Transforming growth factor β1 (TGF-β1) is an immunoregulatory cytokine constitutively expressed in the human gut by various tissue cells, such as epithelial cells and different lamina propria cells.¹ It acts as an anti-inflammatory cytokine by modulating T cell and dendritic cell function (for example, by induction of regulatory T cells and IgA production in plasma cells, but also by controlling inflammatory effector cells such as macrophages and eosinophils).^2–4 On the other hand, TGF-β1 is involved in regulating tissue repair processes by affecting epithelial cells, fibroblasts, and by inducing extracellular matrix production.⁵ This dual function of TGF-β1 is confirmed by in vivo experiments using genetically modified animal models. TGF-β1 and TGF-β receptor II (TGF-RII) gene knockout mice develop a lethal multifocal inflammation involving several organs, including the intestine,^6,7 whereas mice over expressing TGF-β1 in different organs develop tissue fibrosis.^8,9

Mast cells (MC) are known to be involved in both processes. They act as inflammatory effector cells releasing mediators such as histamine, eicosanoids, and tumour necrosis factor α (TNF-α) in the course of inflammatory conditions (for example, allergic reactions, parasitic infections, and inflammatory bowel disease) and, as shown recently, in the irritable bowel syndrome.^10–12 Secondly, MC accumulate at sites of wound healing and tissue fibrosis.^10,11,13 In the normal human gut they are primarily located in the mucosa where they represent approximately 2–3% of all lamina propria cells. At sites of inflammation or fibrosis, their number may be substantially increased and their distribution may expand to other layers of the bowel wall such as the epithelial layer or the muscularis propria.^11,13

Regulation of human MC growth and mediator release is dependent on cytokines such as stem cell factor (SCF), interleukin (IL)-3, -4, and others, which either alone or in combination promote MC growth and maturation, and enhance mediator release in response to IgE dependent and IgE independent triggering agents.^10,14–18 However, our knowledge of endogenous inhibitors of human MC function is poor. In the murine system, TGF-β1 has been proposed as an inhibitor of MC growth.^19–22 Whether TGF-β1 also regulates MC mediator release is unclear as experiments in mice yielded conflicting results.^21–24 In humans, no data are available as yet but TGF-β1 may also affect human MC functions, as reported by Olsson et al who found that TGF-β1 has chemotactic and growth inhibitory effects in the immature human leukaemic mast cell line HMC-1 and in human cord blood derived MC.²⁵ The reason for the inconsistent data about TGF-β1 effects on MC may be that different types of MC derived from different species, from different body sites, and of different maturity grades were used in previous experiments. Here we examined for the first time the effects of TGF-β1 on in vivo maturated human MC isolated from intestinal tissue and purified to homogeneity. The aim of the study was to define the function of TGF-β1 in regulating MC growth, protease, and surface receptor expression, and mediator release.

METHODS

Isolation, purification, and culture of human intestinal MC

MC were isolated from surgery tissue specimens (small and large intestine, bowel resection because of cancer, macroscopically normal intestinal tissue) using a four step enzymatic dispersion method, and subsequently purified by positive selection of c-kit expressing cells using magnetic cell separation (MACS system, Miltenyi Biotech, Bergisch-Gladbach, Germany), as described previously.^15,16 MC (purity 60±35%) were cultured at a density of 1–2×10⁵ cells/ml in medium supplemented with recombinant human SCF (50 ng/ml; Amgen, Thousand Oaks, California, USA), IL-4 (2 ng/ml; Novartis, Vienna, Austria), activated TGF-β1 (5 ng/ml; PeproTech, Rocky Hill, New Jersey, USA), or with a combination of these cytokines. To block TGF-β1 bioactivity, a neutralising anti-TGF-β1 monoclonal antibody (mAb) (clone 9016.2; R&D Systems, Minneapolis, Minnesota, USA) was added to the culture medium every fourth day. During the culture period, MC purity increased independent of the cytokines added from 60±35% to 98±2%, with fibroblasts remaining the only contaminating cells. MC purity and recovery (expressed as per cent of MC numbers at the start of the culture) were determined by cell counting using trypan blue staining (Sigma Chemicals, St Louis, Missouri, USA) and differentiation of cells on cytospin smears stained with May-Grünwald/Giemsa (Merck, Darmstadt, Germany). To confirm the MC recovery data, histamine content per well was measured by RIA (Coulter-Immunotech, Krefeld, Germany) after washing the cells and subsequent cell lysis.

Mediator release assays

Following culture, MC were washed and seeded at a density of 0.5 to 1×10⁵ cells/ml in culture medium without antibiotics or cytokines. MC were stimulated by IgE receptor (IgER) crosslinking using the mAb29C6 (100 ng/ml, 90 or 360 minutes; Hoffmann-La Roche, Nutley, New Jersey, USA). Histamine release into supernatants was measured by RIA and expressed as specific histamine release according to the following formula:

(histamine released after IgER crosslinking−histamine released spontaneously in response to buffer control)/total cellular histamine content as determined after cell lysis.

Cysteinyl-leukotrienes (cys-LT) were measured by RIA as described previously.^15,16 Prostaglandin D2 (PGD2) was measured by PGD2 methoxime EIA (Cayman Chemical Company, Ann Arbor, Michigan, USA) and TNF-α was measured by ultrasensitive ELISA (BioSource Europe, Nivelles, Belgium). Release of cys-LT, PGD2, and TNF-α into the supernatant was expressed as specific mediator release obtained by subtraction of mediator release in response to buffer control.

RNA isolation and reverse transcriptase-polymerase chain reaction (RT-PCR)

Total RNA was prepared using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and RT-PCR was carried out as described previously.^16,18 The following primer sequences were used: TGF-RI: 5′-AAT TCC TCG AGA TAG GCC GT-3′, 5′-TGC GGT TGT GGC AGA TAT AG-3′; TGF-RII: 5′-AGG GAC CTC AAG AGC TCC AAT AT-3′, 5′-CAC TGC ATT ACA GCG AGA TGA CA-3′; β-glycan: 5′-CCT GTC ATT CCC AGC ATA CAA CT-3′, 5′-ATC ACC TGA CAC CAG ATC TTC ATA-3′; endoglin: 5′-TGT CTC ACT TCA TGC CTC CAG CT-3′, 5′-AGG CTG TCC ATG TTG AGG CAG T-3′²⁶; cytosolic phospholipase A2 (cPLA2): 5′-ATT CTG GAT TGT GCT ACC TAC-3′, 5′-AAC CAG AAA CGC CCA AAA CTC-3′)²⁷; cyclooxygenase (COX)-1: 5′-TGC CCA GCT CCT GGC CCG CCG CTT-3′, 5′-GTG CAT CAA CAC AGG CGC CTC TTC-3′; COX-2: 5′-TTC AAA TGA GAT TGT GGG AAA ATT GCT-3′, 5′-AGA TCA TCT CTG CCT GAG TAT CTT-3′²⁸; haematopoietic prostaglandin synthase (PGDS): 5′-CTT CAC CAG AGC CTA GCA AT-3′, 5′-GGA TTA TCT GGC AGG CTG AT-3′)²⁹; and GAPDH: 5′-ACC ACA GTC CAT GCC ATC AC-3′; 5′-TCC ACC ACC CTG TTG CTG TA-3′ (all synthesised by Life Technologies, Eggenstein, Germany).

Immunocytochemistry

Immunocytochemistry was performed using mAbs against human tryptase (clone G3 (0.24 µg/ml), human chymase (0.24 µg/ml), both from Chemikon, Temecula, California, USA), the human nuclear cell proliferation associated antigen Ki-67 (clone MIB-1, dilution 1:50; Dianova, Hamburg, Germany), and appropriate isotype control mAbs (mouse IgG1; Southern Biotechnology, Birmingham, Alabama, USA) as primary antibodies (overnight incubation at 4°C) and the streptavidin-biotin detection system (Histostain-Plus kit; Zymed, San Francisco, California, USA). Percentage of apoptotic MC was assessed using the terminal deoxynucleotidyl transferase mediated dUTP nick end labelling (TUNEL) in situ cell death detection assay for immunocytochemistry (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer’s instructions.

Flow cytometry

For each condition, 2.5–5×10⁴ MC (purity 98±2%) were washed (400 g, five minutes) and suspended in 100 µl phosphate buffered saline supplemented with 0.1% bovine serum albumin, 0.1% sodium acid, 250 µg/ml rabbit IgG, and mAbs directed against TGF-RII (2.5 µg/ml; R&D Systems), c-kit (clone YB5.B8, 2.5 µg/ml; Pharmingen), IgER α-chain (clone 29C6, 2.5 µg/ml), or the appropriate isotype control mAbs (goat IgG (Dianova) and mouse IgG1 (R&D Systems)). After incubation for 30 minutes at 4°C, cells were washed twice and stained for 30 minutes at 4°C with FITC conjugated secondary antibodies (Southern Biotechnology). Apoptotic cells were visualised using FITC conjugated annexin V/propidium iodide double staining (R&D Systems) according to the manufacturer’s instructions.

Western blot

To obtain whole cell extracts, 0.5×10⁶ MCs were lysed in extraction buffer containing 25 mM Tris HCl, pH 7.5, 0.5 mM EDTA, 0.5 mM EGTA, 0.05% Triton X-100, 10 mM β-mercaptoethanol, supplemented with the protease inhibitor cocktail Complete Mini (Roche Diagnostics, Germany). Cell extracts (10–20 µg protein each) were separated on a 12% sodium dodecyl sulphate (SDS) polyacrylamide gel and blotted onto a nitrocellulose membrane (Schleicher and Schuell, Einbeck, Germany) in 0.1% SDS, 20% methanol, 400 mmol/l glycine, 50 mmol/l Tris HCL, pH 8.3, at 4°C for four hours at 40 V by electroblotting using the Trans-Blot Cell system (Bio-Rad Laboratories, Munich, Germany). Membranes were blocked with 5% skimmed milk in phosphate buffered saline containing 0.1% Tween overnight. Membranes were probed with anti-COX1, anti-COX2, anti-PGDS (all Cayman Chemicals), anti-cPLA2 (Upstate Biotech, Lake Placid, New York, USA), and anti-actin Ab (Santa Cruz, California, USA). The antigen-antibody complexes were visualised using an electrochemiluminescence detection system, as described by the manufacturer (NEN Life science, Boston, Massachusetts, USA). To confirm that equal amounts of protein were used for the western blots of COX-1 and 2, membranes were stripped (Restore western blot stripping buffer; Perbio, Rockford, Illinois, USA) for 30 minutes at 37°C and probed again with the anti-actin Ab.

Statistics

All data are expressed as mean (SD) if not indicated otherwise. Significance of differences was assessed using the Wilcoxon signed rank test. A p value of <0.05 was considered statistically significant.

RESULTS

Expression of TGF-β receptors in human intestinal MC

Using RT-PCR, we detected expression of transcripts for TGF-RI, TGF-RII, and the accessory TGF-R endoglin in freshly purified (purity >95%, not shown) as well as in cultured MC (fig 1A). TGF-RII protein expression was confirmed by flow cytometry (fig 1B). IL-4 supplementation to the culture medium decreased the amount of endoglin mRNA whereas IgER crosslinking induced by incubation with mAb29C6 for 90 minutes did not affect expression of any TGF-R mRNAs.

TGF-β1 inhibits growth of cultured human intestinal MC

Addition of TGF-β1 to the culture medium caused a marked decrease in MC recovery from 228.9 (159.4)% to 88.8 (88.4)% in the presence of SCF, and from 449.3 (260.5)% to 264.9 (241.8)% in the presence of SCF and IL-4 (fig 2A). Histamine recovery was reduced in a similar fashion (fig 2B).

TGF-β1 added to the culture medium caused a marked reduction in MC proliferation, as assessed by anti-Ki-67 staining. In accordance with our previous studies,^16,18 we found 12.4 (9)% Ki-67⁺ MC in the presence of SCF and 21.3 (18.0)% in the presence of SCF and IL-4. TGF-β1 almost abolished MC proliferation when added to cultures containing SCF (1.0 (0.9)% Ki-67⁺ cells) or SCF and IL-4 (6.0 (1.4)%) (fig 3A). Light microscopy analysis of May-Grünwald/Giemsa stained cytospin preparations revealed an increase in MC displaying morphological alterations associated with apoptosis (for example, decreased cell size, increased cell density, nuclear condensation) when cells were cultured in the presence of TGF-β1 (not shown). TUNEL assay experiments performed with cytospin preparations, as well as annexin V and propidium iodide double staining for flow cytometry in order to detect apoptotic MC, confirmed this observation (fig 3B, C).

The growth inhibitory effect of TGF-β1 appeared in a concentration dependent fashion, with maximal inhibition at approximately 3 ng/ml and an ED₅₀ of approximately 0.3 ng/ml (n = 2) (fig 4A). Possible toxic effects of exogenous TGF-β1 were largely excluded using an anti-TGF-β1 mAb at different concentrations that abolished the inhibitory effect of TGF-β1 on MC growth (fig 4B). The observation that maximal effective concentrations of the anti-TGF-β1 mAb were not able to further augment MC recovery compared with SCF control conditions indicated that no endogenous TGF-β1, either MC or serum derived, accounted for the effects we observed.

TGF-β1 increases the frequency of chymase expressing MC

Two human MC subtypes have been defined according to their protease content: the tryptase positive, chymase negative MC (MC_T) and the tryptase and chymase double positive MC (MC_TC).^10,30 We analysed these subtypes by immunocytochemistry using antibodies directed against the two proteases after culture of the MC with different cytokines (n = 5). Independent of the cytokines added to the culture medium, virtually all MC expressed tryptase. Confirming previous data, chymase expression was downregulated by IL-4 from 35.8 (20.9)% of MC expressing the MC_TC phenotype in the presence of SCF to 16.3 (7.7)% MC_TC in the presence of SCF and IL-4.¹⁶ However, TGF-β1 significantly increased the proportion of chymase positive MC to 52.5 (21.6)% in the presence of SCF, and to 33.3 (10.9)% in the presence of SCF and IL-4, respectively (fig 4C).

TGF-β1 changes the mediator release profile in human intestinal MC

TGF-β1 not only affected MC growth but also mediator release induced by mAb29C6, causing IgER crosslinking. When added to the culture medium, TGF-β1 decreased specific histamine release from 37.2 (15.1)% to 10.6 (11.9)% when MC were cultured with SCF and from 70.5 (16.0)% to 33.3 (20.7)% when MC were cultured with SCF and IL-4 (fig 5A). Similar results were obtained with respect to cys-LT and TNF-α production that was dramatically decreased in MC cultured with TGF-β1 (fig 5B, D). In contrast, TGF-β1 significantly increased PGD2 production from 23.3 (15.0) to 40.9 (25.5) ng/10⁶ MC (SCF) and from 23.1 (21.4) to 42.0 (31.1) ng/10⁶ MC (SCF and IL-4, fig 5C).

TGF-β1 reduces expression of receptors for IgE and SCF in human intestinal MC

We also studied expression of receptors involved in MC regulation, such as c-kit and IgER, in MC cultured with and without TGF-β1. As shown in fig 6, TGF-β1 strongly downregulated expression of the SCF receptor c-kit (n = 3). Furthermore, TGF-β1 slightly reduced expression of the α chain of the high affinity IgER (fig 6).

TGF-β1 promotes expression of COX-1 and COX-2 in human intestinal MC

To study the mechanism of TGF-β1 induced enhancement of PGD2 production in MC triggered by IgER crosslinking, we analysed expression of key enzymes regulating PGD2 synthesis by semiquantitative RT-PCR and western blot analysis. We detected mRNAs encoding for cPLA2, COX-1, COX-2, and PGDS in MC cultured for 14 days in the presence of SCF (data not shown). IgER crosslinking induced by mAb29C6 for 90 minutes caused upregulation of COX-2 mRNA whereas mRNAs encoding for other enzymes remained unchanged (data not shown). Consistent with PGD2 release data, the presence ofTGF-β1 during MC culture enhanced expression of all four enzymes known to regulate PGD2 synthesis in activated MC (fig 7A). For COX-1 and COX-2, mRNA data were confirmed at the protein level by western blot (fig 7B) but the amounts of cPLA2 and PGDS proteins were not altered by TGF-β1.

DISCUSSION

In the present study, we have provided for the first time evidence that TGF-β1 controls human intestinal MC functions. Our data show that in the presence of TGF-β1, MC growth is retarded, and the profile of mediators released from MC triggered by IgER crosslinking is modulated. Interestingly, we found that many classical MC mediators, such as histamine, cys-LTs, and TNF-α, thought to be responsible for the proinflammatory effects of MC, were released to a lesser extent in the presence of TGF-β1, while only one mediator, PGD2, was upregulated following stimulation of MC by IgER crosslinking. The observation that TGF-β1 suppresses human intestinal MC functions in vitro agrees with previous reports indicating that TGF-β1 is a unique anti-inflammatory cytokine constitutively expressed in the intestine,¹ most likely to maintain ongoing inflammatory reactions triggered by bacteria and other luminal antigens at low levels. This concept was supported by the finding that deletion of TGF-β1 and TGF-RII genes caused severe colitis, in which MC, as well as many other cells, may be involved.^6,7 Obviously, TGF-β1 controls the mucosal immune system not only by inducing IgA production and generating T_H3 regulatory cells but also by controlling inflammatory effector cells such as macrophages,^2,3 eosinophils,⁴ and MC, as shown here, which were all found in the normal lamina propria of the human intestine.

Previous reports on the effects of TGF-β1 in MC were mostly restricted to the murine system.^19–24 Such data cannot be transferred easily to the human system because of the well documented heterogeneity of MC derived from different species and body sites.¹⁰ Moreover, most studies only examined TGF-β1 effects on MC proliferation, not on mediator release, the main effector function of MC. Here, we have extended these data by demonstrating that the recovery rate of tissue derived human MC was strongly reduced in the presence of TGF-β1. The growth inhibitory effect of TGF-β1 was accompanied by both a reduction in MC proliferation and enhancement of MC apoptosis, and appeared in a concentration range similar to TGF-β1 levels found in human plasma.³¹ Such TGF-β1 effects have been reported previously for epithelial, endothelial, and haematopoietic cells, including eosinophils and murine MC, in all of which TGF-β1 causes cell cycle arrest.^3,22 Induction of apoptosis by TGF-β1 was also found in human lymphoma cell lines, leukaemia cells, and normal blood eosinophils.^3–5 The mechanism of MC growth inhibition by TGF-β1 is not clear.^19,20 According to our data, it is likely that TGF-β1 exerts its effect on human intestinal MC to a large extent by downregulating c-kit, the receptor for SCF, the unique mast cell growth factor in rodents and humans.^10,11 Finally, our data agree with the clinical observation that some mastocytosis patients have TGF-RI negative MC, a condition that has been associated with a poorer prognosis compared with those having TGF-RI positive MC.³²

Previous studies on the effect of TGF-β1 on mediator release yielded conflicting results. It was reported that TGF-β1 did not affect histamine, beta-hexosaminidase, or cys-LT release following IgE dependent activation,^21,22 that it inhibited antigen induced release of histamine and TNF-α dependent cytotoxicity,²³ or that it even enhanced IgE dependent cys-LT production without affecting beta-hexosaminidase release in murine MC.²⁴ Here, we investigated for the first time the effects of TGF-β1 on mediator release in mature human MC. We found that TGF-β1 switches the normal MC response to IgER crosslinking from unselective release of many inflammatory mediators to selective release of PGD2. MC releasing preferentially PGD2 has not been reported previously, and may contribute to the known immunomodulatory effects of TGF-β1 as PGD2 is the ligand of the recently discovered receptor CRTH2 which selectively induces chemotaxis in T_H2 cells, basophils, and eosinophils.³³

To address possible mechanisms causing selective upregulation of PGD2 production in activated MC pretreated with TGF-β1, we studied expression of IgER α chain and key enzymes known to be involved in PGD2 synthesis. We found slight downregulation of IgER α chain expression following culture with TGF-β1 that might contribute to the fact that release of mediators such as histamine, cys-LT, and TNF-α was reduced by TGF-β1. The TGF-β1 induced increase in PGD2 production in MC activated by IgER crosslinking seems to be directly related to enhanced expression of COX-1 and 2 mRNA and protein following TGF-β1 pretreatment of MC. cPLA2 and PGDS may also contribute to TGF-β1 dependent enhancement of PGD2 production as mRNAs encoding for these enzymes were enhanced in MC challenged with TGF-β1 although protein levels remained unchanged within an observation period of 90 minutes. Such a time period may be too short and, therefore, does not exclude the possibility that increased cPLA2 and PGDS protein levels occur at later time points (for example, following repetitive stimulation of MC for hours to days).

Interestingly, TGF-β1 not only affected growth, survival, and mediator release but also the phenotype of MC. Addition of TGF-β1 to human MC cultures enhanced the percentage of MC expressing both tryptase and chymase (MC_TC). This MC subtype, also named “connective tissue-type MC”, is found mainly in the skin and intestinal submucosa. In contrast, MC expressing tryptase but no chymase (MC_T), named “mucosa-type MC”, that predominate in the lung and intestinal mucosa,^10,30 were reduced by TGF-β1. Presumably, TGF-β1 induced chymase expression rather than selective apoptosis in MC_T because it was shown that TGF-β1 induced de novo expression of mouse MC protease 1, a chymase-like protease, in developing murine bone marrow derived MC.³⁴ Such findings strongly suggest that mature human tissue MC maintain the capacity to change their phenotype depending on exogenous factors such as TGF-β1 or IL-4, which have opposite effects in this respect. Consistent with our in vitro findings, MC that accumulate in TGF-β1 rich environments, such as intestinal strictures, predominantly belong to the MC_TC subtype.^11,13 Finally, the profile of mediators we observed after culture of MC in the presence of TGF-β1 may reflect alteration of the MC_T/MC_TC ratio in our MC populations.

Our data might have several clinical implications as TGF-β1 seems to be a unique regulator of the responsiveness of inflammatory cells such as MC towards different triggering agents. Proinflammatory stimuli such as lipopolysaccharide, TNF-α, and IL-1β, which are released during bacterial infection, were found to upregulate expression of Smad7, an intracellular protein inhibiting TGF-β1 signalling.³⁵ This in turn may facilitate activation of MC and other inflammatory cells required to initiate an innate immune response. In previous in vivo experiments, it has been clearly shown that MC activation is necessary for host defence against bacterial infections because their mediators, including TNF-α, allow the recruitment of neutrophils needed for bacterial clearance at sites of infection.³⁶ In the absence of inflammatory signals, TGF-β1, apart from its effects on the specific immune response, may limit innate immune responses unless they are required. This hypothesis would imply that breakdown of the TGF-β1 pathway leads to uncontrolled inflammation. Indeed, Monteleone et al reported such a condition recently by showing that Smad7 is upregulated in inflammatory bowel disease.^37,38 Whereas reduced levels of TGF-β1 or impaired TGF-β1 signalling may lead to uncontrolled inflammation, conditions with normal TGF-β1 levels are likely characterised by a controlled tissue MC number and activity, and by support of a “T_H2 milieu” caused by preferential release of PGD2 in MC following activation. In contrast, enhanced TGF-β1 levels may occur during chronic inflammation, possibly in the sense of some form of negative feedback mechanism, paralleled by ongoing tissue repair processes. This may contribute to fibrotic tissue transformation typically occurring in the course of chronic inflammation (for example, in Crohn’s disease, where bowel fibrosis and strictures cause serious clinical problems). Clearly, further studies are necessary to confirm the pathophysiological role of impaired TGF-β1 activity at mucosal sites.