Expression and Functional Activity of the IL-8 Receptor Type CXCR1 and CXCR2 on Human Mast Cells¹

The role of mast cells as central effector cells in immediate type hypersensitivity reactions and also in diverse acute and chronic inflammatory diseases is well recognized. An increase in the number of mast cells has furthermore been demonstrated in diverse allergic inflammatory and fibrotic diseases such as in acute and chronic urticaria (1), psoriasis (2), scleroderma (3), and wound healing (4). Little is known, however, about factors that induce the recruitment of mast cell precursors from the blood into, and the migration of mature mast cells within, the tissue. Directed migration requires the presence of chemotactic factors and the expression of specific cell surface receptors on the target cells. Besides stem cell factor (SCF) (5) and IL-3 (6), members of the β subgroup of chemokines, like macrophage chemotactic protein-1 (MCP-1)³ and macrophage inflammatory protein-1α (MIP-1α) (7), have previously been described to be chemotactic for murine mast cells. RANTES, another β chemokine, has been reported as a further chemotactic factor for murine mast cells (7) and human cord blood-derived cultured mast cells (8). Chemokines are low m.w. proteins, characterized by the presence of four conserved cysteine residues. In β chemokines, the two cysteine residues are adjacent, whereas in α chemokines, like IL-8, melanoma growth stimulatory activity (MGSA), and neutrophil-activating peptide-2 (NAP-2), they are separated by a single amino acid (9).

One of the best studied molecules among the chemokines is IL-8, which is a potent chemoattractant for neutrophils and which induces also a change in shape, chemotaxis, and Ca²⁺-mobilization in these cells (10). These biologic activities depend on the presence of specific cell surface receptors that belong to the G protein-coupled receptor family.

Two types of IL-8 receptors have been reported. CXCR1, formerly known as IL-8 receptor type A, exclusively binds IL-8 with high affinity (0.2–4.0 × 10⁻⁹ M) and MGSA and NAP-2 with low affinity (450 × 10⁻⁹ M), whereas CXCR2, formerly known as IL-8 receptor type B, binds all three ligands with high affinity (11). IL-8 receptors are found on neutrophils, basophils, T-lymphocytes, monocytes, and keratinocytes (12, 13, 14, 15, 16). So far, no published data are however available regarding IL-8 receptors on mast cells.

In recent studies, we have shown that human mast cells produce the α chemokine IL-8 (17), but nothing is known about IL-8 activation or IL-8 receptor expression by these cells. The aim of the present study was therefore to investigate the expression of IL-8 receptors on human mast cells using the human leukemic mast cell line HMC-1, which resembles human mast cells in most properties (18, 19). Furthermore, we have studied the specific binding of IL-8 and its homologues MGSA and NAP-2 to both receptors and confirmed the significance of these findings by examining IL-8 receptors on normal human skin mast cells.

HMC-1 cells, which are immature human leukemic mast cells (kindly provided by Dr. Butterfield, Minneapolis, MN) (20), were cultured in Iscove’s medium (Seromed, Berlin, Germany), supplemented with 10% FCS (Seromed) and 10⁻⁵ M monothioglycerol (Sigma, Deisenhofen, Germany).

The promyelocytic leukemia cell line HL-60 was grown in RPMI 1640 medium containing 10% FCS (Seromed) (21).

IL-8, MGSA, and NAP-2 were purchased from PeproTech, London, U.K.

The specific IL-8 receptor Abs, CXCR1 (clone 9H1) and CXCR2 (clone 10H2), were kindly provided by Dr. Chuntharapai (South San Francisco, CA) (22).

For the induction of granulocytic differentiation, HL-60 cells (2 × 10⁵/ml) were incubated with 1 μM all-trans-retinoic acid and 1.25% DMSO (both from Sigma) for 6 days, as described before (23). Differentiation was measured by the ability of the cells to reduce nitroblue tetrazolium (NBT) (Sigma) (24).

RT-PCR analysis was performed, as previously described (25). Briefly, 3 μg of total cellular RNA were transcribed into cDNA using random priming. The specificity of the amplification products was verified by restriction analysis with two enzymes, BamHI and XbaIII (Life Technologies, Berlin, Germany). Primer sequences used were sense 5′-CAG ATC CAC AGA TGT GGG AT and antisense 5′-TCC AGC CAT TCA CCT TGG AG for CXCR1 and sense 5′-CTT TTC TAC TAG ATG CCG C and antisense 5′-GAA GAA GAG CCA ACA AAG G for CXCR2.

HMC-1 cells were preincubated for 30 min at 4°C with human AB-serum (Behringwerke AG, Marburg, Germany) to block nonspecific binding of the mAbs. Thereafter, cells were stained with saturating concentrations of CXCR1 and CXCR2 Abs or an isotype-matched nonrelevant mouse mAb (Dianova, Hamburg, Germany) for 30 min at 4°C. Cells were then labeled with a dichlorotriazinylaminofluorescein-conjugated F(ab′)₂ fragment of goat anti-mouse IgG (Dianova) for 30 min at 4°C, fixed in PBS/1% paraformaldehyde, and analyzed by flow cytometry (Coulter, Epics XL, Krefeld, Germany).

The specific activity was 500 Ci/mmol for radioiodinated IL-8 (Amersham Buchler, Braunschweig, Germany) and 2200 for MGSA (DuPont, Dreilichen, Germany).

Cells were suspended at 2.6 × 10⁶ cells/ml in 50 μl binding buffer (RPMI medium; with 25 mM HEPES; 0.1% NaN₃; 1% BSA) and incubated for 90 min on ice with 1 × 10⁻⁸ to 5 × 10⁻¹¹ M ¹²⁵I-labeled IL-8 or MGSA and 100-fold excess of unlabeled IL-8 or MGSA. After incubation, cell suspensions were layered on top of (10 + 3) dibutyl phthalate/olive oil (Merck/Sigma) and centrifuged at 12,000 × g for 3 min. The tube tips with the cell pellet were cut, and both the remaining aqueous phase and the tube tips were counted in a Packard autogamma counter 5650 (Frankfurt, Germany). Each experiment was done seven times, and each concentration was tested in duplicate.

As judged by the Scatchard plots, binding to the IL-8 receptors is biphasic for IL-8 and MGSA, suggesting that two binding sites are involved. If one assumes them to be independent, the Scatchard data can be analyzed by employing the following equation:

where [L]_b and [L]_f are the concentrations of bound and free ligands and [R_i]_f and [R_i]_T the free and total concentration of the receptors R_i (I = 1, 2). K₁ and K₂ denote these receptor affinities for ligand binding in M⁻¹. The law of mass conversion can be used to calculate the concentration of free receptors as follows:

Finally, the mass action law and the law of mass conversion dictate the following relationship between the concentrations of free and bound ligands:

K₁, K₂, [R₁]_T and [R₂]_T were used as free parameters.

Cells (5 × 10⁶ cells/ml) were incubated in 50 μl assay buffer for 90 min on ice, with 1 × 10 nM ¹²⁵I-labeled IL-8 in the presence of 0.1- to 1000-fold excess of unlabeled IL-8, MGSA, or NAP-2, as indicated in Figure 4. The number of bound ligands was determined as described above. Results were expressed as the mean of four independent experiments.

The above data were analyzed by a simple binding model. It employs the mass action law to calculate the molar concentration of bound fraction of ¹²⁵I-labeled ligands [L]_b:

where [L]_T is the total concentration of radioactively labeled ligands. The concentrations of free binding sites can be calculated by:

where [L_c]_T is the total concentration of cold ligand applied. K_c1 and K_c2 denote the equilibrium binding constants for the two IL-8 binding sites, respectively. n is the Hill coefficient, which is a measure of cooperativity.

For the Ca²⁺-mobilization assay, HMC-1 cells were suspended in modified Gey’s buffer (138 mM NaCl, 6 mM KCl, 1 mM MgSO₄, 1 mM NaHPO₄, 5 mM NaHCO₃, 5.5 mM glucose, 1 mM CaCl₂, 20 mM HEPES, 0.1% w/v BSA, pH 7.4) at 5 × 10⁶/ml and loaded with 2 μM fura 2-AM (Calbiochem, Bad Soden, Germany) for 30 min, washed twice, and resuspended at 2 × 10⁶/ml in a cuvette under constant stirring, as described before (26). The emission ratio at 339/490 nm was followed with time on a spectrofluorophotometer RF-540 (Shimadzu, Kyoto, Japan). Each measurement was standardized by inducing cell lysis with 0.1% Triton X-100 (100% fura 2-AM saturation) and subsequent quenching of the fluorescence by addition of 1 mM EGTA to the lysed cells.

Migration of HMC-1 cells was measured in 48-well microBoyden chambers (NeuroProbe, Cabin John, MD), as described before (27). Eight micrometer-pore-sized polycarbonate filters (Costar, Bodenheim, Germany) were coated with laminin (10 μg/ml; Life Technologies) at room temperature overnight and were then air dried. The chemoattractants IL-8, MGSA, and NAP-2 were diluted in RPMI 1640/1% BSA (assay medium). Bottom wells were filled with 26 μl of the chemoattractants or the assay medium as negative control, and 50 μl of the cell suspension (1 × 10⁶ cells/ml) were added to the top wells. In some experiments, IL-8 was added to the cell suspension in the top and to medium in the bottom wells at equal concentrations to differentiate chemokinesis from chemotactic activity. Following a 90-min incubation at 37°C, 5% CO₂, the top of the filters was carefully scraped off from nonmigrating cells, fixed, stained with hematoxylin, mounted on slides, and coverslipped. Migration was quantitated microscopically by counting the number of cells at the reverse side of the filters. Results were expressed as the mean number of cells in five high power fields (×400) in triplicate samples.

Relative F-actin contents were analyzed by flow cytometry, as described before (28). Briefly, HMC-1 cells (5 × 10⁶/ml) were stimulated with IL-8, MGSA, or NAP-2 (final concentration 1 to 1000 nM) for different time periods (as indicated) and were then fixed in 4% paraformaldehyde for 20 min at room temperature. Thereafter, the cells were resuspended in PBS/0.02% saponin and stained with the F-actin-specific probe 7-nitrobenz-2-oxa-1,3-diazol-(NBD)-phallacidin (3.3 × 10⁻⁸ M) (Molecular Probes, Eugene, OR) for 30 min. The F-actin contents were measured on an EPICS XL flow cytometer (Coulter Electronics). The relative F-actin contents were expressed as ratio of the mean fluorescence intensities (MFI) of chemokine-stimulated cells to unstimulated cells and were plotted against time.

The identification and ultrastructural localization of IL-8 receptors in counterflow elutriation-enriched human skin mast cells were performed by postembedding immunoelectron microscopy. The mast cell-enriched cell preparation was fixed in 4% paraformaldehyde for 10 min at room temperature. After intense washing in PBS, cells were centrifuged in molten agar (1% in 0.1 M PBS at 45°C) to form cell pellets that were sliced and processed as small blocks. Dehydration was done in a graded series of ethanol solutions at 4°C. The dehydrated cell pellets were transferred to LR-White resin (London Resin, Berkshire, U.K.) without accelerator (29). For infiltration, specimens were changed 4 × with LR-White resin during 48 h at 4°C. Polymerization was induced under oxygen-deficient conditions in polypropylene capsules by adding the manufacturer’s accelerator for 2 h at 4°C, in the proportion of two droplets to 10 ml resin. Thereafter, blocks were allowed to complete polymerization at room temperature for at least 2 days. Semithin sections (0.5 μm) were mounted on glass slides and stained with toluidine blue for quick evaluation. Ultrathin sections (70 nm) were picked up and placed on formvar-coated nickel grids.

Immunostaining was performed on 50-μl droplets in a moist chamber using CXCR1 and CXCR2 mAbs (clone 9H1 and clone 10H2, respectively) and colloidal gold (10 nm)-labeled goat anti-mouse IgG (AuroProbe, Amersham). Briefly, ultrathin sections were blocked with PBS/5% BSA for 30 min at room temperature and were then incubated with the primary Ab (9H1, 0.25 μg/ml; and 10H2, 4 μg/ml) overnight at 4°C. After washing, sections were incubated with the colloidal gold-conjugated second Ab (1:20) for 1 h at room temperature, followed by washing and poststaining with 5% aqueous uranyl acetate (Merck, Darmstadt, Germany) for 15 min. Staining specificity was checked by substituting the primary Ab with an inappropriate isotype Ab. Specimens were examined with a Zeiss EM906 transmission electron microscope at 80 kV (Oberkochem, Germany).

RT-PCR analysis revealed that HMC-1 cells express mRNA for both types of the IL-8 receptor. As shown in Figure 1,A, the expected CXCR1 PCR product with 297 bases and in Figure 1,B the CXCR2 product with 967 bases were detected in HMC-1 cells as well as in all-trans-retinoic acid and DMSO-differentiated HL-60 cells (Fig. 1,B), which served as positive control. The CXCR1 product was not cut by BamHI but shows two products from 194 and 103 kb length after digestion with XbaIII, as expected by the sequence (Fig. 1,A), whereas BamHI (544 and 423 kb) and XbaIII (824 and 143 kb) cut the CXCR2 PCR product twice (Fig. 1 B).

The percentage of positive cells that expressed CXCR1 and CXCR2 receptors on HMC-1 cells was studied next by flow cytometry. Almost 70% of cells stained positively for CXCR1, but only 15% expressed the CXCR2 receptor type (Fig. 2,A). To detect IL-8 receptors possibly expressed intracellularly, HMC-1 cells were prefixed, permeabilized with saponin, and stained with the mAbs 9H1 and 10H2 to the CXCR1 and CXCR2 receptors. Under these conditions, the CXCR2 receptor was detectable in the majority of HMC-1 cells (>60% of cells). The number of CXCR1 type positive cells was also increased (to almost 100%) under these experimental conditions, but to a lesser extent compared with CXCR2 (Fig. 2 B). These results show that the CXCR1 receptor is expressed preferentially on the surface of HMC-1 cells, while the CXCR2 receptor is predominantly expressed intracellularly. We could not detect a translocation of the receptor to the cell surface in response to IL-8 itself (data not shown).

To check whether HMC-1 cells express IL-8 receptor protein that could bind IL-8 and account for its biologic activities, we measured the binding isotherm of ¹²⁵I-labeled IL-8 and also of ¹²⁵I-labeled MGSA. As shown in Figure 3, the corresponding Scatchard plots are clearly biphasic.

Equations 1, 2, 3a, 3b from Materials and Methods were employed to fit the experimental data in Figure 3 A and B The data are appropriately reproduced by this procedure. The individual values are listed in Table I.

They show that two binding sites exist on the cell’s surface that significantly differ in their affinity for the respective ligand. For both ligands, the number of low affinity binding sites (lbs) per cell (and thus also the surface density) exceeds that of high affinity binding sites (hbs) by nearly an order of magnitude, and the total number of IL-8 binding sites is an order of magnitude larger than that of MGSA.

Figure 4 depicts the results of competition binding experiments, demonstrating that IL-8, MGSA, and NAP-2 compete with ¹²⁵I-labeled IL-8 in a dose-dependent manner. At high concentrations, all these ligands replace 95 to 99% of ¹²⁵I-labeled IL-8. The data were analyzed by fitting equation No. 4 to the experimental data. This yields the solid lines in Figure 4 and K_c2 values of 7.0 × 10⁷, 6.0 × 10⁷, and 3 × 10⁷ M⁻¹ for IL-8, MGSA, and NAP-2, respectively. As expected, the inflexion points of the competition curves predominantly reflect the cold ligands binding to lbs, due to its significantly higher concentration in the sample. Hence, K_cl could not be independently determined with sufficient accuracy. Therefore, we fixed K_cl = K₁ for the fits to the competition curves of IL-8 and MGSA. For NAP-2, we solely considered binding to the lbs. For IL-8 and MGSA, the equilibrium constants K_c2 are very close to those observed with the Scatchard plots. This underscores the consistency of our analyses.

It should be mentioned that appropriate fitting to the above data requires a Hill coefficient of 2. This indicates some cooperativity, but the concrete mechanism remains unclear.

In HMC-1 cells loaded with the fluorescent Ca²⁺ indicator fura-2 AM, IL-8 applied at 1 × 10⁻⁸ M to 2 × 10⁻⁷ M induced a dose-dependent rapid and transient increase in cytosolic free calcium [Ca²⁺]_i that peaked within 5 sec and returned to baseline values within 20 sec (Fig. 5,A). In the presence of 3 mM EGTA as an extracellular Ca²⁺ chelator, IL-8 still induced a similar rise in [Ca²⁺]_i, indicating release from intracellular stores (not shown). At 1 × 10⁻⁷ M, the increase of [Ca²⁺]_i induced by IL-8 was of similar amplitude as that induced by MGSA, whereas the response to NAP-2 was considerably weaker (Fig. 5,B). After stimulation with IL-8, the cells were refractory to a second challenge with IL-8, MGSA, or NAP-2 at equimolar concentrations, as shown for MGSA (Fig. 5,B). No cross-desensitization between NAP-2 or MGSA and IL-8 was observed, however, when the sequence of stimuli was reversed (Fig. 5 B). Other chemokines, such as MCP-1, macrophage inflammatory protein (MIP)-1α, and RANTES, did not induce any measurable changes in [Ca²⁺]_i, at up to micromolar concentrations (not shown) and were therefore not further studied.

To analyze whether IL-8 might act as a stimulus for migration of human mast cells, these effects were studied using the chemotaxis assay in modified Boyden chambers. HMC-1 cells showed a distinct dose-dependent chemokinetic and a much stronger chemotactic response not only toward IL-8, with maximal effects at 10⁻⁷ M (Fig. 6,A) but also toward MGSA and less so toward NAP-2, as demonstrated only for the most effective concentration (10⁻⁷ M in each case) (Fig. 6 B).

Since polymerization of actin is thought to be a prerequisite for chemotaxis (28), F-actin contents of HMC-1 cells were analyzed after stimulation with IL-8, MGSA, and NAP-2 (1 to 1000 nM) for different time periods at 37°C, using flow-cytometry. Stimulation of the cells with 100 nM IL-8 and MGSA caused polymerization of actin within 10 s, addition of NAP-2 (100 nM) within 20 s (Fig. 7,A). Each chemokine caused a concentration-dependent polymerization of F-actin, with a maximal effect at 100 nM, as shown for IL-8 in Figure 7,B (not shown for MGSA and NAP-2). At the maximally stimulating dose of IL-8 and MGSA, there was a 1.6- to 1.7-fold increase of F-actin contents in HMC-1 cells, followed by a fast decline to values lower than baseline by 120 s. Addition of NAP-2 (100 nM) resulted in a weaker response, with a maximally 1.3-fold increase of F-actin contents (Fig. 7 A).

Postembedding immunoelectron microscopy was used to identify expression of both types of IL-8 receptors and their ultrastructural localization in isolated human skin mast cells. As shown in Figure 8,A, CXCR1 receptors were detectable on the surface of skin mast cells. A more intense staining was noted on narrow surface folds. The cytoplasm and the mast cell-specific granules showed no immunoreactivity with the mAb 9H1, the mAb against CXCR1. With the mAb 10H2, the mAb against CXCR2, a completely different staining pattern was observed (Fig. 8 B). CXCR2 receptors were not detectable on the cell surface and in the cytoplasm but were clearly expressed in mast cell-specific granules.

The background label for both mAbs used was minimal, and no immunoreactive structures were seen in cells processed in the presence of a corresponding isotype-matched irrelevant mAb (Fig. 8 C).

IL-8 belongs to the α chemokine family and mediates its biologic actions through binding to different receptors that are termed CXCR1 and CXCR2. Until now, IL-8 receptors on mast cells have not been described although many other cell types have been shown to express low and high affinity IL-8 receptors (12, 13, 14, 15, 16) and a possible binding of ¹²⁵I-labeled IL-8 to perivascular mast cells in human skin was suggested already in 1992 by Rot (30). The present findings demonstrate the expression of both types of IL-8 receptor in human skin mast cells and in the HMC-1 cell line. Furthermore, the receptors are shown to be functional, based on findings of a dose-dependent chemotactic activity of IL-8 and its homologues toward human mast cells, data that were further supported by our findings of actin polymerization after chemokine stimulation.

Evidence for biologic effects of IL-8 and its homologues on mast cells is already provided by the data on the induction of Ca²⁺ flux since the rapid and transient rise in Ca²⁺ is in fact considered to be an early event after stimulation of leukocytes with chemotactic agents (10, 31). In particular, we found that the stimulation with IL-8 abrogated the Ca²⁺ response of HMC-1 cells to MGSA or NAP-2 whereas MGSA or NAP-2 pretreatment only partly inhibited the response to IL-8. These data are in agreement with desensitization data in neutrophils (32). Like these authors, we speculate that a faster internalization rate of the CXCR2 receptor induced more strongly by IL-8 than by MGSA could be a reason for these findings. Moreover, the higher number of the CXCR1 receptors detected by Scatchard plot analysis might be involved in the phenomenon observed.

Recent studies by Legler et al. (33) failed to demonstrate intracellular Ca²⁺ flux in response to IL-8 or other α and β chemokines in HMC-1 cells. Whether the differences to our data are due to the HMC-1 subclones used, to culture conditions, or to experimental approaches is presently not clear.

While migration of neutrophils in response to IL-8 is a well-studied phenomenon, little is known about the mechanisms inducing mast cell migration. HMC-1 cells have been reported to be chemotactically activated by stem cell factor (8), C3a, and C5a (34) whereas no effect of any chemokine, including IL-8, RANTES, and MCP-1, has so far been observed. In cord blood-derived mast cells, RANTES induced, however, a chemotactic response (8).

In contrast to data reported by Nilsson et al. (8), we could detect a marked, dose-dependent chemokinetic and an even stronger chemotactic response of HMC-1 cells toward IL-8, and less so toward MGSA and NAP-2. A possible explanation for these divergent results might be the matrices used in the experimental system. While Nilsson et al. (8) used nitrocellulose filters for mast cell chemotaxis, we have used polycarbonate filters. Possibly, the type of filters influences binding of the matrix protein and the attachment of the cells and chemokines (35). Similar differences have been described in the mouse system (7, 36).

The concentration-dependent, rapid polymerization of F-actin in HMC-1 cells after stimulation with IL-8, MGSA, and NAP-2 supports our chemotaxis data. Globular monomeric G-actin polymerization to the filamentous F-actin has in fact been postulated to be a prerequisite for cell migration and chemotaxis (37).

Competition experiments indicate that IL-8, MGSA, and NAP-2 compete with the respective radiolabeled ligand for binding, in accordance with data obtained for neutrophils (38).

Surprisingly, MGSA fully dissociates IL-8 despite the much lower surface density of the binding sites for the two MGSA receptor molecules. A similar observation was also made by Petersen et al. (39) when comparing the Scatchard analysis of IL-8 and NAP-2 binding to surface receptors on polymorphonuclear neutrophils. All these findings suggest that more than one IL-8 binding site is in close proximity to a particular MGSA binding site so that occupation of the latter might inhibit binding to the former by steric hindrance. Moreover, one might speculate that all interacting binding sites provide one functional entity.

The present data and findings in the literature underline that there is a marked heterogeneity of chemokine receptor expression with regard to species and types of cells studied (39, 40, 41). Most observations describing two high affinity IL-8 binding sites were primarily obtained from transfected cells expressing each receptor separately (22). In human mast cells, both IL-8 receptor types with a high and a low affinity binding site are on the other hand described here by various techniques. Results comparable to ours have been reported by Chuntharapai and Kim in neutrophils that also express both IL-8 receptor types, namely a high (CXCR2) and a low affinity (CXCR1) IL-8 receptor (22). These authors compared their results with observations made for other hemopoietic growth factors like GM-CSF or IL-3, for which also a high and a low affinity receptor exist. The high affinity receptor shows cross-reactivity to other growth factors whereas the low affinity receptor is restricted to only one ligand. This supports our assumption of the biologic importance of the high affinity intracellular CXCR2 receptor’s specific up-regulation in human mast cells, as discussed below.

Scatchard plot and competition binding studies for the chemokines studied here were performed only with HMC-1 cells and not with purified skin mast cells since it was not possible to obtain sufficient numbers (5 × 10⁷ - 2 × 10⁸ cells) in high purity for each experiment. Therefore, we have more recently examined the dose-dependent induction of histamine secretion in human skin mast cells by IL-8 or MGSA and found no release at all, not even when cells were preincubated with IL-3 or GM-CSF (Welker et al., manuscript in preparation). A strong chemotactic stimulus is thus not necessarily also a strong stimulus for mediator release as shown for RANTES and other chemokines in basophils and skin mast cells, possibly due to different signal transduction pathways via G-proteins (42, 43).

Using immunoelectron microscopy and specific Abs, we could confirm the expression of both IL-8 receptor subtypes on normal human skin mast cells. HMC-1 cells revealed a weak staining for the CXCR2 receptor, with a clear increase of detection rate upon cell permeabilization, while the level of CXCR1 increased only slightly. Since these experiments were done at 4°C, internalization of the receptor is not a very plausible explanation, and, instead, a basic intracellular expression of the CXCR2 type is more likely, also in view of the findings with human skin mast cells on immunoelectron microscopy. In contrast to the FACS data, CXCR1 was, however, found only in an extracellular location, and no expression of extracellular staining was noted for the CXCR2 receptor on human skin mast cells. The divergent findings with HMC-1 and skin mast cells might be due to different sensitivities of both methods or to the well known heterogeneity of different types of mast cells. Nevertheless, these data demonstrate unequivocally the intracellular existence of the CXCR2 IL-8 receptor, which is to our knowledge the first report of an intracellular chemokine receptor expression.

There are, however, reports on high affinity intracellular receptors of FMLP, a potent neutrophil chemotactic factor, in the specific granules and secretory vesicles of neutrophils. After stimulation with FMLP, these receptors are translocated to the cell membrane, explaining their fast and extensive up-regulation during inflammatory reactions (44). Correspondingly, the CXCR2 receptor in human skin mast cells was also located intracellularly in specific mast cell granules on immunoelectron microscopy. Whether classical mast cell degranulating stimuli might similarly translocate the CXCR2 receptor from specific mast cell granules to the cell surface is thus an interesting speculation. In view of this hypothesis, it may be important that MGSA and NAP-2 mediate their biologic effects mainly through the CXCR2 receptor type, supporting IL-8 effects or mediating other as yet unknown effects once the CXCR2 receptor is specifically up-regulated. MGSA and IL-8 have in fact been shown to induce Ca²⁺ influx by divergent signaling pathways after binding to CXCR2 receptor (45). Furthermore, the specific up-regulation of CXCR2 could be linked not only with well known biologic effects like chemotaxis, but also with melanocyte growth and down-regulation of collagen synthesis, which are known to be mediated through these receptors (46). Elevated IL-8 serum or tissue levels and increased mast cell numbers have furthermore been found in patients with various fibrotic diseases (2, 3, 47, 48).

The high number of IL-8 receptors expressed by human mast cells might thus have important functional implications, particularly regarding the activation and recruitment of mast cells into inflamed tissue. Possible other functions of IL-8 on human mast cells, like stimulation or modulation of cell growth and differentiation, will have to be explored in future studies.

We thank Prof. T. Diamantstein, deceased head of the Department of Immunology, Freie Universität Berlin, for his friendly support and advice during the binding assay studies. We also thank Virgilia Odenwald at the same department for her excellent technical assistance.