Peripheral blood monocytes are characterized by a unique phenotypic plasticity and are able to differentiate into a number of morphologically and functionally diverse cell types in vivo, including the wide range of heterogeneous tissue macrophages, myeloid DCs and multinucleated osteoclasts. The distinct differentiation pathways can be recapitulated in vitro: culturing purified human monocytes for several days in the presence of human serum results in the generation of macrophages (Figure 1a), whereas they develop into myeloid DCs in the presence of the granulocyte-macrophage colony-stimulating factor and IL-4 19. Both cell types are characterized by a unique transcriptome (examples of marker gene expression are shown in Figure 1b) and their development from primary monocytes proceeds without cell division 2021. To confirm the absence of proliferation-dependent or -independent DNA synthesis, we measured the incorporation of [³H]-thymidine during the first 4 days of monocyte to DC differentiation. As shown in Figure 1c, we did not detect significant nucleotide incorporation during the analyzed time period. Similar results were obtained using 5-Bromo-2'-deoxy-uridine (BrdU) incorporation and subsequent immunostaining (Figure S1 in Additional file 1). In line with several earlier studies 222325, proliferative activity ranged between 0 and 2% during the first 3 days of culture depending on the donor. Differentiating monocytes thus present an ideal post-mitotic cellular model to study epigenetic processes.

In order to assess occurrence and extent of active DNA demethylation during monocytic differentiation, we performed genome-wide methylation analyses using methyl-CpG immunoprecipitation (MCIp), a fractionation technique that is based on the salt concentration-dependent affinity of methylated and non-methylated DNA fragments towards an MBD-Fc fusion protein 2627. We refined and adapted the MCIp approach (schematically shown in Figure S2A in Additional file 1) for global promoter methylation analyses as recently described 28. DNA samples from in vitro-differentiated monocyte-derived macrophages and DCs were separated into methylated (mCpG) and unmethylated (CpG) pools via MCIp (Figure S2B in Additional file 1; two biological replicates). Cell type-specific differences in the DNA methylation pattern were then identified by co-hybridization of either both hypermethylated or both hypomethylated DNA subpopulations to custom-designed 244 K human promoter oligonucleotide arrays (Figure 2) covering 5-kb regions around 17,000 known promoters of protein-coding genes. DNA fragments enriched in the methylated fraction of a given cell type are depleted in the corresponding unmethylated fraction. Therefore, the signal intensities in CpG pool and mCpG pool hybridizations complement each other ('mirror-image' approach; Figure 2; Figure S2 in Additional file 1) and allow the identification of differentially methylated regions (DMRs). In total, the microarray analyses revealed 45 regions that were hypomethylated in DCs compared to macrophages. In line with previous findings, most DMRs were of low CpG content and all residual sites were of intermediate CpG content (data not shown). To validate and quantify methylation differences, 28 representative regions (including 21 DMRs, 6 control regions selected from array results and one additional region) were selected for matrix-assisted laser desorption/ionisation-time of flight (MALDI-TOF) mass spectrometry (MS) analysis of bisulfite-treated DNA (for information on amplicons and MALDI-TOF MS results for all samples see Additional files 2 and 3). In total, 22 out of 25 regions detected with both assays (88%) were concordant between MCIp-microarray and MALDI TOF MS data. Figure 3 and Figure S3 in Additional file 1 show several examples for the high consistency of both approaches. Classical bisulfite-sequencing experiments of three representative regions also confirmed the targeted and reproducible demethylation of defined CpG residues in DCs (Figure S4 in Additional file 1). An annotated, complete list of DMRs is given in Figure 4, which also provides the position of each DMR relative to the transcription start site (TSS) of neighboring genes, local CpG/GC content, as well as corresponding mRNA expression data. Interestingly, DMRs were always methylated in monocytes, indicating that all observed methylation differences resulted from demethylation. We did not observe a single case of differentiation-associated de novo DNA methylation. Thus, most (if not all) DMRs are actively demethylated during DC differentiation.

The positional annotation of DMRs in Figure 4 demonstrates that active demethylation processes were not limited to proximal promoter regions. Regardless of genomic localization, demethylation of DMR proceeded in a highly reproducible fashion during monocyte differentiation using cells from several different individuals, as exemplified by the promoter-proximal CCL13 DMR, the promoter-distal, intergenic CD207 DMR and the intragenic CLEC10A DMR in Figure S5 in Additional file 1. The high reproducibility between different donors suggests that active CpG demethylation is a strictly targeted, non-random event. Furthermore, active DNA demethylation processes did not proceed synchronously during monocyte to DC development, with some CpGs being demethylated early (between 18 and 42 h) and others considerably later (> 51 h) (Figure S5 in Additional file 1 and data not shown). Most DMRs contained CpGs that were demethylated during the first 51 h, a period during which we never observed significant proliferation of DCs. The reproducibility of CpG demethylation and the presence of DMR-specific demethylation kinetics suggest sequence-specific targeting mechanisms that are likely mediated through DNA-binding factors either directly or indirectly.

We also correlated the presence of DMR with mRNA expression data obtained by whole genome microarray analyses of monocyte differentiation time courses (three biological replicates). As shown in Figure 4, about half of the DMR-associated genes were up-regulated during monocyte to DC differentiation. As a prime example for active demethylation at a proximal promoter, the CCL13 DMR was studied in more detail as shown in Additional file 1 (detailed characterization of the CCL13 promoter and Figure S6). The data suggest that CpG methylation in this particular case may contribute to transcription repression by preventing the binding of a yet unknown nuclear factor and that active demethylation at this site may be necessary for high level transcription in DCs. However, a correlation between demethylation and increased transcription was not universally observed as transcription levels of many DMR-associated genes remained largely unchanged during differentiation, as measured by microarray analysis. Increases in gene expression at demethylated genes also did not correlate with the local CpG or GC content, which was not significantly different between both groups of genes (P > 0.1, Mann Whitney U-test).

Previous studies in other systems suggested a strong link between lineage-specific CpG demethylation events and changes in activating histone marks, including histone H3 lysine 4 (H3K4) methylation 282930. Since the above studies were done in proliferating cells, it was unclear whether the observed demethylation processes were active or passive. To determine whether similar correlations exist in a setting of post-proliferative monocytes that can only actively demethylate cytosine residues, we performed chromatin immunoprecipitation (ChIP) time course experiments studying the dynamics of histone modifications at selected DMRs, representative of the three possible genomic positions relative to the TSS (proximal promoter/intergenic/intragenic). As shown in Figure 5a and Figure S7 in Additional file 1, all seven actively demethylated regions tested exhibited increased H3K4 methylation or H3/4 acetylation during differentiation. As expected, H3K4 trimethylation was exclusively measured close to transcription start sites (CCL13, CLEC10A, DNASE1L3, P2RY6), whereas promoter-distal sites only acquired H3K4 mono- and dimethylation, which represents a signature indicative of putative enhancers 31.

We next asked whether promoter-distal DMRs display enhancer activity. Properties of generic enhancers include their ability to increase transcriptional activity in a heterologous context, which can be studied using traditional reporter gene assays. We recently developed a reporter vector that completely lacks CpG dinucleotides 32 and utilized this system to test for heterologous enhancer activity of seven selected DMRs (STAT5A, CD207, CBR3, ADPGK, RAP1GAP, ALKBH5, RPS3A) that are located in intergenic areas (between -4,700 to -1,100 bp from the nearest TSS). Transient transfections were performed in untreated, myeloid THP-1 cells using unmethylated (CpG) or in vitro SssI-methylated (mCpG) reporter plasmids. STAT5A, CBR3, ALKBH5 and RPS3A fragments did not show enhancer activity in THP-1 cells (data not shown), which may relate to the fact that this cell line lacks DC-specific transcriptional regulators. As shown in Figure 5b, the remaining regions (ADPGK, CD207, RAP1GAP) significantly enhanced the activity of the basal (CpG-free) EF1 promoter and completely lost enhancer activity when methylated, suggesting that their enhancing activity is critically dependent on their CpG methylation status.

Collection of blood cells from healthy donors was performed in compliance with the Helsinki Declaration. All donors signed an informed consent. Blood sampling, the leukapheresis procedure and subsequent purification of peripheral blood monocytes was approved by the local ethical committee (reference number 92-1782 and 09/066c).

Peripheral blood monocytes were separated by leukapheresis of healthy donors followed by density gradient centrifugation over Ficoll/Hypaque and subsequent counter current centrifugal elutriation in a J6M-E centrifuge (Beckman Coulter GmbH, Krefeld, Germany) as previously described 47. Monocytes were > 85% pure as determined by morphology and expression of CD14 antigen. Supernatants of monocyte cultures were routinely collected and analyzed for the presence of IL-6, which was usually low, indicating that monocytes were not activated before or during elutriation. To generate immature DCs, 1 × 10⁶ monocytes/ml were cultured in RPMI 1640 medium (Thermo Scientific, Bonn, Germany) supplemented with 10% FCS (Biowhittaker, Verviers, Belgium), 20 U/ml IL-4 (Promokine, Heidelberg, Germany) and 280 U/ml granulocyte-macrophage colony-stimulating factor (Berlex, Seattle, WA, USA). For generating macrophages, 1 × 10⁶ monocytes/ml were seeded in RPMI 1640 medium (HyClone) supplemented with 2% human pooled AB-group serum (Cambrex IEP GmbH, Wiesbaden, Germany) and cultured on teflon foils. THP-1 (human monocytic leukemia cell line) and U937 cells (human leukemic monocyte lymphoma cell line) were grown in RPMI 1640 plus 10% FCS (PAA, Pasching, Austria). RPMI 1640 was routinely enriched with 2 mM L-glutamine (Biochrome, Berlin, Germany), MEM non-essential amino acids (Invitrogen, Darmstadt, Germany), sodium pyruvate (Invitrogen), MEM vitamins (Invitrogen), 50 U/ml penicillin/streptomycin (Invitrogen), and 50 nM 2-mercaptoethanol (Invitrogen). The human cervical carcinoma cell line HeLa was maintained in Dulbecco's modified Eagle's medium plus 10% FCS.

Genomic DNA was prepared using the Blood and Cell Culture Midi Kit from Qiagen (Hilden, Germany). DNA concentration was determined with the ND-1000 NanoDrop spectrophotometer (Thermo Scientific, Bonn, Germany) and quality was assessed by agarose gel electrophoresis.

Total cellular RNA was isolated using the RNeasy Mini Kit (Qiagen). RNA concentration was measured with the ND-1000 NanoDrop Spectrophotometer (Thermo Scientific) and quality was controlled on agarose gels or using the Agilent Bioanalyzer (Böblingen, Germany).

Labeling, hybridization and scanning of high quality RNA was performed using the Agilent microarray platform according to the manufacturer's instructions. In brief, 200 to 1,000 ng of high-quality RNA were amplified and Cyanine 3-CTP-labelled with the One colour Low RNA Input Linear Amplification Kit. Labeling efficiency was controlled using the NanoDrop spectrophotometer and 1.65 μg labeled cRNA were fragmented and hybridized on Whole Human Genome Expressionarrays (4 × 44 K Agilent). Microarrays were washed and subsequently scanned with an Agilent scanner. Raw data were extracted with Feature Extraction 9.5.1 software and analyzed using GeneSpring GX 10.0.2 (Agilent). Data were normalized to the 75th percentile and baseline-transformed to the median of freshly isolated monocyte samples. Microarray data have been submitted and are available from the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) repository (accession number [GEO:GSE19236]).

Preparation of cross-linked chromatin was performed as described previously 48 with some modifications. Briefly, cells were treated with 1% formaldehyde solution for 7 minutes at room temperature and quenched by 0.125 M glycine. After washing with phosphate-buffered saline including 1 mM phenylmethylsulfonylfluoride, 2 × 10⁶ cells were resuspended in 50 μl lysis buffer 1A (L1A: 10 mM HEPES/KOH pH 7.9, 85 mM KCl, 1 mM EDTA pH 8.0) and lysed by adding 50 μl lysis buffer 1B (L1A + 1% Nonidet P-40) for 10 minutes on ice. Cross-linked chromatin was sheared to an average DNA fragment size around 400 to 600 bp using a Branson Sonifier 250 (Danbury, CT, USA). After centrifugation, 4 μl of the lysate were used as 5% input. After pre-clearing with 50 μl Sepharose CL-4B beads (blocked with 0.2% bovine serum albumin and 5 μg sheared salmon sperm for 1 h at 4°C) for 2 h, chromatin samples were immunoprecipitated overnight with 2.5 μg rabbit anti-RNA polymerase II CTD repeat YSPTSPS (phospho S5), anti-monomethyl Histone H3 (Lys4) (ab5131, ab8895, respectively; Abcam, Cambridge, UK), anti-dimethyl-Histone H3 (Lys4), anti-trimethyl-histone H3 (Lys4), anti-acetyl-Histone H3, anti-acetyl-Histone H4 or anti-IgG (07-030, 05-745, 06-599, 06-866 or 12-370, respectively; Millipore, Schwalbach/Ts., Germany). Before precipitation, protein A-Sepharose beads (GE Healthcare, Munich, Germany) were treated with 2 μg sheared salmon sperm DNA for 1 h at 4°C. Immunocomplexes were then recovered by incubation for 2 h with the blocked beads at 4°C. After reverse cross-linking, DNA was purified using the QIAquick PCR purification kit (Qiagen) according to the manufacturer's instructions except that the samples were incubated with phosphate buffer for 30 minutes and that they were eluted with 100 μl elution buffer. Enrichment of specific DNA fragments in the immunoprecipitated material was determined by quantitative PCR on the Realplex Mastercycler using the Quantifast SYBR Green PCR Kit (Qiagen). Oligonucleotide sequences are given in Additional file 4.

Production of the recombinant MBD-Fc protein and MCIp were carried out as previously described 2627 with modifications. Briefly, genomic DNA of DCs and macrophages was sonicated to a mean fragment size of 350 to 400 bp using a Branson Sonifier 250. Four micrograms of each sample were rotated with 200 μl protein A-Sepharose 4 Fast Flow beads (GE Healthcare) coated with 70 μg purified MBD-Fc protein in 2 ml Ultrafree-MC centrifugal devices (Millipore) for 3 h at 4°C in a buffer containing 250 mM NaCl (buffer A). Beads were centrifuged to recover unbound DNA fragments (250 mM fraction) and subsequently washed with buffers containing increasing NaCl concentrations (300, 350, 400, 450, 500 mM; buffers B to F). Densely CpG-methylated DNA was eluted with 1,000 mM NaCl (buffer G) and all fractions were desalted using the QIAquick PCR Purification Kit (Qiagen). The separation of CpG methylation densities of individual MCIp fractions was controlled by quantitative PCR using primers covering the imprinted SNRPN and a region lacking CpGs (Empty), respectively. Fractions containing unmethylated DNA (250 to 350 mM NaCl) or methylated DNA (400 to 1,000 mM NaCl) fractions were pooled before subsequent labeling.

Unmethylated (CpG) and methylated (mCpG) pools of both cell types were labeled with Alexa Fluor 5-dCTP (DCs) and Alexa Flour 3-dCTP (macrophages) using the BioPrime Total Genomic Labeling System (Invitrogen) as indicated by the manufacturer. Hybridization on 244K Custom-Oligonucleotide-Microarrays (containing about 17,000 promoter regions (-4,000 to + 1,000 bp relative to the TSS) as well as few regions tiled over large genomic intervals)) and washing was performed as recommended by the manufacturer (Agilent). Images were scanned immediately using a DNA microarray scanner (Agilent) and processed using Feature Extraction Software 9.5.1 (Agilent) with a standard comparative genomic hybridization protocol (including linear normalization). Processed signal intensities were then imported into Excel 2007 for further analysis. Probes with abnormal hybridization behavior (extremely high or extremely low signal intensities in one of the channels) were excluded. To detect DMRs, log₁₀ ratios of individual probes from both comparative genome pool hybridizations were subtracted. A more detailed description of the global methylation assay (MCIp and hybridization) is given in 28. Microarray data have been submitted and are available from the NCBI GEO repository (accession number [GEO:GSE19395]).

We chose a set of genomic regions based on the MCIp microarray results and designed 48 amplicons for bisulfite conversion. Genomic sequences were extracted from the UCSC genome browser 49 and PCR primers were designed using the Epidesigner web tool 50. For each reverse primer, an additional T7 promoter tag for in vitro transcription was added, as well as a 10-mer tag on the forward primer to adjust for melting temperature differences. All primers were purchased from Sigma-Aldrich (Munich, Germany; for sequences see Additional file 2). Sodium bisulfite conversion was performed using the EZ DNA Methylation Kit (Zymo Research, Orange, CA, USA) with 1 μg of genomic DNA and an alternative conversion protocol (Sequenom, San Diego, CA, USA). Amplification of target regions was followed by treatment with shrimp alkaline phosphatase, reverse transcription and subsequent RNA base-specific cleavage (MassCLEAVE, Sequenom) as previously described 38. Cleavage products were loaded onto silicon chips (spectroCHIP, Sequenom) and analyzed by MALDI-TOF MS (MassARRAY Compact MALDI-TOF, Sequenom). Methylation was quantified from mass spectra using the Epityper software v1.0 (Sequenom). Methylation ratios for all samples are given in Additional file 3

Proliferation capacity of cells was measured using [³H]-thymidine incorporation. Cells were seeded in 96-well microtiter plates and pulsed with 0.5 μCi [methyl-³H]-thymidine per well (Hartmann Analytics, Braunschweig, Germany) for 20 h. Cells were harvested onto UniFilter plates using a Wallac harvester and incorporated [³H]-thymidine was determined with a Wallac Betaplate counter (all from PerkinElmer, Gaithersburg, MD, USA).

Differentially methylated regions (ranging from 800 to 1,000 bp) were PCR-amplified from human genomic DNA and cloned into the CpG-free pCpGL-CMV/EF1 vector 32 by ligation replacing the cytomegalovirus (CMV) enhancer with the DMRs. Primer sequences are given in Additional file 4. Inserts were verified by sequencing. Luciferase reporter constructs were either mock-treated or methylated in vitro with SssI methylase for 4 h at 37°C and purified with the Plasmid Quick Pure Kit (Macherey-Nagel, Dueren, Germany) or using the Endofree Plasmid Kit (Qiagen). THP-1 and HeLa cells were transfected as described 51. The transfected cells were cultivated for 48 h and harvested. Cell lysates were assayed for firefly and Renilla luciferase activity using the Dual Luciferase Reporter Assay System (Promega, Mannheim, Germany) on a Lumat LB9501 (Berthold Detection Systems GmbH, Pforzheim, Germany). Firefly luciferase activity of individual transfections was normalized against Renilla luciferase activity.