Presentation of IFN-γ to Nitric Oxide-Producing Cells: A Novel Function for Mast Cells¹

The T cell- and NK cell-derived cytokine IFN-γ exerts a wide range of immune-regulatory and inflammatory activities. These include promotion of Th1 cell differentiation, up-regulation of MHC molecules, activation of macrophages, and stimulation of expression of the inducible form (type II) of NO synthase (1, 2). IFN-γ may also influence immediate hypersensitivity reactions and other processes dependent on mast cells. For example, it inhibits IgE-mediated mast cell degranulation (3, 4, 5, 6), TNF-α-mediated cytotoxicity (7), and clonal proliferation of rodent connective tissue type mast cells (8). In cell culture systems employing density-gradient fractionated mast cells, it is uncertain whether the effects of IFN-γ are direct or occur via the small proportion of contaminating nonmast cells. In the case of IgE-mediated degranulation of mouse and rat peritoneal mast cells, the inhibitory action of the cytokine is certainly indirect and the active intermediate has been identified as NO (9, 10). IL-4 enhances degranulation of mouse and rat peritoneal mast cells (10, 11), and, at least in the rat, this is due to inhibition of NO synthesis by accessory cells (10).

Although IFN-γ does not cause purified rodent peritoneal mast cells to produce NO and does not inhibit degranulation by a direct action, it is not clear whether this is due to lack of mast cell expression of the IFN-γ receptor (IFN-γR). The IFN-γR is expressed on all nucleated cells studied to date. It is ligand and species specific, composed of three subunits of which the α-chain has high affinity for IFN-γ, and is linked to the Jak-STAT signal transduction pathway (1, 2).

Unexpectedly, we found that biotinylated IFN-γ bound equally well to IFN-γR knockout (KO)³ (3) and wild-type (WT) purified mast cells, indicating a unique binding site. One possible candidate is cell surface proteoglycans. In common with other cytokines and growth factors, human IFN-γ has been shown to bind to proteoglycans of the extracellular matrix, and more specifically to the glycosaminoglycans (GAGs) heparin, heparan sulfate, and chondroitin sulfate (12, 13, 14, 15, 16). The binding interaction between human IFN-γ and heparan sulfate has been well characterized: it involves two groups of carboxyl-terminal basic amino acids on the cytokine and two sulfated domains of heparan sulfate, such that the two domains of the GAG directly bind the two carboxyl-terminals of an IFN-γ dimer (13, 17, 18, 19). Although soluble GAGs can inhibit IFN-γ activity, IFN-γ bound to immobilized GAGs is biologically active (12, 14, 15, 16, 20, 21). In the present study, we examined two important questions pertaining to mast cell-bound IFN-γ. First, is the cell-bound cytokine functionally active? Second, what is the identity of the binding site? To test for functional activity, we used mouse peritoneal macrophages that respond to IFN-γ by induced expression of type II NO synthase and high level NO synthesis (22, 23). As a control, to examine whether the IFN-γ effect was direct, we used IFN-γR KO responder cells. The results showed that mast cell-bound IFN-γ did induce NO release by macrophages, and the effect was dependent on IFN-γR expression by the responder cells. Further studies, using GAG-selective enzymes, indicated that chondroitin sulfate B (dermatan sulfate) is the IFN-γ binding site on mast cells. These studies show that mast cells have the capacity to bind IFN-γ via GAGs and to present it in a functionally active form to NO-producing cells. This represents a novel cytokine-presenting function for mast cells. Since mast cells are resident in many tissues and are also recruited to inflammatory sites (24, 25), they may provide a site for cytokine sequestration and for cytokine transport into tissues. The mast cell may serve as a useful paradigm for further studies of cytokine sequestration and presentation by inflammatory cells.

IFN-γR KO and IFN-γR WT 129 Sv/Ev mice were bred from stock obtained from B & K Universal (Hull, U.K.) and were used with kind permission of Dr. S. Huang (Institute of Molecular Biology, Zürich, Switzerland) (26). Phenotype was checked by the capacity of peritoneal cells to respond to IFN-γ by generating nitrite.

Cells were obtained by peritoneal lavage of mice with HBSS (Life Technologies, Paisley, Scotland) and sedimented by centrifugation at 260 × g for 5 min. The cell pellets were pooled (typically from four to six mice), resuspended in 7.5 ml of 72.5% isotonic Percoll (Sigma, Poole, U.K.), and overlaid with 2 ml of complete DMEM (Life Technologies) containing 5% FCS (Life Technologies), 4 mM l-glutamine, and 50 μg/ml gentamicin (cDMEM). The gradient was centrifuged at 400 × g for 15 min. Mast cells were recovered from the pellet and mast cell-depleted peritoneal cells from the Percoll-DMEM interface. All cells were suspended in cDMEM. The mast cell fraction was >98% pure and contained <2% macrophages, whereas the interface cells comprised <2% mast cells and >98% macrophages, as determined by staining in 0.02% toluidine blue and by Giemsa staining of cytospin preparations. The cells were cultured in cDMEM in conical sterile plastic tubes at 37°C with 5% CO₂ in air, usually for 24 h.

Human mast cells of the HMC-1 cell line (27) were grown in IMDM (Life Technologies) containing 10% FCS and subcultured weekly at 1:10.

[³H]Serotonin (5-[1,2-N-³H]hydroxytryptamine creatinine sulfate (sp. act., 27 Ci/mmol; DuPont-NEN, Dreiech, Germany) was added (1 μCi/ml) for the final 2 h of peritoneal mast cell culture. At the termination of culture, the mast cells were washed three times with cDMEM and resuspended (10⁴/ml) in challenge medium (cDMEM buffered to pH 7 with HEPES). The cells (100 μl) were added in duplicate to 100 μl of challenge medium (background control), rat monoclonal anti-mouse IgE (200 ng/ml, kindly provided by Dr. F. D. Finkelman, Uniformed Services University of the Health Sciences, Bethesda, MD), or 0.05% Triton X-100 to lyse the cells. The cells were incubated for 30 min, after which they were centrifuged at 350 × g for 2 min and 100 μl of supernatant medium was removed for scintillation counting. Percentage of specific release of [³H]serotonin from mast cells was measured as (a − b)/c, in which a = cpm for anti-IgE-treated cells, b = mean cpm for control cells, and c = cpm for lysed cells.

Murine rIFN-γ (R&D Systems, Oxon, U.K.) was biotinylated by incubating 25 μg of protein in 0.25 ml PBS with 3 μl of biotin-X-N-hydroxysuccinimide ester (10 mg/ml in DMSO; Calbiochem, Nottingham, U.K.) for 4 h at 37°C. The reaction was quenched with equimolar glycine, and the solution was dialyzed extensively against PBS. This biotinylated IFN-γ retained biological activity. The biotinylated IFN-γ was added (1:10) to 50 μl of purified mouse peritoneal mast cells (2 × 10⁵/ml in PBS containing 0.1% FCS) for 30 min on ice. The cells were washed twice in ice-cold PBS and resuspended in 50 μl of PBS containing streptavidin-PE (1:50) for 30 min on ice, washed twice, and resuspended in 0.5 ml of ice-cold PBS. As negative control, the biotinylated IFN-γ was omitted. The cells were analyzed by flow cytometry (FACScan; Becton Dickinson, Mount View, CA).

NO synthesis was measured as accumulation of nitrite (a stable product of NO) in the culture medium by the Griess reaction (28). Samples and NaNO₂ standards (150 μl) were added in duplicate to 96-well microtiter plates. Then, 50 μl of Griess reagent (1% sulfanilamide and 0.1% N-(1-napthyl) ethylenediamine dihydrochloride in 45% acetic acid) was added to each well, and the samples were incubated for 10 min at room temperature. Absorbances at 570 nm were read on an automatic plate reader (MR 600; Dynatech Instruments, Torrance, CA). The values of nitrite concentration in the culture samples were obtained from the standard curve.

An ELISA for measuring GAG binding to IFN-γ was adapted from a previous study (29). Ninety-six-well microtiter plates (Falcon Probind ELISA plates; Fred Baker, Runcorn, U.K.) were coated with carrier-free murine rIFN-γ (2 μg/ml; Peprotech, London, U.K.) in carbonate buffer, pH 9.6, for 18 h at 4°C. The plates were washed three times with PBS containing 0.05% Tween 20 (PBS-T; Sigma, Poole, Dorset, U.K.). GAGs were then added in triplicate to the top row of wells and serially diluted down the plate in PBS-T. The plates were incubated for at least 1 h at room temperature, and then heparin-BSA-biotin (Sigma) was added at a final concentration of 2 μg/ml for at least 1 h. Positive controls were wells with no GAG added, and negative controls were wells with BSA-biotin (Sigma) in place of heparin-BSA-biotin. The plates were washed three times in PBS-T, and extravidin-HRP (1:2000; Sigma) was added for 1 h. Plates were washed again before addition of 50 μl of o-phenylenediamine substrate solution (Sigma). The reaction was terminated by addition of 20 μl of 3 M HCl, and absorbance was read at 490 nm, reference 630 nm. GAGs used were: heparan sulfate (bovine trachea; Sigma H7640); chondroitin sulfate A (bovine trachea; Sigma C8529); chondroitin sulfate B (porcine intestinal mucosa; Sigma C3788); chondroitin sulfate C (shark cartilage; Sigma C4383); and hyaluronic acid (bovine trachea; Sigma H0902).

Data are presented as the mean ± SEM of several experiments or as representative experiments, and statistical analyses are by Student’s t test or Mann-Whitney U test, as appropriate.

It is known that IFN-γ induces NO synthesis and inhibits Ag-induced mast cell degranulation in mouse peritoneal cell populations. To examine definitively the cellular target for IFN-γ in mast cell inhibition, we added the cytokine for 24 h to cocultures of IFN-γR KO mast cells with IFN-γR WT accessory cells (gradient interface cells, >98% macrophages), or to cocultures of IFN-γR WT mast cells with IFN-γR KO accessory cells. The ratio of mast cells to accessory cells in these experiments was 1:10. Only in cultures containing IFN-γR WT accessory cells did IFN-γ inhibit anti-IgE-induced mast cell serotonin release (Fig. 1,A) and induce nitrite production (Fig. 1,B). IFN-γ was without effect on either parameter in cultures containing IFN-γR WT mast cells plus IFN-γR KO accessory cells (Fig. 1). Thus, macrophages, not mast cells, are the target for IFN-γ-induced NO synthesis in peritoneal cell populations and are responsible indirectly for the IFN-γ-dependent inhibition of mast cell degranulation.

Although IFN-γ did not act directly on mast cells to induce inhibitory levels of NO (Fig. 1) (9), we nevertheless examined whether the cytokine bound to these cells. Purified IFN-γR WT or KO mast cells were incubated with biotinylated IFN-γ for 30 min at 37°C and analyzed by flow cytometry. The cytokine bound to both mast cell phenotypes (Fig. 2). This shows that IFN-γ interacts with a site on mast cells, which is distinct from the IFN-γR. In two of three experiments, there was greater binding of IFN-γ to WT compared with KO mast cells. This probably reflects additional binding to the IFN-γR on the WT cells.

We examined whether mast cell-bound IFN-γ retained functional activity in relation to activation of NO synthesis by mouse peritoneal macrophages. Purified IFN-γR WT mouse mast cells or HMC-1 human mast cells were incubated with a range of concentrations of murine IFN-γ for 30 min at 37°C. The mast cells were washed three times and then cocultured with mouse peritoneal cells (responder cells, >98% macrophages) for 24 h. Free IFN-γ was added to responder cells alone as a positive control. Both types of mast cells induced nitrite production by the responder cells to a degree dependent on the loading concentration of IFN-γ (Fig. 3). Mast cells that had not been treated with IFN-γ failed to induce nitrite production; loading with IFN-γ concentrations of 1 ng/ml or above induced a significant response (Fig. 3). Free IFN-γ at concentrations of 0.01 ng/ml and above induced nitrite production (Fig. 3).

Coculture experiments with mast cells and macrophages confirmed that IFN-γ presentation does not require IFN-γR expression by mast cells, but is dependent upon IFN-γR expression by responding macrophages. IFN-γR WT or KO mast cells were incubated with IFN-γ for 30 min, washed, and added to IFN-γR WT or KO responder cells. IFN-γ-treated mast cells induced nitrite production only when the responder cells were IFN-γR WT, whereas the capacity to present IFN-γ was independent of the mast cell phenotype (Fig. 4). In parallel experiments, we showed that macrophages could also present IFN-γ to responder cells. Again, we found that nitrite production was dependent largely on the expression of the IFN-γR on the responding population, although low levels of nitrite were produced by IFN-γ-pulsed WT macrophages (data not shown).

To examine whether mast cell presentation of IFN-γ required cell contact, IFN-γ-loaded mast cells (10⁵ in 0.25 ml) were added either directly to responder cells (10⁶ in 0.25 ml), or the two cell populations were separated by placing the mast cells inside a transwell (0.45 μm pore size; Corning Costar, High Wycombe, U.K.) that was inserted into the macrophage suspension in the base of a conical tube. The transwells allow transfer only of soluble molecules, not cells, through a semipermeable membrane. Fig. 5 shows that physical separation of the two cell types abolished NO induction by the responder macrophages over 24 h of culture. Thus, the IFN-γ effect is dependent on cell-cell contact. Moreover, it is unlikely that IFN-γ shed from cells loaded with only 10 ng/ml and subsequently washed extensively would reach the concentrations required in solution to induce NO production (Fig. 3).

Mast cell binding and presentation of IFN-γ were independent of expression of IFN-γR and required cell-cell contact. Therefore, we speculated that presentation might be via GAGs. In the first series of experiments, purified mouse mast cells, or peritoneal macrophages prepared as gradient interface cells (both at 10⁵/ml), were incubated with heparinase (1 U/ml; Sigma) or chondroitinase ABC (1 U/ml; Sigma) for 1 h at 37°C in cDMEM. After washing, they were loaded with IFN-γ at 10 ng/ml for 30 min, washed again, then added (2 × 10⁵/ml) to responder peritoneal cells (2 × 10⁶/ml) for 24 h, in a final volume of 0.25 ml. Heparinase significantly inhibited the capacity of the macrophages (but not mast cells) to present IFN-γ to the NO-producing cells. However, chondroitinase ABC inhibited the capacity of mast cells, but not macrophages, to present IFN-γ (Fig. 6). In a second series, purified mast cells (10⁵/ml) were incubated with chondroitinase ABC (1 U/ml), chondroitinase AC (1 U/ml; Sigma), or chondroitinase B (1 U/ml; Sigma) for 4 h and then loaded with IFN-γ (10 ng/ml) for the final 30 min. The cells were then washed and added to responder cells as previously. Chondroitinase ABC inhibited nitrite production to ∼20% of control values, whereas chondroitinase B inhibited to ∼40% of control values. There was no significant difference between the effects of chondroitinase ABC and chondroitinase B. Chondroitinase AC produced a small but not significant inhibition (Fig. 7). These experiments show that IFN-γ binds to chondroitin sulfate B on mouse peritoneal mast cells. In this form, the cytokine is functionally active. Macrophages could also bind and present IFN-γ, but this was via heparan sulfate (Fig. 6).

The capacity of various GAGs to inhibit binding of heparin-BSA-biotin to murine IFN-γ was tested by ELISA. Fig. 8 shows a representative experiment from a series summarized in Table I. In all experiments, the positive control inhibitor heparin-BSA reduced binding completely at low concentrations. Heparan sulfate gave significant inhibition in only one of three experiments, whereas chondroitin sulfate B and hyaluronic acid inhibited in all experiments (mean maximum inhibition over all experiments being 75% and 57.5%, respectively). Chondroitin sulfate A and chondroitin sulfate C gave a small but significant inhibition in one of four and one of three experiments, respectively. These assays confirm that chondroitin sulfate B binds to murine IFN-γ.

IFN-γ inhibits degranulation of rodent peritoneal mast cells (3, 4, 5, 6), and this activity is through induction of NO synthesis by accessory cells (nonmast cells, largely macrophages) in peritoneal populations (9, 10). NO is a known inhibitor of mast cell and basophil degranulation (30, 31, 32). In the present study, using a novel approach, we confirm that IFN-γ does not target mast cells directly to inhibit IgE-mediated degranulation. In coculture experiments, the cytokine inhibited mast cell serotonin release and induced NO synthesis, only when accessory cells expressed the IFN-γR. IFN-γ exhibited both of these activities regardless of whether the mast cells were IFN-γR WT or KO. Therefore, IFN-γ acts at its receptor on accessory cells to induce NO synthesis, and mast cells are suppressed by NO regardless of whether or not they express the IFN-γR. We were unable to confirm reports that mast cells produce NO (31, 33, 34, 35), at least after incubation with IFN-γ.

Although mast cells did not respond to IFN-γ to produce detectable nitrite, we examined whether IFN-γ could bind to purified mast cells. Surprisingly, we found that biotinylated IFN-γ bound equally to both IFN-γR WT and KO cells, indicating that binding is independent of IFN-γR. Additional experiments revealed that the bound IFN-γ retained functional activity, since purified mast cells loaded with IFN-γ were able to stimulate NO synthesis by peritoneal macrophages. This presentation of IFN-γ to NO-producing cells was seen regardless of whether the mast cells expressed the IFN-γR, but showed an absolute requirement for IFN-γR expression on the responder cells and was dependent on cell-cell contact. These experiments show that the IFN-γ binds to mast cells via an interaction independent of the IFN-γR and that, in this state, it can be presented functionally to its receptor on an adjacent cell. Human HMC-1 mast cells were also able to present murine IFN-γ to mouse peritoneal NO-producing cells, showing that this phenomenon is not restricted to rodent mast cells.

We considered proteoglycans as a possible alternative cell surface binding site for IFN-γ. These consist of a protein core with GAG side chains of repeating disaccharide units. Cytokine and growth factor interactions with GAGs have been reported and shown to be important in their functional regulation (36, 37, 38). We therefore investigated whether mast cells might bind IFN-γ through GAGs. Purified mouse peritoneal mast cells exposed to chondroitinase ABC and chondroitinase B had a significantly reduced capacity to present IFN-γ, whereas heparinase and chondroitinase AC had little effect. We found that peritoneal macrophages were also capable of presenting IFN-γ in an autologous manner, but this was heparinase sensitive and chondroitinase ABC resistant. Thus, mouse mast cells appear to use predominantly chondroitin sulfate B (dermatan sulfate), whereas macrophages use heparan sulfate, for functional presentation of IFN-γ. We confirmed by ELISA that IFN-γ had binding activity for dermatan sulfate.

Utilizing endothelial cell layers and immobilized purified GAGs, others have shown that human IFN-γ binds to heparin, heparan sulfate, and chondroitin sulfate, sulfation of the GAGs being essential for in this interaction (12, 13, 14, 15, 16, 39). Binding is thought to be an ionic interaction between positively charged basic amino acids near the C terminus of IFN-γ and the negatively charged sugar residues on GAGs (16). The biological activity of IFN-γ bound to immobilized GAGs or endothelium has been inferred by some and demonstrated by others (12, 14, 15, 16). In the present study, we have demonstrated, for the first time, that dermatan sulfate on the surface of a motile cell type (mast cells) can bind IFN-γ and present it to macrophages, inducing an immunologically important response. Consistent with our findings, it has recently been shown that C-terminal basic peptides of murine IFN-γ interact with heparin, heparan sulfate, and chondroitin sulfate (16). Chondroitin sulfate has also been implicated in the biology of other cytokines and growth factors: chondroitin sulfate B (dermatan sulfate) binds IL-7, IL-12, IL-6, IL-1α and -β, and human hepatocyte growth factor; promotes fibroblast growth factor-2 function; and regulates heparin affin regulatory peptide activity (37, 40, 41, 42). In addition, chondroitin sulfate present on neutrophils binds PF4, while GM-CSF exists in a biologically active form in association with chondroitin sulfate (43, 44).

The fact that cytokines and growth factors can bind to GAGs has profound functional significance. First of all, GAGs can sequester cytokines and growth factors from the extracellular fluid, providing a mechanism for the spatial and temporal regulation of these mediators. This has been demonstrated for GAGs in the extracellular matrix of bone marrow stroma, model basement membrane (Matrigel), human vascular endothelium, and human arterial smooth muscle cell extracellular matrix (12, 14, 15, 16, 45). These GAG-bound cytokines may be presented either to specific receptors on the same cell in an autocrine manner or to specific receptors on other cells in a juxtacrine manner (46). Second, one GAG may bind multiple molecules of a given cytokine or growth factor, resulting in the concentration of these factors in the local microenvironment and encouraging clustering and the possibility of receptor activation (18, 47). Third, interaction with GAGs can provide protection from proteolytic cleavage, prolonging mediator biological t_1/2 (38, 48). Finally, in the case of fibroblast growth factor, interaction with the GAG heparan sulfate is a prerequisite for the delivery of a growth-stimulatory signal (38).

Our results, showing that mast cells can act as a reservoir of stored and functionally active IFN-γ, have implications both for the biological activity of the cytokine and for the role of mast cells in inflammation and immunity, as well as for the understanding of cell-cell interactions during immune responses. Mast cells are motile (49, 50) and are recruited to sites of inflammation (24, 25). The retention of cytokines on their cell surface may result in the spatial and temporal modulation of cytokine action. The aggregation of IFN-γ on mast cell surface dermatan sulfate could change the cytokine microenvironment and extend IFN-γ biological t_1/2. In addition, the phenomenon we describe, whereby mast cells can present IFN-γ to NO-producing cells, may lead to mast cell stabilization (9, 10), representing a possible mechanism of negative feedback control of mast cell activation. ELISA results showed that murine IFN-γ could bind dermatan sulfate, but with lower affinity than to heparin, although our coculture experiments illustrate that the off rate from mast cell surface dermatan sulfate is sufficiently slow to allow presentation to occur. It could be envisaged that heparin released from degranulating mast cells would remove IFN-γ from cell surface dermatan sulfate, given the relative affinities of these two GAGs for IFN-γ. Again, this could be a mechanism of regulation of IFN-γ action and NO release.

To our knowledge, this is the first report that motile cells are capable of sequestering a cytokine and presenting it to adjacent cells. It may provide new insight into the role of mast cells in promoting or extending the multiple effects of IFN-γ in inflammatory and immune processes (1, 2). It could be envisaged that cytokines may be sequestered at their site of secretion and carried to other parts of the same anatomical site or to distant sites.