Functional epigenomics: chromatin complexity untangled

Abstract

The development of an epigenetics-focused, CRISPR-based high-content functional genomics screening platform provides insight into chromatin regulation and uncovers a potential strategy to treat an aggressive type of leukemia.

Functional genomics is a powerful discovery method to assess in parallel the role of many to all genes present within a cell type for involvement in specific phenotypes or biological processes. The integration of CRISPR-based gene-editing technology into functional genomics screening workflows has profoundly expanded opportunities to systematically examine gene functions in different cell types¹. CRISPR screens frequently use a ‘pooled’ approach in which a library of single-guide RNAs (sgRNAs) is introduced in bulk into cells in such a way that an individual cell integrates a specific sgRNA and is accordingly genetically perturbed. High-throughput readouts commonly employed in pooled screens include cell fitness (such as proliferation or a similar selective pressure) or expression of a biomarker (for example, a receptor on an immune cell) using fluorescence-activated cell sorting (FACS) approaches (Fig. 1a). In either case, counting of the sgRNA sequences present in the total cell population before and after the challenge of interest is used to compare the two conditions and can reveal specific genetic perturbations acting as positive and negative regulators of the underlying process.

Fig. 1 |. CRISPR–ChIP functional genomics platform for analysis of epigenetic pathways regulating H3K4me3 and H3K79me2.

a, Schematic showing common high-throughput functional genomics readouts of viability and FACS. High-content CRISPR–ChIP readout described in Gilan et al.⁴ is also shown. The CRISPR library is indicated by circles, with the differently colored arrows representing unique sgRNAs. b, Summary of H3K4me3 CRISPR–ChIP screen. The four top hits in the screen are essential core components of complexes that synthesize H3K4me3 and are shown above the dotted line. The proteins in gray shown below the dotted line are the enzymes that generate H3K4me3 but are not detected in the screen, likely because of functional redundancy. c, Summary of H3K79me2 CRISPR–ChIP and viability screens. Top hits in the H3K79me2 CRISPR–ChIP (top) and viability (bottom) screens are shown. DOT1L was identified in both screens, whereas MLLT0 is required for H3K79me2 generation but not for cell viability. In MLL–AF9-rearranged leukemia cells, DOT1L is proposed to exist in both a native complex with MLLT10 and a disease-promoting complex with the MLL–AF9 fusion protein and MENIN. Depletion of MLLT10, which has no effect on cell viability, renders cells vulnerable to a MENIN inhibitor, a situation that may be modeled in the clinic by combining MENIN and DOT1L inhibitors to treat MLL–AF9-rearranged leukemia.

An exciting area of functional genomics is the adaptation of complex biological readouts amenable to high-throughput screening¹. Epigenetic regulation of chromatin biology is a multifactorial process with fundamental roles in human health and disease^2,3. However, to date, a method to functionally investigate the gene networks regulating epigenetic mechanisms at chromatin using CRISPR-based screening technology has not been described. In their work, Gilan et al.⁴ develop a functional epigenomic platform termed ‘CRISPR–ChIP’ for identifying genes involved in determining a specific chromatin state and demonstrate in proof-of-concept experiments how this system can be used to uncover insights into the molecular pathways underpinning regulation of two key activating histone modifications (Fig. 1a).

In eukaryotes, DNA is compacted by histones and other proteins into a higher-order structure termed chromatin³. A complex molecular network at chromatin regulates genomes, with DNA-templated processes such as transcription and DNA repair being fundamentally affected by chromatin structure and dynamics. A major mechanism for chromatin regulation involves the reversible methylation of lysine residues on histone proteins⁵. Different histone methylation events are thought to underlie the establishment of discrete chromatin functional states and thereby regulate the extent of accessibility of DNA to transacting factors. For example, two important methylation events associated with actively transcribed chromatin are trimethylation of histone H3 at lysine 4 (H3K4me3) and dimethylation of histone H3 at lysine 79 (H3K79me2)⁵. Chromatin immunoprecipitation (ChIP) and newer methods are used to measure the relative levels of histone modifications such as H3K4me3 and H3K79me2 at chromatin at specific loci as well as across the genome⁶. In ChIP, the levels of a specific histone modification or protein at specific DNA sites is determined by immunoprecipitating the epitope of interest (such as H3K4me3) from chromatin sheared into small DNA fragments, and then quantifying the relative amount of DNA present in the immunoprecipitate by targeted PCR or high-throughput sequencing approaches.

Gilan et al.⁴ used a ChIP-based method to perform their CRISPR-guided functional genomics experiments. In their system, a chromatin reporter region is integrated by viral transduction into the cells of interest; the reporter harbors a known cis-regulatory element that, upon integration in the genomic DNA, is shown to be physiologically chromatinized with H3K4me3 and H3K79me2 enrichment. Importantly, the sgRNA targeting sequences are incorporated 50–100 base pairs upstream of the cis-regulatory element and thus, depending on the level of DNA shearing, can be pulled down in ChIP experiments targeting the cis-regulatory element. Comparison of sgRNA sequences present in genomic DNA (input) and ChIP samples allows the authors to determine enrichment or dropout of individual sgRNAs based on the specific ChIP assay (for example, H3K4me3 ChIP). One potential concern with the approach is that the site of integration for the chromatin reporter is random (though it is naturally inclined to integrate within accessible chromatin) and thus may introduce bias into the readout. The authors mitigate this issue by including several essential controls, ranging from using six different guides per gene to showing that the epigenetic landscape at the reporter mirrors the chromatin environment at the endogenous regulatory element.

To benchmark the CRISPR–ChIP method, Gilan et al.⁴ tested a focused epigenetic library targeting 1,444 chromatin and transcription regulatory proteins in the K562 leukemia cell line and assay for H3K4me3 ChIP. In this screen, the top hits were the four nonredundant and core essential components shared by all complexes that synthesize H3K4me3 in somatic cells, demonstrating the robustness of the method⁷ (Fig. 1b). Notably, although some weaker hits that would be predicted to affect H3K4me3 generation were identified, including components of the PAF transcription elongation complex, surprisingly, components of the RNA polymerase II (RNAPII) complex were not. Using orthogonal approaches, the authors provide convincing evidence that H3K4me3 deposition is indeed not dependent on RNAPII activity, suggesting that transcription and H3K4me3 enrichment can be decoupled. Another aspect of the screen was the absence of the main enzymes that generate H3K4me3 (MLL1–4 and SET1A/B) as hits, almost certainly due to redundancy amongst these enzymes^3,5,7. High homology between chromatin factor family members is common, and thus essential histone-modifying activities that are encoded in redundant genes will be likely missed in the current rendition of the CRISPR–ChIP method. Future generations of the approach, in which genes are depleted in a pairwise fashion, would circumvent the redundancy issue and could reveal previously unknown functional synergies and vulnerabilities.

In a second proof-of-principle experimental series, Gilan et al.⁴ turned their focus to the regulation of H3K79me2 in MLL-rearranged (MLL-r) leukemic cells, specifically those harboring the MLL–AF9 neomorphic fusion^3,8. Performing H3K79me2 CRISPR–ChIP, the authors identified DOT1L and MLLT10 as the top two hits (Fig. 1c). DOT1L is the sole known enzyme generating H3K79 methylation, and MLLT10 is a key DOT1L-interacting protein required for generating the bulk of H3K79me2^5,8. Beside these top two hits, several other proteins involved in various aspects of transcription elongation regulation, and previously linked to H3K79me2, also scored as positive hits. Together, these results, like the H3K4me3 screen, highlight the capability of CRISPR–ChIP to identify known, nonredundant regulators of histone modifications on endogenous chromatin. In both the H3K4me3 and H3K79me2 screens, there were several weak to strong positive hits that are not elaborated upon; do these hits have the potential to reveal unexpected epigenetic-regulatory modes, or do they represent background noise (or a combination of the two)? As further experience using CRISPR–ChIP is gained, these types of questions are likely to be addressed.

An advantage of the CRISPR–ChIP workflow is the ability to screen for two readouts in parallel, for example screening H3K79me2 ChIP side by side with cell viability (Fig. 1c). Performing such a screen in MLL-r leukemic cells, the authors discovered that although MLLT10 is essential for global H3K79me2 generation, it is not essential for viability. In contrast, DOT1L is needed for both H3K79me2 generation and cell viability (Fig. 1c). Focusing on this discrepancy, the authors provide strong evidence that DOT1L is incorporated into two complexes: one in which DOT1L is targeted to oncogenic genes through interactions with the neomorphic MLL–AF9 fusion protein, and a second, physiologic DOT1L complex, characterized by the presence of MLLT10, that regulates the bulk of H3K79me2 (Fig. 1c). Depletion of MLLT10, which disrupts the native DOT1L complex but on its own does not impinge on MLL-r leukemia cell viability, renders the MLL–AF9-rearranged cells vulnerable to MENIN inhibition⁹. The authors posit that excess DOT1L released from the MLLT10 complex leads to increased MENIN–MLL–AF9–DOT1L complex formation, thus rendering cells dependent on the activity of this complex to generate H3K79me2 at key oncogenic targets. In this context, combination treatment with MENIN and DOT1L inhibitors shows strong synergistic effects in suppressing the proliferation of MLL–AF9-rearranged leukemic cells, hinting that this strategy may be therapeutically beneficial in the clinic.

Gilan et al. have established and validated CRISPR–ChIP as a discovery method that can be used to elucidate molecular principles governing chromatin biology⁴. Although technical issues related to the location of the sgRNA being close to the reporter necessitate limited sonication to maintain fairly large DNA fragments, the overall design of the system is straightforward and can be adopted for broad use in the field. One can envision subsequent versions of the approach that enable testing of gene combinations, expanded libraries allowing genome-wide exploration of chromatin regulation, and incorporation of higher-content read-outs, such as full epigenome analyses with multiple histone modifications, using profiling technology such as CUT&RUN and CUT&TAG^6,10. In summary, the development of CRISPR–ChIP provides a powerful and unbiased tool that can be used to uncover epigenetic principles relevant for human health and disease at a functional genomics level.