXR). Together with cofactors the complex binds to PPAR response elements (PPRE) in promoter regions of target genes and regulates gene expression.
PPARs are activated by numerous endogenous and exogenous fatty acids and derivatives. 1 PPARα and PPARγ are primarily expressed in the suprabasal layers of the epidermis, whereas PPARβ/δ is present throughout the entire epidermis. 1,2 In skin, PPARs modulate inflammation, lipid synthesis, keratinocyte differentiation and proliferation. [1][2][3][4][5][6] Furthermore, PPARs are involved in the homeostasis of skin appendages including hair follicles and sebaceous glands. 5,7,8 Activation of all three PPARs improves barrier recovery after acute disruption. [1][2][3][4]9 Moreover, PPAR expression is modified in various inflammatory skin conditions that entail epidermal barrier impairment. Depending on the experimental set-up, PPARα expression is decreased or remains unchanged in atopic dermatitis (AD) and psoriasis, respectively. 3,10-13 PPARβ/δ has been shown to be upregulated in psoriasis. 10,12,[14][15][16] Although information on PPARγ is contradictory in AD, it was shown to be consistently reduced in psoriasis. 10,[12][13][14]17,18 In a model of acute epidermal barrier impairment, PPARα and PPARγ expression is decreased, whereas PPARβ/δ is not altered in human skin. 19 In contrast, a more recent study reports that acetone-mediated barrier perturbation increases PPARD and diminishes PPARA and PPARG expression levels in organotypic skin cultures. 20 In agreement with expression data, PPARβ/δ activation induces psoriasis-like skin symptoms, which were shown to be ameliorated by PPARβ/δ antagonism. 14,21,22 Furthermore, PPARγ activation was shown to ameliorate skin lesions in patients with psoriasis. 4,[23][24][25] Accordingly, a novel and specific PPARγ modulator was proven to be anti-inflammatory and anti-proliferative and to restore differentiation in a psoriasis-like mouse model. 26 Together, these data exemplify the diverse role of PPARs in modulating skin inflammation and underscore their importance in skin barrier homeostasis. Thus, in the present study we established reproducible models of epidermal barrier impairment closely mimicking human skin by using human epidermal equivalents (HEE). In these models, we assessed cytokine and PPAR alterations at gene expression level after epidermal barrier impairment and evaluated the modulatory impact of cytokine changes on PPARs.
16 h later, HEEs were lifted to the air-liquid interface by aspiration of the medium inside the inserts. Thereafter, culture media were changed daily until harvesting. HEEs were grown at a humidity of 50-60%, at 37°C and 5% CO 2 .

| Morphological analysis
Human epidermal equivalents were fixed in 4% formaldehyde, paraffin-embedded and 6 µm sections were stained with haematoxylin & eosin. Sections were analysed using an Olympus BH-2 light microscope (Olympus, Shinjuku, Japan) equipped with a ProgRes C10plus camera (Jenoptik, Jena, Germany) and

| Transepithelial electrical resistance (TEER) measurements
Transepithelial electrical resistance measurements were performed using an Epithelial Volthommeter (World Precision Instruments, Sarasota, FL) according to manufacturer's instructions.
Measurements were recorded using fresh 0.5 ml of CnT-02-03D medium on top of the transwell and 1 ml below the transwell.

| RNA Isolation and RT-PCR
Total RNA from cultured keratinocytes and HEEs was isolated using TRIZOL reagent (Gibco BRl, Life Technology). DNA-free kit (Ambion, Carlsbad, CA) was used to remove contaminating gDNA from RNA preparation according to the manufacture's protocol.

| Statistical analysis
Statistical analyses were performed using GraphPad Prism 6 software (GraphPad Software, LaJolla, CA). Data are presented, if not other specified, as mean ± SEM. Statistical significance as determined using Student's paired two-tailed t test or one-way analysis of variance, followed by Bonferroni post hoc test was performed with significance determined as a p-value <0.05.

| SDS perturbs epidermal barrier function in HEEs more efficiently than acetone
First, we established an in vitro model to study epidermal barrier impairment. We generated human epidermal equivalents (HEEs) from primary human keratinocytes cultured at a humidity of 50-60% that closely mimic human epidermis. 27,28 On day 12 of culture, we exposed HEEs to 1% sodium dodecyl sulphate (SDS) or acetone, two agents commonly used to inflict barrier perturbation in skin. 19,29 Haematoxylin & eosin (H&E) staining proves that SDS and acetone do not have toxic effects on the integrity of HEEs ( Figure 1A). To assess whether SDS or acetone perturbs epidermal barrier function, we (i) employed Lucifer Yellow (LY) permeability assay and (ii) measured transepithelial electrical resistance (TEER), two techniques used to evaluate the outside-in (mainly the stratum corneum) and inside-out (tight junctions and stratum corneum) barrier competence, respectively. While SDS, but not acetone, rendered HEEs permeable to LY ( Figure 1B), both SDS and acetone induced a striking decrease of TEER, indicating barrier impairment ( Figure 1C). These data show that both SDS and acetone application successfully perturbs epidermal barrier function, yet to a different degree. Indeed, TEER is a highly sensitive method reflecting the ionic conductance of the paracellular pathway in HEEs.
LY assay is less sensitive than TEER measurements due to the size of the molecule (C 13 H 10 Li 2 N 4 O 9 S 2 ). 30 Thus, our results show that acetone only mildly disrupts the epidermal barrier, whereas SDS has a more deleterious effect.

| In HEE SDS-mediated-not acetonemediated-barrier perturbation leads to changes of cytokines implicated in AD and psoriasis
Next, we examined mRNA expression levels of inflammatory genes known to be produced by keratinocytes after barrier impairment in SDS-or acetone-exposed HEEs. The expression of IL1B and IL1A was increased 6 and 24 h after SDS treatment, respectively ( Figure 2A). TNFA gene expression was enhanced 6 h (11.9-fold±5.6; p = 0.0119) and remained unchanged 24 h post-SDS-mediated barrier perturbation. As illustrated in Figure 2A,

| SDS-mediated barrier impairment results both in PPARG downregulation and PPARD upregulation, while acetone-mediated barrier perturbation only leads to PPARG downregulation
Since in other tissues proinflammatory cytokines have been shown to modulate PPAR expression, we next measured the mRNA level of all three PPAR isotypes. 31,32 Notably, PPARD was increased 5.7fold (±0.6; p = 0.0027) 24 h but not 6 h after SDS exposure. In contrast, barrier impairment due to acetone application did not result in a significant upregulation of PPARD. While PPARG remained un-  Figure 2C). Thus, epidermal barrier perturbation inflicted by SDS treatment recapitulates PPAR gene expression changes in psoriasis. 10,12,[14][15][16] In contrast, acetoneinduced barrier impairment drives PPARG downregulation, which has been inconsistently reported in AD. 13,18 Together, these findings show that PPARG, in contrast to PPARD and PPARA, is consistently downregulated after SDS-and acetone-induced barrier impairment. These data suggest that PPARG expression is a highly sensitive marker for the fitness of the epidermal barrier and serves as the PPAR isoform most closely linked to the integrity of the epidermal barrier.

| IL-1β and TNFα treatments trigger a SDS-like cytokine response in HEEs
Since IL1B is upregulated in both SDS-and acetone-mediated barrier impairment, we treated human primary keratinocytes and HEEs with IL-1β (Figures 3, S1). In HEEs, IL-1β treatment for 6 h and 24 h led to an increase of IL1B and TNFA mRNA levels ( Figure 3A).
IL1A and TSLP expression levels were upregulated 2.6-fold (±0.5) and 7.0-fold (±1.2) 6 h after IL-1β treatment, respectively, and remained unchanged after 24 h ( Figure 3A). IL33 mRNA was not significantly altered in HEEs treated with IL-1β ( Figure 3A). Similar to IL-1β, TNFA mRNA levels were increased after barrier impairment by either SDS or acetone. TNFα treatment for 6 h and 24 h increased IL1B and TNFA expression ( Figure 3B). IL1A mRNA was upregulated in HEEs when treated with TNFα for 24 h. Similar to F I G U R E 2 Cytokine and PPAR expression in HEEs with impaired barrier function. (A) HEEs were exposed to SDS 1% and harvested 6 h and 24 h later. mRNA expression levels of cytokines were assessed by RT-PCR. Combined data from 9 independent experiments are presented as mean ± SEM. (B) HEEs were exposed to acetone and harvested 6 h and 24 h later. mRNA expression levels of cytokines were assessed by RT-PCR. Combined data from 5 independent experiments are presented as mean ± SEM. (C) HEEs were exposed to SDS 1% or acetone and harvested 6 h and 24 h later. mRNA expression levels of PPARA, PPARD and PPARG were assessed by RT-PCR. Combined data from 9 independent experiments for SDS-treated HEEs and from 5 independent experiments for acetone-treated HEEs are presented as mean ± SEM. Data were analysed using a paired Student's t test for all experiments.

| IL-1β decreases PPARG and increases PPARD gene expression in HEEs, whereas TNFα only reduces PPARG mRNA levels
We next explored in more detail the role of IL-1β and TNFα in primary human keratinocytes and in our HEEs on PPAR expression ( Figure 3C and Figure S1). In IL-1β-treated primary human keratinocytes, PPARD gene expression was increased at 1 h, 3 h and 6 h  Figure 4A). In contrast, blocking of TNFα signalling in HEEs with infliximab, a monoclonal antibody directed towards TNFα, did not abrogate SDS-induced PPARG downregulation ( Figure 4B). Taken together, these findings demonstrate that IL-1β, but not TNFα inhibition prevents the downregulation of SDS-mediated barrier impairment resulting in PPARG decrease ( Figure 4A). Moreover, blocking of IL-1R signalling in HEEs by anakinra significantly diminished SDS-induced PPARD upregulation, in contrast to TNFα treatment ( Figure 4B). Together, these data strongly suggest that IL-1β, but not TNFα, modulates PPARG and PPARD expression after SDS-mediated barrier impairment.

| DISCUSS ION
We report two models of acute epidermal barrier perturbation closely mimicking impaired barrier function in human skin. We utilized SDS and acetone to induce barrier impairment. SDS is a detergent that perturbs the cutaneous barrier and has been utilized to F I G U R E 4 Blocking of IL1 and TNFα signalling in SDS-mediated barrier impairment in HEEs. (A) HEEs were pretreated with anakinra or saline 24 h and immediately before SDS-mediated barrier impairment. HEEs were harvested 6 h and 24 h after SDS application. mRNA expression levels of PPARA, PPARD and PPARG were assessed by RT-PCR. Combined data from 5 independent, paired experiments are presented as mean ±SEM. Values are given as % of respective ctrl HEEs. As ctrl HEEs were used (1) saline-pretreated HEEs exposed to PBS for saline-pretreated HEEs exposed to SDS (groups: "6 h SDS" and "24 h SDS") and (2) anakinra-pretreated HEEs exposed to PBS for anakinra-pretreated HEEs exposed to SDS (groups: "ANA +6 h SDS" and "ANA +24 h SDS"). (B) HEEs were pretreated with infliximab or saline 24 h and immediately before SDS-mediated barrier impairment. HEEs were harvested 6 h and 24 h after SDS application. mRNA expression levels of PPARD and PPARG were assessed by RT-PCR. Combined data from 3 independent, paired experiments are presented as mean ± SEM. Values are given as % of respective ctrl HEEs exposed to PBS. As ctrl HEEs were used (1) saline-pretreated HEEs exposed to PBS for saline-pretreated HEEs exposed to SDS (groups: "6 h SDS" and "24 h SDS") and (2) infliximab-pretreated HEEs exposed to PBS for infliximab-pretreated HEEs exposed to SDS (groups: "INF +6 h SDS" and "INF +24 h SDS"). Data were analysed using a paired Student´s t test for all experiments. *p = <0.05; **p = <0.01. IL1, interleukin-1; TNFα, tumor necrosis factor α; PPARD, peroxisome proliferator-activated receptor β/δ; PPARG, peroxisome proliferator-activated receptor γ instigate irritant contact dermatitis in human skin. 19,[33][34][35][36] Acetone, an organic solvent, has commonly been used to perturb the cutaneous barrier in mice. 29,[37][38][39] Both compounds impair barrier function by removal and/or disturbance of SC intercellular lipid domains that are essential to maintain the epidermal permeability barrier. 19,[37][38][39][40][41][42] As shown in Figure 1C, SDS and, to a lesser extent, acetone decreased TEER, demonstrating that both substances impair barrier function in HEEs with SDS producing a more profound impairment.
Our models reliably recapitulate and extent previous data demonstrating upregulation of IL1A and IL1B and a marked increase of TNFA following cutaneous barrier disruption by sequential tape stripping or SDS treatment of normal skin (Figure 2A,B). 34,35,43,44 Yet, in these prior studies, kinetics of inflammatory mediator mRNA expression was incompletely investigated. Furthermore, whole skin tissue samples, as utilized in most of these reports, do not allow delineating the contribution of the distinct cutaneous compartments and cell types to epidermal cytokine response after epidermal barrier impairment. Thus, changes in gene expression levels, that is in the epidermal compartment might remain undetected. 45 By focussing on solely the epidermal compartment, we here describe changes that can be exclusively attributed to keratinocytes without interference of immune cells, dermal and subcutaneous tissue. This is of particular interest, since keratinocytes as first line of defense sense danger signals of "barrier impairment" and consequently orchestrate the (inflammatory) response to overcome barrier disturbances. 46,47 Thus, our models allow investigating the impact of graded barrier impairment on cytokine gene expression, since SDS treatment resulted in a more pronounced barrier perturbation than acetone treatment ( Figure 1B,C). Acetone treatment for instance resulted in an increase of IL1B expression levels whereas IL33 mRNA levels were reduced 24 h after barrier impairment. IL1A and TNFA gene expression was not significantly altered ( Figure 2B). Since acetone-mediated barrier impairment inflicts a less pronounced barrier perturbation, these data demonstrate that IL1B and IL33 are highly sensitive markers of epidermal barrier impairment in human epidermis ( Figure 2B). Furthermore, these findings show that varying degrees of barrier impairment (SDS vs. acetone) lead to distinct cytokine profiles.
Intriguingly, IL33 and TSLP mRNA expression was dampened 24 h after SDS and acetone treatment, whereas it was triggered 6 h after SDS treatment only (Figure 2A,B). TSLP and IL33 expression were reported to be increased in human and murine skin as early as 6 h after tape stripping. 43,[48][49][50] These results were acquired in fullthickness skin samples without investigating temporal kinetics or graded epidermal barrier impairment. By contrast, we document a decrease of TSLP and IL33 gene expression levels following an initial increase after profound epidermal barrier impairment (Figure 2A).
Additionally, we report decreased mRNA expression levels of both cytokines 24 h after milder epidermal barrier perturbation, in absence of the earlier (6 h) upregulation ( Figure 2B). Both TSLP and IL-33 were proposed to function as alarmins that alert the immune system and spur a Th2 immune response following tissue damage. 49,51,52 Moreover, IL-33 can exert immunosuppressive functions by induction of T-regs after barrier impairment, thereby preventing exaggerated skin inflammation. 50 Furthermore, IL-33 and TSLP were suggested to directly dampen epidermal barrier function by downregulation of FLG expression. [53][54][55] Thus, it is likely that IL33 and TSLP upregulation only occurs shortly after strong epidermal barrier impairment, which requires an involvement of immune cells to protect skin against pathogens or to dampen exaggerated skin inflammation. 50,56,57 Along these lines, it is tempting to speculate that downregulation of IL33 and TSLP might correspond to the termination phase of epidermal barrier recovery after acute impairment. In this scenario mRNA downregulation of IL33 and TSLP potentially might contribute to epidermal barrier restoration by enhancing expression levels of critical epidermal proteins including FLG and CLDN1. 54,58 Moreover, our results show a role of TNFα in IL33 downregulation after epidermal barrier impairment ( Figure 3B). 59 This modulatory function is probably indirect because it requires at least 18 h to take place.
Barrier impairment mediated not only by SDS but also by acetone led to reduced PPARG gene expression levels, thereby underscoring the role of PPARγ in epidermal homeostasis ( Figure 3C). 26,[60][61][62] These data indicate that PPARγ signalling closely correlates with epidermal barrier fitness. Thus, it is not surprising that PPARγ ligands promote epidermal barrier recovery. 61 In line, PPARγ activation increases PPARG mRNA levels. 26 These results are in agreement with findings demonstrating reduced PPARG expression in inflamed skin lesions 10,12,14 and with the beneficial effects of PPARγ ligands in patients with psoriasis and in a murine model of this disease. 4,26 Moreover, IL-1β and TNFα cytokine treatment decreased PPARG gene expression levels ( Figure 3C) and abrogation of IL-1 signalling using anakinra, but not of TNFα, restored normal PPARG mRNA levels after barrier impairment induced by SDS ( Figure 4A). Thus, PPARG downregulation might mainly and directly result from upregulation of IL-1β signalling pathway in keratinocytes. In addition, IL-1β enhances PPARD mRNA levels. This together with the findings that SDS-mediated barrier impairment induced a much greater increase of IL1B gene expression than barrier impairment produced by acetone, implicates that only a profound disturbance of the epidermal barrier triggers PPARD mRNA expression via increased IL-1β levels ( Figure 2). In addition, blocking of IL-1 signalling mitigates the increase of PPARD expression levels observed in SDS-treated HEEs ( Figure 4). These data demonstrate that IL-1β modulates PPARG and PPARD expression at transcriptional level in human keratinocytes in an autocrine/paracrine manner. Moreover, the here presented data suggest that PPARA gene expression might be hardly implicated in epidermal barrier recovery. Furthermore, both in psoriasis and AD IL1B gene expression levels are increased. 63,64 Thus, it is tempting to speculate that IL-1β contributes to PPAR expression changes observed in these diseases. 10,12,[14][15][16][17] In this study, we present an organotypic model to investigate the response of epidermal keratinocytes following barrier perturbation. We demonstrate that IL-1β and IL-33 are pertinent markers of epidermal barrier impairment. Furthermore, we show that keratinocyte-derived IL-1β modulates PPARG and PPARD gene expression in human epidermis. In summary, this work may form the basis for future investigations studying the impact of abrogation of IL-1β signalling on PPARγ and its effects on normalization of barrier homeostasis in common inflammatory skin disorders such as atopic dermatitis and psoriasis.

ACK N OWLED G EM ENTS
We thank Birgit Moser and Viktoria Gapp for their technical assistance. This work was supported by the EU action SKINBAD, by a research grant from Laboratories Expanscience to MS and SD and by a grant #28039 from the Austrian Science Fund (FWF) to SD.

CO N FLI C T O F I NTE R E S T
None.