DNA methylation-histone modification relationships across the desmin locus in human primary cells

ChIP-on-chip: global H3 and H4 histone acetylation patterns across 500 kb of chromosome 2q35 encompassing the DES locus

The DES locus contains three muscle specific/preferentially expressed genes, DES APEG1 and SPEG, which are flanked by two housekeeping genes DNPEP and GMPPA. This region also includes the DES LCR which lies 9–18 kb 5' of DES and which is composed of 5 muscle-specific DNaseI HS numbered 1–5 in a 3' to 5' direction [6]. Analysis of the DNA sequence that underlies the elements of the DES LCR shows sub-regions of high (83–100%) sequence conservation between rodents (mouse/rat) and humans within or near each HS site designated with the letters a, b, c or d [6]. Expression of DES increases upon fusion of unfused (UF) skeletal myoblasts to fused (F) myotubes [22] (see Additional file 1, Figures S1 and S2).

The ENCODE region ENr331 which spans 500 kb of chromosome 2q35 does not include the DES LCR as it terminates -5.7 kb 5' of DES. However, ENr331 does span DES and the more downstream genes SPEG, GMPPA, ACCN4, CHPF, INHA, SLC4A3 (Figure 1). A small spotted microarray consisting of ~500 bp overlapping PCR products was designed to extend the area covered by ENr331 to encompass the DES LCR (-18 kb to -8.3 kb 5' of the DES TSS). Note: the PCR primers used to extend the region of the array are shown in Additional file 1, Table ST1.

Native chromatin without cross-linking, with a size range of mono-tetranucleosomes was isolated from cultures of human primary myoblasts, myotubes and PBMCs and subjected to immunoprecipitation with antibodies against histones H3 and H4 acetylated at multiple lysine residues, namely lysines 9 and 14 on H3 and lysines 5, 8, 12 and 16 on H4, which have been associated with a transcriptionally competent chromatin state [23, 24]. This allowed a complete analysis of the histone H3 and H4 acetylation pattern of this entire area in unfused (UF) myoblasts, fused (F) myotubes and PBMCs (Figure 1).

The peaks of enrichment of H3 lysine acetylation throughout the region analysed corresponded to the TSS of most of the genes including DES (Figure 1A). The DES LCR was enriched for H3 acetylation approximately 2–5 fold above non-muscle PBMCs in both myoblasts and myotubes (Figure 1A and Additional file 1, Figure S3).

In contrast to H3 acetylation, which exhibits a relatively tight peak of enrichment localized to the DES promoter, H4 acetylation displays abundance across the entire transcribed region of DES (Figure 1B; panels UF and F) and extends into intergenic areas. This was also the case with other individual genes within the region such as CHPF, which exhibited a broad enrichment across its entire gene body (Figure 1B and Additional file 1, Figure S3). A ~5-fold enrichment of H4 acetylation was observed at the DES LCR and promoter compared with PBMCs (Figure 1B and Additional file 1, Figure S3). Similarly to H3 acetylation, there was very little enrichment of histone H4 acetylation in PBMCs across the area analysed (Figure 1; panels PBMC). See Additional data file 1, Figure S3 and S3' for detailed analysis of individual genes with the region.

The current genome browser build hg18 only illustrates SPEG. However it must be noted that APEG1 is one of its isomers and is transcribed from an internal separate promoter [6]. The presence of the internal (APEG1) promoter accounts for the presence of a peak of acetylation seen on the microarray at this position (Figure 1 and Additional file 1, Figure S3).

DES and the DES LCR show high levels of tissue-specific but a selective histone modification pattern

Having obtained a general overview of histone H3 and H4 acetylation patterns across the DES locus and surrounding 500 kb region, we next sought to obtain a more detailed map of epigenetic status specifically across the DES LCR, as this constitutes a crucial regulatory element. Chromatin used for the microarray analysis was analysed by Q-PCR to assess enrichment for selected regions within DES and the DES LCR.

Histone H3 and H4 acetylation showed a marked enrichment of between 5 to 15-fold above PBMC levels in both myoblasts and myotubes at all of the positions tested within the DES LCR (Figure 2A and 2B; amplicons A-E). Element HS3c exhibited the highest overall level of histone acetylation within the LCR in both myoblasts and myotubes reaching levels of ~15-fold enrichment (Figure 2A and 2B; amplicon C). We also observed enrichment of H3 acetylation across the DES TSS (Figure 2A; amplicons G-J) but with the highest level present 3' of the first exon at position +0.8 kb (Figure 2A; amplicon I). This level of enrichment decreased by almost 2-fold at the second exon (Figure 2A; amplicon J). Histone H4 acetylation was enriched at similar levels (~10–15 fold) across parts of the LCR and DES (Figure 2B; amplicons C-E and G-J).

PBMCs did not show any enrichment of histone H3 and H4 acetylation across either DES or its LCR reflecting these elements muscle-specific nature and function (Figure 2A and 2B, amplicons A-J). The only enrichment of both histone H3 and H4 acetylation in the PBMC sample was seen at the TSS of DNPEP and HTATIP2, which is as expected for ubiquitously expressed genes (Figure 2A and 2B; amplicons DNPEP and HTATIP2). We also investigated the chromatin structure between HS1 of the LCR and the DES TSS, a region at a position approximately -4 kb 5' of DES and found no enrichment of histone H3 or H4 acetylation in any of the samples (Figure 2A and 2B; amplicon F).

The microarray data correlates well with the higher resolution native ChIP (nChIP) results both showing similar trends in H3 and H4 acetylation. However, the Q-PCR and microarray sets of data are technically very different and therefore quantitative comparisons between the two are not possible.

Histone H3K4me2 marks the DES LCR

Histone methylation patterns have been shown to mark different regions of a transcribing gene and to distinguish between promoters and distal enhancers [3, 25]. We therefore assessed the presence of histone H3K4me2 and H3K4me3 at the same regions analysed for global H3 and H4 histone acetylation (Figure 2A and 2B). This also allowed the correlation between histone acetylation and methylation to be examined. H3K4me2 displayed a widespread distribution across the region being enriched up to 20-fold above background levels at elements HS4d to HS1 within the DES LCR (Figure 2C; amplicons B-E). The unfused myoblasts exhibited a 20-fold enrichment across the first 2 kb of DES (Figure 2C; amplicons G-J) whereas myotube cultures showed the same level of enrichment but had a high peak of ~40-fold of H3K4me2 at the end of exon 1 of DES (Figure 2C; amplicon I). PBMCs did not show any enrichment across the DES LCR or DES but showed a marked abundance of ~25-fold above background of H3K4me2 at the control housekeeping gene promoter of HTATIP2 as expected from previous work [26].

H3K4me3 was not enriched at any site within the DES LCR within any of the cell types tested (Figure 2D; amplicons A-E). However, there was a ~40-fold enrichment of this mark above background levels before the start of the second exon of DES in both myoblasts and myotubes (Figure 2D; amplicons H-J). This decreased sharply in the myoblast sample by the second exon compared to that observed in the myotube culture, which remained high at this position (Figure 2D; amplicon J).

DNA methylation status of CpG sites within DES and DES LCR

DES is approximately 8.3 kb in length with 9 exons and is associated with a small compact CpG island of 550 bp that spans its first exon and has a CpG count of 74, a CpG percentage of 27.0% and a ratio of observed to expected CpG dinucleotides of 1.08. Due to the tissue-specific expressed nature of this gene we were able to correlate the histone modification patterns we observed within DES and the DES LCR with the underlying DNA methylation pattern in different transcriptionally active and inactive states. DNA methylation was analysed using sodium bisulphite conversion coupled with PCR/sequencing in adult human primary myoblasts, myotubes and PBMCs at CpG rich areas throughout the LCR and DES (Figure 3; amplicons I-VII) and which corresponded to regions of high sequence conservation between rodents (mouse/rat) and human [6].

Regions analysed within the DES LCR included HS5 and HS4d (Figure 3; amplicons I and II). The two CpG sites within HS5 within the region of rodent/human homology were fully methylated in all the cell types analysed (Figure 3B; panel I). HS4d contained 3 CpG dinucleotides all of which were unmethylated in the myotubes (Figure 3B; panel II, F), showed a mixed pattern of methylation in myoblasts (Figure 3B; panel II, UF) but were ~96% methylated in PBMCs (Figure 3B; panel II, PBMCs).

Within element HS3 only sub-regions HS3c and HS3b were analysed, as HS3a does not contain any CpG dinucleotides. All the CpG sites within HS3c and b in the unfused myoblasts were unmethylated (Figure 3, panel III and IV; UF). Within HS3b in the myotubes the majority of the CpG sites were unmethylated apart from 1 out of the total of 5 CpG sites analysed, which exhibited 90% methylation (Figure 3B, panel IV; F). All the CpG sites except for one within HS3c showed a high percentage of methylation (~40–100%) in PBMCs (Figure 3B, panel III and IV; PBMC) demonstrating that this region is highly methylated compared to muscle cells.

The DES TSS genomic region was analysed to determine where the region of transition occurs between unmethylated and methylated DNA in both a 5' and 3' direction (Figure 3). The 10 CpG sites within the +0.7 kb-+1.5 kb region exhibit a mixed pattern of methylation in both muscle and non-muscle cell types (Figure 3B; panel VII), which indicates that the methylation-free CpG area of DNA appears to end +0.7 kb 3' of the DES TSS. Analysis of a region ~0.5 kb upstream of the DES TSS at the start of the CpG island, showed that all seven CpG sites in this area lack methylation (Figure 3B; panel VI) in all cell types including PBMCs. Therefore, this unmethylated DNA is not related to the transcriptional status of DES. The extent of the island is restricted to the GC dense region spanning the first exon of DES (Figure 3A; CpG). These results also demonstrate that the gene body of DES is methylated starting before the start of the second exon even in expressing muscle cell types (Figure 3B; panel VII). Gene body methylation has now been reported in several systems [27–30].

The unmethylated CpG island spanning the DES promoter is marked by H3K27me3 in PBMCs

DNA methylation analysis showed that the CpG island spanning the DES promoter was unmethylated in both muscle and non-muscle cell types (Figure 3). Given that methylation-free CpG islands are usually associated with an open chromatin structure this poses the question as to how the DES promoter region is repressed in non-expressing cells? The absence of muscle specific transcription factors in non-muscle cells could be sufficient to prevent transcription of DES in non-muscle cells. However, the DES promoter does possess ubiquitous transcription factor binding sites such as Sp1 [31] that would be expected to be functional and therefore suggesting that additional mechanisms may contribute to the complete silencing of DES in non-muscle cells.

A possible candidate epigenetic mark for suppressing expression from the DES promoter in non-muscle tissues is H3K27me3, which has been implicated in marking unmethylated DNA for silencing [18]. In addition, methylation of H3K27 has been universally linked to gene repression [21, 32–34] unlike H3K9 methylation which has been found at both inactive and active genes [12, 35, 36]. Therefore, we determined whether H3K27me3 was present across the LCR and DES in non-muscle PBMCs and muscle cells. The only region that was found to be enriched by 9–15 fold for H3K27me3 above background levels was across the DES unmethylated CpG island in the PBMC sample (Figure 4; amplicons G-I). This peak of enrichment was tightly localized to the methylation-free CpG island and dropped to background levels by the second exon of DES (Figure 4; amplicon J). No significant enrichment above background was seen in the myoblast/myotube cultures (data not shown). Interestingly, this mark was not seen at either the -4 kb region between the LCR and the DES TSS or across the LCR in PBMCs (Figure 4; A-E). In addition, H3K27me3 was not observed at the CpG poor promoter of HBB that is also silent in PBMCs (Figure 4A). These observations provide strong support for the concept that H3K27me3 is a histone modification that marks silent chromatin at specific genomic locations [21].

Mammalian Cell culture

Adult human primary myoblast cells were a generous gift from Serge Braun, Association Française contre les Myopathies, Evry, France (formerly at Transgene SA, Strasbourg, France) and were isolated from a skeletal muscle biopsy obtained after informed consent and in accordance with local French regulatory approval from a healthy 14 year old Caucasian male. These myoblast cells were cultured in Dulbecco's modified eagles medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2% Ultroser^® G (Pall BioSepra, Cergy, France), 5 ml penicillin/streptomycin (pen/strep100 μg/ml) and 5 ml glutamine. To obtain fused myotubes, myoblasts were grown to confluency and transferred to fusion media consisting of DMEM, 10% FCS, 5 ml pen/strep (100 μg/ml), and 5 ml glutamine.

Whole blood samples were obtained by informed consent from normal healthy human volunteers in accordance with King's College London Ethical Committee guidelines and procedures. Whole blood was cultured for 72 hrs in 500 ml of RPMI (Roswell Park Memorial Institute) 1640, 5 ml 1 mg/ml Phaseolus Vulgaris Phytohemagglutinin (PHA-P 1 mg/ml, St. Louis, MO, USA), 5 ml pen/strep (100 μg/ml) and 5 ml glutamine. Peripheral blood mononuclear cells (PBMCs) were then isolated after lysis of red blood cells with RBC lysis buffer (0.1 mM EDTA, 155 mM NH₄Cl, 10 mM KHCO₃).

Native Chromatin Immunoprecipitation (nChIP)

nChIP experiments were performed at least twice using polynucleosomes from independently prepared chromatin preparations as previously described [30]. Both ChIP-on-chip and Q-PCR was conducted on the same input material. ChIP analysis for various histone modifications were conducted as before [30]. Equivalent amounts of recovered nChIP input and immunoprecipitated (IP) DNA were analysed by quantitative real-time PCR (Q-PCR) using the QuantiTect SYBR Green PCR Master Mix (Qiagen, Crawley, UK) system on a LightCycler™ 1.2 (Roche Applied Science, Lewes, UK) as described [40]. Primers used for PCR amplification are shown in Additional file 1, Table ST1. A melting curve program was performed and showed that all PCR primers were specific for the target sequences (Additional data file 1, Figures S4B and B'). Fold enrichment of selected regions of DNA within IP samples was calculated by subtracting the IP Ct (cycle threshold) value from the input Ct value and was presented as a ratio of 2^-CtIP to 2^{-Ct input} [30, 40]. All primers were optimized to amplify the same amount of DNA with equivalent efficiency (Additional data file 1, Figures S4C and C').

ChIP-on-chip analysis

A high resolution DNA microarray was constructed across 500 kb of chromosome 2 using 500 bp overlapping PCR products as previously described [41]. Briefly, primer pairs were designed using the Primer3 software and Web site with Repeat Masking. All primers had a universal amino-linked tag sequence (5'-TGACCATG-3') attached to the 5' end of the sense strand primer, to allow covalent binding to the microarray slide. PCR reactions were conducted on a Flexigene Techne PCR machine using BAC clone accession number AC053503 DNA as a template which encompasses the DES locus and surrounding regions. DNA was denatured at 94°C for five minutes followed by 30 cycles of 94°C for 30 seconds, 65°C for 30 seconds and 72°C for 60 seconds. A final extension step at 72°C for five minutes was performed. A 5 μl aliquot of the reaction was run on a 2.5% agarose gel to confirm product specificity. Amino-liked specific PCR products were spotted onto the glass array slides in duplicate by the microarray facility at the Wellcome Trust Sanger Institute, Hinxton, Cambridge.

Approximately 300 ng of nChIP DNA sample, input and IP were labelled using Cy-5 and Cy-3 dye modified dCTP respectively. These samples were pooled precipitated with human Cot-1 DNA (Roche Applied Sciences, Lewes, UK) and resuspended in 100 μl of hybridization buffer (10 ml 20×SSC, 50 ml deionised formamide, 10 ml 100 mM Tris-HCl pH 7.4, 5 g dextran sulphate, 100 μl TWEEN 20) and left at 65°C to allow the labelled DNA pellets to dissolve.

The samples were loaded onto the microarray slides in the Tecan HS 400™ Pro hybridization machine (Tecan Group Ltd., San Jose, USA) for a 72 hour hybridization reaction which included subsequent washing steps.

Hybridised microarray slides were scanned on an Axon 4000B scanner (Molecular Devices, Concord, Canada). The photon multiplier tube (PMT) levels used for detection were adjusted and tailored for each individual array. Arrays were analysed using Genepix 4.0. Software (Molecular Devices, Concord, Canada). Single image TIFF files from Genepix 4.0 were imported into the 'Spot' program for analysis [42]. The spot file was then uploaded into a custom made Excel based macro, which gave average enrichment ratios for each region represented on the array. The DNA chip data was normalized over the entire microarray which not only contained the ENr331 region but also other ENCODE areas under investigation, specifically loci from chromosomes 10, 11 14, 16 and X. The array also had smaller tiles from chromosomes 2, 5, 7 and 9. This allows accurate comparison of the histone modification patterns between different cell types. Even though diverse patterns of histone modifications can be compared between different ChIP samples, absolute levels of histone H3 and H4 acetylation cannot be evaluated as antibody affinities will vary [43]. The ChIP-on-chip experiment was conducted in duplicate for each spot on the array and repeated with an independent biological sample. The average ratio for each set of duplicates was then calculated and the mean was taken between independent experiments.

DNA bisulphite conversion and sequencing

Total genomic DNA (2 μg) from myoblasts, myotubes and PBMCs, was digested with 5 μl Hind III (100 units) and 5 μl 10× buffer in a reaction volume of 50 μl for 3 hrs at 37°C. The digested DNA was then denatured by adding 3 M NaOH to a final concentration of 0.3 M and incubated at 37°C for 15–20 min. The denatured DNA was added to 1.1 ml of pre-warmed freshly prepared sodium bisulphite solution (4.5 M sodium bisulphite, 2.5 mM hydroquinone, 0.225 M NaOH, pH5), covered with 200 μl of mineral oil and incubated for 5 hrs at 55°C in the dark. The bisulphite treated DNA was purified using the QiaexII gel extraction kit (Qiagen). The sample was then desulphonated by adding 3 M NaOH to a final concentration of 0.3 M and incubated at 37°C for 15–20 min. The sample was then neutralized by adding 6 M ammonium acetate to a final concentration of 3 M and DNA precipitated by adding 3 volumes of 100% ice-cold ethanol. PCR amplification of selected regions with nested pairs of primers (see Additional file 1, Table ST1) was carried out on the bisulphite converted DNA using the same conditions as above but for 35 cycles. PCR products were cloned into the pCR 2.1 TOPO vector (Invitrogen, Paisley, UK) and sequenced using M13 forward and reverse primers.