Analysis of Il36a induction by C/EBPβ via a half-CRE•C/EBP element in murine macrophages in dependen

Interleukin-36α is a novel member of the IL-1 cytokine family that is highly expressed in epithelial tissues and several myeloid-derived cell types after induction. The transcription factor (TF) C/EBPβ binds specifically to an essential half-CRE•C/EBP motif in the Il36a promoter to induce Il36a expression upon LPS stimulation. C/EBPs regulate gene expression by binding to recognition sequences that can contain 5′-cytosine-phosphate-guanine-3′ dinucleotides (CpG), whose methylation can influence TF binding and gene expression. Herein we show that the half-CRE•C/EBP element in the Il36a promoter is differentially methylated in the murine RAW264.7 macrophage cell line and in primary murine macrophages. We demonstrate that C/EBPβ binding to the half-CRE•C/EBP element in the Il36a promoter following LPS stimulation is insensitive to CpG methylation and that methylation of the CpG in the half-CRE•C/EBP element does not alter LPS-induced Il36a promoter activity which correlated with similar Il36amRNA copy numbers and pro-IL-36α protein amount in both cell types. Taken together, our data indicate that C/EBPβ binding to the half-CRE•C/EBP element and subsequent gene activation occurs independently of the CpG methylation status of the half-CRE•C/EBP motif and underlines the potential of C/EBPs to recognize methylated as well as unmethylated motifs.

The interleukin (IL)-36 cytokines constitute a subfamily of the IL-1 cytokine family and consist of three agonistic cytokines (IL-36α, IL-36β IL-36γ), and the IL-36 receptor antagonist (IL-36RA) that play important roles in host immunity [1]. IL-36 cytokines are expressed by a variety of cells in different tissues, such as macrophages, dendritic cells, keratinocytes, and lung epithelial cells [2,3,4,5,6,7]. They bind to the IL-36 receptor (also referred to as IL-1R-like 2) that is widely expressed on many different cells, including murine bone marrow-derived dendritic cells, CD4+ T cells, mononuclear phagocytes, and various epithelial cells of skin, lung, and digestive tract tissues [4, 8, 9]. Similar to the receptor for IL-1, the IL-36 receptor recruits the transmembrane ILl-1 receptor accessory protein upon ligand binding that subsequently activates intracellular signaling via mitogen-activated protein kinases and nuclear factor-κB (NF-κB) activated transcription leading to inflammatory responses [10,11,12]. At the transcriptional level IL-36α and IL-36γ themselves are regulated by NF-κB, C/EBPβ, and by the transcription factor T-box expressed in T cells, respectively [2, 3].

DNA methylation at cytosine residues by DNA methyltransferases is an epigenetic modification involved in mammalian development, lineage identity as well as transcriptional regulation [13]. Traditionally, DNA methylation of gene promoters was considered to silence transcription either directly by inhibiting binding of certain transcription factors (TF) to their binding sites [14,15,16,17] or indirectly by binding of TFs containing a methyl-CpG (mCpG)-binding domain (MBD), that in turn recruits histone deacetylases which subsequently promote local chromatin condensation [18]. In recent years, a growing number of TFs lacking MBDs have been shown to bind to their transcription factor binding sites (TFBS) even when they are methylated, arguing against the paradigm that CpG methylation always represses transcription. Examples include transcriptional activators that contain a C2H4 zinc-finger domain like Kaiso and Krueppel-like factor 4 [19, 20], proteins with a helix-turn-helix DNA-binding domain like Zhx1/2 and Pax6 [21, 22], helix-loop-helix domain-containing TFs like c-MYC and ARNT2 [23, 24], and TFs belonging to the family of bZIP proteins like C/EBPs [25, 26].

C/EBP transcription factors belong to the bZIP family of transcription factors and are involved in tissue-specific gene expression, proliferation, differentiation, and inflammation [27]. In particular, C/EBPβ has been shown to regulate inflammatory genes like the cytokines IL-6, IL-12 p40, and IL-36α [3, 28, 29], the chemokines IL-8 and macrophage inflammatory protein-1α [28, 30] as well as the proinflammatory genes for inducible NO synthase (NOS2) and cyclooxygenase-2 [31, 32] in macrophages. DNA binding and subsequent gene expression by C/EBPβ requires the formation of homodimers or heterodimers with other C/EBP family members or members of the CREB/ATF family via the bZIP domain [33]. The consensus sequence as well as composite sites recognized by C/EBPs can contain a central CpG dinucleotide [27]. Several studies showed that C/EBPs are able to bind these sequences in vitro and in vivo with similar or even with increased affinities if the central CpG dinucleotide is methylated [17, 25, 26, 34]. As a consequence, such methylation can generate C/EBP binding sites at CRE-like sequences leading to activation of a subset of differentiation-specific genes by C/EBPs whilst inhibiting activation of these genes by CREB members [25]. The structural basis for this insensitivity to the methylation status of the bound DNA has been recently unravelled and was shown to be mediated via a so-called methyl-Arg-G triad [35].

Recently, we reported that C/EBPβ confers transcriptional activation of Il36a via an essential half-CRE•C/EBP site that contains a central CpG dinucleotide [3]. For IL-36 cytokines, there is to our knowledge, no information on epigenetic regulation, in particular on CpG methylation and its possible impact on transcription factor binding, available. Therefore, we herein analyzed the methylation level of this element in two different murine macrophage types and its impact on C/EBPβ binding and transcriptional activation of Il36a.

Differential DNA methylation of a half-CRE•C/EBP element in the Il36apromoter in murine macrophages

We have previously shown that LPS-induced Il36a mRNA expression in murine macrophages is essentially mediated by binding of C/EBPβ to a half-CRE•C/EBP element within the Il36apromoter even though classical C/EBP recognition sites are present in the promoter sequence [3]. In this study, we aimed to dissect the specificity of C/EBPβ binding to the half-CRE•C/EBP element in relation to the level of methylation.

Detailed in silico analysis of the Il36a promoter region comprising 1120 bp upstream of the transcriptional start site revealed a GC content of 43.21%, the absence of any CpG islands but the presence of nine CpG dinucleotides (Fig. 1A). Isolation of genomic DNA and subsequent bisulfite sequencing of the region containing the CpG in the half-CRE•C/EBP element showed low methylation of 9.70 ± 10.01% in RAW264.7 cells (Fig. 1B and D). In contrast, the methylation level of this site in BMDMs was 66.95 ± 8.63% (Fig. 1C and D), revealing a significant mean difference in methylation between both cell types (mean difference = 57.2%; 95% CI: 44.5, 67.5).

A Schematic representation of the Il36a gene structure (Genebank entry KM205447.1), CpG islands, GC content (upper part), and the promoter region indicating CpG sites (lollipops, lower part). The position of Il36a on mouse chromosome 2 (red line), the half-CRE•C/EBP element (orange box) and the amplified region for bisulfite sequencing analysis on the (−) strand are indicated. B, C The DNA methylation status for the region indicated in (A) was determined from RAW264.7 (B) or BMDMs (C) by bisulfite sequencing analysis. Each line represents sequencing results of an individual clone (open circle, unmethylated CpG; filled circle, methylated CpG). The CpG in half-CRE•C/EBP element is indicated by grey arrows. D Quantification of DNA methylation in RAW264.7 (light gray dots) and BMDMs (dark gray dots) from n = 3 (RAW264.7) or n = 4 (BMDM) experiments. Individual data points and summary measurements (mean ± SD) are plotted on the left-hand side of the panel; effect size (mean differences, black dot) with bootstrapped 95% confidence intervals and resampling distribution are shown on the right-hand side of the panel.

Full size image

In vitro binding of C/EBPβ to the methylated half-CRE•C/EBP element in the Il36a promoter

Since it was demonstrated that CpG methylation of half-CRE elements enhances binding of C/EBP members and can activate tissue-specific gene expression, the significantly differential methylation status of the CpG in the half-CRE•C/EBP might lead to differential binding activities of C/EBPβ in RAW264.7 macrophages and primary BMDMs. Therefore, we next analyzed the binding of C/EBP and CREB members to the unmethylated and methylated half-CRE•C/EBP element in nuclear extracts by EMSA using a radiolabeled oligonucleotide spanning the region from −314 to −290 relative to the TSS of the murine Il36a gene.

We first studied the in vitro complex formation using nuclear extracts of RAW264.7 cells stimulated with LPS for 4 h. In EMSA with the unmethylated probe, DNA-protein interactions appeared as two major complexes in unstimulated and LPS-stimulated cells. Whereas constitutive DNA binding activity of the slower migrating complexes was nearly unchanged (arrowheads, Fig. 2A), the binding activity of the faster-migrating complexes increased in extracts from LPS-treated cells. Supershift experiments with antibodies against C/EBPβ, C/EBPδ, CREB-1, and ATF-4 indicated that only C/EBPβ and to a certain extent C/EBPδ are present in the inducible complex. Using the methylated probe, a similar inducible protein-DNA complex formation was observed. In comparison to the unmethylated probe, constitutive DNA binding activity was not detectable. The methylated probe also bound C/EBPβ and C/EBPδ but not CREB-1 nor ATF4 (Fig. 2A). Similar results were obtained with nuclear extracts from unstimulated and LPS-stimulated BMDMs. Irrespective of the use of unmethylated or methylated probe, LPS treatment induced a protein-DNA complex that supershifted with antibodies against C/EBPβ and to a minor extent by anti-C/EBPδ antibodies. Again, the formed complexes did not contain a detectable amount of CREB-1 nor ATF4 (Fig. 2B). However, in the extracts of both cell types, the complexes formed with the methylated probe seemed to form more specific complexes with less background as compared to the unmethylated probe which is best emphasized by the reduced constitutive protein binding in the extracts from RAW264.7 macrophages using the methylated probe.

This suggests that C/EBPβ is able to bind unmethylated as well as methylated half-CRE•C/EBP sequences and that DNA methylation of the half-CRE•C/EBP element enhances specificity of protein binding.

DNA binding of C/EBPβ is not enhanced by methylation

To test whether CpG methylation enhances binding of C/EBPs to DNA probes in vitro and to further analyze if there is any preferential binding to the unmethylated or methylated half-CRE•C/EBP oligonucleotides the protein-DNA complex formation was assessed with different approaches. First, we performed competition EMSA. For this, a radiolabeled oligonucleotide containing the C/EBP consensus motif (5′-TTGCGCAA-3′) was incubated with nuclear extracts of unstimulated and LPS-stimulated RAW264.7 cells in the presence of increasing amounts of unlabeled unmethylated or unlabeled methylated half-CRE•C/EBP-Il36a oligonucleotides used before. As shown in Fig. 3A both oligonucleotides concentration-dependently decreased protein-DNA complex formation with comparable inhibitory effects.

Next, we determined dissociation constants (KD) of the murine C/EBPβ-bZIP domain (Fig. 3B) to double-stranded half-CRE•C/EBP-Il36a oligonucleotides with unmethylated or methylated CpG using fluorescence polarization assays (Fig. 3C). The C/EBPβ DNA binding domain bound the unmodified oligonucleotides with a KD of 0.08 ± 0.03 µM (Fig. 3D). Under the same conditions, the methylated oligonucleotides were bound with a KD of 0.10 ± 0.03 µM (Fig. 3D), revealing a nonsignificant mean difference between unmethylated and methylated oligonucleotides (mean difference = 0.0157 µM; 95% CI: −0.0213, 0.0438).

Together, these data further indicate, that C/EBPβ can recognize the half-CRE•C/EBP site located in the Il36a promoter irrespectively of the methylation status of central CpG in vitro.

Methylation of the CpG in the half-CRE•C/EBP element does not inhibit Il36a promoter activity

Next, we assayed the relevance of CpG methylation of the half-CRE•C/EBP element in a system more resembling the in vivo situation in the nucleus. To evaluate whether the binding of C/EBPβ to the methylated half-CRE•C/EBP element is biologically relevant we performed transient transfections using RAW264.7 cells. The Il36a promoter region, comprising the essential regulatory elements for gene induction between −357 and −45 relative to the TSS, was cloned into a reporter plasmid where all CpG dinucleotides have been deleted from the plasmid backbone (pCpGL) [36], (Fig. 4A). The CpG in the half-CRE•C/EBP site in the pCpGL-Il36a–357/−45 plasmid was enzymatically methylated in vitro using SssI (CpG) methylase and S-adenosylmethionine. We confirmed methylation efficacy by restriction analysis. HpyCH4IV linearized the unmethylated plasmid completely (asterisk, Fig. 4B) whereas the methylated plasmid was not linearized, indicating complete methylation (Fig. 4B).

Next, we transfected the methylated and unmethylated plasmids into RAW264.7 cells and determined luciferase activity in untreated cells or cells stimulated with 100 ng/ml LPS after 8 h. As shown in Fig. 4C, stimulation with LPS led to an increased luciferase induction (fold change LPS vs. ctrl) in cells transfected with unmethylated and methylated reporter plasmid, respectively. We did not observe a significant difference in luciferase induction between cells transfected with the unmethylated or methylated plasmid (mean difference = 6.94 RLU; 95% CI: −7.47, 19.7).

These data show that CpG methylation of the half-CRE•C/EBP site does not significantly influence Il36a promoter activity in LPS-stimulated RAW264.7 cells.

Methylation of the CpG in the half-CRE•C/EBP element does not influence IL-36α expression

Next, we analyzed whether the marginal expression differences in cells transfected with the methylated reporter plasmid were reflected by Il36a promoter activation in its ‘in vivo’ context in the nucleus. For this we directly compared expression of Il36a induced by LPS in RAW264.7 cells and BMDMs, which have significantly different methylation level of the half-CRE•C/EBP site (Fig. 1). Cells were stimulated with LPS for 8 h and isolated total RNA was subjected to absolute quantification by qRT-PCR. As shown in Fig. 5A, Il36a induction was detectable after stimulation of RAW264.7 cells as well as BMDMs with LPS. We did not observe a significant difference in the fold change induction of Il36a mRNA between LPS-stimulated RAW264.7 cells and BMDMs (mean difference = −101.4; 95% CI: −2276.3, 1748.2). Furthermore, Western blot analysis of pro-IL-36α in cell lysates of RAW264.7 cells and BMDMs also revealed similar levels of the cytokine after stimulation with LPS (Fig. 5B) supporting the hypothesis that DNA methylation of the half-CRE•C/EBP site does not influence Il36a expression.

Il36a is expressed in a variety of cell types and in different tissues at different levels [11]. Yet information on epigenetic regulation, in particular on CpG methylation and its possible impact on transcription factor binding are missing. We have previously shown that in murine macrophages LPS-induced Il36a mRNA expression requires binding of C/EBPβ to a half-CRE•C/EBP element within the Il36a promoter [3]. Here we observed differential methylation of this element in RAW264.7 cells and primary murine macrophages, leading to the question of whether this difference in methylation points to epigenetic regulation of Il36a by affecting TF binding and subsequent Il36a gene expression.

Regulation of gene expression by methylation of CpG motifs is one of several universal epigenetic mechanisms [37]. The majority of data linking CpG methylation of the promoter regions and gene expression is derived from studies of genes with CpG islands (a region longer than 500 bp with an observed CpG/expected CpG ratio of 0.65) in the promoter region [38]. CpG islands are not present in the Il36a promoter. However, the promoter harbors a number of CpGs proximal to the transcriptional start site, including one located within the essential half-CRE•C/EBP element. Such proximal, non-island CpGs are nowadays considered important for regulation of gene expression and examples of promoters affected by methylation of these non-island CpGs include e.g., Il2, NOS2, MMP13, Il1B, and Il18BP[39,40,41,42,43]. While in most cases, methylation of the CpG sites results in transcriptional inactivation, for other genes methylation increased binding of TFs and thus resulted in stronger gene expression. Examples include Bglap-rs1 and MMP9 [25] as well as PAX2 [44]. This regulation at the epigenetic level may serve and control cell/tissue type-specific physiological functions or lead to the reactivation of early developmental genes in malignancy.

Several lines of evidence presented herein suggest that methylation of the half-CRE•C/EBP element in the Il36a promoter has no influence on Il36a mRNA expression. Gel shift experiments and fluorescence polarization assays demonstrated that C/EBPβ is able to bind the unmethylated as well as methylated form of the CRE•C/EBP element in vitro, which is in agreement with earlier studies demonstrating binding of C/EBPβ to methylated consensus sequences [25]. Interestingly, we also observed the binding of C/EBPδ to both probes although overall binding was lower compared to C/EBPβ. This is in contrast to studies using methylation binding arrays showing that methylation inhibits binding of C/EBPδ [26]. Furthermore, we did not detect the formation of C/EBPβ/ATF4 heterodimers indicating that the heterodimers are not formed after LPS stimulation because ATF4 is presumable not induced in sufficient amounts. However, for lysates from both cell types we detected less unspecific binding when the methylated probe was used, indicating that methylation interferes with the binding of some unidentified TFs recognizing the unmethylated probe. Therefore, one could speculate that methylation modulates specificity of Il36a mRNA expression under certain circumstances. Indeed, Yin et al. demonstrated reduced binding for the majority of bZIP family members to their motifs upon methylation (Fig. 5C in [17]). This supports the hypothesis that methylation of the half-CRE•C/EBP element in BMDMs may increase the specificity for C/EBPβ, one of the TFs in this family that tolerates methylation within the central CpG.

In contrast to the study by Rishi et al., we did not observe enhanced C/EBPβ binding to the methylated motif in vitro [25]. Since Il36a activation is C/EBPβ-dependent our data suggest that tolerance of a transcriptional regulator for CpG methylation might help to overcome otherwise inhibitory epigenetic modifications. In line with this, our findings are consistent with previous reports showing that in vivo binding by C/EBPβ tolerates CpG methylation [34]. Furthermore, it is also in accordance with recently published data that compared relative affinities for C/EBPβ in vitro binding using methylated and unmethylated libraries [45]. Accordingly, we analyzed the specific sequence context of the Il36a half-CRE•C/EBP element by comparing methylated and unmethylated relative affinities with those of a consensus C/EBP motif and a consensus CRE motif, respectively, based on published data for binding of C/EBPβ homodimers [45]. Such a comparison revealed a relatively low binding affinity (compared to a consensus C/EBP motif) for the half-CRE•C/EBP motif that, however, did not differ between methylated and unmethylated sequences (Supplementary Fig. 1A). In contrast, in vitro binding of C/EBPβ to a consensus CRE motif was enhanced when methylated, confirming previous data [25]. This methylation insensitivity is also reflected in the energy logos that are derived from the oligomer enrichment tables (Supplementary Fig. 1B). In agreement, induction of luciferase expression upon LPS stimulation, the Il36a mRNA expression, and the amount of pro-Il-36α at the protein levels did not differ significantly in the two cell types despite the observed difference in methylation. However, since only 66% of the cells in the BMDM population had a methylated half-CRE•C/EBP element, we cannot exclude that incomplete methylation influenced the effect size of the qRT-PCR and Western Blot assays. Taken together, our data suggest that methylation of this motif does not influence endogenous promoter activity. Therefore, we speculate that asymmetry of the half-CRE•C/EBP element and/or the nucleotides flanking the central CpG in the different binding motifs might determine whether there is no effect or enhanced binding if the CpG is methylated.

Altogether, the data presented herein emphasize the potential of C/EBPβ to recognize methylated as well as unmethylated binding sites. Structural analysis or in silico modelling need to be performed in the future to unequivocally elucidate the binding mechanism of C/EBP to the half-CRE•C/EBP element present in the Il36a promoter. Furthermore, our data indicate that Il36a is most likely not regulated epigenetically by CpG methylation of the half-CRE•C/EBP element in the proximal promoter region of the gene in murine macrophages.