Human Keratinocytes Constitutively Express Interleukin-18and Secrete Biologically Active Interleukin-18 AfterTreatment with Pro-Inflammatory Mediators andDinitrochlorobenzene

Shubhada M. Naik, Georgetta Cannon, Guido J. Burbach, Sareeta R. Singh, Robert A. Swerlick,Josiah N. Wilcox* , John C. Ansel and S. Wright CaughmanEmory Skin Diseases Research Core Center, Department of Dermatology and *Division of Hematology/Oncology, Department of Medicine, EmoryUniversity School of Medicine, Atlanta, GA 30322, U.S.A.

Interleukin-18 is a potent inducer of interferon-γ by keratinocyte-secreted interleukin-18 is biologicallyactive, in that conditioned media from phorbol 12-activated T cells, macrophages, and monocytes and is

synthesized as an inactive precursor. Pro-interleukin- myrisate 13-acetate, lipopolysaccharide and DNCB-treated human keratinocytes induce interferon-γ18 must be cleaved by interleukin-1-β-converting

enzyme for secretion of the biologically active form. expression by peripheral blood mononuclear cells.This bioactivity is neutralized by anti-interleukin-18,We report that among selected non-bone marrow

derived skin cells, interleukin-18 mRNA is constitu- but not anti-interleukin-12 antibodies. By immuno-histochemistry, interleukin-18 protein is detected intively expressed by human keratinocytes and not by

dermal microvascular endothelial cells, dermal basal keratinocytes of normal human skin, but itsexpression is markedly upregulated in suprabasalfibroblasts, or melanocytes. Interleukin-18 mRNA and

intracellular protein levels are neither changed in keratinocytes in psoriasis. These findings indicate thathuman keratinocytes are a source of biologically func-human keratinocytes nor induced in human dermal

microvascular endothelial cells, dermal fibroblasts, tional interleukin-18 and thus are capable of playingan initiating part in the local interferon-γ-dependentor melanocytes by exposure to pro-inflammatory

stimuli. Exposure of human keratinocytes to phorbol inflammatoryprocesses through expression, activation,and secretion of interleukin-18. Key words: cytokines/12-myrisate 13-acetate, lipopolysaccharides or the

contact sensitizer DNCB results in the secretion of inflammation/skin/Th1/Th2. J Invest Dermatol 113:766–772, 1999immunoprecipitable interleukin-18 protein. Human

T as IFN-γ-inducing factor (IGIF), is a potent inducer of IFN-γby activated T cells (Okamura et al, 1995). Since its initialcharacterization, IL-18 has been found to have multiple othereffects upon various cells involved in the inflammatory response(Okamura et al, 1995; Tsutsui et al, 1996; Ushio et al, 1996; Puren

he epidermal tissue of the skin is the first line ofdefense against exposure to physical, microbial, andchemical agents that cause local cellular injury(Kupper, 1990; Kalish, 1995). During the evolutionof an inflammatory response, leukocyte subsets are

recruited and extravasate through the locally modified endothelium, et al, 1998). IL-12 and IL-18 play important parts in the developmenta process that is orchestrated by local cytokine production (Springer, of T helper type 1 (Th1) cells (Okamura et al, 1995) and are1995; Ebnet et al, 1996). The activation of macrophages and T synergistic in the induction of IFN-γ by T cells (Okamura et al,lymphocytes during the inflammatory responses elicits a con- 1995; Robinson et al, 1997).comitant release of cytokines by these cells, including interferon IL-18 exhibits an IL-1 signature-like sequences (Ushio et al,(IFN)-γ (Romagnani, 1994). 1996) and contains trefoil structures similar to the IL-1 family of

The newly described cytokine interleukin (IL-18), also known cytokines (Murzin et al, 1992; Bazan et al, 1996). IL-18 and IL-1,however, share only 15–18% homology at the amino acid leveland exhibit different biologic activities that are transmitted throughspecific receptors (Parnet et al, 1996; Matsumoto et al, 1997;Manuscript received April 7, 1999; revised June 30, 1999; accepted for

publication July 22, 1999 Robinson et al, 1997; Torigoe et al, 1997; Dinarello et al, 1998).Reprint requests to: Emory Skin Diseases Research Core Center, The murine (Okamura et al, 1995) and human IL-18 (Ushio et al,

Department of Dermatology, 5001 Woodruff Memorial Bldg., Emory 1996) cDNA share a 65% homology. Human IL-18 (hIL-18) isUniversity School of Medicine, Atlanta, GA 30322, U.S.A. Email: synthesized as an inactive precursor form (pro-hIL-18) of 23 kDascaughm@emory.edu

that does not contain a known signal peptide sequence (OkamuraAbbreviations: HDMEC, human dermal microvascular endothelialet al, 1995). Like the activation of IL-1β (Thornberry et al,cells; HDF, human dermal fibroblasts; HM, human melanocytes; ICE,1992), murine pro-IL-18 has been shown to be activated byinterleukin-1β converting enzyme; PBMC, peripheral blood mono-

nuclear cells. IL-1β-converting enzyme (ICE or Caspase 1) by cleavage at the

0022-202X/99/$14.00 · Copyright © 1999 by The Society for Investigative Dermatology, Inc.

766

VOL. 113, NO. 5 NOVEMBER 1999 HUMAN KERATINOCYTES PRODUCE BIOACTIVE IL-18 767

(LPS), dimethylsulfoxide (DMSO), or 0.001% DNCB in DMSO (all fromN-term-Asp35 to secrete a biologically functional form of 18.8 kDaSigma) for 16 h. Ten micrograms of total cellular RNA isolated from(Ghayur et al, 1997; Gu et al, 1997).treated or untreated cells was analyzed by northern blot using either humanTo date IL-18 has been reported to be produced by activatedIL-18 or β-actin cDNA as probes and were subjected to autoradiographymacrophages such as Kupffer cells, monocytes (Okamura et al,on X-OMAT film (Eastman Kodak Co., Rochester, NY). The human IL-1995; Ushio et al, 1996), osteoblasts (Udagawa et al, 1997), and 18 cDNA was generated by reverse transcription–polymerase chain reaction

dendritic and Langerhans cells (Stoll et al, 1998) in addition to T from PBMC total RNA using primers corresponding to the nucleotidescells. Non-bone marrow derived cells of the skin have been 178–201 (59-ATGGCTGCTGAACCAGTAGAAGAC-39) and 735–759shown to express a wide array of pro-inflammatory cytokines either (59-CTAGTCTTCGTTTTGAACAGTGAAC-39) of the published hIL-

18 cDNA (Okamura et al, 1995) using the Superscript pre-amplificationconstitutively or after a variety of stimuli (Kupper, 1990; Lugersystem (Life Technologies, Inc.) according to the manufacturer’s protocol.and Schwarz, 1990). Consequently, we investigated the profile ofIL-18 cDNA amplified by reverse transcription–polymerase chain reactionIL-18 expression in cultured human cutaneous cells, includingfrom total PBMC RNA was sequenced to confirm identity with thekeratinocytes, dermal microvascular endothelial cells (HDMEC),published hIL-18 cDNA sequence (Okamura et al, 1995) and used todermal fibroblasts (HDF), and melanocytes (HM). Several significantprobe northern blots.reports related to the expression of ICE and bioactive IL-18

prompted us to investigate the effects of pro-inflammatory sub-Polyribosome profile analysis Polysome fractions were obtained asstances on the expression of IL-18 in human keratinocytes, described (Rogers and Munro, 1987; Caruccio and Ross, 1994; Schalinske

HDMEC, HDF, and HM. These reports included secretion of IL- et al, 1998). In brief, whole cell lysates were prepared from untreated1β in response to urushiol (the immunogenic moiety of poison human keratinocyte cells (1 3 107 cells) by scraping the cells in 500 µl ofivy) via activation by ICE (Boehm et al, 1996; Zepter et al, 1997), ice-cold lysis buffer comprised of 150 mM KCl, 10 mM MgCl2, 150 mgmodulation of ICE by activators and inhibitors of protein kinase cycloheximide per ml, 0.5 mg heparin per ml (all from Sigma), 10 mM

Tris–Cl (pH 7.2) (Life Technologies, Inc.), 20 mM DTT, 0.5% NP-40 (allC such as phorbol 12-myristate 13-acetate (PMA) and 1-(5-from United States Biochemical Corp., Cleveland, OH), and 100 Uisoquinolinylsulfonyl)-2-methylpiperazine (Zepter et al, 1997), andRNAsin per ml (Promega Corp., Madison, WI) using diethylpyro-production of bioactive IL-18 by murine keratinocytes (Stollcarbonate (Sigma) treated and autoclaved water. Cells were left on ice foret al, 1997).10 min and then centrifuged for 10 min at 735 3 g to separate nuclei.We report that among the non-bone marrow derived cells of Supernatants were stored on ice. A linear 10–60% (by weight) sucrose

the skin, only human keratinocytes constitutively express IL- gradient was poured in 12 ml Beckman poly-allomer tubes (14 3 89 mm)18 mRNA and protein. On exposure to selected inflammatory containing ultra-pure sucrose (Life Technologies, Inc.) in 20 mM HEPESmediators and the contact sensitizer DNCB IL-18 mRNA levels (pH 7.2), 250 mM KCl, 10 mM MgCl2, 20 mM DTT, 150 µg cycloheximidewere not modulated further in human keratinocytes or induced per ml, 0.5 µg heparin per ml, and 10 U RNAsin per ml. A 50 µl aliquot

of the post-nuclear human keratinocyte supernatant was saved as a controlde novo in HDMEC, HDF, or HM. In addition, upon treatmentand the remainder was applied gently on to the linear sucrose gradient andwith pro-inflammatory mediators and DNCB, human keratinocytescentrifuged at 110,000 3 g in a swinging bucket rotor for 4 h at 4°C.secrete biologically active IL-18 that is capable of inducing IFN-γTotal RNA was isolated from the unfractionated control sample and fromproduction by human peripheral blood mononuclear cells (PBMC).1.25 ml fractions that were collected from the gradient. The RNA samplesImmunohistochemical studies reveal that IL-18 protein is constitu- were then analyzed for the presence of IL-18 mRNA by northern blot.

tively expressed in basal keratinocytes in normal human skin andthat its expression is markedly increased in suprabasal keratinocytes Immunoprecipitation and western blot analysis Equal amounts ofin psoriasis. These studies thus demonstrate that keratinocyte- total protein (50 µg) from treated or untreated cells were analyzed byderived IL-18 may play an important part in IFN-γ-dependent and western blot analysis. Polyclonal N-terminal or C-terminal anti-humanT cell-mediated inflammation in the skin. IL-18 antibodies (Research Diagnostics Inc., NJ) at a 1:100 dilution and

donkey–anti-goat IgG horseradish peroxidase-conjugated antibodies (SantaMATERIALS AND METHODS Cruz Biotechnology Inc., Santa Cruz, CA) at a 1:1500 dilution were used

as primary and secondary antibodies respectively. Actin was used as anCell culture Human PBMC, isolated as previously described (Kanofinternal protein standard for loading control and was detected by mouseet al, 1991) and THP.1, a human monocytic cell line (ATCC, Rockville,anti-actin (Chemicon International, Inc., Temecula, CA) primary antibodiesMD), were cultured in RPMI-1640 media containing 10 mM HEPES,at a 1:1000 dilution and goat anti-mouse Ig (BioRad) at 1:5000 dilution.1 mM sodium pyruvate, 2 mM L-glutamine, 4500 mg glucose per l andEqual volumes of cell culture supernatants were used with anti-IL-181500 mg sodium bicarbonate per l (Vita Cell, ATCC). PBMC wereantibodies at a concentration of 2 µg per ml for immunoprecipitation bysupplemented with 10% human serum (Irvine Scientific, Santa Ana, CA)standard methods, except using a modified immunoprecipitation bufferand THP.1 with 10% fetal bovine serum (Atlanta Biologicals Inc., Norcross,containing 1% Triton X-100, 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mMGA). Human keratinocytes, HM, HDMEC, and HDF were isolated fromethylenediamine tetraacetic acid, 2 mM ethyleneglycol-bis(β-aminoethylneonatal foreskins at the Emory Skin Diseases Research Core Center asether)-N,N,N9,N9-tetraacetic acid, 0.2 mM Na3VO4, 0.2 mM phenyl-previously described (Boyce and Ham, 1985; Kubota et al, 1988). Humanmethylsulfonyl fluoride, and 0.5% NP-40. The immunoprecipitated pro-keratinocytes were cultured in keratinocyte growth media supplementedteins were subjected to a western blot analysis using anti-IL-18 antibodieswith bovine pituitary extract (Clonetics, Walkersville, MD). HDMECor an irrelevant control antibody. Blots were developed using the chemi-were cultured in MCDB 131 media (Life Technologies, Inc., Grand Island,luminescent detection reagent ECL (Amersham Life Science Inc.) andNY) supplemented with 30% normal human serum (Irvine Scientific),exposed to X-OMAT film (Eastman Kodak Co.) to obtain a fluorograph.N 6,29-O-dibutyryl-cyclic adenosine monophosphate (5 3 10–5 M; Sigma),

100 U per ml penicillin, 0.25 mg per ml amphotericin B, and 10 µg perml streptomycin (all from Life Technologies, Inc.). HM were cultured in Immunohistochemistry Paraffin-embedded tissue sections of 4 µm

thickness from biopsied human psoriatic lesions and normal skin wereMelanocyte Growth Medium Bulletkit (Clonetics). HDF were cultured inDulbecco’s modified Eagle’s medium (DMEM) supplemented with 3 mM used after deparaffinization and rehydration according to the standard

protocol (Jaffe and Raffeld, 1996). Sections were fixed in 0.4% formalde-L-glutamine, 100 U per ml penicillin, 0.25 mg per ml amphotericin B,and 10 µg per ml streptomycin (all from Life Technologies, Inc.), and 10% hyde and immersed in 0.1 M Na-citrate (Sigma) at 100°C for 1 min for

antigen unmasking. The tissue sections were subjected to peroxidasefetal bovine serum (Hy-Clone Laboratories Inc., Logan, UT). Cell cultureswere maintained at 37°C in humidified incubators with 5% CO2. Human immunostaining using VectaStain elite ABC kit (Vector Laboratories Inc.,

Burlingame, CA) according to the manufacturer’s recommendations.keratinocytes were passaged at 60–70% confluence to avoid differentiation,using subculture reagents (Clonetics). Experiments with primary cells Normal rabbit serum from the VectaStain kit was used to avoid nonspecific

binding of the secondary biotinylated antibody. Normal goat serum(human keratinocytes, HM, HDMEC, and HDF) were conducted withsubconfluent cells at passage 3. (Accurate Chemical & Scientific Corp., Westbury, CT) was used as a first

step negative control and goat–anti-human IL-18 or goat–anti-human ICE(p20 subunit) antibodies (all from Research Diagnostics Inc.) were used asNorthern blot analysis This was performed for IL-18 mRNA levels

after treatment of cells with a variety of pro-inflammatory mediators. primary antibodies at a 1:20 dilution. The primary antibodies were detectedusing biotinylated rabbit–anti-goat secondary antibodies and a preformedThese mediators were added directly to cells after a fresh media change

and consisted of 10 ng per ml PMA, 50 ng per ml lipopolysaccharide avidin and biotinylated horseradish peroxidase complex from the VectaStain

768 NAIK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

elite ABC kit. Sections were stained with the AEC substrate (VectorLaboratories Inc.), counterstained with Mayer’s hematoxylin, coverslippedand photographed with an Olympus C35AD-4 camera, mounted on anOlympus BH2 microscope.

IFN-γ enzyme-linked immunosorbent assay (ELISA) for IL-18bioassay Secretion of bioactive IL-18 from human keratinocytes aftervarying treatments was determined by the ability of human keratinocytesupernatants to induce IFN-γ secretion from human PBMC. Humankeratinocytes were left untreated or stimulated for 2 h with 10 ng PMAper ml, 10 ng phorbol 12,13-dibutyrate per ml, 50 ng LPS per ml, DMSO,or 0.001% DNCB in DMSO. To avoid a carry over of these substancesto PBMC after the 2 h treatment period, human keratinocytes were washedfour times with solution A (pH 7.4) containing 30 mM HEPES-NaOHbuffer, 10 mM glucose, 3 mM KCl, 130 mM NaCl, 1 mM Na2HPO4•7H2O,and 0.0033 mM phenol red (all from Sigma), and then replenished withnormal keratinocyte growth media. Cells were then cultured for 16 h in2.5 ml media per 100 mm tissue culture plate. Culture supernatants were

Figure 1. Human keratinocytes constitutively expressed IL-18harvested and 40% of the supernatant (1 ml) was used to treat PBMC.mRNA. Northern blot analyses of 10 µg total RNA isolated from eitherPBMC used in the bioassay were pre-stimulated for 16 h with concanavalin(A) human keratinocytes, or (B) HDMEC and HDF using a Klenow filled-A (0.5 mg per ml, Sigma) and then centrifuged and resuspended in RPMIin 32P-labeled IL-18 cDNA to identify the IL-18 mRNA (upper panels) ormedia with 20% human serum. 1 3 106 cells per ml were used for a 24 hβ-actin cDNA to demonstrate even loading (lower panels).(A) Humantreatment with 1 ml of either RPMI media or keratinocyte growth mediakeratinocytes that were either untreated (lanes 1 and 4 ) or treated withalone as controls; RPMI media with the appropriate concentration ofPMA (lane 2), LPS (lane 3), DMSO (lane 5) or DNCB (lane 6). (B)rhIL-18 (Research Diagnostics Inc.), rhIL-12 (R&D Systems Inc.,HDMEC and HDF that were either untreated (lanes 7 and 10) or treatedMinneapolis, MN) or culture supernatants from treated or untreated humanwith PMA (lanes 8 and 11) and LPS (lanes 9 and 12). Data are representativekeratinocytes, to bring the final concentration of human serum to 10%. Aof three separate experiments.human IFN-γ-specific ELISA was used to quantitate IFN-γ secreted by

the PBMC in the media, according to the manufacturer’s recommendations(R&D Systems and Endogen Inc., Woburn, MA) To determine if IFN-γsecretion was specifically induced by IL-18 in the human keratinocytesupernatants, neutralizing antibodies to either rhIL-18 (ResearchDiagnostics Inc.) or to rhIL-12 (Transduction Labs, Lexington, KY) wereadded to human keratinocyte supernatants at the appropriate concentra-tions and incubated for 30 min at 37°C prior to human keratinocytesupernatant exposure to PBMC. rhIL-18 or rhIL-12 preincubated, withor without the neutralizing antibodies, was used as control treatmentsof PBMC.

Data processing The autoradiographs of the northern blots and fluoro-graphs of western blots were scanned on a La Cie flat bed scanner (La CieLtd, Beaverton, OR) utilizing Adobe Photoshop software (Adobe Systems, Figure 2. IL-18 mRNA in human keratinocytes is associated withInc., Mountain View, CA). Subsequently, all of the digitized images were polyribosomes. Total RNA isolated from whole cell lysates of untreatedimported into and labeled in Microsoft Power Point (Microsoft Corp., human keratinocytes (5 3 106 cells) was analyzed by northern blot analysisRedmond, WA) and printed on a high-resolution laser printer. Each to detect presence of IL-18 mRNA in either an aliquot of the lysate thatautoradiograph and fluorograph figure represents a computer-generated was unfractionated (lane 1) or lysates that were fractionated on a 10–60%image and each is typical of the original data in the context of relative linear sucrose gradient (lanes 2–10). The RNA isolated from eitherband and background densities. the unfractionated (Unfx) or the various fractions containing the ribo-

nucleoprotein complexes (RNP) and the polyribosomes associatedRESULTS complexes (polyribosomes) are indicated. Data are representative of three

separate analyses.Human keratinocytes but not HDMEC, HDF, or HM con-stitutively express IL-18 in vitro The profile of constitutive IL-18 mRNA is associated with polyribosomes in humanand induced IL-18 mRNA expression by cultured HK, HDMEC, keratinocytes To determine if the constitutively expressed IL-18HDF, and HM was determined. Human keratinocytes displayed a mRNA in human keratinocytes is also being translated into IL-18robust constitutive IL-18 mRNA expression (Fig 1A). In contrast, protein, we determined the polyribosome profile of IL-18 mRNAconstitutive IL-18 mRNA could not be detected in either HDMEC, in cultured untreated human keratinocytes. As shown in Fig 2,HDF (Fig 1B) or HM (data not shown). Although previously IL-18 mRNA was found to be associated with polyribosomeimplicated in the expression and secretion of bioactive IL-18 in fractions and not within the ribonucleoprotein fraction. Associationvarious cell types, treatment of human keratinocytes, HDMEC, of IL-18 mRNA with polyribosomes indicates that IL-18 mRNA isHDF, and HM with PMA and LPS resulted in no change in their actively being translated into IL-18 protein in cultured, unstimulatedrespective constitutive IL-18 mRNA levels. In addition, human human keratinocytes.keratinocytes treated with DNCB had no effect on RNA levels. Avariety of pro-inflammatory cytokines, growth factors, chemokines, PMA, LPS, and DNCB induce human keratinocytes to

secrete IL-18 Western blot analysis of total cellular proteins fromand neuropeptides, including tumor necrosis factor-α, IL-1β, IL-4,IL-6, IL-8, IL-12, vascular endothelial growth factor, granulocyte human keratinocytes was performed using anti-IL-18 antibodies

(Fig 3A). Lysates from untreated cells as well as cells treated withmacrophage colony-stimulating factor, basic fibroblast growth fac-tor, transforming growth factor-α, substance P, substance K, and PMA, DMSO, or DNCB displayed constitutive IL-18 protein

expression. These findings corroborate our data demonstratingthe contact sensitizer DNCB failed to induce IL-18 mRNAexpression de novo in HDMEC (data not shown). Thus in the constitutive polyribosome association of IL-18 mRNA in cultured

human keratinocytes. We next determined if IL-18 protein couldhuman skin, human keratinocytes but not HDMEC, HDF, or HMexpress constitutive levels of IL-18 mRNA. Moreover, IL-18 be detected in the culture supernatants of human keratinocytes using

immunoprecipitation–western blot analysis. Anti-IL-18 antibodiesexpression was not modulated further in human keratinocytes orinduced de novo in HDMEC, HDF, or HM by pro-inflammatory immunoprecipitated a protein from the culture supernatants of

human keratinocytes that were treated with either PMA, LPS, ormediators including PMA and LPS or the contact sensitizer DNCB.

VOL. 113, NO. 5 NOVEMBER 1999 HUMAN KERATINOCYTES PRODUCE BIOACTIVE IL-18 769

Figure 3. IL-18 protein is constitutively expressed and secreted byhuman keratinocytes after treatment with PMA, LPS, or DNCB.Total protein from human keratinocyte lysates (A) or immunoprecipitatedproteins from culture supernatants of THP.1, HK, and HDMEC usinganti-IL-18 antibodies (B) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and specific proteins were detected bychemiluminescence as described in methods. IL-18, actin, and the molecularweight markers are indicated. These data are representative of threedifferent experiments. (A) Western blot analysis of IL-18 protein in humankeratinocyte lysates. Immunodetection of IL-18 protein (upper panel) andactin (lower panel) as an internal control is shown. Each lane was loadedwith 50 µg of total proteins from human keratinocytes that were eitheruntreated (lane 1) or treated with PMA (lane 2), DMSO (lane 3), or DNCB(lane 4) and subjected to western blot analysis. (B) IP/western blot analysisof IL-18 protein from THP.1, HK, and HDMEC supernatants. Twomicrograms per milliliter of anti-IL-18 antibodies were used to Figure 4. Recombinant human IL-18 and a factor in treated humanimmunoprecipitate proteins from 2.5 ml culture supernatants of THP.1 keratinocyte conditioned media induce IFN-γ expression by human(lanes 1 and 2), HK (lanes 3–7), and HDMEC (lanes 8–12) that were either PBMC. (A) rhIL-18 induces IFN-γ production by human PBMC in auntreated (lanes 1, 3, and 8) or treated with DMSO (lanes 4 and 9), DNCB concentration-dependent fashion. IFN-γ-specific ELISA was performed to(lanes 5 and 10), LPS (lanes 6 and 11), or PMA (lanes 2, 7, and 12). detect production of IFN-γ by human PBMC that were either untreatedImmunoprecipitated proteins were then subjected to western blot analysis (lane 1) or treated rhIL-18 for 24 h as follows: 1 ng per ml (lane 2); 10 ngfor detection of IL-18 protein. per ml (lane 3); 100 ng per ml (lane 4), and 1000 ng per ml (lane 5). Values

are expressed as pg per ml of IFN-γ produced by PBMC and representthe mean6SD of triplicate analyses. (B) Selected human keratinocyteconditioned media induce IFN-γ production by human PBMC. ELISADNCB, but no IL-18 protein was detected in the supernatants offor IFN-γ expression was performed on normal keratinocyte media (KGM)untreated or DMSO-treated human keratinocytes (Fig 3B). Theor human keratinocyte culture supernatants (CS) prior to exposure tohuman keratinocyte secreted protein co-migrated with IL-18 PBMC (left panel: –PBMC) or on CS after treatment of PBMC with

secreted from THP.1 cells upon treatment with PMA (Fig 3B, selected human keratinocyte conditioned media for 24 h (right panel:lane 2) at the expected molecular weight (µ18.8 kDa). THP.1 cells 1PBMC). ELISA results performed without PBMC (left panel) to determinehave been previously shown to secrete IL-18 after PMA stimulation whether human keratinocytes themselves secrete IFN-γ are displayed as(Akita et al, 1997). As expected, anti-IL-18 antibodies failed to follows: lane 1, KGM; lane 2, human keratinocyte CS after treatment with

IL-18 (10 ng per ml); lane 3, human keratinocyte CS after treatment withimmunoprecipitate this protein from the culture supernatants ofPMA (10 ng per ml). ELISA results on PBMC supernatants (right panel )either treated or untreated HDMEC (Fig 3B, lanes 8–12) as theseafter treatment of PBMC with human keratinocyte CS are displayed ascells do not express IL-18 mRNA constitutively or after stimulation.follows: lane 4, untreated human keratinocyte CS; lane 5, PMA (10 ng perUse of an irrelevant first step antibody for immunoprecipitation orml for 16 h) treated human keratinocyte CS; lane 6, LPS (10 ng per ml)second step antibody for western blot failed to detect any proteins. treated human keratinocyte CS; lane 7, DNCB (0.001%) treated human

These results indicate that, although human keratinocytes express keratinocyte CS; lane 8, DMSO treated human keratinocyte CS. Values ofIL-18 mRNA and protein constitutively, they secrete immunode- detected IFN-γ are expressed as pg per ml as determined by internaltectable IL-18 into the culture supernatants only when treated with rhIFN-γ standards and represent the mean6SD of triplicate analyses.inflammatory mediators such as PMA, LPS, and the contactsensitizer DNCB.

concentration-dependent manner by secreting IFN-γ (Fig 4A).Next, we determined whether IFN-γ was present in the growthHuman keratinocytes secrete a factor that induces IFN-γ

production by human PBMC In order to address whether the medium or whether treatment of human keratinocytes by eitherrhIL-18 or PMA resulted in the secretion of IFN-γ into the humanimmunodetected IL-18 protein secreted in the culture supernatants

of treated human keratinocytes was bioactive and capable of keratinocyte culture supernatant. As shown in Fig 4(B) (lanes 1–3),no IFN-γ was detected in either untreated or treated humaninducing IFN-γ production by human PBMC, we first confirmed

that concanavalin A-stimulated PBMC responded to rhIL-18 in a keratinocyte supernatants prior to supernatant exposure to PBMC.

770 NAIK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

IL-18 antibodies did not cross-react with IL-12 that could poten-tially be present in the supernatant. These data confirm thespecificity of the neutralizing effect of anti-IL-18 antibodiesobserved and demonstrate that the IFN-γ-inducing factor secretedby human keratinocytes upon PMA treatment is indeed IL-18.

IL-18 is expressed in vivo in the skin In vivo expression ofIL-18 in normal human and psoriatic skin was determined usingimmunohistochemical techniques. IL-18 expression in normal skin(Fig 6A) was restricted to basal and a few suprabasal keratinocytesand co-localized with ICE expression, whereas control serum failedto show any staining. In contrast, IL-18 expression in psoriatic skin(Fig 6B) was seen predominantly in suprabasal keratinocytes. Againexpression of ICE co-localized with IL-18, whereas control serumfailed to show any staining.

DISCUSSION

Inflammation is a crucial component of the immediate, non-specific immune response to injury or infection, but uncontrolledinflammation results in pathologies such as psoriasis and arthritis.The inflammatory profile of many skin diseases is due in part to Tcell production of pro-inflammatory cytokines such as IFN-γ(Wood et al, 1992; Breathnach, 1993). IFN-γ has wide-rangingimmunomodulatory effects, including the promotion of macro-phage activation, the induction of major histocompatibility complexclass II and adhesion molecule expression, and the promotion ofFigure 5. Anti-hIL-18 but not anti-hIL-12 antibodies neutralize theTh1 and the suppression of Th2-mediated immune responses. InIFN-γ-inducing factor secreted by human keratinocytes. IFN-γ-addition, IFN-γ functions as a priming signal for the induction ofspecific ELISA was performed to assess the ability of neutralizing antibodies

to IL-18 and IL-12 to abrogate the ability of human keratinocyte CS and nitric oxide synthase by endothelial cells (Farrar and Schreiber,rhIL-18 to induce IFN-γ expression by PBMC. PBMC were exposed for 1993) and acts synergistically with other pro-inflammatory cytokines24 h to rhIL-18 and either untreated or PMA-treated (10 ng per ml) such as tumor necrosis factor-α and numerous interleukins.human keratinocyte CS as indicated below. rhIL-18 and human keratinocyte As an IFN-γ-inducing factor, IL-18 is emerging as an importantCS were incubated with neutralizing anti-IL-18 antibodies (8 µg per ml) pro-inflammatory cytokine in localized inflammatory and immuno-or anti-IL-12 antibodies (1 µg per ml) for 30 min at 37°C prior to addition logic diseases. Mice treated with anti-IL-18 receptor antibodiesto PBMC. Variable treatments of PBMC with CS, rhIL-18, and antibodies

have decreased Th1 responses and mice deficient in IL-18 haveare displayed in tabulated form above the ELISA data and are as follows:defective Th1 responses (Takeda et al, 1998; Xu et al, 1998). Thelane 1, untreated; lanes 2–4, rhIL-18 (100 ng per ml); lanes 3 and 7, anti-persistent expression of IL-18 receptor on Th1 cells emphasizes itsIL-18 antibody; lanes 4 and 8, anti-IL-12 antibodies; lane 5, untreated

human keratinocyte CS; lanes 6–8, PMA-treated (10 ng per ml) human importance in the development and perhaps the maintenance ofkeratinocyte CS. Values of detected IFN-γ are expressed as pg per ml as Th1 functions, making it a potential target for clinical interventiondetermined by internal rhIFN-γ standards and represent the mean6SD of (Dinarello et al, 1998; Xu et al, 1998).triplicate analyses. We have shown that among the non-hematopoietic cells of the

human skin, including human keratinocytes, HDMEC, HDF, andHM, IL-18 mRNA is expressed constitutively and only by humanThese data eliminate the possibility of IFN-γ production by humankeratinocytes. We have demonstrated that pro-inflammatory sub-keratinocytes themselves. IFN-γ was induced and expressed bystances, such as PMA and LPS, do not further modulate the IL-18PBMC when treated with culture supernatants from humanmRNA levels in human keratinocytes nor induce IL-18 mRNAkeratinocytes that were either treated with PMA, LPS, DNCBexpression de novo in HDMEC, HDF, or HM. The contact sensitizer(Fig 4B, lanes 5–7), or phorbol 12,13-dibutyrate (data not shown).DNCB also did not affect mRNA levels in human keratinocytes.The magnitude of IFN-γ induction by PBMC after exposure toWe have demonstrated that human keratinocytes constitutivelytreated human keratinocyte supernatants was similar to that inducedtranslate mRNA into protein, but do not secrete immunoprecipit-by rhIL-18. Moreover, the culture supernatants from untreatedable IL-18 protein into the culture supernatant until exposure toor DMSO-treated human keratinocytes failed to induce IFN-γpro-inflammatory agents such as PMA and LPS or the contactproduction by PBMC (Fig 4B, lanes 4 and 8) These results indicatesensitizer DNCB. We have also shown that the culture supernatantsthat the culture supernatants from treated human keratinocytesfrom stimulated human keratinocytes specifically induce IFN-γcontain an IFN-γ-inducing factor.production by PBMC and that this induction is inhibited byneutralizing anti-hIL-18 antibodies, but not by neutralizing anti-The IFN-γ-inducing factor secreted by human keratinocytes

is IL-18 To determine whether the IFN-γ-inducing factor hIL-12 antibodies. Our in vitro findings are buttressed by immuno-histochemical studies of normal and psoriatic skin, demonstratingsecreted by human keratinocytes was indeed IL-18, we assessed

the ability of IL-18 neutralizing antibodies to abrogate the induction constitutive and upregulated expression, respectively, of IL-18protein in human epidermis in vivo.of IFN-γ production in PBMC by rhIL-18 or culture supernatants

from human keratinocytes treated with PMA. As shown in Fig It would be premature to extrapolate from our limited DNCBdata that contact sensitizers in general lead to the secretion of5(A) (lanes 3 and 7 ), anti-rhIL-18 antibodies inhibited the induction

of IFN-γ by PBMC that were either treated with the rhIL-18 or IL-18. To date, we have tested additional compounds, includingnickel sulfate, cobalt chloride, and sodium lauryl sulfate and havewith the culture supernatants from PMA-treated human keratino-

cytes. Anti-hIL-12 antibodies, however, failed to inhibit IFN-γ seen no effect on IL-18 secretion by human keratinocytes. Furtherstudies exploring various dosages and different hapten-generatinginduction by PBMC that were treated with rhIL-18 or the IFN-

γ-inducing factor secreted into the culture supernatants by PMA- sensitizers vs. irritants are under way. Nevertheless, similar to theability of urushiol, also a hapten-generating contact sensitizer, totreated human keratinocytes (Fig 5A, lanes 4 and 8). Antibodies

to hIL-12, but not to hIL-18, neutralized rhIL-12-mediated IFN- induce IL-1β secretion by human keratinocytes through ICEactivation (Boehm et al, 1996; Zepter et al, 1996) it is intriguingγ induction by PBMC (data not shown) indicating that anti-

VOL. 113, NO. 5 NOVEMBER 1999 HUMAN KERATINOCYTES PRODUCE BIOACTIVE IL-18 771

Figure 6. In vivo expression of IL-18 inthe human skin. Biopsied human skinsections from (A) normal human skin and(B) lesional psoriatic skin were subjectedto immunohistochemistry for detection ofin vivo expressed IL-18 and ICE proteins.(A) Immunodetection of normal humanskin using primary antibodies to: (1) hIL-18; (2) ICE p20 subunit; and (3) normalgoat serum control. (B) Immunodetectionof skin sections from psoriatic lesional skinusing primary antibodies to: (1) hIL-18; (2)ICE p20 subunit; and (3) normal goatserum control. Each panel is representativeof skin biopsies from three normal andthree psoriasis patients.

Barker JN, Mitra RS, Griffiths CE, Dixit VM, Nickloff BJ: Keratinocytes as initiatorsthat DNCB is capable of inducing IL-18 secretion in humanof inflammation. Lancet 337:211–214, 1991keratinocytes, which in other cell types requires ICE activation as Bazan JF, Timans JC, Kaselein RA: A newly defined interleukin-1? Nature 379:

well. Previous reports of the processing of IL-18 by ICE in vitro 591–???, 1996Boehm KD, Yun JK, Strohl KP, Elmets CA: In-situ changes in the relative abundance(Ghayur et al, 1997) and of ICE activation in keratinocytes (Boehm

of human epidermal cytokine messenger RNAs following exposure to theet al, 1996; Zepter et al, 1997), together with our observation thatpoison ivy/oak contact allergen, urushiol. Exp Dermatol 5(3):150–160, 1996ICE colocalizes with IL-18 expression in the normal as well as the Boyce ST, Ham RG: Cultivation, frozen storage, and clonal growth of normal

psoriatic skin, suggest that ICE could be responsible for the human epidermal keratinocytes in serum free media. J Tissue Culture Methods9:83–93, 1985proteolytic activation of IL-18 in human keratinocytes. Further

Breathnach SM: The skin system and psoriasis. Clin Exp Immunol 91:343–345, 1993investigations are critical to understand the potential role of IL-18Caruccio N, Ross J: Purification of a human polyribosome-associated 39 to 59in the manifestation of psoriasis and other inflammatory skin exoribonuclease. J Biol Chem 269(50):31814–31821, 1994

diseases. Although originally characterized as a novel IFN-γ- Cooper KD, Hammerberg C, Baadsgaard O, et al: IL-1 activity is reduced in psoriaticskin: decreased IL-1 alpha and increased nonfunctional IL-1 beta. J Immunolinducing factor, IL-18 has more recently been shown to induce144(12):4593–4603, 1990the expression of other pro-inflammatory factors such as tumor

Dinarello CA, Novick D, Puren AJ, et al: Overview of interleukin-18: more thannecrosis factor-α and IL-1β in freshly isolated PBMC (Puren et al, an IFN-γ-inducing factor. J Leuk Biol 63(6):658–664, 19981998). Thus IL-18 may serve to augment cutaneous inflammation Ebnet K, Kaldijian EP, Anderson AO, Shaw S: Orchestrated information transfer

underlying leukocyte endothelial interactions. Annu Rev Immunol 14:155–through several pathways. IL-18-mediated events could potentially177, 1996contribute to the high levels if IL-6 and IL-8 exhibited by human

Farrar MA, Schreiber RD: The molecular cell biology of interferon-gamma and itskeratinocytes in psoriasis (Barker et al, 1991), as tumor necrosis receptor. Annu Rev Immunol 11:571–611, 1993factor-α and IL-1β are known to induce IL-6 and IL-8 (Cooper Ghayur T, Banerjee S, Hugunin M, et al: Caspase-1 processes IFN-γ-inducing factor

and regulates LPS-induced IFN-γ production. Nature 386:619–623, 1997et al, 1990). Our observation that IL-18 expression and distribu-Gu Y, Kuida K, Tsutsui H, et al: Activation of interferon-γ inducing factor mediatedtion is markedly upregulated in psoriatic suprabasal keratinocytes

by interleukin-1β converting enzyme. Science 275:206–209, 1997supports its potential role in this disease. Jaffe ES, Raffeld M: Identification of cells in tissue sections. In: Coligan JE, KruisbeekOur findings that human keratinocytes produce IL-18 mRNA AM, Margulies DH, Shevach EM, Strober W (eds). Current Protocols in

Immunology. New York: John Wiley, 1996, pp. 5.8.1–5.8.8and functional protein underscore the broader role that humanKalish RS: Poison ivy dermatitis: pathogenesis of allergic contact dermatitis tokeratinocytes play in the induction of inflammatory responses in

urushiol. Prog Dermatol 29(3):1–7, 1995the skin. These studies also implicate IL-18 as a potential target Kanof ME, Smith PD, Zola H: Preparation of human mononuclear cell populationsmolecule in designing new strategies for therapeutic interventions and subpopulations. In: Coligan JE, Kruisbeek AM, Margulies DH, Shevach

EM, Strober W (eds). Current Protocols in Immunology. New York: John Wiley,in the treatment of inflammatory skin diseases.1991, pp. 7.1.1–7.1.7

Kubota Y, Kleinman H, Martin GR, Lawley TJ: Role of laminin and basementmembrane in the differentiation of human endothelial cells into capillary-likestructures. J Cell Biol 107(4):1589–1598, 1988The technical assistance of Neera Bahl in cell culture, Katherine A. Casper in

Kupper TS: The role of epidermal cytokines in immunophysiology. In: Shevach E,ELISA, Sue Manos in immunohistochemistry, Mark Stipetic in sequencing, and Oppenheim J (eds). The Role of Cells and Cytokines in Immunity and Inflammation.Patricia A. Kowalczyk in editing is gratefully acknowledged. This work was New York: Oxford University Press, 1990, pp. 285–305

Luger TA, Schwarz T: Evidence of an epidermal cytokine network. J Invest Dermatolsupported by PHS grant AR 41206 and AR 42867 to SWC. GJB is supported95(Suppl. 6):100S–104S, 1990by a Deutsche Forschungsgemeinschaft fellowship.

Matsumoto S, Tsuji-Takayama K, Aizawa Y, Koide K, Takeuchi M, Ohta T,Kurimoto M: Interleukin-18 activates NF-κB in murine T helper type 1 cells.Biochem Biophys Res Commun 234(2):454–457, 1997

Murzin AG, Lesk AM, Chothia CJ: beta-Trefoil fold: Patterns of structure andREFERENCESsequence in the Kunitz inhibitors interleukins-1 beta and 1 alpha and fibroblast

Akita K, Ohtsuki T, Nukada Y, et al: Involvement of caspase-1 and caspase-3 in the growth factors. J Mol Biol 223(2):531–543, 1992production and processing of mature human interleukin 18 in monocytic Okamura H, Tsutsui H, Komatsu T, et al: Cloning of a new cytokine that induces

IFN-γ production by T cells. Nature 378(6552):88–91, 1995THP.1 cells. J Biol Chem 272(42):26595–26603, 1997

772 NAIK ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Parnet P, Garka KE, Bonnert TP, Dower SK, Sims JE: IL-1Rrp is a novel receptor- Takeda K, Tsutsui H, Yoshimoto T, et al: Defective NK cell activity and Th1like molecule similar to the type I interleukin-1 receptor and its homologues response in IL-18-deficient mice. Immunity 8(3):383–390, 1998T1/ST2 and IL-1R AcP. J Biol Chem 271(8):3967–3970, 1996 Thornberry NA, Bull HG, Calaycay JR, et al: A novel heterodimeric cysteine

Puren AJ, Fantuzzi G, Gu Y, Su MS-S, Dinarello CA: Interleukin-18(IFNγ-inducing protease is required for interleukin-1 beta processing in monocytes. Naturefactor) induces IL-8 and IL-1β via TNFα production from non-CD141 human 356:768–774, 1992blood mononuclear cells. J Clin Invest 101(3):711–721, 1998 Torigoe K, Ushio S, Okura T, et al: Purification and characterization of the human

Robinson D, Shibuya K, Mui A, et al: IGIF does not drive Th1 development but interleukin-18 receptor. J Biol Chem 272(41):25737–25742, 1997synergizes with IL-12 for interferon-γ production and activates IRAK and Tsutsui H, Nakanishi K, Matsui K, Higashino K, Okamura H, Miyazawa Y, Kaneda K:NF-κB. Immunity 7(4):571–581, 1997 IFN-gamma-inducing factor (IGIF) up-regulates Fas ligand-mediated cytotoxic

Rogers J, Munro H: Translation of ferritin light and heavy subunit mRNAS is activity of murine natural killer cell clones. J Immunol 157(9):3967–3973, 1996regulated by intracellular chelatable iron levels in rat hepatoma cells. Proc Natl Udagawa N, Horwood NJ, Elliot J, et al: Interleukin-18 (interferon-gamma-inducingAcad Sci USA 84(8):2277–22981, 1987 factor) is produced by osteoblasts and acts via granulocyte/macrophage colony-

Romagnani S: Lymphokine production by human T cells in disease states. Annu stimulating factor and not via interferon-gamma to inhibit osteoclast formation.Rev Immunol 12:227–257, 1994 J Exp Med 185(6):1005–1012, 1997Schalinske KL, Chen OS, Eisenstein RS: Iron differentially stimulates translation of Ushio S, Namba M, Okura T, et al: Cloning of the cDNA for human IFN-γ-mitochondrial aconitase and ferritin mRNAs in mammalian cells. Implications

inducing factor, expression in Escherichia coli, and studies on the biologicfor iron regulatory proteins as regulators of mitochondrial citrate utilization. Jactivities of the protein. J Immunol 156(11):4274–4279, 1996Biol Chem 273(6):3740–3746, 1998

Wood LC, Jackson SM, Elias PM, Grunfeld C, Feingold KR: Cutaneous barrierSpringer TA: Traffic signals on the endothelium for lymphocyte recirculation andperturbation stimulates cytokine production in the epidermis of mice. J Clinleukocyte emigration. Annu Rev Physiol 57:827–872, 1995Invest 90(2):482–487, 1992Stoll S, Mueller G, Kurimoto M, et al: Production of IL-18 (IFN-γ-inducing factor)

Xu D, Chan WL, Leung BP, et al: Selective expression: and functions of interleukinmessenger RNA and functional protein by murine keratinocytes. J Immunol18 receptor on T helper (Th) type 1 but not type 2 cells. J Exp Med159(1):298–302, 1997188(8):1485–1492, 1998Stoll S, Jonuleit H, Schmitt E, et al: Production of functional IL-18 by different

Zepter K, Haeffner A, Soohoo LF, et al: Induction of biologically active IL-1β-subtypes of murine and human dendritic cells (DC): DC-derived IL-18converting enzyme and mature IL-1β in human keratinocytes by inflammatoryenhances IL-12-dependent TH1 development. Eur J Immunol 28(10):3231–

3239, 1998 and immunologic stimuli. J Immunol 159(12):6203–6208, 1997