Expression of Multiple Chemokine Genes by a Human Mast Cell Leukemia*

The chemokines are a large group of cytokines that are recognized to be important mediators of inflamma- tion. In this study we show that the human mast cell leukemia line HMC-1 is a source of multiple chemokines, including 1-309, monocyte chemoattractant protein 1, macrophage inflammatory protein-la, macrophage inflammatory protein-lp, RANTES, and interleukin-8. 1-309 and MCP-1 transcripts are expressed at low levels in unstimulated HMC-1. However, phorbol ester treat- ment up-regulates these and other chemokine transcript levels and also up-regulates chemokine protein synthesis and secretion. Induction of chemokine transcripts in HMC-1 requires de nouo protein synthesis. We compared the effects of anti-inflammatory glucocorti- coids on the expression of chemokine genes in HMC-1 to their effects in activated T-cells. We find that methyl- prednisolone reduces MCP-1 but not other chemokine transcripts in HMC-1, even though there are distinct and more general effects on chemokine transcripts in activated T-cells. These effects are attributed to inhibi- tion of transcription rather than transcript stability. Our results suggest that human mast cells may be a source of multiple chemokines, that glucocorticoids may inhibit the expression of only a subset of these chemokines,

. On the other hand, the only known source of PF4 is platelets, and the only known source of 1-309 is activated T-cells (3,4).
Murine mast cells have recently been recognized as a source of cytokines that are also produced by activated T-cells, including IL-3, IL-4, IL-5, IL-6, granulocyte-macrophage colonystimulating factor (GM-CSF), tumor necrosis factor-a (TNF-a), and interferon-y (IFN-y) (5-7). Among chemokines, the murine CC molecules TCA3, MIP-la, MIP-lp, and MCP-1/JE were found to be produced by growth factor-dependent and -independent mast cell lines (8). Only a limited number of reports are available on the expression of cytokines by human mast cells, primarily because of difficulty in obtaining sufficient quantities of purified cells for analysis. However, it has recently been shown that dispersed mast cells from human foreskin and respiratory tract express TNFa and IL-4 proteins upon stimulation with anti-IgE (9,10).
The human mast cell leukemia HMC-1 is a cell line that was established from the peripheral blood of a patient with mast cell leukemia and exhibits many characteristics of immature mast cells (11). Notably, these cells contain low levels of histamine, are stained metachromatically by toluidine blue, and contain chloroacetate esterase, aminocaproate esterase, and tryptase activities. However, they do not express cell surface FceR, a property that they share with mucosal mast cells from Dichinella spiralis-infected mice (121, primary human mast leukemia cells (131, and immature mast cells established from human fetal liver (14). Although in the absence of FceR HMC-1 cannot be activated by antigen, the cells can still be activated by treatment with phorbol esters and calcium ionophore, as can normal or transformed murine FceR+ mast cell lines (8). These cells can therefore serve as a useful system to begin a n examination of mast cell expression of human chemokines. We show in this study that upon stimulation, HMC-1 cells produced an array of chemokines that is broader than that produced by stimulated human T lymphocytes. We further compared the effects of anti-inflammatory glucocorticoids on chemokine gene expression in HMC-1 and T lymphocytes and observed differential sensitivity in the two cell types. Our data suggest that glucocorticoids may inhibit the expression of only a subset of mast cell-derived chemokines and argue that chemokine genes are differentially regulated in mast cells and T-cells.
MATERIALS AND METHODS Cell Culture-Peripheral blood mononuclear cells were isolated from Leukopaks obtained from the Red Cross blood bank (Charlotte, NC) by centrifugation through Ficoll-Hypaque (Lymphocyte Separation Me-13893 dium; Organon Teknika Corp., Durham, NC). Cells were washed and cultured at an initial density of 2 x lofi cells/ml in RPMI 1640 (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal bovine serum (Irvine Scientific, Santa Ana, CA), 2 mM L-glutamine, 10 m M HEPES (pH 7.3), and antibiotics (penicillin at 50 unitdml and streptomycin a t 50 pg/ml; Life Technologies, Inc.) and were stimulated with 1 pg/ml aCD3 antibody 64.1 (Bristol-Meyers Squibb, Seattle WA) and 50 ng/ml PMA (Sigma). After 3 days of primary activation, the cells were expanded at an initial density of 1 x lofi cells/ml in IL-2 (Genzyme Corp., Cambridge, MA) a t 10 unitdml. After two to three rounds of stimulation and expansion, the activated T-cells thus obtained were restimulated as indicated with one or more of the following reagents: 1 pg/ml aCD3 antibody, 50 ng/ml PMA, 1 p~ methylprednisolone, 10 p~ actinomycin-D (Boehringer Mannheim), and 10 p~ cycloheximide (Sigma). The polyclonal ap T-cell line LPL-1 was generated by stimulating leukopak peripheral blood mononuclear cells with 1:lOOO PHA-P (Difco). After 1 week, the blasts were transferred to 10% conditioned medium and restimulated after 3-4 days with PHA-P (4). HMC-1 cells were routinely maintained as described elsewhere (11). Briefly, the cells were cultured in 75-or 150-cm' flasks (Corning Inc., Corning, N Y ) in Iscove's medium (Life Technologies, Inc.) containing 10% iron supplemented calf serum (Flow Laboratories, McLean, VA), antibiotics, 2 mM I,glutamine, and 1.2 rnM a-thioglycerol (Sigma). Cells were harvested and fed every 3-4 days. HMC-1 cells were treated as indicated with one or more of the following reagents: 30 ng/ml PMA, 500 ng/ml ionomycin (Sigma), 10 p~ actinomycin D, 10 p~ cycloheximide, and methylprednisolone, hydrocortisone, dexamethasone, and progesterone at concentrations ranging from 0.1 to 10 p~. All cells were maintained a t 37 "C in a humidified atmosphere of 5% CO,.
Northern Blot Analysis-Total cellular and nuclear RNA preparations were obtained by the single-step method using acidic guanidinium thiocyanate (15). In some cases, cytoplasmic RNA was prepared from the supernatant of cells lysed using Nonidet P-40 lysis buffer (16). Samples (5 or 10 pg) of total RNA were electrophoresed through 1.5% agarose gels containing 2.2 hf formaldehyde, 20 mM MOPS (pH 7.0), 5 my NaOAc, and 1 mhf EDTA (17). After electrophoresis, RNA was stained with ethidium bromide, washed, photographed, blotted by capillary transfer using 20 x SSC onto Hybond-N nylon membranes (Amersham Corp.), and cross-linked to membranes by W irradiation ( U V Stratalinker 1800, Stratagene, CA). Blots were prehybridized a t 50 "C in 50% formamide, 4.8 x SSC, 5 x Denhardt's solution, 50 rnM HEPES (pH 7.3), 0.5% SDS, 200 pg/ml denatured salmon sperm DNA, and hybridized overnight with :lZP-labeled probes at 42 "C in the same buffer. Blots were initially washed in 1 x SSC, 0.5% SDS a t 23 "C followed by high stringency washes in 0.1 x SSC, 0.1% SDS a t 50 "C, and were exposed to Kodak X-AR5 film. Transcript levels were quantified using a Betascope (Betagen, Waltham, MA). Blots were stripped by two or more additions of boiling 0.1% SDS for 15 min incubations each.
Metabolic Labeling-HMC-1 cells (2 x lofi cells/ml) were stimulated with PMA for various time intervals as indicated. After stimulation, the cells were washed and preincubated for 30 min in methionine-and cysteine-free Dulbecco's modified Eagle's medium plus 10% dialyzed fetal bovine serum, and were then labeled for 2 h in the presence of 50 pCi/ml each of ["Slmethionine and ["Slcysteine (Expre""S:'"S; DuPont NEN). Supernatants were harvested, centrifuged for 5 min a t 13,000 rpm in Eppendorf tubes a t 4 "C, and stored at -80 "C.
Immunoprecipitation and SDS-Polyacrylamide Gel Electrophoresis-Aliquots of labeled culture supernatant (200 pl) were precleared with 5 pl of rabbit preimmune serum or a control monoclonal antibody (mAb), followed by 100 pl of pansorbin (Calbiochem) as a 10% (v/v) suspension in NET buffer (150 m M NaCI, 5 mM EDTA, 50 mM Tris, pH 8.0) plus 0.1% Nonidet P-40. Subsequent immunoprecipitation was performed using 5 pl of rabbit preimmune serum, control mAb, rabbit anti-1-309 serum (4), or anti-MCP-1 mAb (PeproTech, Rocky Hill, NJ), followed by 100 pl of protein A-Sepharose (Pharmacia LKB Biotechnology AB, Uppsala, Sweden) as a 30 mg/ml suspension in NET buffer plus 0.1% Nonidet P-40 and 0.25% gelatin. Beads were washed three times in NET buffer, and immunoprecipitated proteins were eluted by boiling for 3 min in 40 pl of Laemmli sample buffer. Alternatively, total proteins were analyzed by precipitation with ice-cold trichloroacetic acid. All samples were electrophoresed through 15% discontinuous SDS-polyacrylamide gels according to the method of Laemmli (18). Proteins were visualized by fluorography (19) using Kodak X-AR5 film. the expression of chemokine transcripts in the mast cell line HMC-1 (Fig. 1). Unstimulated HMC-1 cells express moderate levels of 1-309 and MCP-1 transcripts and very low levels of RANTES transcripts, but do not express detectable levels of MIP-la, MIP-1P, and IL-8 transcripts. The detected basal level of 1-309 and MCP-1 transcripts in part reflects serum responsiveness, since both of these transcripts are moderately elevated by incubation of HMC-1 in fresh serum containing medium (data not shown).

Expression
HMC-1 cells were stimulated with the phorbol ester PMA and the calcium ionophore ionomycin alone and in combination (Fig. 1). PMAwas found to up-regulate 1-309, MCP-1, and RAN-TES transcripts and to induce the expression of MIP-la, MIPlp, and IL-8 transcripts. In all instances, de nouo protein synthesis was required for transcript induction, as induction was abolished by pretreatment with cycloheximide (data not shown). Ionomycin alone had no effect on chemokine transcript expression and did not potentiate the effects of PMA. Although chemokine transcripts were readily detected in resting and stimulated HMC-1 cells, IL-2 transcripts were not detected under any condition tested (data not shown). This result is consistent with previous data analyzing other human and murine mast cell samples, which identified transcripts encoding a number of cytokines, but not IL-2.
The observed sizes for the various chemokine transcripts in HMC-1 were consistent with previous reports examining expression in other cell types (1,2). terization in activated T-cells suggested that the two species represent differentially polyadenylated forms of 1-309 tran-Examination of a peripheral blood T-cell line LPL-1 revealed a distinct pattern of chemokine transcript expression (Fig. 1). Unstimulated T-cells expressed FtANTES transcripts exclusively. Following stimulation of T-cells with the mitogen PHA, 1-309, MIP-la, and MIP-10 transcripts were all induced to high levels, whereas MCP-1 and IL-8 transcripts remained undetectable. A similar pattern of chemokine expression by activated T-cells has been observed previously (4, 20, 21). De novo protein synthesis was required for the induction of 1-309 transcripts, but, consistent with previous data (211, was not required for the induction of MIP-la and MIP-1P transcripts (data not shown). Notably, although PMA treatment was an effective inducer of chemokine transcripts in HMC-1, PMA treatment has been shown to be ineffective at inducing chemokine expression in T-cells (4, 21). Taken together, these data indicate that human mast cells can be a source of multiple chemokines and that the array of chemokines expressed by mast cells may be broader than that expressed by activated T-cells.
Secretion of Chemokine Proteins by Human Mast Cell Leukemia HMC-I-Because HMC-1 could express high levels of chemokine transcripts, we sought to determine whether these cells were sources of secreted chemokine peptides. HMC-1 cells that were either unstimulated or stimulated for 4 or 8 h were metabolically labeled with a 2-h pulse of [3sS]methionine and [3sSlcysteine, and culture supernatants were harvested and immunoprecipitated using anti-1-309 and anti-MCP-1 antibodies (Fig. 2). Low levels of 16-and 8-kDa 1-309 peptides, and low levels of 15-and 11-kDa MCP-1 peptides were detected in supernatants of unstimulated cells. The 16-and 8-kDa 1-309 species are the expected sizes of glycosylated and nonglycosylated 1-309, respectively (22). The two forms of MCP-1 are presumed to represent differentially glycosylated forms of this molecule, which would be consistent with previous studies (23, 24).

M. Miller, personal communication.
cally up-regulated by PMA stimulation. Additional forms of 1-309 were detected, including a prominent 12-kDa form. However, the degree of 1-309 protein heterogeneity varied in different experiments (data not shown). Similarly, a novel 14-kDa MCP-1 species was also secreted. These distinct 1-309 and MCP-1 species are thought to represent differentially glycosylated forms of the respective proteins, but their precise structures are uncertain.
Differential Glucocorticoid Sensitivity of Chemokine Expression in HMC-1 and T-cells-Anti-inflammatory glucocorticoids are known to have potent down-regulatory effects on cytokine expression in T-cells and in some other cell types (25)(26)(27)(28)(29)(30)(31)(32)(33). To understand the effects of glucocorticoids on HMC-1 chemokine expression, cells were incubated with various corticosteroids, including the glucocorticoids methylprednisolone, hydrocortisone, and dexamethasone, and the nonglucocorticoid progesterone, with or without simultaneous PMAinduction. In unstimulated cells, basal expression of 1-309 and MCP-1 transcripts was assayed as a function of treatment with a range of corticosteroid concentrations (Fig. 3A). 1-309 transcripts were unaffected by treatment with 0.1-10 p~ doses of methylprednisolone, hydrocortisone, or dexamethasone. On the other hand, levels of MCP-1 transcripts were detectably inhibited by as little as 0.1 PM doses of these compounds and were maximally inhibited by 1 p~ doses. Significant inhibition was observed with either 2 h (data not shown) or 4 h (Fig. 3A) of treatment. Progesterone, which does not possess glucocorticoid activity, failed to inhibit the induction of MCP-1 transcripts. Thus, inhibition of MCP-1 transcript levels is a specific property of glucocorticoids and is not a generalized property of corticosteroids. Of the three glucocorticoids tested, methylprednisolone and dexamethasone were more potent inhibitors than hydrocortisone; methylprednisolone was chosen for further experiments.
Glucocorticoid effects on PMA-inducible expression of chemokine transcripts in HMC-1 were investigated next (Fig. 3B). Cells were transferred into medium containing fresh serum and were stimulated with PMA in the presence or absence of 10 V M methylprednisolone for 4 or 8 h. As a control, cells were transferred ints medium containing fresh serum and were harvested after 4 or 8 h with no further treatment. With fresh serum alone, both 1-309 and MCP-1 transcripts are mildly and transiently induced, resulting in a decay in transcript levels between 4 and 8 h after transfer. As observed previously (Fig.  l), PMA treatment resulted in dramatic increases in 1-309 and MCP-1 transcripts and induction ofMIP-la, MIP-1P, and RAN-TES transcripts. Although MCP-1 transcripts were still induced by PMA treatment in the presence of methylprednisolone, the induced levels were decreased by 48 and 73% after 4 and 8 h, respectively, relative to cells treated with PMA alone. In contrast, the levels of 1-309, MIP-la, MIP-lP, and RANTES mRNAs were unaffected.
To evaluate the mechanism by which methylprednisolone reduces MCP-1 mRNAlevels, two experiments were performed. First, we isolated nuclear and cytoplasmic RNA fractions to analyze basal transcript levels in HMC-1 (Fig. 4A). Analysis of untreated and PMA treated control cells indicated that the fractionation protocol was effective. For example, the nuclear fractions displayed relatively low levels of 0.55-kb 1-309 transcripts and higher levels of a transcript that was slightly larger than the 2.4-kb transcripts found in the cytoplasm. Whereas the cytoplasmic 2.4-kb transcript is known to represent a fully spliced form that utilizes a distal polyadenylation site, we assume that the larger nuclear form represents an unspliced 1-309 transcript that utilizes the proximal polyadenylation site. Such a transcript is predicted to be 2.8 kb long based upon the 1-309 genomic structure (34). Unspliced and partially spliced  ; lanes 11-13). Concentrations of each compound tested were 10. 1, and 0.1 PM. A Northern blot of RNA samples was sequentially hybridized with the indicated ""P-labeled 1-309 and MCP-1 probes. B, total RNA was isolated from HMC-1 cells that were either unstimulated or stimulated with PMA in the presence or absence of methylprednisolone ( M P ) , as indicated. A Northern blot of RNA samples was sequentially hybridized with the indicated :'2P-labeled probes.
forms of MCP-1 transcripts were also detected in the nuclear but not the cytoplasmic fractions.
Analysis of nuclear and cytoplasmic fractions from cells treated for 2 h with 10 p.1~ methylprednisolone showed levels of 1-309 transcripts to be unchanged in both compartments. However, levels of MCP-1 transcripts in both compartments were significantly reduced (60% reduction of nuclear, 488 reduction of cytoplasmic). These results suggest that methylprednisolone-mediated inhibition of MCP-1 mRNA is primarily a nisolone for 2 h and were then cultured further in the presence of the transcription inhibitor actinomycin-D (Fig. 4B). MCP-1 mRNA levels were found to decay with similar half-lives in methylprednisolone treated and untreated cells. Taken together, the results of these experiments argue that the methylprednisolone-induced decrease in MCP-1 mRNA levels results from an inhibition of MCP-1 gene transcription.
We next examined the effects of methylprednisolone treatment on chemokine transcript induction in T-cells stimulated with PMA plus anti-CD3 mAb ( Fig. 5A and data not shown). Strikingly, glucocorticoid treatment resulted in a 60-75% reduction in 1-309 transcripts and smaller ( 2 0 4 0 9 ) reductions in MIP-la, MIP-lP, and RANTES transcripts. Under these conditions, IL-2 transcript levels were reduced hy 75-95': (data not shown). Of note, the methylprednisolone-induced reduction in 1-309 transcript levels was specific to T-cells, since this reduction was not observed in HMC-1, even though methylprednisolone treatment was capable of down-regulating MCP-1 transcripts in these cells.
In order to evaluate the mechanism hy which methylprednisolone inhibits 1-309 transcript accumulation in T-cells. activated T-cells were cultured for 2 h in the presence or absence of methylprednisolone and were then cultured for additional time in the presence of actinomycin-D (Fig. 93). 1-309 transcript levels were found to decay with similar half-lives independent of methylprednisolone treatment, arguing that the effect of methylprednisolone must be at the level of 1-309 gene transcription. Consistent with this interpretation, both nuclear and cytoplasmic mRNA species were reduced when T-cells were activated in the presence of methylprednisolone (data not shown). Taken together, these results argue that the expression of individual chemokine genes in a mast cell line and in activated T-cells are differentially sensitive to glucocorticoid treatment. DISCUSSION Although the chemokines are now recognized as an important group of inflammatory mediators that can be produced by a variety of cell t-ypes following appropriate stimulation. mast cell expression of chemokines has not previously heen evaluated in a comprehensive fashion. In this report. we showed that the human mast cell leukemia line HMC-1 is a source of a wide array of chemokines, including the CC chemokines 1-309.
MCP-1, MIP-la, MIP-1P, and RANTES and the C-X-C chemokine IL8. Our results therefore confirm and extend a recent study by Moller et al. (35), who demonstrated inducihle IL-8 expression in HMC-1. These ohservations argue that mast cells, in addition to releasing potent mediators such as histamine and arachadonic acid metabolites immediately following stimulation, may be induced to synthesize and secrete a wide array of inflammatory cytokines at later times. Although mast cells have been shown to produce an array of c-ytokines generally associated with T-cells (fi-8), our data suggest that the range of chemokines produced by mast cells may he broader than that produced by T-cells. Since our study examines a mast cell leukemia that does not express FcrRs, it will he important to extend our results to normal human mast cells activated via FccR cross-linking in future studies.
Previous studies have shown that anti-inflammatory glucocorticoids are ineffective at inhibiting histamine release from human lung, intestine, or skin mast cells ( 3 6 ) . However, effects on mast cell cytokine synthesis have not been examined in detail. We find that glucocorticoids inhibit basal and inducihle transcription of the MCP-1 gene in HMC-I, hut fail to inhibit transcription of other C-C chemokines. MCP-1ME expression was previously shown to be inhibited hy glucocorticoids in stimulated human fibrosarcoma cells and synoviocytes 137,381, in stimulated murine vascular smooth muscle cells and fihroblasts (39, 401, and in ischemic rat kidneys (39), hut was not inhibited by glucocorticoids in stimulated human vein endothelial cells (39). This suggests that inhihition could he cell typeor stimulus-specific. Along this line, we note that glucocorticoids inhibited 1-309 expression in activated T-cells, hut not in HMC-1. Because MIP-la, MIP-la, and RANTES transcripts are unaffected by methylprednisolone in HJIC-I and are only mildly reduced in activated T-cells, we conclude that in both cell types, glucocorticoids are not globally effective inhibitors of the expression of chemokines, an important class of inflammatory cytokines.