4.1. BD1-Selective Inhibitors.

BD1 and BD2 each consist of four antiparallel α-helixes (αZ, αA, αB, and αC) with different lengths linked by two hydrophobic loops (ZA and BC) that together form a hydrophobic Kac recognition pocket. Amino acid differences in the binding pockets of BD1 and BD2 lead to narrower binding regions and differences in the polarity and hydrophobicity of BD2. Poor selectivity among BET proteins and between BD1 and BD2 has incurred unwanted side effects and reduced the efflcacy of BET bromodomain inhibitors in clinical trials.⁵⁸ Thus selectively targeting only BD1 or BD2 in BRD4 can provide better specificity and avoid unnecessary bioreactions triggered by the other domain inhibition as well as poor clinical evaluation due to the concurrent targeting of both BD1 and BD2 with differences in functions and structures. Gilan et al. have developed respective BD1- and BD2-specific inhibitors for better disease treatment and efflcacy based on the functional and structural differences between BD1 and BD2.¹⁴ Because BD1 and BD2 within the same protein are structurally and functionally more diverse compared with the analogous BD1 or BD2 found in different BET proteins,² it is in fact easier to develop inhibitors selective for BD1 or BD2 than for individual BET family members.

Many BD1-selective inhibitors have been discovered, including Olinone (1)^59,60 and MS436 (2),⁶⁰ with the 3D crystal structure of the latter with BRD4-BD1 also shown on the right (Figure 3A). ZL0580 (3) (IC₅₀ = 163 nM), a BRD4 BD1-selective inhibitor, was discovered by structure-guided drug design.⁶¹ Overlay analysis reveals that 3 can access the Kac-binding pocket in BD1 with binding notably different from that of JQ1. Compound 3 induces the epigenetic suppression of HIV in multiple in vitro and ex vivo models. Another BD1-selective inhibitor, LT052 (4),⁶² with more than 100-fold selectivity for BRD4-BD1 over BRD4-BD2 and exhibiting a stable level in liver microsomal metabolic analysis inhibits monosodium urate (MSU)-induced apoptosis in human THP-1 monocytic cells and improves symptoms of acute gouty arthritis via BRD4/NF-κB/NLRP3 inflammasome signaling pathways. Chromone derivatives ZL0513 (5) and ZL0516 (6) both show high binding afflnity for BD1 with a value of 67 and 84 nM, respectively. In vivo pharmacokinetics (PK) profiles between compounds 5 and 6 are similar (half life 3 to 4 h, maximum concentration ~600 ng/mL, AUC ~5000 ng·h/mL), and the oral bioavailability of compounds 5 and 6 of around 35−38% (20 mg/kg) is generally acceptable in rats.⁶³ GSK778 (7), identified by the functional and structural comparison between BD1 and BD2 showing 130 times higher afflnity for BD1 than BD2, inhibits not only the growth and vitality of human cancer cell lines but also the clone-forming ability of human primary AML cells.¹⁴ Further chemical modification of compound 7 leads to another potent, selective, and cell-permeable inhibitor of BD1 (GSK789 (8)), displaying more than 1000-fold selectivity for BD1 over BD2 and achieving a submicromolar inhibition of monocyte chemoattractant protein-1 (MCP-1) release in human whole blood (pIC₅₀ = 6.5).⁶⁴ Starting with I-BET151 guided by structural information leads to imidazoquinoline compound 9 with 150-fold BD1 selectivity over BD2.⁶⁵ Compound 9 reduces the production of IL-6 and MCP-1 in LPS-stimulated human peripheral blood mononuclear cells (PBMCs) and whole blood and suppresses MV4–11 lymphoblast leukemia cell growth (pIC₅₀ = 7.0), indicating that BD1-selective inhibitor compound 9 is sufflcient to maintain the anti-inflammatory and antiproliferative activity of pan-BET inhibition. Compound 10, another BD1-selective inhibitor (28 times higher afflnity for BRD4-BD1 than BRD4-BD2), displays similar selectivity against the BD2 of BRD2 and BRD3 (25 and 33 times, respectively).⁶⁶ This study also reveals that the underlying BD1 selectivity lies in the flexibility of the BD1-conserved YNKP motif and the substitution of structured waters in the Kac-binding site, which, unlike typical pan-BET inhibitor-suppressed MYC expression, leads to enhanced MYC expression in compound 10-treated MM.1S multiple myeloid cells. This unexpected result needs to be further investigated.

Open in a separate window

Figure 3.

Reported novel selective inhibitors 1−20. (A) Chemical structures of BRD4 BD1-selective inhibitors. The docking pose of compounds 2 and 6 with BRD4−BD1 in ribbon representation is shown on the right. Key residues, including Pro82, Gln85, Lys91, Tyr97, and Asn140, are shown in green sticks, and the amide ketone of compound 2 interacts with the critical conserved residue Asn140 to form a hydrogen bond. Similarly, in the docking pose of compound 6, key residues Asn140, Tyr97, Gln85, and Pro82 are green. The carbonyl O atom of compound 6 interacts with the conserved residue Asn140, forming a direct H bond, and with Tyr97, forming an indirect hydrogen bond mediated by a water molecule. (B) Reported BRD4 BD2-selective inhibitors. (C) Chemical structures of BRD4-selective inhibitors.

4.2. BD2-Selective Inhibitors.

Representative BD2-selective inhibitors identified to date are shown in Figure 3B. The tetrahydroquinoxaline derivative GSK340 (11) with BD2 selectivity displays good solubility in both thermodynamic and kinetic solubility assays.⁶⁷ In addition, compound 11 is cell-permeable and reduces the release of the MCP-1 cytokine in human PBMCs and whole blood. BY27 (12) is a potent BD2-selective inhibitor with 7 times higher selectivity for BD2 over BD1.⁶⁸ In MV4–11 cells, compound 12 reduced the c-MYC protein level and increased the p21 expression in a dose-dependent manner. Compared with I-BET762, which caused 11% weight loss at 60 mg/kg, compound 12 is less toxic (exhibiting 7% weight loss at 150 mg/kg) at high doses and has 67% inhibition of tumor growth in a MV4–11 mouse xenograft model. Currently, ABBV-744 (13),⁶⁹ a highly effective inhibitor of the BD2 domain with more than 300 times higher selectivity for BD2 over BD1, is synthesized and shows moderate preclinical PK profiles, including a moderate to high volume of distribution, a 2.0−4.4 h half life, and moderate clearance.⁷⁰ The structural modification of an ABBV-075 analog leads to compound 13. 2,6-Dimethylphenyl ether in compound 13 contributes to BD2 selectivity, with a tertiary alcohol on the central phenyl ring suppressing the activity against BRD4-BD1 and a fluorine atom on the phenyl ether improving the pharmacokinetic properties of compound 13. In a myelodysplastic SKM-1 cell-derived xenograft mouse model, compound 13 displays 83% tumor growth inhibition with minimal (2%) weight loss over 21 days at a dose of 18.8 mg/kg, whereas ABBV-075 has similar tumor growth inhibition with 7% weight loss at the maximally tolerated dose of 1 mg/kg, suggesting that more selective BET inhibitors may improve the tolerability, albeit modestly. Compound 13 is currently in phase I clinical evaluation. GSK046 (iBET-BD2, 14), although less effective than 8 in the human cancer cell lines examined, exhibits more than 300 times higher selectivity for BD2 over BD1 and appears to ameliorate inflammatory disease in preclinical models.¹⁴ A recent study shows that compound 14 has physicochemical and pharmacokinetic properties suitable for use in vivo, and it retains the ability to effectively inhibit the release of MCP-1 cytokines (pIC₅₀ = 7.5) in the cellular and whole blood environment.⁷¹ Interestingly, the inhibition of BD1 or BD2 leads to the suppression of MCP-1 despite the fact that these two types of inhibitors have distinct selectivity profiles. Another study applying a template hopping approach based on the orthogonal template (compound 14) leads to the synthesis of compounds GSK620 (15) and GSK549 (16) with high selectivity and low levels of inhibition against indicators of hepatoxicity.⁷² SJ018 (17) and SJ432 (18), two tetrahydroquinoline analogs, inhibit the expression of MYC and downstream targets but display no rebound effects, as compared with JQ1 in pediatric tumor cell lines, suggesting that BD2-selective inhibitors are effective candidates for the design of antitumor drugs for MYC-driven pediatric malignancies.⁷³

4.3. BRD4-Selective Inhibitors.

Besides BD1- and BD2-selective inhibitors, the specific targeting of BRD4 rather than other BET proteins is also emerging. The ubiquitously expressed BRD2, BRD3, and BRD4 have a broad range of functions and participate in housekeeping and lineage-specific gene regulation in somatic and progenitor/stem cells by interacting with distinct partner proteins. BRD2 associates with E2F transcription factors to regulate E2F-dependent cell differentiation and proliferation genes and interacts with RNA Pol II to engage in transcription elongation in embryonic kidney cells through hyperacetylated nucleosomes due to the lack of a CTM.^74–76 Similar to BRD2, BRD3 is mainly involved in the regulation of gene transcription related to the development of embryonic stem cells, but unlike BRD4, it interacts with RNA Pol II in a P-TEFb-independent manner.⁷⁷ The germ-cell-specific BRDT associates with multiple interactors during spermatogenesis and oogenesis.^78,79 Mice treated with the pan-BET inhibitor JQ1 had a reduced spermatozoa number and reduced motility, indicating that BRDT could be an ideal target for contraceptive drug development.^80,81 Considering the intricate physiological functions of BET family proteins, it is important to improve the target selectivity of BRD4 inhibitors in specific biological processes and disease progression without incurring unintended side effects, especially in the treatment of cancer and inflammation.⁵⁸ Two BRD4-selective inhibitors reported in recent years are shown in Figure 3C. ZL0454 (19) and ZL0420 (20) are two BRD4-selective inhibitors displaying 30−60-fold selectivity over BRD2, 50−90-fold selectivity over BRD3, and 70−120-fold selectivity over BRDT.⁸² Molecular docking shows that compounds 19 and 20 occupy the Kac-binding pocket of BRD4 in a similar way. In vivo experiments further found that 19 and 20 almost completely blocked the TLR3 (toll-like receptor 3)-dependent expression of innate immune genes, including ISG54, ISG56, IL-8, and Groβ in human primary small airway epithelial cells (hSAECs). Given the shared functions between BRD4 and the other BET family proteins yet to be completely defined, it is difflcult to predict what types of diseases are most suited for BRD4-selective targeting. Nevertheless, therapy with fewer unintended targets has the potential to improve the clinical outcomes.

4.4. Dual Pharmacophores.

In recent years, multitarget approaches targeting different proteins and inhibiting kinases have received increasing attention in anticancer drug development.⁸³ A previous study that reported that the CDK inhibitor Dinaciclib could interact with the Kac-binding pocket of BRDT provided the first indication that kinase inhibitors might be used to rationally design a new generation of BRD4 inhibitors.⁸⁴ Since then, on the basis of the idea that polypharmacological inhibitors with multitargets would have a more robust action, many dual BET/kinase inhibitors have been described.^85,86 Some of the recently discovered dual-target inhibitors are shown in Figure 4.

Open in a separate window

Figure 4.

Chemical structures of dual-target inhibitors 21−31.

Histone deacetylases (HDACs) reduce the acetylation level of lysine residues where BET proteins recognize and promote factor recruitment to targeted gene loci, thereby regulating gene transcription.^2,87–89 It is logical to cotarget BET proteins and HDACs for the more efflcient repression of oncogenic gene transcription and tumor cell proliferation.^90,91 Compound 21, a highly potent dual BET/HDAC inhibitor, was reported with balanced activity against BRD4 BD1 and HDAC1.⁹² Compound 22 shows the inhibition of BRD4 at a rate of 88% at 10 μM and strong HDAC3 inhibition with an IC₅₀ value of 5 nM.⁹³ Compound 23, another dual BRD4/HDAC inhibitor, induces autophagic cell death in colorectal cancer (CRC) cells at nanomolar levels.⁹⁰ Compound 24 induces HL-60 AML cell death with c-MYC-inhibitory activity and suppresses cell growth in BET inhibitor-resistant cells.⁹⁴ Compounds 25 and 26 also exhibit potent antiproliferative activity in human leukemia cell lines.^95,96

Because BRD4 plays a critical role in DNA repair (Figure 1D), the inhibition of BRD4 shows synthetic lethality with poly-(ADP-ribose) polymerase-1 (PARP1) inhibitors, resulting in HR deficiency and providing a strategy for cancer treatment.^35,97 Compound 27, designed with the conception of synthetic lethality, is the first dual BRD4/PARP1 inhibitor reported that triggers breast cancer cell apoptosis and induces cell-cycle arrest in the G1 phase.⁹⁸ A recent study reports that polo-like kinase 1 (PLK1) upregulates BRD4-controlled MYC transcription and protein stability, with MYC, in turn, enhancing the PLK1 gene transcription, thus forming a PLK1-MYC feed-forward circuit in double-hit lymphoma (DHL) cells.⁹⁹ Via the chemical modification of the pre-existing dual BRD4/PLK1 inhibitor BI-2536 (28), the dual BET/PLK1 inhibitor WNY0824 (29) was shown to suppress CRPC cell proliferation as well as the tumor growth in an enzalutamide-resistant CRPC xenograft mouse model by blocking BRD4-dependent androgen-receptor- and MYC-driven transcription programs.¹⁰⁰ Fine-tuning the selectivity of inhibitors can also be achieved through structural guidance. Compound 30 is identified as a well-balanced dual BRD4/PLK1 inhibitor, illustrating that the methylated amide and the cyclopentyl group of the compound 28 scaffold are important for maintaining bromodomain-binding activity but are less critical for PLK1-inhibitory activity.¹⁰¹ Another study also chose compound 28 as the starting compound but changed PLK1 inhibition to anaplastic lymphoma kinase (ALK) targeting with the derived inhibitor compound 31, maintaining the BRD4-inhibitory activity but shifting the kinase selectivity from diminished PLK1 inhibition to enhanced ALK inhibition, which shows the remarkable suppression of ALK^F1174L- and BRD4-dependent MYCN-driven gene transcription in pediatric neuroblastoma cells.¹⁰² These strategies provide new directions for the development of novel dual-target inhibitors via the chemical modification of existing pharmacophores targeting, respectively, BET proteins and kinases. However, this new line of research also poses significant challenges in how to properly balance the selectivity between BET proteins and specific kinases in designing dual-target inhibitors with enhanced antitumor activity. A report documenting that PLK inhibition, rather than BRD4 inhibition, is the main action for the anticancer activity of compound 28 in MV4–11 cells may provide some mechanistic hints about this specific compound.¹⁰³

It should be mentioned that many dual-target inhibitors were discovered through their unexpected BET-targeting activity. For instance, TG101209 and BI-2536, originally reported as JAK2 inhibitor and PLK1 inhibitor, respectively, were found to have strong inhibitory potential against BRD4 through binding to the Kac pocket, providing a rationale for the development of dual-target inhibitors.¹⁰⁴ As the products of a new technology, dual-target inhibitors display great potential in cancer therapy, even though they are still in the preclinical stage. Combining two pharmacophores cotargeting BET proteins and HDACs based on their pharmacological activity and safety evaluation, for example, 21 synthesized by linking (+)-JQ1 to HDAC inhibitors SAHA and CI-944,⁹² has become a popular trend in the discovery of dual-target inhibitors.¹⁰⁵ For dual-target inhibitors that have already been developed, structural modification may result in new inhibitors with different biological activities, such as compounds 29, 30, and 31, optimized from compound 28.^96,102

Additional kinase inhibitors to be considered include those involved in the PI3K-AKT-mTOR pathway that regulates MYC expression, stability, post-translational modification, and MYC protein synthesis¹⁰⁶ as well as CDK9, which phosphorylates the Pol II CTM at Ser2, resulting in the stimulation of transcription elongation, and is associated with BRD4 hyperphosphorylation in NMC.^40,83,107 CK2, which phosphorylates BRD4, implicated in chromatin targeting and factor recruitment,⁸ is another promising candidate for the future development of potent dual-target inhibitors for cancer therapy.

4.5. PROTACs.

PROTACs, initially proposed by Deshaies and colleagues,¹⁰⁸ are an effective way to inactivate protein function via compound-mediated proteolysis. PROTACs targeting BET proteins have distinct pharmacological effects compared with the corresponding small-molecule inhibitors because they can disrupt the function of the entire protein rather than just the bromodomain activity.^46,109 A sustained loss of BET proteins has superior antiproliferative effects in vitro while showing better anticancer activity in vivo with no obvious toxicity in a mouse tumor regression study.¹¹⁰

The currently developed BET degraders use ligand binding to the cereblon (CRBN) or von Hipple-Lindau (VHL) component of an E3 ubiquitin ligase. Compounds 32−38 are CRBN-based degraders, whereas compounds 39−45 are based on the VHL component (Figure 5A). Among them, dBET1 (32) containing an eight-atom linker is the first degrader displaying remarkable selectivity in depleting BET proteins with more potent inhibition compared with JQ1 against MV4–11 cells, DHL-4 lymphoma cells, and primary blasts from leukemia patients.¹¹¹ ARV-825 (33) with a different linker type also achieves an effective degradation of BRD4 (DC₅₀ < 1 nM), resulting in strong and lasting c-MYC inhibition in comparison with pan-BET inhibitors JQ1 and OTX015.¹¹² BETd-246 (34) induces BET protein degradation at low nanomolar concentrations within an hour in TNBC cells by downregulating many proliferation/apoptosis-related genes, leading to growth inhibition and apoptosis.¹¹³ Another degrader, BETd-260 (35), shows picomolar potency (IC₅₀ = 51 pM) in inhibiting RS4–11 lymphoblastic leukemia cell growth.¹¹⁰

Open in a separate window

Figure 5.

Reported PROTAC degraders and PPI inhibitors. (A) PROTAC degraders based on CRL4^CRBN (32−38) and CRL2^VHL E3 ligase (39−45). (B) Chemical structures of 46−48.

QCA570 (36) was designed and synthesized based on the 1,4-oxazepine structure with more than 1000 times higher potency compared with 33. Significantly, in both the MV4–11 and RS4–11 acute leukemia xenograft models, compound 36 shows complete and long-lasting tumor regression at 5 mg/kg without apparent toxicity.¹¹⁴ A recent study highlights that DP1 (37) not only achieves rapid, stable, and persistent BRD4 degradation in SU-DHL-4 lymphoma cells but also has significant inhibitory effects on immunodeficient mice with xenotransplanted lymphoma cells.¹¹⁵ Cellular degradation assays showed that ZXH-3–26 (38) exhibits a DC_50/5h of ~5 nM and BRD4 selectivity without inducing significant inhibition or degradation of BRD2 and BRD3,¹¹⁶ a result similar to that observed with AT1 (39) displaying BRD4-selective depletion but negligible activity against BRD2 and BRD3 at 1 to 3 μM of 39 treatment.¹¹⁷ ARV-771 (40), a small-molecule pan-BET degrader, reduces the protein levels of androgen receptor (AR) signaling in a CRPC mouse xenotransplantation model, providing the first reported BET degrader activity against a solid tumor.¹¹⁸ Compound 40 also presents superior pharmacological properties compared with those of 33 in mantle cell lymphoma (MCL) Mino cells.¹¹⁹ MZ1 (41), a JQ1-based BRD4 degrader, exhibits BRD4-targeting preference in cell-based assays but comparable in vitro binding afflnities to individually purified BRD2, BRD3, and BRD4 bromodomains measured by isothermal calorimetry (ITC), suggesting that MZ1-induced proximity and accessible surface lysine residues on BRD4, versus BRD2 and BRD3, might be differentially contacted by the VHL E3 ligase module in vivo and are presumably regulated by BET interactors that are unique to each family member.¹²⁰ Adding a cyclizing linker to compound 40 leads to a macrocyclic PROTAC, macroPROTAC-1 (42), with cellular activity similar to that of compound 41, thus providing a new strategy for drug design.¹²¹

The targeting effects between triazolodiazepine JQ1 and tetrahydroquinoline I-BET762 and the optimal length of linkers connecting the BET-binding ligand JQ1 or I-BET762 to the VHL ligand VH032 for degrader design were examined by synthesizing a series of I-BET762-derived degraders, including MZP-54 (43), MZP-55 (44), and MZP-61 (45),¹²² for comparison with MZ1-related compounds, with the results showing that JQ1-based PROTACs are more effective at inducing BET protein degradation, even though JQ1 alone is not as potent as I-BET762 in BET protein binding. This parallel comparison indicates that stronger inhibitors may not necessarily produce more effective PROTAC degraders, and the linker length is critical in PROTAC design. It is important to note that PROTAC binding to incidental or unintended targets through substrate or E3 ligase association does not guarantee substrate degradation.¹²³ These BET degraders are effective in inhibiting cancer cell proliferation in vivo and in vitro. In mice, compounds 34, 35, and 36 show good tolerance without obvious toxicity.¹²⁴

BET-targeting PROTACs are still in the preclinical stage, and thus their therapeutic potential is presently unknown. For the current degraders, the main challenge is to improve the bioavailability index through precise modification without significantly increasing the molecular size. More selective and tissue-specific degraders are also needed for clinical trials. A strategy involving antibody−PROTAC conjugation provides a proof of concept for tissue-specific BRD4 degradation.¹²⁵ A trastuzumab−PROTAC conjugate (Ab-PROTAC) with blocked degradation activity by an antibody linker that can be selectively released to produce an active PROTAC demonstrates the significant potential for tissue-specific target degradation. It should be mentioned that targeting BET proteins for degradation is not limited to ubiquitination. The lysosomal/autophagy pathway provides cargo-specific degradation through a different molecular mechanism.¹²⁶

Funding for the authors’ research was sponsored in part by the National Natural Science Foundation of China, grant numbers 81922064 (to L.O.), 81874290 (to L.O.) and 81903502 (to J.Z.); the National Science and Technology Major Project of the Ministry of Science and Technology of China, grant number 2018ZX09735005 (to L.O.); the National Major Scientific and Technological Special Project for ‘Signficant New Drugs Development’, grant number 2018ZX09201018–021 (to L.O.); the Sichuan Science and Technology Program, grant number 2019YFS0003 (to L.O.); the Sichuan Science and Technology Program, grant number 2020YJ0091 (to J.L.); the China Postdoctoral Science Foundation, grant number 2020M673268 (to J.Z.); the U.S. National Institutes of Health (NIH), grant number 1RO1CA251698–01 (to C.-M.C.); and the Cancer Prevention and Research Institute of Texas (CPRIT), grant numbers RP180349 and RP190077 (to C.-M.C.).