Targeting Super-Enhancers as a Therapeutic Strategy for Cancer Treatment

Super-enhancers (SEs) refer to large clusters of enhancers that drive gene expressions. Recent data has provided novel insights in elucidating the roles of SEs in many diseases, including cancer. Many mechanisms involved in tumorigenesis and progression, ranging from internal gene mutation and rearrangement to external damage and inducement, have been demonstrated to be highly associated with SEs. Moreover, translocation, formation, deletion, or duplication of SEs themselves could lead to tumor development. It has been reported that various oncogenic molecules and pathways are tightly regulated by SEs. Moreover, several clinical trials on novel SEs blockers, such as BET inhibitor and CDK7i, have indicated the potential roles of SEs in cancer therapy. In this review, we highlighted the underlying mechanism of action of SEs in cancer development and the corresponding novel potential therapeutic strategies. It is speculated that targeting SEs could complement the traditional approaches and lead to more effective treatment for cancer patients.


INTRODUCTION
The hallmarks of cancer, such as aberrant proliferation, invasion, metastasis, and apoptotic evasion, are closely related to aberrant gene expression (Hanahan and Weinberg, 2011). Therefore, genetic and epigenetic changes are fundamental mechanisms of cancer (Bradner et al., 2017). Promoters refer to sites to which the basal transcription machinery is recruited, usually located within 100-1,000 bp upstream of the transcription start sites (TSS). Since a promoter usually induces basal or limited levels of gene expression, higher levels of gene expression require highly regulated promoter-enhancer interactions (Carter et al., 2002).
Enhancers refer to transcription factors (TFs) that bind to DNA regulatory elements. They play key roles in the regulation of cell-type-specific gene expression, over both short and long distances, independent of their position and orientation with respect to TSS (Banerji et al., 1981;Bulger and Groudine, 2011). They exhibit three main characteristics. First, they often contain conserved DNA sequences and are located in open chromatin regions without nucleosomes, which allows for binding of RNA polymerase, TFs, and co-activators. Second, enhancers are typically enriched with a post-translational modification histone mark, such as acetylation at H3 lysine 27 (H3K27ac) and mono-methylation at H3 lysine 4 (H3K4me1). Third, unlike promoter sequences, enhancers can be located distantly from the TSS of their target genes (from less than 10 kb to more than 1 Mb) (Hnisz et al., 2013).
Super-enhancers (SEs) comprise of a set of enhancers spanning across a long range of genomic DNA, with some individual constituent enhancers exhibiting stronger transcriptional activation ability than others (Hnisz et al., 2013;Shin, 2018). SEs exhibit a similar mechanism of action as normal enhancers. Binding of TFs to enhancers facilitates enhancer interaction with the basal transcription machinery, RNA polymerase II, and promoters in a gene-specific manner, which is mediated by "looping" of the loaded enhancer to the cognate promoter. Then, the basal transcription machinery is recruited to promoters, which initiates downstream transcription (Sengupta and George, 2017).
In the past decade, increasing evidence has revealed that SEs play a vital role in tumorigenesis, indicating that SEs could be one of the promising therapeutic targets for cancer treatment. Indeed, BRD4, one of the bromodomain and extra-terminal domain (BET) protein family members, binds acetylated histones at TFs, TSS, and SEs, brings them together, and mediates transcriptional co-activation and elongation via RNA polymerase II and a mediator (Hajmirza et al., 2018). Their inhibition disrupts the communication between SEs and their target promoters along with a subsequent cell-specific-repression of oncogenes, which is considered to be the main mechanism of sensitivity to BET inhibitor (BETi) (Donati et al., 2018). Besides, CDK7 inhibitor, another kind of SE blocker, functions by inhibiting phosphorylation of RNA polymerase II (Nilson et al., 2015), and has been proved to significantly inhibit tumor growth (Figure 1). Here, we review the regulation and roles of SEs in various cancers to elucidate possible therapeutic targets for cancer treatment and provide potential future directions for the studies on SEs. Table 1 summarizes the regulation and roles of SEs in hematological malignancies.

SEs IN HEMATOLOGICAL MALIGNANCY
Several mechanisms of tumorigenesis in hematopoietic system have been proved to be associated with SEs, including mutation, fusion and expression of specific genes, activation of pathways, and infection of Epstein-Barr virus (EBV).
Several mechanisms of tumorigenesis in hematopoietic system have been proved to be associated with SEs, including mutation, fusion, and expression of specific genes, activation of pathways, and infection of EBV (Herranz et al., 2014;Kandaswamy et al., 2016;Katerndahl et al., 2017;Liau et al., 2017;Ma et al., 2017;Nitulescu et al., 2017). Generally, SEs were thought to promote tumorigenesis and malignancy, based on the reports that SEs upregulate oncogenes whereas broad H3K4me3 peaked at tumor suppressor genes (Hnisz et al., 2013;Chen et al., 2015). However, Cao et al. (2017) showed some overlaps in these two types of elements, suggesting that SEs could play both roles. Therefore, researchers should pay attention to the possible downregulation of suppressor genes when using BETi.
BET inhibitor is a hot topic in SE research. BETi, such as I-BET151 and JQ1, downregulated SE-associated genes, and suppressed proliferation and promoted apoptosis in AML multiple myeloma, acute leukemia, lymphoma, and primary effusion lymphoma (Loven et al., 2013;Dawson et al., 2014;Pelish et al., 2015;Gopalakrishnan et al., 2016). Besides, after determining dose-limiting toxicity of OTX015 in phase 1 clinical study, researchers reported that for further phase 2 studies, the once-daily recommended dose for oral, single agent use of OTX015 in patients with acute leukemia or lymphoma is 80 mg, on a 14 days on/7 days off schedule (Amorim et al., 2016;Berthon et al., 2016). Researchers also reported some new BETi, such as BI894999, and other SE inhibitors, including THZ1, NCD38, PLX51107, GSIs, and CBP30 Knoechel et al., 2015;Sugino et al., 2017;Garcia-Carpizo et al., 2018;Gerlach et al., 2018;Ott et al., 2018). These could provide novel alternatives or synergetic BETi drugs for cancer treatment. Besides, researchers have found some feedback regulations between SEs and corresponding genes. For example, STAT5 and PEPII promote the expression of corresponding SEs in B-cell acute lymphoblastic leukemia (Katerndahl et al., 2017), which indicated a better response to SE blockers in this cancer. Intriguingly, Zhang et al. (2017), Garcia-Carpizo et al. (2018) drew contradictory conclusions with each other about whether CREBBP, an acetyltransferase, promotes or suppresses cancer development. However, their conclusion came from different tumor models (diffuse large B cell lymphoma and follicular lymphoma for Zhang; leukemia and lymphoma for Garcia) and involved different downstream genes (BCL6, MEF2B, and MEF2C for Zhang; GATA1 and MYC for Garcia). Thus, the inconsistent results might be obtained due to the diverse functions of downstream genes controlled by corresponding SEs. Table 2 summarizes the regulation and roles of SEs in nervous system neoplasms.

SEs IN NERVOUS SYSTEM NEOPLASMS
In nervous system tumors, SEs could be regulated by gene rearrangement, single nucleotide polymorphism (SNP), binding of TFs, or modification of enzymes (Northcott et al., 2014;Valentijn et al., 2015;Dhar et al., 2018;He et al., 2018). Besides, SEs themselves could also be modified or translocated to new loci and exhibit different activities, leading to novel mechanisms for SEs-related tumorigenesis (Francis et al., 2014;Johnson et al., 2016). Furthermore, several downstream genes and related pathways of SEs have been discovered. Decaesteker et al. (2018) identified a novel core regulatory circuitry constituent (TBX2) in high-risk neuroblastoma, which was regulated by SEs. Dhar et al. (2018) reported that some SEs suppressed medulloblastoma and provided a unique tumor-suppressive mechanism in which MLL4, a H3K4 methyltransferase, is necessary to maintain broad H3K4me3 and SEs at tumor suppressor genes.
Several researchers have reported various SEs inhibitors that control cancer development. Chipumuro et al. (2014) found that CDK7i could lead to significant tumor regression in high-risk neuroblastoma mouse model without introducing systemic toxicity, which implied striking therapeutic selectivity. Henssen et al. (2016) reported that, in preclinical MYCN-driven neuroblastoma models, concurrent MYCN repression was observed in OTX015-treated samples, which could not be FIGURE 1 | Schematic representation of the functions of enhancer and super-enhancer (SE) in the regulation of gene expression, mediated by "looping." BRD4 binds to acetylated lysines (Ac) in enhancer, SE, and transcription factors (TF), bringing them together and mediating transcriptional co-activation and elongation via RNA polymerase II (RNA pol II) and mediator (Med) (Sengupta and George, 2017;Donati et al., 2018;Hajmirza et al., 2018). CDK7 can activate RNA Pol II by promoting its phosphorylation (Nilson et al., 2015). CDK7i, CDK7 inhibitor; BETi, BET inhibitor; p, phosphate group.  abrogated by ectopic MYCN expression. In addition, OTX015 treatment significantly suppressed tumor cell proliferation and improved survival of mice. Decaesteker et al. (2018) found that JQ1 coupled with THZ1 prevented the cell growth, proliferation, and differentiation in neuroblastoma primary cultured cell through strong repressive effects on CRC gene expression and p53 pathway response. Similarly, by gene mapping and integrating data with drug interaction databases, Mack et al. (2018) identified and validated dependency of ependymoma to SEs, which was responsive to SE inhibition. Meng et al. (2018) reported that THZ1 inhibits growth and proliferation of glioblastoma cells both in vitro and in vivo. In addition, CDK7 inhibition via CRISPR-Cas9 or RNA interference significantly disrupted GBM cell growth.
These results indicated that inhibitors of SEs could be promising candidates for cancer treatment. Table 3 summarizes the regulation and roles of SEs in visceral organ tumors.

SEs IN VISCERAL ORGAN TUMORS
Most SEs promote the forming and malignancy of visceral organ tumors. For example, SEs lead to the overexpression of ERG, leading to overexpression of target genes that drive development of prostate cancer (Babu and Fullwood, 2017). Moreover, SEs activate MAPK signaling pathway to inhibit apoptosis and promote proliferation of colon cancer (Nakamura et al., 2017).
However, in some cases they also suppressed cancer development. For example, TBX4, which is highly associated with SEs, was downregulated in lung cancer-associated fibroblasts (Horie et al., 2018). DNA hypermethylation suppressed some SEs in squamous cancer cells, whose downstream genes, such as SMGs, CUL3, and ZFP36L2, are important tumor-suppressors specific to the OSCC subtype .
Gene mutation, gene fusion, and aberrant expression of oncogenes or TFs activate SEs and ultimately lead to tumorigenesis (Johnatty et al., 2015;Garcia-Carpizo et al., 2016;Babu and Fullwood, 2017;Andricovich et al., 2018;Peng et al., 2018;Xie et al., 2018). Besides, translocation, 5hmC modification, methylation profile shifts, or 3D contact domain formation of SEs also lead to cancer development (Xiang et al., 2014;Hu H. et al., 2017;Weischenfeldt et al., 2017). The effect of SE blockers have been tested in various cancers, including the testing of THZ1 in esophageal cancer, iBET72 in cervical cancer, and JQ1 in cervical cancer, colorectal carcinoma, colon cancer, and squamous cell carcinoma (Dooley et al., 2016;Togel et al., 2016;Jiang Y.Y. et al., 2017;Nakamura et al., 2017). Thus, they could potentially be used as biomarkers or therapeutic targets in the future. Table 4 summarizes the regulation and roles of SEs in other tumors.

SEs IN OTHER CANCERS
Super-enhancer-associated mechanisms involved in other cancers include genomic rearrangements in pheochromocytomas, genome epigenetic reprogramming and recurrent tandem duplication in breast cancer, and ultraviolet radiation and aberrant activation of proto-oncogene in squamous cell carcinoma Yang et al., 2015;Betancur et al., 2017;Glodzik et al., 2017;Shen et al., 2017;Dwight et al., 2018). Besides, Drier et al. (2016) found that translocation of SEs upregulated MYB in adenoid cystic carcinoma. Similarly, Dwight et al. (2018) reported that translocation of SEs promoted the expression of TERT in pheochromocytomas and paragangliomas. Besides, formation of SEs upregulated several genes, such as CYP24A1, GJA5, SLAMF7, and ETV1, in squamous cell carcinoma , and deletion of SEs could lead to downregulation of MYC in several tumors. (Dave et al., 2017) In addition, SEs that are flanked by strong topologically associating domains (TAD) may be exploited as a functional unit to promote gene expression, and strong TAD boundaries and SE elements are frequently co-duplicated in cancer cells (Gong et al., 2018). All these studies further deepened our understanding about the mechanism of action of SE in cancers and provided novel possible targets for anti-tumor therapy.
Several researchers have widely used SE blockers to illustrate the implication of SEs in cancers. Wang et al. (2015), Sahni et al. (2017) Gong et al., 2018 carcinoma model to explore the efficacy of JQ1 and THZ1, and found significant inhibition of proliferation and enhancement of apoptosis. JQ1 was also reported to reduce the cell proliferation in melanoma model (Gelato et al., 2018). However, it is noteworthy that sometimes BETi could still function without participation of SEs. Sahni et al. (2017) reported that in triple-negative breast cancer, the mitosis regulator, LIN9, was often amplified and overexpressed. Although, it lacked a related SE, BETi could decrease its expression and inhibit mitosis in triple-negative breast cancer. Donati et al. (2018) reported that BRD4 participates in the activation and repair of DNA damage checkpoints and telomere maintenance. Therefore, in addition to blocking the function of SE, BETi can also inhibit tumorigenesis via other mechanisms, such as hindering the repair of DNA damage. This raised a question about the reliability of SEs' roles in cancers illustrated by BETi mechanism of action in previous studies. Did the function of BETi really arise from the inhibition of SEs, or actually they work through some other pathways? We still cannot draw a definite conclusion.

REGULATION OF SEs BY TUMOR-ASSOCIATED VIRUSES
Some virus-induced SE alteration in abovementioned cancers were discussed in corresponding parts of this review. Other studies, that did not involve specific tumors, found that SEs generally promoted tumorigenesis in virus-related cancers. The carcinogenic potential of viruses could arise from aberrant activation of host genes as well as integration of viral genes, both requiring the participation of SEs (Dooley et al., 2016;Gunnell et al., 2016;Jiang S. et al., 2017;Chen et al., 2018;Warburton et al., 2018). These results implied potential application of SE blockers to treat patients with high cancer risk from virus infection.

FUTURE DIRECTIONS
Super-enhancers refer to a class of regulatory regions with unusually strong enrichment of the binding sites for transcriptional co-activators. Although the roles of individual SEs vary with downstream genes, their overall effect in a particular tumor is relatively stable. In most of the cancer cases, SEs act as oncogenes to promote tumor growth. Therefore, SEs could be a promising therapeutic target in those cancers.
Several studies to investigate the potential of SEs as therapeutic targets have been conducted. Some researchers reported the existence of positive feedback loops. For example, Ets2 and Elk3 genes in squamous cell carcinoma can upregulate specific SEs (Yang et al., 2015). This not only intensified the function of the downstream genes, but also indicated potentially more sensitive responses to SE blockers. Besides, CDK7i showed striking selectivity in regression of neuroblastoma cells, without significant systemic toxicity (Chipumuro et al., 2014). In addition, some key downstream molecules and pathways of SEs are involved in various tumors, making them promising therapeutic targets for multiple cancers (Xu et al., 2017). Importantly, the effect of BETi has been proven in some virus-induced tumors models. For example, JQ1 and iBET72 could inhibit proliferation of cervical neoplasia induced by HPV (Dooley et al., 2016). Thus, SE blockers may have the potential to treat virusinduced cancer as well as patients with high cancer risk from virus infection. According to these studies, SEs have good prospects as potential therapeutic targets for cancers because of their strong potency, high selectivity, and broad applicability (Figure 2).
However, although the roles of SEs have been validated in many cancer cells, the degree of their involvement is still controversial. The main mechanism of action of BETi is considered to be blocking of the interaction between SEs and BRD4, which is a co-activator that can bind acetylated histones in SEs and TFs and directly interact with the mediator complex and elongation factors (Dawson et al., 2011). However, some target genes of BETi, such as LIN9 gene in triplenegative breast cancer, do not possess any corresponding SEs sites (Sahni et al., 2017). In addition, some researchers have proposed a non-transcriptional role of BRD4 in activation and repair of DNA damage checkpoints and telomere maintenance, opening new perspectives on the use of BETi in cancer (Donati et al., 2018). Similarly, other than inhibiting SEs, THZ1 can also cause defects in Pol II phosphorylation, co-transcriptional capping, promoter proximal pausing, and productive elongation (Nilson et al., 2015). Therefore, the roles of SEs in the above-mentioned cancers, illustrated by BETi and CDK7i, are still unclear. Besides, it is noteworthy that targeting SEs for cancer treatment might cause significant side effects because some tumor suppressor genes will also be suppressed when blocking SEs. Therefore, more studies and better understanding of mechanisms are urgently needed before SEs could be utilized as therapeutic targets to treat specific cancers.
Therefore, for future studies of SEs, the focus could be: (1) SEs are cell-type-specific and have the potential to be used for identification of different subtypes of cancer. Therefore, their application in precision medicine is promising. In the future studies, researchers could try to use them to distinguish subtypes of cancer and give more precise treatment strategy.
(2) For cancers that lack known genetic drivers and are recalcitrant to therapeutic development, SE sequencing and investigation may provide novel directions for significant breakthroughs. (3) In addition to blocking BRD4, other mechanisms of BETi are worth studying to validate the previous conclusions or obtain new explanation for the reported results. (4) The interactions between SEs and virus infection need further research. A better understanding of them might benefit not only the virus-related cancer patients, but also those who have high risk of cancer due to virus infection. (5) Epigenetic changes on SEs can also lead to significant differences in phenotypes. Therefore, the combination of SEs and epigenetics (such as DNA methylation and acetylation) could be good research topics. (6) Some BETi have been put into clinical trials, such as OTX015 in multiple myeloma and acute leukemia. The practical application of BETi in other tumors awaits further exploration.

AUTHOR CONTRIBUTIONS
YH and WL searched and wrote the manuscript. QL revised the manuscript and supervision of the entire project.