IL-17 Is Produced by Nickel-Specific T Lymphocytes and Regulates ICAM-1 Expression and Chemokine Production in Human Keratinocytes: Synergistic or Antagonist Effects with IFN-γ and TNF-α¹

Allergic contact dermatitis (ACD)³ is a form of delayed-type hypersensitivity reaction that results from a dysregulated immune response to small chemical substances, the haptens, penetrating into the skin. In the sensitization phase, epidermal Langerhans cells or dermal dendritic cells capture the haptens and migrate from the skin to regional lymph nodes where they can efficiently prime hapten-specific naive T cells (1, 2). These activated T cells acquire then a propensity to recirculate in the skin thanks to the expression of skin-homing receptors such as the cutaneous lymphocyte-associated antigen (CLA) which allows preferential interactions with E-selectin present on activated endothelial cells of the skin microvasculature (3). Further contacts of sensitized subjects with the causative allergen induce a rapid recruitment and expansion of memory hapten-specific T cells in the skin and eventually give rise to the disease. A primary function of T cells is the release of cytokines that regulate the magnitude and duration of both the immune and inflammatory responses. IL-2, IFN-γ, and TNF-α released by CD4⁺ and CD8⁺ T cells exert mostly proinflammatory effects, whereas IL-4 and IL-10 released by CD4⁺ Th2 and other T-regulatory cells inhibit the reaction (4, 5, 6). In particular, IFN-γ and TNF-α are primarily involved in amplifying inflammation by stimulating resident skin cells to synthesize chemokines that attract inflammatory cells and membrane molecules important for the retention and activation of T cells. Epidermal keratinocytes are a major target of IFN-γ and TNF-α. These cytokines induce the release of IL-8, monocyte Chemotactic protein (MCP)-1, and RANTES from keratinocytes in vitro (7, 8). Moreover, IFN-γ and, to a lower degree, TNF-α stimulate the expression of ICAM-1 and MHC class II Ags on keratinocytes both in vitro and in vivo (7, 9, 10). ICAM-1 plays a critical role in T lymphocyte adhesion to keratinocytes by acting as the ligand for LFA-1- and Mac-1-bearing leukocytes (11). In addition, ICAM-1 expression can be important for an efficient lysis of keratinocytes by cytotoxic T cells (12). In ACD, keratinocyte ICAM-1 expression is markedly increased and correlates with the infiltration of LFA-1-positive cells (13). Similarly, in the lymph derived from skin sites affected by ACD, large amounts of soluble ICAM-1 (sICAM-1) can be measured (14). Moreover, fluid from suction blisters raised over ACD lesions, but not on normal skin, contains T cell chemoattractants, with keratinocytes being the likely source of these factors (15, 16).

IL-17 is a novel cytokine, apparently secreted only by CD4⁺-activated memory T cells, that induces the release of cytokines and prostaglandins from stromal cells and macrophages (17, 18, 19). Both TNF-α and IFN-γ had an additive effect on the IL-17-induced secretion of IL-6, and the combination of IL-17 plus TNF-α was effective in promoting GM-CSF release by synovial fibroblasts (18). Furthermore, IL-17 was shown to enhance cell surface expression of ICAM-1 on human fibroblasts (17). Thus, a broad set of effects are induced by IL-17, and its action can be potentiated by other cytokines. In the present study, we investigated whether hapten-specific skin-homing T cells produce IL-17, and the influences of IL-17 alone or in combination with IFN-γ or TNF-α on the immune activation of keratinocytes.

Recombinant human (rh) IFN-γ and TNF-α were obtained from Genzyme (Cambridge, MA); rhIL-17 from R&D Systems (Abingdon, Oxon, U.K.). Mouse anti-human IFN-γ receptor (IgG1) and polyclonal rabbit anti-human TNF-α were from Genzyme; anti-TNF-α receptor type I (htr 9, IgG1) mAb was from Biomedicals (Augst, Switzerland); goat anti-IL-17 polyclonal Ab was purchased from R&D Systems and rabbit anti-IL-17 polyclonal Ab was from Peprotech (Rocky Hill, NJ). The HECA-452 mAb (rat IgM) was kindly provided by Dr. Louis J. Picker (Lab. of Experimental Pathology, Dept. of Pathology, University of Texas Southwestern Medical Center, Dallas, TX). Phycoerythrin (PE)-conjugated anti-CD8 (SK1, IgG1), FITC-conjugated anti-HLA-DR (L243, IgG2a), anti-CD4 (SK3 and SK4, IgG1) and anti-CD28 (Leu-28, IgG1) mAbs were purchased from Becton Dickinson (San Jose, CA); FITC-conjugated anti-CD54 (84H10, IgG1), anti-CD80 (MAB104, IgG1), PE-conjugated anti-TCR-αβ (BMA031, IgG2b), FITC-conjugated anti-TCR-γδ (IMMU510, IgG1), and anti-CD3 (UCHT-1, IgG1) mAbs were from Immunotech (Marseille, France). Anti-CD40 (BE-1, IgG1) and anti-MHC class I (W6-32, IgG2a) mAbs were from Ancell (Bayport, MN) and Dako (Glostrup, Denmark), respectively; PE-conjugated anti-CD86 (IT2.2, IgG2b) was from PharMingen (San Diego, CA). F(ab′)₂ fragments of FITC-conjugated anti-mouse Ig came from Silenus (Hawthorn, Australia), and PE-conjugated anti-rat IgM from PharMingen. Control unconjugated and FITC-conjugated mouse IgG1, IgG2a were purchased from Becton Dickinson, and rat IgM from PharMingen.

Biopsies of normal skin and 48-h positive patch test reactions to 5% NiSO₄ from two patients allergic to nickel were either frozen or used to isolate nickel-specific T cell lines. To this end, skin was washed with PBS and then placed in culture in RPMI 1640 complemented with 2 mM glutamine, 1 mM sodium pyruvate, 1% nonessential amino acids, 0.05 mM 2-ME, 100 U/ml penicillin, 100 μg/ml streptomycin (all from Life Technologies, Chagrin Falls, OH) (complete RPMI), 5% FCS (HyClone, Logan, UT), and 20 U/ml rIL-2 (kindly provided by Chiron Italia, Siena, Italy) for 12 days, with fresh medium and IL-2 replaced every 3 days. T cell clones were prepared from short term nickel-specific CD4⁺ and CD8⁺ T cell lines obtained from the peripheral blood as previously described (6). T cell lines were cloned by limiting dilution (0.6 cell/well) in the presence of 2 × 10⁵ PBMC, 20 U/ml rIL-2, and 1% PHA (Life Technologies) in U-bottom 96-well microplates. T cell cultures were performed in complete RPMI supplemented with 10% FCS and 3% human plasma. Clones were grown by adding rIL-2 (20 U/ml) twice a week and were periodically stimulated with 1% PHA in the presence of feeder cells. Skin-derived T cell lines and resting T cell clones were assayed for Ag specificity after extensive washing to remove IL-2, using autologous EBV-transformed B cell lines as APC and 10 μg/ml NiSO₄ (Sigma, St. Louis, MO). After 48 h, the cultures were pulsed overnight with 5 μCi/ml [³H]thymidine (Amersham, Little Chalfont, U.K.) at 37°C and then harvested onto fiber-coated 96-well plates (Packard Instruments, Groningen, The Netherlands). Radioactivity was measured in a beta counter (Topcount, Packard Instruments). Supernatants from the nickel-specific CD4⁺ T cell clone, OC2, were collected after 48 h of culture (10⁶ cells/ml in 24 wells) in complete RPMI supplemented with 5% human serum in the presence of coated anti-CD3 (1 μg/ml) and soluble anti-CD28 (1 μg/ml) mAbs, filtered, and then stored at −80°C.

Samples of normal human skin were obtained from adult donors (n = 3) undergoing abdominoplasty or mammoplasty surgery. Epidermal cell suspensions were prepared as previously described (20). In brief, epidermal sheets were separated from dermis using 0.5% dispase (Boehringer Mannheim, Mannheim, Germany) and then disaggregated to single-cell suspensions using 0.25% trypsin (Biochrom, Berlin, Germany). Keratinocyte primary cultures were established by seeding epidermal cells (1.2–2 × 10⁴ cells/cm²) on a feeder layer of irradiated 3T3/National Institutes of Health fibroblasts (2 × 10⁴ cells/cm²) and cultured in a modified Green’s medium, as described before (20). At 70–80% confluence, keratinocytes were detached with 0.05% trypsin/0.02% EDTA and then aliquoted and cryopreserved in liquid nitrogen. Second- or third-passage keratinocytes were used in all experiments, with cells cultured in the serum-free medium, keratinocyte growth medium (Clonetics, San Diego, CA), for at least 3–5 days before the experiments were performed. During incubation with cytokines, hydrocortisone was omitted from the culture medium. Keratinocyte cultures were devoid of any contaminating leukocyte as assessed by flow cytometry analysis using mAbs against dendritic cell, monocyte, and T cell markers (not shown). The HaCaT keratinocyte cell line (gift from Dr. N. Fusenig, Deutsches Krebsforschungszentrum, Heidelberg, Germany) and A431 keratinocytes were cultured in DME medium supplemented with 10% FCS.

The immunophenotype of the nickel-specific T cell lines and clones was evaluated by double-color flow cytometry analysis using anti-CD4, anti-CD8, anti-TCR-αβ, and anti-TCR-γδ PE- or FITC-conjugated mAbs. CLA expression was studied using the HECA-452 mAb followed by PE-conjugated anti-rat IgM. Keratinocytes were stained with FITC-conjugated mAbs or unconjugated primary mAbs followed by FITC-conjugated anti-mouse Ig. In control samples, staining was performed using isotype-matched control Abs or omitting primary mAbs. Cells were analyzed with a FACScan equipped with Cell Quest software (Becton Dickinson, Mountain View, CA).

Media conditioned for 48 h by T cell clones stimulated with 10 ng/ml PMA and 1 μg/ml ionomycin (Sigma) in 24 wells at 10⁶ cells/ml were harvested and filtered. IL-4, IFN-γ, and TNF-α content was measured with commercially available sandwich ELISA kits (R&D Systems), following the manufacturer’s instructions. IL-17 release was determined in supernatants by using microtiter plates coated with 4 μg/ml goat anti-IL-17 diluted in 50 mM carbonate buffer, pH 9.0. Staining was performed with 2 μg/ml rabbit anti-IL-17 in PBS containing 1% BSA followed by an anti-rabbit IgG horseradish peroxidase-conjugated Ab (Santa Cruz Biotech, Santa Cruz, CA). Chemokines and sICAM-1 were measured on cell-free supernatants collected after a 48-h treatment of subconfluent keratinocyte cultures performed in 6-well plates. The following ELISA kits were used: ICAM-1 Predicta kits from Genzyme; RANTES, IL-8, and MCP-1 Quantikine kits from R&D Systems. An ELISA reader model 3550 UV Bio-Rad (Hercules, CA) was used. T cell and keratinocyte cultures were conducted in triplicate for each condition.

Total cellular RNA was extracted from skin samples, T cells, and keratinocytes using the acid guanidinium thiocyanate-phenol-chloroform method (21). For Northern blot experiments, 15 μg of RNA were fractionated on 1% formaldehyde-agarose gel, blotted to nylon membranes (Amersham), and fixed by UV irradiation. Blots were hybridized with PCR-amplified probes, corresponding to ICAM-1, MCP-1, and IL-8 (22, 23, 24), and labeled by random priming (Boehringer Mannheim) with [³²P]dCTP. After 20 min of prehybridization in Quickhyb solution (Stratagene, La Jolla, CA), denatured probe was added, and hybridization was conducted for 1 h at 68°C. Blots were washed under highly stringent conditions and subjected to autoradiography. Equal loading and integrity of RNA were assessed either by ethidium bromide staining of the gels or hybridizing the membrane with a probe specific for 28S rRNA. For RT-PCR analysis, 0.5–1 μg of total RNA was converted in cDNA using oligo(dT) primers and then amplified with a GeneAmp RNA PCR kit (Perkin-Elmer, Roche Molecular Systems, Branchburg, NJ) according to the manufacturer’s instructions. The following synthetic oligonucleotides were used: for IFN-γ amplification, primers TGCAGGTCATTCAGATGTAG and AGCCATCACTTGGATGAGGG (306-bp amplification product); for TNF-α, ATGAGCACTGAAAGCATGATCCGG and CTACAACATGGGCTACAGGCTTGT (295-bp amplification product); for IL-17, primers TGGAGGCCATAGTGAAGG and GGCCACATGGTGGACAAT (415-bp amplification product) (17); for IL-17 receptor extracellular domain, primers CTAAACTGCACGGTCAAGAAT and ATGAACCAGTACACCCAC (833-bp amplification product); for IL-17 receptor intracellular domain, primers ATGGACAGGTTCGAGGAG and TTCACGATGCCGGTTCCC (276-bp amplification product) (25); for RANTES, primers TCATTGCTACTGCCCTCTGGG and CGTCGTGGTCAGAATCTGGG (373-bp amplification product) (26). As an internal control for the amount of RNA used, the glyceraldehyde-3-phosphate dehydrogenase housekeeping gene was used with primers TGAAGGTCGGAGTCAACGGATTTGGT and CATGTGGGCCATGAGGTCCACCAC (expected cDNA product, 983 bp). For semiquantitative analysis of RANTES, RNA concentrations, primers, and PCR cycles were titrated to obtain standard curves to verify linearity and to permit analysis of signal strength.

Supernatants were collected from T cell clones stimulated or not for 48 h with PMA and ionomycin. Supernatants corresponding to 10⁶ cells and rhIL-17 were immunoprecipitated with goat anti-IL-17 polyclonal Ab and protein G-Sepharose beads (Pharmacia Biotech, Uppsala, Sweden) before analysis on 12% SDS-PAGE. After tranferring to PVDF membrane, immunoprecipitates were probed with goat anti-IL-17 Ab at 1:500 dilution. Blots were developed with anti-goat horseradish peroxidase-conjugated IgG (Santa Cruz), using the ECL-Plus immunodetection system (Amersham) followed by autoradiography.

Wilcoxon’s signed rank test was used (SigmaStat, Jandel, San Rafael, CA). p values ≤ 0.05 were considered significant.

In the first set of experiments, we determined whether skin affected with ACD to nickel and hapten-specific T cells expressed IL-17. Skin biopsies from positive patch tests to nickel performed in two allergic patients showed the simultaneous presence of mRNA for IFN-γ, TNF-α, and IL-17. IL-17 mRNA could also be detected in skin affected by psoriasis, a chronic inflammatory disease with prominent CD4⁺ T cell infiltration, thus suggesting that IL-17-producing T cells can accumulate in the skin during different inflammatory conditions. In contrast, normal skin did not express TNF-α and IL-17 mRNAs and showed limited IFN-γ mRNA expression (Fig. 1). Nickel-specific T cell lines generated from the same ACD biopsies were >90% CD4⁺ and showed, upon 12-h stimulation with autologous APC and NiSO₄, a similar pattern of cytokine expression, which included IFN-γ, TNF-α, and IL-17 (Fig. 1). To investigate IL-17 production at the single-cell level, T cell clones specific for nickel were prepared from the peripheral blood of allergic patients and characterized for their Ag specificity, surface phenotype, cytokine release profile, and expression of the skin-homing receptor, CLA. Eight TCR-αβ⁺,γ/δ⁻CD4⁺, and two CD8⁺ T cell clones were included in the study. They were strictly nickel specific, as assessed in proliferation assays performed in the presence of autologous APC cells and NiSO₄ (Table I). According to the relative IFN-γ and IL-4 release following activation with PMA and ionomycin, four Th1, two Th0, two Th0/Th2, one Tc0, and one Tc1 clones were defined. All CD4⁺ clones released also substantial amounts of TNF-α, with the two CD8⁺ clones being less productive. In addition, with the exception of the TM35, all nickel-specific T cell clones analyzed were found to express the CLA receptor (Table I), thus possessing the potential to recirculate in the skin. Supernatants from four of eight activated CD4⁺ clones contained IL-17 protein, ranging from 0.5 to 1.9 ng/10⁶ cells, as assessed by ELISA (Fig. 2,A), whereas none of the two CD8⁺ clones released detectable amounts of IL-17. IL-17 release, albeit at slight lower levels, was also observed when CD4⁺ T cell clones were stimulated with soluble anti-CD3 mAb and PMA, coated anti-CD3 and soluble anti-CD28, or autologous APC and NiSO₄ (data not shown). The IL-17-releasing clones comprised three Th1 and one Th0/2 clones, and were isolated from two different nickel-allergic patients. RT-PCR analysis confirmed that IL-17 mRNA was expressed after 5 h of activation with PMA/ionomycin by the four IL-17 releasing CD4⁺ clones (Fig. 2,B). In contrast, activated CD8⁺ T cells did not show detectable IL-17 transcripts even after performing a nested PCR analysis. IL-17 has been shown to be secreted as a mixture of glycosylated and nonglycosylated homodimers which dissociate in two compounds of 22 and 15 kDa, respectively, under reducing conditions (18). Therefore, supernatants from two activated CD4⁺ T cell clones (OC2 and TM34) with the higher IL-17-releasing capacity were immunoprecipitated with a specific anti-IL-17 Ab and then analyzed on reduced SDS-PAGE. Both glycosylated and nonglycosylated forms were released by OC2 and TM34 cells, with the levels of IL-17 detected by immunoprecipitation-SDS-PAGE comparable with those measured by ELISA (Fig. 2 C).

Next, we evaluated whether normal human keratinocytes and keratinocyte cell lines expressed the IL-17 receptor. IL-17 receptor cDNA has been recently cloned and characterized from a human T cell library. It does not share homology with previously identified cytokine receptor families and exhibits a broad tissue distribution (25). RT-PCR analysis performed with two couples of primers specific for the extracellular and intracellular portion revealed that normal keratinocytes as well as A431 and HaCaT keratinocytes express constitutively the IL-17 receptor. Treatment of normal keratinocytes with TNF-α, IFN-γ, or IL-17 for 14 h did not seem to change IL-17 receptor gene expression (data not shown).

The capacity of IL-17 to modulate the expression of cell surface-immunomodulatory molecules on keratinocytes was investigated at both protein and mRNA levels using IL-17 alone or in combination with IFN-γ and/or TNF-α. Treatment of cultured human keratinocytes for 24–72 h with IFN-γ (20–200 U/ml) induced de novo expression of ICAM-1 and HLA-DR as well as up-regulated MHC class I and CD40 molecules, whereas TNF-α (50 ng/ml) promoted weak ICAM-1 expression but did not induce MHC and CD40 molecules, as previously reported (7, 27). Incubation of keratinocytes with IL-17 alone (50 ng/ml) had no effects on the expression of ICAM-1, HLA-DR, MHC class I, or CD40. However, IL-17 could stimulate a marked and selective enhancement of IFN-γ-induced membrane ICAM-1 (Fig. 3). This activity was dose dependent, it was confirmed using different doses of IFN-γ (data not shown), and it was specific, given that it could be completely abolished with an anti-IL-17 Ab but not with an irrelevant matched isotype Ab (Fig. 3). The cooperative action of IL-17 in increasing the IFN-γ-induced ICAM-1 was similar to or higher than that of TNF-α, and a dramatic increase of membrane ICAM-1 expression was observed upon keratinocyte treatment with a mixture of IFN-γ, IL-17, and TNF-α (Fig. 3). On the other hand, IL-17 did not change keratinocyte expression of MHC and CD40 molecules induced by IFN-γ, and ICAM-1 expression stimulated by TNF-α (Fig. 3). To more closely reproduce the in vivo situation, culture supernatant from IL-17-releasing nickel-specific CD4⁺ T cells was tested for its effect on keratinocyte ICAM-1 expression. Supernatant from the Th1-like clone, OC2, activated with coated anti-CD3 and soluble anti-CD28 mAbs, was capable of strongly up-regulating ICAM-1 on keratinocytes, an effect that could be markedly inhibited by treating keratinocytes with a blocking anti-IFN-γ receptor mAb. Addition to the supernatant of neutralizing anti-TNF-α or anti-IL-17 Ab resulted in a less prominent, but still significant, reduction of ICAM-1 expression. Blocking the activity of IFN-γ and IL-17 or of all three cytokines further inhibited, although not completely, ICAM-1 on keratinocytes (Table II).

In the following experiments, we investigated whether the increase of membrane-bound ICAM-1 expression induced by IL-17 was associated with changes in the release of sICAM-1, a form that derives from proteolytic cleavage of the membrane-bound form rather than from an alternative splicing (22). As shown in Fig. 4, supernatants from IL-17-stimulated keratinocytes did not contain significant amounts of sICAM-1, whereas cultures treated with IFN-γ (200 U/ml) or TNF-α (50 ng/ml) for 48 h released large amounts of sICAM-1. In keeping with the membrane ICAM-1 data, IL-17 augmented dose dependently the release of sICAM-1 induced by IFN-γ, but not by TNF-α, an effect that could be suppressed by the addition of IL-17-neutralizing Ab. Again, sICAM-1 production was maximal when keratinocytes were stimulated with a mixture of IFN-γ, IL-17, and TNF-α. Consistent with the results at the protein level, Northern blot analysis revealed substantial amounts of ICAM-1 mRNA in keratinocytes upon treatment with IFN-γ for 14 h, and addition of 50 ng/ml IL-17 to the cultures resulted in a significant increase in ICAM-1 mRNA content (Fig. 5, lane 5 vs lane 2), further augmented when the three cytokines were used in combination (Fig. 5, lane 8). Finally, the same cytokines, even when used together, were ineffective in inducing the costimulatory molecules, B7-1 and B7-2, on keratinocyte cell surface (not shown).

Keratinocytes can synthesize and release several chemokines active on a variety of leukocytes (28). Among the C-X-C chemokines, IL-8 is a powerful chemoattractant for neutrophils. RANTES and MCP-1 belong to the C-C family of chemokines, with RANTES being chemotactic for T cells and eosinophils and MCP-1 mostly active on T cells, monocytes, and dendritic cells (29, 30). Chemokine production by keratinocytes is strongly regulated by proinflammatory cytokines. Both IFN-γ and TNF-α promote RANTES release, whereas IL-8 and MCP-1 are preferentially induced by TNF-α and IFN-γ, respectively (7, 8). IL-17 has been reported to induce IL-8 release by skin and rheumatoid synovial fibroblasts (17, 18), and therefore we studied the effects of IL-17 on chemokine synthesis by keratinocytes. IL-17 alone increased IL-8 mRNA (Fig. 5, lane 3), and dose dependently augmented IL-8 release by keratinocytes, with a potency higher than that of IFN-γ (Fig. 6,A). IL-17 also synergized with both TNF-α and IFN-γ in inducing IL-8 (Fig. 5, lanes 5 and 6), with maximal mRNA expression (Fig. 5, lane 8) and protein release (Table III) when keratinocytes were stimulated with the three cytokines together. Addition of anti-IL-17 Ab could abolish completely IL-17 activity. On the other hand, IL-17 had no effect on MCP-1 production (Fig. 5) and did not influence IFN-γ- or TNF-α-induced MCP-1 release (Fig. 6,B and Table III). Finally, IL-17 inhibited expression of RANTES mRNA induced by IFN-γ and/or TNF-α (Fig. 5, lanes 5 and 6). Northern blot analysis was not sensitive enough to detect RANTES mRNA, and thus semiquantitative RT-PCR analysis was performed. A dose-dependent suppressive activity of IL-17 on RANTES production was confirmed at the protein level (Fig. 6,C), it was also evident when keratinocytes were stimulated with IFN-γ plus TNF-α and could be reversed with the addition of anti-IL-17 Ab (Table III).

The present study shows that nickel-specific CD4⁺ T cells expressing the CLA skin homing receptor can synthesize and release IL-17 and that IL-17 may exert important regulatory effects on human keratinocytes with respect to the cutaneous expression of ACD. Interestingly enough, IL-17 was not detected in all the nickel-specific T cell clones studied, but only in some CD4⁺ T cell clones with the characteristics of the Th1 (3 clones) or Th0/2 (1 clone) subset. An extensive analysis of more than 50 nickel-specific CD4⁺ T cell clones isolated from ACD skin or peripheral blood confirmed that ∼50% of the clones produce IL-17, with the IL-17-releasing clones belonging to the Th1 (50%), Th0 (32%) or Th2 (18%) subset. Thus, IL-17 production does not appear to segregate into a distinct Th subset (C. Albanesi, A. Cavani, S. Sebastiani, C. Scarponi, and G. Girolomoni) manuscript in preparation). The two CD8⁺ T cell clones failed to synthesize IL-17, in agreement with previous results on activated peripheral blood CD8⁺ T cells (18). IL-17 receptor gene expression was shown to be constitutive in cultured normal keratinocytes and in human keratinocyte cell lines, consistently with its broad tissue distribution (25). Because the majority of nickel-specific T cells examined released IFN-γ and TNF-α, IL-17 activity on keratinocytes was examined also in combination with these two cytokines.

A prominent feature of a variety of immune-mediated skin disorders, including ACD and psoriasis, is the intraepidermal presence of lymphocytes, and ICAM-1 provides a major adhesion pathway by which T lymphocytes bind to keratinocytes and are thus retained in the epidermis (11). In addition, ICAM-1 can serve as a relevant accessory signal for activation of both CD4⁺ and CD8⁺ T lymphocytes (12, 31). ICAM-1 is induced on keratinocytes by a limited set of cytokines, especially by IFN-γ and TNF-α. Here we showed that IL-17 did not induce ICAM-1 on keratinocytes per se but efficiently cooperated with IFN-γ in promoting the synthesis and membrane expression of ICAM-1. IL-17 was as potent as TNF-α in enhancing IFN-γ-induced ICAM-1, and very high levels of this adhesion molecule were measured on keratinocytes when the three cytokines were used in combination. In parallel, IL-17 also increased IFN-γ-induced release of sICAM-1 in supernatants from keratinocyte cultures. The function of sICAM-1 has not been thoroughly investigated, but evidence exists suggesting that sICAM-1 can regulate LFA-1-mediated events, providing a mechanism that may prevent adhesion or promote deadhesion between leukocytes and ICAM-1-positive cells (22). A role for sICAM-1 in vivo is suggested by the finding that the lymph derived from skin sites affected by ACD contains elevated levels of sICAM-1 (14).

ACD is characterized by an inflammatory infiltrate mainly constituted by CD4⁺ and, to a lower extent, by CD8⁺ T lymphocytes, monocytes, and dendritic cells. In the acute phase, but not in chronic ACD, neutrophils can also be present in the epidermis (32). The recruitment and activation of these cells are under the control of chemokines released by resident skin cells, and many studies indicate that activated keratinocytes are an important source of these factors (7, 8, 15, 16, 28). In ACD to urushiol, keratinocytes were shown to produce IL-8 at 24–48 h after hapten challenge, a time that correlated with the migration of T cells toward the epidermis (33), and IL-8 neutralization suppressed the development of delayed-type hypersensitivity reactions in rabbit skin (34). Another chemokine important for the expression of ACD is MCP-1. Transgenic mice with basal layer keratinocytes expressing constitutively MCP-1 showed enhanced contact hypersensitivity responses coupled with an exaggerated infiltration of dendritic cells and Langerhans cells in the skin (30). In addition, Abs against MCP-1 could almost completely abolish immigration of T cells and monocytes in rats undergoing cutaneous delayed-type hypersensitivity reactions (35). Finally, certain chemokines, such as RANTES, other than attract inflammatory cells, can mediate the adhesion of CD4⁺ T lymphocytes to keratinocytes, thus retaining the targeted leukocytes in the epidermis (36). An interesting result of our study was the selective regulatory activity of IL-17 on chemokine production by keratinocytes. In fact, IL-17, directly or in synergism with IFN-γ and/or TNF-α, stimulated IL-8 synthesis, whereas it inhibited IFN-γ- and TNF-α-induced production of RANTES and had no effect on MCP-1. Through selective modulation of chemokine release by keratinocytes, IL-17 may regulate the recruitment of distinct leukocyte and T cell subsets (37, 38, 39, 40), and thus ultimately direct the outcome of cell-mediated immune responses in the skin. When IFN-γ, TNF-α, and IL-17 were used in combination, they induced a considerably high production of IL-8 compared with the lower amounts of RANTES and MCP-1. Although T lymphocytes bear low levels of IL-8 receptors and the ability of IL-8 to induce T cell migration is still controversial (29, 41, 42, 43), it has been reported that IL-8 is selectively involved in the enhanced migration of CLA⁺ T cells across activated endothelium (44). Moreover, IFN-γ and TNF-α up-regulate IL-8 receptors on T cells (45), and IL-8 released from TNF-α-activated keratinocytes promotes directed migrational responses of both neutrophils and T lymphocytes (16). IL-8 may also act indirectly by stimulating the release of T cell chemoattractants from neutrophils (46). Therefore, IL-8 produced by endothelial cells and subsequently by keratinocytes may have a role in the homing of specific T cells to inflamed skin (33). The question remains of why the high production of IL-8 by keratinocytes is not associated with a strong neutrophil infiltration in the skin during ACD.

The mechanisms by which IL-17 modulates TNF-α and IFN-γ activities have not been investigated. It is unlikely that this modulation is exerted at the receptor level because IFN-γ and TNF-α did not appear to modify IL-17 receptor gene expression and, conversely, IL-17 did not change keratinocyte membrane expression of IFN-γ and TNF-α receptors (not shown). Studies with a variety of human cell types have demonstrated that TNF-α and IFN-γ coregulate in a synergistic manner the expression of many inflammatory genes via the independent activation of two distinct transcription factors, STAT1 and NF-κB, activated preferentially by IFN-γ and TNF-α, respectively (47). IFN-γ-induced STAT1 and TNF-α-induced NF-κB synergistically stimulate the transcription of ICAM-1 and RANTES genes (47). IFN-γ plus TNF-α can also synergistically induce the production of IL-8, but in this case it appears that activator protein 1 (AP-1) and NF-κB-like factor binding elements are involved in conferring responsiveness (48). The direct and cooperative action of IL-17 in inducing IL-8 synthesis and enhancing IFN-γ-dependent ICAM-1 expression by keratinocytes may be related to the ability of IL-17 to activate NF-κB and AP-1 (19, 49). On the other hand, IFN-γ-and/or TNF-α-dependent RANTES production may be down-regulated by IL-17 through the activation of inhibitory factor(s), and the upstream region of RANTES gene presents inhibitory regulating elements the deletion of which results in enhanced promoter activity (50). Likewise, IL-17 on keratinocytes, IL-4 and IL-13 inhibit IFN-γ- and TNF-α-induced RANTES release from endothelial and smooth muscle cells (51, 52).

In conclusion, our results suggest that IL-17 is an important element in regulating the outcome of skin immune responses to haptens, and probably of other T cell-mediated skin immune responses, with synergistic or antagonist effects on IFN-γ- and TNF-α-stimulated keratinocyte activation.

We thank L. Picker for donation of the HECA-452 mAb, N. Fusenig for the HaCaT cell line, and S. Pastore for critical reading of the manuscript.