Proc. Natl. Acad. Sci. USA Vol. 88, pp. 4220-4224, May 1991 Medical Sciences
Human dermal mast cells contain and release tumor necrosis factor a, which induces endothelial leukocyte adhesion molecule 1 (skin/cytokines/endothelium/inflammation)
LAURENCE J. WALSH*, GIORGIO TRINCHIERIt, HEIDI A. WALDORF*, DIANA WHITAKER*, AND GEORGE F. MURPHY* *Department of Dermatology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104; and tWistar Institute of Anatomy and Biology, Philadelphia, PA 19104
Communicated by Hilary Koprowski, January 30, 1991
ABSTRACT Tumor necrosis factor a (TNF-a) is a proinflammatory cytokine that mediates endothelial leukocyte interactions by inducing expression of adhesion molecules. In this report, we demonstrate that human dermal mast cells contain sizeable stores of immunoreactive and biologically active TNF-a within granules, which can be released rapidly into the extracellular space upon degranulation. Among normal human dermal cells, mast cells are the predominant cell type that expresses both TNF-a protein and TNF-a mRNA. Moreover, induction of endothelial leukocyte adhesion molecule 1 expression is a direct consequence of release of mast cell-derived TNF-a. These findings establish a role for human mast cells as "gatekeepers" of the dermal microvasculature and indicate that mast cell products other than vasoactive amines influence endothelium in a proinflammatory fashion.
binant TNF-a (500 pg/ml) (R & D Systems, Minneapolis) or with other recombinant cytokines (see Results) for 18 hr at 220C prior to use for TNF-a immunolabeling. Polyclonal antibodies to the mast cell granule-specific serine proteases tryptase and chymase were used to identify dermal mast cells (13, 14), while polyclonal antibodies to factor XIIIa (Calbiochem) were used to identify dermal dendritic cells (15). Monoclonal H4/18 antibody (courtesy of M. A. Gimbrone, Jr., Harvard Medical School, Boston) was used to detect ELAM-1 (16, 17). Lineage-specific monoclonal antibodies (CD3, CD15, CD22, CD57) (Becton Dickinson) were used to analyze peripheral blood cell populations. In all cases, isotype-specific, irrelevant monoclonal antibodies or nonimmune sera were used as negative controls. Monolayer cultures of endotoxin-stimulated monocytes (18) were used as a positive control for all TNF-a immunolabeling studies. rhTNF-a (R & D Systems) was used for control ELAM-1 induction in foreskin explants and cultured endothelial cells (5, 16, 19), while polyclonal rabbit anti-human TNF-a antibodies (Genzyme) were used to neutralize TNF-a bioactivity in blocking studies. Tissues. Tissue specimens included neonatal foreskin (n = 7), normal adult skin (forearm, n = 4; breast, n = 3), normal oral mucosa (n = 3), and sequential biopsy samples of delayed cutaneous hypersensitivity to 2,4-dinitrochlorobenzene obtained 1, 2, 4, 6, 24, and 48 hr after challenge from four healthy male volunteers according to a previous protocol (20). Neonatal fores-kins from elective circumcision procedures were used for organ culture and for cell culture studies. Immunohistochemistry. Acetone-fixed, cryostat sections (4 ,um) or cytocentrifuge preparations underwent reaction with primary antibodies or appropriate irrelevant controls (3). For single labeling studies, biotinylated species-specific second layer reagents were followed by avidin-biotin-peroxidase complex (Vector Laboratories), as described (3). 3,3'Diaminobenzidine (DAB) was used as chromogen, and preparations were lightly counterstained with hematoxylin prior to dehydration and mounting. For double immunohistochemical labeling, preparations were incubated with primary antisera, followed by species-specific biotinylated antibody (Vector Laboratories) together with affinity-purified fluorescein isothiocyanate-conjugated antibody (Cappel Laboratories). Alkaline phosphatase-conjugated streptavidin (Amersham) and horseradish peroxidase-conjugated anti-fluorescein isothiocyanate antibody (DAKO, Carpinteria, CA) were used as third layer reagents (1 mM levamisole added to buffer; refs. 21 and 22). Control experiments for endogenous avidin-binding activity, endogenous peroxidase, and endogenous alkaline phosphatase were routinely performed. For
Mast cells are granule-containing secretory cells that reside mostly in connective tissue compartments. After activation by immunologic or nonimmunologic stimuli, mast cells release preformed mediators, including vasoactive amines and serine proteinases (1, 2). In vitro studies recently have demonstrated that human mast cell degranulation is linked to the induction of endothelial leukocyte adhesion molecule 1 (ELAM-1) on dermal microvascular endothelium (3, 4). This effect is abrogated by antibodies to tumor necrosis factor a (TNF-a), a known inducer of ELAM-1 (5), raising the possibility that human dermal mast cells may contain preformed TNF-a and release this mediator upon degranulation. While previous studies have described TNF-a-like cytotoxic factors and TNF-a mRNA in cultured human and rodent basophil/mast cell lines (6-11), the following questions remain unresolved: (i) do human dermal mast cells produce and secrete TNF-a; (ii) do other dermal cells constitutively contain this cytokine; and (iii) what are the biological effects of endogenously produced TNF-a on responding cells in the dermal microenvironment. In this report, we demonstrate that mast cells are a principal source of TNF-a in normal human dermis. We also provide direct evidence that mast cell degranulation in vivo and in vitro releases TNF-a, which in turn directly induces ELAM-1 on dermal microvascular endothelium.
MATERIALS AND METHODS Antibodies and Cytokines. B154.2, a monoclonal antibody that reacts with recombinant human TNF-a (rhTNF-a) but not other cytokines (12) was used for immunohistochemical detection of TNF-a. To confirm the specificity of this reagent, B154 ascites (diluted 1:500) was absorbed with recom-
Abbreviations: DAB, 3,3'-diaminobenzidine; ELAM-1, endothelial leukocyte adhesion molecule 1; HUVE, human umbilical vein endothelial cell; MS, morphine sulfate; TNF-a, tumor necrosis factor alpha; rhTNF-a, recombinant human TNF-a; IL, interleukin.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 4220
Medical Sciences: Walsh
et
al.
double immunofluorescence studies, specimens were incubated with first and second layer reagents as described above, and Texas Red-conjugated streptavidin (Amersham) was used as a third layer reagent. Microdensitometry. ELAM-1 staining intensity of the 10 most reactive vessel profiles in each tissue block was quantified by using a video-based computerized image analysis system (Microcomp DS, Southern Micro Instruments, Atlanta) attached to an Olympus BH-2 microscope (23). Density values were calculated from cross-sectional transmittance measurements through regions of maximal DAB reaction product on postcapillary venules (x400; blinded intraobserver variation, 97%). Similar results were obtained when endotoxinstimulated monocytes were exposed to MS (P < 0.0001; Fig. 4B, bar 4). Nonadherent cells (71-77% CD3+ T lymphocytes, 17-24% CD22+ B lymphocytes) secreted only small quantities of cytokine in response to MS. Mast Cell-Derived TNF-a Induces ELAM-1 Expression. Cell-free mast cell supernatants of dermal mast cell suspensions exposed to secretagogue induced ELAM-1 expression on dermal endothelium in target foreskin explants in which endogenous mast cell mediator release had been prevented by cromolyn sodium as demonstrated (3, 4). This effect was identical to that produced by rhTNF-a (5) and was abrogated by neutralizing antibody to TNF (Fig. 5 A and B). In target explants, mast cells retained TNF-a immunoreactivity and did not display structural alterations suggestive of degranulation. When HUVEs were used as targets, both mast cell supernatants and rhTNF-a induced ELAM-1 expression (Fig. 5C). As with foreskin organ cultures, the effect of mast cell supernatants was abrogated by neutralizing antibodies to TNF-a (Fig. 5D). MS alone did not induce ELAM-1 expression on HUVEs. Mast Cells Release TNF-a During Delayed Hypersensitivity Reactions. We next examined TNF-a and neutral protease expression by mast cells in vivo during cutaneous delayed hypersensitivity reactions elicited by challenge of sensitized subjects with dinitrochlorobenzene. Before challenge, mast cells coexpressed chymase, tryptase, and TNF-a, as noted above. Mast cell immunoreactivity for TNF-a was markedly diminished 4-6 hr after challenge (data not shown), the same time point at which we previously observed ultrastructural evidence of mast cell degranulation (20). Of note, after the diminution of perivascular mast cell immunoreactivity for TNF-a, ELAM-1 expression was detected on dermal postcapillary venules and perivascular lymphocyte infiltrates were observed (23). Expression of TNF-a mRNA in Dermal Mast Cells. TNF-a mRNA was rarely detected in dermal mast cells in normal skin; however, low levels of message were present in basal keratinocytes (data not shown). Mast cells containing TNF-a B
A
±
lp
D
FIG. 5. Induction of ELAM-1 by mast cell-derived TNF-a. (A) Endothelial cells lining dermal vessel of explant exposed to supernatant derived from purified mast cells show strong H4/18 reactivity, indicating ELAM-1 induction. (B) In explant treated with mast cell supernatant and blocking antiserum to TNF-a, abrogation of ELAM-1 induction is apparent. Endothelial nuclei are weakly counterstained with hematoxylin. (C) Cultured umbilical vein endothelial cells exposed to mast cell supernatant display marked ELAM-1 induction. (D) Parallel culture treated with blocking antiserum does not show H4/18 reactivity. (A and B, x200; C and D, x250.)
4224
Medical Sciences: Walsh et al.
message were more numerous in lesions of cutaneous delayed hypersensitivity reactions 48-96 hr after challenge (Fig. 2D), a period during which reappearance of granular immunoreactivity for TNF-a and mast cell proteases occurred (Fig. 2 E and F) (23).
Proc. Natl. Acad. Sci. USA 88 (1991) mast cells or to inhibit endothelial cell responsiveness to this cytokine should now be explored. We thank N. M. Schechter and M. A. Gimbrone, Jr., for their generous gifts of antibodies, and D. Cines for providing cultured umbilical vein endothelial cells. This work was supported by Grants AR3%74 and CA40358 from the National Institutes of Health.
DISCUSSION We demonstrate here that human dermal mast cells contain the potent proinflammatory cytokine TNF-a as a preformed mediator within granules. Degranulation of dermal mast cells, elicited by exposure to either immunologic stimuli (e.g., in delayed hypersensitivity reactions) or nonimmunologic stimuli (e.g., MS) results in rapid dissipation of intracellular stores of this cytokine. After degranulation in vivo, TNF-a mRNA is increased and reconstitution of TNF-a occurs. TNF-a appears to represent the principal, if not the sole, mast cell-derived mediator responsible for the induction of ELAM-1 on both cultured endothelial cells and dermal postcapillary venules. Thus, mast cell-derived TNF-a may represent an important determinant of leukocyte-endothelial cell adhesive interactions during the genesis of cellular inflammation in skin. Because mast cells are located in close proximity to vessels and are the only normal dermal cells that contain sizeable quantities of preformed TNF-a, they are ideally poised to serve as "gatekeepers" of the dermal microvasculature. During our previous studies of mast cell degranulation in vitro, we observed that mast cell degranulation induced by a variety of secretory stimuli was associated with subsequent induction of ELAM-1 in superficial dermal venules (3, 4). The kinetics of this induction were identical to those following addition of recombinant IL-1 or TNF-a to skin organ cultures (5). Moreover, induction could be abrogated by preincubation of explants with the mast cell inhibitor cromolyn sodium and with neutralizing antiserum to TNF-a (but not to IL-1) (3). These findings suggested that ELAM-1 induction was elicited by mast cell-dependent mechanisms that involved local release of TNF-a. The present study provides direct evidence that human dermal mast cells contain preformed TNF-a, which, upon release, is capable of eliciting ELAM-1 expression following mast cell degranulation. In addition to ELAM-1 induction, other known effects of TNF-a on endothelium include induction of vascular cell adhesion molecule 1 (29), increased expression of intercellular adhesion molecule 1 (19, 30), and class I and II major histocompatibility complex antigens (19). Such molecules also may promote adhesion between endothelial cells and various peripheral blood cell populations (31), and their coincident expression with ELAM-1 is likely to collaborate in the regulation of leukocyte-microvascular interactions during early phases of inflammation (5, 16, 23). The temporal association between mast cell degranulation, ELAM-1 induction, and leukocyte accumulation in experimentally induced cutaneous inflammation (23) now may be in part explained on the basis of local TNF-a release by perivascular mast cells. Although it is tempting to speculate that mast cell degranulation and concordant release of TNF-a may be a common pathway for cutaneous inflammation in general, it is likely that inductive events will prove to be considerably more complex than is indicated by the data presently at hand. In summary, this study indicates that mast cells are an important source of the proinflammatory cytokine TNF-a. Our findings establish a central role for mast cells as gatekeepers of the microvasculature. Accordingly, therapeutic strategies designed either to prevent TNF-a release from
1. Metcalf, D. D. & Kaliner, M. (1981) CRC Crit. Rev. Immunol. 3, 23-74. 2. Van Loveren, H., Teppema, J. S. & Askenase, P. W. (1990) in Skin Immune System, ed. Bos, J. D. (CRC, Boca Raton, FL), pp. 171-193. 3. Klein, L. M., Lavker, R. M., Matis, W. L. & Murphy, G. F. (1989) Proc. Nati. Acad. Sci. USA 86, 8972-8976. 4. Matis, W. L., Lavker, R. M. & Murphy, G. F. (1990) J. Invest.
Dermatol. 94, 492-495.
5. Messadi, D. V., Pober, J. S., Fiers, W., Gimbrone, M. A. & Murphy, G. F. (1987) J. Immunol. 139, 1557-1562. 6. Jadus, M. R., Schmunk, G., Djeu, J. Y. & Parkman, R. (1986) J. Immunol. 137, 2774-2783. 7. Okuno, T., Takagaki, Y., Pluznik, D. H. & Djeu, J. Y. (1986)
J. Immunol. 136, 4652-4658.
8. Young, J. D., Liu, C. C., Butler, G., Cohn, Z. A. & Galli, S.
(1987) Proc. NatI. Acad. Sci. USA 84, 9175-9179.
9. Tharp, M. D., Kasper, C., Thiele, D., Charley, M. R., Kennerly, D. A. & Sullivan, T. J. (1989) J. Invest. Dermatol. 93, 423-428. 10. Steffen, M., Abboud, M., Potter, G. K., Yung, Y. P. & Moore, A. S. (1989) Immunology 66, 445-450. 11. Gordon, J. R. & Galli, S. J. (1990) Nature (London) 346, 274-276. 12. Cuturi, M. C., Murphy, M., Costa-Giomi, M. P., Weinmann, R., Perussia, B. & Trinchieri, G. (1987) J. Exp. Med. 165,
1581-1594.
13. Craig, S. S., Schechter, N. M. & Schwartz, L. B. (1989) Lab.
Invest. 60, 147-157.
14. Schechter, N. M., Choi, J. K., Slavin, D. A., Deresienski, D. T., Sayama, S., Dong, G., Lavker, R. M., Proud, D. & Lazarus, G. S. (1986) J. Immunol. 137, 962-970. 15. Cerio, R., Spaull, J., Oliver, G. F. & Wilson Jones, E. (1990) Am. J. Dermatopathol. 12, 221-233. 16. Pober, J. S., Bevilacqua, M. P., Mendrick, D. L., Lapierre, L. A., Fiers, W. & Gimbrone, M. A. (1986) J. Immunol. 136,
1680-1687. 17. Bevilacqua, M. P., Stengelin, S., Gimbrone, M. A. & Seed, B.
(1989) Science 243, 1160-1165.
18. Andersson, U. & Matsuda, T. (1989) Eur. J. Immunol. 19, 1157-1160. 19- Pober, J. S., Gimbrone, M. A., Lapierre, L. A., Mendrick, D. L., Fiers, W., Rothlein, R. & Springer, T. A. (1987) J.
Immunol. 137, 1893-18%.
20. Lewis, R. E., Buchsbaum, M., Whitaker, D. & Murphy, G. F. (1989) J. Invest. Dermatol. 93, 672-677. 21. Graham, R. C. & Karnovsky, M. J. (1966) J. Histochem.
Cytochem. 14, 291-302. 22. Leary, J. J., Brigati, D. J. & Ward, D. C. (1983) Proc. Natl. Acad. Sci. USA 80, 4045-4049. 23. Waldorf, H. A., Walsh, L. J., Schechter, N. M. & Murphy, G. F. (1991) Am. J. Pathol. 138, 477-486. 24. Benyon, R. C., Lorreman, M. A. & Church, M. K. (1987) J. Immunol. 138, 861-867. 25. Enerback, L. (1974) Histochemistry 42, 301-313. 26. Walsh, L. J., Stritzel, F., Yamazaki, K., Bird, P. S., Gemmell, E. & Seymour, G. J. (1989) Arch. Oral Biol. 34, 679-683. 27. Craig, S. S., Schechter, N. M. & Schwartz, L. B. (1988) Lab. Invest. 58, 682-691. 28. Billingham, M. E. (1987) Br. Med. Bull. 43, 350-370. 29. Osborn, L., Hessian, C., Tizard, R., Vassalo, C., Luhoneskyj, S., Chi-Rosso, G. & Lobb, R. (1989) Cell 59, 1203-1211. 30. Dustin, M. L., Rothlein, R., Bhan, A. K., Dinarello, C. A. & Springer, T. A. (1986) J. Immunol. 137, 245-254. 31. Walsh, L. J., Lavker, R. M. & Murphy, G. F. (1990) Lab. Invest. 63, 592-600.
Human dermal mast cells contain and release tumor necrosis factor a, which induces endothelial leukocyte adhesion molecule 1 (skin/cytokines/endothelium/inflammation)
LAURENCE J. WALSH*, GIORGIO TRINCHIERIt, HEIDI A. WALDORF*, DIANA WHITAKER*, AND GEORGE F. MURPHY* *Department of Dermatology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104; and tWistar Institute of Anatomy and Biology, Philadelphia, PA 19104
Communicated by Hilary Koprowski, January 30, 1991
ABSTRACT Tumor necrosis factor a (TNF-a) is a proinflammatory cytokine that mediates endothelial leukocyte interactions by inducing expression of adhesion molecules. In this report, we demonstrate that human dermal mast cells contain sizeable stores of immunoreactive and biologically active TNF-a within granules, which can be released rapidly into the extracellular space upon degranulation. Among normal human dermal cells, mast cells are the predominant cell type that expresses both TNF-a protein and TNF-a mRNA. Moreover, induction of endothelial leukocyte adhesion molecule 1 expression is a direct consequence of release of mast cell-derived TNF-a. These findings establish a role for human mast cells as "gatekeepers" of the dermal microvasculature and indicate that mast cell products other than vasoactive amines influence endothelium in a proinflammatory fashion.
binant TNF-a (500 pg/ml) (R & D Systems, Minneapolis) or with other recombinant cytokines (see Results) for 18 hr at 220C prior to use for TNF-a immunolabeling. Polyclonal antibodies to the mast cell granule-specific serine proteases tryptase and chymase were used to identify dermal mast cells (13, 14), while polyclonal antibodies to factor XIIIa (Calbiochem) were used to identify dermal dendritic cells (15). Monoclonal H4/18 antibody (courtesy of M. A. Gimbrone, Jr., Harvard Medical School, Boston) was used to detect ELAM-1 (16, 17). Lineage-specific monoclonal antibodies (CD3, CD15, CD22, CD57) (Becton Dickinson) were used to analyze peripheral blood cell populations. In all cases, isotype-specific, irrelevant monoclonal antibodies or nonimmune sera were used as negative controls. Monolayer cultures of endotoxin-stimulated monocytes (18) were used as a positive control for all TNF-a immunolabeling studies. rhTNF-a (R & D Systems) was used for control ELAM-1 induction in foreskin explants and cultured endothelial cells (5, 16, 19), while polyclonal rabbit anti-human TNF-a antibodies (Genzyme) were used to neutralize TNF-a bioactivity in blocking studies. Tissues. Tissue specimens included neonatal foreskin (n = 7), normal adult skin (forearm, n = 4; breast, n = 3), normal oral mucosa (n = 3), and sequential biopsy samples of delayed cutaneous hypersensitivity to 2,4-dinitrochlorobenzene obtained 1, 2, 4, 6, 24, and 48 hr after challenge from four healthy male volunteers according to a previous protocol (20). Neonatal fores-kins from elective circumcision procedures were used for organ culture and for cell culture studies. Immunohistochemistry. Acetone-fixed, cryostat sections (4 ,um) or cytocentrifuge preparations underwent reaction with primary antibodies or appropriate irrelevant controls (3). For single labeling studies, biotinylated species-specific second layer reagents were followed by avidin-biotin-peroxidase complex (Vector Laboratories), as described (3). 3,3'Diaminobenzidine (DAB) was used as chromogen, and preparations were lightly counterstained with hematoxylin prior to dehydration and mounting. For double immunohistochemical labeling, preparations were incubated with primary antisera, followed by species-specific biotinylated antibody (Vector Laboratories) together with affinity-purified fluorescein isothiocyanate-conjugated antibody (Cappel Laboratories). Alkaline phosphatase-conjugated streptavidin (Amersham) and horseradish peroxidase-conjugated anti-fluorescein isothiocyanate antibody (DAKO, Carpinteria, CA) were used as third layer reagents (1 mM levamisole added to buffer; refs. 21 and 22). Control experiments for endogenous avidin-binding activity, endogenous peroxidase, and endogenous alkaline phosphatase were routinely performed. For
Mast cells are granule-containing secretory cells that reside mostly in connective tissue compartments. After activation by immunologic or nonimmunologic stimuli, mast cells release preformed mediators, including vasoactive amines and serine proteinases (1, 2). In vitro studies recently have demonstrated that human mast cell degranulation is linked to the induction of endothelial leukocyte adhesion molecule 1 (ELAM-1) on dermal microvascular endothelium (3, 4). This effect is abrogated by antibodies to tumor necrosis factor a (TNF-a), a known inducer of ELAM-1 (5), raising the possibility that human dermal mast cells may contain preformed TNF-a and release this mediator upon degranulation. While previous studies have described TNF-a-like cytotoxic factors and TNF-a mRNA in cultured human and rodent basophil/mast cell lines (6-11), the following questions remain unresolved: (i) do human dermal mast cells produce and secrete TNF-a; (ii) do other dermal cells constitutively contain this cytokine; and (iii) what are the biological effects of endogenously produced TNF-a on responding cells in the dermal microenvironment. In this report, we demonstrate that mast cells are a principal source of TNF-a in normal human dermis. We also provide direct evidence that mast cell degranulation in vivo and in vitro releases TNF-a, which in turn directly induces ELAM-1 on dermal microvascular endothelium.
MATERIALS AND METHODS Antibodies and Cytokines. B154.2, a monoclonal antibody that reacts with recombinant human TNF-a (rhTNF-a) but not other cytokines (12) was used for immunohistochemical detection of TNF-a. To confirm the specificity of this reagent, B154 ascites (diluted 1:500) was absorbed with recom-
Abbreviations: DAB, 3,3'-diaminobenzidine; ELAM-1, endothelial leukocyte adhesion molecule 1; HUVE, human umbilical vein endothelial cell; MS, morphine sulfate; TNF-a, tumor necrosis factor alpha; rhTNF-a, recombinant human TNF-a; IL, interleukin.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 4220
Medical Sciences: Walsh
et
al.
double immunofluorescence studies, specimens were incubated with first and second layer reagents as described above, and Texas Red-conjugated streptavidin (Amersham) was used as a third layer reagent. Microdensitometry. ELAM-1 staining intensity of the 10 most reactive vessel profiles in each tissue block was quantified by using a video-based computerized image analysis system (Microcomp DS, Southern Micro Instruments, Atlanta) attached to an Olympus BH-2 microscope (23). Density values were calculated from cross-sectional transmittance measurements through regions of maximal DAB reaction product on postcapillary venules (x400; blinded intraobserver variation, 97%). Similar results were obtained when endotoxinstimulated monocytes were exposed to MS (P < 0.0001; Fig. 4B, bar 4). Nonadherent cells (71-77% CD3+ T lymphocytes, 17-24% CD22+ B lymphocytes) secreted only small quantities of cytokine in response to MS. Mast Cell-Derived TNF-a Induces ELAM-1 Expression. Cell-free mast cell supernatants of dermal mast cell suspensions exposed to secretagogue induced ELAM-1 expression on dermal endothelium in target foreskin explants in which endogenous mast cell mediator release had been prevented by cromolyn sodium as demonstrated (3, 4). This effect was identical to that produced by rhTNF-a (5) and was abrogated by neutralizing antibody to TNF (Fig. 5 A and B). In target explants, mast cells retained TNF-a immunoreactivity and did not display structural alterations suggestive of degranulation. When HUVEs were used as targets, both mast cell supernatants and rhTNF-a induced ELAM-1 expression (Fig. 5C). As with foreskin organ cultures, the effect of mast cell supernatants was abrogated by neutralizing antibodies to TNF-a (Fig. 5D). MS alone did not induce ELAM-1 expression on HUVEs. Mast Cells Release TNF-a During Delayed Hypersensitivity Reactions. We next examined TNF-a and neutral protease expression by mast cells in vivo during cutaneous delayed hypersensitivity reactions elicited by challenge of sensitized subjects with dinitrochlorobenzene. Before challenge, mast cells coexpressed chymase, tryptase, and TNF-a, as noted above. Mast cell immunoreactivity for TNF-a was markedly diminished 4-6 hr after challenge (data not shown), the same time point at which we previously observed ultrastructural evidence of mast cell degranulation (20). Of note, after the diminution of perivascular mast cell immunoreactivity for TNF-a, ELAM-1 expression was detected on dermal postcapillary venules and perivascular lymphocyte infiltrates were observed (23). Expression of TNF-a mRNA in Dermal Mast Cells. TNF-a mRNA was rarely detected in dermal mast cells in normal skin; however, low levels of message were present in basal keratinocytes (data not shown). Mast cells containing TNF-a B
A
±
lp
D
FIG. 5. Induction of ELAM-1 by mast cell-derived TNF-a. (A) Endothelial cells lining dermal vessel of explant exposed to supernatant derived from purified mast cells show strong H4/18 reactivity, indicating ELAM-1 induction. (B) In explant treated with mast cell supernatant and blocking antiserum to TNF-a, abrogation of ELAM-1 induction is apparent. Endothelial nuclei are weakly counterstained with hematoxylin. (C) Cultured umbilical vein endothelial cells exposed to mast cell supernatant display marked ELAM-1 induction. (D) Parallel culture treated with blocking antiserum does not show H4/18 reactivity. (A and B, x200; C and D, x250.)
4224
Medical Sciences: Walsh et al.
message were more numerous in lesions of cutaneous delayed hypersensitivity reactions 48-96 hr after challenge (Fig. 2D), a period during which reappearance of granular immunoreactivity for TNF-a and mast cell proteases occurred (Fig. 2 E and F) (23).
Proc. Natl. Acad. Sci. USA 88 (1991) mast cells or to inhibit endothelial cell responsiveness to this cytokine should now be explored. We thank N. M. Schechter and M. A. Gimbrone, Jr., for their generous gifts of antibodies, and D. Cines for providing cultured umbilical vein endothelial cells. This work was supported by Grants AR3%74 and CA40358 from the National Institutes of Health.
DISCUSSION We demonstrate here that human dermal mast cells contain the potent proinflammatory cytokine TNF-a as a preformed mediator within granules. Degranulation of dermal mast cells, elicited by exposure to either immunologic stimuli (e.g., in delayed hypersensitivity reactions) or nonimmunologic stimuli (e.g., MS) results in rapid dissipation of intracellular stores of this cytokine. After degranulation in vivo, TNF-a mRNA is increased and reconstitution of TNF-a occurs. TNF-a appears to represent the principal, if not the sole, mast cell-derived mediator responsible for the induction of ELAM-1 on both cultured endothelial cells and dermal postcapillary venules. Thus, mast cell-derived TNF-a may represent an important determinant of leukocyte-endothelial cell adhesive interactions during the genesis of cellular inflammation in skin. Because mast cells are located in close proximity to vessels and are the only normal dermal cells that contain sizeable quantities of preformed TNF-a, they are ideally poised to serve as "gatekeepers" of the dermal microvasculature. During our previous studies of mast cell degranulation in vitro, we observed that mast cell degranulation induced by a variety of secretory stimuli was associated with subsequent induction of ELAM-1 in superficial dermal venules (3, 4). The kinetics of this induction were identical to those following addition of recombinant IL-1 or TNF-a to skin organ cultures (5). Moreover, induction could be abrogated by preincubation of explants with the mast cell inhibitor cromolyn sodium and with neutralizing antiserum to TNF-a (but not to IL-1) (3). These findings suggested that ELAM-1 induction was elicited by mast cell-dependent mechanisms that involved local release of TNF-a. The present study provides direct evidence that human dermal mast cells contain preformed TNF-a, which, upon release, is capable of eliciting ELAM-1 expression following mast cell degranulation. In addition to ELAM-1 induction, other known effects of TNF-a on endothelium include induction of vascular cell adhesion molecule 1 (29), increased expression of intercellular adhesion molecule 1 (19, 30), and class I and II major histocompatibility complex antigens (19). Such molecules also may promote adhesion between endothelial cells and various peripheral blood cell populations (31), and their coincident expression with ELAM-1 is likely to collaborate in the regulation of leukocyte-microvascular interactions during early phases of inflammation (5, 16, 23). The temporal association between mast cell degranulation, ELAM-1 induction, and leukocyte accumulation in experimentally induced cutaneous inflammation (23) now may be in part explained on the basis of local TNF-a release by perivascular mast cells. Although it is tempting to speculate that mast cell degranulation and concordant release of TNF-a may be a common pathway for cutaneous inflammation in general, it is likely that inductive events will prove to be considerably more complex than is indicated by the data presently at hand. In summary, this study indicates that mast cells are an important source of the proinflammatory cytokine TNF-a. Our findings establish a central role for mast cells as gatekeepers of the microvasculature. Accordingly, therapeutic strategies designed either to prevent TNF-a release from
1. Metcalf, D. D. & Kaliner, M. (1981) CRC Crit. Rev. Immunol. 3, 23-74. 2. Van Loveren, H., Teppema, J. S. & Askenase, P. W. (1990) in Skin Immune System, ed. Bos, J. D. (CRC, Boca Raton, FL), pp. 171-193. 3. Klein, L. M., Lavker, R. M., Matis, W. L. & Murphy, G. F. (1989) Proc. Nati. Acad. Sci. USA 86, 8972-8976. 4. Matis, W. L., Lavker, R. M. & Murphy, G. F. (1990) J. Invest.
Dermatol. 94, 492-495.
5. Messadi, D. V., Pober, J. S., Fiers, W., Gimbrone, M. A. & Murphy, G. F. (1987) J. Immunol. 139, 1557-1562. 6. Jadus, M. R., Schmunk, G., Djeu, J. Y. & Parkman, R. (1986) J. Immunol. 137, 2774-2783. 7. Okuno, T., Takagaki, Y., Pluznik, D. H. & Djeu, J. Y. (1986)
J. Immunol. 136, 4652-4658.
8. Young, J. D., Liu, C. C., Butler, G., Cohn, Z. A. & Galli, S.
(1987) Proc. NatI. Acad. Sci. USA 84, 9175-9179.
9. Tharp, M. D., Kasper, C., Thiele, D., Charley, M. R., Kennerly, D. A. & Sullivan, T. J. (1989) J. Invest. Dermatol. 93, 423-428. 10. Steffen, M., Abboud, M., Potter, G. K., Yung, Y. P. & Moore, A. S. (1989) Immunology 66, 445-450. 11. Gordon, J. R. & Galli, S. J. (1990) Nature (London) 346, 274-276. 12. Cuturi, M. C., Murphy, M., Costa-Giomi, M. P., Weinmann, R., Perussia, B. & Trinchieri, G. (1987) J. Exp. Med. 165,
1581-1594.
13. Craig, S. S., Schechter, N. M. & Schwartz, L. B. (1989) Lab.
Invest. 60, 147-157.
14. Schechter, N. M., Choi, J. K., Slavin, D. A., Deresienski, D. T., Sayama, S., Dong, G., Lavker, R. M., Proud, D. & Lazarus, G. S. (1986) J. Immunol. 137, 962-970. 15. Cerio, R., Spaull, J., Oliver, G. F. & Wilson Jones, E. (1990) Am. J. Dermatopathol. 12, 221-233. 16. Pober, J. S., Bevilacqua, M. P., Mendrick, D. L., Lapierre, L. A., Fiers, W. & Gimbrone, M. A. (1986) J. Immunol. 136,
1680-1687. 17. Bevilacqua, M. P., Stengelin, S., Gimbrone, M. A. & Seed, B.
(1989) Science 243, 1160-1165.
18. Andersson, U. & Matsuda, T. (1989) Eur. J. Immunol. 19, 1157-1160. 19- Pober, J. S., Gimbrone, M. A., Lapierre, L. A., Mendrick, D. L., Fiers, W., Rothlein, R. & Springer, T. A. (1987) J.
Immunol. 137, 1893-18%.
20. Lewis, R. E., Buchsbaum, M., Whitaker, D. & Murphy, G. F. (1989) J. Invest. Dermatol. 93, 672-677. 21. Graham, R. C. & Karnovsky, M. J. (1966) J. Histochem.
Cytochem. 14, 291-302. 22. Leary, J. J., Brigati, D. J. & Ward, D. C. (1983) Proc. Natl. Acad. Sci. USA 80, 4045-4049. 23. Waldorf, H. A., Walsh, L. J., Schechter, N. M. & Murphy, G. F. (1991) Am. J. Pathol. 138, 477-486. 24. Benyon, R. C., Lorreman, M. A. & Church, M. K. (1987) J. Immunol. 138, 861-867. 25. Enerback, L. (1974) Histochemistry 42, 301-313. 26. Walsh, L. J., Stritzel, F., Yamazaki, K., Bird, P. S., Gemmell, E. & Seymour, G. J. (1989) Arch. Oral Biol. 34, 679-683. 27. Craig, S. S., Schechter, N. M. & Schwartz, L. B. (1988) Lab. Invest. 58, 682-691. 28. Billingham, M. E. (1987) Br. Med. Bull. 43, 350-370. 29. Osborn, L., Hessian, C., Tizard, R., Vassalo, C., Luhoneskyj, S., Chi-Rosso, G. & Lobb, R. (1989) Cell 59, 1203-1211. 30. Dustin, M. L., Rothlein, R., Bhan, A. K., Dinarello, C. A. & Springer, T. A. (1986) J. Immunol. 137, 245-254. 31. Walsh, L. J., Lavker, R. M. & Murphy, G. F. (1990) Lab. Invest. 63, 592-600.
