Different interaction of mast cells with human endothelial cells and fibroblasts

ARTICLE

Mast cells (MC) have been implicated in the pathogenesis of a number of chronic inflammatory conditions that result in vascular injury and fibrosis such as keloids, hypertrophic scars, systemic sclerosis (SSc), and chronic graft-versus-host disease [1]. In chronic graft-versus-host disease, MC undergo changes in numbers and morphology that correlate with the course of fibrosis development [2]. Recent studies have reported several results suggesting MC involvement in SSc [3-7]. Vascular endothelial cell (EC) injury has been known to occur during the early stages of the disease, before the sclerosis is clinically detectable [8, 9]. From a study of experimental chronic graft-vs-host disease and SSc, increased MC activity has been shown [10]. A possibility has been proposed that a link between the activation of both EC and fibroblasts (FB) may be provided by the family of heparin-binding growth factors [10]. These and other growth factors may be responsible for EC proliferation and excess collagen production by FB. Peripheral blood mononuclear cell supernatants from GVHD display histamine releasing activity when cocultured with rat MC [11]. The GVHD supernatant decreased FB [³H]-thymidine uptake, but the presence of MC in the culture abrogated this inhibitory effect. Skin biopsies of involved areas in GVHD patients revealed significantly reduced numbers of MC and showed signs of MC degranulation. It was concluded that complex interactions between immunocompetent cells, MC, and FB probably play a role in GVHD pathogenesis. Concomitant study of salivary gland biopsy tissues and sera for the expression of E-selectin and its potent activator tumor necrosis factor alpha (TNF alpha), using immunostaining and enzyme-linked immunosorbent assay [7]. E-selectin was overexpressed in SSc patients, and TNF alpha was detected in MC. MC-derived TNF alpha was concluded to contribute to EC activation in SSc. Although the cause of the disease is unknown, evidence indicates that the fibrosis is associated with excessive accumulation of collagen in the skin and of the organs [12], the result of increased collagen synthesis by FB due to a defect in FB regulation [13]. This observation raises the intriguing possibility that MC react with EC and FB. In the present study, the possible interactions of MC with EC and FB were studied utilizing MC-EC and MC-FB coculture system.

Materials and methods

Endothelial cell and fibroblast culture

EC were isolated from human umbilical vein by adapting the method of Jaffe et al. [14]. The vein was washed with HBSS (pH 7.4) and digested with 0.2% collagenase (type II, Sigma, St. Louis, USA) in HBSS for 15 min at 37° C and rinsed with HBSS. The cells, collected by centrifugation at 250 g for 5 min, were washed and cultured in Medium199 with 20% FCS, 200 mug/ml glutamine, 100 U/ml penicillin, 50 mug/ml gentamicine, 30 mug/ml endothelial cell growth factor (Sigma, St. Louis, USA) and 100 mug/ml heparin in plastic dishes at 37° C with 5% CO₂ in air. Cells were fed twice weekly, transferred at confluence, and used for experiments from three to eight passages.

Human skin FB were obtained from the dorsum of the forearm of normal volunteer donors, and cultured with ordinary expansion method. The cells were cultured with alpha-modified Dulbeco's minimum essential medium (alpha-MEM, Gibco Laboratories, Grand Island, NY) supplemented with 5% FCS at 37° C with 5% CO₂ in air. Cells were fed twice weekly, transferred at confluence, and used for experiments from three to five passages.

Conditioned medium and MC/9 mast cell line culture

Spleen cells from C3H mice (0.5 x 10⁶ cells/ml) were cultured with alpha-MEM containing Con A (2 mug/ml) in 75-cm² tissue culture flasks [15]. The cells were incubated at 37° C in humidified 5% CO₂/85% air. After incubation for 45 hrs, the medium was collected, centrifuged for 20 min at 1,000 x g, filtered through a 0.45-mum Millipore filter, and used as conditioned medium.

The murine MC line, MC/9 was maintained in alpha-MEM supplemented with 10% FCS, 50 muM 2-mercaptoethanol and 5% conditioned medium [16]. 2 x 10⁶ cells/ml of MC were cultured with alpha-MEM supplemented with 10% FCS, 50 muM 2-mercaptoethanol and 5% conditioned medium for 4-5 days. The cells were centrifuged and the supernatants were collected and kept at ­ 20° C until use. 2 x 10⁶ cells/ml of MC were sonicated, and centrifuged at 2,500 rpm for 15 min to remove the pellet. The supernatant was stored at ­ 70° C until use as sonicated MC.

Adhesion assay

The adhesion assay using ⁵¹Cr was performed as described [17]. Briefly, MC were labeled with 150 muCi/ml of ⁵¹Cr for 12 hrs before addition to the assay wells. Cells were then separated from free ⁵¹Cr by centrifugation at 400 x g for 10 min at room temperature. The cells were resuspended in alpha-MEM supplemented with 10% FCS and 5% conditioned medium, and the cell number were adjusted to 2 x10⁶/ml. 100 mul of this suspension were placed in each assay well which contained 2 x 10⁴ EC or FB /well. The plate was maintained in a CO₂ incubator for 4 hrs unless otherwise indicated. After incubation, medium containing nonadherent cells was taken out. After washing 3 times with PBS, adherent cells were harvested on one set of filters. Radioactivity associated with the filters was assayed using a gamma counter.

Proliferation assay

EC or FB were plated in 96-well-flat-bottom plates at 2 x 10⁴ cells/well in a final volume of 200 mul added with MC, MC supernatant or sonicated MC, and incubated for 72 hrs in a CO₂ incubator. 0.5 muCi/well of ³H-thymidine was added for the last 18 hrs, and cells were harvested and counted in a scintillation counter. Results were expressed as the mean cpm of triplicate cultures.

Cytotoxic assay

MC was tested for cytotoxic activity in 4-hr, ⁵¹Cr release assays as described [18]. Briefly, EC or FB (2 x 10⁴/well) were treated with 1 muCi of ⁵¹Cr for 16 hrs. Subsequent to washing, MC (2 x 10⁵/well) were mixed with EC or FB in a plate final volume of 200 mul/well. After 4 hrs incubation at 37° C in a humidified atmosphere containing 5% CO₂, 100 mul of supernatants were collected and the ⁵¹Cr content was measured in a gamma counter. Assays were performed in triplicate and percentage cytotoxicity was calculated as percentage cytotoxicity = [(experimental release ­ spontaneous release)/(maximum release ­ spontaneous release)] x 100. Spontaneous release represents the amount of ⁵¹Cr released by EC or FB without effector cells, and maximum release represents the amount of ⁵¹Cr released by EC or FB treated with 1% Triton-X. A trypsin inhibitor (alpha1-anti-trypsin, Sigma, St. Louis, USA) was used to block the cytotoxic activity of MC.

Collagen and protein synthesis assays of fibroblasts

Collagenase or pronase releasable ³H-proline assay were used for the synthesis of collagen and protein in FB [19]. FB were plated in 24-well-flat-bottom plates at 1 x 10⁵ cells/well (confluent) in a final volume of 500 mul and cultured for 2-3 days with alpha-MEM supplemented with 5% FCS. After washing twice with warmed (37° C) Hank's balanced salt solution, the cells were cultured with alpha-MEM without FCS and with 50 mug/ml of ascorbic acid and 50 mug/ml of aminoproprionitrile in a final volume of 500 mul with MC or MC supernatants or without them. At the same time, 20 muCi/ml of 2,3,4,5-³H-proline (New England Nuclear, Boston, MA) were added. After 24 hrs, the cells were removed by a rubber policeman, and cells and supernatant were collected on ice and precipitated overnight with cold 10% trichloracetic acid (TCA) in the presence of 0.02% unlabeled proline and 4 mg/ml bovine serum albumin (Cohn fraction V, Sigma, St. Lois, MO). After washing 5 times in 5% TCA and 0.01% proline at 4° C, the precipitate was dissolved in 0.2 N NaOH, and exposed for 1 hr at 37° C in Tris-CaCl₂ buffer (0.05 M Tris-HCl at pH 7.6, 0.005 M CaCl₂, 0.02% NaN₃) either to 125 U/50 ml of bacterial collagenase (Type III, Sigma, St. Lois, USA), to pronase (Calbiochem, La Jolla, CA), or to buffer without enzyme. All incubations were done in the presence of N-ethelmaleimide (2.5 mM). Tubes were then placed on ice and again precipitated for 1 hr with 10% TCA and 0.5% tannic acid. The precipitate was discarded and a 0.5 ml aliquot of scintillation counting fluid. Samples were counted in a Beckman scintillation counter for 2 min.

Statistical analysis

Student's unpaired t-test was used to determine whether the observed differences in experiments were statistically significant.

Results

Mast cell adhesion to endothelial cells and fibroblasts

The time courses of MC adhesion to EC (Fig. 1a) and FB (Fig. 1b) were examined. The cpm/well showed time-dependent and significant increase compared to that of addition time in both EC and FB. 6 hrs after the addition, it almost reached a plateau in both EC and FB. Dose dependent adhesion of MC to EC (Fig. 1c) and FB (Fig. 1d) were examined. The adhesion showed significant dose dependent increase in both EC and FB.

The alteration of endothelial cell and fibroblast proliferation by mast cells and their derivatives

The ³H-Thymidine uptake of EC showed a dose dependent decrease when sonicated MC and MC supernatant were added (Fig. 2a). The ³H-Thymidine uptake of FB showed dose-dependent increase when cultured with MC, but showed no significant change when cultured with sonicated MC and MC supernatant (Fig. 2b).

The cytotoxic activity of mast cells to endothelial cells and fibroblasts

In a 4 hrs ⁵¹Cr release assay to EC, MC, MC supernatant and sonicated MC induced dose dependent cytotoxic activity to EC (Fig. 3a). The specific percent lysis of EC with MC addition were 37.0 ± 5.2 (X20) and 24.0 ± 3.9 (X10). The lysis with MC supernatant addition were 12.2 ± 15.3 (20%) and 8.1 ± 2.4 (10%), and those with sonicated MC addition 26.0 ± 8.2 (20%) and 29.3 ± 4.5 (10%). Trypsin inhibitor (TI) (Fig. 3b) inhibited the cytotoxicity of MC derivatives to EC. The percent specific lysis with TI was only 0.5 ± 4.0. The lysis with MC (83.6 ± 7.5) was inhibited to 13.3 ± 1.4 with the addition of TI. The lysis with MC supernatant (79.0 ± 5.0) was inhibited to 22.1 ± 2.0, and those with sonicated MC (77.0 ± 1.6) was inhibited to 17.3 ± 1.6 with the addition of TI, respectively. In a 4 hrs ⁵¹Cr release assay to FB, MC induced a negligible cytotoxicity (0.23-0.80% cell lysis), while sonicated MC material induced 9.74 to 13.34% specific cell lysis (Fig. 3c), all of which showed no significant difference among them.

The effect of mast cells on collagen and protein synthesis in fibroblasts

FB added with MC and MC supernatant showed increased collagen and protein synthesis (Fig. 4). Of note was the result that FB added MC showed 9.95 times collagen synthesis and 11.0 times protein synthesis compared with FB with medium only, respectively. The ratios of collagen/protein synthesis in medium only, MC supernatant and MC coculture showed no significant difference.

Discussion

The possible interaction of MC with EC and FB were studied with four kinds of in vitro experiments: the adhesion of MC to EC and FB, the effects of MC to proliferation of EC and FB, cytotoxic activity of MC to EC and FB, and the effect of MC to collagen and protein synthesis in FB. The results showed different effects of MC interaction on EC and FB. The effect of MC on EC showed a proliferation inhibiting and a cytotoxic effect. On the other hand, the effect of MC on FB showed a proliferation stimulating and a collagen/protein synthesis stimulating effect.

Dose and time dependent MC adhesion to EC and FB were found. MC have been known to adhere to FB [20, 21]. The results of this study showed that MC adhere to EC as well as FB. The critical role of MC in regulating the expression of EC adhesion molecules, ICAM-1 and VCAM-1, have been indicated [22]. The incubation of human dermal microvascular endothelial cells and human umbilical vein endothelial cells with activated MC or MC conditioned medium (MCCM) markedly increased ICAM-1 and VCAM-1 surface expression, time and dose dependently. The ICAM-1 and VCAM-1 upregulation by MCCM was neutralized by antibody to tumor necrosis factor alpha (TNF-alpha). It has been demonstrated that mucosal-type MC lines express functional alpha4 integrins that can mediate adhesion to VCAM-1 and mucosal adressin cell adhesion molecule-1 (MAdCAM-1) [23]. Rat mucosal-type MC lines expressed high levels of alpha4 integrins on their surface and bound to CHO cells transfected with VCAM-1 or MAdCAM-1. Anti-alpha4 mAbs inhibited the specific adhesion of the MC to VCaM-1 or MAdCAM-1. MC, MC supernatant and sonicated MC inhibited the proliferation of EC. These inhibitions were induced by the cytotoxic activity of MC to EC. Addition of MC and sonicated MC showed significantly higher cytotoxicity than MC supernatant. These results indicated that membrane bound molecules or proteases in MC granules induced the cytotoxicity. The cytotoxicity was dramatically inhibited by trypsin inhibitor, results which indicated that the cytotoxicity was mediated by some kind of protease in granules, not by membrane bound molecules. MC are known to contain many kinds of protease in their granules [24-27]. Co-culture of rat peritoneal MC and bovine EC showed dose and time dependent inhibition of EC growth [28]. MC also generate a granule-associated molecule similar to TNF [29, 30]. MC express spontaneous cytotoxicity in vitro against the TNF-sensitive murine fibrosarcoma line WEHI-164, and this cytolytic activity is inhibited by the addition of antibodies to TNF [29, 30]. Accordingly, some kinds of protease contribute to cytotoxic activity of MC to EC.

The cytotoxicity of MC to EC is in contrast to the protective activity of MC to FB [31, 32]. Collagen and protein synthesis of FB was activated by MC membrane bound molecules, because the addition of MC, not MC supernatant showed a remarkable increase. A study reported that tryptase increased the proliferation and type I collagen production of human dermal FB [33]. The increase of FB proliferation and the production of type I collagen by the FB with tryptase was significantly reduced by antitryptase IgG antibody. The number of FB does not increase if FB were exposed to lysates of MC, or to MC derived conditioned medium, or if the two cell types are separated from one another [31]. This in vitro increase of FB is dependent on the number of MC initially seeded with the FB. The extracellular domain of W (c-kit) receptors was shown to be necessary for the attachment of MC to FB [21]. It has been indicated that human MC attach to FB independently of beta 1- or alpha v-integrins as well as of c-kit receptor-mediated mechanisms [34]. MC attached to collagen I and fibronectin, laminin, collagen IV and vitronectin, but not to collagens III and VI or hyaluronic acid. Adhesion to fibronectin, collagen I and laminin was completely inhibited by mAbs blocking beta 1-integrins, whereas adhesion of MC to vitronectin was inhibited by anti-alpha v-chain mAbs. FB induce the accumulation of IL-3 mRNA in connective tissue MC [32], which may mean that the production of IL-3 plays a role in the survival of this type of MC on the FB monolayer. MC activation has been reported to enhance significantly 3T3 FB proliferation and collagen production, which indicated a direct MC involvement in the fibrotic process [35]. Accordingly, these deferences induce a different activity of MC between EC and FB.

MC have been implicated in the pathogenesis of a number of chronic inflammatory conditions such as keloids, hypertrophic scars and chronic graft-versus-host disease (GVHD). Double immunofluorescent staining of MC for the presence of surface IgE receptors and cytoplasmic granules (avidin) revealed IgE receptor-bearing cells that lacked avidin-binding granules at the time when MC were not apparent on light microscopy [2]. By electron microscopy, reappearing MC have the morphology of immature dermal MC. MC were studied during the induction of chronic GVHD induced in mice [36]. Serial skin biopsies were taken over 26 days, during which time changes occurred resembling SSc, dermal fibrosis and mononuclear cell infiltrate. Ultrastructural analysis showed that MC in GVHD skin were present but underwent degranulation [36]. Some MC showed only pale expanded sacks, indicating granule depletion. Cellular activation occurred in many GVHD MC as shown by increased cytoplasmic activity, with numerous Golgi complexes, ribosomes, granular endoplasmic reticulum, and small vesicles. It has been reported that MC stimulated 3T3 fibroblast migration and proliferation into an in vitro model of a wound, obtained by producing a midline cut in a confluent 3T3 monolayer, and by scraping the cells from half of the monolayer [37]. MC have been shown to participate in wound healing by synthesizing and releasing basic fibroblast growth factor (bFGF) [38]. In active stages of granulation tissue, histologically characterized by prominent capillary proliferation, large numbers of bFGF-positive macrophages and MC were located within granulation tissue. Histamine involvement in the pathogenesis of chronic GVHD has been reported [11]. PBMC supernatants from chronic GVHD patients displayed histamine releasing activity when cocultured with rat MCs. Skin biopsies of involved areas in chronic GVHD patients revealed significantly reduced numbers of MC and showed signs of MC degranulation compared with biopsies from controls. Immunocompetent cell supernatants from humans with chronic GVHD increased basal histamine release by MCs and reduced FB proliferation. These findings indicated that complex interactions between immunocompetent cells, MCs, and FB probably play a role in chronic GVHD pathogenesis.

Several reports indicated the involvement of MC in the formation of microvascular and connective tissue abnormalities of SSc skin tissues [3-7]. MC counts in clinically involved skin of patients with early stages of SSc were significantly greater than those in clinically uninvolved skin of the same patients [3, 4]. The number of MC was increased in both involved and uninvolved skin and in both early and late disease [5]. An 85% increase of MC was noted in involved papillary dermis and a 152% increase in involved reticular dermis in patients with early SSc [3]. MC increased in the papillary dermis with fine collagen bundles and decreased in the papillary dermis with homogeneous collagen bundles [4]. There was an increase in the number of degranulated MC in the involved skin of patients with both early and late disease and in the not-yet-involved skin of patients with early disease. Increases in MC number and degranulation precede clinically apparent dermal fibrosis in SSc [5]. MC_T cells were present in 12 specimens and comprised between 8% and 100% of the total MC [6]. Extracellular tissue deposits of tryptase-positive and/or chymase-positive granular material were observed in 8 specimens, suggesting possible MC degranulation. E-selectin was overexpressed in SSc, and TNFalpha was detected in MC [8].

Vascular abnormalities and tissue fibrosis are characteristic of a number of chronic inflammatory conditions that result in vascular injury and fibrosis including keloids, hypertrophic scars, SSc, and chronic GVHD. The bulk of accumulated evidence suggests a role for cell mediated immune mechanisms in the pathogenesis. The result of this study suggests the presence of MC effect on EC damage and FB stimulation. It is likely that the immune process is relentlessly aimed at the destruction of microvessels leading to the clinically recognized state of chronic organ ischemia and tissue under perfusion.

Article accepted on 28/10/99

REFERENCES

1. Fernex M. The mast cell system: its relation to atherosclerosis, fibrosis and eosinophils. Williams and Wilkins, Baltimore, 1968.

2. Choi KL, Giorno R, Claman HN. Cutaneous mast cell depletion and recovery in murine graft-versus-host disease. J Immunol 1987; 135: 4093-101.

3. Hawkins RA, Claman HN, Clark RA, Steigerwald JC. Increased dermal mast cell populations in progressive systemic sclerosis: a link in chronic fibrosis? Ann Intern Med 1985; 102: 182-6.

4. Nishioka K, Kobayashi Y, Katayama I, Takijiri C. Mast cell numbers in diffuse scleroderma. Arch Dermatol 1987; 123: 205-8.

5. Seibold JR, Giorno RC, Claman HN. Dermal mast cell degranulation in systemic sclerosis. Arthritis Rheum 1990; 33: 1702-9.

6. Irani AM, Gruber BL, Kaufman LD, et al. Mast cell changes in scleroderma. Presence of MC_T cells in the skin and evidence of mast cell activation. Arthritis Rheum 1992; 35: 933-9.

7. Hebbar M, Lassalle P, Janin A, et al. E-selectin expression in salivary endothelial cells and sera from patients with systemic sclerosis. Role of resident mast cell-derived tumor necrosis factor alpha. Arthritis Rheum 1995; 38: 406-12.

8. Campbell PM, LeRoy EC. Pathogenesis of systemic sclerosis: a vascular hypothesis. Arthritis Rheum 1975; 4: 351-68.

9. Kahaleh MB, Sherer GK, LeRoy EC. Endothelial injury in scleroderma. J Exp Med 1979; 149: 1326-35.

10. Claman HN. On scleroderma. Mast cells, endothelial cells, and fibroblasts. JAMA 1989; 262: 1206-9.

11. Levi-Schaffer F, Segal V, Barak V, Rubinchik E, Nagler A. Regulation of the functional activity of mast cells and fibroblasts by mononuclear cells in murine and human chronic graft-vs-host disease. Exp Hematol 1997; 25: 238-45.

12. Buckingham RB, Prince RK, Rodnan GP, Taylor F. Increased collagen accumulation in dermal fibroblast cultures from patients with progressive systemic sclerosis (scleroderma). J Lab Clin Med 1978; 92: 5-21.

13. LeRoy EC. Increased collagen synthesis by scleroderma skin fibroblasts in vitro: a possible defect in the regulation or activation of the scleroderma fibroblast. J Clin Invest 1984; 54: 880-9.

14. Jaffe EA, Nachman RL, Becker CG, Minick CR. Culture of human endothelial cells derived from umbilical veins. Identification by morphological and immunological criteria. J Clin Invest 1973; 52: 2745-56.

15. Razin E, Cordo-Cardo C, Good RA. Growth of a pure population of mouse mast cells in vitro with conditioned medium derive from concanavalin A-stimulated splenocytes. Proc Natl Acad Sci USA 1981; 78: 2559-61.

16. Thompson-Snipes L, Dhar V, Bond MW, et al. Interleukine-10: a novel stimulatory factor for mast cells and their progenitors. J Exp Med 1991; 173: 507-10.

17. Dastych J, Costa J, Thompson HL, Metcalfe DD. Mast cell adhesion to fibronectin. Immunology 1991; 73: 478-84.

18. Kotasek D, Vercellotti M, Ochoa AC, et al. Mechanism of cultured endothelial injury induced by lymphokine-activated killer cells. Cancer Research 1988; 48: 5528-32.

19. Peterkofsky B, Diegelmann R. Use of a mixture of proteinase-free collagenases for the specific assay of radioactive collagen in the presence of other proteins. Biochemistry 1971; 10: 988-94.

20. Kaneko Y, Takenawa J, Yoshida O, et al. Adhesion of mouse mast cells to fibroblasts: adverse effect of stee. (Sl) mutation. J Cell Physiol 1991; 147: 224-30.

21. Adachi S, Ebi Y, Nishikawa S, et al. Necessity of extracellular domain of W (c-kit) receptors for attachment of murine cultured mast cells to fibroblasts. Blood 1992; 79: 650-6.

22. Meng H, Tonnesen MG, Marchese MJ, Clarke RA, Bahou WF, Gruber BL. Mast cells are potent regulators of endothelial cell adhesion molecules ICAM-1 and VCAM-1 expression. J Cell Physiol 1995; 165: 40-53.

23. Palecanda A, Briskin MJ, Issekutz TB. Rat mast cell lines bind to the vascular cell adhesion molecule-1 (VCAM-1) and the mucosal addressin cell adhesion molecule-1 (MAdCAM-1). J Immunol 1997; 158: 2904-10.

24. Ghildyal N, McNeil HP, Stechschulte S, et al. IL-10 induces transcription of the gene for mouse mast cell protease-1, a serine protease preferentially expressed in mucosal mast cells of Trichinella spiralis-infected mice. J Immunol 1992; 149: 2123-9.

25. Ghildyal N, McNeil HP, Gurish MF, et al. Transcriptional regulation of the mucosal mast cell-specific protease gene, MMCP-2, by interleukine-10 and interleukine-3. J Biol Chem 1992; 267: 8437-73.

26. McNeil HP, Reynolds DS, Schiller V, et al. Isolation, characterization, and transcription of the gene encoding mouse mast cell protease 7. Proc Natl Acad Sci USA 1992; 89: 11174-8.

27. Sommerrhoff CP, Ruoss SJ, Caughey GH. Mast cell proteoglycans modulate the secretagogue, proteoglucanase, and amidolytic activities of dog mast cell chymase. J Immunol 1992; 148: 2859-66.

28. Ryan US, Clements E, Habliston D, Ryan JW. Isolation and culture of pulmonary artery endothelial cells. Tissue Cell 1987; 10: 535-54.

29. Tharp MD, Kasper C, Thiele D, et al. Studies of connective tissue mast cell-mediated cytotoxicity. J Invest Dermatol 1989; 93: 423-8.

30. Young JD, Liu CC, Butler G, et al. Identification, purification, and characterization of a mast cell-associated cytolytic factor related to tumor necrosis factor. Proc Natl Acad Sci USA 1987; 84: 9175-9.

31. Dayton ET, Caulfield JP, Hein A, et al. Regulation of the growth rate of mouse fibroblasts by IL-3-activated mouse marrow-derived mast cells. J Immunol 1989; 142: 4307-13.

32. Razin E, Leslie KB, Schrader JW. Connective tissue mast cells in contact with fibroblasts express IL-3 mRNA. An analysis of single cells by polymerase chain reaction. J Immunol 1991; 146: 981-7.

33. Abe M, Kurosawa M, Ishikawa O, Miyachi Y, Kido H. Mast cell tryptase stimulates both human dermal fibroblast proliferation and type I collagen production. Clin Exp Allergy 1998; 28: 1509-17.

34. Trautmann A, Feuerstein B, Ernst N, Brocker EB, Lein CE. Heterotypic cell-cell adhesion of human mast cells to fibroblasts. Arch Dermatol Res 1997; 289: 194-203.

35. Levi-Schaffer F, Rubinchik E. Activated mast cells are fibrogenic for 3T3 fibroblasts. J Invest Dermatol 1995; 104: 999-1003.

36. Claman HN, Choi KL, Sujansky W, Vatter AE. Mast cell "disappearance" in chronic murine graft-vs-host disease (GVHD)-ultrastructural demonstration of "phantom mast cells". J Immunol 1986; 137: 2009-13.

37. Levi-Schaffer F, Kupietzky A. Mast cells enhance migration and proliferation of fibroblasts into an in vitro wound. Exp Cell Res 1990; 188: 42-9.

38. Murata M, Hara K, Saku T. Dynamic distribution of basic fibroblast growth factor during epulis formation: an immunohistochemical study in an enhanced healing process of the gingiva. J Oral Path Med 1997; 26: 224-32.