In late 2023, the world’s first CRISPR-Cas9-based gene editing therapy, Casgevy, was approved by drug regulators in the United Kingdom and by the U.S. Food and Drug Administration (FDA) as a potential cure for sickle cell anemia and β-thalassemia, representing a global milestone for modern medicine. These inherited hemolytic diseases are caused by distinct mutations in the b-globin gene (HBB), which encodes the β subunit of hemoglobin A (HbA), also known as adult hemoglobin. Whereas sickle cell disease is the consequence of a single missense mutation (E6V) in HBB, β-thalassemia-associated mutations down-regulate HBB expression, causing an imbalance in α and β chains that disrupts red blood cell generation. HBB is located in a gene cluster that also includes HBG1/2, encoding fetal γ-globin chains. Constituents of the β-globin gene cluster compete for access to a collection of strong upstream transcriptional enhancers, referred to as the locus control region (LCR), consisting of multiple DNase I hypersensitivity sites that recruit specific transcription factors (TFs) (Fig. 1) (1). Notably, although technically feasible with clustered regularly interspaced short palindromic repeats (CRISPR)-based tools, Casgevy does not actually correct any disease-associated HBB mutations. Instead, its mechanism of action involves the replenishment of fetal Hb (HbF) by causing down-regulation of B-cell Lymphoma 11A (BCL11A) that suppresses HBG1/2 expression. The specific target of Casgevy is the binding site for the GATA1 TF at an erythroid-specific BCL11A enhancer (2), which is disrupted by CRISPR-Cas9-mediated genome editing in isolated CD34+ hematopoietic stem cells (HSCs) ex vivo. Reintroduction of the genetically-modified HSCs by intravenous infusion confers robust therapeutic benefit in sickle cell disease and β-thalassemia, due to the elevated levels of HbF in patient red blood cells (3). Thus, Casgevy also represents an example of an effective enhancer-specific treatment, in which perturbation of enhancer-promoter transcriptional cooperation reinstates a separate promoter-promoter competition (Fig. 1A, B), and underscores the value of interrogating regulatory element interactions.

As exemplified by the human β-globin locus, appropriate control of gene expression in eukaryotes involves a complex interplay of cis-regulatory elements (CREs) and their protein components. RNA polymerase II (Pol II)-dependent utilization of CREs that are located proximal or distal to the transcription start site (TSS) is the basis of transcriptional regulation of gene expression during development, homeostasis, disease, and context-dependent responses (4). In addition, both promoters, the CREs that are closest to the TSS (typically < 1 kb), and remotely positioned (> 1 kb from TSS) CREs, or enhancers, engage particular TFs based on the presence of specific DNA motifs and harbor an assortment of transcriptional co-factors as well as architectural proteins that collectively contribute to physical and functional CRE interactions, which dictate lineage-determining and cell type-specific gene expression. In this review, we summarize recent insights into CRE cooperation and competition, including enhancer-promoter, enhancer-enhancer, and promoter-promoter interactions. We also discuss the involvement of TFs and co-factors in 3D genome architecture impacting CRE cooperation and competition and briefly consider the molecular mechanisms that facilitate regulatory element interplay in transcription.

Despite separation in the linear genome that is often substantial, cooperation between transcriptional enhancers and their target gene promoters orchestrates cell-type specific and spatiotemporal gene expression patterns (5). Physical interaction between enhancers and promoters is regarded as a fundamental aspect of enhancer-mediated transcriptional regulation. Methods based on expression quantitative trait loci (eQTL) that link genetic variants to gene transcript levels and high-throughput derivatives of chromatin conformation capture (3C) that identify physical contacts between genomic regions have been instrumental in identifying examples of enhancer-promoter interaction (EPI). While most EPIs lack functional validation, many investigations of specific loci have confirmed the regulatory significance of EPI-based looping events. Indeed, the role of looping in altering chromatin topology and in facilitating the assembly of local transcription hubs is well accepted; however, the detailed molecular mechanisms that contribute to EPI and their specificity remain largely elusive. We first outline the importance of EPI in allowing cooperative activation of target promoters as well as pertinent features of enhancer function from the point of view of enhancer transcription.

Contact between enhancers and promoters

Mammalian genomes consist of over a billion bp of DNA that can extend more than 2 m linearly, requiring substantial compaction to fit within the nucleus while retaining a highly organized arrangement that is amenable to regulation (6). 3D genome structure affects various cellular functions, including transcription, and is relatively stable but flexible. Next-generation sequencing (NGS) coupled with 3C-based methodologies and super-resolution imaging tools have allowed for the identification of some fundamental properties of the 3D genome, such as the polymeric nature of chromatin, topologically associated domains (TADs), and EPI dynamics (7). Proper chromatin looping between enhancers and promoters is crucial in development, as it controls cell fate commitment by licensing gene expression through regulatory element cooperation, whereas enhancer mutations that alter EPIs can adversely impact gene expression and lead to developmental abnormalities or diseases, referred to as enhanceropathies (5). Recent advances in molecular biology tools enabling forced looping events indicate that EPIs often have causative roles in enhancer-promoter cooperation rather than being a consequence of the collaboration; however, the converse may be true in some cases (8, 9).

Forced looping events involving enhancers and promoters can be achieved in eukaryotic model systems with genome engineering tools, such as synthetic zinc-fingers (ZFs) and CRISPR-Cas technology (Fig. 2) (10). At the β-globin gene locus, tethering the lim domain binding 1 (LDB1) dimerization domain to either the HBB promoter (11) or the promoters of the HBG1/2 genes (12) via a customized ZF is sufficient to drive expression in a selective fashion (Fig. 2A). This effect is analogous to the normal function of LDB1 in mediating long-distance LCR-promoter interactions as a component of cofactor complexes (Fig. 1C). The forced DNA looping approach also provided early experimental evidence that epigenetic manipulation could overcome the developmental silencing of γ-globin genes to significantly increase HbF levels in sickle cell disease (13). Subsequent application of CRISPR-based technologies to sequence-specific DNA looping has yielded consistent results and further insights. Utilization of orthogonal dCas9 proteins from Streptococcus pyogenes (Sp) and Streptococcus thermophilus (St) linked to separate heterodimerization domains allowed for the establishment of stable promoter-enhancer connections that supported gene upregulation in bacteria (14). Inducible DNA looping strategies, such as chromatin loop reorganization using CRISPR-dCas9 (CLOuD9) (15) and light-activated dynamic looping (LADL) (16), have been applied in mammalian cells, confirming that controlled modification of loop structure alters gene expression (Fig. 2B). While the former approach was only able to achieve modest gene upregulation in its initial version (15), the latter showed a more robust effect on gene expression concomitant to forced DNA looping that was reversible upon ligand washout, at least following short-term induction (16). Coupling of bioorthogonal chemistry with CRISPR technology has enabled the manipulation of chromatin contacts involving multiple genomic loci, permitting the investigation of transcriptional hubs (Fig. 2C) (17). The bioorthogonal reaction-mediated programmable chromatin loop (BPCL) system employs single-guide RNAs (sgRNAs) outfitted with a unique sequence that is specifically recognized by a single-stranded clickable oligonucleotide adaptor (ssCOA). The ssCOAs also feature a photocleavable linker. Thus, the association of two genomic loci that are targeted by the dCas9/sgRNA complex can be stabilized by bioorthogonal adaptor ligation and dissociated by light illumination. This technique was used to induce the interaction of pluripotency gene promoter and enhancer elements either as independent enhancer-promoter pairs or as multi-component chromatin hubs, demonstrating stable but reversible contacts and consistent, albeit modest, effects on gene expression (17). Collectively, these studies involving forced chromatin looping between enhancers and promoters have substantiated the role of EPI in target gene expression and provide methods for future exploration of the regulatory mechanisms that underlie transcriptional activation (10).

Delivery of basal transcriptional components from enhancers to promoters

It has been proposed that enhancers recruit general transcription factors (GTFs) as well as Pol II and then deliver the machinery, collectively referred to as the preinitiation complex (PIC), to their cognate promoters by looping-mediated physical interaction (18-20). However, there is currently only limited evidence to support this transfer model of eukaryotic gene transcription. To better understand enhancer-promoter cooperation, biochemical analysis of the requisite components and the molecular mechanisms controlling the spatiotemporal kinetics of EPIs and associated transcriptional activation would be particularly instructive (21). Nevertheless, high resolution microscopy approaches and transcriptional assays have already afforded some insights. By visualizing PIC assembly at the single-molecule level, a study employing yeast nuclear extracts and the upstream activating sequence (UAS)-promoter system uncovered the dynamic behavior of GTFs/Pol II and some aspects of the early sequence of molecular events at these regulatory elements (22). This investigation demonstrated that Pol II is recruited to the UAS/enhancer along with GTFs and subsequently transferred as a mostly pre-assembled PIC to the core promoter, suggesting sequential enhancer-promoter transcriptional activation. Similarly, combined cryoelectron microscopy (cryo-EM) and cryoelectron tomography (cryo-ET) analyses of yeast PICs assembled on divergent promoters indicated that an incomplete PIC initially forms on the UAS/enhancer before transfer to the core promoter and concomitant recruitment of the remaining GTFs (23). However, these in vitro studies did not explicitly determine the relationship of PIC assembly dynamics to transcriptional activation at the specific regulatory elements. In the human MCF7 breast cancer cell model, we showed that ligand-induced enhancers are activated temporally before promoters (24). In addition, we observed more than three-fold higher enhancer transcription by PRO-seq upon target promoter deletion, likely due to the accumulation of transcriptional machinery at the enhancer, supporting the notion, albeit indirectly, that the capacity for transcriptional activation is transferred from an enhancer to its cognate promoter(s) in a sequential fashion.

eRNA

A principal mechanism by which enhancers confer transcriptional activity to their target promoter(s) through looping involves transcriptional events at enhancers themselves. Contemporary genomic approaches have revealed that mammalian genomes are pervasively transcribed, producing a vast array of non-coding RNA (ncRNA) species that include transcripts originating from enhancers, or enhancer RNAs (eRNAs) (25, 26). Transcribed from functional enhancers with activating epigenetic marks, eRNAs are typically short, bidirectional, non-polyadenylated, and unstable (27). Accumulating evidence suggests that eRNA transcripts play functional roles in enhancer-promoter cooperation. Depletion of eRNA transcripts by various methods attenuates target gene expression, although the mechanistic basis for this effect remains a subject of debate (28-30). As eRNAs show heterogeneity in sequence composition, length, secondary structure, and directionality, generalizations regarding the functional intent of tens of thousands of enhancers known to transcribe eRNAs are somewhat equivocal. Nevertheless, molecular mechanism studies of eRNA function have identified at least three, non-mutually exclusive roles impacting either cooperation or competition between regulatory elements that involves their protein constituents: i) facilitating and/or stabilizing EPIs by interaction with looping factors, such as cohesin (29), CTCF (31), and Mediator (32, 33); ii) promoting recruitment of TFs and cofactors, including YY1 (34), BRD4 (35), and CBP/p300 (36, 37); and iii) enabling the release of paused Pol II by associating with the negative elongation factor (NELF) complex to allow transcriptional elongation (38, 39). Interrogation of eRNA complexes by mass spectrometry has revealed a large number of interacting partners relevant to gene expression, including proteins involved in chromatin modification, DNA looping, RNA processing, and transcription regulation (40). For example, over 30 high-confidence interactors have been reported for Bloodlinc eRNA, which is synthesized from an erythroid-specific super-enhancer (41). Given the diversity of potential protein partners, eRNAs may serve as scaffolds, baits, or even decoys. The interaction of eRNAs with multiple proteins might be explained, at least in part, by the concept of liquid-liquid phase separation (LLPS), a biophysical process involved in the assembly of membraneless organelles (25, 42) that may also apply to EPI regions harboring eRNA. In this regard, eRNAs modified co-transcriptionally with m6A (N6-methyladenosine) were shown to promote biomolecular condensate formation in a m6A reader protein-dependent manner (43). Whether other functions ascribed to m6A-eRNAs, including effects on DNA methylation, histone modification, chromatin state, and downstream gene transcription, actually involve an LLPS-related mechanism is unclear (44-46).

At promoters, eRNAs have been implicated in PIC recruitment and the eventual transition from paused transcription to elongation. Transcription initiation enabled by the promoter-bound PIC is often halted after the synthesis of a short, nascent RNA. The latter regulatory step is mediated by the pausing components DRB sensitivity-inducing factor (DSIF) and NELF. In mouse primary neurons, membrane depolarization-induced eRNAs facilitate the release of NELF from paused Pol II at a set of neuronal genes, thereby allowing Pol II to engage in productive transcriptional elongation (38). NELF interacts with eRNAs via its RNA-recognition motif (RRM) motif, which also binds nascent mRNAs. Accordingly, competition between eRNAs and nascent mRNAs for NELF binding has emerged as a mechanism for pause release. The ability of an eRNA to detach NELF from paused Pol II seems to be neither sequence nor structure dependent. Rather, biochemical assays have indicated that transcript length (200 nt threshold) and guanosine content are important determinants of the NELF-dissociating capacity of eRNAs (39). Furthermore, eRNA-mediated transcriptional release may also be affected by LLPS, although direct evidence is lacking. Treatment of cells with the aliphatic alcohol 1,6-hexanediol (1,6-HD) that disrupts many cellular membraneless organelles (47) resulted in a transcript profile consistent with promoter-proximal pausing of Pol II, characterized by reduced sequence tag density over gene bodies and an accumulation of reads at TSSs, as determined by PRO-seq (48).

Phase separation in transcription

LLPS is now appreciated as a common biological mechanism responsible for all membraneless organelles in eukaryotic cells, thereby impacting various cellular processes. Although biomolecular condensates formed by phase separation are also thought to be an important feature of eukaryotic gene transcription, definitive evidence for their involvement remains elusive, largely due to limitations in tools for specific and selective perturbation (49). The phase separation of proteins is concentration dependent and can be modulated by various environmental or experimental parameters, including temperature (50), ionic strength (51), and pH (52). Both physical and chemical properties of nucleic acids impact their capacity for phase separation (53). Biomolecular condensates support locally enhanced catalytic activities due to the enrichment of enzymes and substrates (54). Many RNA- and chromatin-binding proteins, including TFs, cofactors, and epigenetic regulators with intrinsically disordered regions (IDRs), have been shown to undergo phase transition in biochemical experiments and/or in cultured cells. The liquid de-mixing of IDR-containing proteins is dynamically modulated by RNAs, which alter multivalent interactions (43). Phase separation is favored by poorly translated mRNAs, and many long non-coding RNAs (lncRNAs) have been reported to enhance the generation of phase-separated condensates and/or impact their dynamics (55, 56). eRNAs, which represent one class of lncRNAs, interact with numerous proteins that are involved in transcription. Because eRNAs lack conserved sequences and many eRNA-binding protein partners, such as CBP and BRD4, can associate with a broad spectrum of RNAs, eRNA-protein interactions are likely non-sequence specific (57). Thus, eRNAs are a key feature of the biomolecular condensates proposed to form at interacting regulatory elements and transcriptional hubs. eRNAs have a high m6A content, which was documented by methylation-inscribed nascent transcript sequencing (MINT-seq) (43). m6A-modified eRNAs may further promote phase separation at actively transcribing enhancers by recruiting the m6A-reader YTHDC1. In addition, m6A-modified eRNAs might augment DNA demethylation and chromatin accessibility at proximal or interacting genomic regions via an LLPS-mediated mechanism involving the m6A-reader FXR1 and the DNA 5-methylcytosine dioxygenase TET1, as has been reported for some sites that harbor m6A transcripts (58). Whereas low levels of RNA generated upon transcription initiation at regulatory elements, such as eRNAs, promote condensate formation, elevated RNA production during transcription elongation results in condensate dissolution (59). This finding indicates that LLPS may be involved not only in eRNA-mediated target gene activation but also in transcriptional suppression by feedback regulation.

Clusters of enhancers that control highly expressed, cell-type-specific genes, originally referred to as LCRs, were discovered more than 30 years ago (60). The human β-globin LCR, which drives sequential transcriptional activation of the γ-globin and β-globin genes during erythroid development (Fig. 1), is a well-studied example (61). Located 6 to 20 kb away from the β-globin gene cluster, the LCR consists of 5 DNase I hypersensitivity sites (HSs) that contain TF binding motifs for NF-E2 and GATA1 among others (Fig. 1C) (62). Upon binding these LCR sites, GATA1 interacts with CBP to promote histone acetylation and also recruits p45/NF-E2 to form a chromatin loop, both of which are necessary to drive γ-globin genes expression (63). The switch from γ-globin to β-globin involves the collaboration of BCL11A, SOX6, and GATA1, which interact with the LCR to alter chromatin loop architecture in the β-globin gene cluster (61). The developmental function of each HS in the LCR is unique and may not be essential. HS2 maintains equal expression of γ-globin and β-globin during development and may contribute to the formation of the β-globin TAD by conferring an active chromatin structure (i.e., high H3K27ac) on neighboring CTCF sites (64, 65). HS3 drives robust fetal expression of the γ-globin gene but is seemingly decommissioned subsequently, as its deletion does not significantly decrease activity of the β-globin gene in adults (64, 66). HS4 has no impact on globin gene transcript levels during embryonic erythropoiesis; however, it promotes γ-globin and β-globin gene expression during definitive erythropoiesis within the fetal liver and also contributes to activation of the β-globin gene in adult red blood cells (67). In contrast, HS1 appears to play a minor role that has a negligible effect on globin gene activity throughout development (64). While deletion of the entire mouse β-globin LCR causes a dramatic reduction in β-globin transcript levels, the chromatin accessibility of the β-globin gene cluster and its developmentally-specific expression pattern are unaffected, indicating that these regulatory features are separable (68).

The property of the LCR in which individual enhancers are functionally distinct but can collaborate with other constituent enhancers as a single regulatory unit is now appreciated as broadly relevant to eukaryotic transcription control (69). Enhancer clusters re-identified in the past decade by largely arbitrary bioinformatic thresholds applied to genome-wide histone modification or TF/co-factor binding data have been dubbed super-enhancers or stretch enhancers, with the former terminology gaining the most traction in the gene transcription field (70, 71). Super-enhancers (SEs) are large clusters of enhancers with high TF density and robust gene regulatory capacity. Several algorithms, such as Rank Ordering of Super-Enhancers (ROSE), are widely used to identify SEs by detecting disproportionately high levels of TFs and activating histone modifications based on ChIP-seq data (72, 73). Compared to typical enhancers, SEs tend to span larger genomic regions, show increased enrichment of H3K4me1/2 as well as H3K27Ac, and contain more Pol II, TFs, mediator complex and other cofactors, histone acetyltransferases p300 and CBP, and chromatin architectural proteins such as cohesin. Collectively, these features confer a greater capacity to activate transcription locally, in the form of eRNA, and at target gene promoters (74). The dense accumulation of transcriptional machinery at SEs has distinct biophysical properties consistent with phase separation (75). Many SE-regulated genes show cell-type specificity, and Gene Ontology indicates that they are predominantly associated with biological processes pertaining to the identity of the corresponding cell and tissue (76). SEs also contribute to transcriptional dysregulation in the pathogenesis of various diseases (77). In particular, cancer-specific SEs, which can arise from mutational events, serve as important drivers of oncogene expression to promote each stage of cancer development (78). Despite a general appreciation of their distinct properties and (patho)physiological significance, major conceptual questions remain largely unresolved, such as whether individual, constituent enhancers have independent roles and if there is a synergistic effect of the clustered arrangement. Extensive studies of the β-globin LCR, which qualifies as a SE, have provided strong evidence for cooperation between several enhancers in gene expression control albeit with specific roles for some. Genetic manipulation of other SEs has further demonstrated that the different functional contributions of constituent enhancers can be position dependent (79).

The quantification of EPIs and promoter-promoter interactions (PPIs) has been facilitated by an improved derivative of Hi-C, dubbed Micro-C, which employs micrococcal nuclease (MNase) to achieve individual nucleosome resolution (147 bp) (80). Although substantially less than the number of EPIs (> 20,000), a large amount of PPIs (> 7,000) were detected in mouse embryonic stem cells (mESCs), suggesting their importance as another mechanism of transcriptional regulation. In addition, multiple promoters can compete for a single enhancer. This regulatory strategy, called promoter competition or enhancer retargeting, seems to occur predominately with promoters and enhancers residing in the same TAD (81, 82). Following the discovery of promoter competition at the chicken globin gene cluster (83), various other examples of this phenomenon, involving Hox genes, ANT-C in Drosophila, human globin genes, and the oncogene MYC, have been described (3, 84-86).

The developmental expression pattern of human γ-globin and β-globin is the result of promoter competition for the upstream LCR. As noted above, restoring HbF production in adult erythroid cells represents a viable treatment option for inherited hemoglobinopathies, such as β-thalassemia and sickle cell disease (3, 87). Evaluation of the genomic alterations in heterozygous individuals diagnosed with hereditary persistence of fetal hemoglobin (HPFH) or mild forms of β-thalassemia (δβ-thalassemia and β0-thalassemia) exhibiting elevated HbF revealed a common deletion region, which spanned the promoter and the first exon of the HBB gene (88). A series of CRISPR-Cas9-mediated genomic edits at the HBB promoter demonstrated that increased HBG1/2 transcript levels can be attributed to disruption of the HBB promoter as opposed to the gene body, and the increment of HBG1/2 upregulation generally correlated with the extent to which HBB expression was compromised. Furthermore, the contact frequency between the HBG1/2 promoters and the LCR, as measured by Capture-C experiments, was significantly increased in the HBB promoter-deleted cells, indicative of enhancer retargeting (88). Thus, in theory, alternative CRISPR-Cas9 genomic perturbations could achieve the same therapeutic endpoint as Casgevy, eliciting upregulation of HbF, but by a distinct mechanism of action.

The human α-globin locus is also subject to promoter competition involving a distinct LCR. The former features three genes encoding functional α-like proteins, including embryonically expressed HBZ (ξ-globin) and a set of duplicated fetal/adult-specific genes, HBA1 and HBA2 (α1- and α2-globin) (89). Like the β-globin genes, the α-globin transcription units are arranged in the order of their sequential activation during development. Among the cluster of CREs that constitute the upstream LCR, HS-40, which is also known as MCS-R2, serves as the major enhancer driving gene expression at the α-globin locus (90). Genomic variants in an intergenic region between the HBA2 promoter and the HS-40 enhancer can create a de novo promoter-like element that competitively down-regulates the α-globin genes (HBA1 and HBA2), thereby causing α-thalassemia (90). The HS-40 enhancer not only regulates globin-related gene expression but also that of other genes in the same TAD, including NPRL3, which harbors HS-40 as an intronic enhancer, and NME4, located ∼300 kb away. NPRL3 and NME4 can be activated by EPI with HS-40 in erythroid cells, while the other intervening non-globin genes at the locus do not show cell type-specific differences in expression (91). Notably, the absence of one or more α-globin genes including their promoter sequences, as observed in some cases of α-thalassemia, results in NME4 upregulation due to increased interaction of the latter with HS-40 (91).

Consistent with these earlier examples of promoter competition, recent studies have substantiated the capacity of mutations that functionally compromise a preferred promoter to release its cognate enhancer. This process, termed enhancer release and retargeting (ERR), allows the liberated enhancer to engage and activate an alternative promoter in the same TAD, thereby impacting transcriptional activity (24). In MCF7 cells, deletion of the TFF1 promoter, which is the preferred target of the TFF1 enhancer (TFF1e), causes a looping event between TFF1e and a different target promoter, that of the TFF3 gene located 50 kb away. This retargeting results in a >20-fold increase in TFF3 expression at the transcriptional level and may explain the correlation of cancer-associated single nucleotide polymorphisms (SNPs) in the TFF1 promoter with elevated TFF3 expression (24, 92).

Activation of other proto-oncogenes via the ERR mechanism has been demonstrated at several genomic loci. The promoters for MYC and the lncRNA PVT1 compete for interaction with intragenic enhancers located in the PVT1 gene. Silencing of the PVT1 promoter causes MYC transcriptional activation and a concomitant increase in the proliferation of breast cancer cell lines (24, 86). In diffuse large B cell lymphoma (DLBCL) cells, a shared SE controls expression of ZCCHC7 and PAX5, which is negatively correlated. Higher expression of ZCCHC7 has been observed in DLBCL cells harboring a PAX5 promoter mutation, and, similarly, genetic deletion of the PAX5 promoter causes upregulation of ZCCHC7, which may promote lymphoma progression by modulating ribosome biogenesis (93). In B-cell acute lymphocytic leukemia (B-ALL), two different types of recurrent deletions that induce enhancer retargeting have been detected at 13q12.2, which is the site of the FLT3/PAN3 locus. Type I deletions that are situated upstream of FLT3 and include the PAN3 promoter are associated with elevated FLT3 expression. In the absence of the PAN3 promoter, the intragenic PAN3 enhancer alternatively interacts with and activates the FLT3 promoter (94, 95). Type II deletions include the promoter region and exon 1 of the FLT3 gene as well as the PAN3 promoter, resulting in downregulation of these mutated genes and simultaneous upregulation of CDX2 that is located ∼30 kb downstream of FLT3. H3K27ac HiChIP data, which reveals genome-wide interactions among active CREs, has confirmed long-distance looping between the PAN3 intragenic enhancer and the CDX2 promoter in B-ALL cells with Type II deletions (94). This observation of secondary retargeting of the PAN3 enhancer to CDX2 is consistent with a result from genetically modified MCF7 cells harboring deletions of both the TFF3 promoter and the TFF1 promoter. The double mutant MCF7 cells display elevated transcription of TFF2, another gene in the same TAD, suggesting that enhancers can be released from an alternative target gene to retarget still other functional promoters in the neighborhood (24). In gastric cancer, ERR events defined by ENCODE and GTEx eQTL data were identified. Analysis of cis-eQTLs in promoters (+/−3 kb of TSS) and H3K27ac-marked enhancers allowed for the classification of 853 ERR candidate events that affect gene pairs in the same chromosomal region (+/−200 kb) (96). Similar to the proposed mechanism at the TFF1 locus, CTCF promoter binding plays a critical role in determining enhancer-promoter specificity for the identified ERR gene pair CAPN10 and GPR35. An upstream enhancer engages and activates GPR35 upon mutation of the CTCF-binding site in the promoter of its neighboring gene, CAPN10 (96).

Accumulating evidence indicates that the ERR mechanism is not limited to cancer genomes or mammalian gene expression systems. An example of ERR has been reported in mouse mesoderm development in which knockout of either Mesp1 or Mesp2 results in upregulation of the other due to retargeting of a shared enhancer (97). Examination of rare promoter de novo variations (DNVs) has revealed a possible role for ERR in autism spectrum disorder (ASD) (98). Genome-wide analysis of DNV distribution demonstrated their enrichment in TADs containing ASD risk genes. The DNVs are often located in the promoters of non-ASD risk genes and seem to affect not only the expression of those genes but also that of other genes in the same TAD, including ASD risk genes. As an example of ERR in plants, the mechanism was implicated in a study probing tomato GLYCOALKALOID METABOLISM (GAME) gene regulation involving a distal CRE, GAME Enhancer 1 (GE1) (99). In Drosophila, an ERR-like mechanism may explain features of transvection, a phenomenon involving EPI between two homologous chromosomes. In this context, several cases have been documented in which deletion of a promoter for an allele on the same homolog as its enhancer leads to an increase in the promoter activity of the allele on the other chromosome (100).

Promoter competition and enhancer retargeting can be facilitated by epigenetic silencing in addition to genomic alterations. As exemplified at the TFF1 and MYC loci, enhancer retargeting can be prompted by application of CRISPR-based epigenome-editing technologies, such as dCas9-KRAB-mediated promoter inhibition (CRISPRi) (24, 86). Therefore, the functional validation of enhancer retargeting does not require laborious and low-throughput genome editing. In addition, while CRISPRi tools can be applied to individual loci in focused studies, they are amenable to large-scale interrogation of CREs (101, 102). Accordingly, genome-wide functional screening with CRISPRi promoter targeting should help to identify many additional examples of enhancer retargeting-dependent gene regulation. Furthermore, CRISPRi-based perturbation of promoters harboring GWAS-linked SNPs coupled with single-cell RNA sequencing may allow for the high-throughput discovery of causal variants that contribute to disease-related transcriptional changes due to effects that extend beyond the promoters in which they reside (103).

Elucidation of cooperation and competition between regulatory elements of transcriptional units is of fundamental importance to understanding gene expression. Since the discovery of the first enhancer element in simian virus 40 several decades ago, the interplay of CREs, especially enhancer-promoter cooperation, has been extensively studied, yielding many insights that collectively evince the complexity of metazoan transcriptional regulation (104). Although much about enhancer-promoter cooperation remains elusive, it is now clear that physical contact between enhancers and promoters, eRNA-mediated recruitment of regulatory proteins, and release of paused RNA pol II are key mechanisms of target gene activation. The initial evidence of enhancer-enhancer cooperation was provided by detailed characterization of the β-globin LCR, and the concept has re-emerged prominently in recent years in the form of super-enhancers that can be identified bioinformatically based on ChIP-seq data. Phase separation has been implicated in both enhancer-promoter and enhancer-enhancer cooperation, but formal proof of its contribution to any aspect of Pol II-mediated transcriptional regulation is lacking. Promoter-promoter competition was also originally described in studies of different globin gene clusters and their LCRs. The concept has recently been extended to the ERR mechanism that is involved in the aberrant activation of disease-associated genes.

An appreciation of CRE cooperation and competition in transcription affords promising opportunities for therapeutic intervention. Current epigenetic drugs that can affect CREs, such as DNA methyltransferase (DNMT) inhibitors, have limited specificity that restricts their clinical utility (4). Also, while the approval of the enhancer-targeting drug Casgevy demonstrates the potential of CRE-focused gene therapy, CRISPR-Cas9-based genomic alterations pose an intrinsic risk, even in ex vivo applications, due to potential off-target genomic effects. Therefore, further development of durable, tunable, and sequence-specific precision therapies that also lack an inherent capacity to introduce DNA breaks is needed (105).

Recent advances in epigenome editing technologies for programmable modulation of disease-related gene transcription have yielded compelling preclinical results. Lipid nanoparticle delivery of mRNA encoding a transcriptional repressor comprising a zinc-finger protein (ZFP) DNA-binding domain linked to epigenetic effector domains (KRAB and DNMTs) conferred robust and long-lasting silencing of the Pcsk9 gene in mice. The protein encoded by Pcsk9 controls cholesterol levels in the bloodstream, and thus effective Pcsk9 knockdown resulted in a substantial reduction in circulating cholesterol (106). The fusion protein targeted a CpG island (CGI), a DNA element sensitive to methylation, in the Pcsk9 promoter. Given that CGIs are present in most promoters as well as a subset of highly active enhancers and can prompt their physical and functional interplay (107), it is possible that this type of hit-and-run (108) epigenetic silencing strategy could be broadly employed to modulate disease-associated transcription related to enhancer-promoter and enhancer-enhancer cooperation or promoter-promoter competition. Induced proximity through dimeric small molecules is another epigenetic approach with therapeutic potential. In the case of DLBCL, a dimeric small molecule that binds both the TF BCL6 and the epigenetic activator BRD4 can act as a chemical inducer of proximity (CIP) to activate transcription of pro-apoptotic BCL6 target genes (109), resulting in cancer cell death. Alternative configurations are possible for other disease contexts. Finally, eRNAs could prove to be useful targets for decommissioning active enhancers that enable deleterious gene expression in various diseases (110). In theory, effective antisense oligonucleotides and siRNA therapeutics targeting eRNAs could suppress particular disease-associated genes by disturbing enhancer-promoter and/or enhancer-enhancer cooperation with limited collateral effects due to the high tissue and cell-type specificity of eRNAs.

Despite recent insights, many questions regarding CRE interplay are still unresolved. How the specificity of EPIs is achieved remains a major conundrum, even though several mechanisms have been proposed (104). It is also largely unclear how the individual, constituent enhancers of a SE are coordinated within the broader regulatory circuit, and what determines their redundancy or synergy. Another matter of uncertainty is how frequently promoters compete for enhancers and what are the key determinants of this competition. By addressing such lingering questions pertaining to the detailed mechanisms of gene expression, new principles of enhancer biology and transcriptional regulation will emerge that can inform novel therapeutic strategies.

Soohwan Oh is supported by a Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (2022R1C1C1010699) and the Ministry of Science and ICT (MSIT), Korea, under the Information Technology Research Center (ITRC) support program (IITP-2024-RS-2023-00258971), supervised by the Institute for Information & Communications Technology Planning & Evaluation (IITP). The figures were created with BioRender.

The authors have no conflicting interests.