Transcriptional regulation of the mucosal mast cell-specific protease gene, MMCP-2, by interleukin 10 and interleukin 3.

Although transcription of mast cell (MC) secretory granule neutral protease genes has been shown to distinguish MC subclasses in mucosal and serosal environments, the specific cytokines that regulate the expression of these genes have not been determined. To examine cytokine-mediated gene regulation, bone marrow-derived MC (BMMC) differentiated in vitro were obtained by culturing mouse bone marrow progenitor cells in the presence of WEHI-3 cell-conditioned medium, concanavalin A-stimulated splenocyte-conditioned medium (BMMCC), or recombinant (r) interleukin (IL)-3 (BMMCIL-3). All three populations of BMMC expressed the serosal MC-specific transcripts that encode mouse MC serine protease (MMCP)-5, MMCP-6, and MC carboxypeptidase A. However, only BMMCC contained MMCP-2 mRNA, a late expressed gene selectively transcribed by intestinal mucosal MC that proliferate during helminthic infestation in response to the T cell-derived cytokines IL-3, IL-4, and IL-10. When BMMCIL-3 were exposed to rIL-10 in the presence of either rIL-3 or rIL-4, they expressed MMCP-2 mRNA. Not only was the transcription of the MMCP-2 gene in BMMC dependent on continuous exposure of the cells to rIL-10, but the level of MMCP-2 mRNA in these cells could be down-regulated by rIL-3. These studies comparing the effects of two cytokines on the transcriptional regulation of secretory granule protease genes in MC demonstrate that rIL-10 induces BMMCIL-3 to express the mucosal MC protease MMCP-2, that rIL-3 attenuates the rIL-10-induced expression of this gene, and that transcription of the MMCP-2 gene is reversed in the absence of rIL-10.

Although transcription of mast cell (MC) secretory granule neutral protease genes has been shown to distinguish MC subclasses in mucosal and serosal environments, the specific cytokines that regulate the expression of these genes have not been determined. To examine cytokine-mediated gene regulation, bone marrow-derived MC (BMMC) differentiated in vitro were obtained by culturing mouse bone marrow progenitor cells in the presence of WEHI-3 cell-conditioned medium, concanavalin A-stimulated splenocyteconditioned medium (BMMCc), or recombinant (r) interleukin (1L)-3 (BMMCIL-3). All three populations of BMMC expressed the serosal MC-specific transcripts that encode mouse MC serine protease (MMCP)-B, MMCP-6, and MC carboxypeptidase A. However, only BMMCc contained MMCP-2 mRNA, a late expressed gene selectively transcribed by intestinal mucosal MC that proliferate during helminthic infestation in response to the T cell-derived cytokines IL-3, IL-4, and IL-10. When BMMCIL-3 were exposed to rIL-10 in the presence of either rIL-3 or rIL-4, they expressed MMCP-2 mRNA. Not only was the transcription of the MMCP-2 gene in BMMC dependent on continuous exposure of the cells to rIL-10, but the level of MMCP-2 mRNA in these cells could be down-regulated by rIL-3. These studies comparing the effects of two cytokines on the transcriptional regulation of secretory granule protease genes in MC demonstrate that rIL-10 induces BMMCIL-3 to express the mucosal MC protease MMCP-2, that rIL-3 attenuates the rIL-10-induced expression of this gene, and that transcription of the MMCP-2 gene is reversed in the absence of rIL-10.
Phenotypically distinct subclasses of mast cells (MC)' have been identified in the mouse based on the presence or absence in their secretory granules of varied combinations of a car-* This work was supported by Grants AI-22531, AI-23401, AI-23483, AR-35907, HL-36110, and RR-05950 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
As assessed by their content of protein and/or RNA, mouse serosal MC (the connective tissue MC prototype) express abundant levels of mouse MC carboxypeptidase A, mouse MC protease (MMCP)-3, MMCP-4, MMCP-5, and MMCP-6, but no detectable levels of MMCP-1 and MMCP-2. In contrast, mucosal MC in the intestines of helminth-infected mice express MMCP-1 and MMCP-2, but little if any MC carboxypeptidase A, and no detectable MMCP-5 or MMCP-6. Pure but immature populations of mouse MC have been derived by culturing hematopoietic progenitor cells in interleukin (IL)-3-rich conditioned medium (10-14). The transcription phenotype of mouse bone marrow-derived MC (BMMC) obtained with concanavalin A-stimulated splenocyte-conditioned medium (BMMCc) has not been determined. However, cells obtained with WEHI-3 cell-conditioned medium (BMMCw) express those transcripts that encode three of the five proteases found in the secretory granules of mouse serosal MC (MC carboxypeptidase A (3), , and MMCP-6 (5)), but not the mucosal MC-specific protease MMCP-2 (6). Fresh bone marrow cells or BMMC differentiated in vitro can be used to reconstitute both subclasses of MC in MC-deficient WBB6F1-W/W mice in vivo (15)(16)(17)(18)(19). Moreover, mouse BMMCw that are subsequently cocultured with mouse 3T3 fibroblasts become morphologically, histochemically, biochemically, and functionally more similar to serosal MC than to mucosal MC (20-23). c-Kit ligand (24-27) and nerve growth factor (28) are two fibroblast-derived cytokines that induce mouse hematopoietic progenitor cells to express the histochemical phenotype of serosal MC.
Because an antibody to IL-4 prevents the increase in the number of mucosal MC that normally occurs in the intestines of Nippostrongylus brasiliensis-infected mice (29) and because increased numbers of lymphocytes that produce IL-4 and IL-10 are present in the lymph nodes of mice infected with the parasite Leishmania major (30), we considered that IL-4 and/ or IL-10 might induce immature BMMC to transcribe at least one of the protease genes that are selectively expressed by mucosal MC. We now show that BMMC, derived with recombinant (r) IL-3 alone and then maintained with rIL-4 and rIL-10, express abundant levels of MMCP-2 mRNA. In addition, we demonstrate that this high level of MMCP-2 mRNA in rIL-10-treated BMMC can be attenuated by treatment of these cells with rIL-3 and can be completely suppressed by withdrawal of rIL-10 from the culture medium. These findings demonstrate the principle of cytokine regulation of transcription of secretory granule neutral proteases and specifically illustrate the sequential interaction of IL-3 followed by IL-10 in achieving expression of the mucosal MC, subclass-specific gene, MMCP-2.

EXPERIMENTAL PROCEDURES
Cell Culture-Mouse BMMCc, BMMCw, and BMMClr,.3 were obtained by culturing bone marrow cells from the femurs and tibias of 6-16-week-old BALB/c mice (Jackson Laboratories, Bar Harbor, ME) for 2 days to 5 weeks a t 37 "C in a humidified atmosphere of 5% COY in the presence of enriched medium (RPMI 1640 with 100 units/ ml penicillin, 100 pg/ml streptomycin, 10 pg/ml gentamicin, 2 mM L-glutamine, 0.1 mM nonessential amino acids, 50 p M B-mercaptoethanol, and 10% fetal calf serum (GIBCO)) containing either 50% concanavalin A-stimulated splenocyte-conditioned medium (13), 50% WEHI-3 cell-conditioned medium (31, 32), or 100 units/ml rIL-3 (Genzyme, Boston, MA), respectively. BMMCII,.R were also cultured for up to 3 additional weeks in enriched medium containing various combinations of 100 units/ml mouse rIL-3 (Genzyme), rIL-4 (Genzyme), and/or rIL-10 (DNAX Research Institute, Palo Alto, CA). These concentrations of rIL-4 and rIL-10 are optimal for the proliferation of a transformed mouse mast cell line and for generating colonies of mouse mast cells from mesenteric lymph node cells (33). BMMCII..:~ cultured in the presence of rIL-10 alone did not remain viable. Thus, for those studies examining the role of rIL-10 in MC differentiation, the culture medium was supplemented with either rIL-3 or rIL-4. Maximal proliferation and the highest level of MMCP-6 mRNA were obtained when fresh bone marrow cells were cultured in medium containing a t least 50 units/ml rIL-3. Every 7 days, the nonadherent cells in each culture were transferred into a new culture flask that contained fresh medium and the appropriate cytokine(s). Mouse rIL-3 and rIL-4 were expressed in yeast and then purified; mouse rIL-10 was expressed in myeloma NS-1 cells (DNAX Research Institute). Control experiments were performed with mock supernatants from NS-1 cells in those experiments in which rIL-10 was used. WEHI-3 cell-conditioned medium was obtained by culturing WEHI-3 cells (line TIB-68; American Type Culture Collection, Rockville, MD) in enriched medium for 24 h (32). Concanavalin-A-stimulated splenocyte-conditioned medium was prepared by coculturing splenocytes from C57BL/6J and C3H mice (0.5 X lo6 cells/ml from each strain) in enriched medium containing 2 pg/ml concanavalin A in 75cm' Falcon culture flasks a t 37 "C in a humidified atmosphere of 5% COP (13). After 45 h, the splenocytes were removed by a 20-min centrifugation a t 120 X g, and the resulting conditioned medium was filtered through a 0.45-pm Millipore filter (Nalge Sybron Corp., Rochester, NY).
Histochemical and Biochemical Analysis of MC-Cytocentrifuge preparations of the cultured BMMC were stained with Alcian blue followed by safranin (34,35). Samples of the cell lysates of the MC were analyzed for their histamine content with a radioimmunoassay kit (AMAC, Westbrook, ME) (36), and for their carboxypeptidase A activities with a high pressure liquid chromatography assay that measures the generation of phenylalanine from hippuryl-L-phenylalanine (22). One milliunit of carboxypeptidase A activity is defined as that amount of enzyme needed to produce 1 nmol of phenylalanine/ min from hippuryl-L-phenylalanine.
Turnover of mRNA in BMMC and Investigation of the Reversibility of Transcription of the MMCP-2 Gene-The rate of disappearance of various mRNAs in BMMCc after treatment with actinomycin D was used to determine their half-lives. Total RNA was isolated from BMMCc (8.6 X lo6 cells/experiment) maintained in 25 ml of 50% concanavalin A-stimulated splenocyte-conditioned medium containing 5 pg/ml actinomycin D (Sigma) for 1, 2, or 6 h. RNA blots were prepared and probed either with a 24-mer oligonucleotide (5'-ACGGTATCTGATCGTCTTCGAACC-3') that corresponds to a region within 18 S rRNA (44,45) or with the cDNAs that encode MMCP-2, MC carboxypeptidase A, and SG-PG.
To determine whether the expression of MMCP-2 mRNA in BMMC is reversible, mouse bone marrow cells were cultured for 2-3 weeks in the presence of rIL-3, WEHI-3 cell-conditioned medium, or concanavalin A-stimulated splenocyte-conditioned medium. The resulting BMMC were washed and exposed to different culture media for an additional 2-4 weeks. Blots, prepared with RNA isolated from the different populations of MC, were probed with the MMCP-2 cDNA followed by the 18 S rRNA probe.
To determine when these transcripts were expressed during the differentiation process in conditioned media, RNA blots were prepared from fresh bone marrow cells and cells that had been cultured for 2,5, and 10 days in either WEHI-3 cellconditioned medium or concanavalin A-stimulated splenocyte-conditioned medium. The level of SG-PG mRNA was barely detectable in the starting bone marrow cells but increased progressively in both populations of MC during the 10 days of culture (Fig. 2) Days in Culture  FIG. 2. Time course of expression of transcripts that encode different constituents of the secretory granules of MC. Total RNA was isolated from fresh bone marrow cells (time 0) and from cells cultured for 2,5, and 10 days in the presence of 50% concanavalin A-stimulated splenocyte-conditioned medium (CCM) or 50% WEHI-3 cell-conditioned medium (WCM). RNA blots were prepared and probed with radiolabeled cDNAs that encode MMCP-2, MMCP-5, MMCP-6, SG-PG, MC carboxypeptidase A (MC-CPA), and actin. mRNA was below detection at all time points. At day 10 of culture, detectable levels of MMCP-2 mRNA and MMCP-5 mRNA were obtained in BMMCc, but the level of MC carboxypeptidase A mRNA was borderline and that of MMCP-6 mRNA was below detection (Fig. 2).
As assessed by SDS-polyacrylamide gel electrophoresis (data not shown), lysates of 3-week BMMCc contained slightly more 28-32-kDa proteins than lysates of 3-week BMMCw (n = 3), but the amount of these putative granule proteins was considerably less than in mouse serosal MC or in a mature Kirsten sarcoma virus-immortalized MC line. One million BMMCw contained only 275 f 0 ng of histamine and 5.4 f 5.9 milliunits of MC carboxypeptidase A enzymatic activity, and lo6 BMMCc contained only 525 f 141 ng of histamine and 4.5 f 1.9 milliunits of MC carboxypeptidase A enzymatic activity (mean f 1 h range, n = 2). The mean histamine and MC carboxypeptidase A levels per lo6 MC have been reported to be >20,000 ng and 2300 milliunits for mouse serosal MC and >20,000 ng and 1500 milliunits for mature mouse Kirsten sarcoma virus-immortalized MC (20, 22, 46). Thus, BMMC derived with conditioned medium from two different sources possessed distinct transcriptional phenotypes even though both populations of in vitro-derived MC were immature in appearance and by content of mediators.
rIL-3 and rIL-10 Regulation of mRNA Levels in Mouse Mast Cells-To determine if a specific cytokine could induce expression of MMCP-2 mRNA, fresh mouse bone marrow cells were cultured for 2 weeks in the presence of rIL-3 to obtain an immature MC population, designated BMMCIL.3. The BMMCIL.3 were harvested, washed, and cultured for 2 subsequent weeks in enriched medium containing various combinations of rIL-3, rIL-4, rIL-10, and mock supernatant from NS-1 cells. After 1 week, the number of MC (mean f S.D., n = 3-5) increased 1.5 f 0.9-fold in the rIL-3-treated cultures, 5.8 & 3.1-fold in the rIL-3-and rIL-4-treated cultures, 3.0 f 2.2-fold in the rIL-3-and rIL-10-treated cultures, 5.6 f 2.7-fold in the rIL-3-, rIL-4-, and rIL-10-treated cultures, and 1.5 f 1.2-fold in the rIL-4-and rIL-10-treated cultures. After the second week, the number of MC increased 2.7 f 1.8-, 10.7 f 2.3-, 8.0 f 3.6-, 8.0 f 3.3-, and 1.0 f 0.8fold in the corresponding cultures, and all populations of BMMC remained Alcian blue+/safranin-when stained. Blots containing total RNA from these five cultures and a mock supernatant-treated culture were probed with cDNAs that encode MMCP-2, MMCP-5, MMCP-6, MC carboxypeptidase A, actin, and FceRIcu (Fig. 3). All six cultures expressed abundant to modest levels of MMCP-5, MMCP-6, and MC carboxypeptidase A mRNAs. MMCP-2 mRNA was the most abundant protease transcript in BMMC cultured in medium containing both rIL-10 and rIL-4 (lane E ) . This transcript was less prominent in the cells cultured in medium that also contained rIL-3 (lanes C and D ) and was absent in cells cultured without rIL -10 (lanes A , B, and F ) .
Turnover of mRNA in BMMCc and the Reversibility of Transcription of the MMCP-2 Gene-To determine whether the high level of the MMCP-2 transcript in BMMCc (and by analogy in rIL-10-treated BMMCrr,.J relative to BMMCW and BMMCIL.3 was a consequence of increased transcription of the gene or of an abnormally high degree of stabilization of the transcript, the half-lives of MMCP-2 mRNA and other transcripts in BMMCc were assessed after treatment of the cells with actinomycin D. There was no decrease in MMCP-2 mRNA, MC carboxypeptidase A mRNA, SG-PG mRNA, or 18 S rRNA in BMMCc exposed to actinomycin D for 1 or 2 h (Fig. 4). Although there was no apparent change in the level  Total RNA was isolated from BMMCc at time 0, from BMMCc that were cultured for 24 h in the absence of actinomycin D, and from BMMCc that were cultured in the presence of actinomycin D for 1, 2, or 6 h. RNA blots were prepared and analyzed with a radiolabeled 18 S rRNA probe, as well as radiolabeled cDNAs that encode MMCP-2, MC carboxypeptidase A (MC-CPA), and SG-PG. of rRNA in the 6-h actinomycin D-treated BMMCc, there was a substantial reduction in the levels of mRNAs that encode all three constituents of the secretory granule ( n = 2). By quantitation of the radioactivity associated with each RNA blot, it was estimated that the half-lives of MMCP-2 mRNA, MC carboxypeptidase A mRNA, and SG-PG mRNA in these actinomycin D-treated BMMCc were all less than 6 h.
To determine if the expression of MMCP-2 mRNA in BMMC is reversible or if these culture conditions select for a population of MC that constitutively express MMCP-2, 2week BMMCIL.3 were cultured for an additional 4 weeks in medium containing rIL-3 alone (Fig. 5, lane 1 ), rIL-3 for 2 weeks followed by rIL-10 and rIL-3 for 2 weeks (lane 2), rIL-10 and rIL-3 for 4 weeks (lane 3), or rIL-3 and rIL-10 for 2 weeks followed by rIL-3 alone for 2 weeks (lane 4 ) . As depicted in Fig. 5 , the MMCP-2 transcript became prominent when BMMCIL.3 were cultured in the presence of both rIL-3 and rIL-10 for 2-4 weeks. In contrast, MMCP-2 mRNA was not detected in BMMC that were cultured with rIL-3 alone throughout or for 2 weeks after the rIL-lO/rIL-3 treatment. The induction of MMCP-2 mRNA was also observed in BMMCw that were then exposed for 2 weeks to concanavalin A-stimulated splenocyte-conditioned medium, and reversibility was also obtained in BMMCc that were then exposed for 2 weeks to WEHI-3 cell-conditioned medium (data not shown).

DISCUSSION
In this study, we demonstrate that different populations of in vitro-differentiated bone marrow-derived MC (designated BMMC), which all stain Alcian blue+/safranin-, are heterogeneous in terms of transcription of their secretory granule protease genes (Figs. 1-3). We delineate the transcription phenotype of rIL-3-elicited BMMC and demonstrate that rIL-10 can induce BMMCIL.3 to express MMCP-2 mRNA (Figs. 3 and 5 ) , a transcript that is found specifically in mucosal MC. These studies with recombinant cytokines not only demonstrate cytokine regulation at the transcription level, but also add specificity for particular T cell-derived cytokines.
The expression of MMCP-2 mRNA in BMMC can be reversed and is dependent on the continued exposure of the cells to rIL-10 ( Fig. 5) or concanavalin A-stimulated splenocyte-conditioned medium. The demonstration of reversible transcription of a gene that encodes a subclass-specific protease in BMMC indicates that the number and type of granule proteases eventually expressed by a mature tissue-localized mouse MC probably are not predetermined in the bone marrow, but rather is regulated by those cytokines in the tissue microenvironments in which the bone marrow-derived progenitor cells settle, proliferate, and mature.  noted that the MC in different tissue sites of reconstituted MC-deficient mice were histochemically identical with those in the same sites of normal mice. Even though the histochemical properties of MC reflect only the type of post-transcriptional modification of the SG-PG peptide core, these investigators also concluded that the tissue microenvironment was more important than the bone marrow microenvironment in subclass-specific differentiation of mouse MC.
Starting bone marrow cells contained SG-PG mRNA but no detectable amounts of the transcripts that encode MC carboxypeptidase A, MMCP-2, MMCP-5, or MMCP-6. Detectable levels of these protease transcripts were not obtained until day 10 of culture with either conditioned medium (Fig.  2) or with rIL-3 (data not shown). Thus, the SG-PG gene appears to be one of the earliest expressed genes in bone marrow cells, likely including those progenitor cells destined to become MC. A similar finding is obtained when mouse bone marrow cells are cultured in the presence of recombinant mouse c-kit ligand (47). The expression of SG-PG before any of the proteases probably occurs so that acidically charged proteoglycans can participate in the packaging of the basically charged proteases in the granule of mast cells (22, 48) and perhaps even in other hematopoietic cell types that also contain proteoglycan/protease macromolecular complexes (49).
Because concanavalin A-treated splenocytes express some IL-10 mRNA,* because IL-4 and IL-10 are two cytokines that are preferentially expressed by lymph node T lymphocytes during parasitic infection (29), and because the mastocytosis that occurs in parasite-infected intestines of mice is dependent on IL-4 (30), we examined the transcriptional regulation of protease genes in BMMCIL.~ exposed to various combinations of rIL-3, rIL-4, and rIL-10. In preliminary experiments, we were unable to obtain MC when starting mouse bone marrow cells were cultured in the presence of rIL-4, rIL-10, or a combination of rIL-4 and rIL-10. Thus, rIL-3 was used to obtain the BMMC~L.~ precursor population of MC for further study. In addition, because BMMCIL.~ did not remain viable in culture medium that contained only rIL-10, it was not possible to access directly the transcriptional regulation of the MMCP-2 gene by rIL-10 alone. Nevertheless, the fact that rIL-4 could maintain cell viability during rIL-10 action is important since rIL-4 alone or in the presence of rIL-3 (Fig. 3) failed to elicit transcription of the MMCP-2 gene. When BMMCrL.3 were exposed to a combination of rIL-4 and rIL-10, they did not increase in number even though they contained the highest level of MMCP-2 mRNA. This finding suggests that rIL-10 does not preferentially induce proliferation of a minor subpopulation of MC that is exclusively transcribing the MMCP-2 gene. However, because the halflife of the MMCP-2 transcript in actinomycin D-treated BMMCc is <6 h (Fig. 4), it is possible that an individual MC can quickly alter its protease phenotype when the cytokine microenvironment is altered.