Chromatin landscapes and genetic risk in systemic lupus

Enhancers are both cell specific and cell-state specific [19]. Because neutrophils were not among the cells studied in either the ENCODE or Roadmap Epigenomics projects, we sought to create a genomic map for enhancer element locations using normal adult neutrophils. We obtained neutrophils from three healthy adults aged 25–40 using techniques we have described previously [11].

Neutrophils were isolated as described previously [20]. The ChIP assay was carried out according to the protocol of the manufacturer (Cell Signaling Technologies Inc., Danvers, MA, USA) and has been described in our work published previously [11]. Briefly, adult neutrophils were incubated with newly prepared 1% formaldehyde in PBS at room temperature (RT). Crosslinking was quenched by adding 1× glycine. The crosslinked samples were centrifuged, the supernatant discarded, and the pellet washed with cold PBS followed by resuspension in 10 ml ice-cold Buffer A plus DTT, PMSF, and protease inhibitor cocktail. Cells were incubated on ice and then centrifuged at 4 °C to precipitate nucleus pellets, which were then resuspended in 10 ml ice-cold Buffer A plus DTT. The nucleus pellet was incubated with Micrococcal nuclease for 20 minutes at 37 °C with frequent mixing to digest DNA. Sonication of nuclear lysates was performed using a Sonic Dismembrator (FB-705; Fisher Scientific, Pittsburgh, PA, USA) on ice. After centrifugation of sonicated lysates, the supernatant was transferred into a fresh tube. Fifty microliters of the supernatant (chromatin preparation) was taken to analyze chromatin digestion and concentration. Fifteen micrograms of chromatin was added into 1× ChIP buffer plus protease inhibitor cocktail to a total volume of 500 μl. After removal of 2% of chromatin as the input sample, the antibodies were added to the ChIP buffer. The antibodies against respective histone modifications were rabbit polyclonal antibodies against histone H3 acetylated at lysine 27 (H3K27ac) and histone H3 monomethylated at lysine 4 (H3K4me1) from Cell Signaling Technologies. The negative control was normal IgG (Cell Signaling Technologies). After immunoprecipitation, the magnetic beads were added and incubated for another 2 hours at 4 °C. The magnetic beads are covalently coupled to a truncated form of recombinant protein G. They were then collected with a magnetic separator (Life Technologies, Grand Island, NY, USA). The beads were washed sequentially with low and high salt wash buffer, followed by incubation with elution buffer to elute protein/DNA complexes and reverse crosslinks of protein/DNA complexes to release DNA. The DNA fragments were purified by spin columns and dissolved in the elution buffer. The crosslinks of input sample were also reversed in elution buffer containing proteinase K before purification with spin columns. DNA sequencing was then conducted using the Illumina HiSeq 2500 at the next-generation sequencing center in University at Buffalo.

Analysis of the ChIP-seq data was carried out exactly as described previously [11]. MACS2 v2.1.10 [21] was applied for calling regions enriched with histone marks against the input sample, with the parameters “–nomodel --extsize 150 --broad –broad-cutoff 0.1”. Details of these analyses are further described by Jiang et al. [11].

In order to compare data from neutrophils with existing data from CD19⁺ B cells and CD4⁺ T cells, we queried data generated from the Roadmap Epigenomics Project [22]. Raw ChIP-seq data for CD19⁺ primary cells were downloaded from the GEO database [23] [GEO:GSM1027296, GEO:GSM1027287, GEO:GSM1027300, GEO:GSM1027304] for H3K4me1, H3K27ac, H3K4me3, and input control respectively. Raw ChIP-seq data for CD4⁺ T cells were downloaded [GEO:GSM1220567, GEO:GSM1220560, GEO:GSM1102798, GEO:GSM1102805] for H3K4me1, H3K27ac, H3K4me3, and input control respectively. The methods for mapping and region-calling are the same as those was used to analyze neutrophil data.

ENCODE transcription factor binding site (TFBS) data were downloaded from UCSC Genome Browser ENCODE (http://​hgdownload.​cse.​ucsc.​edu/​goldenPath/​hg19/​encodeDCC/​wgEncodeRegTfbsC​lustered/​). Only the TFBS information derived from blood cells was used for analysis. The whole genome was binned to 100 bp bins and intersected with H3K27ac, H3K4me1, or H3K4me3 peak regions, which were used as background. Fisher’s exact test was applied to test the significance of enrichment for TFBS for each transcription factor (TF) within H3K27ac, H3K4me1, or H3K4me3 peaks within linkage disequilibrium (LD) blocks compared with peak regions in the whole genome. The cutoff point for the false discovery rate (FDR) was set to 0.05.

We searched the LD regions near (within 5 kb of) each GWAS index locus for association with histone marks from ChIP-seq. LD blocks were defined for 46 out of the 58 SNPs described in recent GWAS [17, 18] using information from the SNAP database (http://​www.​broadinstitute.​org/​mpg/​snap) [24] by querying data from the 1000 Genomes Project pilot and the HapMap3 database. LD blocks were defined using r ² < 0.9.

We further investigated H3K4me1, H3K4me3, and H3K27ac distal regions relative to transcription start sites. The distal regions typically correspond to cis-acting enhancers located far away from the gene(s) they regulate [25]. Regions containing at least one methylated (H3K4me1 or H3K4me3) and one acetylated histone mark (H3K27ac) were referred to as active enhancers, and those that contained only one methylated histone group or H3K27ac region were referred to as poised enhancers or H3K27ac-active enhancers, respectively. Of note, H3K4me3 appears to be cell-type specific, expressed in cells of the lymphoid lineage, and noted to be an important histone mark for enhancer activity [26].

We found functional elements within 5 kb of 36 of the 46 SNPs in adult neutrophils. Epigenetic evidence for active enhancers were found in 29 LD blocks, poised enhancers in six LD blocks, and H3K27ac-active enhancers in one LD block (Table 1). Using Fisher’s exact test, these regions were significantly enriched for enhancer activity above the genomic background (p < 0.05) (Fig. 1).

Using the same approaches as we used for neutrophils, we interrogated available H3K4me1, H3K4me3, and H3K27ac ChIP-seq data from the Roadmap Epigenomics project for CD4⁺ T and CD19⁺ B cells. We found considerable overlap for the locations of H3K4me1 and H3K27ac peaks between our neutrophil data and resting CD4⁺ T cells. Functional elements were found within 39 of the 46 SNPs of interest in CD4⁺ T cells. Epigenetic evidence for active enhancers was found in 30 LD blocks, and for poised enhancers in nine LD blocks; no LD blocks contained H3K27ac marks alone (Table 2).

In CD19⁺ B cells, we identified functional elements found in 42 of 46 SNPs of interest. Epigenetic evidence for active enhancers was present in 32 LD blocks, and for poised enhancers in 10 LD blocks; no LD blocks contained H3K27ac histone marks alone (Table 3). There are thus more SNPs of interest in SLE with functional elements within lupus-associated LD blocks in CD19⁺ cells than in either CD4⁺ cells or neutrophils (p < 0.05). Representative screenshots from the UCSC Genome Browser with two of the LD blocks of interest are shown in Fig. 2.

While it is difficult at this time to determine exact differences in the acetylation and methylation of patients with SLE compared with healthy controls due to the scarcity of available data, we did perform an analysis of differentially methylated regions as identified by Absher et al. [27] in adult SLE patients. This analysis revealed only one gene (PBX2, on chromosome 6) lying within the same LD block as rs1270942 to be aberrantly methylated in T cells, B cells, and/or monocytes in SLE patients. In addition, analysis of differentially methylated regions identified by Coit et al. [28] in the neutrophils of adult SLE patients compared with healthy adult data from our laboratory and available in Roadmap Epigenomics revealed that none of these regions were located within the LD blocks containing the SLE-associated SNPs. Less than 40% of these regions contained enhancer marks in healthy adult neutrophils. Fewer than 1/3 of these regions contained enhancer marks in healthy adult CD4⁺ and CD19⁺ cells.

We next sought to further test the likely functional significance of H3K4me1, H3K4me3, and H3K27ac enrichment within the SLE-associated LD blocks. We therefore analyzed the TF ChIP-seq data from blood cells obtained from the UCSC Genome Browser ENCODE data portal [29] to determine whether there was significant enrichment for TF binding within the enhancer regions located within SLE-associated LD blocks compared with other regions (Fig. 3). We investigated both promoter and nonpromoter regions within the regions where histone marks (H3K4me1, H3K4me3, and H3K27ac) are associated with enhancer activities. As expected, promoter regions within these LD blocks (defined as (−5 K, 1 K) of transcription start sites) are highly enriched for TF binding sites. Furthermore, we identified more enrichment for TF binding sites in H3K4me3 peak regions than in either H3K27ac or H3K4me1 peak regions in promoter sites in CD19⁺ and CD4⁺ T cells (p < 0.05). Overall, the LD blocks containing lupus-associated SNPs appeared to be in active, dynamic regions of leukocyte genomes as determined by the abundance of transcription factor binding motifs within these regions

Of note, 23 out of 46 SNP regions have shared histone modifications in all three cell types investigated. There are 102 genes located within the 46 SNP LD blocks, which are involved in 26 Panther pathways, including T-cell and B-cell activation, and Jak/STAT signaling pathways (Additional file 1: Table S1). Moreover, Farh et al. [30] demonstrated that many causal variants which map to immune-cell enhancers may gain histone acetylation and transcribe enhancer-associated RNA upon immune stimulation. This could indicate that the majority of SNPs are functional SNPs. However, expression quantitative loci analysis (eQTL) revealed that only one SNP (rs2736340) was associated with a target gene (FAM167A). Similarly, when the SNPs of interest were compared with genes identified by Bennett et al [31]. whose expression was upregulated in SLE patients, only one gene was found to lie in the same LD block as one of the identified SNPs: a phorbolin-1 like gene from the interferon family lies in the same LD block as SNP rs61616683.