MITF Is Necessary for Generation of Prostaglandin D 2 in Mouse Mast Cells*

Mast cells generate eicosanoids that are linked to asthma and other inflammatory diseases. A basic-helix-loop-helix leucine zipper transcription factor termed MITF is essential for the development of mast cells. Although other substances also linked to inflammatory reactions (such as various proteases and serotonin) re-quire MITF for their expression, the role of MITF in eicosanoid generation has not been studied. We examined eicosanoid generation in bone marrow-derived mast cells (BMMCs) of tg/tg mice that lack MITF. Most eicosanoids generated by BMMCs are either prostaglandin (PG) D 2 or leukotriene C 4 . The former is synthe- sized via the cyclooxygenase pathway, whereas the latter is synthesized via the 5-lipoxygenase pathway. In response to stimulation with IgE and antigens, BMMCs of tg/tg mice synthesized leukotriene C 4 normally. How- ever, neither immediate nor delayed PGD 2 production was detected in these BMMCs. This indicates that MITF is a transcription factor that specifically activates the cyclooxygenase pathway, but not the 5-lipoxygenase pathway. Significant decreases in expression of hematopoietic PGD 2 synthase (hPGDS, a terminal synthase for PGD 2 ) were observed at both mRNA and protein levels in tg/tg BMMCs. MITF transactivated the hPGDS gene via a CACCTG motif located in the promoter region. MITF appeared to be essential for generation of PGD 2 by enhancing expression of the hPGDS gene in BMMCs. the In some experiments, the amount of PGD 2 generated was measured in BMMCs preincubated with 1 (cid:4) g/ml indomethacin (Sigma). In these cases, BMMCs were sensitized with anti-DNP IgE in the presence of indomethacin, washed, and then elicited with DNP-HSA. were CodeLink (Amersham Biosciences) extracted with RNeasy (Qiagen, CA). Experimen- tal procedures, including the synthesis of double-stranded cDNA and biotin-labeled cRNA target, and the analysis were performed by Kurabo Co. Ltd. (Osaka, Japan).

Stimulation of BMMCs by cross-linking of Fc⑀RI with IgE and antigens elicits secretory granule exocytosis and immediate eicosanoid generation (15,16). BMMCs generate mainly prostaglandin D 2 (PGD 2 ) and LTC 4 (16). The former is generated via the cyclooxygenase pathway, whereas the latter is generated via the 5-lipoxygenase pathway. The immediate generation of PGD 2 and LTC 4 is completed within 10 min of stimulation. When BMMCs are cultured with Kit ligand (KitL), IL-1␤-and IL-10delayed synthesis of PGD 2 , but not of LTC 4 , occurs 2-8 h after stimulation with IgE and antigen (17)(18)(19)(20). The immediate synthesis of PGD 2 is mediated by constitutively expressed PG endoperoxide H synthase (PGHS)-1, whereas the delayed synthesis of PGD 2 is mediated by inductively expressed PGHS-2. PGH 2 is generated in BMMCs by both PGHS-1 and -2 and is unstable, quickly being metabolized to PGD 2 by hematopoietic PGD 2 synthase (hPGDS) (21). Recently, Stevens and coworkers (22) demonstrated that Ras guanine nucleotide-releasing protein (Ras-GRP) 4 regulates expression of the hPGDS gene and that mast cell lines with low RasGRP4 expression levels show defective generation of PGD 2 but normal generation of LTC 4 . RasGRP4 appeared to be involved in the cyclooxygenase pathway, but not in the 5-lipoxygenase pathway. In our study we examined eicosanoid generation in BMMCs of tg/tg mice and found that PGD 2 , but not LTC 4 , synthesis requires MITF. MITF appears to be involved in the cyclooxygenase pathway but not in the 5-lipoxygenase pathway.
Measurement of ␤-Hexosaminidase Release and Generation of Eicosanoids-The assay for ␤-hexosaminidase release was performed 30 min after elicitation with DNP-HSA. BMMCs were precipitated, and ␤-hexosaminidase activity in the supernatant and in the cell pellets (after lysis by freeze-thawing) was quantitated by spectrophotometric analysis of the hydrolysis of p-nitrophenyl-␤-D-2-acetamido-e-deoxyglucopyranoside (Sigma). The percent release of ␤-hexosaminidase was calculated by the formula [S/(S ϩ P)], where S and P are the ␤-hexosaminidase contents of equal portions of supernatant and cell pellet, respectively. PGD 2 and LTC 4 levels were quantified in the supernatant of elicited BMMCs using enzyme immunoassay kits according to the manufacturer's instructions (Cayman Chemical Company, Ann Arbor, MI). In some experiments, the amount of PGD 2 generated was measured in BMMCs preincubated with 1 g/ml indomethacin (Sigma). In these cases, BMMCs were sensitized with anti-DNP IgE in the presence of indomethacin, washed, and then elicited with DNP-HSA.
Transcript Analysis of Stimulated BMMCs of ϩ/ϩ or tg/tg Mice-The expression profiles of genes in ϩ/ϩ or tg/tg BMMCs stimulated with anti-DNP IgE alone or stimulated with anti-DNP IgE and then elicited with DNP-HSA for 30 min were examined with a CodeLink UniSet Mouse 20K I bioarray (Amersham Biosciences) using 2 g of total RNA extracted with an RNeasy column (Qiagen, Valencia, CA). Experimental procedures, including the synthesis of double-stranded cDNA and biotin-labeled cRNA target, and the analysis of results were performed by Kurabo Co. Ltd. (Osaka, Japan).
Quantification of mRNA Levels by Real-time RT-PCR-RNA was extracted from BMMCs using an RNeasy kit (Qiagen) with DNase I treatment. BMMCs cultured under the following four conditions were used: unstimulated, sensitized with anti-DNP IgE alone, sensitized with anti-DNP IgE and elicited with DNP-HSA for 30 min, or sensitized with anti-DNP IgE and elicited with DNP-HSA for 120 min. The mRNA levels for RasGRP4, PGHS-1, PGHS-2, hPGDS, and glyceraldehyde-3phosphate dehydrogenase genes were verified using a TaqMan Universal PCR Master Mix and Assays-on-Demand primers from Applied Biosystems (Foster City, CA). The primers and probes used for cell cycle regulators were Assays-on-Demand gene products. The mRNA levels for each gene were normalized to that of glyceraldehyde-3-phosphate dehydrogenase mRNA.
Nuclear Run-on Assay-A nuclear run-on assay was used to measure gene transcription rates. BMMCs (2 ϫ 10 7 ) were washed twice in phosphate-buffered saline, lysed in buffer containing 10 mM HEPES (pH 7.4), 10 mM NaCl, 3 mM MgCl 2 , and 0.5% Nonidet P-40, and incubated for 7 min on ice. Nuclei were isolated by centrifugation at 600 ϫ g for 5 min, resuspended in buffer containing 20 mM HEPES (pH 8.3), 5 mM MgCl 2 , 0.1 mM EDTA, and 40% glycerol, and stored until use. Frozen nuclei (100 l) were added to 100 l of buffer containing 20 mM HEPES-KOH (pH 8.0), 25% glycerol, 10 mM MgCl 2 , 0.2 mM KCl, 1.2 mM ATP, 0.6 mM CTP, and 0.6 mM GTP. After the addition of 40 units/ml RNase inhibitor and 100 Ci of [␣-32 P]UTP, the mixture was incubated at room temperature for 45 min. The labeled RNA was passed through a QIAshredder column (Qiagen) and purified with a Qiagen RNeasy kit according to the manufacturer's protocol. The labeled RNA from ϩ/ϩ and tg/tg BMMCs was hybridized to dot blots containing 2 g of purified fragments of hPGDS cDNA, ␤-actin cDNA (as a positive control), or pEF-BOS expression vector (as a negative control) immobilized onto nylon filters. Blots were hybridized for 48 h in Church buffer (7% SDS, 0.5 M phosphate buffer (pH 7.5), 1% bovine serum albumin). The blots were then washed twice in 2ϫ SSC, 0.1% SDS and once in 0.1ϫ SSC, 0.1% SDS. The membranes were then exposed to x-ray films.
Luciferase Assay-The DNA fragment containing the promoter and 5Ј-untranslated region of the hPGDS gene (nt Ϫ1500 to ϩ200, where ϩ1 is a transcription initiation site) was amplified by PCR from genomic DNA of ϩ/ϩ BMMCs with LA-Taq DNA polymerase (Takara, Kyoto, Japan). This fragment was cloned upstream of the luciferase gene, and the reporter plasmid was constructed. Reporter plasmids sequentially deleted in the promoter region or mutated at the CACCTG motif were also constructed by PCR. The sequence of all constructs was verified using an ABI 3100 sequencer (Applied Biosystems). 10 g of a reporter, 1 g of a pEF-BOS expression vector containing MITF cDNA or containing no insert, and 1 g of an expression vector containing the ␤-galactosidase gene were cotransfected into tg/tg BMMCs by electroporation (350 V, 500 F). In each transfection, 5 ϫ 10 6 tg/tg BMMCs were used. The cells were harvested 48 h after transfection, and the soluble extracts were assayed for luciferase and ␤-galactosidase activity. The normalized value by ␤-galactosidase activity was expressed as relative luciferase activity. The effect of MITF on the reporter plasmid was demonstrated by the ratio of the relative luciferase activity with MITF to the relative luciferase activity without MITF (-fold activation).
Electrophoretic Gel Mobility Shift Assay-Production of the fusion protein containing glutathione S-transferase (GST) and MITF was as previously described (23). To examine whether MITF bound to the CACCTG motif mediating the transactivation ability, an oligonucleotide containing this motif was used as a probe. The sequence of the FIG. 2. Amounts of generated PGD 2 and LTC 4 following stimulation by IgE and antigens in ؉/؉ and tg/tg BMMCs. BMMCs were sensitized with 1 g/ml anti-DNP IgE for 2 h, washed, and elicited with 100 ng/ml DNP-HSA. The quantities of PGD 2 and LTC 4 generated were measured 30 min after elicitation. The amounts of PGD 2 and LTC 4 in BMMCs that had not been elicited were also measured. The values represent the mean Ϯ S.E. of three experiments. In some cases, the S.E. was too small to be shown by bars. *, p Ͻ.01 by t test when compared with the amount in non-elicited BMMCs.
After incubation at 37°C for 15 min, the reaction mixture was subjected to electrophoresis at 14 volt/cm at 4°C on a 5% polyacrylamide gel in 0.25ϫ TBE buffer (1ϫ TBE is 90 mmol/liter Tris-HCl, 64.6 mmol/liter boric acid, and 2.5 mmol/liter EDTA, pH 8.3). In some experiments, GST was used as a protein instead of GST-MITF. Non-labeled oligonucleotides containing the CACCTG motif or non-labeled oligonucleotides mutated from CACCTG to CTCCAG were added in a competition assay. Polyacrylamide gels were dried on Whatman 3MM chromatography paper and subjected to autoradiography.

Effect of MITF on the Response of BMMCs Stimulated with
IgE and Antigens-The effect of MITF on the response of BMMCs after stimulation with IgE and antigens had not previously been investigated. We examined this by measuring ␤-hexosaminidase release from BMMCs of tg/tg mice that effectively lacked MITF (5, 23) in response to IgE and antigens. BMMCs were sensitized with anti-DNP IgE, washed, and elicited by various concentrations of DNP-HSA. The released ␤-hexosaminidase was measured 30 min after elicitation. Levels of released ␤-hexosaminidase increased in a dose-dependent manner from 5 to 100 ng/ml DNP-HSA, thereafter reaching a plateau (Fig. 1). The ϩ/ϩ and tg/tg BMMCs showed comparable patterns of ␤-hexosaminidase release after stimulation with IgE and antigens (Fig. 1).
Production of Eicosanoids in BMMCs Derived from tg/tg Mice-We next examined the generation of eicosanoids in BMMCs of tg/tg mice. BMMCs sensitized with anti-DNP IgE were elicited with 100 ng/ml DNP-HSA, at which concentration the effect of DNP-HSA reached a plateau (Fig. 1). The concentrations of generated PGD 2 and LTC 4 (the major synthesized eicosanoids in BMMCs) were measured 30 min after elicitation. The amount of generated PGD 2 was significantly lower in tg/tg BMMCs than in ϩ/ϩ BMMCs (Fig. 2). In contrast, the amount of generated LTC 4 was comparable between ϩ/ϩ and tg/tg BMMCs (Fig. 2).
Because the generation of PGD 2 , but not of LTC 4 , was defective in tg/tg BMMCs, we subsequently concentrated our study on PGD 2 production. We examined the process of PGD 2 generation elicited by 100 ng/ml DNP-HSA (Fig. 2). Because there was a possibility that tg/tg BMMCs responded at a higher concentration of antigen, we examined generation of PGD 2 at various concentrations of DNP-HSA. At all examined concen- trations (even 1 g/ml DNP-HSA), PGD 2 generation was barely detectable in tg/tg BMMCs (Fig. 3A).
The time-dependent pattern of PGD 2 generation elicited with 100 ng/ml DNP-HSA was compared between ϩ/ϩ and tg/tg BMMCs. In ϩ/ϩ BMMCs, PGD 2 generation was maximal within the first 15 min, and the amount of generated PGD 2 did not change over the next 8 h (Fig. 3B). In contrast, PGD 2 generation was barely detectable in tg/tg BMMCs at any time during the period of examination (Fig. 3B).
Delayed PGD 2 Production in tg/tg BMMCs-Next, we examined delayed phase of PGD 2 generation in tg/tg BMMCs. Murakami et al. (17) reported that a delayed phase of PGD 2 generation was hardly detectable in BMMCs cultured in medium containing IL-3 alone. Their observations are consistent with the result shown in Fig. 3B in which BMMCs cultured with IL-3 alone were used. We changed the medium for ϩ/ϩ and tg/tg BMMCs to one containing IL-10, IL-1␤, and KitL, in which ϩ/ϩ BMMCs had been reported to show delayed phase generation of PGD 2 (17)(18)(19)(20). After culturing BMMCs in this medium for 2 h with anti-DNP IgE, 100 ng/ml DNP-HSA was added. As previously reported (17), the presence of IL-10, IL-1␤, and KitL induced generation of PGD 2 in ϩ/ϩ BMMCs without the need for elicitation by DNP-HSA, although in relatively small amounts (0.52 Ϯ 0.02 ng/10 6 cells, Fig. 3C). Elicited ϩ/ϩ BMMCs generated PGD 2 during the first 15 min, followed by a gradual increase of PGD 2 generation over the next 8 h (Fig.  3C). Such patterns of PGD 2 generation were not detected in tg/tg BMMCs (Fig. 3C).
To reveal only the delayed phase of PGD 2 generation, the immediate phase was eliminated by preincubation of BMMCs with 1 g/ml indomethacin for 2 h. This reagent inactivates PGHS-1, which is necessary for the immediate phase of PGD 2 generation; the PGD 2 synthesized in BMMCs preincubated with this reagent reflects the delayed phase of PGD 2 generation (18). BMMCs were preincubated with indomethacin in the presence of IL-10, IL-1␤, and KitL, sensitized with anti-DNP IgE, washed, and then elicited with DNP-HSA. The ϩ/ϩ BMMCs synthesized ϳ2 ng of PGD 2 /10 6 cells 8 h after elicitation with DNP-HSA (Fig. 3D). A similar delayed phase of PGD 2 generation was barely detectable in tg/tg BMMCs (Fig. 3D).
Deficient Expression of hPGDS Gene in tg/tg BMMCs-The deficient PGD 2 generation observed in both immediate and delayed phases suggests that the expression of some gene(s) related to the synthesis of PGD 2 might be defective in tg/tg BMMCs. We examined the expression of genes related to PGD 2 synthesis using BMMCs cultured in IL-3. The expression levels of genes participating in eicosanoid generation between ϩ/ϩ and tg/tg BMMCs were compared using CodeLink UniSet mouse expression bioarrays. BMMCs were examined under the following two conditions, BMMCs sensitized with anti-DNP IgE or BMMCs sensitized with anti-DNP IgE and then elicited with DNP-HSA for 30 min. Under both conditions, the levels of expression of hPGDS mRNA were significantly lower in tg/tg BMMCs than in ϩ/ϩ BMMCs (Table I). Other genes had comparable expression levels between ϩ/ϩ and tg/tg BMMCs under both conditions. We used real-time RT-PCR to confirm the apparently defective hPGDS expression in tg/tg BMMCs. We also quantified the levels of mRNAs from three genes related to PGD 2 generation, RasGRP4, PGHS-1, and PGHS-2. We examined BMMCs under the following four conditions, BMMCs without any stimulation, BMMCs sensitized with anti-DNP IgE, BMMCs sensitized with anti-DNP IgE and then elicited with DNP-HSA for 30 min, or BMMCs sensitized with anti-DNP IgE and then elicited with DNP-HSA for 120 min. Under all conditions examined, the hPGDS gene was expressed in tg/tg BMMCs at levels less than one tenth those observed in ϩ/ϩ BMMCs (Fig. 4). In contrast, the expression levels of RasGRP4 and PGHS-2 genes were comparable between ϩ/ϩ and tg/tg BMMCs under all conditions. The expression level of the PGHS-1 gene in tg/tg BMMCs was approximately one third that found in ϩ/ϩ BMMCs, but this magnitude of reduction was small compared with that observed for the hPGDS gene (Fig. 4).
The expression levels of hPGDS, RasGRP4, and PGHS-1 genes were unchanged under the four experimental conditions, whereas the expression level of the PGHS-2 gene increased after stimulation with anti-DNP IgE (Fig. 4). The PGHS-2 expression level further increased following elicitation by DNP-HSA. The magnitude of change in PGHS-2 expression levels induced by stimulation with IgE and antigens was comparable between ϩ/ϩ and tg/tg BMMCs (Fig. 4).
Expression of hPGDS, PGHS-1, and PGHS-2 genes was examined in terms of protein concentrations using BMMCs without any stimulation. As in the case of mRNA expression, hPGDS protein was detected in ϩ/ϩ BMMCs but not in tg/tg BMMCs (Fig. 5). The amount of PGHS-1 protein detected was comparable between ϩ/ϩ and tg/tg BMMCs, and no PGHS-2 protein was detected in either BMMC type (Fig. 5). a Transcript showing more than 10-fold difference of expression level between ϩ/ϩ and tg/tg BMMCs that received the same stimuli (IgE or IgE/Ag).
b Expression level of PGHS-2 mRNA was too low to detect in this method.
Enhancement of hPGDS Transcription by MITF-Because real-time RT-PCR and immunoblotting indicated the steady state amounts of hPGDS mRNA and protein, we examined directly whether transcription of the hPGDS gene was defective in tg/tg BMMCs using a nuclear run-on assay. The transcription rate of the hPGDS gene was significantly higher in ϩ/ϩ BMMCs than in tg/tg BMMCs (Fig. 7). Then we examined the motif mediating the transactivation effect of MITF. The reporter plasmid containing the promoter region of the hPGDS gene (Ϫ1500 to ϩ200) was transfected with or without MITF. Luciferase activity increased ϳ5-fold when transfected with MITF as compared with transfection without MITF (Fig. 8A). This MITF-related increase in luciferase activity was detected even in the reporter plasmid that contained the promoter region starting at Ϫ200 (Fig. 8A). MITF recognizes a CANNTG (N is any nucleotide) motif. Only one CANNTG motif was present in the promoter region starting at Ϫ200 (CACCTG between Ϫ22 and Ϫ17). To examine whether MITF transactivated the hPGDS promoter via this motif, we constructed a reporter plasmid mutated from CACCTG to CTCCAG (the mutated nucleotides are underlined). The mutated reporter plasmid did not show increased luciferase activity mediated by MITF (Fig. 8A).
The binding of MITF to the CACCTG motif was examined by electrophoretic gel mobility shift assay. When a GST-MITF fusion protein was added to labeled oligonucleotide containing a CACCTG motif, a retarded band was detected (Fig. 8B). This retarded band was not found when GST protein alone was added (Fig. 8B). The intensity of this retarded band was gradually weakened by additions of cold oligonucleotide with the CACCTG motif (10-, 50-, and 100-fold molar excess), but not by adding cold oligonucleotide mutated at the CACCTG motif (Fig.  8B). These results demonstrate the specific binding of MITF to the CACCTG motif.

DISCUSSION
BMMCs generate PGD 2 immediately after stimulation with IgE and antigens. When BMMCs are cultured in a medium containing IL-10, IL-1␤, and KitL, the amount of generated PGD 2 increases further up to 8 h after stimulation. These immediate and delayed phases of PGD 2 generation were not observed in BMMCs derived from tg/tg mice, indicating that MITF is essential for both phases of PGD 2 generation. Among the genes related to eicosanoid generation, hPGDS showed significant decreases in expression level in tg/tg BMMCs. The hPGDS protein converts unstable PGH 2 generated by PGHS-1 and -2 into PGD 2 . As detected by real-time PCR, the amount of hPGDS mRNA in tg/tg BMMCs was approximately one tenth that observed in ϩ/ϩ BMMCs. The hPGDS protein was barely detected in tg/tg BMMCs. A nuclear run-on assay revealed that transcription of the hPGDS gene was defective in tg/tg BMMCs. MITF appeared to transactivate the hPGDS promoter by binding to the CACCTG motif. Urade and coworkers (21,24) reported that mice with a disrupted hPGDS gene were defective in both the immediate and delayed phases of PGD 2 generation. This indicates that hPGDS is a key enzyme for PGD 2 generation. The defects in PGD 2 generation observed in tg/tg BMMCs may be attributable to decreased expression of the hPGDS gene.
Stevens and coworkers (22) recently reported that RasGRP4 regulated the generation of PGD 2 . RasGRP4 is a mast cellspecific guanine nucleotide exchange factor and is thought to act downstream of KIT, a receptor for KitL (25). Overexpression of RasGRP4 in a human mastocytoma cell line increases hPGDS mRNA levels and the amount of generated PGD 2. In- FIG. 4. Quantification of hPGDS, RasGRP4, PGHS-1, and PGHS-2 transcripts in ؉/؉ and tg/tg BMMCs. The mRNA levels of hPGDS, RasGRP4, and PGHS-1 and -2 genes were quantified by realtime PCR. BMMCs were unstimulated (designated by IgE Ϫ Ag Ϫ ), sensitized with 1 g/ml anti-DNP IgE (IgE ϩ Ag Ϫ ), sensitized with 1 g/ml anti-DNP IgE, and elicited with 100 ng/ml DNP-HSA for 30 min (IgE ϩ Ag 30 ) or sensitized with 1 g/ml anti-DNP IgE and elicited with 100 ng/ml DNP-HSA for 120 min (IgE ϩ Ag 120 ). Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous reference mRNA to standardize mRNA concentrations. The amount of each mRNA was divided by the amount found in unstimulated ϩ/ϩ BMMCs, and the resultant value is shown as a relative quantity of mRNA. Values represent the mean Ϯ S.E. of three experiments. In some cases, the S.E. was too small to be shown by bars. hibition of RasGRP4 expression with siRNA in a rat mastocytoma cell line decreases hPGDS protein levels. RasGRP4 appears to act as an upstream regulator for hPGDS transcription. In tg/tg BMMCs, RasGRP4 expression levels were comparable with those observed in ϩ/ϩ BMMCs. However, we have not examined the expression level of RasGRP4 at the protein level, and the possibility that MITF regulates RasGRP4 protein expression remains. Further analyses using an anti-RasGRP4 antibody are needed to address this possibility. An alternative is that MITF acts downstream of RasGRP4. RasGRP4 acts downstream from KIT, and Fisher and colleagues (26) reported that MITF was also a downstream molecule of KIT. The signal from RasGRP4 may activate MITF and subsequently activate expression of hPGDS. RasGRP4 is known to activate the cyclooxygenase pathway, but not the 5-lipoxygenase pathway (22). MITF also activates the cyclooxygenase pathway alone, which supports the hypothesis that MITF is a downstream molecule for RasGRP4.
Murakami et al. (17,18) reported that elicitation with IgE and antigens increased the amount of PGHS-2 mRNA in BMMCs. In fact, the addition of IgE to BMMCs increased the expression level of PGHS-2 mRNA, and elicitation with the antigen resulted in a further increase. The magnitude of the increase in PGHS-2 mRNA levels was comparable between ϩ/ϩ and tg/tg BMMCs. Recently, Inoue et al. (27) reported that a metabolite of PGD 2 named 15-deoxy-⌬ (12, 14)-PGJ 2 bound the nuclear peroxisome proliferator-activated receptor (PPAR) ␥ and that PPAR␥ negatively regulated the expression of PGHS-2 in macrophage cell lines. Because tg/tg BMMCs defective in PGD 2 generation normally enhanced the mRNA level of the PGHS-2 gene after stimulation with IgE and antigens, the negative feedback loop between PGD 2 and PGHS-2 observed in macrophage cell lines does not appear to apply to BMMCs.
In contrast to the changes in PGHS-2 mRNA levels, the changes in PGHS-2 protein levels in tg/tg BMMCs differed from those observed for ϩ/ϩ BMMCs. The PGHS-2 protein was barely detected in ϩ/ϩ BMMCs during the examined period (for 8 h after elicitation). This was consistent with a previous report by Murakami et al. (17) that ϩ/ϩ BMMCs cultured in IL-3 did not express PGHS-2 protein. In contrast to the case of ϩ/ϩ BMMCs, PGHS-2 protein was unexpectedly detected in tg/tg BMMCs, with levels peaking 2 h after elicitation. Expres-FIG. 6. Time-dependent changes in amounts of hPGDS, PGHS-1, and PGHS-2 proteins in BMMCs of ؉/؉ and tg/tg mice. Whole cell extracts of BMMCs established with rmIL-3 were immunoblotted. BMMCs sensitized with 1 g/ml anti-DNP IgE (shown at time 0) or sensitized with 1 g/ml anti-DNP IgE and elicited with 100 ng/ml DNP-HSA for various periods (from 1 min to 8 h) were examined. Two independent experiments were performed, and a representative result is shown.

FIG. 7. Transcriptional defects in the hPGDS gene in tg/tg
BMMCs. Nuclei were isolated from ϩ/ϩ and tg/tg BMMCs, and transcripts were labeled. The labeled RNA from ϩ/ϩ and tg/tg BMMCs was hybridized to dot blots containing purified fragments of hPGDS cDNA, ␤-actin cDNA (as a positive control), or pEF-BOS expression vector (as a negative control) immobilized onto nylon filters.
FIG. 8. Transactivation by MITF through the CACCTG motif in the promoter region of the hPGDS gene. A, luciferase assay using reporter plasmids containing sequentially deleted or mutated versions of the hPGDS promoter. The reporter plasmid and the effector plasmid (expression plasmid containing MITF cDNA or expression vector alone) were transfected, and the relative luciferase activity in the presence of MITF was divided by the relative luciferase activity in the absence of MITF. The divided value is shown as -fold activation. The box in the hPGDS promoter region indicates the CACCTG motif. B, electrophoretic gel mobility shift assay showing that MITF can bind specifically to the CACCTG motif. An oligonucleotide containing the CACCTG motif was labeled. A retarded band formed with the addition of GST-MITF, but not with the addition of GST alone. Excess cold oligonucleotides with the CACCTG motif inhibited retarded band formation, but excess cold oligonucleotides mutated at the CACCTG motif did not. sion of the PGHS-2 gene might be regulated at a translational level, and the mechanism negatively regulating PGHS-2 translation might be defective in tg/tg BMMCs. Several prostanoids, such as PGE 2 , are known to augment the induction of PGHS-2 (24). Another possibility is that tg/tg BMMCs might be more sensitive to such prostanoids than ϩ/ϩ BMMCs.
KitL induces the expression of PGHS-2 protein (17). We recently reported that the expression of KIT and the response to KitL were partially impaired in tg/tg BMMCs (28). However, the addition of KitL induced the expression of PGHS-2 protein to comparable levels in ϩ/ϩ and tg/tg BMMCs (data not shown). The partially deficient signal from KIT appeared to be sufficient for induction of PGHS-2 protein in tg/tg BMMCs. PGD 2 is known to recruit eosinophils (3,4). Recently we found defective eosinophil recruitment in tg/tg mice (29), suggesting that MITF may play an important role in recruitment of eosinophils through production of PGD 2 . MITF appears to be a key transcription factor regulating the function of mast cells.