Tumor necrosis factor alpha (TNFα) in human skin: a comparison of different antibodies for immunohistochemistry

Chapter II

Tumor necrosis factor alpha (TNFα) in human skin: a comparison ofdifferent antibodies for immunohistochemistry

Namkje van der Laan 1

Lou de Leij PhD 2

Wim Buurman PhD 3

Wim Timens PhD 2

Henk Jan ten Duis MD, PhD 1

1 Department of SurgeryUniversity Hospital GroningenThe Netherlands

2 Department of Pathology Laboratory medicine, section Medical BiologyUniversity Hospital GroningenThe Netherlands

3 Department of SurgeryUniversity Hospital MaastrichtThe Netherlands

Archives of Dermatological Research (2001) 293:226-232

Chapter 2

24

Summary

Conflicting results have been reported regarding the localization andpresence of TNFα in normal human skin. To study TNFα expression, we havetested a panel of antibodies directed against human TNFα. First, antibodies weretested for immunoreactivity on cytospots of isolated, LPS stimulated peripheralblood mononuclear cells. Second, antibodies were tested to detect recombinantTNFα on Western blot. Some antibodies were found to be unable to detectrecombinant TNFα on blot. Most antibodies were able however to bind TNFαprotein, whereas they did not bind to other irrelevant proteins that were alsopresent on blot. Finally, antibodies were tested on cryosections of normal humanskin. Antibodies that not reacted with TNFα on blot were incubated with TNFαbefore the staining procedure to see whether these antibodies specifically bindTNFα.

We found that although the antibodies all bind TNFα clear differences instaining patterns are observed. This indicates that these antibodies may recognizedistinct epitopes or different forms of TNFα. Differences found here, but alsothose reported in literature can be the result of the concentration of the antibodyused, the staining procedure or specificity of the antibody itself. So, forunambiguous interpretation of data, it is important to know the characterization ofthe antibodies used.

TNFα in human skin

25

Introduction

Many in vitro experiments have been performed to investigate theproperties of different cells of the skin immune system (SIS), especiallyconcerning their ability to produce cytokines (1). Epidermal keratinocytes havebeen shown to be able to produce a number of cytokines such as interleukin 1, 6and tumor necrosis factor alpha (TNFα) when cultured in the presence of avariety of stimuli. Also other cells of the SIS are able to produce cytokines asshown in in vitro experiments (2-4). About cytokine production of cells of theskin in in vivo situations, much remains unclear. It would be of interest to knowwhich cells produce cytokines such as TNFα under different conditions, sincethis could give information on the role of TNFα in physiological circumstancesand during inflammatory or systemic disease processes. For detection of TNFαmRNA in human skin, in situ hybridization or RT-PCR experiments can beperformed. The data from such experiments does not, however, give informationabout the presence of TNFα at the protein level. Protein can be detected usingWestern blotting, but this technique has the disadvantage that localization of theprotein is unknown. Therefore immunohistochemistry is a useful technique tolocalize cells that contain TNFα protein. Several studies concerning the presenceof TNFα in human skin cryostat sections have been published. These studieshave shown conflicting results regarding the presence and localization of TNFα.Oxholm et al.(5) were the first to detect TNFα in cryosections of normal humanskin. They showed that TNFα staining was localized in cells found in the upperlayers of the epidermis, stratum granulosum, stratum spinosum and stratumcorneum. No staining was observed in the basal cell layer of the epidermis, thestratum basale. In the dermis, only hair follicles were found positive for TNFα.This same pattern has been described by different authors (5-11). In contrast,others have described positive staining only in the basal cell layer (12-14),whereas others reported the absence of TNFα staining of epidermal cells ofnormal human skin (15-20).

Positive staining in the dermal compartment is described for different cells,eccrine sweat ducts, hair follicles and sebaceous glands. To study the role ofTNFα in skin under different pathological conditions (psoriasis, lepra, UVBradiation, trauma), the presence of the cytokine TNFα in the normal ‘resting’state should be well defined. To this end, we determined the specificity andstaining pattern of a panel of different TNFα antibodies in human skin.

Chapter 2

26

Materials and methods

Antibodies

The anti-TNFα antibodies used in this study are summarized in Table 1.Antibodies MR1-2 and MR2-1 (Hbt, Uden, the Netherlands) were used to detectrespectively TNFα receptor I (p55) and II (p75). Both antibodies were used at aconcentration of 25 µg/ml.

Table 1: Anti-TNFα antibodies used in this study

Antibody, clone Subclass Source10F Mouse IgG1 Dr. W. Buurman4H31 Mouse IgG1 Dr. W. Buurman, Hbt, Uden, The Netherlands52B83 Mouse IgG1 Dr. W. Buurman, Hbt, Uden, The Netherlands61E71 Mouse IgG1 Dr. W. Buurman, Hbt, Uden, The NetherlandsD9D10 Mouse IgG1 Eurogenetics, E. Bosman, Tessenderlo, Belgium35G10F3 Mouse IgG1 Genzyme Diagnostics, Abingdon, UKIP-300 Rabbit

polyclonalGenzyme Diagnostics, Abingdon, UK

199 Mouse IgG1 Boehringer Mannheim, Roche, Almere,The Netherlands

J1D9 Mouse IgG1 Ancell, Kordia, Leiden, The Netherlands28401.111 Mouse IgG1 R&D systems, Abingdon, UK

Western blotting

Recombinant TNFα (Boehringer Ingelheim, Germany) diluted indenaturation buffer and mixed with a crude cell extract (21) as a source ofirrelevant proteins, was used to determine the specificity of the antibodies.Samples were heated for 5 minutes at 57°C and loaded on a 15% SDS-polyacrylamide gel. After electrophoresis, the proteins were transferred ontonitrocellulose membranes (Schleicher & Schuell, Keene, NH) using semi-dry blotsystem. After blocking with 4% skimmed milk powder in 0.01 M phosphatebuffered saline (PBS), antibodies were tested at different concentrations in PBScontaining 1% skimmed milk powder and 0.1% Tween-20 (Sigma Chemicals, St.Louis, MO). Different concentrations of the antibodies were tested, ranging from0.125-2 µg/ml. Antibodies were detected with rabbit anti-mouse peroxidase(RαM-PO, P260, Dako, Glostrup, Denmark) diluted 1:5000 in PBS containing1% skimmed milk powder and 0.1% Tween-20 with 5% normal human serum toprevent non specific binding of the conjugate. Antibody IP-300 was detectedwith swine anti-rabbit peroxidase, 1:10.000 (SαR-PO, P217, Dako). Bound

TNFα in human skin

27

antibody was visualized using para-hydroxy-coumaric acid (Sigma) as enhancerand sodium-luminol (Sigma) 1.25 mM in 0.1M Tris-HCl, pH 8.6, as a substrate.Rainbow colored protein molecular weight marker (RPN 756, Amersham,Buckinghamshire, UK) with a molecular weight range of 14300-220000 was usedas a reference.

Preparation of LPS stimulated cells

For use in immunohistochemical staining, peripheral blood mononuclearcells (PBMC) were isolated from heparinized peripheral human blood. Blood wasdiluted with PBS (1:1) and cells were isolated by density centrifugation onlymphoprep (Nycomed, Oslo, Norway) at 2400 rpm for 20 minutes. The PBMCfraction was washed twice in RPMI-1640 (Gibco, Grand Island, N.Y.). Cellswere stimulated with 1 µg/ml LPS (L9143, Sigma) in complete medium, RPMI-1640 supplemented with 2 mM glutamine and 15% normal human serum, for 3hours at 37°C, 5% CO2 in humidified atmosphere. After stimulation, the cellswere washed with phosphate buffered saline (PBS) and 100 µl of the cellsuspension was centrifuged (5 min, 500 rpm) on glass slides using acytocentrifuge apparatus (Shandon, Runcorn, Cheshire, UK). Also nonstimulated cells, incubated for 3 hours in complete medium without LPS werecentrifuged on glass slides. The cytospots were air dried for at least 30 minutes,fixed with acetone for 10 minutes and stored at -20°C until use.

Skin biopsies

Normal skin was obtained from healthy patients undergoing electivesurgery. Informed consent was given and this study was approved by the localethics committee. Biopsies taken at first incision were immediately snap frozen inliquid nitrogen. Skin was obtained from different parts of the body (Table 2). Allbiopsies were immediately snap frozen in liquid nitrogen and stored at -80°C untiluse.

Chapter 2

28

Table 2: Description of patients skin biopsies

Place of the body Age SexThigh 15 FemaleHip 22 MaleJust below knee 40 MaleUpperleg 48 FemaleHip 55 Male

Immunostaining of cytospins and sections

For immunohistochemistry, 6 µm sections of human skin were cut, airdried and fixed for 10 minutes with acetone.

Single step detection

This protocol was used for detection of TNFα on cytospots with alltested antibodies and for antibodies (4H31, 52B83, 35G10F3 and IP-300) thatcould be used at relatively low concentrations on human skin cryosections(respectively 2, 5, 2 µg/ml and 1:100) . First, sections were incubated for onehour with the different TNFα antibodies diluted in PBS. After washing with PBSfor 5 minutes, slides were incubated for 30 minutes with RαM-PO or SαR-PO(DAKO) for IP-300 at a respectively 1: 40 and 1:50 dilution in PBS supplementedwith 1% normal human serum. After washing with PBS, peroxidase reactivitywas visualized using amino-ethylcarbazole (AEC, Sigma) as a substrate. Slideswere counterstained with hematoxylin and mounted in Kaiser’s glycerin.

ABC

This protocol was used for antibodies (10F, 61E71, D9D10, 199, J1D9and 28401.111) that were used at a concentration of 10 or 25 µg/ml on humanskin cryosections. Sections were incubated for one hour with the different TNFαantibodies diluted in PBS. After washing with PBS for 5 minutes, slides wereincubated for 30 minutes with biotinylated goat anti-mouse conjugate (GαM-Ig(H+L)-bio, Southern Biotechnology Associates, Birmingham, Ala.) at a 1:200dilution in PBS supplemented with 1% normal human serum, washed with PBSand subsequently incubated for 30 minutes with streptavidine ABComplex (ABC,Dako). Streptavidine and biotinylated horseradishperoxidase were diluted 1:100in PBS. After washing with PBS, peroxidase reactivity was visualized usingamino-ethylcarbazole (AEC, Sigma) as a substrate. Slides were counterstainedwith hematoxylin and mounted in Kaiser’s glycerin.

TNFα in human skin

29

Immunofluorescence double staining

Antibodies IP-300 and 35G10F3 were used together with antibodiesagainst monocyte / macrophages (CD14 1:40, CD68 1:40, Dako) or endothelialcells (CD31 1:100, vWillebrand factor 1:200, Dako). Sections were incubated forone hour with a mixture of anti-TNFα and a cell specific antibody. After washingwith PBS for 5 minutes, slides were incubated for 30 minutes with a mixture of aFITC and TRITC labeled conjugate (GαM-IgG1-FITC, GαM-IgG2b-FITC,GαM-IgG1-TRITC, Southern Biotechnology Associates or SαR-FITC, SαR-TRITC, Dako) in PBS supplemented with 1% normal human serum. Afterwashing with PBS, sections were incubated for 15 minutes with DAPI(Boehringer Mannheim, 1:200.000) to stain nuclei. Sections were mounted incitifluor and analyzed for double positive cells.

Pre incubation of TNFα antibodies with recombinant TNFα

Three antibodies that were not reactive on Western blot (4H31, D9D10 and61E71) and three antibodies (4H31, 35G10F3 and IP-300) with a distinctivestaining pattern were tested for their TNFα specificity. Before the stainingprocedure, antibody was mixed with an excess (10x, 5x and equal amounts) ofrecombinant TNFα (Boehringer Ingelheim). This mixture was incubated for onehour at room temperature to allow binding of the antibody to TNFα. Then, themixture was applied to the section, for one hour, as antibody incubation.Furthermore, detection of TNFα expression was performed as described above.

Chapter 2

30

Results

Distribution of TNFα in stimulated cells

All selected antibodies stained LPS stimulated peripheral bloodmononuclear cells, whereas they were all negative when applied to unstimulatedcells. The concentrations of the antibodies used, to optimally detect TNFα instimulated cells, differed substantially between the antibodies: 10F, 4H31,35G10F3 and 28401.111 stained faintly positive at a concentration of 1 µg/ml,whereas 52B83, 61E71, 199, J1D9 stained at a concentration of 2 µg/ml andD9D10 needed 10 µg/ml. The concentration of IP-300 was not known and wasused at a dilution of 1:100. With the use of polyclonal antibody IP-300 differentcell types were found to be positive.

Specificity on Western blot

All antibodies were tested for specificity by their ability to detect recombinantTNFα added to a mixture of irrelevant human proteins (crude granulocyteextract) separated by SDS PAGE and analyzed by Western blotting. A maximumconcentration of 20 ng of recombinant TNFα was added, to be detected by theantibodies at a concentration of maximal 2 µg/ml. Several antibodies (4H31,61E71, D9D10 and 199) did not react with any protein present on Western blotunder these conditions. The other antibodies were able to detect TNFα on blot,as indicated by a positive staining of the added recombinant TNFα i.e. as a bandwith an apparent Mw of 17 kD. Three of these antibodies: 52B83, J1D9 and28401.111 show one specific band on Western blot at 17 kD (Fig.1).

Figure 1. Detection of TNFα protein on Western blotting using enhanced chemoluminescence.Lane 1+2, 5+6, 11+12, 13+14, 17+18, 19+20: 2.5 and 5 ng TNFα detected with 1 µg/ml 10F, 1µg/ml 52B83, 0.25 µg/ml 35G10F3, IP-300 1:400, 2 µg/ml J1D9 and 0.5 µg/ml 28401.111. Lane3+4, 7+8, 9+10, 15+16:10 and 20 ng TNFα detected with 2 µg/ml 4H31, 61E71, D9D10 and199.

TNFα in human skin

31

So they bind specifically to the monomeric form of TNFα protein (17kD). After longer exposure of the film 10F, 35G10F3 and IP-300 also seemed tobind to the TNFα dimer and tetramer protein, since a faint positively stainedband at 34 kD and 68 kD could be observed.

Distribution of TNFα in normal human skin

All antibodies were tested on cryosections of normal human skin. First,concentrations as used for detecting TNFα in cytospots and up to 5 µg/ml wereused to detect TNFα in human skin. With these concentrations, 10F, 61E71,199, J1D9 and 28401.111 did not stain any cell type in normal human skin,whereas 4H31, 52B83, D9D10, 35G10F3 and IP-300 showed positive staining innormal human skin. The pattern of staining of these latter antibodies wasdifferent. Also 10 µg/ml and 25 µg/ml of 10F, 61E71, 199, J1D9 and 28401.111were tested on skin, also in combination with the ABC protocol, to see whetherthese antibodies could detect TNFα in human skin at higher concentrations or adifferent protocol than were used for cytospots. Except for 199, all antibodiesstained positive on human skin using higher concentration antibody incombination with a different protocol. In Table 3 results of the detection ofTNFα using the different antibodies are summarized. Most of the antibodiesrevealed a diffuse granular staining pattern of the epidermis. 4H31 showed adifferent, very pronounced staining, only of the basal cell layer (Fig.4b).

Regarding staining of cells and structures in the dermis, most antibodiesdetected TNFα in endothelial cells, perivascular cells and cells scattered throughthe dermis.10F, D9D10, IP-300, J1D9 and 28401.111 reveal diffuse staining(Fig.2a,e. Fig .4f), whereas 52B83, 61E71 and 35G10F3 (Fig. 2b,c,d, Fig. 4d)stain more distinct cells. Different dermal structures like hair follicles, sweatglands and muscle are found positive with most tested antibodies (Table 3 andFig.3). Furthermore, no differences in immunoreactivity were seen betweennormal skin of different parts of the body. Skin was considered normal sinceonly biopsies were taken from healthy skin. Histologically the epidermal anddermal layer looked normal and contained no cellular infiltrates. The epidermallayer contains normal stratified keratinocytes and the dermal layer contains anormal pattern of connective tissue, resting endothelial cells and fibroblasts.

Chapter 2

32

Table 3: An overview of the results, using different antibodies to detect TNFα in normal humanskin.

Antibody Concen-tration

Description

10F 10 µg/ml Some epidermal cells positive; perivascular cells andscattered dermal cells diffusely positive; hair folliclespositive.

4H31 2 µg/ml Only basal cell layer of epidermis positive; no positivecells in the dermal cell layer; hair follicles and sweatglands positive.

52B83 5 µg/ml Faint granular-like staining in the epidermis, few cellsintensely positive; cells in dermis diffusely positive, withsome cells intensely positive; hair follicles, sweat glandsand muscle positive.

61E71 25 µg/ml Diffuse, granular-like staining through different layers ofepidermis, strongest in stratum basale; cells in dermisdiffusely positive, with some cells intensely positive; hairfollicles, sweat glands and muscle positive.

D9D10 10 µg/ml Diffuse, granular-like staining through different layers ofepidermis, strongest in stratum basale; diffuse positivitymainly in cells surrounding vessels; hair follicles, sweatglands positive.

35G10F3 2 µg/ml Diffuse, granular-like staining through different layers ofepidermis; strongest in stratum basale, some scatteredcells intensely positive; vascular and perivascular cells inthe dermis distinctly positive; hair follicles, sweat glandsand muscle (diffusely) positive.

IP-300 1:100 Diffuse granular-like staining through different layers ofepidermis, strongest in stratum basale; cells in thedermis, (peri)vascular and scattered, positive; hairfollicles, sweat glands and muscle positive.

199 25 µg/ml No staining observed.J1D9 10 µg/ml Diffuse, granular-like staining through different layers of

epidermis, strongest in stratum basale; cells in the dermisdiffusely positive; hair follicles faintly positive.

28401.111 25 µg/ml Epidermal and dermal cells faintly positive; hair folliclesand muscle diffusely positive.

TNFα in human skin

33

Distribution of TNF receptors p55 and p75 in normal human skin

The staining pattern of p55 in normal human skin was similar to the stainingpattern we observed with 4H31. Using anti-p55, we observed a clear staining ofthe basal cell layer of the epidermis (Fig.2f). Staining with anti-p75 antibody didnot lead to staining of cells of the epidermis and dermis.

Immunofluorescence double staining

To compare different staining patterns in more detail and identify the cellsthat stain positive for TNFα, double immunofluorescence staining wasperformed. 35G10F3 and IP-300 were used in combination with a cell specificantibody to see which cells are positive for TNFα. Both antibodies showed thatendothelial cells, but also other cells are TNFα positive. Furthermore, 35G10F3and IP-300 both showed that some but not all monocyte / macrophage like cellsare TNFα positive.

Pre incubation of TNFα antibodies with recombinant TNFα

In order to determine the specificity of the three antibodies with a cleardifferent staining pattern and of the antibodies that did not react with TNFα onblot, these antibodies were pre incubated with recombinant TNFα. After preincubation with TNFα, staining of 4H31, 61E71, D9D0, 35G10F3 or IP-300 wasabolished (Fig.4a, c, e).

Chapter 2

34

Figure 2. Staining patterns of TNFα using different TNFα antibodies (a-e) and the TNF receptor I,p55 (f) on cryosections of normal human skin. a) Antibody 10F: diffuse staining in the epidermallayer, b) Antibody 52B83: diffuse staining in the epidermal layer, clear staining of endothelial cells inthe dermal layer c) Antibody 61E71: diffuse staining in the epidermal layer, strongest in stratumbasale, d) Antibody 35G10F3: staining of endothelial cells in the dermal layer, e) Antibody28401.111: faint staining in the epidermal layer, f) Antibody MR1-2.

Figure 3. Positive staining of dermal structures a) Hair follicle - 10F, b) Muscle - 61E71 c) Sweatglands - 35G10F3.

TNFα in human skin

35

Figure 4. TNFα detection with (a, c, e) and without (b, d, f) preincubation of recombinant TNFαprotein, using (a, b ) 4H31, (c, d) 35G10F3 and (e, f) IP-300 (a-f 400x).

Chapter 2

36

Discussion

This study shows that TNFα is present in normal human skin, but thatdepending on which TNFα antibody is used, different staining patterns arefound. Furthermore, we found that although the tested antibodies are able todetect TNFα they are not necessarily useful for detecting TNFα in human skin.

Cryosections of normal human skin stained positively with all antibodies,except 199. Although TNFα antibodies 10F, 4H31, 52B83, 61E71, D9D10,35G10F3, IP-300, J1D9 and 28401.111 are able to detect TNFα in human skin,the staining pattern of these antibodies is different (Table 3 and Fig. 2, 3 and 4).35G10F3 and 4H31 revealed a clear distinct staining pattern, at a lowconcentration (2 µg/ml), which might imply that these antibodies have a highaffinity for TNFα. Antibodies 52B83, 61E71 and IP-300 are able to detectdistinct positive cells, but 10F, D9D10, J1D9 and 28401.111 reveal a diffusestaining pattern which is rather difficult to interpret.

In this study, we determined that although the antibodies reveal verydifferent staining patterns they all bind to TNFα. From the experiments with LPSstimulated PBMCs, in which we demonstrated that all antibodies reacted withstimulated, but not with unstimulated cells, we concluded that all antibodies areprobably reactive to TNFα protein. To confirm that the antibodies reactspecifically with TNFα and not with other proteins, Western blotting wasperformed. The antibodies 10F, 52B83, 35G10F3, IP-300, J1D9 and 28401.111detected only TNFα and none of the other proteins present. Therefore theseantibodies are judged to be specific for TNFα. Antibodies 4H31, 61E71, D9D10and 199 turned out not to be suitable for TNFα detection on blot. This may bedue to the changed availability of epitopes on blot. Conditions such as heating at57°C, although rather mild as compared to the usually used 100°C, and thepresence of SDS is known to induce conformational changes of the TNFαprotein, which may not be reversed during the rest of the procedure leading todestruction of epitopes.

Because specific reactivity for TNFα could not be confirmed for 4H31,61E71 and D9D10 on Western blot, an additional experiment with pre incubationof recombinant TNFα was performed. Abolishment of TNFα staining after thepre incubation confirmed that the antibodies are specific for TNFα. In addition,pre incubation of the antibodies 35G10F3 and IP-300 with recombinant TNFαand the double immunofluorescence experiments showed that although theseantibodies specifically bind TNFα and detect TNFα in endothelial cells andmonocytes, still different staining patterns are found.

These different staining patterns and differences as observed by using theother antibodies may be due to the fact that the antibodies recognize different

TNFα in human skin

37

epitopes or different forms of the TNFα protein. TNFα is synthesized as a pro-protein of 26 kD that is cleaved at the membrane by a metalloproteinase to a 17kD monomeric form (22;23) (24). The biological active TNFα is a homotrimerthat can mediate its effects after binding the membrane bound receptor TNF RI(p55) and TNF RII (p75) (25-27). Some antibodies may stain accumulation ofthe TNFα pro-protein, but also other forms such as membrane or receptorbound TNFα may be recognized by antibodies.

In this study, binding to mono, dimer or tetramer protein on Western blotdoes not correspond with a specific staining pattern on cryosections and can notexplain the different TNFα staining patterns on human skin. The staining patternof TNF receptor p55 was very similar to the pattern of 4H31.This may indicatethat 4H31 recognizes TNFα bound to its receptor and explain the staining patternof this antibody.

A staining pattern, similar to the one found by us, was described by Orteuet al. (28) for 35G10F3. Using the polyclonal antibody IP-300, Pierard et al. (11)showed positive staining of the upper layers, Terajima et al. (20) did not detectpositive staining in the epidermis, whereas we found staining in especially thebasal layer of the epidermis. So, even with the use of the same antibody differentstaining patterns are described. Discrepancies between our and earlier findingsusing 52B83 (19), 61E71 (9), D9D10 (29) and between our results and literaturein general may be explained by different concentrations of the antibody used, useof different protocols or different specificity of the antibodies used.

Our results confirm the conflicting results found in literature regarding thelocalization of TNFα in normal human skin using different antibodies. Therefore,the immunohistochemical approach should be used with care for studying TNFαproduction in human skin. It is not clear what underlies differences in stainingpattern of the different antibodies, and this makes comparison between studiesdifficult. Clear descriptions of the concentrations, protocols, and controls maymake it possible to compare different studies, but even then, for unambiguousinterpretation of data, it is important to know the characterization of theantibodies used. Elucidation of origin of differences in staining patterns maycontribute to interpretation of data.

In summary, we have shown that TNFα is present in normal human skin,but that depending on which TNFα antibody is used, different staining patternsare found. Furthermore, we found that although the tested antibodies are able todetect TNFα they are not necessarily useful for detecting TNFα in human skin.

Chapter 2

38

References

1. Luger, T. A. Epidermal Cytokines. Acta Derm.Venereol.Suppl.Stockh. 1989;151:61-76.

2. Kock, A., Schwarz, T., Kirnbauer, R., Urbanski, A., Perry, P., Ansel, J. C., and Luger, T. A.Human Keratinocytes Are a Source for Tumor Necrosis Factor Alpha: Evidence for Synthesisand Release Upon Stimulation With Endotoxin or Ultraviolet Light. J.Exp.Med. 1-12-1990;172(6):1609-14.

3. Ansel, J., Perry, P., Brown, J., Damm, D., Phan, T., Hart, C., Luger, T., and Hefeneider, S.Cytokine Modulation of Keratinocyte Cytokines. J.Invest.Dermatol. 1990;94(6 Suppl):101S-7S.

4. Bos, J. D. and Kapsenberg, M. L. The Skin Immune System: Progress in Cutaneous Biology.Immunol.Today 1993;14(2):75-8.

5. Oxholm, A., Oxholm, P., Staberg, B., and Bendtzen, K. Immunohistological Detection ofInterleukin I-Like Molecules and Tumour Necrosis Factor in Human Epidermis Before and AfterUVB-Irradiation in Vivo. Br.J.Dermatol. 1988;118(3):369-76.

6. Oxholm, A., Diamant, M., Oxholm, P., and Bendtzen, K. Interleukin-6 and Tumour NecrosisFactor Alpha Are Expressed by Keratinocytes but Not by Langerhans Cells. APMIS1991;99(1):58-64.

7. Oxholm, A. Epidermal Expression of Interleukin-6 and Tumour Necrosis Factor-Alpha in Normaland Immunoinflammatory Skin States in Humans. APMIS Suppl. 1992;24:1-32.

8. Kolde, G., Schulze, Osthoff K., Meyer, H., and Knop, J. Immunohistological and ImmunoelectronMicroscopic Identification of TNF Alpha in Normal Human and Murine Epidermis.Arch.Dermatol.Res. 1992;284(3):154-8.

9. Hoefakker, S., Caubo, M., van-'t-Erve, E. H., Roggeveen, M. J., Boersma, W. J., van-Joost, T.,Notten, W. R., and Claassen, E. In Vivo Cytokine Profiles in Allergic and Irritant ContactDermatitis. Contact.Dermatitis. 1995;33(4):258-66.

10. Paquet, P., Nikkels, A., Arrese, J. E., Vanderkelen, A., and Pierard, G. E. Macrophages andTumor Necrosis Factor Alpha in Toxic Epidermal Necrolysis. Arch.Dermatol. 1994;130(5):605-8.

11. Pierard, G. E., Nikkels, Tassoudji N., Arrese, J. E., Pierard, Franchimont C., and Leveque, J. L.Dermo-Epidermal Stimulation Elicited by a Beta-Lipohydroxyacid: a Comparison With SalicylicAcid and All-Trans-Retinoic Acid. Dermatology. 1997;194(4):398-401.

12. Walsh, L. J., Trinchieri, G., Waldorf, H. A., Whitaker, D., and Murphy, G. F. Human DermalMast Cells Contain and Release Tumor Necrosis Factor Alpha, Which Induces EndothelialLeukocyte Adhesion Molecule 1. Proc.Natl.Acad.Sci.U.S.A. 15-5-1991;88(10):4220-4.

13. Chu, C. Q., Field, M., Andrew, E., Haskard, D., Feldmann, M., and Maini, R. N. Detection ofCytokines at the Site of Tuberculin-Induced Delayed-Type Hypersensitivity in Man.Clin.Exp.Immunol. 1992;90(3):522-9.

14. Kristensen, M., Chu, C. Q., Eedy, D. J., Feldmann, M., Brennan, F. M., and Breathnach, S. M.Localization of Tumour Necrosis Factor-Alpha (TNF-Alpha) and Its Receptors in Normal and

TNFα in human skin

39

Psoriatic Skin: Epidermal Cells Express the 55-KD but Not the 75-KD TNF Receptor.Clin.Exp.Immunol. 1993;94(2):354-62.

15. Ruco, L. P., Stoppacciaro, A., Pomponi, D., Boraschi, D., Santoni, A., Tagliabue, A., Uccini, S.,and Baroni, C. D. Immunoreactivity for IL-1 Beta and TNF Alpha in Human Lymphoid andNonlymphoid Tissues. Am.J.Pathol. 1989;135(5):889-97.

16. Griffiths, C. E., Barker, J. N., Kunkel, S., and Nickoloff, B. J. Modulation of LeucocyteAdhesion Molecules, a T-Cell Chemotaxin (IL-8) and a Regulatory Cytokine (TNF-Alpha) inAllergic Contact Dermatitis (Rhus Dermatitis). Br.J.Dermatol. 1991;124(6):519-26.

17. Nickoloff, B. J., Karabin, G. D., Barker, J. N., Griffiths, C. E., Sarma, V., Mitra, R. S., Elder, J.T., Kunkel, S. L., and Dixit, V. M. Cellular Localization of Interleukin-8 and Its Inducer, TumorNecrosis Factor-Alpha in Psoriasis. Am.J.Pathol. 1991;138(1):129-40.

18. Castagnoli, C., Stella, M., Berthod, C., Magliacani, G., and Richiardi, P. M. TNF Production andHypertrophic Scarring. Cell Immunol. 1993;147(1):51-63.

19. Turner, G. D., Ly, V. C., Nguyen, T. H., Tran, T. H., Nguyen, H. P., Bethell, D., Wyllie, S.,Louwrier, K., Fox, S. B., Gatter, K. C., Day, N. P., White, N. J., and Berendt, A. R. SystemicEndothelial Activation Occurs in Both Mild and Severe Malaria. Correlating DermalMicrovascular Endothelial Cell Phenotype and Soluble Cell Adhesion Molecules With DiseaseSeverity. Am.J.Pathol. 1998;152(6):1477-87.

20. Terajima, S., Higaki, M., Igarashi, Y., Nogita, T., and Kawashima, M. An Important Role ofTumor Necrosis Factor-Alpha in the Induction of Adhesion Molecules in Psoriasis.Arch.Dermatol.Res. 1998;290(5):246-52.

21. Roozendaal, C., Zhao, M. H., Horst, G., Lockwood, C. M., Kleibeuker, J. H., Limburg, P. C.,Nelis, G. F., and Kallenberg, C. G. Catalase and Alpha-Enolase: Two Novel GranulocyteAutoantigens in Inflammatory Bowel Disease (IBD). Clin.Exp.Immunol. 1998;112(1):10-6.

22. Pennica, D., Nedwin, G. E., Hayflick, J. S., Seeburg, P. H., Derynck, R., Palladino, M. A., Kohr,W. J., Aggarwal, B. B., and Goeddel, D. V. Human Tumour Necrosis Factor: PrecursorStructure, Expression and Homology to Lymphotoxin. Nature 20-12-1984;312(5996):724-9.

23. Kriegler, M., Perez, C., DeFay, K., Albert, I., and Lu, S. D. A Novel Form of TNF/Cachectin Isa Cell Surface Cytotoxic Transmembrane Protein: Ramifications for the Complex Physiology ofTNF. Cell 8-4-1988;53(1):45-53.

24. Gearing, A. J., Beckett, P., Christodoulou, M., Churchill, M., Clements, J., Davidson, A. H.,Drummond, A. H., Galloway, W. A., Gilbert, R., Gordon, J. L., and et, al. Processing of TumourNecrosis Factor-Alpha Precursor by Metalloproteinases. Nature 18-8-1994;370(6490):555-7.

25. Smith, R. A. and Baglioni, C. The Active Form of Tumor Necrosis Factor Is a Trimer.J.Biol.Chem. 25-5-1987;262(15):6951-4.

26. Jones, E. Y., Stuart, D. I., and Walker, N. P. Structure of Tumour Necrosis Factor. Nature 16-3-1989;338(6212):225-8.

27. Bazzoni, F. and Beutler, B. The Tumor Necrosis Factor Ligand and Receptor Families.N.Engl.J.Med. 27-6-1996;334(26):1717-25.

Chapter 2

40

28. Orteu, C. H., Poulter, L. W., Rustin, M. H., Sabin, C. A., Salmon, M., and Akbar, A. N. TheRole of Apoptosis in the Resolution of T Cell-Mediated Cutaneous Inflammation. J.Immunol. 15-8-1998;161(4):1619-29.

29. de-Graaf, J. H., Tamminga, R. Y., Dam, Meiring A., Kamps, W. A., and Timens, W. ThePresence of Cytokines in Langerhans' Cell Histiocytosis. J.Pathol. 1996;180(4):400-6.