Superenhancers as master gene regulators and novel therapeutic targets in brain tumors

Transcriptional deregulation, a cancer cell hallmark, is driven by epigenetic abnormalities in the majority of brain tumors, including adult glioblastoma and pediatric brain tumors. Epigenetic abnormalities can activate epigenetic regulatory elements to regulate the expression of oncogenes. Superenhancers (SEs), identified as novel epigenetic regulatory elements, are clusters of enhancers with cell-type specificity that can drive the aberrant transcription of oncogenes and promote tumor initiation and progression. As gene regulators, SEs are involved in tumorigenesis in a variety of tumors, including brain tumors. SEs are susceptible to inhibition by their key components, such as bromodomain protein 4 and cyclin-dependent kinase 7, providing new opportunities for antitumor therapy. In this review, we summarized the characteristics and identification, unique organizational structures, and activation mechanisms of SEs in tumors, as well as the clinical applications related to SEs in tumor therapy and prognostication. Based on a review of the literature, we discussed the relationship between SEs and different brain tumors and potential therapeutic targets, focusing on glioblastoma.


INTRODUCTION
Brain tumors, including glioblastoma, medulloblastoma, diffuse intrinsic pontine glioma, meningioma, ependymoma, and other rare brain tumors, account for approximately 3% of cancer cases worldwide 1 . While brain tumors are relatively rare, they still deserve research attention because of their significant mortality and morbidity at all ages 2 . Currently, treatments for brain tumors include surgery, chemotherapy, and radiotherapy. Despite substantial advances in chemotherapy for different brain tumors, chemotherapy resistance and relapse are still challenges. Delivering therapeutic agents to primary brain tumors is particularly challenging because of the unique brain-blood barrier (BBB) and brain-tumor barrier (BTB), resulting in poor response to treatments 3,4 . Moreover, brain tumor cells breach the BBB and adhere to the BTB, and this diffuse mode of invasion is the fundamental reason for brain metastases and relapse 4 . Therefore, it is necessary to study pathogenesis further and find more effective targets to improve the prognosis of patients with brain tumors.
Gene transcription is a complex and highly coordinated process. Transcriptional dysregulation mediated by epigenetic modifications in tumors has attracted significant attention. Over the last decades, epigenetic modifications such as DNA methylation and histone modification have been identified as critical drivers of several types of brain cancer 5 . Enhancers are a class of regulatory DNA sequences that function as cis-regulatory elements to enhance the transcription of target genes occupied by coactivators and transcription factors (TFs) 6,7 . With the expanding concept of enhancers, "superenhancers (SEs)" have been proposed, which span several kilobases and are enriched with a higher density of TFs, coactivators, and epigenetic modifications 8 . Compared to typical enhancers, SEs can drive higher gene expression and participate in many biological processes 9 . It is worth mentioning that oncogenes regulated by SEs in tumor cells are not expressed in normal cells, suggesting that SEs play a critical role in the occurrence and development of tumors 10 . As tumor-associated variants are significantly enriched in SEs, identifying SEs improves the understanding of the mechanisms of tumorigenesis and provides new insights into the diagnosis, treatment, and prognosis of tumors 11 . Overall, SE-driven transcriptional disorders are associated with the progression of human cancers, including brain tumors. This review will summarize the characteristics and identification, unique organizational structures, and activation mechanisms of SEs in tumors, as well as the clinical applications related to SEs in tumor therapy and prognostication. Furthermore, we will discuss SE-associated genes, transcriptional regulatory mechanisms, and therapeutic targets in different brain tumors.

OVERVIEW OF SUPERENHANCERS Characteristics and identification of SEs
In 2013, Young et al. first proposed superenhancers: large clusters of transcriptional enhancers that drive the expression of cellidentity genes 10 . SEs were initially discovered in embryonic stem cells (ESCs), with ESC-specific master transcription factors Oct4, Sox2, and Nanog binding to enhancer elements and recruiting mediators to activate the gene transcription program 12 . Subsequently, SEs were identified in other cells, including cancer cells 13,14 . As novel epigenetic regulatory elements, SEs precisely regulate oncogene transcriptional activation during tumorigenesis and may be potential therapeutic targets 15,16 .
SEs are bound by many cell-type-specific TFs and recruit a series of mediators, transcriptional coactivators, and chromatin regulators (CBP/p300 and cohesin) as well as RNA polymerase II (RNA pol II), thus forming an SE-promoter loop to initiate downstream transcription 17 . In addition, histone modifications are important features of SEs. Histone H3 lysine 27 acetylation (H3K27ac) and histone H3 lysine 4 methylation (H3K4me1), which make the chromatin structure looser and provide accessible TF binding sites, are labels of active SEs 18,19 . H3K27ac and H3K4me1 are required for enhancers to activate target gene transcription, and these active chromatin markers can mediate the recruitment of epigenetic readers, including BET proteins 20,21 . In conclusion, these key components are integral parts of SE organization and the functional basis of SEs (Fig. 1).
The role of SE-derived RNAs Enhancers regulate the transcription of target genes and are actively transcribed into enhancer RNAs (eRNAs), and this ability to be transcribed is a general feature of functionally active enhancers 41 . Correspondingly, SEs are also transcribed into seRNAs, including long noncoding RNAs (lncRNAs), microRNAs (miRNAs), and circular RNAs (circRNAs), which play an important role in gene expression and epigenetic regulation [42][43][44] .
A growing number of studies have revealed the critical roles of seRNAs in regulating target gene transcription. Young's group proposed an RNA-mediated feedback control model of transcriptional condensates. In the model, low levels of RNAs mediate the formation of transcriptional condensates, while high levels of RNAs cause condensate dissolution, thus affecting the transcriptional output 45 . The seRNAs act synergistically with SEs to regulate gene expression by maintaining the SE-promoter chromatin loop 46 . For example, the seRNA HPSE interacts with hnRNPU and p300 and then facilitates chromatin looping between the SE and HPSE promoter to enhance HPSE expression, thus promoting cancer progression 47 .
Although the role of seRNAs in transcriptional regulation in both cis and trans is clear, the molecular mechanism is still controversial. Three models of seRNA mechanisms have been reported: interacting with TFs or coactivators, driving enhancerpromoter looping, and transferring to the cytoplasm to mediate various cell activities 48 . While most studies have demonstrated that seRNAs promote enhancer-promoter looping, Panigrahi et al. showed that seRNAs do not affect chromatin looping 49 . They concluded that mutual costimulation of enhancer and promoter transcription is not dependent on seRNAs 49 . Further studies are required to explain these conflicting observations regarding the molecular mechanism of seRNAs.
SeRNAs that are generated from genetic alterations or somatic mutations of SEs during tumorigenesis and are termed oncogenic seRNAs 46 (Fig. 1). Oncogenic seRNAs are significantly upregulated in several types of tumors and associated with carcinogenic processes, including cell proliferation, apoptosis, autophagy, and epithelial-mesenchymal transition (EMT). In hepatocellular carcinoma (HCC), HCCL5, an SE-driven lncRNA, is significantly overexpressed and promotes cell growth, invasion, and metastasis 50 .  1 Schematic diagram of the structure and characteristics of SEs. SEs recruit TFs, mediators, RNA pol II, histone modifiers, and other chromatin regulators, activating the expression of downstream genes. The enhancer-promoter loop that is formed with the help of seRNAs contributes to the transcription of target genes. Blocking SEs with BRD4 or CDK7 inhibitors is considered a viable antitumor approach. SE superenhancer, TF transcription factors, BRD4 bromodomain protein 4, CDK7 cyclin-dependent kinase 7.
Recently, the SE-lncRNA FASRL has been identified in HCC, and FASRL can increase fatty acid synthesis and lipid accumulation, thereby exacerbating HCC progression 51 . The novel SE-lncRNA LOC100506178 was found to promote nasopharyngeal carcinoma metastasis by interacting with the TF hnRNPK and controlling the expression of hnRNPK, accelerating the EMT process 52 . In addition, SEs can promote the transcription of pri-miRNAs and recruitment of the Drosha/Dgcr8 complex, thereby increasing the expression of cell-specific miRNAs 43 . In HCC, the YY1/p65/p300 complex increases QK1 expression through interaction with the SEpromoter loop, thus promoting HCC malignancy and increasing circRNA formation and EMT 53 . SEs usually contain discrete loci with seRNA expression peaks, and seRNA signals in cancer samples have been identified to have clinical relevance 54 . Although some studies have reported a correlation between SEderived ncRNAs and prognostic indicators, most were metaanalyses, and thus, further clinical validation is required 55 .

Unique organizational structures of SEs
Studies on SEs consider not only the linear structure of the genome but also the three-dimensional (3D) structure of the genome. In the eukaryotic genome, DNA-protein complexes fold 3D chromatin loops, called topologically associating domains (TADs), which are structural units for transcriptional regulation 56 . SEs can be enriched in the chromatin loop inside TADs, and SEcontaining TADs have more chromatin interactions 57 . TADs insulate promoters from enhancers and superenhancers, and the disruption of TADs alters regulatory circuits and leads to oncogene activation 58,59 . The architectural protein CTCF interacting with the cohesin complex defines the boundary for TADs and SEs, and this interaction is specifically required for the formation of chromatin loops, allowing orderly gene regulation 58,60,61 . In T-cell acute lymphoblastic leukemia, loss of CTCF induces a TAD fusion event, which leads to a direct interaction between the MYC promoter and a distal SE, thus activating MYC 62 . Especially in the absence of cohesin, SEs tend to form links with each other, resulting in extensive SE fusion 63 . Therefore, an accurate understanding of the 3D genome chromatin structure is essential for SE-mediated transcriptional regulation.
Liquid-liquid phase separation (LLPS), a dynamic physicochemical process, forms a membraneless organelle termed a condensate to organize biological processes within cells 64 . In 2017, the proposed phase separation model limited SE-specific gene regulation to the membraneless organelle 65 . The intrinsically disordered regions (IDRs) of TFs, the transcriptional coactivators bromodomain protein 4 (BRD4) and mediator complex subunit 1 (MED1), and RNA Pol II can form phase-separated condensates at SE regions, leading to the transcriptional bursting and simultaneous activation of SE-driven genes 66,67 . The phase-separated condensate can separate SEs from other chromatin regions and concentrate SE-associated transcriptional processes at key regions, which explains why the effect of transcription regulation of SEs is greater than the overall effect of individual typical enhancers 66 . In leukemias, NUP98-HOXA9, a NUP98 fusion oncoprotein containing IDRs, can form LLPS condensates. LLPS of NUP98-HOXA9 is critical for leukemogenesis because it not only induces CTCFindependent chromatin loops but also leads to the formation of a "superenhancer"-like binding pattern at leukemogenic genes to regulate transcriptional activity 68,69 . The pharmacological inhibition of LLPS condensates can suppress metastasis and chemoresistance in osteosarcoma, representing a novel therapeutic strategy 70 . In addition, seRNAs also contribute to the organization of the phase-separated condensate and play a role in transcriptional activation 71 . Current studies on phase separation mainly focus on the role of biomolecular condensates assembled in tumors. However, such studies have not considered the intrinsic mechanism of the dynamic phase separation process, which still needs further investigation. Therefore, the phase separation model, which can ensure the precise regulation of genes, provides a novel perspective to elucidate the formation and transcriptional regulatory mechanisms of SEs (Fig. 2).
The gene transcription process requires a variety of transcriptional activators at specific DNA regulatory elements. To coordinate transcription programs in normal and malignant cells, a small group of master TFs forms an interconnected core regulatory circuitry (CRC) by directly co-occupying their SEs and each other's SEs, allowing close contact between SEs and their target promoters 8,72 . The activities of SEs and TFs affect each other. On the one hand, SEs coordinate with TFs to regulate the gene expression program, and the activity of SEs is affected by TF enrichment. On the other hand, the expression of TFs is often regulated by the activity of SEs, indicating positive feedback regulation between SEs and TFs 8 . The interconnected autoregulatory loop containing SEs can regulate specific cell-type transcriptional programs. To date, the CRC model and associated master TFs have been identified in multiple cancer types, including esophageal cancer 72 , acute myeloid leukemia 73 , multiple myeloma 74 , and osteosarcoma 70 . Because of the important role of CRC in malignant tumors, targeting the core TFs may suppress tumor growth.

Functional activation of SEs during tumorigenesis
Cancer cells proactively construct SEs via mechanisms such as genetic mutation, single-nucleotide polymorphism (SNP), chromosomal rearrangement, and viral infection 11 (Fig. 3). During tumorigenesis, the functional activation of SEs leads to the dysregulation of transcriptional programs, making tumors highly dependent on gene expression regulators 75 .
Genetic mutations lead to SE activation by creating new binding sites for TFs, changing SE copy numbers, and changing the spatial structure of the genome. In T-cell acute lymphoblastic leukemia cells, there is a 12 bp insertion upstream of the TAL1 oncogene, which generates an SE region through the binding of MYC and the recruitment of CBP and other core TFs, driving the aberrant expression of key oncogenes 76 (Fig. 3). In addition to insertion, focal amplification of SEs is a common mechanism of SE activation. For instance, two focal amplifications of SEs located on the 3' side of MYC in endometrial carcinoma and lung cancer are close to the MYC promoter region and are associated with aberrant expression of MYC 77 .
Some specific tumor-related SNPs in regulatory elements are correlated with the activity of SEs. The SNP rs2168101 has been demonstrated to affect neuroblastoma susceptibility; it is located in the LMO1 superenhancer and changes a GATA-binding motif into a TATA motif, thus inhibiting the activity of the SE and decreasing the expression of LMO1 78 (Fig. 3). Similarly, in diffuse large B-cell lymphoma, two SNPs (rs9831894 and rs6773363) of susceptibility loci have been identified and are a part of a tumorspecific SE 79 .
It has been shown that structural variations can disrupt 3D genome organization and active SE regions, known as "superenhancer hijacking", which has been reported in neuroblastoma, leukemia, and colorectal cancer [80][81][82][83][84][85] . Superenhancer hijacking has been described as a novel mechanism for tumorigenesis in which chromosomal rearrangements cause enhancers to transfer to nearby oncogenes, thereby leading to high expression of oncogenes and the tumor initiation 82 . In colorectal cancer, IGF2 is a target of enhancer hijacking, which is mediated by the formation of a contact domain comprising IGF2 and a lineagespecific superenhancer 82 (Fig. 3). In addition, Wang et al. developed a computational framework called NeoLoopFinder to predict enhancer hijacking by identifying the chromatin interactions induced by structural variations 80 . They found that enhancer hijacking-driven oncogenes such as MYC, ETV1, PVT1, and CDK12 are significantly upregulated in cancer cells. Viral infections have also been shown to induce SE formation and drive aberrant transcription of genes involved in the initiation and progression of tumors. Epstein-Barr virus (EBV), human papilloma virus (HPV), human T-cell leukemia virus (HTLV), and human hepatitis B virus (HBV) are common oncogenic viruses, and their capacity to activate SE-associated genes has been reported in several tumors. HTLV-1 is associated with adult T-cell leukemia/lymphoma (ATLL) and specifically encodes the TF HBZ. HBZ binds to ATLL-specific BATF3 superenhancers, leading to the expression of BATF3 and other downstream genes and promoting the proliferation of ATLL cells 86 (Fig. 3). In HPV-infected cervical cancer, integration of the viral genome into the host genome generates a superenhancer that drives the overexpression of viral E6 and E7, promoting the growth of cancer cells 87 .

CLINICAL APPLICATIONS RELATED TO SES IN CANCER Small-molecule inhibitors targeting SEs for tumor therapy
The high transcriptional activity of superenhancers mediates transcriptional addiction in aggressive tumors, making oncogenes extremely susceptible to transcriptional alterations 88 . Thus, targeting SE complexes with small-molecule inhibitors is a promising strategy for tumor therapy (Fig. 1). Unlike genetic mutations, epigenetic alterations are reversible, and many drugs targeting epigenetic regulators have exhibited therapeutic potential in clinical trials 89 . The sensitivity of SEs to emerging smallmolecule inhibitors has been confirmed in various tumors, suggesting that SEs might be antitumor targets 90 . Some drugs have entered clinical trials (Table 1), with the hope that they can be applied clinically. BRD4, a member of the bromodomain and extraterminal domain protein (BET) family, plays a role in SE organization and oncogene expression regulation 91 . Mechanistically, BRD4 binds to acetylated lysines in enhancers, SEs, and TFs, bringing them together and mediating transcriptional activation and elongation via RNA pol II and mediators 92,93 . Inhibition of BRD4 disrupts the communication between SEs and their target promoters, resulting in subsequent repression of oncogenes 91 . KDM6A, a histone demethylase, is frequently mutated, which promotes tumorigenesis. KDM6A loss was found to regulate aberrant activation of SEs of oncogenes, ultimately leading to pancreatic cancer development. KDM6A-deficient pancreatic cancer is sensitive to BET inhibitors (such as JQ1 and I-BET151), which can decrease the expression of SE-associated genes and suppress tumor growth 94 . Another BET inhibitor, OTX015, is effective against mouse and human MYCN-driven neuroblastoma in models, as it can selectively disrupt the binding of BRD4 and SEs and lead to the repression of MYCN expression 95 .
In addition, cyclin-dependent protein kinases (CDKs) are protein-serine/threonine kinases that play an essential role in regulating the cell cycle and transcription 96 . During transcriptional activation, BRD4 binds to SEs, followed by the recruitment of the TFIIH/CDK7 initiation complex and P-TEFb elongation complex containing CDK9 75 . Cyclin-dependent kinase 7 (CDK7), an important CDK, can phosphorylate the RNA pol II C-terminal domain (CTD) at serine 5 (Ser5) and Ser7, leading to transcriptional initiation 97 . CDK7 also phosphorylates and activates CDK9/cyclin T, and CDK9 increases the phosphorylation of RNA pol II at Ser2, thus promoting transcriptional elongation 98 . Therefore, given that CDK7 and CDK9 are directly or indirectly involved in cellular transcriptional regulation, inhibition of CDK7 and CDK9 may interfere with transcription 99 . Covalent CDK7 inhibitors (such as THZ1 and THZ2) result in the downregulation of oncogene transcription and can serve as SE blockers to inhibit the expression of SE-driven oncogenes 100 . THZ2 is a newly developed CDK7 inhibitor with a fivefold longer half-life than THZ1, and its  105,106 . The notable antitumor effects of JQ1 and THZ1 have been demonstrated in multiple types of tumors, including triple-negative breast cancer, diffuse large B-cell lymphoma, and pancreatic cancer 88,107,108 . Simultaneous inhibition of BRD4 and CDK9 with JQ1 and LDC067 suppresses cell growth and migration in medulloblastoma 109,110 . The coinhibition of these two molecules also shows a similar antiproliferative effect in malignant rhabdoid tumors 109,110 . Combination therapy with JQ1 and the novel CDK7 inhibitor YKL-5-124 shows a synergistic effect in neuroblastoma and can delay resistance to BRD4 inhibition 111 . In addition, combination of a histone deacetylase (HDAC) inhibitor and JQ1 or OTX015 results in stronger repression of oncogenes and higher expression of tumor suppressor genes, suggesting that these epigenetic drugs have synergistic antitumor effects 112 . Similarly, when THZ1 is combined with an HDAC inhibitor, sensitivity notably increases 105,113 . Given the potential therapeutic effects of small-molecule inhibitors, rational combinational strategies may be more effective for cancer treatment.
Despite significant inhibitory effects in previous studies, some side effects of BRD4 or CDK7 inhibitors have been reported in animal experiments and phase I clinical trials, such as heart toxicity, gastrointestinal toxicity, fatigue, thrombocytopenia, and impairment of muscle function [114][115][116] . Furthermore, drug resistance is a challenge related to small-molecule inhibitors. In MYCNdriven neuroblastoma cells, upregulation of the multidrug transporters ABCB1 and ABCG2 results in resistance to THZ1 117 . In tumors with resistance to BRD4 inhibitors, hyperphosphorylation of BRD4 and epigenetic plasticity contribute to decreased sensitivity of JQ1 118,119 . The safety and efficacy of these smallmolecule inhibitors, including potential side effects, off-target effects, and drug resistance, need to be further studied before clinical application.

Utility of SEs in tumor prognostication
SEs can also serve as biomarkers for evaluating the prognosis of tumor patients. In HCC, the SE-associated lncRNA HCCL5 regulates malignant biological behavior and is associated with the prognosis of patients with HCC 50 . Further analysis revealed that the aberrant SE landscape in HCC is the result of extensive reprogramming. The key components of SEs (BRD4, CDK7, p300, and MED1) are overexpressed in HCC, which is associated with the poor prognosis of patients 120 . In addition, SEs with broad and high H3K27ac signals have been identified in nasopharyngeal carcinoma and were found to be associated with overexpression of the ETV6 oncogene and poor prognosis 121 . Xu et al. defined an SEassociated gene risk signature to predict the response to chemotherapy in patients with diffuse large B-cell lymphoma, which may help clinicians make more appropriate treatment decisions 122 . Moreover, in our previous studies, we constructed prognostic models based on SE-associated genes to predict overall survival for osteosarcoma and multiple myeloma patients, which may be helpful for clinical treatment 123,124 . Therefore, oncogenic SEs may promote the malignancy of tumors and have utility in predicting the clinical outcome of tumor patients.

THE ONCOGENIC ROLES AND REGULATORY MECHANISMS OF SES IN BRAIN TUMORS SEs in glioblastoma
Glioblastoma (GBM), a high-grade glioma (World Health Organization grade IV), is the most common malignant brain tumor and has a 5-year survival rate of 5.6% 125 . Despite multimodal treatment, including surgical resection, radiation, and standard therapy with temozolomide (TMZ), the prognosis is universally poor due to therapy resistance and recurrence [126][127][128] . Therefore, it is urgent to explore the molecular mechanisms underlying GBM progression and develop novel therapeutic strategies to improve the prognosis of GBM patients.

Identification of SEs in GBM.
In the past few years, several novel epigenetic markers contributing to the pathogenesis of GBM have been reported. In 2014, a population-based single-cell wholegenome sequencing methodology was applied to characterize genomic heterogeneity in EGFR-amplified GBM 129 . Researchers identified a translocation of a superenhancer to the 5' promoter region of TERT, which can activate TERT 129 . Moreover, another study described 3D genome information through Hi-C sequencing in glioblastoma stem cells (GSCs) 130 . In GSCs, genomic structural variants lead to SE-promoter interactions, such as physical interactions between the two SEs (SE1 and SE2) and JAK1 130 . In addition, integration of Hi-C and chromatin data revealed strong H3K27ac signals in stemness genes and identified a region overlapping SOX2 as an SE locus 130 . Currently, the majority of biopsies are preserved as archived formalin-fixed paraffinembedded (FFPE) samples. Zhao et al. assessed FFPE tissue with antibody-guided chromatin tagmentation with sequencing (FACTseq), the first highly sensitive method to describe histone modifications in FFPE tissues 131 . Using FACT-seq of H3K27ac in FFPE human GBM samples, the researchers identified 492 diseasespecific SEs (in genes such as EGFR, ETV1, and CDK6) 131 . In summary, the development of novel technologies allows the identification of complete SE landscapes and histone modifications in the whole genome, which aids the investigation of the epigenetic regulation and heterogeneity of GBM.

SE-mediated aberrant transcriptional programs.
Research focused on SE-driven aberrant transcriptional programs in GBM is gradually increasing. Since CDK7 can phosphorylate RNA Pol II to initiate transcription, CDK7 expression is significantly increased in GBM and is associated with poor prognosis 98,132 . The CDK7 inhibitor THZ1 can disrupt global gene transcription and preferentially target SE-associated genes in GBM cells. For example, five highly expressed SE-associated genes (WNT7B, FOSL1, FOXL1, ZMIZ1, and PHC2) were found to be associated with sensitivity to THZ1, and their knockdown inhibited the proliferation of GBM cells (Fig. 4a) 132 . In addition, GBM cells harbor a superenhancer in the Mcl-1 locus, leading to an increased level of Mcl-1, a member of the antiapoptotic Bcl-2 family of proteins 133 . Pharmacological inhibition of the SE by THZ1 decreased Mcl-1 mRNA and protein levels. A similar study identified the SE-derived lncRNA TMEM44-AS1, which was upregulated and correlated with malignant phenotypes in GBM 134 . When Myc directly binds to the promoter and SE of TMEM44-AS1 and colocalizes with MED1, H3K27ac, and RNA pol II, TMEM44-AS1 is activated 134 . The Myc inhibitor Myci975 can inhibit the growth of GBM cells by inhibiting the Myc/TMEM44-AS1 feedback loop (Fig. 4b) 134 . Together, these findings suggest that SE-mediated aberrant transcriptional programs are associated with GBM progression, and targeting these oncogenic SEs may be an effective therapeutic strategy for GBM.
SE-driven mechanisms of chemoresistance. Resistance to TMZ occurs in almost all patients during chemotherapy, leading to a poor prognosis. High expression of O6-methylguanine-DNA methyltransferase (MGMT) is one of the leading causes of TMZ resistance because MGMT is a DNA repair factor that can reverse DNA damage caused by TMZ 135 . MGMT expression is negatively associated with methylation modification in the MGMT promoter region 136 . Thus, the higher the methylation degree of the MGMT gene promoter region in GBM patients is, the better the effect of TMZ. Apart from methylation, MGMT expression can be regulated by other factors. It was found that the enhancer (K-M enhancer) located between the Ki67 and MGMT genes can increase MGMT expression and is activated in TMZ-resistant GBM tissues 137 . Deleting the K-M enhancer reduces MGMT and Ki67 expression, thus increasing sensitivity to TMZ 137 . Moreover, another study demonstrated that aberrant transcriptional regulation of SEs is a cause of resistance to TMZ 138 . The H3K27ac status regulates cisregulatory elements such as SEs. However, the RET finger proteinhistone deacetylase 1 (RFP-HDAC1) complex deacetylates H3K27ac marks on core histones, decreasing gene transcription levels 138 . Disruption of the HDAC complex can increase TMZ sensitivity in GBM cells by affecting the H3K27ac status ( Fig.  4c) 138,139 . In the future, more in-depth studies are needed to investigate whether the epigenetic regulatory mechanism underlying TMZ resistance is mediated by histone acetylation and which SEs drive transcriptional reprogramming. GBM resistance to TMZ may be related to abnormal SE activity, providing new ideas for overcoming GBM drug resistance.
HDACs regulate SE activity. Multiple lines of evidence have reported that the malignant phenotype of GBM is regulated by SEs and HDACs. HDAC1, together with BRD4, RNA pol II, and other key components, is enriched in SEs 10 . SLC30A3 is significantly reduced in GBM tissues and can inhibit the growth and metastasis of GBM cells 140 . Overexpression of HDAC1 decreases the H3K27ac level in the SE region of SLC30A3, which inhibits the expression of SLC30A3 140 . HDAC1 is predominantly associated with transcriptional repression, while the transcriptional regulation mechanism of HDAC7 is complex and related to both activation and repression. In stem-like breast cancer cells, HDAC1 and HDAC3 regulate the expression of HDAC7 141 . In turn, HDAC7 maintains H3K27ac levels by binding to SEs, thereby promoting the expression of SE-associated oncogenes 141 . HDAC1/3 knockdown increases the overall H3K27ac level, but HDAC7 knockdown decreases the H3K27ac level, and these changes are only observed in cancer stem cells and not in other cells 141 . These results suggest functional differences between HDAC1/3 and HDAC7. Inhibition of HDAC1/2/3 disrupts SE activity by increasing the accessibility of TF sites and disrupting chromatin loops, an indirect effect 105,142-144 . In GBM models, the HDAC inhibitors panobinostat and romidepsin have been shown to target SEs and elicit metabolic reprogramming through H3K27ac modification, thereby suppressing GBM progression 143,145 . Overall, different types of HDACs have different regulatory effects on SE activity. The molecular mechanisms underlying HDAC inhibitormediated SE activity are not yet fully understood. Some studies have investigated the regulatory mechanisms of HDAC inhibitors: (1) reshaping epigenetic markers by promoting H3K27 acetylation 144,146,147 ; (2) decreasing the binding of RNA pol II at SE regions [143][144][145] ; (3) inducing SE looping defects by reducing contact with SE components and promoters 142 ; and (4) reducing the expression of core regulatory circuitry TFs to repress SE activity 72 . These epigenetic alterations cause the global disruption of SEs and thus inhibit SE-associated transcript and gene expression 143,144 . Based on their antitumor effects, HDAC inhibitors should be able to exert synergistic effects in combination with SE inhibitors (such as THZ1 and JQ1) and chemotherapies in neuroblastoma, glioblastoma, and other tumors 72,105,112,113,146 .
SE-mediated core regulatory circuitry. SE-mediated core regulatory circuitry plays a crucial role in human cancers, including medulloblastoma, neuroblastoma, and glioblastoma 8,[148][149][150] . An active chromatin landscape was mapped in GSCs, identifying SEassociated genes and core TFs that establish SEs and maintain GSC identity 151 . The study found that many genes defining GSC identity are highly expressed and regulated by SEs, such as SOX2, SEPT9, CXXC5, CDK6, SALL3, and EGFR, which are required for tumor cell proliferation 151 . In addition, another study depicted the core TFs and their interconnected regulatory circuitry in GBMs, showing that many motifs of SE-driven TFs (SOX2, JUN, and RFX2) are enriched in SE regions (Fig. 4d) 152 . The stem cell TF SOX2 can bind to the ELOVL2 superenhancer in GSCs to drive the expression of ELOVL2, which is critical for the maintenance of cancer stem cells 153 . Furthermore, nine SE-associated lncRNAs were identified in GBM tissues, and LINC01094 is one of the lncRNAs whose disease-specific expression is regulated by tumor-associated core TFs and SEs 152 . SEs and the core regulatory circuitry in GBM cooperatively regulate transcriptional programs, which is significant for GBM progression.
Roles of seRNAs in gliomas. Some seRNA-regulated immunerelated genes (IRGs) relevant in the tumor microenvironment were identified in glioma 154 , and they were used to construct a prognostic model that can predict the survival outcomes of patients with glioma. These glioma-specific seRNAs that regulate IRGs are located in adjacent sites and coexpressed in tumors, suggesting that they have potential as therapeutic targets 154,155 .
Promising antitumor effects of targeting BET proteins. The BET inhibitor I-BET151 inhibits the proliferation and eliminates the tumorigenicity of GBM cells [156][157][158] . The antiproliferative activity is equivalent to that of TMZ, suggesting that I-BET151 may be beneficial for GBM patients with TMZ resistance 156 . In addition, another BET inhibitor, OTX015, was found to cross the BBB and selectively penetrate tumor tissues in a preclinical study, indicating its potential efficacy in GBM 159 . In addition, dBET6, a chemical degrader of BET proteins, can inhibit the proliferation and selfrenewal ability of GBM cells by significantly reducing BET protein occupancy, RNA pol II activity, and active histone markers, and dBET6 is superior to first-generation BRD4 inhibitors such as JQ1 160,161 . Interestingly, GBM cells exhibit sex-specific responses to the BRD4 inhibitor JQ1, which is correlated with sex-specific gene expression patterns. The sensitivity of male patient-derived GBM cells to BET inhibitors is higher than that of female patient-derived GBM cells 162 .
As newly recognized epigenetic elements, numerous SEs have been identified in GBM and have been shown to be related to tumor progression ( Table 2). As described above, novel epigenetic mechanisms that regulate the malignant biological behaviors of GBM cells have been revealed, providing potential treatments. In the future, targeting SEs may be a promising complement to traditional therapies for GBM treatment.
SEs in pediatric brain tumors SEs in medulloblastoma. Medulloblastoma (MB) is the most common malignant brain tumor in children and adolescents younger than 15 years old 163 . Previous genomic studies have revealed four molecular subgroups of MB (WNT, sonic-hedgehog (SHH), Group 3, and Group 4) with different biological and clinical behaviors 164,165 . Among these subgroups, Group 3 and Group 4 MBs account for the majority of cases and have the poorest outcomes 166 .
A study of MB illustrated that oncogenic drivers were restricted to Group 3 and Group 4 167 . The growth factor independent 1 family oncogenes GFI1 and GFI1B are oncogenes of Group 3 and Group 4 MBs that can cooperate with MYC to promote MB formation in vivo 167 . Genomic structural variants occur in the GFI1 or GFI1B coding sequences proximal to enhancers and superenhancers, leading to oncogenic activation of GFI1 and GFI1B 167 . In addition, scientists have identified some subgroup-specific SEs in MBs that can activate oncogenes, including ALK, MYC, SMO, and ETV4 148 . In Group 3 MB, an SE-driven transcriptional regulatory network consisting of 14 SE-associated genes exists 168 . A BET inhibitor (JQ1) and a CDK7 inhibitor (THZ1) showed synergistic inhibitory effects, with therapeutic potential for Group 3 MB 168 . Moreover, some adult SHH MB patients present with truncating mutations in the chromatin reader BRPF1, which are absent or rare in pediatric patients 169 . Mutated BRPF1 increases the accessibility of a subset of SEs associated with key genes involved in cerebellum development and chromatin remodeling, promoting tumorigenesis in adult SHH MB 170 .
Superenhancers not only activate oncogene expression but are also associated with tumor suppressor gene activation. Brainspecific knockout of the H3K4 methyltransferase MLL4 in mice can induce MB 171 . Mechanistically, MLL4 loss leads to MB by reducing broad H3K4me3 and SE signals, downregulating tumor suppressors (Dnmt3a and Bcl6), and extensively damaging epigenomic signatures 171 . Therefore, MLL4-established superenhancers play an important role in tumor suppression in normal cells. The antitumor role of SEs has also been reported in other types of tumors, such as chronic myelogenous leukemia and breast cancer 172 . Superenhancers can drive either oncogene expression or tumor suppressor expression, playing dual roles in tumor progression. SE-associated genes and possible mechanisms in MB are shown in Table 2.
SEs in diffuse intrinsic pontine glioma. Diffuse intrinsic pontine glioma (DIPG) is a fatal pediatric brain tumor with a median survival time of 11 months 173 . Because of its diffuse growth and location, DIPG is inoperable, and the standard therapy is radiation therapy, but its effectiveness is extremely limited 174 . The characterization and development of drugs are challenging due to the scarcity of DIPG samples.
Approximately 85% of DIPGs are characterized by mutation of lysine 27 to methionine in histone 3 (H3K27M), which leads to oncogenic transcription dysregulation and increased stem-like potential and proliferation 105,[173][174][175] . H3K27M glioma cells exhibit greater proliferation potential due to the aberrant oncogenic program 176 . Together with RNA pol II and H3K27ac, H3K27M localizes to transcriptionally active regions and can drive SE formation 177,178 . H3K27M inhibits the enzymatic activity of PCR2 through its interaction with the E2H2 subunit, resulting in the loss of H3K27me3 at SE regions 179,180 . H3K27M leads to a more accessible chromatin configuration in key regulatory regions, which can expose binding site motifs for key TFs, including ASCL1 and NEUROD1 181 . These TFs then bind to SEs to activate neurogenesis and NOTCH pathway-related genes, ultimately contributing to glioma formation 180,181 . Recent studies of H3K27M DIPG in vitro and in vivo have identified some important druggable targets and found some effective small molecules to reverse epigenetic alterations, including the HDAC inhibitor panobinostat 182 , the H3K27me2/3 inhibitor GSKJ4 182 , the BET inhibitor JQ1 178 , and the selective E2H2 inhibitor EPZ6438 183 .
The analysis of SEs in DIPG has revealed potential cell-identity genes, supporting the idea that DIPG originates from a precursor cell of the oligodendroglial lineage. These SE-associated genes include genes classically associated with oligodendrocyte precursor cells (such as SOX10) and genes expressed by oligodendroglial lineage cells during differentiation (such as OLIG1, MYRF, MYT1, and MBP) 105 . DIPG is vulnerable to BRD4 and CDK7 blockade, which can impair DIPG cell growth 105 . For example, the BRD4 inhibitor JQ1 can induce neuron-like differentiation and delay tumor growth in a mouse model of DIPG, and the CDK7 inhibitor THZ1 can disrupt transcription and inhibit DIPG growth 105,184 . Hyperacetylation in DIPG favors the action of BDR4 and leads to enhancer landscape reprogramming, activating SE-driven oncogenes in DIPG 174 . In addition, CBP is related to transcription activation, and its activity can be enhanced by BRD4 185 . The combination of JQ1 and a CBP inhibitor (ICG-001) can reverse the detrimental SE programs activated by BET or CBP 174 . Collectively, these findings suggest that SE-driven oncogenes in DIPG can be targeted with SE blockers. SEassociated genes and possible mechanisms in DIPG are shown in Table 2.

SEs
in other rare pediatric brain tumors. DNA 5-hydroxymethylcytosine (5hmC), one of the molecular alterations in GBM, can recruit DNA-binding proteins and is essential for GBM tumorigenesis 186 . 5hmC preferentially localizes to the enhancers and superenhancers of tumor-specific genes in glioblastoma, activating disease-specific gene expression programs 187 . 5hmC alteration is observed not only in adult brain tumors but also in pediatric brain tumors according to a recent study; 5hmC is located in TF binding sites and SE regions and is crucial to cell identity 187,188 . Thus, the epigenetic alterations of SEs in adult and pediatric brain tumors are somewhat similar, and 5hmC can be used to identify these aberrant regulatory elements.
Atypical rhabdoid/teratoid tumors (AT/RTs) are aggressive and lethal cancers that can be diagnosed at a young age 189 . The SWItch/Sucrose Non-Fermentable (SWI/SNF) chromatinremodeling complex is an essential regulator of pluripotency in human embryonic stem cells. Nevertheless, mutation of its core subunit SMARCB1 can lead to AT/RTs due to the disruption of enhancer accessibility 190 . With noticeable H3K27ac features in enhancer regions, SMARCB1 is required for the integrity of the SWI/SNF complex 191 , which is inactivated abnormally in most rhabdoid tumors 192 . Loss of SMARCB1 reduces genome-wide targeting at enhancers, thus impairing the functions of the SWI/ SNF complex. However, the small amount of residual SWI/SNF complexes preferentially bind to SEs, including some shared by all subtypes, such as SPRY1, and other lineage-specific SEs, such as SOX2, in brain-derived rhabdoid tumors, which is crucial for maintaining aberrant cell identity. Then, these SEs drive oncogenic transformation by locking cells into a poorly differentiated and highly proliferative state 191 . Further research found that knockdown of SMARCB1 prevents silencing in SE regions, thus leading to transcriptional upregulation in human embryonic stem cells 193 . Therefore, SMARCB1 and its regulatory effects on SEs in AT/RT are part of a novel AT/RT tumorigenesis mechanism.
Embryonal tumor with multilayered rosettes (ETMR) is a sporadic and difficult-to-treat brain tumor in infants and young children, with rapid progression and only 10-20% overall survival 194,195 . Because there are few research models, mechanistic and therapeutic studies of this rare disease are extremely limited. Recently, an oncogenic circuit driven by hijacked superenhancers in ETMRs was revealed. The C19MC-TTYH1 gene fusion and MYCN DNA interactions create superenhancers. Then, the interaction between C19MC-TTYH1 superenhancers and MYCN enhancers fortifies the C19MC-LIN28A-MYCN circuitry, driving the expression of embryonically restricted DNMT3B6 to promote a primitive malignant epigenetic state in ETMRs 196 . Interestingly, the BET inhibitor JQ1 can downregulate key components of the C19MC-LIN28A-MYCN circuit, including MYCN, LIN28A, and DNMT3B6, disrupting the circuit and inducing ETMR cell death 196 . The unique SE-dependent oncogenic circuit protects the ETMR and is vulnerable to BET inhibition. Therefore, inhibition of BET may be a promising therapeutic strategy for this orphan disease. SEassociated genes and possible mechanisms in these pediatric brain tumors are shown in Table 2.

SEs in other brain tumors
Meningioma is one of the most common intracranial tumors. Most meningioma patients can be cured by surgical resection, but approximately 20% of patients experience an aggressive clinical course with tumor recurrence or progression 197 . A comprehensive investigation of the genomic landscape has revealed the overall genomic instability in aggressive meningioma 198 . Upregulation of the SE-associated HOXD gene is associated with meningioma aggressiveness 198 (Table 2).
Ependymoma is a rare disease that can arise throughout neuraxis 199 . In children, ependymomas mainly occur intracranially, while in adults, the spine is the most common location of ependymomas 199 . The challenges of ependymomas are resistance to chemotherapy and lack of effective molecular targets. More than 60% of ependymomas harbor a ZFTA-RELA (ZFTA fus ) gene fusion 200 . ZFTA fus contributes to an oncogenic transcriptional program because it binds TF motifs and recruits transcriptional coactivators (BRD4, EP300, CBP, RNA pol II), thus driving SE gene expression in ependymoma 201 . To identify SE-associated genes that ependymoma cells depend on, Mack et al. analyzed the chromatin landscapes of ependymomas 202 . In two cohorts of ependymoma specimens, they identified that the vast majority of SEs were tumor-specific and enriched in oncogenes. Among the genes, EPHB2 and CCND1 have been previously proven to be ependymoma-related oncogenes 203,204 . In addition, ependymoma is sensitive to the BET inhibitor JQ1, which can inhibit the proliferation of ependymoma cells 202 (Table 2).

CONCLUDING REMARKS
The discovery of SEs is a novel breakthrough in the field of epigenetics. SEs are core regulatory elements that maintain the identity of cancer cells and drive cancer cells to become highly addicted to oncogene transcription. SEs regulate the expression of oncogenes that facilitate proliferation, migration, invasion, and even drug resistance, thus promoting tumor malignancy. With the accumulation of relevant research, the role of SEs in brain tumors is becoming increasingly clear. However, the composition of tumor-specific SEs and their potential molecular mechanisms remain to be further investigated.
Transcriptional activators (such as BRD4 and CDK7) are highly enriched in oncogenic SE regions, and their inhibition preferentially affects SE-associated genes in tumor cells. Some inhibitors targeting transcriptional activators have been evaluated in clinical trials. Moreover, combining SE inhibitors with other chemotherapeutic drugs can suppress tumor growth, providing a new strategy for cancer treatment. Since transcription is a basic biological process common to all nucleated cells, targeting SEassociated transcription may lead to general toxicity. Although inhibitors targeting SEs, including JQ1, THZ1, and THZ2, have been studied in many tumors, their potential side effects and off-target effects need further study.
As described above, the development of various brain tumors is closely associated with SEs and downstream oncogenes, and SEs may serve as master gene regulators and novel therapeutic targets. Further studies on the underlying mechanisms of SE activation will shed light on the complex pathogenesis of brain tumors. With persistent SE research efforts, breakthroughs in the treatment of malignant brain tumors are expected.