APMIS 99: 58-64, 1991
Interleukin-6 and tumour necrosis factor alpha are expressed by keratinocytes but not by Langerhans cells ~
ANNEMETTE OXHOLM'. 3, MARCUS DIAMANT4, PETER OXHOLM'. and KLAUS BENDTZEN4 'Department of Dermatology and *Division of Oral Pathology, Department of Stomatology, University of California, San Francisco, USA, 'Bartholin Institute, Kommunehospitalet and 4Laboratory of Medical Immunology TTA, Righospitalet, University Hospital, Copenhagen, Denmark
Oxholm, A., Diamant, M., Oxholm, P. & Bendtzen, K. Interleukin-6 and tumour necrosis factor alpha are expressed by keratinocytes but not by Langerhans cells. APM IS Y9; 58-64, 199 1. The presence of human cytokines was examined in parallel skin biopsies and epidermal single cell preparations obtained from normal individuals. Using biotin-avidin-peroxidaseand immunofluorescence techniques and antibodies against recombinant cytokines, a granular intercellular/membrane-associated staining for interleukin-6 (IL-6) and tumour necrosis factor alpha (TNF alpha), but not IL- 1 alpha or beta, was observed. An epidermal cytoplasmic staining pattern was also detected, which was most pronounced using the anti-rIL-6 antiserum. In the epidermal single cell preprations, membrane-associated staining was detected for both IL-6 and T N F alpha. Double staining revealed that CDl-positive Langerhans cells (LC) failed to express any of the examined cytokines. In vitro binding of rIL-6 or rTNF alpha to skin sections and epidermal single cell preparations indicated that the cell surface-associated IL-6 and T N F alpha originally demonstrated on keratinocytes were truly membrane-bound. Finally, co-cultivation of epidermal cells with an IL-6 responsive cell line, B9, and testing of epidermal cell supernatants in this assay, indicated that the in vivo membrane-bound IL-6 had biological activity. Key words: Cytokines; keratinocyte membrane; Langerhans cell; IL-6-activity; T N F alpha. Annemette Oxholm, Department of Dermatology, Bispebjerg Hospital, Bispebjerg Bakke, DK-2400 Copenhagen NV, Denmark.
In v i m studies have shown that keratinocytes produce and secrete different cytokines, such as interleukin (1L)-1 alpha/beta, tumour necrosis factor (TNF) alpha, IL-6, interferon gamma and granulocyte-macrophage colony-stimulating factors, and possess receptors for and respond to these molecules (8, 9, 24). The biological significance and the potential role of keratinocyte cytokines in cutaneous immunity have been the subject of several recent reviews ( 10, 14-16).The location of these cytokines in vivo has recently been elucidated by immunohistology using anti-cytokine antibodies, and by in situ hybridization techniques (2, 4, 17-21). We previously demonstrated the presence of Received March 26, 1990. Accepted June 25, 1990. 58
IL-6-like molecules and TNF alpha, but not IL- 1 alpha/beta, in normal and psoriatic epidermis (17-20). In normal skin, the cytokines were distributed in the upper epidermal layers. In UV-irradiated skin, however, the staining intensity and distribution of these cytokines were increased, and in lesional psoriatic skin, the staining of IL-6 was more pronounced and extended throughout the epidermis. Furthermore, we detected selective modulation of epidermal cytokine expression after topical administration of a vitamin D3 analogue (20). Grussman et al. (4) substantiated these findings by detecting high levels of IL-6 expressed in psoriatic skin using a similar technique. In addition, they observed membrane-bound IL-6 on cultured keratinocytes. Recently, Didierjean el al. (2) and Rornero et al. (21) were successful in demonstrating IL- I beta and IL- 1 alpha in normal and psoriatic human epidermis and the distribu-
OXHOLM ef a/
tion pattern was analogous to that of IL-6 and TNF alpha (19, 20). To discriminate between cytokines located to the intercellular epidermal compartment and to cell membranes, we have extended our previous investigations of skin sections from normal healthy individuals. Parallel in vivo and in vitro immunohistologic investigations of single cell preparations and skin sections were performed, along with double staining techniques, to examine whether keratinocytes and/or Langerhans cells (LC) express these cytokines. Finally, the bioactivity of IL-6 bound to membranes of epidermal cells was investigated. MATERIALS AND METHODS Skin Sources Three healthy and unmedicated individuals, one female (35 years) and two males (35 and 41 years) had biopsies taken and suction blisters induced at least three times at intervals of three weeks or longer. Skin Samples Punch biopsies (4 mm) of normal skin from the buttocks were obtained using infiltration anesthesia with 2% lidocaine in a ring around the biopsy site. Suction blisters were induced concurrently with the biopsies. One suction cup of acrylic plastic with a 50 mm hole was placed on the upper buttock. A negative pressure of 350 mm Hg induced a blister after 75 - 90 min. The epidermal sheet was removed using sterile forceps and scissors and immediately transferred to 10 ml of 0,l M phosphate-buffered saline (PBS) (pH 7.2). Trypsination of Epidermal Sheets The epidermal sheets were washed in PBS and treated for 45 min at 37 "C with 2 ml 0.25% trypsin (Sigma Chem. Co., St. Louis, Mo., U.S.A.). The reaction was stopped by adding 2 ml fetal calf serum (FCS). After removing cell debris, the cell suspension was washed three times in Hanks' balanced salt solution (HBSS) containing 10% FCS (13). Using this procedure, about 1-1.5 x lo6 epidermal cells from each sheet were obtained; the cell viability determined by trypan blue exclusion exceeded 90% in all cases. Epidermal Cell Cultures Separated epidermal cells were diluted to 1 x lo6 viable cell/ml in tubes containing RPMI 1640 (with penicillin/streptomycin and L-glutamine) and maintained at 37 "C in a 5% C 0 2 atmosphere. Supernatants were decanted for testing of IL-6 activity after incubation for 24 h and 72 h, respectively. Cells for cytokine staining were collected at 0 h, 24 h and 72 h. They were mounted on glass slides, air dried for 18 h at 20 "C and
then frozen at -80 "C. In initial experiments, Actinomycin D (Sigma) ( 1 Fg/ml, 1 h, 37 "C) added to the cells (to stop protein synthesis) immediately after separation and before air dryingdid not alter the specific staining for cytokines.
Cytokines Human recombinant IL-I alpha (10' U/mg) and human rIL- 1 beta( lo7U/mg) were kindly donated by Dr S. Gillis (Immunex Corp., Seattle, WA, USA) and human rIL-6 ( lo9U/mg) by Dr Hirano (Osaka, Japan). Human rTNFalpha (4x lo7U/mg) was kindly donated by Dr G. R . Adolph (Boehringer, Vienna, Austria). In each case, the cytokines were tested for bioactivity using mouse thymocytes (rIL-1 alpha and rIL-I beta), B9 hybridoma cells (rIL-6) (12), and L-M fibroblasts (rTNF alpha). Bioactivities of the cytokines were compared with corresponding international interim reference preparations (Natl Inst Biol Standard Control, London, UK). Production of Specific Polyclonal Antisera to Human Monokines The antisera to human rIL- 1 alpha, rIL- 1 beta, rIL-6, and rTNF alpha were generated by repeated immunizations of high-responder rabbits (Dakopatts, Glostrup, Denmark) with 5- I0 pg of the purified cytokines (5) and tested for reactivity with their respective cytokines by Ouchterlony double diffusion technique, ELISA and immunoblotting. Biochemical and biological neutralization tests using several different human cytokines confirmed the monospecificity of these antibodies. The rabbit antiserum against crude supernatants of Staph. albus-activated human blood monocytes was produced by Dr C. A. Dinarello (3). This antiserum was purified by sequential absorption with antigens from leukocyte supernatants obtained after 30 min of incubation with serum and staphylococci, with leukocyte supernatantsobtained after 18 h of incubation in 2.5 pg/ml of cycloheximide,which inhibits the synthesis of various monokines, and with fresh human AB serum. This absorbed antiserum (anti-MK) reacts with both major formsof IL-1 (IL-I alphaandbeta)and,in addition, with IL-6 but not with IL-2, interferon gamma, TNF alpha or TNF beta (Bendtzen, data not published). Biotin-Avidin Techniquefor Demonstration of Cytokines Bound to Tissue Sections and Single Cell Preparations Skin sections, 4-6 pm, were cut in a cryostat. The samples were air dried for 5 min. Both sections and single cells mounted on slides were fixed in acetone for 1 min. The slides were placed in PBS in a humidity chamber for 20 min with 20% normal swine (Dakopatts, Copenhagen, Denmark) or 10% normal goat serum (Vektor Laboratories, Burlingame, CA, USA). Excess serum was removed, and the slides were then treated for 30 min (20 "C)with antiseraat two-folddilutionsfrom 1 :40to 1:640. The specimens were then washed in PBS (3 x 5 min) and incubated for 30 min (20 "C) with biotinylated swine 59
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
(1:300 in PBS) (Dakopatts) or goat anti-rabbit immunoglobulin (Ig) ( 1 :400 in PBS) (Vector). After washing in PBS (3 x 5 rnin), incubation was performed with ABComplex (Dakopatts or Vector) for 30 rnin (20 'C) and then washed as mentioned above. Specimens were then incubated with 0.038% 3-amino-9-ethylcarbaol (Sigma) in acetate buffer (pH 5) and 0.014% hydrogen peroxide for 5 min, washed in water for 10 min, counterstained with hematoxylin for 2 1/2 rnin and, finally, washed in water for 10 min. The samples were mounted with Gurraquamount (BDH chemicals, Poole, Dorset, UK) or Crystal mount (Biomeda, Foster City, CA, USA).
tone-fixed cryosections and epidermal single cell preparations. In step 1, incubation with either recombinant cytokines (0.1 pg rIL-6 (= IO'U) or 1 pg rTNF alpha (= 4 x 104U)per specimen) or PBS was performed for 1 h at 20 "C. In steps 2 , 5 and 7, washing was carried out (3 x 5 min) in PBS. In step 3, samples were incubated for 20 rnin at 20 "C with normal goat serum (Vector) and excess serum was removed thereafter. In steps 4 and 6, samples were incubated with anti-IL-6 (1: IOO), antiTNF alpha (1 :IOO), preimmune serum ( I :100) or PBS for 1 h at 20 "C. In step 8, samples were identically incubated with the ABComplex (Vector) and further processed until final mounting, as described above.
Indirect Immunofluorescence (IF)for Demonstration of Membrane-bound Cytokines in Single Cell Preparaiions After fixation in acetone for 1 min, the specimens were washed in PBS and treated with anti-cytokine antibody as above. The specimens were then washed in PBS (3 x 5 min), incubated for 30 min in a humidity chamber with diluted ( 1 :20) fluorescein isothiocyanate (FIT)-conjugated swine Ig against rabbit Ig, washed again in PBS (3 x 5 min) and mounted with Gel mount (Biomeda).
Bioassay for IL-6 Activity IL-6 activity was determined using the ILd-dependent mouse hybridoma cell line B 13.29 clone B9, as decribed (12). The B9 cells were maintained in RPMI 1640 with 25 x lo-' M Hepes Buffer (Gibco Biocult, Paisly, UK) containing 400 IU/ml of penicillin, 400 g/ml of streptomycin (Gibco Biocult), 2 x 10.' M of L-glutamin (Sigma), 5 x lo" M of 2-mercapto-ethanol and 5% heat-inactivated fetal calf serum (Gibco, Biocult) plus 1 ng/ml of human rIL-6. Before assay, B9 cells were washed three times in IL-6 free medium. The cells, 5 x 103/200pl, were cultured with serial dilutions of the test sample. After 68 h, 0.5 pCi/ml of 'H-thymidine (Radiochemical Centre, Amersham, UK) was added. Five hours later, the cells were harvested on glass fiber filters using a semiautomatic cell harvester (Skatron, Lierbyen, Norway) and the 'H-thymidine incorporation into DNA measured in a liquid scintillation counter (Tri-Carb, Pacard Instruments Co., Rockville, MD, US). All assays were performed in triplicate. One U/ml of IL-6 activity was defined as the concentration giving half-maximal 'H-thymidine incorporations. rIL-6 was used as an internal standard in all assays.
Double Staining for CDI (T6)-Positive LC and Epidermal Membrane bound Cytokines Double labelling was performed by applying an indirect IF staining for the CDl antigen and the biotin-avidin-peroxidase staining for the cytokin antigens. All specimens were air dried for 5 min and fixed in acetone for 1 min. The specimens were then incubated in a humidity chamber for 20 rnin with 10% normal goat (Vector)or 20% swine (Dakopatts) serum in PBS. Excess serum was removed, and the specimens incubated for 1 h with rabbit anti-cytokine antiserum (1:80). Hereafter, the specimens were washed (3 x 5 min) in PBS, incubated with mouse monoclonal antibody against CD1 (OKT6) (1: 100) (Ortho Diagnostic Systems, Raritan, New Jersey, USA), washed again (3 x 5 min) in PBS, incubated for 30 min at 20 "C with a goat (Dakopatts) or horse (Vector) FIX-labelled anti-mouse Ig (1:20), washed 3 x 5 rnin in PBS and, finally, incubated for 30 min at 20 "C with biotinylated swine ( 1:400) (Dakopatts) or goat (1:300) (Vector) anti-rabbit Ig diluted in PBS. After washing (3 x 5 rnin), the specimens were processed with the ABComplex as described above. Fluorescence Microscopy The slides were read in a Reichert Polyvar Microscope (Reichert, Vienna, Austria), equipped with a 200 W high pressure mercury lamp, and an incident illumination system with interference primary filters, beam splitters and secondary filters adapted for F I T (Reichert). In vitro Binding of rIL-6 or rTNF Alpha to Epidermal Cells The schedule for the in vitro binding experiments and their controls is given in Table 1. Parallel experiments (I-IV) were performed with ace60
RESULTS
Staining,for Cytokines and CDI (T6)-Positive LC in Skin Sections and Single Cell Preparations Sections:In the upper epidermal layers (stratum granulosum and spinosum) similar staining was obtained with the anti-rIL-6, anti-rTNF alpha and anti-MK antisera. The staining was located intercellularly as an irregular granular pattern, in some areas associated with membranes of single cells, in others with groups of cells. Using the anti-rTNF alpha and anti-MK antisera, it was difficult to discriminate the epidermal cytoplasmic staining from the unspecific staining obtained with preimmune serum. However, the antiserum to rIL-6 caused a characteristic cytoplasmic staining pattern observed in scattered groups of cells in the
OXHOLM et a/
stratum spinosum and granulosum (Fig. 1). Maximum dilutions gving rise to specific staining were for the anti-rTNF antiserum in the range 1:40 190, for the anti-MK antiserum 1:160 - 1:320, and for the anti-rIL6 antiserum 1:160 - 1:640 with the 1 :80 dilution giving good discrimination from the preimmune serum in all cases. There were also differences in the dermal staining by the different anti-cytokine antisera. While there was no unspecific staining of intercellular connective tissue structures using anti r-IL-6 antiserum, a prominent specific staining of fibroblasts and endothelial cells was detected. The anti-TNF alpha and anti-MK antisera caused a diffuse unspecific staining of the connective tissue which made interpretation of any specific staining of dermal structures difficult. There was no specific epidermal or dermal staining by the anti-IL-1 alpha or -beta antisera. Single cell preparations: IF was superior to the immunoperoxidase technique in discriminating specific from unspecific staining. However, the sensitivity of the IF technique was lower compared with the immunoperoxidase technique. In both techniques the primary layer antibody could be diluted to 1 :160 - 1:640, with the dilutions 1:80 1 : 160 being optimal. Both techniques revealed membrane fluorescence/staining of IL-6 and TNF alpha in 30 - 50%of the cells (Fig. 2). Again, there was no staining for IL- 1 alpha or beta. Staining was not only confined to cells with squamous morphology but also to small basaloid cells. Membrane-related staining was apparent at all times (0 h, 24 h and 72 h). Double staining experiments: LC remained negative both in single cell preparations and in sections when stained with all the anti-cytokine antisera (Fig. 3).
Specificity Controls Staining was abolished in experiments where specimens were treated with antibodies absorbed by preincubating the anti-rTNF alpha antiserum with rTNF alpha (4 x 104Uper 0,5 pl antiserum) and the IL-6 and the anti-MK antisera with rIL-6 (lOsU per 0,5 pl antiserum) at 20 "C for 1 h followed by incubation overnight at 4 "C. The specific staining was lost if either the primary layer (the anti-rIL-6, -rTNF alpha or anti-MK antiserum) or the secondary layer antibody was omitted, or if the primary antibody was replaced by Tris buffer, normal rabbit serum (Dakopatts) or
preimmune serum. Autofluorescence was evaluated and found negligible in specimens without added FIT-conjugated antibody.
Fig. I. Interleukin-6 demonstrated in skin section by biotin-avidin-peroxidasetechnique, using anti-rIL-6 antibody (original x 800). Fig. 2. Membrane-bound interleukin-6 on separated epidermal cells demonstrated by indirect IF technique using anti-rIL-6 anti-serum (original x 800). Fig. 3. Double staining of CD I -positive Langerhans cells (LC), demonstrated by indirect immunofluorescence, and interleukin-6 (IL-6) expressed on the membranes of separated epidermal cells, demonstrated by biotin-avidin-peroxidase technique. LC do not stain for IL-6 (onginal x 800). 61
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
TABLE 1. Stepwise procedure for demonstrating in vitro binding of cytokines to epidermal single cells and to skin
sect ions Steps
Experiment 1
2
3
4
5
6
7
8
I
recomb. cytokine
wash
normal goat serum
anticytokine antiserum
wash
anticytokine antiserum
wash
ABcompl.
I1
recomb. cytokine
wash
normal goat serum
anticytokine antiserum
wash
PBS
wash
ABcompl.
111
PBS
wash
normal goat serum
anticytokine antiserum
wash
PBS
wash
ABcompl.
PBS
wash
normal goat serum
preimmune serum
wash
PBS
wash
ABcompl.
~~
IV
__________________
Schedule for the four parallel experiments examining the in vitro ability of IL-6 and TNF alpha to bind either skin sections or epidermal single cell preparations.
In vitro Binding for IL-6 and TNF Alpha In experiments I and 111, the specific staining for IL-6 and TNF alpha (Table 1) was demonstrated in both skin sections and in epidermal single cell preparations. Specific staining patterns for IL-6 and TNF alpha were, however, lost in experiments I1 and IV. IL-9 Bioactivity of Cell-Associated but not Supernatant Materials Supernatants decanted at 0 h, 24 h and 72 h, tested in serial dilutions, did not reveal measurable IL-6 acitivity when tested in the B9 cell assay. However, epidermal cells, when co-cultivated with B9 cells, stimulated ’H-thymidine incorporation in these cells. A typical experiment is illustrated in Table 2.
DISCUSSION In previous studies (1 7-20) we described in vivo expression of “IL-6-like molecules” and TNF alpha in the epidermis. We were, however, only able to give indirect evidence for the presence of IL-6. In the present investigation we used a specific antiserum to rIL-6, and this gave a staining pattern identical to that seen using the anti-MK antiserum, although in this case the cytoplasmic staining was more pronounced. The epidermal location of TNF alpha and IL-6 is similar to that described for IL- 1 alpha and IL- 1 beta by Didierjean et al. ( 2 ) and Romero et al. (21). They demonstrated immunohistological staining in the upper epidermal layers, and the pattern was primarily membranous or intercellular. In contrast to these investigators,
TABLE 2. IL-6 bioactivity of epidermal cells tested in the B9 cell assay Culture conditions/well (200 ul)
’H-Thymidine incorporation mean cpm (range)
lo5 Viab. epiderm. cells lo5 Viab. epiderm. cells + 5 x lo3 B9 cells 5 x lo4 Viab. epiderm. cells 5 x lo4 Viab. epiderm. cells + 5 x lo3 B9 cells 5 x lo3 B9 cells 5 x lo3 B9 cells + 20U/ml IL-6
307.3 (256-380) 5094.2 (4016-5652) 136.7 (104-194) 4399.3 (3578-4880) 2686.0 (1408-4356) 26835.3 (25828-27 172)
Illustration of one experiment made in triplicate. Co-cultivation of trypsin-separated epidermal cells with B9 cells in the IL-6 biological assay. As controls were included epidermal cells without B9 cells and B9 cells without epidermal cells; c.p.m. = counts per minute. viab. epiderm. = viable epidermal cells. The experiment was carried out in triplicate and the mean values given.
62
OXHOLM el a1
we have not been able to visualize epidermal IL- 1 alpha or beta. An explanation might be that high concentrations of antibody were required, and staining was not greatly above background (21). If Romero et al. (2 1) omitted methanol, which resulted in higher background staining, IL- 1 beta activity could not be distinguished. In our hands, however, methanol treatment abolished staining. An additional explanation could be different affinities of the antibodies used in their studies and in the present study. The expression of IL-6 in lesional psoriatic skin has recently been described also by Grossman et al. (4).In accordance with our studies ( 18,20),they found IL-6 primarily located in the cytoplasmic areas and, to a lesser degree, associated with membrane. They also found that IL-6 was expressed in all epidermal layers. However, the authors gave no description of the staining of normal skin (4). Both IL- 1 and IL-6 activate cells of the myeloid and monocytic lineages, and the former appears to be a potent stimulator of LC functions (24). LC have been shown to produce IL-I-like activities (22), but evidence for production of other cytokines by this dendritic cell has not been published. Even so, Didierjean et al. ( 2 ) failed to locate IL- 1 alpha or beta bound to LC in skin sections from normal skin. In our study, LC failed to express cell-associatedTNF alpha and IL-6 both in normal skin and after isolation by trypsinization. This was in sharp contrast to the keratinocytes. One could speculate that LC need induction, for example by membrane-bound cytokines on adjacent keratinocytes, to produce these cytokines. The studies on single cell preparations allow us to focus on membrane-bound cytokineson freshly isolated epidermal cells. However, direct comparison with the staining of skin sectionsis impossible, because single cells were handled differently. It appeared that the keratinocytes, and not the LC, expressed IL-6 and TNF alpha. The binding experiments indicate that IL-6 and TNF alpha, expressed in vivo both on single cells and in skin sections, are firmly bound to the keratinocyte membranes. This is underscored by the fact that IL-6 and TNF alpha, after addition in vitro, were only loosely bound (cytokine-anti-cytokine antibody complexes were removed during washing in step 5 ) while expression of the endogenous and membrane-associated cytokines was unaffected even by repeated washing procedure. The demonstration of IL-6 firmly associated
with the membranes of keratinocytes is in agreement with recent observations by Kupper et a/. (1 I). They found that digitonin-prepared membranes contained significant amounts of material which was antigenically similar to IL-6. However, these membranes did not exert IL-6 bioactivity (1 1). In contrast to Kirnbauer et al., who detected IL-6 release from freshly isolated epidermal cells (7), we were not able to detect IL-6 activity in epidermal cell supernatants. This could be explained by differences in the epidermal cell preparation techniques and/or the handling of the supernatants. For example, the trypsin treatment might have digested IL-6 produced by the cells. However, the staining for IL-6 on the epidermal cell membranes was readily demonstrated after trypsin treatment. Furthermore, trypsin-separated epidermal cells were capable of stimulating B9 cells, most likely because of bioactive IL-6 in or on their membranes. The results from the present study confirm and substantiate our previous observations of epidermal in vivo expression of IL-6 and TNF alpha. Epidermis consists of tightly adherent sessile cells, and a possible biological activity of membranebound cytokine on one of these cell types could therefore be of both physiological and pathophysiological significance. The combination of morphological and biological assays may help to clarify inflammatory and proliferative processes in the skin. We thank Jette Pedersen for excellent technical assistance. Funding was provided by The Spies Foundation, The Danish National Association Against Rheumatic Diseases, Auge Bangs Foundation and The Danish Medical Research Council.
REFERENCES I . Beutler, B., Milsark, I. W.& Cerami, A.: Cachectin/ tumor necrosis factor: Production,distribution, and metabolic fate in vivo. J. Immunol. 135: 3972-3977, 1985. 2. Didierjean, L., Salomon, D., Merot, Y., Siegenthaler, G., Shaw, A., Dayer, J. M. & Saurat, J. H.: Localization and characterization of the interleukin 1 immunoreactive pool (IL-1 alpha and beta forms) in normal human epidermis. J. Invest. Dermatol. 92: 809-8 16, 1989. 3. Dinarello, C. A , , Renfer, L. & Wolx S. M.: Human
63
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
leukocytic pyrogen: Purification and development of a radioimmunoassay. Roc. Natl. Acad. Sci. USA 74: 4624-4627, 1977. 4. Grossman, R. M., Krueger, J., Yourish, D., GranelliPiperno, A . , Murphy, D. P., May, L. T., Kupper, T. S., Sehgal, P. B. & Gottlieb, A. B.: Interleukin 6 is expressed in high levels in psoriatic skin and stimulates proliferation of cultured human keratinocytes. Roc. Natl. Acad. Sci. USA 86: 6367-6371, 1989. 5. Harboe, N . M. G. & Ingild, A.: Immunization, isolation of immunoglobulins and antibody titre determination. Scand. J. Immunol. 17, Suppl. 10: 345-351, 1983. 6 . Hauser, C., Saurat, J. H., Schmitt, A., Jaunin, F. & Dayer, J. M.: Interleukin 1 is present in normal human epidermis. J. Immunol. 136: 331 7-3323, 1986. 7. Kirnbauer, R., Kock, A . , Schwartz, T., Urbanski, A . , Krutmann, J., Borth, W., Damm, D . , Shipley, G . , Ansel, J. C. & Luger, T.: IFN-beta 2, B cell differentiation factor 2, or hybridoma growth factor (IL-6)is expressed and released by human epidermal cells and epidermoid carcinoma cell lines. J. Immunol. 142: 1922-1928, 1989. 8. Kupper, T. S., Dower, S., Birchall, N . , Clark, S. & Lee, F.: Interleukin 1 binds to specific high affinity receptors on human keratinocytes and induces granulocyte macrophage colony stimulating factor mRNA and protein: A potential autocrine role for IL- 1 in epidermis. J. Clin. Invest. 82: 1787-1792, 1988. 9. Kupper, T. S.: Production of cytokines by epithelial tissues. A new model for cutaneous inflammation. Am. J. Dermatopathol. I f : 69-73, 1989. 10. Kupper, T. S.: Mechanisms of cutaneous inflammation. Interactions between epidermal cytokines, adhesion molecules, and leukocytes. Arch. Dermatol. 125: 1406-1412, 1989. 11. Kupper, T. S., Min, K., Sehgal, P., Mizutani, H . , Birchall, N., Ray, A. & May, L.: Production of IL-6 by keratinocytes. Implications for epidermal inflammation and immunity. Ann. N.Y. Acad. Sci. 557: 454-464, 1989. 12. Landsdorp, P. M., Aarden, L. A., Calafat, J. & Zeiljemaker, W . P.: A growth factor-dependent B-cell hybridoma. Cum, Top. Microbiol. Immunol.
64
132: 105-113, 1986. 13. Larsen, C. G., Ternowitz. T.,Larsen, I;. & ThestrupPedersen, K.: Epidermis and lymphocyte interactions during an allergic patch test reaction. Increased activity of ETAF/IL- 1, epidermal derived lymphocyte chemotactic factor and mixed skin lymphocyte reactivity in persons with type IV allergy. J. Invest. Dermatol. 90: 230-233, 1988. 14. Mizel, S. B.: The interleukins, FASEB J. 3: 23792388, 1989. 15. O’Garra,A.: Interleukinsand the immune system 1. Lancet i: 943-946, 1989. 16. O’Garra,A.: Interleukinsand the immune system 2. Lancet i: 1003-1005, 1989. 17. Oxholm, A , . Oxholm, P., Staberg, B. & Bendtzen, K.: Immunohistological detection of interleukin 1-like molecules and tumour necrosis factor in human epidermis before and after UVB-irradiation in vivo. Br. J. Dermatol. 118: 369-376, 1988. 18. Oxholm, A , , Oxholm, P., Staberg, B. & Bendtzen, K.: Interleukin-6 in the epidermis of patients with psoriasis before and during PUVA treatment. Acta Derm. Venereol. (Stock.) 69: 195-199, 1989. 19. Oxholm, A., Oxholm, P.,Permin, H. & Bendtzen, K.: Epidermal tumour necrosis factor alpha and interleukin 6-like activities in AIDS-related Kaposi’s sarcoma. APMIS 97: 533-538, 1989. 20. Oxholm, A., Oxholm, P., Staberg, B. & Bendtzen, K.: Expression of interleukin-6-like molecules and tumour necrosis factor after topical treatment of psoriasis with a new vitamin D analogue (MC 903). Acta Derm. Venereol. (Stock.) 69: 385-390, 1989. 21. Romero, L . I., Ikejima, T. & Pincus, S. H.: In situ localization of interleukin- 1 in normal and psoriatic skin. J. Invest. Dermatol. 93: 5 18-522, 1989. 22. Sauder, D. N., Dinarello, C. A. & Morhenn, V.: Langerhans cell production of interleukin-1. J. Invest. Dermatol. 82: 695-607, 1984. 23. Sauder, D. N . , Wong, D., Mckenzie, R . et al.: The pluripotent keratinocyte: Molecular characterization of epidermal cytokines. Clin. Res. 36: 692 A, 1988. 24. Steinman, R. M.:Cytokines amplify the function of accessory cells. Immunol. Let. 17: 197-202, 1988.
Interleukin-6 and tumour necrosis factor alpha are expressed by keratinocytes but not by Langerhans cells ~
ANNEMETTE OXHOLM'. 3, MARCUS DIAMANT4, PETER OXHOLM'. and KLAUS BENDTZEN4 'Department of Dermatology and *Division of Oral Pathology, Department of Stomatology, University of California, San Francisco, USA, 'Bartholin Institute, Kommunehospitalet and 4Laboratory of Medical Immunology TTA, Righospitalet, University Hospital, Copenhagen, Denmark
Oxholm, A., Diamant, M., Oxholm, P. & Bendtzen, K. Interleukin-6 and tumour necrosis factor alpha are expressed by keratinocytes but not by Langerhans cells. APM IS Y9; 58-64, 199 1. The presence of human cytokines was examined in parallel skin biopsies and epidermal single cell preparations obtained from normal individuals. Using biotin-avidin-peroxidaseand immunofluorescence techniques and antibodies against recombinant cytokines, a granular intercellular/membrane-associated staining for interleukin-6 (IL-6) and tumour necrosis factor alpha (TNF alpha), but not IL- 1 alpha or beta, was observed. An epidermal cytoplasmic staining pattern was also detected, which was most pronounced using the anti-rIL-6 antiserum. In the epidermal single cell preprations, membrane-associated staining was detected for both IL-6 and T N F alpha. Double staining revealed that CDl-positive Langerhans cells (LC) failed to express any of the examined cytokines. In vitro binding of rIL-6 or rTNF alpha to skin sections and epidermal single cell preparations indicated that the cell surface-associated IL-6 and T N F alpha originally demonstrated on keratinocytes were truly membrane-bound. Finally, co-cultivation of epidermal cells with an IL-6 responsive cell line, B9, and testing of epidermal cell supernatants in this assay, indicated that the in vivo membrane-bound IL-6 had biological activity. Key words: Cytokines; keratinocyte membrane; Langerhans cell; IL-6-activity; T N F alpha. Annemette Oxholm, Department of Dermatology, Bispebjerg Hospital, Bispebjerg Bakke, DK-2400 Copenhagen NV, Denmark.
In v i m studies have shown that keratinocytes produce and secrete different cytokines, such as interleukin (1L)-1 alpha/beta, tumour necrosis factor (TNF) alpha, IL-6, interferon gamma and granulocyte-macrophage colony-stimulating factors, and possess receptors for and respond to these molecules (8, 9, 24). The biological significance and the potential role of keratinocyte cytokines in cutaneous immunity have been the subject of several recent reviews ( 10, 14-16).The location of these cytokines in vivo has recently been elucidated by immunohistology using anti-cytokine antibodies, and by in situ hybridization techniques (2, 4, 17-21). We previously demonstrated the presence of Received March 26, 1990. Accepted June 25, 1990. 58
IL-6-like molecules and TNF alpha, but not IL- 1 alpha/beta, in normal and psoriatic epidermis (17-20). In normal skin, the cytokines were distributed in the upper epidermal layers. In UV-irradiated skin, however, the staining intensity and distribution of these cytokines were increased, and in lesional psoriatic skin, the staining of IL-6 was more pronounced and extended throughout the epidermis. Furthermore, we detected selective modulation of epidermal cytokine expression after topical administration of a vitamin D3 analogue (20). Grussman et al. (4) substantiated these findings by detecting high levels of IL-6 expressed in psoriatic skin using a similar technique. In addition, they observed membrane-bound IL-6 on cultured keratinocytes. Recently, Didierjean el al. (2) and Rornero et al. (21) were successful in demonstrating IL- I beta and IL- 1 alpha in normal and psoriatic human epidermis and the distribu-
OXHOLM ef a/
tion pattern was analogous to that of IL-6 and TNF alpha (19, 20). To discriminate between cytokines located to the intercellular epidermal compartment and to cell membranes, we have extended our previous investigations of skin sections from normal healthy individuals. Parallel in vivo and in vitro immunohistologic investigations of single cell preparations and skin sections were performed, along with double staining techniques, to examine whether keratinocytes and/or Langerhans cells (LC) express these cytokines. Finally, the bioactivity of IL-6 bound to membranes of epidermal cells was investigated. MATERIALS AND METHODS Skin Sources Three healthy and unmedicated individuals, one female (35 years) and two males (35 and 41 years) had biopsies taken and suction blisters induced at least three times at intervals of three weeks or longer. Skin Samples Punch biopsies (4 mm) of normal skin from the buttocks were obtained using infiltration anesthesia with 2% lidocaine in a ring around the biopsy site. Suction blisters were induced concurrently with the biopsies. One suction cup of acrylic plastic with a 50 mm hole was placed on the upper buttock. A negative pressure of 350 mm Hg induced a blister after 75 - 90 min. The epidermal sheet was removed using sterile forceps and scissors and immediately transferred to 10 ml of 0,l M phosphate-buffered saline (PBS) (pH 7.2). Trypsination of Epidermal Sheets The epidermal sheets were washed in PBS and treated for 45 min at 37 "C with 2 ml 0.25% trypsin (Sigma Chem. Co., St. Louis, Mo., U.S.A.). The reaction was stopped by adding 2 ml fetal calf serum (FCS). After removing cell debris, the cell suspension was washed three times in Hanks' balanced salt solution (HBSS) containing 10% FCS (13). Using this procedure, about 1-1.5 x lo6 epidermal cells from each sheet were obtained; the cell viability determined by trypan blue exclusion exceeded 90% in all cases. Epidermal Cell Cultures Separated epidermal cells were diluted to 1 x lo6 viable cell/ml in tubes containing RPMI 1640 (with penicillin/streptomycin and L-glutamine) and maintained at 37 "C in a 5% C 0 2 atmosphere. Supernatants were decanted for testing of IL-6 activity after incubation for 24 h and 72 h, respectively. Cells for cytokine staining were collected at 0 h, 24 h and 72 h. They were mounted on glass slides, air dried for 18 h at 20 "C and
then frozen at -80 "C. In initial experiments, Actinomycin D (Sigma) ( 1 Fg/ml, 1 h, 37 "C) added to the cells (to stop protein synthesis) immediately after separation and before air dryingdid not alter the specific staining for cytokines.
Cytokines Human recombinant IL-I alpha (10' U/mg) and human rIL- 1 beta( lo7U/mg) were kindly donated by Dr S. Gillis (Immunex Corp., Seattle, WA, USA) and human rIL-6 ( lo9U/mg) by Dr Hirano (Osaka, Japan). Human rTNFalpha (4x lo7U/mg) was kindly donated by Dr G. R . Adolph (Boehringer, Vienna, Austria). In each case, the cytokines were tested for bioactivity using mouse thymocytes (rIL-1 alpha and rIL-I beta), B9 hybridoma cells (rIL-6) (12), and L-M fibroblasts (rTNF alpha). Bioactivities of the cytokines were compared with corresponding international interim reference preparations (Natl Inst Biol Standard Control, London, UK). Production of Specific Polyclonal Antisera to Human Monokines The antisera to human rIL- 1 alpha, rIL- 1 beta, rIL-6, and rTNF alpha were generated by repeated immunizations of high-responder rabbits (Dakopatts, Glostrup, Denmark) with 5- I0 pg of the purified cytokines (5) and tested for reactivity with their respective cytokines by Ouchterlony double diffusion technique, ELISA and immunoblotting. Biochemical and biological neutralization tests using several different human cytokines confirmed the monospecificity of these antibodies. The rabbit antiserum against crude supernatants of Staph. albus-activated human blood monocytes was produced by Dr C. A. Dinarello (3). This antiserum was purified by sequential absorption with antigens from leukocyte supernatants obtained after 30 min of incubation with serum and staphylococci, with leukocyte supernatantsobtained after 18 h of incubation in 2.5 pg/ml of cycloheximide,which inhibits the synthesis of various monokines, and with fresh human AB serum. This absorbed antiserum (anti-MK) reacts with both major formsof IL-1 (IL-I alphaandbeta)and,in addition, with IL-6 but not with IL-2, interferon gamma, TNF alpha or TNF beta (Bendtzen, data not published). Biotin-Avidin Techniquefor Demonstration of Cytokines Bound to Tissue Sections and Single Cell Preparations Skin sections, 4-6 pm, were cut in a cryostat. The samples were air dried for 5 min. Both sections and single cells mounted on slides were fixed in acetone for 1 min. The slides were placed in PBS in a humidity chamber for 20 min with 20% normal swine (Dakopatts, Copenhagen, Denmark) or 10% normal goat serum (Vektor Laboratories, Burlingame, CA, USA). Excess serum was removed, and the slides were then treated for 30 min (20 "C)with antiseraat two-folddilutionsfrom 1 :40to 1:640. The specimens were then washed in PBS (3 x 5 min) and incubated for 30 min (20 "C) with biotinylated swine 59
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
(1:300 in PBS) (Dakopatts) or goat anti-rabbit immunoglobulin (Ig) ( 1 :400 in PBS) (Vector). After washing in PBS (3 x 5 rnin), incubation was performed with ABComplex (Dakopatts or Vector) for 30 rnin (20 'C) and then washed as mentioned above. Specimens were then incubated with 0.038% 3-amino-9-ethylcarbaol (Sigma) in acetate buffer (pH 5) and 0.014% hydrogen peroxide for 5 min, washed in water for 10 min, counterstained with hematoxylin for 2 1/2 rnin and, finally, washed in water for 10 min. The samples were mounted with Gurraquamount (BDH chemicals, Poole, Dorset, UK) or Crystal mount (Biomeda, Foster City, CA, USA).
tone-fixed cryosections and epidermal single cell preparations. In step 1, incubation with either recombinant cytokines (0.1 pg rIL-6 (= IO'U) or 1 pg rTNF alpha (= 4 x 104U)per specimen) or PBS was performed for 1 h at 20 "C. In steps 2 , 5 and 7, washing was carried out (3 x 5 min) in PBS. In step 3, samples were incubated for 20 rnin at 20 "C with normal goat serum (Vector) and excess serum was removed thereafter. In steps 4 and 6, samples were incubated with anti-IL-6 (1: IOO), antiTNF alpha (1 :IOO), preimmune serum ( I :100) or PBS for 1 h at 20 "C. In step 8, samples were identically incubated with the ABComplex (Vector) and further processed until final mounting, as described above.
Indirect Immunofluorescence (IF)for Demonstration of Membrane-bound Cytokines in Single Cell Preparaiions After fixation in acetone for 1 min, the specimens were washed in PBS and treated with anti-cytokine antibody as above. The specimens were then washed in PBS (3 x 5 min), incubated for 30 min in a humidity chamber with diluted ( 1 :20) fluorescein isothiocyanate (FIT)-conjugated swine Ig against rabbit Ig, washed again in PBS (3 x 5 min) and mounted with Gel mount (Biomeda).
Bioassay for IL-6 Activity IL-6 activity was determined using the ILd-dependent mouse hybridoma cell line B 13.29 clone B9, as decribed (12). The B9 cells were maintained in RPMI 1640 with 25 x lo-' M Hepes Buffer (Gibco Biocult, Paisly, UK) containing 400 IU/ml of penicillin, 400 g/ml of streptomycin (Gibco Biocult), 2 x 10.' M of L-glutamin (Sigma), 5 x lo" M of 2-mercapto-ethanol and 5% heat-inactivated fetal calf serum (Gibco, Biocult) plus 1 ng/ml of human rIL-6. Before assay, B9 cells were washed three times in IL-6 free medium. The cells, 5 x 103/200pl, were cultured with serial dilutions of the test sample. After 68 h, 0.5 pCi/ml of 'H-thymidine (Radiochemical Centre, Amersham, UK) was added. Five hours later, the cells were harvested on glass fiber filters using a semiautomatic cell harvester (Skatron, Lierbyen, Norway) and the 'H-thymidine incorporation into DNA measured in a liquid scintillation counter (Tri-Carb, Pacard Instruments Co., Rockville, MD, US). All assays were performed in triplicate. One U/ml of IL-6 activity was defined as the concentration giving half-maximal 'H-thymidine incorporations. rIL-6 was used as an internal standard in all assays.
Double Staining for CDI (T6)-Positive LC and Epidermal Membrane bound Cytokines Double labelling was performed by applying an indirect IF staining for the CDl antigen and the biotin-avidin-peroxidase staining for the cytokin antigens. All specimens were air dried for 5 min and fixed in acetone for 1 min. The specimens were then incubated in a humidity chamber for 20 rnin with 10% normal goat (Vector)or 20% swine (Dakopatts) serum in PBS. Excess serum was removed, and the specimens incubated for 1 h with rabbit anti-cytokine antiserum (1:80). Hereafter, the specimens were washed (3 x 5 min) in PBS, incubated with mouse monoclonal antibody against CD1 (OKT6) (1: 100) (Ortho Diagnostic Systems, Raritan, New Jersey, USA), washed again (3 x 5 min) in PBS, incubated for 30 min at 20 "C with a goat (Dakopatts) or horse (Vector) FIX-labelled anti-mouse Ig (1:20), washed 3 x 5 rnin in PBS and, finally, incubated for 30 min at 20 "C with biotinylated swine ( 1:400) (Dakopatts) or goat (1:300) (Vector) anti-rabbit Ig diluted in PBS. After washing (3 x 5 rnin), the specimens were processed with the ABComplex as described above. Fluorescence Microscopy The slides were read in a Reichert Polyvar Microscope (Reichert, Vienna, Austria), equipped with a 200 W high pressure mercury lamp, and an incident illumination system with interference primary filters, beam splitters and secondary filters adapted for F I T (Reichert). In vitro Binding of rIL-6 or rTNF Alpha to Epidermal Cells The schedule for the in vitro binding experiments and their controls is given in Table 1. Parallel experiments (I-IV) were performed with ace60
RESULTS
Staining,for Cytokines and CDI (T6)-Positive LC in Skin Sections and Single Cell Preparations Sections:In the upper epidermal layers (stratum granulosum and spinosum) similar staining was obtained with the anti-rIL-6, anti-rTNF alpha and anti-MK antisera. The staining was located intercellularly as an irregular granular pattern, in some areas associated with membranes of single cells, in others with groups of cells. Using the anti-rTNF alpha and anti-MK antisera, it was difficult to discriminate the epidermal cytoplasmic staining from the unspecific staining obtained with preimmune serum. However, the antiserum to rIL-6 caused a characteristic cytoplasmic staining pattern observed in scattered groups of cells in the
OXHOLM et a/
stratum spinosum and granulosum (Fig. 1). Maximum dilutions gving rise to specific staining were for the anti-rTNF antiserum in the range 1:40 190, for the anti-MK antiserum 1:160 - 1:320, and for the anti-rIL6 antiserum 1:160 - 1:640 with the 1 :80 dilution giving good discrimination from the preimmune serum in all cases. There were also differences in the dermal staining by the different anti-cytokine antisera. While there was no unspecific staining of intercellular connective tissue structures using anti r-IL-6 antiserum, a prominent specific staining of fibroblasts and endothelial cells was detected. The anti-TNF alpha and anti-MK antisera caused a diffuse unspecific staining of the connective tissue which made interpretation of any specific staining of dermal structures difficult. There was no specific epidermal or dermal staining by the anti-IL-1 alpha or -beta antisera. Single cell preparations: IF was superior to the immunoperoxidase technique in discriminating specific from unspecific staining. However, the sensitivity of the IF technique was lower compared with the immunoperoxidase technique. In both techniques the primary layer antibody could be diluted to 1 :160 - 1:640, with the dilutions 1:80 1 : 160 being optimal. Both techniques revealed membrane fluorescence/staining of IL-6 and TNF alpha in 30 - 50%of the cells (Fig. 2). Again, there was no staining for IL- 1 alpha or beta. Staining was not only confined to cells with squamous morphology but also to small basaloid cells. Membrane-related staining was apparent at all times (0 h, 24 h and 72 h). Double staining experiments: LC remained negative both in single cell preparations and in sections when stained with all the anti-cytokine antisera (Fig. 3).
Specificity Controls Staining was abolished in experiments where specimens were treated with antibodies absorbed by preincubating the anti-rTNF alpha antiserum with rTNF alpha (4 x 104Uper 0,5 pl antiserum) and the IL-6 and the anti-MK antisera with rIL-6 (lOsU per 0,5 pl antiserum) at 20 "C for 1 h followed by incubation overnight at 4 "C. The specific staining was lost if either the primary layer (the anti-rIL-6, -rTNF alpha or anti-MK antiserum) or the secondary layer antibody was omitted, or if the primary antibody was replaced by Tris buffer, normal rabbit serum (Dakopatts) or
preimmune serum. Autofluorescence was evaluated and found negligible in specimens without added FIT-conjugated antibody.
Fig. I. Interleukin-6 demonstrated in skin section by biotin-avidin-peroxidasetechnique, using anti-rIL-6 antibody (original x 800). Fig. 2. Membrane-bound interleukin-6 on separated epidermal cells demonstrated by indirect IF technique using anti-rIL-6 anti-serum (original x 800). Fig. 3. Double staining of CD I -positive Langerhans cells (LC), demonstrated by indirect immunofluorescence, and interleukin-6 (IL-6) expressed on the membranes of separated epidermal cells, demonstrated by biotin-avidin-peroxidase technique. LC do not stain for IL-6 (onginal x 800). 61
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
TABLE 1. Stepwise procedure for demonstrating in vitro binding of cytokines to epidermal single cells and to skin
sect ions Steps
Experiment 1
2
3
4
5
6
7
8
I
recomb. cytokine
wash
normal goat serum
anticytokine antiserum
wash
anticytokine antiserum
wash
ABcompl.
I1
recomb. cytokine
wash
normal goat serum
anticytokine antiserum
wash
PBS
wash
ABcompl.
111
PBS
wash
normal goat serum
anticytokine antiserum
wash
PBS
wash
ABcompl.
PBS
wash
normal goat serum
preimmune serum
wash
PBS
wash
ABcompl.
~~
IV
__________________
Schedule for the four parallel experiments examining the in vitro ability of IL-6 and TNF alpha to bind either skin sections or epidermal single cell preparations.
In vitro Binding for IL-6 and TNF Alpha In experiments I and 111, the specific staining for IL-6 and TNF alpha (Table 1) was demonstrated in both skin sections and in epidermal single cell preparations. Specific staining patterns for IL-6 and TNF alpha were, however, lost in experiments I1 and IV. IL-9 Bioactivity of Cell-Associated but not Supernatant Materials Supernatants decanted at 0 h, 24 h and 72 h, tested in serial dilutions, did not reveal measurable IL-6 acitivity when tested in the B9 cell assay. However, epidermal cells, when co-cultivated with B9 cells, stimulated ’H-thymidine incorporation in these cells. A typical experiment is illustrated in Table 2.
DISCUSSION In previous studies (1 7-20) we described in vivo expression of “IL-6-like molecules” and TNF alpha in the epidermis. We were, however, only able to give indirect evidence for the presence of IL-6. In the present investigation we used a specific antiserum to rIL-6, and this gave a staining pattern identical to that seen using the anti-MK antiserum, although in this case the cytoplasmic staining was more pronounced. The epidermal location of TNF alpha and IL-6 is similar to that described for IL- 1 alpha and IL- 1 beta by Didierjean et al. ( 2 ) and Romero et al. (21). They demonstrated immunohistological staining in the upper epidermal layers, and the pattern was primarily membranous or intercellular. In contrast to these investigators,
TABLE 2. IL-6 bioactivity of epidermal cells tested in the B9 cell assay Culture conditions/well (200 ul)
’H-Thymidine incorporation mean cpm (range)
lo5 Viab. epiderm. cells lo5 Viab. epiderm. cells + 5 x lo3 B9 cells 5 x lo4 Viab. epiderm. cells 5 x lo4 Viab. epiderm. cells + 5 x lo3 B9 cells 5 x lo3 B9 cells 5 x lo3 B9 cells + 20U/ml IL-6
307.3 (256-380) 5094.2 (4016-5652) 136.7 (104-194) 4399.3 (3578-4880) 2686.0 (1408-4356) 26835.3 (25828-27 172)
Illustration of one experiment made in triplicate. Co-cultivation of trypsin-separated epidermal cells with B9 cells in the IL-6 biological assay. As controls were included epidermal cells without B9 cells and B9 cells without epidermal cells; c.p.m. = counts per minute. viab. epiderm. = viable epidermal cells. The experiment was carried out in triplicate and the mean values given.
62
OXHOLM el a1
we have not been able to visualize epidermal IL- 1 alpha or beta. An explanation might be that high concentrations of antibody were required, and staining was not greatly above background (21). If Romero et al. (2 1) omitted methanol, which resulted in higher background staining, IL- 1 beta activity could not be distinguished. In our hands, however, methanol treatment abolished staining. An additional explanation could be different affinities of the antibodies used in their studies and in the present study. The expression of IL-6 in lesional psoriatic skin has recently been described also by Grossman et al. (4).In accordance with our studies ( 18,20),they found IL-6 primarily located in the cytoplasmic areas and, to a lesser degree, associated with membrane. They also found that IL-6 was expressed in all epidermal layers. However, the authors gave no description of the staining of normal skin (4). Both IL- 1 and IL-6 activate cells of the myeloid and monocytic lineages, and the former appears to be a potent stimulator of LC functions (24). LC have been shown to produce IL-I-like activities (22), but evidence for production of other cytokines by this dendritic cell has not been published. Even so, Didierjean et al. ( 2 ) failed to locate IL- 1 alpha or beta bound to LC in skin sections from normal skin. In our study, LC failed to express cell-associatedTNF alpha and IL-6 both in normal skin and after isolation by trypsinization. This was in sharp contrast to the keratinocytes. One could speculate that LC need induction, for example by membrane-bound cytokines on adjacent keratinocytes, to produce these cytokines. The studies on single cell preparations allow us to focus on membrane-bound cytokineson freshly isolated epidermal cells. However, direct comparison with the staining of skin sectionsis impossible, because single cells were handled differently. It appeared that the keratinocytes, and not the LC, expressed IL-6 and TNF alpha. The binding experiments indicate that IL-6 and TNF alpha, expressed in vivo both on single cells and in skin sections, are firmly bound to the keratinocyte membranes. This is underscored by the fact that IL-6 and TNF alpha, after addition in vitro, were only loosely bound (cytokine-anti-cytokine antibody complexes were removed during washing in step 5 ) while expression of the endogenous and membrane-associated cytokines was unaffected even by repeated washing procedure. The demonstration of IL-6 firmly associated
with the membranes of keratinocytes is in agreement with recent observations by Kupper et a/. (1 I). They found that digitonin-prepared membranes contained significant amounts of material which was antigenically similar to IL-6. However, these membranes did not exert IL-6 bioactivity (1 1). In contrast to Kirnbauer et al., who detected IL-6 release from freshly isolated epidermal cells (7), we were not able to detect IL-6 activity in epidermal cell supernatants. This could be explained by differences in the epidermal cell preparation techniques and/or the handling of the supernatants. For example, the trypsin treatment might have digested IL-6 produced by the cells. However, the staining for IL-6 on the epidermal cell membranes was readily demonstrated after trypsin treatment. Furthermore, trypsin-separated epidermal cells were capable of stimulating B9 cells, most likely because of bioactive IL-6 in or on their membranes. The results from the present study confirm and substantiate our previous observations of epidermal in vivo expression of IL-6 and TNF alpha. Epidermis consists of tightly adherent sessile cells, and a possible biological activity of membranebound cytokine on one of these cell types could therefore be of both physiological and pathophysiological significance. The combination of morphological and biological assays may help to clarify inflammatory and proliferative processes in the skin. We thank Jette Pedersen for excellent technical assistance. Funding was provided by The Spies Foundation, The Danish National Association Against Rheumatic Diseases, Auge Bangs Foundation and The Danish Medical Research Council.
REFERENCES I . Beutler, B., Milsark, I. W.& Cerami, A.: Cachectin/ tumor necrosis factor: Production,distribution, and metabolic fate in vivo. J. Immunol. 135: 3972-3977, 1985. 2. Didierjean, L., Salomon, D., Merot, Y., Siegenthaler, G., Shaw, A., Dayer, J. M. & Saurat, J. H.: Localization and characterization of the interleukin 1 immunoreactive pool (IL-1 alpha and beta forms) in normal human epidermis. J. Invest. Dermatol. 92: 809-8 16, 1989. 3. Dinarello, C. A , , Renfer, L. & Wolx S. M.: Human
63
INTERLEUKIN-6 AND TNF ALPHA IN KERATINOCYTES
leukocytic pyrogen: Purification and development of a radioimmunoassay. Roc. Natl. Acad. Sci. USA 74: 4624-4627, 1977. 4. Grossman, R. M., Krueger, J., Yourish, D., GranelliPiperno, A . , Murphy, D. P., May, L. T., Kupper, T. S., Sehgal, P. B. & Gottlieb, A. B.: Interleukin 6 is expressed in high levels in psoriatic skin and stimulates proliferation of cultured human keratinocytes. Roc. Natl. Acad. Sci. USA 86: 6367-6371, 1989. 5. Harboe, N . M. G. & Ingild, A.: Immunization, isolation of immunoglobulins and antibody titre determination. Scand. J. Immunol. 17, Suppl. 10: 345-351, 1983. 6 . Hauser, C., Saurat, J. H., Schmitt, A., Jaunin, F. & Dayer, J. M.: Interleukin 1 is present in normal human epidermis. J. Immunol. 136: 331 7-3323, 1986. 7. Kirnbauer, R., Kock, A . , Schwartz, T., Urbanski, A . , Krutmann, J., Borth, W., Damm, D . , Shipley, G . , Ansel, J. C. & Luger, T.: IFN-beta 2, B cell differentiation factor 2, or hybridoma growth factor (IL-6)is expressed and released by human epidermal cells and epidermoid carcinoma cell lines. J. Immunol. 142: 1922-1928, 1989. 8. Kupper, T. S., Dower, S., Birchall, N . , Clark, S. & Lee, F.: Interleukin 1 binds to specific high affinity receptors on human keratinocytes and induces granulocyte macrophage colony stimulating factor mRNA and protein: A potential autocrine role for IL- 1 in epidermis. J. Clin. Invest. 82: 1787-1792, 1988. 9. Kupper, T. S.: Production of cytokines by epithelial tissues. A new model for cutaneous inflammation. Am. J. Dermatopathol. I f : 69-73, 1989. 10. Kupper, T. S.: Mechanisms of cutaneous inflammation. Interactions between epidermal cytokines, adhesion molecules, and leukocytes. Arch. Dermatol. 125: 1406-1412, 1989. 11. Kupper, T. S., Min, K., Sehgal, P., Mizutani, H . , Birchall, N., Ray, A. & May, L.: Production of IL-6 by keratinocytes. Implications for epidermal inflammation and immunity. Ann. N.Y. Acad. Sci. 557: 454-464, 1989. 12. Landsdorp, P. M., Aarden, L. A., Calafat, J. & Zeiljemaker, W . P.: A growth factor-dependent B-cell hybridoma. Cum, Top. Microbiol. Immunol.
64
132: 105-113, 1986. 13. Larsen, C. G., Ternowitz. T.,Larsen, I;. & ThestrupPedersen, K.: Epidermis and lymphocyte interactions during an allergic patch test reaction. Increased activity of ETAF/IL- 1, epidermal derived lymphocyte chemotactic factor and mixed skin lymphocyte reactivity in persons with type IV allergy. J. Invest. Dermatol. 90: 230-233, 1988. 14. Mizel, S. B.: The interleukins, FASEB J. 3: 23792388, 1989. 15. O’Garra,A.: Interleukinsand the immune system 1. Lancet i: 943-946, 1989. 16. O’Garra,A.: Interleukinsand the immune system 2. Lancet i: 1003-1005, 1989. 17. Oxholm, A , . Oxholm, P., Staberg, B. & Bendtzen, K.: Immunohistological detection of interleukin 1-like molecules and tumour necrosis factor in human epidermis before and after UVB-irradiation in vivo. Br. J. Dermatol. 118: 369-376, 1988. 18. Oxholm, A , , Oxholm, P., Staberg, B. & Bendtzen, K.: Interleukin-6 in the epidermis of patients with psoriasis before and during PUVA treatment. Acta Derm. Venereol. (Stock.) 69: 195-199, 1989. 19. Oxholm, A., Oxholm, P.,Permin, H. & Bendtzen, K.: Epidermal tumour necrosis factor alpha and interleukin 6-like activities in AIDS-related Kaposi’s sarcoma. APMIS 97: 533-538, 1989. 20. Oxholm, A., Oxholm, P., Staberg, B. & Bendtzen, K.: Expression of interleukin-6-like molecules and tumour necrosis factor after topical treatment of psoriasis with a new vitamin D analogue (MC 903). Acta Derm. Venereol. (Stock.) 69: 385-390, 1989. 21. Romero, L . I., Ikejima, T. & Pincus, S. H.: In situ localization of interleukin- 1 in normal and psoriatic skin. J. Invest. Dermatol. 93: 5 18-522, 1989. 22. Sauder, D. N., Dinarello, C. A. & Morhenn, V.: Langerhans cell production of interleukin-1. J. Invest. Dermatol. 82: 695-607, 1984. 23. Sauder, D. N . , Wong, D., Mckenzie, R . et al.: The pluripotent keratinocyte: Molecular characterization of epidermal cytokines. Clin. Res. 36: 692 A, 1988. 24. Steinman, R. M.:Cytokines amplify the function of accessory cells. Immunol. Let. 17: 197-202, 1988.
