Transcriptional Regulation of Mouse Mast Cell Protease-2 by Interleukin-15*

Mast cells (MCs) play a critical role in innate and adaptive immunity through the release of cytokines, chemokines, lipid mediators, biogenic amines, and proteases. We recently showed that the activities of MC proteases are transcriptionally regulated by intracellularly retained interleukin-15 (IL-15), and we provided evidence that this cytokine acts as a specific regulator of mouse mast cell protease-2 (mMCP-2). Here, we show that in wild-type bone marrow-derived mast cells (BMMCs) IL-15 inhibits mMCP-2 transcription indirectly by inducing differential expression and mMCP-2 promoter binding of the bifunctional transcription factors C/EBPβ and YY1. In wild-type BMMCs, C/EBPβ expression predominates over YY1 expression, and thus C/EBPβ preferentially binds to the mMCP-2 promoter. In IL-15-deficient BMMCs, the opposite is found: YY1 expression predominates and binds to the mMCP-2 promoter at the expense of C/EBPβ. Hypertranscription of the mMCP-2 gene in IL-15-deficient BMMCs is associated with histone acetylation and, intriguingly, with methylation of non-CpG dinucleotides within the MCP-2 promoter. This suggests a novel model of cytokine-controlled protease transcription: non-CpG methylation maintains a chromosomal domain in an “open” configuration that is permissive for gene expression.

Mast cells (MCs) 3 are key cellular mediators of host innate defense (1). Upon activation, MC secretory granules release a number of preformed inflammatory molecules that include cytokines, histamine, serglycin proteoglycans, and several MC-specific proteases (chymases, tryptases, and carboxypeptidase A) (2,3). Upon release into the extracellular space, these mediators regulate a broad variety of responses ranging from acute phase immediate reactions over recruitment of specific cell populations to sites of infection to long term tissue-remodeling reactions (1).
In the mouse, several mast cell proteases (mMCPs) have been identified and partially characterized, including the chymases mMCP-1, -2, -4, -5, and -9 and the tryptases mMCP-6 and -7 (4). Proteases play an important role in extracellular matrix remodeling, extravascular coagulation, and fibrinolysis as well as angiogenesis. Our recent investigation of the mechanisms that regulate mMCP gene expression has shown that intracellular cytokines can regulate mMCP activities (5). Murine MCs express constitutive and lipopolysaccharide-inducible IL-15 and store IL-15 intracellularly, where it colocalizes with tumor necrosis factor-␣ and proteases in MC granula. IL-15 Ϫ/Ϫ MCs exhibit markedly elevated chymase activity, leading to increased bactericidal activity and processing of neutrophil-recruiting chemokines. In particular, intracellular IL-15 operates as a specific negative transcriptional regulator of mMCP-2 (5).
In this report, we show that transcriptional repression of mMCP-2 by intracellular IL-15 is indirect. IL-15-mediated repression is effected by changing the expression levels of the transcription factors (TFs) C/EBP␤ and YY1 and thus the amount of each TF bound to the mMCP-2 promoter. C/EBP␤ binding is correlated with repression of mMCP-2 in wild-type (WT) BMMCs whereas YY1 binding is correlated with activation of mMCP-2 in IL-15 Ϫ/Ϫ BMMCs. In IL-15 Ϫ/Ϫ BMMCs, activation of mMCP-2 is associated with histone acetylation throughout the promoter and the body of the gene. Finally, we show that the genomic DNA of the active mMCP-2 locus is hypermethylated at non-CpG dinucleotides, and we suggest a novel role for non-CpG-methylation in gene activation.
Plasmid Construction-The full-length mMCP-2 promoter reporter construct (Ϫ1311 to ϩ28, p1311) as well as the 5Ј truncated constructs (Ј591 to ϩ28 (⌬p591), Ϫ397 to ϩ28 (⌬p397), Ϫ189 to ϩ28(⌬p189) were amplified by PCR from mouse genomic DNA using primers described in supplemental "Experimental Procedures". Inserts were cloned upstream from the ␤-lactamase reporter gene into the pGeneBlazer TOPO TA reporter vector (Invitrogen) according to the instructions of the manufacturer. Mouse IL-15 SSP (with short signal peptide)-expressing vector was generated by recombination of pEntry-SSP-IL-15 with expression destination vector pDest 53 (Invitrogen). The human IL-15-expressing vector was generated as described previously (6).
Real Time Quantitative PCR and Western Blotting-Total RNA from WT or IL-15 Ϫ/Ϫ BMMCs was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's recommendations, and RNA quality controls were determined using routine agarose gel control of the 28 S/18 S ratio. Five micrograms of total RNA were treated with RNase free DNase I (Invitrogen) and were reverse-transcribed to cDNA using oligo(dT) [12][13][14][15][16][17][18] primer and SuperScriptII reverse transcription (Invitrogen) in the presence of 10 units of RNase inhibitor (Invitrogen) according to the manufacturer's recommendations. cDNAs of BMMCs were routinely diluted 5-fold, and 2-l aliquots were used in 10-l light cycler PCRs containing 5 l of QuantiTect SYBR Green PCR kit (Qiagen) and 5 M genespecific primers described in supplemental "Experimental Procedures". The PCR was performed in a Light Cycler 3 (Roche Diagnostics) for 45-50 cycles of a four-step program: initial denaturation of 95°C for 12 min, 95°C for 10 s, 63°C-53°C touchdown (0.5°C per step) for 10 s, 72°C for 15 s, and fluorescence read step for 1 s. Quantitative PCRs were performed twice in triplicate; threshold cycles were obtained using cycler software version 3.0 (Roche Diagnostics). Mouse 18 S RNA was used for normalization of expression values. Relative quantification was calculated using the Relquant software (Roche Diagnostics), according to calibrator normalized quantification procedure with correction of differences in PCR efficiencies between target and reference. The difference in PCR amplification efficiencies was determined by amplification of target and reference genes (18 S RNA) in five dilution steps of P815 mastocytoma cell line cDNA in triplicate to generate a fit coefficient file according to the recommendation of Light Cycler 3.0 (Roche Diagnostics) instructions. For Western blots, nuclear extracts obtained from 10 7 WT or IL-15 Ϫ/Ϫ BMMCs were electrophoresed in Novex 4 -12% gradient SDS-bis-Tris gels (Invitrogen). The separated proteins were electroblotted onto nitrocellulose membranes (Schleicher and Schuell) in a transfer buffer consisting of 20 mM Tris-HCl, 150 mM glycine, and 20% methanol. The membranes were incubated sequentially in synthetic 2ϫ blocking solution (Roti-Immunoblock; Carl Roth) in 1ϫ phosphate-buffered saline for 1 h, in 1:250 dilution (0.75 g/ml) anti-C/EBP␤ IgG, clone sc-150 or anti-YY1 IgG, clone sc-281 (Santa Cruz Biotechnology), or 1:1000 of anti-␤-actin IgG clone 13E5 (Cell Signaling Technology), in 2ϫ blocking solution with 0.1% Tween 20 for 1 h and 3 ϫ 5 min in wash buffer (1ϫ phosphate-buffered saline with 0.1% Tween 20). For visualization of bound primary antibodies, membranes were incubated in 1:10,000 goat polyclonal anti-rabbit-IRDye IgG 800 nm (Biomol) for 1 h following three 5-min incubations of membranes in wash buffer. Proteins were detected using the Odyssey imaging system (Li-Cor Biosciences). Mouse polyclonal anti-␤-actin IgG purchased from Santa Cruz was routinely used to control equal sample loadings.
GeneBlazer Promoter Activity Assay-0.5 ϫ 10 6 COS-7 cells were incubated for 48 h with 2 g of mMCP-2 reporter plasmid and 3 l of FuGENE transfection reagent (Roche Diagnostics) at 37°C. ␤-Lactamase reporter expression in cell lysates was determined using the fluorescence resonance energy transfer (FRET)-based GeneBlazer in vitro detection system (Invitrogen) according to the manufacturer's instructions. Briefly, replicate samples of cells were washed once in cold 1ϫ Hanks' balanced salt solution (HBSS), lysed by three freeze-thaw cycles in 100 l of Hanks' balanced salt solution, and centrifuged at 13,000 rpm for 5 min at 4°C to separate cell debris. Fifty microliters of cell lysate was mixed with 50 l of 1:200 diluted Lytic Blazer-h-FRET B/G substrate in homogeneous lysis buffer (Invitrogen) and incubated in the dark for up to 3 h at room temperature. Fluorescence was then quantified on a TECAN infinite M200 fluorescence plate reader using 409/25 nm excitation, and emission was detected via 460/40 nm (blue) and 530/30 nm (green) bandpass filters. Emission intensities from blank wells were subtracted from wells containing cell lysates. Ratiometric fluorescence intensities of blue to green have been calculated as a value for mMCP-2 promoter activity. Lysates of untransfected COS-7 cells and Hanks' balanced salt solution alone served as negative controls, and transfection of COS-7 cells with the human ubiquitin promoter construct was routinely performed as positive control.
Chromatin Immunoprecipitation (ChIP) Assays-ChIP has been performed as described previously (7). Briefly, 6 ϫ 10 6 mast cells (WT or IL-15 Ϫ/Ϫ ) were washed twice with phosphate-buffered saline and cross-linked for 10 min with 1% formaldehyde in serum-free medium at 65°C. Cross-linking was stopped with 1.25 M glycine, and genomic DNAs were sonificated (7). For precipitation of acetylated chromatin, we used 2 l of polyclonal rabbit antibody specific for acetylated histone 3 (Upstate). The transcription factor assays were carried out with 5 g of anti-C/EBP␤ antibody or anti-YY1 antibody (both from Santa Cruz Biotechnology). As negative controls, we used rabbit normal serum (Dianova) or anti-muscle-specific kinase antibody (Abcam).
Bisulfite Sequencing-Genomic DNA was purified from WT and IL-15 Ϫ/Ϫ BMMCs with the Easy-DNA kit (Invitrogen). One microgram of genomic DNA was bisulfite-converted using the EpiTect bisulfite kit (Qiagen) according to the manufacturer's instructions. The DNA was ethanol-precipitated and resuspended in 20 l of TE buffer. For bisulfite sequencing, PCR was performed with 1 l of the converted genomic DNA solution from WT and IL-15 Ϫ/Ϫ BMMCs as template with primer pairs described in supplemental "Experimental Procedures". PCR products were cloned into the pGEM-T Easy vector (Promega), and a minimum of 15 clones from each sample was cycle sequenced with the BigDye Terminator kit (version 3.1; PE Biosystems) and an ABI automated DNA sequencer (PE Biosystems).
Statistical Analysis-The statistical significance threshold values for differential expression with the individual methods were calculated for genes showing changes Ͼ2-fold. Furthermore, to be regarded as differentially expressed between IL-15 Ϫ/Ϫ and WT BMMCs, a gene should differ in the same direction between a minimum of two different samples (taken at two time points) when analyzed by at least two different methods. Real time quantitative reverse transcription (RT)-PCR and Western blotting were performed in triplicate, and the values were imported into Excel worksheets for determination of means and standard errors. Student's t test for unpaired samples was calculated with p values of Ͻ0.05 accepted as significant.

IL-15 Regulates Repression of mMCP-2 Promoter Activity-
To investigate the role of IL-15 in transcriptional repression of the mMCP-2 promoter, we generated a series of constructs where either the full-length mMCP-2 promoter (p1311) or 5Ј truncations of the promoter (⌬p591, ⌬p397, and ⌬p189) were linked to the ␤-lactamase reporter gene (Fig. 1, A and B). Using a FRET-based GeneBlazer in vitro assay, we then measured the ␤-lactamase activity driven by these constructs in COS-7 cells in the presence or absence of (mouse/human) IL-15. As shown (Fig. 1, B and C) each of the truncations showed significantly higher ␤-lactamase activity compared with p1311, indicating that the p1311 promoter contains an additional element(s) that operates independently of IL-15, resulting in a lower basal mMCP-2 activity. Importantly, all constructs showed a significant decrease in ␤-lactamase activity when exogenous mouse or human IL-15 was coexpressed in the cells, showing that the minimal mMCP-2 fragment downstream of nucleotide Ϫ591 is sufficient for IL-15-regulated transcriptional repression of the mMCP-2 promoter.
IL-15 Regulates C/EBP␤ and YY1 Binding to the Minimal mMCP-2 Promoter-A bioinformatic search for potential TF binding sites within the mMCP-2 promoter (Ϫ591 to ϩ28) revealed a high number of putative C/EBP␤ and YY1 binding sites along with binding sites for Tst-1, Sp1, Oct-1, NF-b, MITF, MZF-1, Lmo2, and c-Ets-1 TFs (Fig. 2 and supplemental Figs. 1 and 2). We decided to focus on the bifunctional C/EBP␤ and YY1 TFs that are known to act either as suppressors or as activators of gene activity depending on the context (cell type, relative concentration, and interacting partners) (8 -12). To test whether C/EBP␤ and YY1 do indeed bind within the mMCP-2 promoter, we performed a ChIP assay with BMMCs derived from WT or IL-15 Ϫ/Ϫ mice with TF-specific antibodies. The primers used for amplification of the ChIPped material were taken from the limits that define the ⌬p189 promoter, which we had previously shown to be the minimal promoter exhibiting IL-15-dependent regulation (Fig. 1C). As shown in Fig. 3, we observed reciprocity in the binding of the C/EBP␤ and YY1 to the ⌬p189 truncated promoter that was dependent on the presence or absence of IL-15. In WT BMMCs, C/EBP␤ antibodies immune precipitated ⌬p189-containing chromatin,  NOVEMBER 20, 2009 • VOLUME 284 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 32637
whereas hardly any amplification was observed when the ChIP was performed with anti-YY1 antibodies (Fig. 3). In contrast, when IL-15 Ϫ/Ϫ BMMCs were used, anti-YY1 immunoprecipi-tated ⌬p189-containing chromatin, whereas anti-C/EBP␤ antibodies did not. These data indicate that the IL-15-dependent transcriptional repression of ⌬p189 (Fig. 1, A-C) correlates with the IL-15dependent binding of C/EBP␤. Conversely, activity of ⌬p189 correlates with IL-15-dependent binding of YY1. Using antibodies specific for intracellular IL-15, we could prove that IL-15 does not bind the mMCP-2 promoter directly (data not shown).

IL-15 Modulates C/EBP␤ and YY1 Expression in Mouse MCs-We
next investigated whether the IL-15dependent binding of C/EBP␤ and YY1 to ⌬p189 (Fig. 3) results from changes in the expression levels of C/EBP␤ and YY1 that are dependent upon IL-15 activity. Accordingly, we determined the expression of C/EBP␤ and YY1 at the transcript and the protein levels in WT and IL-15 Ϫ/Ϫ BMMCs. As shown in Fig.  4A (upper panel), quantitative realtime RT-PCR using PCR primers specific for C/EBP␤ showed significant overexpression in WT compared with IL-15 Ϫ/Ϫ BMMCs. We also observed the typical "doublet" of the C/EBP␤ protein in WT nuclear BMMCs extracts, whereas the C/EBP␤ doublet was barely detectable in IL-15 Ϫ/Ϫ BMMCs (Fig. 4A, lower panels). The converse was true for YY1. In IL-15 Ϫ/Ϫ BMMCs, quantitative realtime RT-PCR demonstrated significant overexpression of YY1 in IL-15 compared with WT BMMCS (Fig. 4B, upper panel). Consistent with the RT-PCR data, we found that YY1 protein expression in IL-15 Ϫ/Ϫ BMMC nuclear extracts was higher than that observed in WT BMMC (Fig. 4B, lower panels). These data suggest that IL-15 represses mMCP-2 expression indirectly through regulation of the C/EBP␤ and YY1 transcription factors which bind to the mMCP-2 promoter resulting in either its repression (C/EBP␤) or activation (YY1). (Figs.  2-4) and likely to operate via binding of C/EBP␤ to the mMCP-2 promoter, we decided to test the effect of exogenously expressing C/EBP␤ in COS-7 cells that were cotransfected with the p1311 ⌬p591, ⌬p397, and ⌬p189 which we had previously shown to exhibit IL-15-dependent regulation (see Fig. 1). As a control, the C/EBP␤-expressing plasmid was cotransfected into COS-7 cells with an irrelevant ubiquitin promoter construct. As shown in Fig. 5, exogenous expression of mouse C/EBP␤ results in specific sup-FIGURE 2. mMCP-2 promoter sequence (؊591 to translation start site) is shown with sites predicted to bind C/EBP␤ and YY1 based on bioinformatic analysis (supplemental Fig. 1). Transcription factor binding sites are shown from the transcription start indicated by an arrow to Ϫ502 bp. Potential and predominant in number are the C/EBP␤ and YY1 sites located throughout the region. Potential MITF, MZF1, Tst-1, c-ETS-1(p54), Lmo-complex, Sp1, and NF-b, which in many positions overlap with C/EBP␤ or YY1, are also shown. Fiftyseven methylated mainly non-CpG cytosine nucleotide sites (only one CpG) in IL-15 Ϫ/Ϫ BMMCs with a threshold frequency higher than 40 percentage points in sequenced clones, as indicated in Fig. 7, are shaded. The transcription start site is denoted by a black arrow, and the translation start is underlined. WT and IL-15 Ϫ/Ϫ BMMCs were analyzed by ChIP using monoclonal Abs to C/EBP␤, YY1, or with an equal amount of unspecific (antimuscle-specific kinase receptor, Musk) Abs, or isotype-matched control Abs (rabbit IgG), or without Abs, all used as negative control. Immunoprecipitated DNA was analyzed by PCR using mouse mMCP-2-specific primers, which amplify the Ϫ189 to ϩ28 bp (⌬p189) region. Each panel displays one experiment representative of three that were performed. Input, DNA purified from chromatin that has not been subjected to ChIP. M, 100-bp ladder markers. ated with gene activity (17). Methylation of cytosine residues can also occur in the context of CpA, CpT, and CpC dinucleotides as has been observed in cultured embryonic stem cells; however, the role of non-CpG methylation in epigenetic regulation of gene activity is less well explored (18,19). To investigate the methylation status of individual cytosines in the proximal region of the mMCP-2 promoter between nucleotides Ϫ591 and ϩ28 in WT and IL-15 Ϫ/Ϫ BMMCs, bisulfite genomic conversion and DNA sequencing were employed. Sequencing of at least 15 clones each from WT and IL-15 Ϫ/Ϫ BMMCs identified 57 hypermethylated non-CpG cytosines in IL-15 Ϫ/Ϫ BMMCs with an average degree of methylation being at least 90 percentage points higher than in WT cells (Fig. 7). Because non-CpG methylation correlates with the high mMCP-2  . IL-15 prevents non-CpG methylation of cytosines in downstream mMCP-2 promoter from ؊591 to the transcription start. Genomic DNA of WT and IL-15 Ϫ/Ϫ BMMCs were bisulfite-converted, amplified with primers covering the region Ϫ591 to ϩ28, cloned and sequenced (n Ͼ 15). The degree of methylation was calculated for every single cytosine of the sequence (x axis). The difference between the methylation rate of IL-15 Ϫ/Ϫ and WT BMMCs has been computed and plotted. Fifty-seven cytosines are hypermethylated in the IL-15 Ϫ/Ϫ BMMCs (difference Ͼ40 percentage points threshold). Different colors of the columns represent alternative cytosine-containing dinucleotides. As shown in Fig. 2B, many of the methylated cytosines in IL-15 Ϫ/Ϫ BMMCs occur at C/EBP␤ and YY1 binding sites. expression in IL-15 Ϫ/Ϫ BMMCs, this could represent an epigenetic mechanism that keeps the mMCP-2 chromosomal domain in an "open," active, configuration and/or may prevent binding of suppressor transcription factors.

DISCUSSION
The choice between IL-15-dependent repression or activation of the mMCP-2 gene in MCs is indirect and appears to be mediated by the binding of either the C/EBP␤ (for repression) or the YY1 (for activation) TFs to the mMCP-2 promoter (Figs.  2 and 3). Dependence on IL-15 is founded upon the observation that IL-15 regulates the expression of C/EBP␤ and YY1, where the levels of C/EBP␤ are elevated in WT cells and reduced in IL-15-deficient cells (Fig. 3), whereas the opposite is found for YY1 (Fig. 3). The binding sites for these two TFs lie in close proximity to each other in the mMCP-2 promoter. Within the ⌬p591 promoter fragment there are potentially 27 C/EBP␤ and 39 YY1 binding sites, several of which are overlapping (Fig. 2). We suggest that in WT cells the elevated levels of C/EBP␤ lead to its binding to the mMCP-2 promoter (Fig. 5) with little or no binding of YY1, resulting in repression of mMCP-2. In IL-15deficient MCs, the reverse occurs where YY1, whose levels are elevated, binds to the mMCP-2 promoter, with little or no binding of C/EBP␤, resulting in activation of mMCP-2 (Figs. 3 and 4). In addition to the reciprocal effects of IL-15 on the expression levels of C/EBP␤ and YY1, and their resultant binding to the mMCP-2 promoter, the close proximity of the C/EBP␤ and YY1 binding sites (some sites overlap) indicates that there may be a competition of the TFs for their recognition sequences within the mMCP-2 promoter, where binding of one TF to the mMCP-2 promoter occurs at the expense of the other.
To explore whether changes in chromatin structure are involved in the IL-15-dependent regulation of the mMCP-2 gene, we undertook ChIP analysis with antibodies to histone modifications, AcK9H3 and Me(3)K4H3 (Fig. 6). We could show that the promoter and the body of the gene are hypoacetylated in WT BMMCs compared with IL-15-deficient BMMCs, where the promoter and gene are hyperacetylated. These data are consistent with many previous studies which have shown that acetylation of histones is an epigenetic marker of gene activity that is involved in the cell-to-cell inheritance of patterns of gene activity (20).
Our DNA methylation analysis revealed that in IL-15-deficient BMMCs, where the mMCP-2 gene is hyperactive, we find that its promoter (Ϫ581 to ϩ2) is hypermethylated at both CpG and non-CpG dinucleotides (Fig. 7). This is, to our knowledge, the first time that non-CpG methylation in mammals has been correlated with gene activity; CpG methylation has previously been correlated with the allele-specific expression of the maternal Igf2r gene (17) although, in the vast majority of cases, CpG methylation is associated with gene inactivity. What could be the purpose of this cytosine methylation at the promoter of the active mMCP-2 gene in IL-15 Ϫ/Ϫ BMMCs? We suggest that the methylation at the active mMCP-2 gene acts to inhibit the binding of suppressor transcription factors, such as C/EBP␤, or a protein complex. It is known that the binding of both C/EBP␤ and YY1 is affected by DNA methylation (21,22). Based on our data, we suggest that the mechanism responsible for the dramatic 10 3 -fold increase in mMCP-2 expression in IL-15-deficient BMMCs occurs in two steps. First, there is the preferential (competitive) binding of YY1 to the mMCP-2 promoter at the expense of C/EBP␤ (Figs. 1-4). Second, YY1 recruits the chromatin-remodeling machinery to "open" the chromatin so that it is transcribed by the basal transcriptional machinery, and this open chromatin state is epigenetically inherited through cell division, via histone acetylation. In addition, stable epigenetic (cell-to-cell) inheritance of this open chromatin state might be reinforced by the cytosine methylation at CpG and non-CpG sites within the promoter (Figs. 6 and 7). Our future work will be focused on testing this model for regulation of mMCP-2 by IL-15. Conclusively, our study delineates a novel concept in cytokine-mediated control and fine tuning of intracellular protease activities in primary mast cells.