Inhibition of members of the bromodomain and extraterminal (BET) family of proteins has proven a valid strategy for cancer chemotherapy. All BET identified to date contain two bromodomains (BD; BD1 and BD2) that are necessary for recognition of acetylated lysine residues in the N-terminal regions of histones. Chemical matter that targets BET (BETi) also interact via these domains. Molecular and cellular data indicate that BD1 and BD2 have different biological roles depending upon their cellular context, with BD2 particularly associated with cancer. We have therefore pursued the development of BD2-selective molecules both as chemical probes and as potential leads for drug development. Here we report the structure-based generation of a novel series of tetrahydroquinoline analogs that exhibit >50-fold selectivity for BD2 versus BD1. This selective targeting resulted in engagement with BD-containing proteins in cells, resulting in modulation of MYC proteins and downstream targets. These compounds were potent cytotoxins toward numerous pediatric cancer cell lines and were minimally toxic to nontumorigenic cells. In addition, unlike the pan BETi (+)-JQ1, these BD2-selective inhibitors demonstrated no rebound expression effects. Finally, we report a pharmacokinetic-optimized, metabolically stable derivative that induced growth delay in a neuroblastoma xenograft model with minimal toxicity. We conclude that BD2-selective agents are valid candidates for antitumor drug design for pediatric malignancies driven by the MYC oncogene.

Significance:

This study presents bromodomain-selective BET inhibitors that act as antitumor agents and demonstrates that these molecules have in vivo activity towards neuroblastoma, with essentially no toxicity.

The targeting of chromatin-modifying proteins has become an attractive avenue for drug design, specifically for antitumor agents (1–3). Consequently, there are considerable ongoing efforts to develop chemical matter that demonstrates specificity for such proteins. The bromodomain and extraterminal (BET) family of proteins, which includes BRD2, BRD3, BRD4, and BRDT, contains two domains (BD1 and BD2) that are required for the recognition of acetylated lysine residues in histones (4, 5). These bromodomains (BD) demonstrate considerable amino acid and structural homology, and hence, it has been difficult to develop small molecules with significant selectivity toward either BD1 or BD2 (6). Therefore, most clinical candidates that target BET inhibit both BD with similar efficiencies. This includes the exemplar compound (+)-JQ1 (1), based upon a benzodiazepine scaffold, which is currently the basis for at least four molecules in clinical trials (see http://clinicaltrials.gov).

However, biochemical and cellular studies have revealed that independently modulating these domains results in different downstream sequelae, arguing that the recognition of acetylated lysine residues by BD1 and BD2 is specific and leads to the regulation of disparate gene sets (7–11). For example, BRDT-BD1 (but not BRDT-BD2) is required for spermatogenesis (8). Similarly, isothermal titration calorimetry (ITC) studies indicate that BD1 preferentially binds to histone H3 sequences, whereas BD2 has higher affinity for histone H4 and acetylated lysine peptides derived from cyclin T1 (12). BRD3-BD1 has been shown to be important for the expression of erythroid and megakaryocyte-specific genes through interactions with the acetylated transcription factor (TF) GATA1 (7). In contrast, BRD4-BD2 recruits TWIST, and BRD4-BD1 serves to anchor the complex to chromatin (9). Because there are greater sequence homologies among the BET BD1 BDs and BD2 BDs, as compared with BD1 versus BD2, this suggests that the separate bromodomains have different biological functions. We hypothesized therefore, that targeting either BD1 or BD2 may be exploited in anticancer drug design.

We have developed BD-selective BET inhibitors (BETi) chemical probes that target BD1 and BD2s, and assessed their activity in pediatric tumor models. Using the tetrahydroquinoline (THQ) scaffold (6, 13) and a combination of structure-based iterative drug design, coupled with biochemical analyses, we have generated highly potent, BD2-selective BETi. These molecules: bind BET in cultured cells; downregulate MYC and associated targets; are cytotoxic to tumor, but less so to nontumorigenic cells; and are minimally toxic in vivo and induce antitumor activity. While BD-selective probes have been synthesized previously (14–16), our studies seek to validate the activity of such compounds toward pediatric tumors with dire prognoses.

Reagents

The list of cell lines, their sources, culture conditions, and antibodies used in this study are reported in the Supplementary Information (Supplementary Tables S1–S3). Three neuroblastoma lines were used for subsequent studies based upon their MYC expression status: SJ-N-AS—Amplified c-MYC, no expression of MYCN; SK-N-SH—Expresses c-MYC, no expression of MYCN; IMR32—No expression of c-MYC, MYNC amplified (Supplementary Fig. S1). Cell lines were used directly without further testing. All synthetic reagents were of ACS grade and purchased from Sigma-Aldrich, Combi-Blocks or Strem Chemicals. Female CB17SCID mice [CB17/Icr-Prkdc(scid)/IcrIcoCrl; Charles River Laboratories] were used for therapeutic studies and were housed in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited facility with food and water provided ad libitum.

Biophysical analyses

Time-resolved fluorescence energy transfer (TR-FRET) was undertaken using commercially available kits (Cayman Chemical). Surface plasmon resonance (SPR) was undertaken on a Pioneer optical biosensor (ForteBio) using polyHis-tagged BD domains (BRD2-BD1 amino acids 74–194; BRD2-BD2 amino acids 348–455). More detail is presented in the Supporting Information. ITC experiments were performed on an iTC200 (MicroCal Panalytical) and peak areas were integrated, normalized, and fitted using the independent sites model in Origin software (MicroCal Panalytical). BETi selectivity was validated using BROMOscan analysis (DiscoveRX).

Cytotoxicity assays

Cytotoxicity of BETi was determined using either Alamar Blue or Cell Titer Glo (see Supporting Information for further details).

Fluorescence recovery after photobleaching

Fluorescence recovery after photobleaching (FRAP) was determined in U2OS cells expressing BRD4 protein fused with GFP. Microscopy images were acquired using a Yokogawa CSU-X spinning disc confocal microscope with multiple prebleach images of nuclei collected to establish a baseline. A single laser pulse lasting 0.09 seconds was then targeted onto the nucleus and single section images were collected at 200 ms intervals. Fluorescence intensity plots were generated using Slidebook 6 × 64 software (3i Technologies), and FRAP determined as described previously (17).

Microarray

Total RNA was hybridized to an Affymetrix Human Gene 2.0 ST Array (Thermo Fisher Scientific) and signals normalized using the multiarray average algorithm (18). Datasets were subsequently analyzed by a one-factor ANOVA model (Partek Genomics Suite) where drug plus concentration were used as the treatment factor. A FDR threshold of <0.05 was used to identify differentially expressed transcripts and were analyzed for functional enrichment using the Enrichr (19) and DAVID (20) bioinformatics databases.

Chromatin immunoprecipitation sequencing

Chromatin immunoprecipitation sequencing (ChIP-seq) was performed as described previously (21). Differential histone 3, lysine 27 acetylation (H3K27ac) read enrichment was identified using DESeq2 (22) on normalized read depth at the union of all reproducible H3K27ac sites. Principal component analysis was performed using the prcomp function in R. CentriMo (23) and was used to identify enriched TF motifs within differentially enriched H3K27ac sites.

BRD2-BD1 and BRD2-BD2 expression

cDNAs encoding the human BRD2-BD1 (residues 67–200) and BRD2-BD2 (amino acids 348–455) domains were expressed from pET28a(+) containing a N-terminal His-tag. Detailed methods are provided in the Supporting Information.

Crystallographic analyses

Structure of BRD2-BD1/SJ432 complex was obtained by soaking apo crystals in 1.5 mmol/L SJ432 for 2 days. BRD2-BD2/SJ432 complexes were preformed in solution and then crystallized. Crystals were grown using the sitting drop vapor diffusion method at 18°C and all diffraction data were collected at the SERCAT beam lines 22-BM and 22-ID at the Advanced Photon Source. The BD1/SJ432 and BD2/SJ432 structures were solved by molecular replacement using, respectively, BD1 (PDB 4UYH) and BD2 (PDB 5IG6) of BRD2 as search models, and refined and optimized using PHENIX and COOT (24, 25). Data collection statistics are summarized in the Supporting Information (BRD2-BD1/SJ432—PDB 2DVQ; BRD2-BD2/SJ432—PDB 2E3K).

Immunoblotting analyses

Immunoreactivity of protein extracts to desired antibodies was undertaken using standard approaches. See Supporting Information for antibodies used and their respective working dilutions.

Pharmacokinetic studies

Pharmacokinetic (PK) studies were conducted by SAI Life Sciences Ltd using female athymic nude mice (ACTREC). Molecules were administered intraperitoneally, and at time intervals ranging from 15 minutes to 24 hours, animals were humanely sacrificed and the levels of free drug present in the plasma and brain tissue were determined. All data points were conducted in triplicate.

Preclinical studies

Six- to eight-week-old CB17SCID female mice were injected into the flank with 1 × 106 SK-N-AS cells resuspended in Matrigel matrix (Corning) and tumors were allowed to grow until they reached approximately 225 mm3. SJ432, formulated in 5% 1-methyl-2-pyrrolidinone, 5% Solutol HS15 (Sigma Biochemicals), and 90% saline, was administered intraperitoneally daily for 14 days. JQ1, given by the same route and schedule, was formulated in 10% (2-hydroxypropil)-β-cyclodextrin solution (Sigma Biochemicals), 10% DMSO, and 80% saline. Ten mice per group were used. Tumors were measured using digital calipers and volumes were calculated [V = (L × W2)/2]. Toxicity was assessed primarily by weight loss, but also by daily examination by individuals with no knowledge of the treatment protocol. All animal studies were approved by the St. Jude Children Research Hospital Institutional Animal Care and Use Committee.

Rational design of BD2-selective BETi

Previously, we reported that amino acid residue variations between BD1 and BD2 induce differences in the water networks that could be exploited by heteroaryl-substituted THQ to achieve BD-selectivity (6, 13). However, the identified lead compound, SJ599 (2), showed only modest BD2-selectivity and the 2-furan group would be a liability for in vivo use. On the basis of our analysis of the cocrystal structure of 2 bound to BRD2-BD2 (PDB: 5EK9), we hypothesized that meta-substituted phenyl substituents (3–6) or indole (7) could stabilize the water network present in BD2 (Fig. 1A–C). Unfortunately, no improvement in BD2-selectivity was obtained (Fig. 1C), although the m-acetamide (3) and m-aniline (4) analogs demonstrated increased potency toward BRD2-BD2. Increasing the steric bulk on the acetamide (8–12), improved BD2-selectivity, resulting in higher lipophilicity and decreased ligand efficiency. Previously, we found that replacing the isopropyl-carbamate at R1 with an aryl group enhanced BD2-selectivity. Therefore, we generated analogs, holding the m-acetamide at R2 fixed, and varying the R1 position (9, 13–20). These compounds were more potent toward BRD2-BD2, with selectivity, as compared with BRD2-BD1, ranging from 6.9- to 66.5-fold.

Figure 1.

A and B, Docking studies of THQ analogs of 1. C, Synthetic route to aryl substituted THQ and BETi BD binding affinities for derived analogs. Reagents and conditions: (a) TFA, CH2Cl2, quantitative; (b) R1-Br, K2CO3, BrettPhos Palladacycle Gen. 3, BrettPhos, THF 100°C.

Figure 1.

A and B, Docking studies of THQ analogs of 1. C, Synthetic route to aryl substituted THQ and BETi BD binding affinities for derived analogs. Reagents and conditions: (a) TFA, CH2Cl2, quantitative; (b) R1-Br, K2CO3, BrettPhos Palladacycle Gen. 3, BrettPhos, THF 100°C.

Close modal

On the basis of potency toward BD2 (Kd = 14 nmol/L), and differential activity against BD1 (∼67-fold), we investigated 22 (SJ018) in more detail using three orthogonal biophysical assays: ITC, SPR, and a BD-binding assay (BROMOscan) (Fig. 2A–D; Supplementary Table S4). Excellent agreement between the outputs was observed and the latter confirmed on-target BET inhibition, indicating that modifications made to the THQ scaffold did not alter BET specificity. To assess on-target engagement of SJ018 to BET in cells, we used FRAP (17) and compared our results with those obtained when using the pan-BD BETi, JQ1 (1). In FRAP, a molecule that inhibits binding of BET to chromatin increases the fluorescence recovery process as the protein is no longer bound to histones and can diffuse much faster. Hence, this approach provides a direct measure of in vivo displacement of BET from chromatin. SJ018 was more effective than JQ1 in increasing FRAP (Fig. 2E), suggesting that BD2-binding alone is enough to evict BRD4 from chromatin.

Figure 2.

Biophysical characterization of SJ018. A, Chemical structure, biophysical parameters, and growth inhibition data. B, SPR analysis with BRD2-BD2. C, BROMOscan analysis using 1 μmol/L SJ018. D, ITC analysis with BRD2-BD2. E, Comparison of BETi-modulated FRAP in U2OS cells using GFP-labeled BRD4. Top, photobleaching recovery in nucleus after a pulse of laser light (area circled in yellow). Bottom, quantitation of FRAP datasets following incubation of cells with SJ018 (gray), JQ1 (gold), or DMSO (blue).

Figure 2.

Biophysical characterization of SJ018. A, Chemical structure, biophysical parameters, and growth inhibition data. B, SPR analysis with BRD2-BD2. C, BROMOscan analysis using 1 μmol/L SJ018. D, ITC analysis with BRD2-BD2. E, Comparison of BETi-modulated FRAP in U2OS cells using GFP-labeled BRD4. Top, photobleaching recovery in nucleus after a pulse of laser light (area circled in yellow). Bottom, quantitation of FRAP datasets following incubation of cells with SJ018 (gray), JQ1 (gold), or DMSO (blue).

Close modal

Cytotoxicity studies comparing SJ018 and JQ1

Having established intracellular binding of SJ018 to BRD4, we evaluated the cytotoxicity of this molecule compared with JQ1 in a diverse panel of pediatric tumor cell lines (Fig. 2A). SJ018 was more active against all evaluated cells types, as compared with JQ1, except for the ALL lines, Nalm16, and Loucy. Toxicity was generally time-dependent, with activity increasing with longer drug exposure times. We used extended periods of treatment in these assays because it was not clear how long it would take for compounds that modified chromatin to induce their cytotoxic action. Consistent with our previous studies, little toxicity was observed with the THQ BETi scaffold against two nontumorigenic cell lines (BJ—fibroblast; LHCN-M2—myoblast), even after long drug exposure times, arguing that BETi demonstrate inherent antitumor selectivity.

Modulation of gene expression by SJ018 and JQ1

To assess the molecular consequences of selective BD2 inhibition, we exposed three different and diverse pediatric tumor cell lines (MV4–11—AML; HDMB03—Group 3 medulloblastoma (MB); Kelly—neuroblastoma) to SJ018 or JQ1 at six different concentrations (25–4,000 nmol/L), and performed microarray analyses after 3 hours. Data were pooled across cell lines and normalized to eliminate cell line–specific effects. Genes differentially expressed relative to DMSO, following drug treatment at each concentration, were identified using ANOVA (FDR < 0.05). Consequently, data obtained from these three experiments were independent of cell type and better represented the most consistent changes of expression induced by each drug and the respective dose.

Changes in transcription following treatment with SJ018 occurred at significantly lower concentrations than that observed with JQ1. At the lowest dose tested (25 nmol/L), 628 of the 651 differences (97%) were exclusively due to SJ018, 20 were common (3%), and only 3 (<1%) were unique to JQ1 (Fig. 3A). Importantly, it required 100 nmol/L JQ1 to induce statistically significant changes in most (545/628 = 87%) of the genes perturbed by SJ018 at 25 nmol/L. At 100 nmol/L, JQ1 induced changes in a total of 2,042 genes, indicating that potency alone, as opposed to BD-selectivity, was unlikely to account for the differential gene expression pattern associated with this compound. As the dose of each agent increased, the pattern of gene expression changes converged. At 4,000 nmol/L, both drugs differentially expressed 68% of all perturbed genes (6,917/10,216).

Figure 3.

Microarray studies comparing the changes in gene expression induced after 3 hours by increasing doses of SJ018 or JQ1 across three pediatric cancer cell lines (MV4–11, Kelly, and HDMB03). A, Venn diagrams depicting the number of significant changes in gene expression and the number of transcripts modulated over the dose range (FDR < 0.05). B, Scatter plots comparing all gene changes between the two drugs. All datasets were normalized to gene expression obtained from treatment of the relevant cell line with DMSO.

Figure 3.

Microarray studies comparing the changes in gene expression induced after 3 hours by increasing doses of SJ018 or JQ1 across three pediatric cancer cell lines (MV4–11, Kelly, and HDMB03). A, Venn diagrams depicting the number of significant changes in gene expression and the number of transcripts modulated over the dose range (FDR < 0.05). B, Scatter plots comparing all gene changes between the two drugs. All datasets were normalized to gene expression obtained from treatment of the relevant cell line with DMSO.

Close modal

Yet, it is important to note that there was still strong concordance between gene expression changes at each drug concentration when looking at all genes, not just those meeting the criteria for statistical significance (Fig. 3B). The concordance improved steadily (as reflected by Pearson R-squared values) with increasing molecule concentration, consistent with the hypothesis that SJ018 behaved more like a pan-BD inhibitor at higher concentrations. However, while the slope of the regression line at 25 nmol/L was not unity (slope = 1.64), the R-squared was 0.86, indicating that most of the genes that were differentially expressed by SJ018 were also perturbed by JQ1. Therefore, the extent to which SJ018 selectively perturbed BD2-responsive genes at low drug concentration was modest.

To further investigate changes in RNA expression induced by SJ018 and JQ1, the list of genes that significantly changed (FDR < 0.05) were divided into two groups: (i) “25 nmol/L,” which contained all genes altered at the lowest concentration tested (25 nmol/L), and was primarily driven by SJ018 and (ii) “≥200 nmol/L,” which contained all genes perturbed by SJ018, JQ1, or both at concentrations greater than or equal to 200 nmol/L (Supplementary Tables S5 and S6). Analysis using Enrichr (19, 26) indicated that the 25 nmol/L gene set was significantly enriched for biological processes associated with modification of chromatin, namely, histone lysine acetyltransferase and histone deacetylase activities, as well as genes involved in RNA transcription. In contrast, the ≥200 nmol/L gene set was significantly enriched for functions associated with the cell cycle, DNA repair and replication, RNA metabolism (ncRNA, rRNA, tRNA), stress response, and ubiquitin-associated proteolysis. This also included chromatin modification genes with methyltransferase activity. Furthermore, analysis of both gene sets predicted alteration of two different TF protein–protein interaction networks (Supplementary Tables S7 and S8). Both contained TP53 and FOXP3, but the 25 nmol/L set contained histone acetyltransferases (CLOCK, HDAC8), while the ≥200 nmol/L set contained MYC, and other stimulus-response genes (ESR1, ESR2, STAT1, SMAD2).

To better explore the differences between the two drugs, a two-factor (compound and concentration) ANOVA was performed on the batch-adjusted datasets after removing the untreated and DMSO samples. This analysis identified 382 transcripts (291 unique genes) that were differentially expressed between SJ018 and JQ1 (FDR < 0.05). The magnitude of the ratio of expression (fold changes) in this set was small and ranged from 0.86 to 1.25 (Supplementary Table S9). The top DAVID (27) function enrichment cluster contained the terms “Transcription,” “Transcription regulation,” and “DNA-binding” (Supplementary Table S10). “DNA Methylation and Transcriptional Repression Signaling” was the top pathway returned by Ingenuity Pathway Analysis (IPA) Canonical Pathway (Supplementary Fig. S2; ref. 28). Moreover, the top gene identified by IPA as an upstream regulator was CLOCK, which was also identified earlier in the comparison between genes expressed at low and high drug concentrations.

ChIP-seq studies with SJ018 and JQ1

To further explore differences between the epigenetic changes induced by SJ018 and JQ1, we undertook ChIP-seq studies (29) on drug-treated MV4–11 cells to identify loci across the genome that are perturbed by BETi exposure. H3K27ac, a marker for transcriptionally active chromatin (30), was monitored as microarray showed BETi induced both activation and repression of gene expression. ChIP-seq experiments showed high concordance between triplicate samples, and principal component analysis showed clear delineation of BETi treatment groups (Supplementary Figs. S3 and S4).

Interestingly, both compounds at 25 nmol/L caused a similar change in the number of H3K27ac sites that demonstrated either a decrease, or increase, in ChIP read enrichment, relative to DMSO (Fig. 4A; 1,259 vs. 1,359 H3K27ac sites at FDR < 0.01 for SJ018 and JQ1, respectively). Moreover, JQ1 induced 71.2% more changes at 1 μmol/L (7,553 vs. 12,952 H3K27ac sites at FDR < 0.01 for SJ018 and JQ1, respectively), even though both compounds showed a similar number of differentially expressed genes and higher concordance at this concentration (Fig. 3). These observations indicate that JQ1 disrupts chromatin to a greater extent than SJ018.

Figure 4.

ChIP-seq experiments identifying areas of active chromatin in MV4–11 cells following 3 hours treatment with either SJ018 or (+)-JQ1. A, Number of H3K27ac sites perturbed by SJ018 or JQ1 at 25 nmol/L and 1 μmol/L. B, Histograms indicating that both SJ018 and JQ1 exhibit consistent, but more pronounced changes in H3K27ac activated and repressed sites at 1 μmol/L versus 25 nmol/L. C, ChIP-seq H3K27ac read enrichment plot at the NT5C2 gene locus, where chromatin was found to be selectively activated by JQ1. D, ChIP-seq H3K27ac read enrichment plots at the LINC01565 and RPN1 gene loci, where chromatin was found to be selectively repressed by SJ018.

Figure 4.

ChIP-seq experiments identifying areas of active chromatin in MV4–11 cells following 3 hours treatment with either SJ018 or (+)-JQ1. A, Number of H3K27ac sites perturbed by SJ018 or JQ1 at 25 nmol/L and 1 μmol/L. B, Histograms indicating that both SJ018 and JQ1 exhibit consistent, but more pronounced changes in H3K27ac activated and repressed sites at 1 μmol/L versus 25 nmol/L. C, ChIP-seq H3K27ac read enrichment plot at the NT5C2 gene locus, where chromatin was found to be selectively activated by JQ1. D, ChIP-seq H3K27ac read enrichment plots at the LINC01565 and RPN1 gene loci, where chromatin was found to be selectively repressed by SJ018.

Close modal

As consistent activation or repression of a site is an indicator of an on-target effect, both molecules demonstrated near perfect perturbation: 952 of 953 (99.9%) of SJ018-modulated sites and 1,171 of 1,173 (99.8%) of JQ1-modulated sites. Moreover, we observed consistently stronger perturbations at 1 μmol/L compared with 25 nmol/L for both drugs: 902 of 952 (94.7%) of SJ018-modulated sites and 1,071 of 1,171 (91.4%) of JQ1-modulated sites (Fig. 4B). However, with SJ018 at 25 nmol/L, a larger percentage of sites showed decreases in H3K27ac enrichment versus increases, as compared with JQ1 (59.5% vs. 38.7%). At 1 μmol/L of each compound, this disparity was reduced (58.8% vs. 53.4%). Examples where chromatin was found to be selectively activated by JQ1 or repressed by SJ018 include the NT5C2 gene locus and LINC01565 and RPN1 gene loci, respectively (Fig. 4C and D). Finally, an analysis of TF motif enrichment at H3K27ac sites indicated that the E26 transformation-specific family of TFs (ETV6, ELK3, ELF5, ERG, GABPA, etc.), RUNX family TFs (RUNX2 and RUNX3), AP-2 family TFs (TFAP2A, TFAP2B, and TFAP2C), and ZIC family TFs (ZIC1, ZIC3, and ZIC4) demonstrated the greatest increases. However, at both 25 nmol/L and 1 μmol/L, JQ1 sites that exhibited increased H3K27ac enrichment were also highly enriched for CCAAT-enhancer-binding family TF motifs (CEBPB, CEBPE, CEBPG, and CEBPD). This was not observed with SJ018. In summary, these ChIP-seq experiments suggest that SJ018 induces greater selectivity with respect to chromatin modification than JQ1.

Design, synthesis, and biological activity of a metabolically stable, PK-optimized, BD2-selective BETi

While SJ018 demonstrated excellent selectivity for BD2 versus BD1, it contained significant liabilities that would preclude its use in vivo. These were eliminated by replacement of the ester on the A ring with a nitrile group, and substitution of the m-phenylacetamide (D ring) with 2,3-pyrazole. The resulting molecule, SJ432 (Fig. 5A), maintained potency and selectivity for BD2 versus BD1 in the TR-FRET assay for both BRD2 and BRD4 (6 nmol/L and 2 nmol/L, respectively, >80-fold more potent that that seen with BD1), was considerably more water soluble than SJ018 (∼40 μmol/L vs. <0.1 μmol/L), and was stable in biological samples (Supplementary Table S11 and S12). BD2-selectivity toward BRD2, 3, 4, and BRDT was independently confirmed using BROMOScan and SPR (Supplementary Table S13 and S14; Supplementary Fig. S5–S7). The cytotoxicity of SJ432 toward MV4-11 was within <3-fold of SJ018 (8 nmol/L vs. 3 nmol/L), and almost 2-fold higher than JQ1; in contrast, the enantiomer of SJ432, SJ432-neg, was >42-fold less potent (Supplementary Fig. S8).

Figure 5.

Crystal structures of the BRD2 complexes with SJ432. A, SJ432 chemical structure and indicated selectivity for BD2 with BRD2 and BRD4. B, BRD2-BD1: SJ432 (orange). C, BRD2-BD2: SJ432 (blue). D, BRD2-BD1: acetylated histone H4 peptide (magenta, PDB 2DVQ). E, BRD2-BD2: acetylated histone H4 peptide (green, PDB 2E3K). F, Overlay of the BRD2-BD1:SJ432 complex (orange/yellow) and BRD2-BD2 (blue/cyan).

Figure 5.

Crystal structures of the BRD2 complexes with SJ432. A, SJ432 chemical structure and indicated selectivity for BD2 with BRD2 and BRD4. B, BRD2-BD1: SJ432 (orange). C, BRD2-BD2: SJ432 (blue). D, BRD2-BD1: acetylated histone H4 peptide (magenta, PDB 2DVQ). E, BRD2-BD2: acetylated histone H4 peptide (green, PDB 2E3K). F, Overlay of the BRD2-BD1:SJ432 complex (orange/yellow) and BRD2-BD2 (blue/cyan).

Close modal

Structural analysis of SJ432 binding to BRD2-BD1 and BRD2-BD2

To better understand the basis of BD-selectivity for SJ432, we undertook X-ray crystallography of the complexes with isolated BRD2-BD1 and BRD2-BD2 domains. Following previous studies (31, 32), we obtained well-diffracting crystals from constructs encompassing residues 67–200 of BD1 and 348–455 of BD2, and determined high-quality complex structures at 1.5 Å and 1.05 Å, respectively (Supplementary Table S15; Supplementary Fig. S9; PDB 2DVQ and PDB 2E3K). SJ432 binds BD1 and BD2 similarly (Fig. 5B and C) and the three component moieties each engage three distinct pockets. The pyrazole moiety occupies an open pocket bounded by Trp97/370 and Leu108/381. The ring is oriented such that the two nitrogen atoms are exposed to the solvent, but the ring is flipped by 180° due to crystal packing in the one of the three BD1 complex structures in the asymmetric unit, reflecting the loose interactions of this moiety (Supplementary Fig. S10).

The central THQ moiety binds within a tight conserved hydrophobic pocket bounded by Pro98/371 (BD1/BD2), Phe99/372, and Ile162/Val435 on one side, and Val103/376, Leu108/381, and Leu110/383 on the opposite side. The methyl group at the C2 position of THQ ring contributes to the latter interactions, but also extends out to make additional van der Waals interactions with Tyr113/386 and Tyr155/428. The oxygen atom of the acetyl group at the N1 position forms a hydrogen bond with the ND2 nitrogen of Asn156/429 and the methyl group makes van der Waals interactions with Ile162/Val435, Val103/376, Pro98/371, and Phe99/372. Comparison with the published structures of BD1 and BD2 in complex with their cognate peptides reveals that the THQ moiety largely occupies the conserved binding pocket for the acetylated lysine side chain in which the terminal acetyl oxygen forms a hydrogen bond with the ND2 nitrogen of Asn156/429 (Fig. 5D and E).

The 4-(methylamino) benzonitrile moiety of SJ432 also binds to a conserved pocket bounded by Trp97/370, Pro98/371, Asp161/434, Met165/438, and Ile162/Val435. In addition, there is a hydrogen bond interaction between the linker nitrogen atom and the main chain carbonyl oxygen of Leu108/381 that is mediated by two bridging water molecules. However, the moiety adopts a small but significantly different conformation that suggests a tighter binding in the BD2 complex (Fig. 5F). Specifically, His433 forms a π–π stacking interaction with the phenyl ring in the BD2 complex whereas its counterpart Asp160 does not in the BD1 complex. The smaller Val435 in BD2, as compared with Ile162 in BD1, allows the moiety to penetrate deeper into the pocket. In addition, the conformations of Trp370 and Pro371 in BD2 create a slightly smaller pocket than Trp97 and Pro98 in BD1 that is also apparent when comparing the apo structures. One caveat to the His433/Asp160 difference is that a crystallographic ethylene glycol molecule occupies the His433 location in the BD1 structure that may prevent the equivalent interaction with Asp160. His433 forms a hydrogen bond with amide nitrogen atom of Val435 in the BD2 complex, and ethylene glycol and a water molecule make hydrogen bind interactions with amide nitrogen atoms of Asp161 and Ile162 in the BD1 complex (Supplementary Fig. S10). However, there are three complexes in the asymmetric unit of the BD1 structure and Asp160 adopts the same orientation in all three including one that has water molecules and not ethylene glycol at this location.

Our data indicate that the selectivity of SJ432 for BD2 over BD1 relates to the different peptide specificities. The almost identical interactions of the THQ core with BD1 and BD2, together with the loose interactions of the pyrazole moiety, suggest that the selectivity resides in the different binding modes of the benzonitrile moiety. Comparison with the peptide complexes of BD1 and BD2 reveals that the benzonitrile binds adjacent to the peptide binding locale. The specificity can be rationalized by Asp160 in BD1 and His433 in BD2 that both have central roles in binding the respective peptides, but have different contributions in binding the benzonitrile. Thus, whereas His433 forms the favorable π–π stacking interaction with the phenyl ring, a similar interaction would not expected for Asp160. We also note that His433 contributes to the water bridging hydrogen-bonding interaction between the linker nitrogen atom and the main chain carbonyl oxygen of Leu381 via a third water molecule (Fig. 5E). Computational analysis revealed that in the BD2/SJ432 complex, where the benzonitrile moiety more tightly engages the pocket, SJ432 demonstrates a lower intrinsic strain energy than in the BD1/SJ432 complex (28.9kcal/mol vs. 31.3 kcal/mol, respectively). Furthermore, ΔG for SJ432 with BD2 was calculated to be −67.11 kcal/mol, as compared with −65.21 kcal/mol for BD1. These structural and computational studies provide a detailed rationale for the selectivity of SJ432 with BD2.

Modulation of MYC expression and downstream target genes by SJ432

Because BETi are known to modulate MYC expression, we assessed the ability of SJ432 to alter c-MYC and MYCN in a series of neuroblastoma cell lines. Three lines were chosen (SK-N-AS—c-MYC–amplified; SK-N-SH—c-MYC–overexpressed, but not amplified; and IMR32—MYCN-amplified) and were exposed to SJ432 or JQ1. SJ432 was more effective than JQ1 at reducing levels of MYC protein in neuroblastoma cell lines (see Fig. 6; Supplementary Fig. S11). This occurs at both lower drug concentrations and over shorter time frames as compared with JQ1. It should also be noted that SJ432 did not induce a rebound effect in c-MYC expression, in contrast to JQ1 (e.g., see the 0.04 μmol/L data points in Fig. 6A and B, and 0.80 μmol/L value in Supplementary Fig. S11). In addition, c-MYC target genes that were either upregulated (CDKN1A-p21 and GADD45H) or downregulated (CCND2-cyclin D2 and ODC1) regulated following exposure to SJ432 were readily observed in SK-N-AS following treatment (Fig. 6C). Finally, the upregulation of HEXIM1, a validated target for BET inhibition, was more robust and durable in SJ432-treated cells (Fig. 6D).

Figure 6.

Biological activity of SJ432. A, Western blot analysis indicating loss of c-MYC protein in SK-N-SH cells after 16 hours exposure to either SJ432 or JQ1. B, Quantitation of c-MYC levels from A. C, Modulation of MYC targets by SJ432 in SK-N-AS cells. D, Upregulation of HEXIM1 by SJ432 or JQ1 in SK-N-AS cells. E, Loss of BRD4, and to a lesser extent BRD2, following exposure of cells to SJ432 or JQ1. F, Quantitation of BRD4 and BRD2 protein levels from E.

Figure 6.

Biological activity of SJ432. A, Western blot analysis indicating loss of c-MYC protein in SK-N-SH cells after 16 hours exposure to either SJ432 or JQ1. B, Quantitation of c-MYC levels from A. C, Modulation of MYC targets by SJ432 in SK-N-AS cells. D, Upregulation of HEXIM1 by SJ432 or JQ1 in SK-N-AS cells. E, Loss of BRD4, and to a lesser extent BRD2, following exposure of cells to SJ432 or JQ1. F, Quantitation of BRD4 and BRD2 protein levels from E.

Close modal

We also analyzed the levels of BRD2 and BRD4 in SK-N-AS cells following drug treatment. Both compounds induced the rapid loss of both proteins, with continued and extended suppression of BRD4 (Fig. 6E and F). Similar results were also observed in a c-MYC–amplified human pediatric Group 3 MB line, HDMB03 (33; Supplementary Fig. S12). Treatment of cells with MG132 did not reduce the loss of BRD4, indicating that this event is not mediated via the proteasome (Supplementary Fig. S13).

PK parameters for SJ432

Having developed SJ432, we assessed its PK parameters in female athymic/nude mice (Supplementary Table S16). These results indicated that the molecule was readily bioavailable, and yielded greater Cmax and AUC0→∞ values (∼4.5-fold and ∼4-fold, respectively), but reduced clearance (∼3.5-fold), as compared with JQ1. The differences in these values indicated that SJ432 levels would be maintained at therapeutic concentrations for longer than JQ1 in animals and should therefore be more effective in preclinical studies (see Supplementary Fig. S14). We also noted that the brain penetration for SJ432 was >100-fold lower than that for the pan BETi (Supplementary Table S16), arguing that SJ432 would not demonstrate the inherent neurotoxicity seen with JQ1. No serious adverse toxicity was seen in these animals and hence MTD studies were undertaken. These experiments indicated that 15 mg/kg was minimally toxic when given intraperitoneally using a daily dosing schedule for 14 consecutive days.

Antitumor activity of BD2-selective BETi

We assessed the activity of SJ432 toward the pediatric neuroblastoma xenograft SK-N-AS. Female CB17SCID mice bearing flank tumors, were dosed intraperitoneally with SJ432 daily for 14 days. We observed a pharmacologic dose response that resulted in significant tumor growth delays (Fig. 7A). Direct comparison of the activity of SJ432 (15 mg/kg) with JQ1 (50 mg/kg) resulted in a significant reduction in tumor volumes by the former, and increased animal survival, with median survivals of 31 (P = 0.0002 vs. control) and 26 (P = 0.0064 vs. control) days, respectively (Fig. 7B and C). More impressively however, no serious toxicities were observed in SJ432-treated mice, with essentially no loss in body weight for the duration of the study (Fig. 7D). This contrasts with studies with JQ1 where mice demonstrated morbidity, weight loss, and a failure to thrive.

Figure 7.

In vivo activity of SJ432. In all experiments, 10 mice were used per group. A, Flank SK-N-AS tumor volumes in cohorts of mice treated with increasing daily doses of SJ432 (5, 10, or 15 mg/kg). Statistical significance versus control tumors is indicated by asterisks (blue asterisk, 15 mg/kg vs. control; red asterisk, 10 mg/kg vs. control; green asterisk, 5 mg/kg vs. control). B, Comparison of tumor volumes of animals bearing SK-N-AS xenografts following treatment with either JQ1 (50 mg/kg; blue line) or SJ432 (15 mg/kg; red line). Asterisks above the data points indicate statistical significance as compared with control animals (red asterisk, 15 mg/kg SJ432 vs. control; blue asterisk, JQ1 vs. control). Asterisks (red) below the data points indicate statistical significance between JQ1- and SJ432-treated animals. C, Kaplan–Meier curves demonstrating increased survival of animals bearing SK-N-AS tumors treated with SJ432 (15 mg/kg; red line) as compared with JQ1 (50 mg/kg; blue line), or control mice (black line). Median survivals were 22.5, 26, and 31 days for control, JQ1-treated, and SJ432-treated mice, respectively. Significance for these datasets using log rank test were as follows: SJ432 vs. control, P < 0.0007; JQ1 vs. control, P = 0.067; SJ432 vs. JQ1, P = 0.0791. D, Weights of mice from the study presented in B and C. E, Western blot analyses indicating the levels of c-MYC, HEXIM1, BRD4, and BRD2 protein in SK-N-AS xenografts harvested either 4 hours or 24 hours after SJ432 dosing. C, control, untreated tumors. F, Quantitation of the levels of protein in E. Data were corrected for gel loading differences using β-actin expression.

Figure 7.

In vivo activity of SJ432. In all experiments, 10 mice were used per group. A, Flank SK-N-AS tumor volumes in cohorts of mice treated with increasing daily doses of SJ432 (5, 10, or 15 mg/kg). Statistical significance versus control tumors is indicated by asterisks (blue asterisk, 15 mg/kg vs. control; red asterisk, 10 mg/kg vs. control; green asterisk, 5 mg/kg vs. control). B, Comparison of tumor volumes of animals bearing SK-N-AS xenografts following treatment with either JQ1 (50 mg/kg; blue line) or SJ432 (15 mg/kg; red line). Asterisks above the data points indicate statistical significance as compared with control animals (red asterisk, 15 mg/kg SJ432 vs. control; blue asterisk, JQ1 vs. control). Asterisks (red) below the data points indicate statistical significance between JQ1- and SJ432-treated animals. C, Kaplan–Meier curves demonstrating increased survival of animals bearing SK-N-AS tumors treated with SJ432 (15 mg/kg; red line) as compared with JQ1 (50 mg/kg; blue line), or control mice (black line). Median survivals were 22.5, 26, and 31 days for control, JQ1-treated, and SJ432-treated mice, respectively. Significance for these datasets using log rank test were as follows: SJ432 vs. control, P < 0.0007; JQ1 vs. control, P = 0.067; SJ432 vs. JQ1, P = 0.0791. D, Weights of mice from the study presented in B and C. E, Western blot analyses indicating the levels of c-MYC, HEXIM1, BRD4, and BRD2 protein in SK-N-AS xenografts harvested either 4 hours or 24 hours after SJ432 dosing. C, control, untreated tumors. F, Quantitation of the levels of protein in E. Data were corrected for gel loading differences using β-actin expression.

Close modal

An analysis of tumor material harvested from SJ432-treated mice, either 4 or 24 hours after the last dose of a 14-day consecutive treatment, demonstrated a reduction in the level of c-MYC, loss of BRD4, and upregulation of HEXIM1 (Fig. 7E). At 24 hours after the last dose of drug, BRD4 was essentially undetectable and HEXIM1 was increased by approximately 5.4-fold (Fig. 7F). Levels of BRD2 were transiently elevated in the tumor harvested at 4 hours postdosing, but essentially returned to control levels after 24 hours. These results are consistent with our in vitro studies using this same drug and cell line (Fig. 6D and E).

All BET proteins contain two BDs, indicating that the duplication and architecture of these domains are required for their biological function. However, the previous targeting of BET has focused on the development of small molecules that interact with these proteins without regard to BD-selectivity. Because evidence in the literature indicates differential effects of the different BD domains (7–11), we have generated BD2-selective BETi based upon a THQ scaffold. Using SJ018 as a tool compound, we evaluated the molecular and cellular consequence of selective inhibition of the BD2 domain. Microarray studies indicated that modulation of expression of different gene sets was observed in comparison with JQ1 in three tumor cell lines of different histologic origin (AML, neuroblastoma, and MB). By selecting a diverse panel of lines for these analyses, the derived datasets would be agnostic of cell of origin and more reflective of drug activity. Further validation using ChIP-seq confirmed changes in H3K27ac in chromatin, indicating that these molecular approaches allowed discrimination and identification of genes and pathways that are specifically regulated by BD2 interaction with histones.

During the course of this project, another THQ-based BETi, IBET726, was reported (34, 35). This molecule demonstrated <100 nmol/L potency and 5-fold and 8-fold selectivity for BD2 versus BD1 in BRD2 and BRD4, respectively. In addition, recent work by Law and colleagues (36) identified compounds with comparable BD2-selectivity with SJ018. However, no cytotoxicity or in vivo data was presented for the series of compounds. While molecule potency, rather than BD-selectivity, may have been used for compound selection and design by these investigators, our results argue that the former is not the only consideration that should be evaluated for development of this class of inhibitors.

In growth inhibition assays, we observed that the THQ analogs were universally active (Fig. 2A), and generally more potent than JQ1. This occurred in essentially all tumor histotypes and indicated that BD2-selective compounds have broad spectrum activity. It is not known whether this is a consequence of the selective targeting of BD2, an inherent susceptibility of the pediatric lines to such molecules, or a superior chemicophysical and biological profile associated with the THQ scaffold. However, our results argue that the development of such agents for pediatric tumors is warranted. While numerous clinical trials with BETi are ongoing in adults diagnosed with a variety of malignancies, we are aware of only one being undertaken in children (NCT03936465). Coupled with our in vivo data demonstrating the efficacy of SJ432 with minimal toxicity, we urge the medical community to consider the use of BD2-selective BETi in pediatric patients.

The THQ analogs induced cytotoxicity in pediatric lines and potently suppressed c-MYC and MYCN expression. Whether these effects are directly mediated via the inhibition of these cellular pathways is unclear, but our data argue that BD2-selective compounds are more efficacious than JQ1 at modulating MYC protein levels. Furthermore, we observed no reexpression of c-MYC protein, consistent with the lack of a rebound effect. Because upregulation of BET proteins following exposure to BETi has been observed (37), it is likely that this may act as a potential mechanism of drug resistance, whereby cells transiently overcome the effects of BET inhibition by enhanced transcription/translation of genes encoding these proteins (e.g., c-MYC). In the neuroblastoma lines we used, we observed an upregulation of c-MYC following JQ1 exposure, but not with SJ432. Presumably, the former arises from increased expression of BETs following JQ1 treatment, but we conclude that resistance to THQ-derived molecule is unlikely to occur by this same mechanism. We also noted that BRD4 was rapidly lost following SJ432 exposure, whereas BRD2 was not. It is not known whether the modulation of BET proteins by this drug is important for biological activity, but THQs may have inherent targeting of BRD4. This would explain the potent antitumor activity of these drugs with minimal off-target toxicity.

In summary, our data indicate that dual targeting of BD domains in BET is not necessary to induce cytotoxicity and antitumor activity. Hence, the use of BD-specific assays, rather than those that assess binding to BET proteins as a whole, should be sufficient to identify potent, novel BETi. We have exploited these approaches and developed unique reagents that are effective in vitro and in vivo. These data suggest that selective BD targeting should allow the design of more effective, and potentially, less toxic BETi with antitumor activity.

P.J. Slavish reports a patent for US20200079739A1 pending. W.R. Shadrick has a patent for US20200079739A1 pending, and support for this research came from the donors to St. Jude Children's Research Hospital (SJCRH), and the pending patent will belong to SJCRH. J.E. Price reports a patent for US20200079739A1 pending. B.M. Young reports a patent for US20200079739A1 pending. R.K. Guy reports a patent for US20200079739A1 pending. A.A. Shelat has a patent for tetrahydroquinoline-based bromodomain inhibitors pending. No potential conflicts of interest were disclosed by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

P.J. Slavish: Conceptualization, data curation, formal analysis, investigation, methodology, writing-original draft, writing-review and editing. L. Chi: Data curation, formal analysis, investigation. M.-K. Yun: Data curation, formal analysis, investigation, visualization, writing-original draft, writing-review and editing. L. Tsurkan: Data curation, formal analysis, investigation. N.E. Martinez: Data curation, formal analysis, investigation. B. Jonchere: Data curation, formal analysis, investigation. S.C. Chai: Data curation, formal analysis, investigation. M. Connelly: Data curation, formal analysis, investigation. M.B. Waddell: Data curation, formal analysis, investigation. S. Das: Data curation, formal analysis, investigation. G. Neale: Data curation, formal analysis, investigation, methodology. Z. Li: Data curation, investigation. W.R. Shadrick: Conceptualization, data curation, formal analysis, investigation. R.R. Olsen: Data curation, formal analysis, investigation. K.W. Freeman: Conceptualization, formal analysis, supervision, investigation. J.A. Low: Data curation, formal analysis, investigation. J.E. Price: Data curation, investigation. B.M. Young: Conceptualization, data curation, formal analysis, supervision, investigation, methodology. N. Bharatham: Data curation, formal analysis, investigation. V.A. Boyd: Data curation, investigation. J. Yang: Data curation, investigation, methodology. R.E. Lee: Conceptualization, supervision, methodology. M. Morfouace: Data curation, formal analysis, investigation. M.F. Roussel: Conceptualization, formal analysis, supervision, project administration, writing-review and editing. T. Chen: Formal analysis, supervision, methodology. D. Savic: Data curation, formal analysis, investigation, methodology, writing-review and editing. R.K. Guy: Conceptualization, formal analysis, supervision, methodology, project administration, writing-review and editing. S.W. White: Conceptualization, formal analysis, supervision, visualization, methodology, writing-original draft, project administration, writing-review and editing. A.A. Shelat: Conceptualization, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. P.M. Potter: Conceptualization, data curation, formal analysis, supervision, funding acquisition, writing-original draft, project administration, writing-review and editing.

We would like to acknowledge Dr. Aaron Pitre from the Cellular Imaging Shared Resource at St. Jude Children's Research Hospital for his assistance with the FRAP experiments. High-resolution microscopy images were acquired in the Cell & Tissue Imaging Center at St. Jude. This work was supported, in part, by NIH grants (P01 CA096832 to M.F. Roussel; R01 CA225945 to A.A. Shelat and P.M. Potter), a Cancer Center Core grant (NCI, P30 CA021765), and by the American Lebanese Syrian Associated Charities (ALSAC).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Filippakopoulos
P
,
Qi
J
,
Picaud
S
,
Shen
Y
,
Smith
WB
,
Fedorov
O
, et al
Selective inhibition of BET bromodomains
.
Nature
2010
;
468
:
1067
73
.
2.
Johnstone
RW
. 
Histone-deacetylase inhibitors: novel drugs for the treatment of cancer
.
Nat Rev Drug Discov
2002
;
1
:
287
99
.
3.
Cochran
AG
,
Conery
AR
,
Sims
RJ
 III
. 
Bromodomains: a new target class for drug development
.
Nat Rev Drug Discov
2019
;
18
:
609
28
.
4.
Ferri
E
,
Petosa
C
,
McKenna
CE
. 
Bromodomains: structure, function and pharmacology of inhibition
.
Biochem Pharmacol
2016
;
106
:
1
18
.
5.
Zeng
L
,
Zhou
M-M
. 
Bromodomain: an acetyl-lysine binding domain
.
FEBS Lett
2002
;
513
:
124
28
.
6.
Bharatham
N
,
Slavish
PJ
,
Shadrick
WR
,
Young
BM
,
Shelat
AA
. 
The role of ZA channel water-mediated interactions in the design of bromodomain-selective BET inhibitors
.
J Mol Graph Model
2018
;
81
:
197
210
.
7.
Lamonica
JM
,
Deng
W
,
Kadauke
S
,
Campbell
AE
,
Gamsjaeger
R
,
Wang
H
, et al
Bromodomain protein Brd3 associates with acetylated GATA1 to promote its chromatin occupancy at erythroid target genes
.
Proc Natl Acad Sci U S A
2011
;
108
:
1
10
.
8.
Shang
E
,
Nickerson
HD
,
Wen
D
,
Wang
X
,
Wolgemuth
DJ
. 
The first bromodomain of Brdt, a testis-specific member of the BET sub-family of double-bromodomain-containing proteins, is essential for male germ cell differentiation
.
Development
2007
;
134
:
3507
15
.
9.
Shi
J
,
Wang
Y
,
Zeng
L
,
Wu
Y
,
Deng
J
,
Zhang
Q
, et al
Disrupting the interaction of BRD4 with diacetylated twist suppresses tumorigenesis in basal-like breast cancer
.
Cancer Cell
2014
;
25
:
210
25
.
10.
Umehara
T
,
Nakamura
Y
,
Wakamori
M
,
Ozato
K
,
Yokoyama
S
,
Padmanabhan
B
. 
Structural implications for K5/K12-di-acetylated histone H4 recognition by the second bromodomain of BRD2
.
FEBS Lett
2010
;
584
:
3901
08
.
11.
Vollmuth
F
,
Geyer
M
. 
Interaction of propionylated and butyrylated histone H3 lysine marks with Brd4 bromodomains
.
Angew Chem Int Ed Engl
2010
;
49
:
6768
72
.
12.
Vollmuth
F
,
Blankenfeldt
W
,
Geyer
M
. 
Structures of the dual bromodomains of the P-TEFb-activating protein Brd4 at atomic resolution
.
J Biol Chem
2009
;
284
:
36547
56
.
13.
Shadrick
WR
,
Slavish
PJ
,
Chai
SC
,
Waddell
B
,
Connelly
M
,
Low
JA
, et al
Exploiting a water network to achieve enthalpy-driven, bromodomain-selective BET inhibitors
.
Bioorg Med Chem
2018
;
26
:
25
36
.
14.
Picaud
S
,
Wells
C
,
Felletar
I
,
Brotherton
D
,
Martin
S
,
Savitsky
P
, et al
RVX-208, an inhibitor of BET transcriptional regulators with selectivity for the second bromodomain
.
Proc Natl Acad Sci U S A
2013
;
110
:
19754
59
.
15.
Gosmini
R
,
Nguyen Van
L
,
Toum
J
,
Simon
C
,
Brusq Jean-Marie
G
,
Krysa
G
, et al
The discovery of I-BET726 (GSK1324726A), a potent tetrahydroquinoline ApoA1 up-regulator and selective BET bromodomain inhibitor
.
J Med Chem
2014
;
57
:
8111
31
.
16.
Kati
W
. 
ABBV-744: a first-in-class highly BDII-selective BET bromodomain inhibitor [abstract]
.
In
:
Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14–18
;
Chicago, IL. Philadelphia (PA)
:
AACR
; 
2018
.
Abstract nr DDT01-05
.
17.
Philpott
M
,
Rogers
CM
,
Yapp
C
,
Wells
C
,
Lambert
JP
,
Strain-Damerell
C
, et al
Assessing cellular efficacy of bromodomain inhibitors using fluorescence recovery after photobleaching
.
Epigenetics Chromatin
2014
;
7
:
14
.
18.
Irizarry
RA
,
Hobbs
B
,
Collin
F
,
Beazer-Barclay
YD
,
Antonellis
KJ
,
Scherf
U
, et al
Exploration, normalization, and summaries of high density oligonucleotide array probe level data
.
Biostatistics
2003
;
4
:
249
64
.
19.
Chen
EY
,
Tan
CM
,
Kou
Y
,
Duan
Q
,
Wang
Z
,
Meirelles
GV
, et al
Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool
.
BMC Bioinformatics
2013
;
14
:
128
.
20.
Huang da
W
,
Sherman
BT
,
Lempicki
RA
. 
Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources
.
Nat Protoc
2009
;
4
:
44
57
.
21.
Savic
D
,
Ramaker
RC
,
Roberts
BS
,
Dean
EC
,
Burwell
TC
,
Meadows
SK
, et al
Distinct gene regulatory programs define the inhibitory effects of liver X receptors and PPARG on cancer cell proliferation
.
Genome Med
2016
;
8
:
74
.
22.
Love
MI
,
Huber
W
,
Anders
S
. 
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.
23.
Bailey
TL
,
Machanick
P
. 
Inferring direct DNA binding from ChIP-seq
.
Nucleic Acids Res
2012
;
40
:
e128
.
24.
Afonine
PV
,
Grosse-Kunstleve
RW
,
Echols
N
,
Headd
JJ
,
Moriarty
NW
,
Mustyakimov
M
, et al
Towards automated crystallographic structure refinement with phenix.refine
.
Acta Crystallogr D Biol Crystallogr
2012
;
68
:
352
67
.
25.
Emsley
P
,
Lohkamp
B
,
Scott
WG
,
Cowtan
K
. 
Features and development of Coot
.
Acta Crystallogr D Biol Crystallogr
2010
;
66
:
486
501
.
26.
Kuleshov
MV
,
Jones
MR
,
Rouillard
AD
,
Fernandez
NF
,
Duan
Q
,
Wang
Z
, et al
Enrichr: a comprehensive gene set enrichment analysis web server 2016 update
.
Nucleic Acids Res
2016
;
44
:
W90
7
.
27.
Huang
DW
,
Sherman
BT
,
Tan
Q
,
Collins
JR
,
Alvord
WG
,
Roayaei
J
, et al
The DAVID Gene Functional Classification Tool: a novel biological module-centric algorithm to functionally analyze large gene lists
.
Genome Biol
2007
;
8
:
R183
.
28.
Kramer
A
,
Green
J
,
Pollard
J
 Jr
,
Tugendreich
S
. 
Causal analysis approaches in Ingenuity Pathway Analysis
.
Bioinformatics
2014
;
30
:
523
30
.
29.
Johnson
DS
,
Mortazavi
A
,
Myers
RM
,
Wold
B
. 
Genome-wide mapping of in vivo protein-DNA interactions
.
Science
2007
;
316
:
1497
502
.
30.
Creyghton
MP
,
Cheng
AW
,
Welstead
GG
,
Kooistra
T
,
Carey
BW
,
Steine
EJ
, et al
Histone H3K27ac separates active from poised enhancers and predicts developmental state
.
Proc Natl Acad Sci U S A
2010
;
107
:
21931
6
.
31.
Chung
C-w
,
Dean
AW
,
Woolven
JM
,
Bamborough
P
. 
Fragment-based discovery of bromodomain inhibitors Part 1: inhibitor binding modes and implications for lead discovery
.
J Med Chem
2012
;
55
:
576
86
.
32.
Tripathi
S
,
Mathur
S
,
Deshmukh
P
,
Manjula
R
,
Padmanabhan
B
. 
A novel nhenanthridionone based scaffold as a potential inhibitor of the BRD2 bromodomain: Crystal structure of the complex
.
PLoS One
2016
;
11
:
e0156344
.
33.
Milde
T
,
Lodrini
M
,
Savelyeva
L
,
Korshunov
A
,
Kool
M
,
Brueckner
LM
, et al
HD-MB03 is a novel Group 3 medulloblastoma model demonstrating sensitivity to histone deacetylase inhibitor treatment
.
J Neurooncol
2012
;
110
:
335
48
.
34.
Chaidos
A
,
Caputo
V
,
Gouvedenou
K
,
Liu
B
,
Marigo
I
,
Chaudhry
MS
, et al
Potent antimyeloma activity of the novel bromodomain inhibitors I-BET151 and I-BET762
.
Blood
2014
;
123
:
697
705
.
35.
Zhao
Y
,
Yang
C-Y
,
Wang
S
. 
The making of I-BET762, a BET bromodomain inhibitor now in clinical development
.
J Med Chem
2013
;
56
:
7498
500
.
36.
Law
RP
,
Atkinson
SJ
,
Bamborough
P
,
Chung
CW
,
Demont
EH
,
Gordon
LJ
, et al
Discovery of tetrahydroquinoxalines as Bbromodomain and extra-terminal domain (BET) inhibitors with selectivity for the second bromodomain
.
J Med Chem
2018
;
61
:
4317
34
.
37.
Lu
J
,
Qian
Y
,
Altieri
M
,
Dong
H
,
Wang
J
,
Raina
K
, et al
Hijacking the E3 ubiquitin ligase cereblon to efficiently target BRD4
.
Chem Biol
2015
;
22
:
755
63
.