Essential Role of CARD14 in Murine Experimental Psoriasis

Psoriasis is a chronic inflammatory skin disorder that is predominantly characterized by sharply demarcated chronic erythematous plaques (1). Although its etiological mechanisms are largely unknown, recent evidence suggests that the topical application of imiquimod (IMQ) cream causes psoriasis-like skin inflammation in humans and mice (2, 3). Although IMQ is an agonist of mouse TLR7, TLR7-deficient mice display epidermal hyperplasia comparable to that displayed by wild type (WT) mice in an IMQ-induced psoriasis model (4), suggesting that TLR7-independent mechanisms contribute to this pathogenesis. Indeed, accumulating evidence has suggested that both TLR7 and TLR9 are involved in the activation of dendritic cells (DCs) in human psoriasis, indicating that the innate immune signaling pathways mediated by multiple TLRs are critical for the development of psoriasis (5). However, no reports have firmly demonstrated the dual requirement of both TLR7 and TLR9 by using doubly deficient mice for TLR7 and TLR9, for example.

In addition to innate immune recognition and signaling, inflammatory cytokines such as IL-17 and IL-23 hold the key to understanding the pathogenesis of psoriasis (6). mAbs against IL-23p19 (guselkumab and tildrakizumab) showed more clinical benefit than blockage of the IL-12/23p40 subunit in human patients (6, 7). In an experimental setting, an absence of IL-23p19 in vivo abrogated IMQ-induced psoriasis-like skin inflammation (2). Vice versa, intradermal injections of mouse ears with recombinant IL-23 protein (rIL-23) also reproduced many features of psoriasis, such as the upregulation of IL-22 and epidermal hyperplasia via the activation of STAT3 (8, 9). Overall, the IMQ- and IL-23–induced skin inflammation models in mice tightly linked to IL-17 and IL-22 production are useful animal models for psoriasis (8–10).

Despite increasing evidence that IL-23p19 plays a crucial role in autoimmune diseases, the cell types involved and the mechanisms of IL-23p19 production are not fully understood. Langerin⁻ conventional DCs have been shown to be the major sources of IL-23p19 in IMQ-induced psoriasis (11), whereas other reports suggest that Langerin⁺ Langerhans cells (LCs) in mice (12) and humans (13) in psoriasis, as well as epidermal keratinocytes in patients with atopic dermatitis (14) or psoriasis (15), are potential sources of IL-23p19.

Nevertheless, the secreted IL-23p19 in turn can activate IL-17– and IL-22–producing γδ T cells, critical cell types in the pathogenesis of psoriasis (9). IL-23R is highly expressed on γδ T cells, and mutations in IL-23R are associated with psoriasis in humans (9, 16). In murine skin, γδ T cells exist abundantly as dendritic epidermal γδ T cells (DETCs), which are characterized by the expression of the Vγ5⁺ TCR (nomenclature of Heilig and Tonegawa) and participate in tissue surveillance and wound healing (17–19). Vγ4⁺γδ T cells are also the main producers of IL-17 in psoriasis models (9, 19). γδ T cells are infrequent in human skin compared with the number of DETCs in healthy mouse skin, whereas Vγ9Vδ2 T cells are reported to be a proinflammatory subset in psoriasis patients (20).

Recently, caspase recruitment domain family member 14 (CARD14, also known as CARMA2 or BIMP2), encoded at the psoriasis susceptibility locus 2 located in human chromosomal region 17q25.3, was shown to have unique gain-of-function mutations associated with psoriasis (21). This finding was confirmed in a genome-wide association study of psoriasis susceptibility locus 2, which identified several risk-associated variants, including mutations in CARD14 (16). CARD14, like CARD11, is a novel membrane-associated guanylate kinase family member, which functions as an upstream activator of B cell lymphoma 10 (BCL10) and NF-κB signaling (22). Although CARD11 is preferentially expressed in hematopoietic cells, CARD14 is expressed more ubiquitously in certain tissues such as skin (21). CARD14 interacts with BCL10 and MALT1, leading to the formation of the CARD14–BCL10–MALT1 complex (23, 24). This complex activates NF-κB through the IκB kinase complex in response to upstream stimuli (23, 24). Current evidence indicates that CARD14 is a key molecule in psoriasis and pityriasis rubra pilaris, but the mechanisms and correlations with the immune responses are unknown (21, 25).

Because little is known about the immunological role of CARD14 in psoriasiform skin inflammation in vivo, we evaluated its immunological function in murine psoriasis-like models. In this study, we examine the pathophysiological role of CARD14 in IMQ-induced psoriasiform skin inflammation by generating Card14^−/− mice.

The 8–12 wk old Card14^−/− (details described below and in Supplemental Fig. 1), Tlr7^−/−, Tlr9^−/−, and Tlr7^−/−Tlr9^−/− doubly deficient mice (26) used in this study were bred and maintained under specific pathogen-free conditions in our animal facility (National Institutes of Biomedical Innovation, Health and Nutrition). Rag2^−/−Il2rg^−/− mice were obtained from Taconic Biosciences. CD45.1⁺ congenic mice on a C57BL/6 background (Ly5.1-B6) were obtained from the RIKEN BioResource Center (Tsukuba, Japan). Tcrd^−/− mice were from the Jackson Laboratory. C57BL/6J (WT) mice were purchased from CLEA Japan. All experiments were conducted in accordance with the institutional ethical guidelines for the care and use of laboratory animals of National Institutes of Biomedical Innovation, Health and Nutrition.

Card14^−/− mice were purchased from and originally generated by Lexicon Pharmaceuticals using a targeted vector designed to remove part of exon 2 and exon 3 by homologous recombination. PCR was used for routine mouse genotyping with the following primers: P1, 5′-GGGTGTTCCTCTGACTCTCCCAGTTGGATG-3′ (forward); P2, 5′-GCTGACCGCTTCCTCGTGCTTTACGGTATC-3′ (forward); and P3, 5′-CAGTGACTCAAGGAGGGGCAAACGCCTATG-3′ (reverse). Card14^−/− targeted mice were backcrossed at least eight times into the C57BL/6J background, and identified using the marker-assisted speed congenic method (27).

To induce psoriasiform skin inflammation, both mouse ears were treated as described previously for six consecutive days with a topical dose of 62.5 mg of Beselna cream (outside Japan, IMQ cream is sold as Aldara; 3M Pharmaceuticals) containing 5% IMQ (purchased from Mochida Pharmaceutical) (2). Control mice were not treated with the cream. Intradermal injections of 20 μl of PBS, either alone or containing 500 ng of mouse recombinant cytokine proteins (rIL-22 or rIL-23) (BioLegend), were given in both ears of the anesthetized mice with a 30 gauge needle every other day for 5 d, as described previously (8, 28). Ear measurements were made at the center of the ears using a constant-pressure thickness gauge (PG-16J; Teclock). The mice were sacrificed and their tissues were collected.

Bone marrow (BM) cells (5 × 10⁶) were obtained by flushing the femurs and tibias of donor Ly5.1-B6 or WT mice. The 6 wk old mice used for this experiment (WT, Ly5.1-B6, or Card14^−/−) received a lethal dose of x-rays (8.5 Gy), then reconstituted with an i.v. injection of BM cells through the orbital sinus. Chimerism in the peripheral blood was measured as the percentage of donor cells (>90%) in the chimeric mice at least 10 wk after reconstitution, after which applications of IMQ cream were commenced.

The mice were sacrificed on the final day of the experiment. Their ears were collected and fixed in 10 N formalin for subsequent embedding in paraffin. Paraffin tissue sections (4 μm thick) were deparaffinized and stained with a rabbit polyclonal Abs directed against mouse keratin 5 (K5) (1:400; Covance), or a rabbit mAb directed against mouse Ki67 (Thermo Fisher Scientific). Before the primary Ab was added, the sections were treated with 0.01 M citric acid (adjusted to pH 6) for 10 min at 80°C. The secondary Ab was a peroxidase-labeled polymer conjugated to a goat anti-rabbit IgG Ab (Dako). For the chromogen, 3,3′-diaminobenzidine (Dako) was used, producing a brown-colored stain (29). Slides were visualized with an absolute magnification of ×10, and ×20 using a microscope (Axioplan2; Carl Zeiss Vision), and images were analyzed using Photoshop Elements (version 6; Adobe).

Epidermal cell suspensions were prepared by incubating the epidermis with trypsin-EDTA for 45 min at 37°C. A buffer containing 1.6 mg/ml collagenase IV (Worthington) and 1.2 mg/ml hyaluronidase (Sigma) in complete RPMI 1640 medium (Nacalai Tesque) was used to produce dermal cell suspensions. The cells were filtered through 70 μm pore-size cell strainers (BD Falcon) to obtain single-cell suspensions (30). In vitro, the cells were cultured with medium alone, or with 100 ng/ml of rIL-1β (BioLegend) and/or 100 ng/ml of rIL-23 (BioLegend) for 20 h at 37°C.

Both ears with or without IMQ 4 h after application were collected, and homogenized by adding 5 mm diameter Zirconium beads (Funakoshi, Japan) in RLT buffer (RNeasy kit; Qiagen, Germany) and shaking with a MixerMill 300 (Qiagen) at a speed of 20 Hz for 5 min. Total RNA was isolated using TRIzol reagent (Invitrogen) and purified using RNeasy Mini Kit (Qiagen), according to the manufacturer’s instructions. For sorted cells, epidermal and dermal cells were prepared the same as for the FACS sample preparation, and integrated both tissues as one sample. Cells were stained and sorted by FACSAria II. Sorted cells were directly added in TRIzol regent and purified using NH₄OAc/ethanol. RNA was then reverse transcribed to cDNA with reverse transcriptase (ReverTra Ace; Toyobo) and an oligo(dT) primer. The real-time PCR mixture consisted of 10 μl of SYBR Green Master Mix (Roche), forward and reverse primers (200 nM), and 2 μl of cDNA sample in a total volume of 20 μl. Primers used for amplification were: Aqp4, 5′- AGCAATTGGATTTTCCGTTG-3′ (forward), 5′- ATGATAACTGCGGGTCCAAA -3′ (reverse); Bcl10, 5′-AAACTGGAGCACCTCAAAGG-3′ (forward) and 5′-CGGAATTGCACCTAGAGAGG-3′ (reverse); Card14, 5′-TGCATAGCTCCCGTTTCAC-3′ (forward) and 5′-GGAACTTCAGGCTTTCCAGA-3′ (reverse); Ccr7, 5′-GTGGTGGCTCTCCTTGTCAT-3′ (forward) and 5′-AGTCCACCGTGGTATTCTCG-3′ (reverse); Cxcl1, 5′-GACTCCAGCCACACTCCAAC-3′ (forward) and 5′-TGACAGCGCAGCTCATTG-3′ (reverse); Gapdh, 5′-CAAGATTGTCAGCAATGCATCC-3′ (forward) and 5′-CCTTCCACAATGCCAAGTTG-3′ (reverse); Il23p19, 5′-CACCTCCCTACTAGGACTCAGC-3′ (forward) and 5′-CTGCCACTGCGACAGAAC-3′ (reverse); and Pik3ca, 5′-AAGTGTGTGGCTGTGACGAA-3′ (forward) and 5′-CGGCAGCTGAGAGTATAGGC-3′ (reverse). The PCRs were performed in a MicroAmp Optical 384-Well Reaction Plate for 40 cycles (95°C for 15 s, 60°C for 1 min) in the LightCycler 480 System II (Roche). The fluorescent signals (∆ threshold cycle) detected during the threshold cycle were recorded by the software installed on the machine. To standardize the target gene levels with respect to the variability in RNA and cDNA quality, Gapdh was amplified under the same conditions, as an internal control.

Total RNA was isolated from control and IMQ cream–treated (4 h) whole-ear skin from WT, Card14^−/−, Tlr7^−/−, Tlr9^−/−, and Tlr7^−/−Tlr9^−/− mice using the per-cell normalization method (31). Equal aliquots of total RNA from six to eight mice were pooled (total, 10 μg), and treated with DNase with the RNase-Free DNase Set (Qiagen). In brief, 200 ng of total RNA was labeled with Cy3 using the SurePrint G3 Mouse Gene Expression v2 8 × 60K Microarray Kit (Agilent) at the DNA-chip Development Center for Infectious Diseases (Research Institute for Microbial Diseases, Osaka University). After background subtraction, quartile normalization was performed with the limma package (32) in R [R Development Core Team (2011)] (33). IL-23 target genes were obtained from Suárez-Fariñas et al. (34). We extracted 538 genes that were upregulated by IL-23 treatment (fold change > 3) and were downregulated by additional treatment with an antagonist of TLR7 and TLR9 (IMO-3100) or an antagonist of TLR7, TLR8, and TLR9 (IMO-8400) (fold change < −1.5) as the IL-23– target genes for further analysis. The expression intensities of these IL-23 target genes in WT, Card14^−/−, Tlr7^−/−, Tlr9^−/−, and Tlr7^−/−Tlr9^−/− mice, treated with or without IMQ cream, were visualized on a heatmap. The made4 package (35) in R was used to create the heatmap with a dendrogram (method = average, distance = pearson). Gene expression microarray data were deposited on the Gene Expression Omnibus under accession number GSE104603 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE104603).

The following Abs were purchased from BD PharMingen or BD Horizon: anti-CD11c (HL3, allophycocyanin), anti-CD19 (1D3, BV510), anti-CD3ε (17A2, BV605), anti-CD4 (RM4-5, PE-CF594), anti–MHC class II (2G9, FITC), anti-Vγ5 (536, BV421), anti-CD45 (30-F11, PerCP/Cy5.5), anti-mouse γδ TCR (GL3, PE), and anti–IL-17A (TC11-18H10, PE). The following Abs were from BioLegend: anti-CD14 (Sa14-2, BV510), anti-CD207 (4C7, PE), anti-CD3ε (N418, PE/cy7), anti-CD4 (RM4-5, PE/Cy7), anti-CD45 (30-F11, PE/Cy7), anti–IL-17A (TC11-18H10.1, Alexa Fluor 700), anti-TCRβ (H57-597, PerCP/Cy5.5), anti-mouse γδ TCR (GL3, Pacific Blue), anti-Vγ1 (211, FITC), and anti-Vγ4 (UC3-10A6, PE). The following Abs were purchased from eBioscience: anti-CD19 (1D3; Pacific Blue), anti–IL-22 (IL22JOP, allophycocyanin), and anti-TCR γδ (GL3, allophycocyanin). Cells were preincubated with purified rat anti-mouse CD16/CD32 (BioLegend) and LIVE/DEAD staining (aqua or blue; Thermo Fisher Scientific) for 15 min, and then labeled with the appropriate cell-surface Ab for 30 min at 4°C. To stain intracellular cytokines, cells were stimulated with 50 ng/ml PMA (Sigma) and 500 ng/ml ionomycin (Sigma), and treated with GolgiStop (BD Biosciences) for 5 h at 37°C. For intracellular staining, cells were fixed and permeabilized with the BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD Biosciences), then stained with Ab for 30 min at 4°C. A BD LSR II flow cytometer was used for the analysis. A FACSAria flow cytometer was used for cell sorting. The samples were then analyzed with FlowJo software (Tree Star).

Prism software (GraphPad) was used for all statistical analyses. Data were compared with an unpaired t test, one-way ANOVA, or two-way ANOVA. Where significant differences were found, the Tukey–Kramer multiple-comparisons test was used to identify differences between individual groups.

To characterize the role of CARD14 in psoriasiform skin inflammation in vivo, we generated CARD14-deficient (Card14^−/−) mice in a C57BL/6 background (Supplemental Fig. 1). The Card14^−/− mice were viable and did not show gross anatomical abnormalities, and display normal skin T cell numbers in both epidermis and dermis in steady state (Supplemental Fig. 2). In addition, CARD14 did not influence the migratory and T cell priming capacity of cutaneous DCs, as immunization of Card14^−/− mice with T dependent Ag (OVA) with or without IMQ cream as an adjuvant displayed comparable OVA-specific B, αβ CD4 T, and CD8 T cell responses (Supplemental Fig. 3). We used and examined these mice in a well-defined IMQ-induced psoriasis-like model (2, 8). Consistently, psoriasiform skin inflammation was induced by the topical application of Beselna cream containing IMQ, a TLR7 ligand, to both ears of mice for 6 d, which resulted in around 2-fold increased ear thickness in WT mice (Fig. 1A). In sharp contrast, Card14^−/− mice showed significantly less thickness of the IMQ-treated skin over 6 d (Fig. 1A). The significant changes in skin inflammation were confirmed by the histochemical analysis, showing that the numbers of proliferating keratinocytes, which are K5- and Ki67-positive cells in the ear skin of Card14^−/− mice at day 6, were significantly less than those in the WT mice, even though both received IMQ cream (Fig. 1B, 1C).

To examine whether Card14^−/− mice are defective in the effector function of IMQ-induced psoriasiform skin inflammation, we measured the numbers and function of IL-17–producing γδ T cells in the skin, as previous studies have shown that γδ T cells are major IL-17 producers in psoriasis (9). Although WT and Card14^−/− mice show similar numbers of αβ⁺ and γδ⁺ T cells in the epidermis and dermis (Supplemental Fig. 2), the IL-17– and IL-22–producing cells collected from the epidermal sheets of WT mice were significantly increased in IMQ-treated mice. On the contrary, IL-17– and IL-22–producing γδ T cells were significantly reduced in Card14^−/− mice in the IMQ-induced psoriasis model (Fig. 1D, 1E). These results suggest that CARD14 is essential for the formation of the psoriasiform skin inflammation induced by IMQ cream.

As CARD14 has recently been reported to interact with BCL10 and MALT1 to form a complex that activates NF-κB (23, 24), we hypothesized that CARD14 is involved in the innate immune activation signaling downstream of TLRs. To clarify the requirement of TLR signaling in IMQ-induced psoriasiform skin, we examined TLR7-deficient (Tlr7^−/−), TLR9-deficient (Tlr9^−/−), and TLR7 and TLR9 double-deficient (Tlr7^−/−Tlr9^−/−) mice in the same experimental setting as described above. Consistent with a previous report (4), Tlr7^−/− mice developed psoriasiform skin inflammation characterized by the increased epidermal thickening (including parakeratosis) similar to that in WT mice (Fig. 2A–D), although the cream contains 5% of the TLR7 agonist, IMQ.

As MyD88-deficient mice display no IMQ-mediated skin inflammation (11), we then examined the other potential TLRs, such as TLR9, in addition to TLR7. Plasmacytoid DCs are known to express both TLR7 and TLR9, which respond to nucleic acids from self especially when the aggregation of self-DNA forms complexes with LL37 (5). Therefore, we used Tlr9^−/− mice to examine the pathway involved in skin inflammation. TLR9-deficient mice displayed an identical phenotype to that of WT mice (Fig. 2E). By contrast, like the Card14^−/− mice, when we examined the Tlr7^−/−Tlr9^−/− mice we found significantly less ear thickness (epidermal hyperplasia) and less dermal cell infiltration than in WT mice on the last 3 d of treatment (Fig. 2F, 2G). We also examined IL-17– and IL-22–producing cells from the ears of WT and Tlr7^−/−Tlr9^−/− mice that had received IMQ cream and found that γδ T cells producing IL-17 and IL-22, which markedly increased in the IMQ-treated WT mice, were reduced in the IMQ-treated Tlr7^−/−Tlr9^−/− mice (Fig. 2H). Although speculated by previous reports (4, 10), our results clearly show, to our knowledge for the first time, that IMQ-induced psoriasiform skin inflammation requires both TLR7 and TLR9 signaling pathways in mice in vivo. Although CARD14 may well be downstream of TLR7, TLR9, or both, the question of how this dual TLR activation occurs still needs to be answered.

To link between TLR7- and TLR9-mediated and CARD14-mediated immune signaling in vivo, we next focused on IL-23, one of the important key factors for pathogenesis of psoriasis. We examined whether IL-23p19 production or characteristic gene signatures induced by IL-23 are affected in Card14^−/− and Tlr7^−/−Tlr9^−/− mice by IMQ treatment. We collected total RNA from the whole-ear skin of WT mice at various time points after IMQ treatment (0, 2, 4, 8, 24, and 36 h), and determined the Il23p19 mRNA levels with quantitative real-time reverse transcription PCR. Expression of Il23p19 was upregulated at 2 h and peaked at 4 h, then waned by 36 h after IMQ treatment (Fig. 3A). We then measured the expression of Il23p19 mRNA at 4 h after IMQ treatment in Card14^−/−, Tlr7^−/−, Tlr9^−/−, and Tlr7^−/−Tlr9^−/− mice, and found that relative Il23p19 mRNA expressions were significantly downregulated in Card14^−/− and Tlr7^−/−Tlr9^−/− mice (Fig. 3B), whereas CARD14 deficiency does not affect the secretion of IL-12p40 and IL-23p19 by DCs stimulated by a TLR7 or TLR9 ligand in vitro (Supplemental Fig. 4).

Because Il23p19 mRNA expressions at 4 h after IMQ-cream treatment coincided with the phenotype of Card14^−/− and Tlr7^−/−Tlr9^−/− mice (Figs. 1A, 2F), we further analyzed the gene-expression profile of the treated skin at 4 h after IMQ-cream treatment by microarray, and found that the gene expression profiles of the Card14^−/− and Tlr7^−/−Tlr9^−/− mice were similar, but distinct from other control groups such as WT, Tlr7^−/−, or Tlr9^−/− mice (Fig. 3C). When the microarray gene expression patterns were further sorted with a previously described set of genes targeted by IL-23 (34), the profiles in both the Card14^−/− and Tlr7^−/−Tlr9^−/− mice revealed that most of the sorted genes show a similar trend. However, there are 17 unique psoriasis-related genes, which were differentially expressed in responses to IL-23 in the Card14^−/− or Tlr7^−/−Tlr9^−/− mice (Fig. 3D, 3E). Among groups that are treated with IMQ-cream, Aqp4, Bcl10, Ccr7, Cxcl1, and Pi3kca, all of which are relevant to inflammation, cell activation and maturation of DCs and T cells, were remarkably downregulated in Card14^−/− mice compared with those in WT mice, but not in Tlr7^−/−Tlr9^−/− mice (Fig. 4A–E). This indicates that the distinct intra- or intercellular signaling pathway(s) and subsequent gene regulations between TLR7, TLR9, and CARD14 exist.

To better understand the cell-specific relation between TLR7, TLR9, and CARD14 in terms of Il23p19 mRNA expression during IMQ-induced psoriasis in vivo, we sorted the potential IL-23p19–producing cells that include keratinocytes, γδ T cells, Langerin^high LCs, and Langerin^low DCs from the ear skin of mice after IMQ-cream treatment for 4 h. Whereas Card14 mRNA expression has been shown in the skin (25), keratinocytes and Langerin^high LCs showed higher expression of Card14 mRNA than its residual level of γδ T cells and almost no level in Langerin^low DCs (Fig. 4F). Card14 mRNA expressions were not up- or downregulated after IMQ-cream treatment in most cell types, but Langerin^high LCs were downregulated (Fig. 4F). Consistent with the previous finding that Langerin⁺ LCs are potential IL-23 producers in psoriatic mice (12, 13), we observed a significant increase in Il23p19 mRNA expression in Langerin^high LCs in the control treatment group and significantly upregulated by IMQ-cream treatment (Fig. 4G). Furthermore, when the highly upregulated Il23p19 mRNA expressions by Langerin^high LCs were compared between WT, Card14^−/−, and Tlr7^−/−Tlr9^−/− mice, the levels of Il23p19 mRNA expression were significantly reduced in Tlr7^−/−Tlr9^−/− mice, but not in Card14^−/− mice (Fig. 4H). Collectively, these data indicate that IL-23p19–producing Langerin⁺ LCs are characteristic of psoriasis, but their levels differ in the early responses between Card14^−/− and Tlr7^−/−Tlr9^−/− mice, although their whole-skin Il23p19 mRNA expression is similar.

Because it has been known that IL-23 acts directly on T cells, we next examined whether Tlr7^−/−Tlr9^−/− and Card14^−/− mice are sensitive to another established model of psoriasiform dermatitis, induced with rIL-23 (8). Consistent with a previous report (8), rIL-23 injection on days 0, 2, and 4 caused a significant increase in ear thickness, with epidermal acanthosis and dermal inflammatory cell infiltration on day 5. These changes were independent of both TLR7 and TLR9 (Fig. 5A), and the number of γδ T cells producing IL-17 and IL-22 was not different between WT and Tlr7^−/−Tlr9^−/− mice (Fig. 5B). These results demonstrate that both TLR7 and TLR9 are required for the full development of IMQ-induced psoriasiform dermatitis, whereas in the IL-23–induced model does not require a TLR7- and TLR9-mediated innate immune signaling pathway. In sharp contrast, IL-23–induced ear swelling was significantly suppressed in the Card14^−/− mice at day 2, 4, and 5 d after the first rIL-23 injection (Fig. 5C). Consistently, immunohistochemical analysis revealed that K5- and Ki67-positive hyperproliferative cells, and both epidermal hyperplasia and dermal skin inflammation (indicated by infiltrating cells stained with H&E) did not occur in the rIL-23–injected Card14^−/− mice (Fig. 5D, 5E). The data above suggest that CARD14 is essential for IL-23–induced psoriasiform skin inflammation, potentially acting downstream under IL-23R on certain effector cells such as γδ T cells secreting IL-17 and IL-22 to develop psoriasis.

We then went on identifying the subsets of effector γδ T cells that produce IL-17 and IL-22 in the epidermis and dermis of these mice at 6 d after IMQ-cream treatment. Consistent with a previous report, which found that Vγ4⁺ T cells are the main source of IL-17 rather than DETCs (also known as Vγ5⁺ T cells) in psoriatic skin (36), the frequencies of both epidermal and dermal Vγ4⁺ T cells were significantly increased in the IMQ-cream–treated WT mice than untreated controls (Fig. 6A, 6B, left panel). Importantly, those increased Vγ4⁺ T cells secrete IL-17 and IL-22, both of which were significantly reduced in the Card14^−/− mice treated with IMQ cream (Fig. 6A, 6B, right panel). In contrast, there was no significant difference in epidermal Vγ5⁺ T cells and their secretion of IL-17 and IL-22 between WT and Card14^−/− mice treated with IMQ cream (Fig. 6C). Therefore, CARD14 contributes to the increased numbers of IL-17⁺ and/or IL-22⁺ Vγ4⁺ cells that migrate into the skin being distributed in the both the dermis and epidermis during the IMQ-induced inflammatory response.

It has been reported that epidermal CCR6⁺ γδ T cells are responsible for psoriasis in mouse skin (37), but IL-17–producing dermal γδ T cells have also been implicated (9, 36). This discrepancy has been discussed and it has been suggested that dermal γδ T cells migrate into the epidermis during inflammation (38). We have confirmed this proposition and found that the number of Vγ4⁺ T cells producing IL-17 and IL-22 in the IMQ-cream–treated skin were increased predominantly in the epidermis, which, importantly, is prevented in Card14^−/− mice (Fig. 6A, 6B). Although our results suggest that CARD14 is necessary for the IMQ-mediated production of IL-17 and IL-22 by γδ T cells (Figs. 1E, 6A, 6B), further studies are required to clarify how CARD14 is involved in these intracellular signaling pathways of such cell types.

As it has been shown that CARD14 is involved in the innate immune activation of keratinocytes (21, 25), we examined whether Card14^−/− mice are protected from IL-22–induced psoriasis, in which IL-22 activates and proliferates keratinocytes to develop psoriasiform skin inflammation (8). WT mice displayed ear swelling by rIL-22 injection, however, the swelling in Card14^−/− mice was significantly reduced at day 4 and 5 after rIL-22 injection (Fig. 7A), suggesting that CARD14 may directly be associated with keratinocytes acting as effector cell types to develop psoriasis.

To understand the other cell types involved in the effector phase of psoriasis, we examined IMQ-induced psoriasis in TCR δ–deficient (Tcrd^−/−) mice, which lack γδ T cells, and in Rag2^−/−Il2rg^−/− mice, which lack T cells, NK cells, NKT cells, and innate lymphoid cells. Consistent with a previous report (10), Tcrd^−/− and Rag2^−/−Il2rg^−/− mice showed significantly less skin swelling than WT mice after the application of IMQ cream (Fig. 7B). These data suggest that γδ T cells are essential in IMQ-induced psoriasiform skin inflammation.

Overall, our data indicate a potential role for CARD14 in induction and effector function of IL-17⁺Vγ4⁺ cells during psoriasiform skin inflammation. To further examine whether CARD14 is critical in keratinocytes or other hematopoietic cells, we generated reciprocal BM chimeras. BM cells from Ly5.1-B6 or Card14^−/− mice were transplanted into x-ray–irradiated Card14^−/−, Ly5.1-B6, or WT mice. After 10 wk, both ears of the transplanted mice were treated with IMQ cream for six consecutive days. Chimerism was confirmed as the CD45.1⁺/CD45.2⁺ population ratio in the peripheral blood. The donor-derived cell population exceeded 90% in all groups 10 wk after reconstitution (Fig. 7C). Chimerism of the skin, in contrast, was unique where the remaining radio-resistant CD45.2⁺ cells were over 80% in the epidermis and around 40% in the dermis (Fig. 7D). After IMQ-cream treatment, numbers of CD45.1⁺ cells from the donor BM in the epidermis were significantly increased compared to those in the control, but not in the dermis or lymph nodes (LN) (Fig. 7D). Although the total number of CD45.1⁺ cells in the dermis was increased, the ratio between CD45.1⁺ and CD45.2⁺ was not altered (Fig. 7D). These data strongly suggest that radio-sensitive CD45.1⁺ cells moving into the epidermal area are originally from the BM, rather than radio-resistant skin-resident cells, and play a major role in psoriasiform skin inflammation.

Psoriasiform skin inflammation was induced in WT mice transplanted with BM from Ly5.1-B6 mice, and partially induced in Card14^−/− mice transplanted with BM from Ly5.1-B6 mice. On the contrary, irradiated Ly5.1-B6 mice transplanted with Card14^−/− BM showed significantly suppressed swelling (Fig. 7E), suggesting that CARD14 expressed in radio-sensitive, hematopoietic BM-derived cells is essential for the pathogenesis of psoriasis in our IMQ-induced model. Although the possibility cannot be excluded that in radio-resistant, nonhematopoietic cells, including keratinocytes, CARD14 plays a role in psoriasiform skin inflammation, this experiment suggests that CARD14 in BM-derived cells, potentially γδ T cells, is necessary for the development of IMQ-induced psoriasiform skin inflammation.

Finally, we examined the role of CARD14 from donor BM in the induction of IL-17⁺ γδ T cell subsets in the skin. As it has previously been shown that IMQ cream–activated Vγ4⁺ T cells leave the LN and accumulate in inflamed skin (39), we demonstrated the presence of an IL-17⁺ γδ T cell subset in LNs. Donor IL-17⁺Vγ4⁺ T cells derived from Ly5.1-B6 mice accumulated strongly in the IMQ-cream–treated LN at higher frequencies than in IL-17⁺Vγ1⁺ and IL-17⁺Vγ5⁺ T cells (Fig. 7F). In contrast, in the control mice, the frequency of IL-17⁺Vγ4⁺ T cells derived from Card14^−/− mice was significantly reduced (Fig. 7F). Cells that express IL-23R must play an important role in the CARD14-mediated pathogenesis of psoriasiform skin inflammation downstream from the signaling pathway initiated by IL-23. It is unclear which cells express IL-23R, but most cells that express IL-23R are γδ T cells, together with some CD11c⁺ cells (40). We then cultured IL-17–producing cells with IL-1β and/or IL-23 (Fig. 7G). The IL-17⁺ γδ/T cell ratio in the Card14^−/− mice was more strongly reduced than the ratio in the control when the cells were cultured with IL-23 rather than with IL-1β plus IL-23.

The above results demonstrate that CARD14 is required for the optimal production of IL-17 and IL-22 by Vγ4⁺ T cells, when mice are treated by either IMQ cream or rIL-23. To our knowledge, this is the first report that CARD14 is solely responsible for the pathogenesis of psoriasis-like dermatitis in a murine experimental model, whereas human mutations in Card14 have been reported to regulate NF-κB signaling and are strongly correlated in the skin in psoriasis and pityriasis rubra pilaris (21, 25). It has also been reported that CARD14 is expressed in the epithelial cells of skin lesions in psoriasis patients (41). We therefore hypothesized initially that CARD14 would be a downstream signaling molecule for TLR7-mediated NF-κB activation in the epithelial stromal cells, such as keratinocytes. However, our initial hypothesis is incorrect, and instead the results imply that CARD14 acts downstream of the IL-23–mediated immune signaling pathway and is important for the initiation and amplification of inflammation, as CARD14 was essential for both IMQ- and IL-23–induced psoriasiform skin inflammation (Figs. 1A, 5C), whereas TLR7, together with TLR9, was required for IMQ-induced, but not for IL-23–induced, psoriasiform skin inflammation (Figs. 2F, 5A). Although CARD14 is expressed in epithelial keratinocytes as well as Langerin^high LCs, it was also expressed in γδ T cells in the psoriasis-like skin rash induced in our models (Fig. 4F). Although our data suggest that CARD14 is critical for cells that express IL-23R to respond to IL-23 such as γδ T cells (Figs. 1E, 6) to produce IL-17 and IL-22 in vivo after IMQ or rIL-23 treatment, as well as in vitro culture with rIL-23 (Fig. 7G), further studies are required to clarify how CARD14 mediates these intracellular signaling pathways at a molecular level.

Keratinocytes that express CARD14 are also involved in skin inflammation, but require IL-22 because IL-22R is predominantly expressed on keratinocytes but not leukocytes. Consistent with a previous report (9), our data suggest that Card14 expressed in keratinocytes is involved in IMQ-induced as well as rIL-23–induced psoriasiform skin inflammation via IL-22, which is predominantly secreted by CARD14⁺γδ T cells. Based on the above-mentioned experiments, we clearly demonstrated the relative contribution of CARD14 in γδ T cells and keratinocytes to skin inflammation. Even though the experimental conditions differ between human patients and experimental mouse models, numbers of γδ T cells are known to differ in humans and mice (9, 19). Therefore, it is possible that the roles of CARD14 in humans and mice diverge according to the CARD14-expressing cell types or the experimental conditions, including differences in animal resources. Further studies are required to clarify the role of CARD14 in human γδ T cells during psoriasis.

IMQ cream is used as a topical treatment of nonmelanoma skin cancer, and is known to cause side effects such as psoriasis-like skin inflammation (4). It is conceivable that TLR7 activation by IMQ cream and certain stimuli within it leads to cell death and release of DNA by resident keratinocytes, or recruitment of more specialized immune cells, such as neutrophils to undergo neutrophil extracellular traps associated with cell death to release DNA as a TLR9 agonist (42). It is also possible that the skin microbiome–related nucleic acids play a role in this unique TLR9 involvement during IMQ-cream–mediated psoriasiform skin inflammation (43), although further studies seem required.

Skin APCs are a highly heterogeneous population with functionally specialized subsets; single functional markers that uniquely define individual subsets of cells are absent (44). Consistent with a previous study focused on IL-23–producing cells in resident LCs in psoriasis (13), we selectively detected upregulation of Il23p19 in LCs. Although we demonstrated sorting against four major populations in the skin (keratinocytes, γδ T cells, LCs, and DCs), there may have been a lack of other potential IL-23p19–producing APC subsets, especially CD11b⁻ populations including dermal cDC1 (CD11b⁻XCR1⁺) and double-negative DC (CD11b⁻XCR1⁻). Overall, Il23p19 expression was higher in LCs than DCs in our data, whereas the discrepancy in the reported IL-23p19–producing cells may result from the use of different markers and determination of accurate target cells. It should also be noted that we used different methods for RNA purification for skin tissue and cells that were sorted, and demonstrated that sorting using a high electric charge might cause cell damage. It is conceivable that original immunological responses were lost and there may be other endogenous factors involved that are not currently known. These reasons could explain the data discrepancy that requires further clarification in the future.

CARD14 may be a good target for reducing the pathogenesis of psoriasiform dermatitis. The immunological significance of the CARD14-mediated innate and adaptive immune responses should be investigated, because γδ T cells play an important role in various immune disorders, including multiple sclerosis, and infectious diseases (17, 18).

We thank all members of the Laboratory of Adjuvant Innovation, National Institutes of Biomedical Innovation, Health and Nutrition for helpful discussions. We particularly thank Drs. Cevayir Coban, Jun Kunisawa, Patrick M. Lelliott, Reiko Fukui, Takahiro Nagatake, and Teruki Dainichi for support and insightful comments.

The authors have no financial conflicts of interest.