INTRODUCTION

The human hematopoietic system encompasses a wide range of cell types with unique morphologies and functions, involved in processes as disparate as immune defense, nutrient transport, or coagulation. However, like every other organ and tissue in the human body, they all carry the exact same genetic information: 46 chromosomes containing roughly 20,000 protein-coding genes.1,2 If they all share the same genome, what accounts for the full spectrum of cells in the blood, not to mention the entire organism? The answer resides in the precise regulation of gene expression. Although between 40% and 50% of human genes are ubiquitously expressed, a subset of genes that determine cell identity are only expressed in a tissue-specific manner.3–6 These patterns of expression change along hematopoiesis, as cells progressively specialize and commit to certain lineages. To understand why some genes are active and others are silent, ensuring the maintenance of highly specific transcriptional programs, one must look beyond coding sequences, and put the lens on a much less understood part of the genome—regulatory elements.

The concept of gene regulation can be traced back to the model of Jacob and Monod, derived from their studies of the lactose system in bacteria.7 In their seminal publication, they proposed that repressor molecules could bind regulatory elements (operators) on the DNA to regulate the synthesis of proteins through short-lived RNA intermediates. Despite significant advances in the field, this surprisingly prescient model outlined the 2 major modes of transcriptional regulation. On the one hand, molecules that bind the DNA are transcription factors (TFs) that act in trans to control the expression of multiple genes across the entire genome. On the other hand, noncoding DNA regions bound by these factors are cis-regulatory elements (CREs) that are specific to genes in the vicinity. In turn, CREs can be classified into various functional classes, among which enhancers are of particular interest as critical determinants of cell identity.8

This review describes the role of enhancers in transcriptional regulation during hematopoiesis, with a focus on their involvement in malignant transformation. Although key aspects about enhancer biology are presented here, we refer the reader to other excellent reviews for a more in-depth discussion.8–11

ENHANCERS IN TRANSCRIPTIONAL REGULATION Principles of transcriptional regulation

Gene expression starts with transcription, defined as the copying of a DNA sequence into complementary RNA by a member of the RNA polymerase family of enzymes. RNA polymerase (RNA pol) II transcribes all protein-coding and most noncoding genes, whereas RNA pol I and III transcribe ribosomal RNA and certain small noncoding RNAs, respectively.12 Transcription can be divided into 3 distinct phases: initiation, elongation, and termination. It begins at the transcriptional start site (TSS), located at the 5′ end of a gene, and progresses toward its 3′ end. Upon completion of this process, the product of protein-coding genes, known as mRNA, is imported into ribosomes for translation.13 In this final step, a protein is synthesized by sequentially adding amino acids, following the order dictated by sequences of 3 nucleotides (codons) in the mRNA.14

Spatiotemporal regulation of gene expression is mediated by CREs, which include promoters, enhancers, insulators, and silencers (Figure 1A). Originally identified by Monod and colleagues in 1964, a promoter is a start signal at the beginning of a gene that directs RNA pol II to initiate transcription.15,16 The minimal stretch of DNA sufficient to direct this process is known as the core promoter, defined as a 50-bp region around the TSS that docks the preinitiation complex, which consists of RNA poll II together with general transcription factors (GTFs).9,16 Moreover, the rate of initiation can be modulated by TFs, proteins that bind specific DNA motifs and recruit components of the transcription machinery.17,18 To achieve this goal, they rely on 2 types of functional domains: DNA-binding domains recognize TF-binding sites (TFBS), whereas effector domains interact with other proteins, including RNA pol II and transcriptional cofactors (which can be either activators or repressors).19–22 TFBS appear at CREs in dense clusters, arranged with precise order, orientation, and spacing to ensure that TFs can cooperate effectively.17 Various modes of cooperativity, which can involve direct protein-protein interactions,23 DNA-facilitated interactions,24 or other indirect mechanisms, enable a finer control of transcriptional patterns.25

Figure 1.:

Transcription initiation is regulated by enhancers and promoters. (A) Promoters recruit GTFs, which in turn facilitate the binding of RNA Pol II, leading to the formation of the PIC. Transcription from promoters is favored by distal enhancers, which bind sequence-specific TF and COF. (B) Chromatin loops mediated by architectural proteins such as cohesion and CTCF enable contacts between distant enhancers and promoters, while preventing interactions with CREs outside the loop. COF = cofactors; CRE = cis-regulatory element; CTCF = CCCTC-binding factor; GTFs = general transcription factors; PIC = preinitiation complex; Pol II = polymerase II; TFs = transcription factors.

Although core promoters are capable of driving autonomous transcription, they often have low basal activity. In order to reach the expression levels required by the cell, they may thus require input from enhancers.26,27 Enhancers collaborate in the recruitment of RNA pol II by forming loops with target promoters, which can be located kilobases away in the linear genome (Figure 1B).28 In addition, there are a number of other distal CREs that participate in gene regulation, including silencers and insulators. Silencers reduce transcription from their target promoters by bringing repressive TFs, known as repressors (Figure 1A).29 Insulators bind architectural proteins such as CCCTC-binding factor (CTCF) or cohesin that generates loop domains, thereby blocking interaction across domains and favoring those within the same loop.30 Thus, contacts between CREs and their target genes usually take place within these insulated regions, often referred to as topologically associated domains (TADs).31

What is an enhancer?

Enhancers are DNA sequences of a few hundred basepairs that contain TFBS and increase the level of transcription from their target promoters.32,33 Enhancers were discovered in the 1980s through the identification of a 72-bp DNA sequence from the SV40 virus that increased transcription of a reporter gene by ≈200-fold, irrespective of distance and orientation.27,34 The first cellular enhancer was later found in the immunoglobulin heavy chain (IGH) gene locus, within the intron preceding the constant region exons.35,36 The authors noted the striking tissue specificity of this element, which was only active in B cells. These early discoveries established the key properties of enhancers: (1) they augment gene expression of their target genes; (2) act independently of orientation; (3) can function at large distances; and (4) are often tissue-specific. Furthermore, enhancers preserve their function even in a different genomic context, as shown by reporter assays, which also has implications for disease. Successive studies have consistently confirmed these features, and the importance of enhancers for tissue-specific gene regulation in vivo.37,38 Moreover, enhancers are modular and can contribute either additively or synergistically to the transcriptional output of their target genes.39–41 Long-distance interactions are mainly synergistic and confer robustness against comutagenesis, whereas the additivity of short-range enhancers maintains high expression.42

Recent estimates of the number of potential enhancers in the human genome range from a low of 40,00043 to more than a million,44–46 depending on the predictive approaches employed and the number of tissues surveyed. Despite the disparity of these figures, in all cases they greatly exceed the number of promoters detected. Nevertheless, the repertoire of enhancers active in each lineage is only a fraction of this number. Together with the fact that most binding events of TFs take place at enhancers, this points to a pivotal role for enhancers in the regulation of tissue-specific gene expression and cell identity.8 Indeed, enhancers in conserved regions are key regulators of development and disease47,48 and enhancer activity strongly correlates with gene expression in genome-wide studies.49–51 In the hematopoietic system, this notion is further supported by the fact that clustering of chromatin accessibility classifies cell types better than gene expression.52 More recently, multimodal single-cell approaches have demonstrated positive correlation between CRE accessibility and gene expression, with enhancer activation in early hematopoietic stages preceding transcription in more differentiated cells.53,54 Moreover, use of a Venus-YFP reporter in embryonic stem cells (ESCs) undergoing hematopoietic specification provided functional evidence that tissue-specific enhancers were associated with expression of genes in the same stage.55 A recent publication using bacterial methylation labeling showed coordinated enhancer and gene activity throughout enterocyte differentiation.56

The myeloid master regulator CEBPA offers a clear example of tissue-specific regulation, with different enhancers active in each tissue that expresses said gene, and complete absence of enhancer activity in tissues where CEBPA is silent (Figure 2).57 Other enhancers, however, are constitutively active. Thus, 2 broad classes can be distinguished: housekeeping or ubiquitous enhancers are active across tissues, whereas developmental or tissue-specific enhancers are restricted to specific cell types.43,58 Tissue specificity is the result of the recruitment of TFs and cofactors, which in turn depends on (a) the pool of TFs available in a particular cell type, and (b) the accessibility of their binding sites at a given enhancer.22 For instance, binding sites for the ETS, C/EBP, and NF-κB families are accessible in monocyte-specific enhancers, whereas neuronal enhancers are enriched for RFX and SOX proteins.43 Cooperative binding of TFs further narrows both tissue and genomic specificity of developmental enhancers.59 On the one hand, the requirement for simultaneous engagement of multiple TFBSs at enhancers prevents transcriptional noise due to spurious recognition of short motifs. On the other hand, it allows a finer control of transcriptional patterns during differentiation, as specific combinatorial patterns are uniquely expressed in specific cell types. Thus, in hematopoietic stem and progenitor cells (HSPCs), a heptad of TFs (TAL1, LYL1, LMO2, ERG, FLI1, GATA2, and RUNX1) frequently colocalize at CREs of key hematopoietic genes and act in concert to regulate their expression.60,61 At least 4 of these regulators establish protein-protein interactions that stabilize their DNA-binding and facilitate complex formation.

Figure 2.: Control of CEBPA expression by tissue-specific enhancers, adapted from. 57 The CEBPA locus contains multiple putative enhancers, identified here as peaks in H3K27ac ChIP-seq data (left side). The +42 and +34 elements are only active in blood, whereas the +55 peak is present in liver, adipose tissue, and fetal gastrointestinal tissue. Tracks are ranked by CEBPA expression in each tissue, shown on the right in RPKM. H3K27ac data from macrophages and neutrophils in PB were generated in-house; H3K27ac and RNA-seq data from other tissues were obtained from Roadmap.44 ChIP-seq = chromatin immunoprecipitation with sequencing; H3K27ac = histone H3 lysine 27 acetylation; PB = peripheral blood; RPKM = reads per kilobase per million.

Although binding of lineage-determining TFs (LDTFs) establishes the repertoire of tissue-specific enhancers in a cell, not all of them are immediately active. Some of them, known as inducible enhancers, require binding of additional TFs in response to internal or external signals.8,62 This type of enhancer is particularly common in plastic cell types that undergo phenotypic adaptations upon changes in the environment, like macrophages, T cells, or neutrophils. For example, macrophages stimulated with TLR4 ligands activate preexistent enhancers that control genes involved in inflammatory responses.63–65 During macrophage differentiation, these enhancers are primed by combinations of LDTFs, such as PU.1 or C/EBPβ,66,67 and become fully activated upon stimulation by TFs such as NF-κB, IRFs, and AP-1.63–65 Similarly, acquisition of a regulatory phenotype by CD4+ T cells following TCR stimulation largely results from activation of preestablished enhancers by newly attached FOXP3.68 Although most inducible enhancers seem to be previously primed during development, a fraction of them, known as latent enhancers, are created de novo upon reception of external stimuli.64,65,68 Those are roughly 10% of all enhancers induced during macrophage differentiation, but <1% in regulatory T cells. Inducible enhancers are critically dependent on cohesin.69

Anatomy of active and inactive enhancers

Enhancer states can be classified as inactive, primed, poised or active, each of which is associated with distinct epigenetic marks (Box 1; Figure 3). Inactive enhancers are located in compact chromatin and thus are inaccessible to TFs and cofactors, which results in lack of histone modifications. However, pioneer factors have the unique ability to strongly bind DNA wrapped around nucleosomes71–74 and recruit chromatin remodelers75,76 to make the region accessible to other TFs and epigenetic modifiers. Cirillo and colleagues first coined the term “pioneer factors” to describe FOXA (HNF3) and GATA4, after demonstrating they bind nucleosome arrays and open compacted chromatin,77 but multiple other LDTFs have a similar function in the hematopoietic system, including C/EBPβ,78 GATA1,79,80 and PU.1.66 These factors direct the selection of primed enhancers that are ready for further activation (see81 for a review on chromatin priming), exhibit reduced DNA methylation82 and are flanked by nucleosomes with lysine 4 monomethylation at histone 3 (H3K4me1).49,66,83

Figure 3.: Enhancer activation and decommissioning (adapted from 70 and 33). Pioneer factors mediate chromatin remodeling and make the region accessible to other TFs and epigenetic modifiers, turning inactive regions into primed enhancers marked by H3K4me1. Full activation entails recruitment of Pol II and histone acetylases that deposit H3K27ac marks. When enhancers are no longer needed, they can be decommissioned by enzymes that reverse these changes and render chromatin closed. H3K27ac = histone H3 lysine 27 acetylation; H3K4me1 = histone H3 lysine 4 monomethylation; Pol II = RNA polymerase II; TF = transcription factor.

Poised enhancers are a category of primed enhancers associated with lineage specification marked by both H3K4me1 and H3K27me3,84 which is associated with transcriptional silencing and is established by the polycomb repressive complex 2.85 The role of H3K4me1, primarily deposited by MLL3/4 methyltransferases,86 is uncertain, but it may contribute to increased responsiveness to activating signals49,87 and serve as a molecular memory of previous stimulation in the case of inducible enhancers.8,65 This mark is thought to be a key mechanism in the acquisition of inflammatory memory, or trained immunity, which depends on the opening of chromatin domains by TFs like AP-1.88,89 On the contrary, it is plausible that H3K4me1 prevents de novo DNA methylation at poised enhancers, as shown for a similar mark (H3K4me3) at bivalent promoters that also harbor H3K27me3.90

A primed enhancer becomes fully active upon binding of additional TFs and cofactors that further modify the epigenetic landscape.8,33 The histone acetyl transferases (HATs) CREB-binding protein (CBP) and p300 deposit acetylation marks like histone H3 lysine 27 acetylation (H3K27ac)91 and H3K9ac,92 which neutralize the positive charge of lysine residues, thereby decreasing their affinity for DNA and destabilizing the nucleosome to increase chromatin accessibility.93 Indirectly, acetyl groups act as docking sites for bromodomain-containing proteins such as the switch/sucrose non-fermentable (SWI/SNF) chromatin remodeler,94,95 leading to the displacement of nucleosomes and increased chromatin accessibility.96 Moreover, the acetylating activity of CBP/p30097–99 facilitates the recruitment of RNA pol II and GTFs at enhancers100,101 to initiate transcription. Analogously to gene promoters, this results in the production of enhancer RNAs (eRNAs),102 which are often bidirectional and whose biological function remains obscure (see Box 1). Elongation of these transcripts, but also of mRNA at cognate promoters, involves the enlistment of BRD4, which also recognizes H3K27ac, to release RNA pol II from proximal pausing.103–105 Another essential cofactor is Mediator, a large multisubunit complex that associates with enhancers106 to transmit regulatory signals to promoters and stimulate initiation of mRNA transcription.107,108

Finally, active enhancers can be decommissioned by a process that involves TF release, removal of active histone marks, loss of chromatin accessibility, and gain of DNA methylation.70

Altogether, active enhancers are characterized by a number of epigenetic features, including open chromatin109–111; clustered binding of TFs and cofactors such as p30083,112 or Mediator106; and enrichment for H3K27ac and H3K4me1 histone modifications at nearby nucleosomes,51,83,91 with comparatively low H3K4me3 levels (see Box 1 for details). Moreover, these flanking nucleosomes contain certain histone variants that destabilize them, facilitating displacement.51 Another hallmark of enhancers is the bidirectional production of eRNAs at levels that correlate with mRNA synthesis by their target genes.83,102 Although these characteristics have been defined through statistical associations, they have a direct function in enhancer biology, rather than being mere bystander effects.

Box 1:

Epigenetic features associated with enhancer function

Histones: These are small, positively charged proteins that can strongly bind the negatively charged backbone phosphates of DNA through electrostatic interactions.113 Histone proteins consist of a well-ordered globular core (histone fold) flanked by intrinsically disordered tail domains (histone tails).114

Histone posttranslational modifications (PTMs): Histone tail domains contain a large number of sites that can be target of PTMs, which modulate the charge of the tail and thus alter the electrostatic interactions supporting chromatin structure.115 The existence of histone tail PTMs has been known since 1964, when Vincent Allfrey showed that acetylation and methylation are incorporated after synthesis of the polypeptide chain.116 Despite the strong association between histone PTMs and gene expression, their role in transcriptional regulation may be more limited than originally thought. Experiments in drosophila revealed that gene activation occurs in the absence of H3K4 methylation117 and that point mutations in H3K27 only lead to a loss of repression, suggesting that acetylation mainly antagonizes H3K27me3.118

Histone code: the histone code hypothesis proposed by C. David Allis and Brian Strahl suggested that histone tail domains encode a language that could be read, written, or erased by specific proteins.119 Examples of readers are bromodomain-containing proteins that bind acetylated lysines,120 whereas HATs like p300 are writers that mediate acetylation.121 In contrast, histone deacetylases (HDACs) are erasers that removes acetylation.122 Another prediction of this hypothesis was that PTMs may be interdependent and act in combination, which has been confirmed by the integration of multiple chromatin marks into so-called chromatin states.123 These inferred functional associations are a result of specific recognition by reader proteins that contain motifs able to distinguish residues based on their methylated stated and surrounding sequence.

Histone variants: These are paralogues of the so-called canonical histones, with differences that can range from a few amino acids to 50% of their sequence.124 Some examples include H2A.Z and H3.3, both of which are associated with enhancers. While canonical histones assemble into nucleosomes behind the replication fork, variants are incorporated during the cell cycle, in a replication-independent manner.125 The replacement of canonical histones by their variants changes the properties of nucleosomes and their interaction with remodelers and other proteins, thus having an effect on gene expression.

Nucleosome-free regions: Nucleosome eviction or destabilization in nucleosome-free regions is a critical requirement for the binding of TFs to cis-regulatory elements and initiation of transcription.111 These accessible chromatin regions are susceptible to digestion by nucleases, and as such they are also known as DNase hypersensitive sites (DHS).126 Chromatin accessibility is facilitated by several processes, including the replacement of canonical histones with histone variants, the eviction or repositioning of histones by chromatin remodelers, and the covalent modification of histones.127

Enhancer-derived RNAs: eRNAs are generally bidirectional, unspliced, and nonpolyadenylated,102 although a recent study in single cells concluded that this bidirectionality is an artifact of bulk data.128 Three main models have been proposed to explain the role of eRNA in gene regulation, reviewed in more depth.33 First, both the transcription of enhancers and the resulting eRNAs are nonfunctional and merely a byproduct of high RNA pol II concentrations. Second, the act of transcription participates in the remodeling of chromatin, by carrying histone transferases or opening up chromatin, although the resulting eRNAs would be irrelevant. Third, eRNAs themselves have a function, such as the stabilization of enhancer-promoter looping, the binding of TFs, or the sequestration of transcriptional repressors. Although these different possibilities are not mutually exclusive, a recent study provided convincing evidence that, at least in some instances, enhancer transcripts are required for physical interaction between enhancers and promoters.129

Identification and validation of enhancers

Starting with the seminal article of Banerji et al in 1981, the first efforts to identify enhancers relied on reporter assays that exploited the capability of these elements to augment gene transcription, regardless of cellular context, distance, or orientation. While successful, this approach was limited by its low throughput, the inability to determine whether enhancers are active in vivo and in which cell types or tissues. Therefore, more scalable methods to detect novel enhancers have been developed and are currently in use.10 These can be broadly classified as follows:

Biochemical annotations: the biochemical features of active enhancers described in the previous section have been exploited to detect putative enhancers on the basis of annotations from various molecular biology techniques (Box 2). These include DNase I hypersensitivity sequencing (DNAse-seq) and assay for transposase accessible chromatin with sequencing (ATAC-seq) to measure open chromatin, and chromatin immunoprecipitation with sequencing (ChIP-seq) and cleavage under targets and release using nuclease (CUT&RUN) to assay for histone modifications and TF-binding. In particular, p300 binding has shown to be strongly predictive for tissue-specific enhancers.112 The ability to produce bidirectional transcripts has also been exploited to detect active putative enhancers.43,130 Among the various epigenetic marks, H3K27ac is the best predictor for validated active regulatory regions,131 although eRNA discriminates better between genes with high or low expression.132 Such methods have been employed extensively by consortia like ENCODE or Roadmap, on account of their exceptional scalability and their ability to measure enhancer-associated signal in their genomic context across multiple tissues. Nevertheless, the predictions upwards of 1 million putative enhancers contain numerous false positives, whereas enhancers characterized by atypical marks (such as H3K64ac or H3K122ac, located in the histone globular domains133) may be missed. Massively parallel reporter assays (MPRAs): validating predicted regions as bona fide enhancers requires functional characterization, proving they can indeed increase transcription from a reporter gene. This task can be accomplished using MPRAs, such as CRE analysis by sequencing (CRE-seq)134 or self-transcribing active regulatory region sequencing (STARR-seq).38 In CRE-seq, the putative enhancers are inserted upstream of a minimal promoter in barcoded plasmids, whereas in STARR-seq they are inserted in the 3′ UTR of the reporter gene, avoiding the need for barcodes. These techniques can be applied in an unbiased manner or in combination with biochemical annotations. Their main appeal is that they directly test for the intrinsic ability of an enhancer to increase expression, which constitutes the functional basis of their definition. On the other hand, these early MPRAs rely on a single promoter and are conducted outside the original cellular and genomic contexts, ignoring the influence of factors such as enhancer-promoter (E-P) compatibility, chromatin looping, or the available TF repertoire. In recent years, strategies that overcome some of these drawbacks have been developed. For example, in site-specific integration fluorescence-activated cell sorting followed by sequencing, putative enhancers coupled with a reporter are integrated into the HPRT locus of ESCs, ensuring a constant and accessible chromatin environment.55,135 Differentiating cells are next sorted based on reporter expression and candidate regions enriched in populations with high reporter signal are considered functionally active enhancers. Despite its advantages, this approach is still constrained by the use of a single promoter, and the dependency on ESC differentiation trajectories. Targeted genome editing screens: techniques based on clustered regularly interspaced short palindromic repeats (CRISPR) are a relative newcomer to this field, but their potential is becoming increasingly clear. In CRISPR-based screens, guide RNAs (gRNAs) targeted against a collection of enhancers are delivered to a pool of cells and those gRNAs associated with changes in the expression of genes of interest are identified.10 Targeted enhancers can be deleted with Cas9,136 repressed with dead Cas9 (dCas9) either on its own or fused to a repressor like KRAB (CRISPRi),137 or activated by dCas9 with a transcriptional activator, such as VP64 (CRISPRa).138 In most studies so far, the technique has been applied to a single gene, identifying gRNAs that are enriched in a fraction of cells with phenotypic changes related to that gene. It can also be applied to determine which specific regions of a putative enhancer region are essential for gene regulation, as shown for a MYB binding site in a relocated GATA2 enhancer.139 More recent approaches harness the power of single-cell (sc) sequencing to determine which enhancers are perturbed in each cell and their corresponding transcriptomic profiles.140,141 These have received various names, but can be collectively referred to as scCRISPR-seq.142 The main strength of CRISPR-based screens is that they assay for changes in the expression of genes while targeting their putative enhancers in their cellular context, thus overcoming shortcomings of the previous 2 approaches. On the negative side, they are expensive and can be hampered by the low efficiency of gRNAs at certain regions, and by the presence of shadow enhancers that mask the effect of targeting a single enhancer. Furthermore, they are only applicable to cells that grow in culture, of which not all are amenable to genetic manipulation. In vivo validation: only experiments in entire organisms can confirm the role of an enhancer in physiological conditions, not only in the control of gene expression, but also in cellular processes like differentiation. For example, deletion of an H3K27ac-marked region +42 kb downstream of CEBPA resulted in loss of CEBPA expression and neutrophil depletion, establishing such region as a bona fide enhancer critical for neutrophil commitment.57 Nevertheless, such experiments are very costly and limited to a single candidate, so they are exclusively employed for validation. Alternatively, expression quantitative trait loci (eQTLs) located inside enhancers may act as an indirect confirmation that such regions control gene expression in humans.

These methods have their own strengths and weaknesses (Table 1), so they are best used in combination. Biochemical annotations can be used to catalogue enhancers in a given cell type at a low cost, which can then be further validated with MPRAs or CRISPR screens, or more targeted experiments for only a few loci of interest. An emerging technology that may aid in these efforts is the prediction of enhancer activity on the basis of sequence features using deep learning, which has seen some success in Drosophila.143

Table 1 - Summary of Methods Used for the Identification of Enhancers Type of Approach Techniques Advantages Limitations Biochemical annotations ChIP-seq, CUT&RUN, ChIP-exo
DNAseq-seq, ATAC-seq Very scalable and reproducible
Detection of putative enhancers in their context across tissues Lack of functional validation leading to false positives
Constrained by the choice of epigenetic marks MPRAs CRE-seq, STARR-seq, SIF-seq Directly testing the functional definition of an enhancer
Possibility to synthesize tested regions Relatively expensive and complex
Observations are made out of the real cellular context (except SIF-seq)
Ignore enhancer-promoter compatibility Targeted genome editing screens CRISPR-Cas9, CRISPRi, CRISPRa, scCRISPR-seq Functional testing of enhancers in their genomic context
Inference of E-P assignment Expensive
Low efficiency of some gRNAs
Differences in applicability between cell systems In vivo validation Animal models, eQTL Understanding of the role of an enhancer in vivo