Skin expression of IL-23 drives the development of psoriasis and psoriatic arthritis in mice

Psoriasis (PS) is a chronic skin inflammation. Up to 30% of the patients with PS develop psoriatic arthritis (PsA), a condition characterized by inflammatory arthritis that affects joints or entheses. Although there is mounting evidence for a critical role of interleukin-23 (IL-23) signaling in the pathogenesis of both PS and PsA, it remains unclear whether IL-23-induced skin inflammation drives joint disease. Here, we show that mice expressing increased levels of IL-23 in the skin (K23 mice) develop a PS-like disease that is characterized by acanthosis, parakeratosis, hyperkeratosis, and inflammatory infiltrates in the dermis. Skin disease preceded development of PsA, including enthesitis, dactylitis, and bone destruction. The development of enthesitis and dactylitis was not due to high circulating levels of IL-23, as transgenic animals and controls had similar levels of this cytokine in circulation. IL-22, a downstream cytokine of IL-23, was highly increased in the serum of K23 mice. Although IL-22 deficiency did not affect skin disease development, IL-22 deficiency aggravated the PsA-like disease in K23 mice. Our results demonstrate a central role for skin expressed IL-23 in the initiation of PS and on pathogenic processes leading to PsA.

Psoriasis (PS), one of the most prevalent autoimmune diseases, is commonly associated with other conditions, including inflammatory bowel disease and arthritis 1,2 . Up to 30% of the patients with psoriasis have psoriatic arthritis (PsA), a potentially debilitating condition that causes joint damage and pain and have a significant impact in the quality of life 3 . PsA similarly affects both genders and in most populations typically occurs between ages 30 and 50 4 . PsA frequently affects joints of the hands, feet and knees causing joint swelling and pain often with diffuse involvement of the digits (dactylitis), and tendons/ligaments bone insertion sites (enthesis/enthesitis) 5 . A subset of patients develop axial disease with spondylitis and sacroiliitis 5 . Histologically, PsA is characterized by synovial hyperplasia and inflammation with cartilage and bone erosive changes.
The pathogenesis of PsA remains incompletely understood. In 70% of cases PS precedes the onset of PsA by several years 6 . There is evidence for a partial overlap between PS and PsA susceptibility genes, including HLA-Cw6 and several others identified in such as IL-13, NFKBIA and IL-23R 7-10 . Genome-wide association studies (GWAS) have also revealed genes uniquely associated with PsA such as IFNLR1 [7][8][9][10] . The IL-23R gene has been one of the non-MHC genes most strongly and consistently associated with both PS and PsA 11 , and the IL-23/ IL-17 pathway has been implicated in the development of aspects of the disease such as spondyloarthropathy 12 . IL-23 has been detected in the skin and joint synovial tissues of PsA patients 13 and neutralization of IL-23 is effective in reducing PsA disease activity and severity 14 . IL-23 is a dimeric cytokine formed by two subunits, p19 and p40. IL-23 is produced by keratinocytes 15,16 and activated antigen-presenting cells (APCs), including Langerhans cells, macrophages, and dendritic cells (DCs) that induce the differentiation of pathogenic Th17 cells 17 . Mice treated with IL-23 minicircle DNA in vivo developed arthritis 12,18 . Subsequently it was shown that high systemic levels of IL-23 could also lead to development of enthesitis and skin disease 12 . IL-23R + T cells are found in the skin and in the entheses and have been suggested to react to systemic levels of IL-23 and play a pathogenic role in PsA 12 .
Identical T cell clones have been detected both in the skin and synovial tissues of PsA patients 19 suggesting that a shared antigen might be driving immune responses in both sites. However, it remains unclear whether www.nature.com/scientificreports www.nature.com/scientificreports/ skin PS disease processes are required and drive the joint disease. In this report we show that mice expressing increased levels of IL-23 in the skin develop PS and PsA, including dactylitis, enthesitis and joint destruction. Of note, the skin abnormalities preceded development of enthesitis, dactylitis, and bone destruction. Surprisingly, the development of enthesitis and dactylitis was not predicated on circulating levels of IL-23, as transgenic animals and controls had similar levels of this cytokine in circulation. Together these results demonstrate an important role for IL-23 expressed in the skin in the initiation of the pathogenic processes leading to PsA.

Results
Generation of mice expressing IL-23 conditionally in keratinocytes. To generate mice conditionally expressing IL-23 in the skin we first generated mice containing both p40 and p19 subunits of IL-23 (p40-2A-p19) in the ROSA26 locus downstream of a floxed STOP cassette (R23 mice) 20 . R23 mice were intercrossed with mice carrying a tamoxifen inducible Cre recombinase driven by the keratin-specific K14 promoter 21 (K14 CreERT2 ) to generate inducible K23 mice (Fig. 1A). To promote cre-mediated excision of the STOP cassette and expression of IL-23 in skin cells, we treated K23 mice with 3 cycles of tamoxifen in the food (TD.130968) 20 (days 1-7, 14-21, and [28][29][30][31][32][33][34][35]. To examine expression of the IL-23 transgene in the skin we extracted RNA from the ear of WT and K23 mice after 8 weeks of TAM treatment and performed qPCR. As expected, we found that IL-23p40 and IL-23p19 expression was increased in the ears of mice that received TAM in K23 mice compared to WT mice (Fig. 1B).
Conditional expression of IL-23 in keratinocytes induces a psoriasis-like disease. The K14 promoter used to drive expression of the CreER gene is strongly active in dividing cells of epidermis, and targets the stem cell compartment 22 . Non-treated K23 mice did not develop any phenotypes during 52 weeks of observation. However, PS-like skin phenotypes in the tails and ears were observed as early as 5 weeks after initiation of TAM treatment in K23 mice, but not WT controls (Fig. 1C,D). These PS phenotypes included scaling and discoloration of the skin, ear swelling and patchy hair loss (Fig. 1C). Histological analyses of the skin of K23 mice treated with TAM after 6 weeks, compared to WT mice, showed a psoriatic phenotype of acanthosis, parakeratosis, hyperkeratosis, and inflammatory infiltrates in the dermis (Fig. 1E), similar to human PS lesions 1 .
Hallmarks of psoriasis are observed in the epidermis in K23 mice. Most clinical pathology in psoriasis is related to hyperproliferative and disturbed differentiation of epidermal cells. Similarly to what is observed in the skin of PS patients, the basal cell layer of the ear epidermis of K23 mice was hyperproliferative, as indicated by Ki67 staining (Fig. 2A) and cells in the dermis and epidermis showed increased phosphorylation (activation) of STAT3 (pSTAT3) (Fig. 2B). Keratin 6 (CK6) is normally expressed by hyperproliferative epithelial cells 23 whereas keratin 10 (CK10) is expressed by differentiated epithelial cells 24 . Similarly to what is observed in the human psoriatic lesions [25][26][27] , the ear skin of K23 mice had cells co-expressing CK6 and CK10 (Fig. 2C). These results indicate that psoriatic skin is marked by hyperproliferation and aberrant differentiation of keratinocytes in K23 mice after tamoxifen treatment.
To determine if K23 developed a chronic inflammatory disease with similar characteristic to human psoriasis, we analyzed the expression of psoriasis markers in the ear skin of WT and K23 mice 20 weeks after TAM treatment ( Fig. 2D-G). The calcium-binding proteins S100A8 and S100A9 are highly expressed in PS skin [28][29][30] . Increased expression of S100A9 was also observed in the epithelial layer of the ear of K23 mice when compared to WT littermate (Fig. 2D). Podoplanin, a well-known lymphatic endothelial marker, is also expressed in keratinocytes 31 . Similar to human PS lesions, the ears of K23 mice with PS-like lesion had increased expression of podoplanin in the epidermis (Fig. 2E). As observed in human psoriatic lesions, vimentin 32 expression was also increased in the epidermis of K23 mice treated with TAM (Fig. 2F). Finally, psoriatic K23 mice had increased CD31 expression in microvessels of papillary dermis suggesting increased angiogenesis as seen in PS skin (Fig. 2G). Together these results indicate that IL-23 expression in keratinocytes induces marked molecular expression changes in the skin that resemble those observed in human PS during the chronic phase.

Characterization of immune infiltrates in the skin of K23 animals.
To characterize the skin inflammatory infiltrates, we performed flow cytometric analysis using cells dissociated from the ears of K23 and WT littermate mice 20 weeks after TAM treatment (Fig. 3). The flow cytometry gating strategy is shown in Fig. 3A,B. The ears of K23 mice had 10 times more leukocytes than those of WT littermate controls (Fig. 3C). CD3 + TCRαβ + T cells (CD4 + T cells and CD8 + T cells) represented the major cell population that infiltrated the ears of K23 mice (Fig. 3D). A significant decrease in the number TCRγδ T cells was observed in the skin of K23 mice at this time-point (Fig. 3D). Neutrophils and macrophages were the major myeloid cell populations enriched in the skin of K23 mice (Fig. 3E). A significant increase in the number of B cells (Fig. 3D), monocytes, DCs, NK, and eosinophils ( Fig. 3E) was also observed in the psoriatic ears of K23 mice. Together the results indicate that innate and adaptive immune cells accumulate in the ears of K23 mice, similar to what is observed in human psoriatic lesions 33 .
K23 mice develop a PsA-like disease after PS. Up to 30% of the patients with psoriasis have psoriatic arthritis (PsA). As observed in patients with PsA, swelling of the digits (dactylitis) and arthritis with joint and paw swelling was also observed in 13% of K23 mice after 10 weeks of TAM treatment reaching 100% by 25 weeks (Fig. 4A,B). These results indicate that K23 mice develop arthritis after psoriasis (Fig. 4B). More importantly, the joints of K23 mice treated with TAM had increased synovial hyperplasia (Fig. 4C), dactylitis (Fig. 4D), and enthesitis (Fig. 4E), when compared to WT mice. The pronounced inflammation of the joints of K23 mice was associated with bone erosion and bone destruction including the typical distal phalanx acrosteolysis seen in PsA patients (Fig. 4F). Whereas psoriasis was observed in the skin as early as 6 weeks after TAM, synovial hyperplasia and dactylitis was observed 10 weeks after TAM treatment. Enthesitis and joint destruction were observed only after 19 weeks of TAM (Fig. 4G). These results indicate that psoriasis precedes the development of synovial inflammation and bone destruction in K23 mice, as seen in most patients with PsA.

Significant changes in circulating levels of cytokines during disease evolution. It has been
reported that high systemic levels of IL-23 induce psoriasis and spondyloarthritis in mice 12 . To examine if IL-23 could be detected in the serum at distinct phases of disease, we collected serum of K23 mice at 5 and 20 weeks and used LUMINEX technology to quantify it. To our surprise, the circulating levels of IL-23 were not different  (Fig. 5). In addition, we checked the expression of several cytokines (IL-22, IL-17, IFNg, TNF, GMCSF and IL-33) in the skin of K23 mice 8 weeks after TAM treatment by qPCR analysis (Fig. 6). We found that the expression of the cytokines IL-22, IFNg and TNF were significantly increased in the skin of K23 www.nature.com/scientificreports www.nature.com/scientificreports/ mice compared to WT mice (Fig. 6). These results indicate that expression of IL-23 in the skin of K23 mice does not significantly increase the levels of IL-23 in circulation, and suggest that the inflammatory process in the skin increases the circulating and local levels of several cytokines, most markedly IL-22.   www.nature.com/scientificreports www.nature.com/scientificreports/ acanthosis, parakeratosis, and hyperkeratosis (Fig. 7D). These results suggest that IL-22 is not required for development of skin inflammation.

IL-22 deficiency aggravates the PsA-like disease in K23 mice. To study the role of IL-22 in the
PsA-like disease development, we also analyzed development of PsA-like disease in K23/IL-22 −/− and K23/ IL-22 +/+ mice. There was no significantly difference in the incidence of development PsA-like disease in K23/ IL-22 −/− and K23/IL-22 +/+ mice (Fig. 8A). Histological analyses of the digits of K23/IL-22 −/− mice and K23/ IL-22 +/+ mice showed inflammation and bone erosion were present in both groups of mice 25 weeks after TAM treatment (Fig. 8B). However, histological scores focusing on PsA features such as synovial proliferation, enthesitis, bone erosion and joint destruction showed that disease was more severe and destructive in K23/IL-22 −/− when compared to K23/IL-22 +/+ animals (Fig. 8C). Together, these results indicate that ablation of IL-22 was not protective, and appeared to cause a more severe PsA-like disease.

Discussion
In this report we describe a novel mouse model of PS and PsA induced by selective and conditional expression of IL-23 in the skin (K23 mice). Upon expression of IL-23, the animals developed a skin inflammatory disease that presented hallmarks of psoriasis such as acanthosis, parakeratosis, cellular infiltrates in the dermis and epidermis, hyperkeratosis and increased angiogenesis. Several weeks after the development of the skin PS phenotypes K23 mice developed arthritis, enthesitis, and dactylitis. Similar findings occur in humans. Most patients with PsA develop joint disease after developing skin disease 6 .
In addition to the typical clinical and histologic findings of PS and PsA, the skin of K23 mice expressed increased levels of established markers of PS skin disease, including increased levels of pSTAT3 38 , S100A8, S100A9 [28][29][30] and podoplanin 31 . Furthermore, K23 mice developed the typical PsA dactylitis, arthritis, enthesis and acroosteolysis, which is a destruction of the distal phalanx almost only seen in PsA. K23 mice also had increased expression of other key mediators of inflammation and joint destruction such as IL-1β, IL-6, IFNγ and CXCL10.
An association of IL-23 with skin inflammatory disease was first suggested by analysis of mice expressing IL-23-p19 from the CMV-Bactin promoter 39 . These animals presented a systemic inflammatory disease and died prematurely. The skin had a marked inflammatory infiltrate characterized by the presence of lymphocytes, macrophages, and neutrophils in the dermis and epidermis, with focal epithelial necrosis and ulceration. Subsequent experiments demonstrated that IL-23 injection in situ, or systemically via gene therapy, could also induce similar inflammatory phenotypes 12,40,41 . Together with results shown here, these results establish a role for IL-23 in the development of psoriasis, and suggest that IL-23 acts locally in the skin to do so.
The IL-23/IL-17 axis is currently considered to be crucial in the pathogenesis of psoriasis 42 . Human IL-23 is primarily produced by antigen-presenting cells and induces and maintains differentiation of Th17 cells and Th22 cells, a primary cellular source of proinflammatory cytokines such as IL-17 and IL-22, which mediate the epidermal hyperplasia, keratinocyte immune activation and tissue inflammation inherent in psoriasis 43 .
To further study the role of IL-23 in skin inflammation we engineered transgenic mice carrying the subunits of IL-23 (p19 and p40) preceded by a loxP-flanked STOP cassette in the ROSA26 locus (R23 mice). To promote expression of p19 and p40, we crossed these animals to mice expressing Cre recombinase in keratinocytes. All double positive mice died within the first few days of life 44 . To bypass this early lethality, we intercrossed the R23 mice to mice expressing an inducible form of cre recombinase in keratinocytes, and treated the double positive mice, when adult, with TAM. All the mice survived treatment but developed skin inflammation and, unexpectedly, developed localized joint disease. All the hallmarks of the human condition are observed in the mouse model (skin involvement, enthesitis and bone changes) directly implicating IL-23 in the pathogenesis of PsA. The only notable difference between the mouse model and the human disease concerns the development of enthesitis. In human PsA enthesitis occurs prior to development of synovitis and/or bone changes, whereas in K23 mice enthesitis is observed after synovial involvement. While the reasons for this differences are not clear at the moment our results clearly document that the skin disease precedes and drives the joint disease in mice, and as such, they have implications to the understanding of the human disease.
Development of spondyloarthropathy has been observed in animals that express high levels of IL-23 after hydrodynamic injection of IL-23 12 . IL-23R + CD3 + CD4 − CD8 − Sca1 + cells were present in the enthesis and systemically elevated levels of IL-23 were associated with enthesitis and entheseal new bone growth, in the absence of synovitis 12 . Subsequent work confirmed the presence of IL-23R + cells in the entheses 45 . These observations suggested that systemically elevated levels of IL-23 could promote enthesitis by acting on a IL-23r positive cell population present in the enthesis. In contrast with these observations, we show here that elevated systemic levels of IL-23 are not required to drive development of skin or joint disease in K23 mice. IL-23 is readily detected in the skin of K23 mice, but blood levels of IL-23 were not different from those of control mice. IL-23 levels have been reported to be either increased 46   www.nature.com/scientificreports www.nature.com/scientificreports/ IL-22 is induced by IL-23 and has been shown to have both pro-and anti-inflammatory properties. Injection of minicircles encoding IL-22 in mice can induce paw swelling suggesting a possible role of IL-22 in joint inflammation 12 . Additionally, IL-22 levels are significantly elevated in the synovial fluid of PsA patients compared to that of patients with osteoarthritis 35 . In animal models, IL-22 is required for the development of the Th17-mediated skin inflammation 48 , and it has been shown that transgenic mice expressing IL-22 develop a psoriasis-like skin phenotype 49 . Several studies have suggested that IL-22 can act in synergy with IL-17 or interferon γ (IFN γ) through an IL-23-dependent mechanism 40 to amplify the inflammation observed in psoriatic skin 50 . The circulating levels of IL-22 were high in the K23 mice treated with TAM, and thus it was reasonable to hypothesize that IL-22 could have a role in the K23 mice joint inflammation. Our results, however, do not support the hypothesis that elevated levels of IL-22 are directly causative of joint inflammation as its deletion did not prevent paw swelling and bone destruction in K23 mice. In fact, IL-22 deletion increased the severity of the joint disease and damage, suggesting that in this setting IL-22 may have a protective function. This is of great interest given that IL-22 has been reported to have anti-inflammatory properties in organ-specific rodent models of inflammation, such as pancreatitis and pneumonitis 36 . The role of IL-17 in this specific model remains to be examined, but it has been described that injection of minicircles encoding IL-17 into mice also does not promote joint disease 12 .
Two other models have been previously reported to induce a PS-like disease based on genetic manipulation, including skin overexpression of activated STAT3 51 and skin double knockdown of JunB/c-Jun 52 . While both are helpful models to study disease, none clearly addressed the role of IL-23 in disease development. More recently, it was shown that the chronic skin inflammation observed in mice deficient in the neural Wiskott-Aldrich syndrome protein (N-WASP) is dependent on IL-23 producing keratinocytes 16 .
On the basis of the relationship between PS and the occurrence of PsA, it has been hypothesized that the skin immune mechanisms might contribute and ultimately trigger the onset of the articular manifestations 42 . Our present results support this hypothesis. We show that expression of IL-23 in the skin leads to development of skin disease that precedes development of arthritis. Whether this disease is driven by specific autoantigens shared between joint and skin as proposed 42 , or reflects increased generation of autoreactive T cells remains to be elucidated. Mice. R23 mice 20 and IL-22 −/− mice 44 were described before. K14-creERT2 mice (CN 005107) 22 were purchased from The Jackson Laboratory (Bar Harbor, ME). IL-22 tdTomato mice were provided by Dr. Scott Duram (NIH, Bethesda, MD). R23 mice were crossed with K14-creERT2 mice to generate K23 animals. Mice were maintained under specific pathogen-free conditions. Reverse-transcription polymerase chain reaction. Total RNA from tissues was extracted using the RNeasy mini Kit (Qiagen) according to the manufacturer's instructions. Quantitative reverse-transcription polymerase chain reaction (RT-qPCR) experiments were performed as previously described 20,44,53 . Ear thickness. Ear thickness was measured at indicated time using digital calipers. An increase in ear thickness was used to indicate the extent of epidermal proliferation and inflammation.

Flow cytometry.
Ears were collected and processed as described 54 . All cells were first pre-incubated with anti-mouse CD16/CD32 for blockade of Fc γ receptors, then were washed and incubated for 40 min with the appropriate monoclonal antibody conjugates in a total volume of 200 μl PBS containing 2 mM EDTA and 2% (vol/vol) bovine serum. DAPI (Invitrogen) was used to distinguish live cells from dead cells during cell analysis. Stained cells were analyzed on a FACS Canto or LSRII machine using the Diva software (BD Bioscience). Data were analyzed with FlowJo software (TreeStar). The following fluorochrome-conjugated anti-mouse antibodies were used at