[1] 142w Immunohistochemical analyses showed that in human psoriatic lesions, the hyperplastic epidermis showed an increased level of Stat3, particularly in the nuclei of keratinocytes (Fig. 1a). Activation of Stat3 was confirmed by the presence of nuclear staining for phosphorylated Tyr705 Stat3 (PY-Stat3; Fig. 1b). Thus, nuclear staining of Stat3 is considered equivalent to positive staining for nuclear PY-Stat3. More than 90% (n = 19 of 21) of the primary psoriatic lesions examined showed this pattern of Stat3 immunostaining. Stat3 was also activated in adjacent, uninvolved epidermis in some psoriatic patients (see Supplementary Fig. 1 online). The epidermis of normal controls showed weaker, diffuse cytoplasmic staining for Stat3 (Fig. 1c). Lesions from three nonpsoriatic inflammatory skin disorders with characteristic epidermal hyperplasia (chronic dermatitis, prurigo and lichen planus; Fig. 1) showed a Stat3 staining pattern similar to normal epidermis (n = 10 of 14 cases).

[2] 331w In light of the results shown in Figure 1 transgenic mice were generated in which a constitutively active Stat3 cDNA (Stat3C) 28 was expressed under control of the bovine keratin (K) 5 promoter (hereafter referred to as K5.Stat3C). The K5 promoter targets expression of Stat3C to the basal layer of epidermis as well as keratinocyte stem cells 29 . Our recent work showed that basal keratinocytes express Stat3 and that these cells required Stat3 for both survival and proliferation in response to external stimuli 30,31 . Western blot analysis showed that keratinocytes from K5.Stat3C mice expressed the Flag epitope tagged to Stat3C 28 . These mice had increased levels of total Stat3 and PY-Stat3, whereas Stat1 levels were unaffected (Fig. 2a). K5.Stat3C mice were viable, and their skin appeared normal at birth, except for thicker and more prominent subcutaneous blood vessels (Supplementary Fig. 2 online). No substantial histological alterations were found in skin of newborn transgenic mice (data not shown); however, by 2 weeks of age, their skin was reddened and scaly, hyperkeratotic lesions developed in the tail and spread to the dorsum and hind feet in some mice (Supplementary Fig. 2 online). Histologic examination of affected tails showed a marked hyperplasia of the epidermis (acanthosis) with elongation of rete ridges accompanied by loss of the granular layer and confluent parakeratotic scale (Fig. 2b). Further examination showed a dense dermal infiltrate of inflammatory cells, an increased number of capillaries and small nests of neutrophils within the epidermis (Fig. 2b). Keratinocytes in these lesions were highly positive for Ki67 (Fig. 2c), a marker of hyperproliferation. Keratin 1 expression in the suprabasal layer was decreased, and apparently replaced by keratin 6 (data not shown), suggesting an alteration in keratinocyte differentiation similar to that found in human psoriasis 32 . Lesions of the tail spread contiguously to the dorsum and became exacerbated with age (Supplementary Fig. 3 online). These clinical and histologic features in K5.Stat3C mice show a strong resemblance to human psoriatic plaques.

[3] 216w Previous studies showed that Stat3-deficient mice had delayed wound healing [22][23][24] . In contrast, K5.Stat3C mice developed scaly, erythematous lesions following full-thickness wounding (Supplementary Fig. 2 online). Histologic evaluation of wound healing sites in the skin of transgenic mice showed a marked acanthosis with elongation of rete ridges in contrast to the skin of control littermates, in which the re-epithelized epidermis appeared flattened after wound healing (Fig. 2d). The woundinginduced skin lesions seen in K5.Stat3C mice shared other histological features with human psoriatic lesions, including hyperkeratosis, parakeratosis, absence of granular layer, leukocyte infiltrates and dilated blood vessels in the papillary dermis (Fig. 2d). Also, there was sporadic invasion of neutrophils into the epidermis (data not shown). These psoriatic lesions in K5.Stat3C mice did not subside spontaneously and became more severe with time (Supplementary Fig. 3 online). Immunohistochemical analysis showed that expression of the Stat3C transgene in the psoriatic lesions was intense in the cytoplasm and nuclei of suprabasal keratinocytes (Fig. 2e). The strong staining for Stat3 in suprabasal keratinocytes of both human psoriatic lesions and wounding-induced psoriatic lesions in transgenic mice may owe to altered Stat3 turnover as a result of altered epidermal differentiation. The psoriatic lesions that develop in K5.Stat3C mice show a pattern of Stat3 staining very similar to that of human psoriatic plaques.

[4] 197w We next utilized tape stripping, a milder form of epidermal injury, as perturbation of the skin permeability barrier has been reported to be associated with the earliest clinical manifestation of psoriasis 33 . Notably, K5.Stat3C mice developed well-circumscribed scaly lesions as early as 4 d after tape stripping, whereas control mice developed a mild erythematous lesion (Fig. 2f). The tape stripping-induced lesions in K5.Stat3C mice also recapitulated the histological features of human psoriasis, including hyperkeratosis, parakeratosis, acanthosis with elongation of rete ridges, collection of neutrophils (Munro's abscess) in the horny layer (Fig. 2g,h). In addition, we observed mononuclear cell infiltrates, and increased, dilated blood vessels in the papillary dermis (Fig. 2g). In addition to wounding, topical treatment of transgenic mice with 12-O-tetradecanoylphorbol-13-acetate (TPA) for more than 3 weeks resulted in the development of diffuse, scaly erythematous lesions with histologic features virtually identical to those induced by full-thickness wounding or tape stripping (Supplementary Fig. 4 online). Collectively, the data from both experiments indicate that constitutive activation of Stat3 in transgenic mice increased the sensitivity of the epidermis to external stimuli, leading to a psoriatic phenotype, similar to the classic clinical feature known as the Koebner phenomenon 34 .

[5] 177w RT-PCR analysis revealed elevated mRNA for Stat3 but not Stat1 in keratinocytes of K5.Stat3C mice (Fig. 3a), consistent with increased protein levels (Fig. 2a). Furthermore, this analysis revealed upregulation of several psoriasis-related genes, including Vegfa, Tgfa, Icam1 and Nfkbia (Fig. 3a). Vegf production in newborn keratinocytes was found to be increased more than twofold over controls as detected by ELISA (Fig. 3b), consistent with thicker and more prominent subcutaneous blood vessels in newborn transgenic skin (Supplementary Fig. 2 online). Previous studies have shown that elevated VEGF levels are associated with human psoriasis 35 , and transgenic delivery of VEGF to the skin resulted in an increased vascularity and generation of psoriasiform lesions 16 . Tape stripping-induced psoriatic lesions showed that vessels of the K5.Stat3C mice were dilated and tortuous in the papillary dermis (Fig. 3c). In addition, there was a significant (P < 0.05) increase in blood vessel number throughout the dermis by more than twofold over control (Fig. 3d). We also noted thickening of and an increase in the number of subcutaneous blood vessels (Fig. 3e).

[6] 65w Western blot analysis also showed that IκB-α and cyclin D1 proteins were increased in transgenic keratinocytes compared with controls (Fig. 3f). Flow cytometric analysis showed that cell surface Icam-1 was increased (increase in mean fluorescence intensity) in keratinocytes but not spleen cells isolated from newborn K5.Stat3C mice (Fig. 3g). Also, keratinocyte cell surface Icam-1 expression was considerably increased in tape stripping-induced psoriatic lesions (Fig. 3h).

[7] 184w Tape stripping alone did not induce psoriatic changes in skin grafts from K5.Stat3C mice on athymic nude mice (Fig. 4a); however, psoriatic lesions were reproduced by intradermal injection of activated T cells together with tape stripping (Fig. 4a). It is noteworthy that the skin grafts from control mice did not develop psoriatic lesions after injection with activated T cells together with tape stripping (Fig. 4a). These results clearly show a requirement for both activated Stat3 in keratinocytes and activated T cells in the dermis and epidermis for development of tape stripping-induced psoriatic lesions in mouse skin. Furthermore, immunostaining of tape stripping-induced lesions in K5.Stat3C mice showed that there was a considerable number of CD4 + T-cell infiltrates in the upper dermis and epidermis (Fig. 4b), whereas only a few CD8 + T cells were found preferentially in the epidermis (Fig. 4b). Most of the CD4 + T cells expressed CD25 (data not shown), an early activation marker of T cells, suggesting that CD4 + T cells underwent activation and infiltrated the skin during the development of tape stripping-induced psoriatic lesions in the transgenic mice.

[8] 116w The SCID-human skin xenograft model allowed us to investigate further the involvement of activated Stat3 in human keratinocytes during phenotypic conversion of symptomless (PN) skin into psoriatic plaques (PP) after intradermal injection of activated CD4 + T cells 10,11 . Analysis of SCID-human skin grafts 2-3 weeks after injection revealed that keratinocytes in converted lesions (PN to PP), exhibited substantial nuclear staining for PY-Stat3 (Fig. 4c). In contrast, nonconverted skins (PN to PN), which were injected with activated CD8 + T cells, were devoid of nuclear PY-Stat3 in keratinocytes (Fig. 4c). Identical results were reproduced in at least five different converted grafts (PN to PP) and five different nonconverted grafts (PN to PN) (data not shown).

[9] 161w Topical treatment of a Stat3-specific decoy oligonucleotide 36 was successfully targeted to the epidermis and hair follicles as detected by Cy3 fluorescence (Fig. 5a). We did not observe a signal in the epidermis or follicles of mice treated with a mutant decoy (Fig. 5a), which does not bind to Stat3 (ref. 36). Pretreatment of K5.Stat3C mice with the Stat3 decoy inhibited the generation of tape stripping-induced psoriatic lesions, whereas the mutant DNA did not (Fig. 5b,c). It should be noted that we did not find any considerable infiltrates of T cells in decoytreated skin upon tape stripping (data not shown). The Stat3 decoy also reversed preexisting psoriatic lesions induced by TPA treatment. Four treatments of Stat3 decoy but not mutant decoy resulted in attenuation of the persistent epidermal hyperplasia (Fig. 5d). Also, other characteristic features of psoriasis (hyperkeratosis, parakeratosis, loss of granular layer and dilation of capillaries in papillary dermis) subsided after Stat3 decoy but not mutant decoy treatment (Fig. 5d).

[1] 137w It is known that normal re-epithelization processes during wound healing recapitulate pathogenic processes in psoriasis [25][26][27] . Also, it has been proposed that psoriatic keratinocytes show an abnormal, exaggerated woundhealing response . Because it was previously found that wound healing requires activation of keratinocyte Stat3 (refs. 22-24) and Stat3 is activated in the regenerating epidermis at the edge of wounds (S.S., unpublished data), we examined the status of this transcription factor in the lesional epidermis of psoriatic patients. We found that Stat3 was activated in keratinocytes in the majority of the human psoriatic lesions examined. Activation of Stat3 was not merely a consequence of epidermal hyperplasia because keratinocytes from other nonpsoriatic, inflammatory skin disorders displaying epidermal hyperplasia did not show nuclear localization of Stat3. Thus, activation of Stat3 seemed to be specific for psoriasis among nonmalignant dermatoses.

[2] 496w Analyses of K5.Stat3C mice allowed us to conclude that constitutive activation of Stat3 in keratinocytes leads to development of skin lesions that recapitulate human psoriasis both clinically and histologically. In addition, the location of spontaneous psoriatic lesions (tail and lower dorsum) suggested that physical stimulation might have aided in their development. This was confirmed by inducing psoriatic lesions in transgenic mice using various forms of wounding. The induction of psoriatic lesions in K5.Stat3C mice through wounding was reminiscent of the Koebner phenomenon 34 in human psoriasis. Thus, our present data suggest that activation of Stat3 in keratinocytes and subsequent changes in Stat3-regulated genes are required for development of psoriatic lesions in transgenic mice. The presence of activated Stat3 in keratinocytes from both primary and induced human psoriatic lesions supports the hypothesis that Stat3 activation in keratinocytes may also be required for development of psoriasis in humans. Another notable observation was the presence of activated Stat3 in keratinocytes of uninvolved skin of psoriatic patients. This finding also suggests the possibility that in psoriatic patients the Stat3 pathway may be abnormally sensitive to stimulation, which could lead to a predisposition to psoriasis in humans. The observation that T cellmediated psoriatic conversion took place preferentially in the graft skin derived from psoriatic patients 10 is consistent with our hypothesis that psoriatic epidermis may possess an increased sensitivity to stimulation of the Stat3 signaling pathway. The role of keratinocyte abnormalities versus abnormal immune function in the pathogenesis of psoriasis has been elusive in terms of the primary cause of this disease [2][3][4] . In the current study, data derived from transplantation experiments showed a requirement for activated T cells in the skin together with activated Stat3 in keratinocytes for development of tape-stripping-induced psoriatic lesions in transgenic mice. These data in K5.Stat3C mice, together with the data showing activation of Stat3 in both primary and induced psoriatic lesions in human skin, provide compelling evidence that Stat3 provides an important link between activated keratinocytes and immunocytes, which interdependently participate in the pathogenesis of this disease. It has been reported that ICAM-1 expression is increased in psoriatic lesions 38 . Icam-1 expression was constitutively elevated in keratinocytes of K5.Stat3C mice. Icam-1 expression was further elevated by tape stripping as early as 24 h (data not shown), although psoriatic lesions were induced at 4 d after tape stripping. This rapid upregulation of Icam-1 was inhibited by Stat3-decoy treatment (Supplementary Fig. 5 online), indicating that Icam-1 is a target molecule of Stat3 as reported previously 39 . These results suggest that T cells are recruited to the skin and activated by Icam-1 through LFA-1 (ref. 40) during development of tape stripping-induced psoriatic lesions. In addition to these changes, keratinocytes may produce increased levels of chemokines or integrins whose transcription is upregulated by activated Stat3, further aiding Tcell migration and adherence to the epidermis 41 . Therefore, in psoriatic lesions of K5.Stat3C mice, changes in keratinocytes driven by activated Stat3 lead to changes in immune status.

[3] 203w The exact mechanism underlying activation of Stat3 in human psoriatic keratinocytes remains to be determined. A variety of growth factors and cytokines can lead to Stat3 activation in skin, including members of the IL-6 family, members of the EGF family, HGF and others 18,19,22 . A new IL-10-related cytokine, IL-20, which was cloned in silico, stimulates keratinocyte Stat3 in an autocrine fashion and was found to be overexpressed in psoriatic skin 17 . Psoriatic lesions are characterized by the infiltration of T cells that secrete interferon-γ (IFN-γ) 41,42 . Although IFN-γ inhibits proliferation of normal keratinocytes, it has been shown that cultured psoriatic keratinocytes are relatively resistant to this effect 39,43,44 . Because IFN-γ can induce phosphorylation of Stat3, as well as Stat1 (refs. 45,46), it is possible that molecules downstream of Stat3 may be further upregulated in psoriatic keratinocytes in response to IFN-γ, contributing to development and maintenance of the psoriatic phenotype. Thus, growth factors and cytokines produced during wound healing or secreted by activated T cells and other inflammatory cells are likely to be responsible for activation of Stat3 during the development and maintenance of psoriasis. Further study of psoriatic lesions in K5.Stat3C mice will help to identify these critical factors.

[4] 149w Recently, transgenic mice in which VEGF was expressed under control of the K14 promoter were reported to develop spontaneous psoriasiform lesions and to show the Koebner phenomenon 16 . K5.Stat3C mice develop psoriatic lesions in response to wounding considerably faster than VEGF transgenic mice and with clinical features remarkably similar to human psoriasis. These observations support the conclusion that upregulation of other Stat3-regulated genes in addition to VEGF, including ICAM-1, TGF-α, cyclin D1 and IκB-α contribute to the development of psoriatic lesions in these mice. In this regard, Stat3 may be an important regulator of genes that are involved in the development of the psoriatic phenotype. Overall, by examining the clinical, histologic and immunologic characteristics, K5.Stat3C mice seem to recapitulate many of the key features of psoriasis. In addition, our data suggest that targeting Stat3, using oligonucleotide-based therapies or other approaches, may be an effective new treatment for psoriasis.

[1] 65w Patient skin specimens. We obtained skin biopsies of psoriatic lesions from 21 outpatients (aged 24-57 years; 13 males, 8 females) at Osaka University Hospital with informed consent. Additionally, we took biopsies from individuals suffering from nonpsoriatic, inflammatory dermatoses associated with acanthosis, chronic dermatitis (n = 7), prurigo nodularis (n = 2) and lichen planus (n = 5). We obtained normal skin from three healthy volunteers.

[2] 56w Transgenic mice. We excised the Stat3C cDNA 28 from vector pRc/CMV-S3C and ligated it into the pBK5 construct 29. pBK5.Stat3C was digested with EcoRI, purified and used to generate transgenic founders on an FVB background. Transgenic mice were identified by PCR of genomic DNA with primers specific for the gene encoding rabbit β-globin; 5´-TTCAGGATGTTGTTTA-GAATGG-3´ and 5´-CAATAAGAATATTTCCACGCCA-3´.

[3] 14w Groups were compared using one-or two-tailed Student's t-tests. P < 0.05 was considered significant.

[1] 140w Psoriasis affects approximately 2% of the population in Western countries and is one of the most common inflammatory skin disorders 1 . The pathogenesis of psoriasis remains unclear, and it is controversial as to whether psoriasis results from a primary abnormality in epidermal keratinocytes or from deregulation of the immune system [1][2][3][4] , although these two possibilities are not mutually exclusive. The aberrant function of T lymphocytes has been proposed as a potential cause of psoriasis based on several lines of evidence 1,[5][6][7][8][9][10][11][12] . Over the last decade, transgenic mice have been generated with epidermal expression of molecules shown to be elevated in psoriatic lesions [13][14][15][16][17] . Although many of these models have exhibited various aspects of psoriasis including epidermal hyperplasia with leukocytic infiltrates, none have produced bona fide psoriatic plaques with all of the typical histologic or immunophenotypic abnormalities.

[2] 139w Signal transducers and activators of transcription (STATs) are a family of latent cytoplasmic proteins involved in transmitting extracellular signals to the nucleus 18,19 . Stat3, a widely studied member of this family, undergoes activation through phosphorylation of Tyr705 followed by dimerization and nuclear translocation 18,19 . Stat3 has a critical role in various biological activities, including cell proliferation, survival and cell migration [18][19][20] . Stat3 participates in these processes through regulation of genes such as those encoding cyclin D1, Bcl-xL, and others [18][19][20][21] . Stat3 activation in keratinocytes is essential for skin wound healing [22][23][24] , and activated Stat3 is found in the epidermis at the wound edge in humans and mice (S.S., unpublished data). Because of similarities between acute re-epithelization during wound healing and psoriasis, some investigators have postulated that psoriatic plaques represent persistent wound-healing reactions [25][26][27] .

[3] 107w Here, we show that Stat3 is activated in keratinocytes of human psoriatic lesions. Furthermore, transgenic mice expressing a constitutively active form of Stat3 (Stat3C) in keratinocytes developed skin lesions either spontaneously or in response to wounding that closely resembled human psoriasis. Further analyses showed a dual requirement for both activated Stat3 in keratinocytes as well as activated T lymphocytes in skin of transgenic mice for development of psoriatic lesions. These results led to the hypothesis that changes in Stat3-regulated genes in keratinocytes cooperate with activated immunocytes in the development of psoriasis. Our data suggest that Stat3 represents a potential new therapeutic target for treatment of this disease.

[4] 98w We stained paraffin-embedded skin specimens with antibodies specific for Stat3 (Santa Cruz), PY-Stat3 (Cell Signaling Technology), Ki67 (DAKO) and K6 (Babco) as described 31 . For immunofluorescence with the CD31 (PECAM-1), CD54 (ICAM-1), CD3, CD4, CD8, and CD25 (IL2R-α) antibodies (BD Pharmingen), we incubated frozen skin specimens with primary antibodies followed by FITCor Cy3-conjugated secondary antibodies (Jackson ImmunoResearch). In some specimens, nuclei were counterstained with DAPI (Vector Laboratories). Spleen cells or isolated keratinocytes were stained with a CD54-specific antibody followed by FITC-conjugated IgG against hamster IgG (Jackson ImmunoResearch), and analyzed using an EPICS Elite flow cytometer (Beckman Coulter).

[5] 218w Keratinocyte culture, western blotting and ELISA. We prepared protein lysates from cultured keratinocytes or epidermal scrapings and subjected them to western blotting as previously described 31 with antibodies (1:1,000 dilutions) specific to the following: Stat1 (Transduction Laboratories), Stat3, phospho-Stat3 (PY-Stat3), IκB-α, tubulin (Santa Cruz), cyclin D1 (NeoMarkers), Flag (Sigma), and β-actin (Sigma), followed by treatment with horseradish peroxidase-conjugated antibodies against mouse or rabbit IgG (Amersham), and visualized them using chemiluminescence (ECL, Amersham). Supernatants from subconfluent keratinocyte cultures were subjected to ELISA for Vegf (R&D Systems) according to the manufacturer's protocol. Values are presented as the ratio of Vegf to total protein (pg/mg protein). Skin grafts and xenografts. Skin (2 × 1.5 cm) from either control or K5.Stat3C newborn mice was transplanted onto the back of female Balb/c nu/nu mice, sutured and covered for 1 week with occlusive dressings. After 21 d, activated T cells (2 × 10 7 ) in 100 µl PBS were injected intradermally into the grafts. After 5 min, grafted skins received 20 strokes of tape stripping and mice were killed 4 d later. We orthotopically transplanted human skin xenografts onto SCID mice as described 10,11 . After 2-3 weeks, autologous cells were injected intradermally (300 µl) into the xenograft. The chimeric mice were killed 2.5 weeks after intradermal injection and 4-mm punch biopsies taken.

[6] 64w T lymphocyte culture. To obtain activated T cells, we cultured spleen cells from K5.Stat3C mice in RPMI1640 media supplemented with 10% fetal bovine serum for 72 h in the presence of 0.5 µg/ml of Con A (Sigma) and 10 U/ml of IL-2 (R&D Systems). For injection into SCID-human xenografts, we isolated autologous mononuclear cells from heparinized blood and cultured them as described 10,11 .

[7] 98w Stat3 decoy treatment. Stat3 decoy and mutant control decoy oligonucleotides were designed and prepared as reported 36. Before application, we treated dorsal skin with Nair depilatory cream, followed by ten strokes of tape stripping and application of acetone. Stat3 decoy or mutant decoy (20 µl) was applied topically and the skin sealed with an occlusive dressing. After 24 h, the treated sites received an additional fifteen strokes of tape stripping. Cy3-labeled Stat3 decoy and mutant decoy were used to determine delivery in vivo and were generated by conjugation with Cy3 at the 5´ ends of the sense strands.