Many of us remember being taught a simplified doctrine on histone modifications: certain marks tend to appear on transcriptionally inactive, condensed heterochromatin, while others characterize active, open euchromatin. In particular, H3K27me3 tends to mark facultative heterochromatin that may be expressed during development, whereas H3K9me3 is associated with constitutive heterochromatin. However, histone modifications do not reliably distinguish between heterochromatin and euchromatin, as numerous studies have shown, and as Justin Becker and colleagues in the Zaret lab demonstrate in their recent publication in Molecular Cell1. They debut a new technique called gradient-seq that separates chromatin based on physical properties instead of covalent histone modifications, and use this tool to interrogate differences between heterochromatic and euchromatic DNA bearing various histone marks. The researchers began by critically assessing the standard protocol for ChIP-seq, which requires crosslinking and sonicating chromatin, precipitation using antibodies to histone marks or other proteins, and then sequencing. The sonication step is crucial, as only DNA fragments under a certain size can be sequenced. However, dense heterochromatin is resistance to shearing and may be underrepresented in the sequencing reads.
Justin developed the idea of placing crosslinked and sonicated chromatin in a sucrose gradient to separate fragments into lighter, shorter, sonication sensitive portions, and heavier, longer, sonication resistant segments. He then sequenced these purified chromatin fractions (a method termed gradient-seq) and asked whether the gradient had successfully sorted DNA into euchromatin and heterochromatin. As expected, the longest DNA segments possessed characteristics of heterochromatin: traditionally repressive histone marks, low transcriptional activity, and CpG methylation. Sequencing revealed that genes in the heavy DNA chromatin fraction were marked with H3K9me3 and resisted activation during fibroblast reprogramming to hepatocytes, meaning that they are functionally heterochromatic. Thus, the heaviest gradient fraction earned the name “sonication resistant heterochromatin” (srHC). In contrast, the euchromatic DNA fragments were actively transcribed, unmethylated at CpG islands, enriched for traditionally permissive histone marks, and depleted for repetitive elements.
Having reliably separated heterochromatin and euchromatin, the team investigated the true landscape of histone marks on each type of DNA. The srHC fraction contained the majority of H3K9me3 and H3K27me3 marks, which generally did not overlap. Surprisingly, certain subpopulations of euchromatin also bore the “repressive” histone marks H3K9me3 and/or H3K27me3, despite being transcribed. These subpopulations also carried the H3K36me3 mark, which corresponds to transcriptional elongation. The euchromatic H3K9me3 population of DNA was highly enriched for KRAB-containing zinc finger nucleases (KRAB-ZNFs). These genes had an unusual pattern of histone marks: dual marking by H3K9me3 and the elongation mark H3K36me3, H3K9me3 in the gene body, and H3K4me3 at the promoter. Euchromatic DNA with H3K27me3 contained genes for many transcription factors, especially HOX family genes. During reprogramming of fibroblasts to hepatocytes or cholinergic neurons, genes marked by H3K9me3 and/or H3K27me3 in the euchromatin gradient portion were more prone to activation than those in the heterochromatic gradient portion. This demonstrates that the functional capacity of chromatin is more accurately represented by gradient separation than by histone modifications.
Nuclease digestion is often used to separate DNA by chromatin state, given that nucleases cut at euchromatic regions and leave denser, protein-bound heterochromatin untouched. However, when Justin compared DNA digestion by DNase and micrococcal nuclease to the fractions obtained by gradient separation, there were notable differences. The nucleases were unable to digest H3K9me3-marked DNA that is euchromatic by transcriptional activity and sonication sensitivity. Thus, gradient-seq is a more reliable indicator of chromatin state than nuclease digestion.
With a tool in hand to separate euchromatin from heterochromatin, the researchers next investigated which proteins are bound to heterochromatin, especially H3K9me3-marked heterochromatin, which is particularly resistant to transcriptional activation during cellular reprogramming. They performed mass spectrometry on srHC DNA, srHC DNA following H3K9me3 inmmunoprecipitation, and the “gradient top” fraction containing euchromatic DNA and soluble proteins. There were 217 proteins enriched in the srHC fraction, 172 enriched in H3K9me3 srHC, and 1474 unique to or enriched in the gradient top. Many of the proteins bound to H3K9me3 srHC are involved in chromatin organization, control of gene expression, and inhibition of reprogramming; such proteins include linker histones, lamin B1, HDAC2, HNRNPK, and NONO and SFPQ. Intriguingly, the srHC fraction also contained six proteins frequently mutated in Amyotrophic Lateral Sclerosis (ALS). These proteins bear domains that mediate phase separation, which has been shown to be involved in chromatin organization. Mutations in these proteins could, therefore, lead to abnormal phase separation and chromatin dysregulation, which may contribute to ALS pathogenesis. The technique and the data sets that Justin created in this study certainly hold promising leads to inform research on chromatin regulation and disease states. Finally, when many of the proteins enriched in H3K9me3 srHC were knocked down using RNAi, fibroblasts became more easily converted to hepatocytes, indicating that H3K9me3 heterochromatin functions to restrict changes to cell identity.
Sucrose gradient sedimentation separates sonicated chromatin into euchromatin and heterochromatin as characterized by DNA sequencing matched to mRNA sequencing, proteomics, nuclease sensitivity, and impact on cell reprogramming. The technique, called gradient-seq, is more accurate than histone modifications in identifying euchromatin and heterochromatin, and can be used to better characterize chromatin states. (Graphical Abstract from Becker et al. 2017, Molecular Cell 68, 1023)
In summary, the technique of gradient-seq, which couples DNA chromatin separation by size using a sucrose gradient with DNA sequencing, delineates heterochromatin and euchromatin based on the true functional state of chromatin rather than on histone modifications. Justin demonstrates that ChIP-seq data sets are skewed by the exclusion of large, sonication resistant heterochromatin fragments and that histone modifications alone do not reliably predict chromatin state. Gradient-seq can be paired with proteomics, histone modification mapping, or other means of characterizing DNA chromosomes in order to truly interrogate differences between euchromatin and heterochromatin
Becker JS, McCarthy RL, Sidoli S, Donahue G, Kaeding KE, He Z, Lin S, Garcia BA, Zaret KS. Genomic and Proteomic Resolution of Heterochromatin and Its Restriction of Alternate Fate Genes. Mol Cell. 2017 Dec 21;68(6):1023-1037.e15.