NRF2 Augments Epidermal Antioxidant Defenses and Promotes Atopy Free

Atopic dermatitis (AD) is a chronic inflammatory skin disorder associated with cutaneous hyperreactivity to numerous environmental triggers (1). As the filaggrin-null allele (2) and abnormal skin peeling (3) are closely associated with the development of AD, hapten penetration is considered to be a critical first step in promoting sensitization (1). One of the hallmark features of AD is aberrant humoral immunity, notably elevations in serum IgE directed against innocuous environmental Ags; this condition may ultimately lead to the atopic march [i.e., progression to allergic rhinitis and asthma in later years (4)]. Although allergic contact dermatitis (ACD) is a frequent complication of AD, whether AD patients are at an increased risk for developing ACD remains controversial (1). Contact hypersensitivity (CHS) is a CTL-mediated immune response that models ACD. Under these conditions, topical administration of a hapten generates immunological memory; a specific immune response is evoked upon re-encountering the Ag (5). There is a growing body of evidence that suggests that the specific chemical properties of the hapten determine the nature of the immunological memory response and thus the nature of the subsequent allergic reaction (1). A strong example of this principle is the observation that tolerogenic haptens, including 2,4-dinitrothiocyanobenzene (DNTB) or bis(3-carboxy-4-nitrophenyl) disulfide, preferentially induces peripheral tolerance through the generation of a regulatory T cell (Treg) memory (6), whereas topical administration of immunogenic haptens, including 1-fluoro-2,4-dinitrobenzene (DNFB) or 1-chloro-2,4-dinitrobenzene (DNCB), results in disruption of the reactive thiol layer in the stratum corneum (SC) and leads to dinitrophenylation and glutathione (GSH) depletion throughout the entire epidermis. Interestingly, dinitrophenylation induced by the tolerogenic hapten DNTB is limited to the SC (7).

The epidermis maintains intrinsic redox-sensitive pathways that respond efficiently to noxious stimuli in the external environment. The actions of transcription factor NF erythroid 2–related factor 2 (NRF2) are critical for establishing and maintaining the thiol gradient that provides essential cytoprotection (8). The downstream targets of NRF2 include the epidermal differentiation complex, a group that includes the cysteine-rich protein, loricrin (LOR) (9, 10), and phase II detoxifying genes that contribute to the synthesis of GSH (8). Cytoskeleton-associated Kelch-like erythroid cell–derived protein with cap'n’collar homology-associated protein 1 (KEAP1) is a negative regulator of NRF2; this protein senses oxidative/electrophilic stress via its reactive cysteine residues (11), whereas NRF2 promotes the repair of LOR-deficient epidermis by transactivating the antioxidant protein, small proline-rich protein 2 (Sprr2) in the stratum granulosum (SG) (12). These lines of evidence suggest that the epidermal repair response (12) and CHS priming (13) are both closely associated with the penetration of thiol-reactive electrophiles into the SG (7). Therefore, we reasoned that the activation of the KEAP1–NRF2 system might generate intrinsic inflammatory cues that break tolerance and promote type 2 immunological memory, a hallmark of AD.

We found that NRF2 is required for the CHS response and for support of the ensuing skew toward local or systemic type 2 immune responses. Through a series of experiments, we concluded that decreased antioxidant host defense in keratinocytes of Nrf2-deficient (Nrf2^−/−) mice was responsible for impaired memory generation among CTLs. Along these lines, the epidermal samples from patients with Netherton syndrome or peeling skin syndrome (PSS), which are conditions associated with premature desquamation and atopic syndromes (3), exhibited augmented expression of both NRF2 and its downstream target, SPRR2. Our results indicate that the keratinocyte-intrinsic oxidative stress response, as opposed to Ag penetration, may provide critical direction to the early phases of atopic disorders that ensue.

BALB/c mice (8–12 wk of age) were purchased from Charles River Laboratories Japan (Yokohama, Japan), and Nrf2^−/− mice were purchased from RIKEN BioResource Research Center (Tsukuba, Japan) (14). All mice were maintained under specific pathogen–free conditions at the animal facility of the University of Tsukuba, and all procedures were approved by the University of Tsukuba Ethics Committee.

Abdominal skin was shaved and then painted with 25 μl of 0.5% DNFB (D0835; Tokyo Chemical Industry, Tokyo, Japan) in 4:1 acetone/olive oil (AOO). Five days later, the dorsal surfaces of the ears were challenged with 10 μl of 0.3% DNFB. Ear thickness was measured using a digital micrometer (293-230-30 MDC-25-MX; Mitsutoyo, Kawasaki, Japan) at the time points indicated, and the delayed-type reaction (DTR) was evaluated quantitatively. To generate immediate-type reaction (ITR), 20 μl of 1% 2,4,6-trinitro-1-chlorobenzene (TNCB) (C0307; Tokyo Chemical Industry) was applied to the dorsum of the ear, and the ITR was elicited by repetitive application of 20 μl 1% TNCB every other day for 32 d.

For the adoptive transfer of CHS, mice were sensitized with 0.5% DNFB; 25 μl was administered to the abdomen and 10 μl to each paw. Five days later, draining lymph nodes (DLNs) were collected from axillary, brachial, popliteal, and inguinal regions, and single-cell suspensions were generated by physical disruption followed by filtration through a 40-μm screen. Naive mice were injected i.v. with 5 × 10⁷ DLN cells. Two hours later, baseline ear thickness was measured, and the mice were challenged with 10 μl of 0.3% DNFB. Ear thickness was reevaluated 24 h later.

Epidermal sheets were prepared from ear skin that was collected and incubated with 0.5% trypsin (207-19183; FUJIFILM Wako Pure Chemical, Osaka, Japan) in PBS for 30 min at 37°C. Epidermal sheets were stirred in PBS containing 0.05% DNase I from bovine pancreas (DN25; Sigma-Aldrich, St. Louis, MO) and 10% FBS for 15 min at room temperature. Epidermal cell (EC) suspensions were obtained by filtering the epidermal sheet suspension through a 100-μm screen. DNP-conjugated ECs were prepared by incubating the cell suspension with 100 μg/ml 2,4-dinitrobenzene sulfonic acid hydrate (DNBS) (556971; Sigma-Aldrich) for 30 min at 37°C. Naive mice were injected s.c. with 2 × 10⁶ DNP-conjugated ECs suspended in PBS. Five days later, baseline ear thicknesses were measured, and 20 μl of 0.3% DNFB was applied to the ear skin; ear thickness was measured again at 24 h. For intervention studies, ECs were resuspended in EpiLife medium (MEPI500CA; Thermo Fisher Scientific, Waltham, MA) with N-acetylcysteine (NAC) (013-05133; FUJIFILM Wako Pure Chemical; 7 mg/ml, neutralized to [pH 7.4] with NaOH) for 1 h at 37°C.

Monobromobimane (MBB) was used to detect the thiol-rich layer in the SG. Cryosections prepared from the back skin of neonatal mice were stained with 90 μM MBB (B4220; Tokyo Chemical Industry) for 20 min, as described previously (7).

The abdominal skin of the mice was shaved and painted with 25 μl of 0.5% DNFB in AOO or 100 μl of 1% DNTB (T0199; Tokyo Chemical Industry) in AOO. Cells from DLNs were collected 5 d later as described above. Cells in suspension were stained with the following mAbs purchased from BioLegend (San Diego, CA): allophycocyanin-conjugated anti-CD3 (17A2), FITC-conjugated anti-CD4 (GK1.5), and PE-conjugated anti-CD8a (53-6.7). For detection of Tregs, cells in suspension were first stained with the following mAbs purchased from BioLegend; allophycocyanin-conjugated anti-CD3 (17A2), PE/cyanin7–conjugated anti-CD4 (GK1.5), and Brilliant Violet 421–conjugated CD25 (PC61); this was followed by fixation/permeabilization with True-Nuclear Transcription Factor Buffer Set (424401; BioLegend) and staining with PE-conjugated anti-Foxp3 (150D).

Human epidermal keratinocytes, neonatal (HEKn) (C0015C; Thermo Fisher Scientific) were maintained in EpiLife medium (MEPI500CA; Thermo Fisher Scientific) with EpiLife Defined Growth Supplement (s0125; Thermo Fisher Scientific). Lentiviral particles encoding short hairpin RNA targeting Keap1 mRNA (sc-43878-V; Santa Cruz Biotechnology, Dallas, TX) were used. Lentiviral particles encoding a scrambled short hairpin RNA sequence (sc-108080; Santa Cruz Biotechnology) were used as a control. HEKn cells were transduced with these lentiviral particles in accordance with the manufacturer’s instructions, and the knockdown efficacy was assessed at the mRNA level by quantitative real-time PCR. Primers for GAPDH (4310884E; Thermo Fisher Scientific) were used for internal reference.

HEKn were stimulated with 10 μM of tert-butylhydroquinone (112941; Sigma-Aldrich), 100 μg/ml DNBS (556971; Sigma-Aldrich), 1 mg/ml OVA (A5503; Sigma-Aldrich), 1 mg/ml house dust mite (HDM) (LG-5449; Cosmo Bio, Tokyo, Japan), and 1 mg/ml of cedar pollen (CP) (LG-5229; Cosmo Bio). After 30 min, whole cell lysate was dissolved in SDS (NaDodSO₄) extraction buffer (Tris-HCl [pH 6.8], 2% SDS, 0.86 M 2-ME, and 10% glycerol).

Skin samples from healthy patient (Control), AD patient (AD1), Netherton syndrome patients (Netherton syndrome 1, 2), and a PSS patient (PSS1) were subjected to immunohistochemical staining with NRF2 and SPRR2. Written informed consent was provided from all subjects under the approval from the Institutional Review Board at the University of Tsukuba (H29-003; Tsukuba Clinical Research and Development Organization) and the Juntendo University Urayasu Hospital (2013074; the Ethics Committees at Juntendo University Urayasu Hospital).

Skin samples were fixed overnight with 10% buffered formalin. After paraffin embedding, 3-μm sections were deparaffinized and subjected to routine Ag retrieval and blocking procedures. Sections were then incubated overnight with primary Abs at the following dilutions: anti-CD4 (1 μg/ml, 14-9766-80; Invitrogen, Carlsbad, CA), anti-CD8 (1 μg/ml, 14-0808-82; Invitrogen), anti-NRF2 (12, 15) (1:500), and anti-SPRR2 (1:5000; kindly provided by Dr. D. Hohl). After the blocking of endogenous peroxidase with PBS containing 0.3% NaN₃ and 0.01% H₂O₂, the sections were incubated for 1 h with secondary Abs at the following dilutions: biotinylated goat anti-rabbit IgG Ab (3 μg/ml, BA-1000; Vector Laboratories, Burlingame, CA) and biotinylated goat anti-rat IgG Ab (3 μg/ml, BA-9400; Vector Laboratories). Visualization was performed using the ABC Developing System (PK-7100; Vector Laboratories) and peroxidase substrate (SK-4105; Vector Laboratories) according to the standard protocol, followed by light counterstaining with hematoxylin. The numbers of infiltrating mast cells, basophils, and eosinophils in the dermis were quantified by counting toluidine blue–, Alcian blue–, and Congo Red–positive cells, respectively, per 20 random high power fields using a digital microscope (AX80; Olympus, Tokyo, Japan) and FLOVEL image filing system (FLOVEL, Tokyo, Japan). The results were presented as the mean number of stained cells ±SEM per high power field from four mice.

An equal amount of protein was subjected to SDS-PAGE on Mini-PROTEAN TGX Precast Gels (4569036; Bio-Rad Laboratories, Hercules, CA), and the proteins were transferred onto nitrocellulose membranes (10600001; GE Healthcare, Buckinghamshire, U.K.). The membranes were incubated overnight with the following primary Abs; anti-NRF2 (1:200; kindly provided by Dr. Roop) and anti-tubulin (1 μg/ml, 017-25031; FUJIFILM Wako Pure Chemical) followed by a 60 min incubation with HRP-labeled secondary Abs against rabbit or mouse IgG (0.04 μg/ml, sc-2004 or sc-2005; Santa Cruz Biotechnology). Ab binding was visualized and enhanced with SuperSignal West Dura Extended Duration Substrate (34075; Thermo Fisher Scientific) and an image analysis system (LAS4000 mini; Fujifilm, Tokyo, Japan).

Serum IgE levels were measured using a specific ELISA kit (432401; BioLegend) as per manufacturer’s instructions. Serum trinitrophenyl (TNP) IgE levels were measured using Abs directed against TNP–BSA (2 μg/ml, T-5050-10; Biosearch Technologies, Hoddesdon, U.K.) and mouse IgE (0.1 μg/ml, GTX77227; GeneTex, Irvine, CA). EC suspensions were incubated with 100 μg/ml DNBS for 30 min, and EC-derived IL-1α levels were measured using a specific ELISA kit (433404; BioLegend) as per manufacturer’s instructions.

Samples of full-thickness ear skin were collected and homogenized in TRIzol reagent (15596-026; Thermo Fisher Scientific). Total RNA was isolated, and cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit (4368814; Thermo Fisher Scientific). The following primers specific for each target with double-quenched probes were purchased from Integrated DNA Technologies (Coralville, IA): Ifng (Mm.PT.58.41769240), Il4 (Mm.PT.58.7882098), Il13 (Mm.PT.58.31366752), Il10 (Mm.PT.58.13531087), Il1a (Mm.PT.58.32778767), Sprr2d (Mm.PT.58.42683792.g), quinone 1 (Nqo1; Mm.PT.58.10871473), Slc7a11 (Mm.PT.58.29117975), glutamate-cysteine ligase catalytic subunit (Gclc; Mm.PT.58.30656560), thymic stromal lymphopoietin (Tslp; Mm.PT.58.41321689), and IL-1A (Hs.PT.58.40913627). Primers for Gapdh (4352339E; Thermo Fisher Scientific) were used for internal reference. All data were analyzed using a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). The interpretation of quantitative real-time PCR data was conducted by the comparative cycle threshold method.

Single-cell suspensions from DLNs were prepared and stained with mAbs. Isotype-matched Abs were used as controls. A LIVE/DEAD Fixable Violet or Near-IR Dead Cell Stain Kit (Life Technologies, Gaithersburg, MD) was used to distinguish dead cells. Flow cytometric analysis was performed on a Gallios flow cytometer (Beckman Coulter, Fullerton, CA).

Data are presented as means ± SEM. Comparisons were performed with unpaired t tests with Welch correction (between two groups) or two-way ANOVA (time-course experiments) by use of GraphPad Prism 6 software (GraphPad, La Jolla, CA). In all studies, p <0.05 indicated statistical significance.

NRF2 promotes upregulation of critical antioxidants, including Sprrs, in response to electrophililes and mediates the repair of the compromised epidermal barrier at the SG (12). As anticipated, a single topical application of 1% DNFB (200 μg) to ear skin resulted in a significant increase in expression of NRF2 downstream targets, including Sprr2d, the phase II detoxifying enzyme NAD(P)H dehydrogenase, Nqo1, and CHS initiator Il1a that is enriched in the SC (16), as compared with the levels of expression induced by 1% DNTB (200 μg) (Fig. 1A). By contrast, both DNFB and DNTB induced comparable levels of Gclc, a response that may be related to GSH depletion (7). However, compared with DNTB, DNFB promoted significantly increased expression of solute carrier family 7 member 11 (Slca11, xCT; Fig. 1A). This discrepancy suggests that DNFB may have a profound effect on extracellular thiol levels; DNFB has a substantially stronger capacity for thiol oxidation than DNTB. Indeed, extracellular thiol levels rather than intracellular GSH concentrations are the primary factors regulating xCT expression (17). Intriguingly, both DNFB and DNTB treatment generated comparable levels of Tslp (Fig. 1A). These results suggest that the immunogenicity of the reactive hapten, DNFB, may rely directly on NRF2-mediated antioxidant host defense mechanisms, whereas DNTB preferentially induces tolerance, as it is not capable of generating this response (7). To address this hypothesis, we subjected Nrf2^−/− mice to single applications of DNFB or DNTB. In the steady state, the SG in tissue samples from both wild-type (Nrf2^+/+) and Nrf2^−/− mice revealed comparable thiol-rich bands as detected by probing with MBB; this finding refutes the possibility that NRF2 deficiency results in a developmental defect in skin tissue (Fig. 1B). Nonetheless, application of DNFB to Nrf2^−/− mice resulted in significantly decreased expression levels of SPRR2 compared with the responses of their Nrf2^+/+ counterparts (Fig. 1B). By contrast, we observed only minimal DNTB-induced Sprr2 expression in Nrf2^+/+ mice (Fig. 1B). From these findings, we concluded that when compared with the responses to DNTB, DNFB evokes a strong NRF2-mediated antioxidant response that may contribute to its immunogenicity.

CHS experiments performed on Nrf2^−/− mice yielded conflicting results. Kim et al. (18) reported that CHS was attenuated in older (aged 9–22 mo) Nrf2^−/− mice, whereas others reported that NRF2 deficiency resulted in aggravated CHS, a result that stressed the importance of NRF2 in regulating critical components of the inflammatory response such as neutrophil recruitment (19). The latter findings are consistent with other reports that focus on a critical anti-inflammatory role for NRF2 (11). To address this issue, we used Nrf2^−/− mice on the BALB/c background, which are a high IgE responder; we evoked an AD-like phenotype by repetitive topical application of TNCB (20). Interestingly, Nrf2^−/− mice exhibited significantly decreased levels of ITR (Fig. 2A), effector cell infiltration (Fig. 2B), type 2 cytokine expression (Fig. 2C), and total serum IgE (Fig. 2D). However, the levels of TNP-specific IgE (Fig. 2E) in serum samples from Nrf2^−/− mice were comparable to those detected in Nrf2^+/+ controls, suggesting that NRF2 deficiency had a primary impact on the adaptive CTL response (5), as opposed to humoral immunity. Accordingly, Nrf2^−/− mice with CHS secondary to DNFB exhibited significantly attenuated levels of DTR (Fig. 2F) and expression of both type 1 and type 2 cytokines (Il4, Il10, Il13, and Ifng; Fig. 2G), when compared with their Nrf2^+/+ counterparts. Nrf2^−/− mice also demonstrated attenuated CHS against haptens FITC plus dibutyl phthalate or 4-ethoxymethylene-2-phenyl-2-oxazolin-5-one compared with Nrf2^+/+ counterparts (Supplemental Fig. 1).

A series of adoptive transfer experiments were performed to determine the factors contributing to the attenuated CTL response in Nrf2^−/− mice. Transfer of DNFB-primed DLN cells from Nrf2^−/− mice resulted in limited ear swelling compared with observations made in experiments performed in Nrf2^+/+ mice (Fig. 3A); these results indicate that NRF2 primarily regulates the priming phase of CHS. Proliferation of T cells was reduced in DLN cells from Nrf2^−/− mice in response to a single topical application of 0.5% DNFB (125 μg); this treatment also resulted in a smaller fraction of CD25⁺ CD4⁺ Foxp3⁺ Tregs compared with Nrf2^+/+ mice (Fig. 3B). By contrast, T cell counts and the Treg population in DLN cell suspensions from Nrf2^−/− and Nrf2^+/+ mice were comparable to one another after topical application of 1% DNTB (1000 μg) (Fig. 3C). Taken together, these results indicated that Nrf2^−/− mice failed to generate a CTL memory response that is a critical feature underlying the development of CHS (5).

The application of 0.3% DNFB (30 μg) to ear skin induced significantly lower Tslp expression levels in Nrf2^−/− mice compared with Nrf2^+/+ mice, whereas 0.3% DNTB (30 μg) resulted in minimal changes in Tslp throughout (Fig. 4A). Accordingly, Nrf2^−/− EC suspension pulsed with DNBS produced significantly smaller amounts of IL-1α (Fig. 4B), a factor that is critically required for CHS priming (16). When KEAP1 expression was silenced in cultured HEKn, the expression levels of IL-1A were significantly increased, suggesting that NRF2 abundance maintains IL-1A expression (Fig. 4C) Similarly, adoptively transferred DNP-pulsed EC cells from Nrf2^−/− mice evoked a significantly decreased levels of DTR compared with those from Nrf2^+/+ mice; likewise, EC cells from Nrf2^+/+ mice restored CHS in Nrf2^−/− recipients (Fig. 4D). Intervention with NAC likewise restored DTR in the Nrf2^−/− strain (Fig. 4E); bone marrow chimerism revealed that Nrf2^−/− recipients consistently experienced decreased ear swelling compared with Nrf2^+/+ recipients, whereas Nrf2^−/− donors did not (Supplemental Fig. 2). Furthermore, common environmental allergens, such as OVA, HDM, or CP stabilized NRF2 in cultured HEKn, suggesting that the NRF2-mediated antioxidative response in keratinocytes is augmented by direct exposure to allergens (Fig. 4F). These results indicated that NRF2 regulates keratinocyte-derived inflammatory responses that condition APCs toward CTL memory generation; these factors lead to systemic type 2 immune response after repeated provocation (21).

To validate the clinical relevance of our findings, we evaluated the relative levels of NRF2 by immunohistochemistry in patients with genetic disorders that promote skin peeling and that manifest with severe atopy (3). Netherton syndrome (OMIM number 256500) and PSS (OMIM number 609796) are rare, autosomal-recessive genodermatoses. In earlier studies, Allen et al. (22) found that the epidermis of Netherton syndrome patients displayed enhanced absorption of the topical immunosuppressive agent tacrolimus and high levels of keratinocyte-intrinsic cytokines that promote type 2 immunological memory (21), including IL-1α (22) and TSLP (3). We hypothesized that skin peeling (premature desquamation) exposes keratinocytes in the SG to ambient air; this might result in oxidization of the cysteine residues of KEAP1 and thereby initiate atopic sensitization. Epidermal tissue from patients with AD, Netherton syndrome, and PSS exhibited high levels of NRF2 and its downstream target, SPRR2 (12, 15), when compared with tissue from healthy controls (Fig. 5). Our results suggest that NRF2 provides a link between atopy and defective barrier function. Furthermore, our results suggest that the thiol-mediated biochemical responses in differentiated epidermal tissue determine the nature and outcome of cutaneous immunological memory following percutaneous immunization.

Consistent with the previous report by Kim et al. (18), we have found that loss of NRF2 resulted in the attenuation of CHS, decreased generation of IL-1α, a keratinocyte-intrinsic factor necessary for the generation of CTL memory (16), and was responsible for the attenuation of this response. Furthermore, epidermal tissue from patients diagnosed with skin peeling disorders associated with atopic manifestations displayed increased levels of NRF2 and likewise displayed increased levels of its downstream mediator SPRR2. Given the role of the epidermis as a frontline host defense system and the thiol gradient established by the KEAP1/NRF2 system (8), the thiol-based antioxidant system may conveniently be poised to sense xenobiotic haptens and thereby skew the epidermal immune microenvironment toward type 2 immunity following repeated immunizations (21, 23). In alignment with this hypothesis, CHS was inhibited by the supplementation with the antioxidant NAC, a biomolecule that inhibited the thiol-coupling reaction in the early phase of CHS (24).

From a structural perspective, the SC is a macromolecule composed of cross-linked proteins and lipids (25). Functional NRF2 in the SG allows LOR-deficient mice to withstand harsh environmental conditions (9, 12). Thiols are readily oxidized in ambient air to generate disulfide bonds. This autoxidative property stabilizes keratin intermediate filament networks and promotes corneocyte maturation above the SG (26). Recent phylogenic evidence has suggested that the acquisition of terrestrial lifestyles coincided with the emergence of the epidermal differentiation complex (27). Therefore, thiol-enriched, redox-sensitive systems in the outermost layer of the epidermis would provide a definitive evolutionary advantage for terrestrial organisms. This redox exchange not only transactivates LOR and stabilizes SC via extensive disulfide bond formation (9, 10) but also protects the body surface from oxidative insults, including UV rays or electrophilic biomolecules (9, 12, 28).

In the clinical conditions, such as Netherton syndrome, which are characterized by impaired corneocyte cohesion and premature desquamation, the KEAP1–NRF2 system in the upper epidermis could be readily activated by ambient air (8). This may result in continual and/or chronic production of IL-1α from keratinocytes, which would skew local immune responses away from tolerance (3, 22, 29). In the same line of thinking, lipophilicity is a critical chemical property that correlates with irritation resulting from application of DNCB versus DNTB (7). This chemical property may also promote perturbation of the SC lipid bilayers (30) as well as cause irritation, activate the inflammasome, and break tolerance (31). However, given that oral delivery of DNFB (via a nonkeratinized epithelium) induces tolerance (32), it would be reasonable to hypothesize that the epithelial tissue structure is the primary determinant of immune effector function (33). Compared with the intestinal epithelium, the epidermis maintains highly impermeable layers of SC above the tight junctions (26). Functionally, the epidermis primarily provides the impermeability, whereas the intestinal epithelium absorbs nutrients, water, or electrolytes. Therefore, as Pickard et al. (7) have proposed earlier, the KEAP1–NRF2 system appears to constitute the specialized thiol-based “biochemical barrier” located above the tight junctions (26) that is unique to the keratinizing squamous epithelium.

In conclusion, the prototypical type 2 immunological memory generated in this model following repeated Ag exposure may have been instrumental for eradicating xenobiotic stress-related neoantigens that are associated with senescence or carcinogenesis (34–36). The degree of antioxidant host defense achieved through the thiol-based sensor-effector mechanisms (11) could be a functional determinant of the epidermis-intrinsic immunologic memory.

The authors have no financial conflicts of interest.