Differential BET bromodomain inhibition by dihydropteridinone and pyrimidodiazepinone kinase inhibitors

Abstract

BRD4 and other members of the BET bromodomain family of proteins are promising epigenetic targets for the development of novel therapeutics. Among the reported BRD4 inhibitors are dihydropteridinones and benzopyrimidodiazepinones originally designed to target the kinases PLK1, ERK5 and LRRK2. While these kinase inhibitors were identified as BRD4 inhibitors, little is known about their binding potential and structural details of interaction with the other BET bromodomains. We comprehensively characterized a series of known and newly identified dual BRD4-kinase inhibitors against all 8 individual BET bromodomains. Detailed analysis of 23 novel cocrystal structures of BET-kinase inhibitor complexes in combination with direct binding assays and cell signaling studies revealed significant differences in molecular shape complementarity and inhibitory potential. Collectively, the data offer new insights into the action of kinase inhibitors across BET bromodomains, which may aid the development of drugs to inhibit certain BET proteins and kinases differentially.

Graphical Abstract

INTRODUCTION

Bromodomain and extraterminal (BET) proteins constitute a subfamily of bromodomain (BRD)-containing proteins that recognize and interact with histones and other proteins through a conserved acetyllysine (KAc) binding site. BET proteins comprise BRD2, BRD3, BRD4 and BRDT, each containing two N-terminal BRDs in tandem (BD1 and BD2) (Supplementary Fig. S1). Both individual bromodomains show high sequence homology with 73% and 65% identity for BD1 and BD2 across BETs, respectively. By contrast, intra-domain sequence conservation of BD1 and BD2 is relatively low (sequence identity 36–43%).¹ BD1 mainly interacts with acetyl-lysines of histones and regulates epigenetic mechanisms, whereas BD2 also interacts with acetylated lysine residues of other proteins and mediates sequence specific DNA binding interactions.^2–4

Many small molecule BET inhibitors have been reported in the last decade, predominantly targeting BRD4 for cancer and immunoinflammation.⁵ Currently, 11 BRD4 inhibitors are in clinical trials for blood cancers and solid tumors as well as diabetes and pulmonary arterial hypertension.⁶ The majority of reported BRD4 inhibitors target BD1 and BD2 indiscriminately. Recently, inhibitors that selectively target BD2 over BD1 revealed an important role of BD2 in the recruitment of BET proteins for gene expression, and may provide more effective therapeutic options by avoiding the toxicity associated with inhibition of BD1.^7–9 Beyond prototypical BRD4 inhibitors, a series of compounds capable of concurrently inhibiting the KAc site of BRD4 and the ATP site of certain kinases has been discovered.^{10, 11} Among these dual BRD4-kinase inhibitors are the diaminopyrimidine JAK2 inhibitors fedratinib and TG101209¹², the dihydropteridinone PLK1 inhibitors volasertib and BI2536^{13, 14}, and the benzopyrimidodiazipinones LRRK2-IN-1¹⁵ and ERK5-IN-1.^{16, 17} While these kinase inhibitors have been studied in the context of BRD4 inhibition, their interaction with other BET bromodomains is unknown. Such information, however, may aid the development of BET inhibitors designed to simultaneously target certain BET bromodomains and kinases differentially, depending on the disease application.

Here, we comprehensively characterized a series of dual BRD4-kinase inhibitors comprising dihydropteridinone, pyrimidodiazipinone and benzopyrimidodiazipinone chemophores against all 8 BET bromodomains using a combination of large-scale cocrystal structure determination, direct binding, cell inhibition and binding energy predictions. A total of 23 novel crystal structures of BET bromodomains liganded with kinase inhibitors were determined and analyzed in detail. Combined, the data reveal significant differences in molecular shape complementarity, binding affinity, and inhibitory potential of kinase inhibitors across BET bromodomains.

RESULTS

Assessment of binding affinity of kinase inhibitors for BRD4 and BRDT

Kinase inhibitors containing dihydropteridinone (BI2536, volasertib) or benzopyrimidodiazepinone (ERK5-IN-1, LRRK2-IN-1, XMD8-92) moieties were evaluated for inhibitory potential against all 8 isolated BET bromodomains (Fig. 1A). Another kinase inhibitor, the pyrimidodiazepinone PLK1 inhibitor Ro3280¹⁸, was discovered as a novel BET inhibitor during these studies. To accurately assess the binding affinity of compounds for isolated bromodomains, we employed differential scanning fluorimetry (DSF) and isothermal microcalorimetry (ITC) as primary methods. Orthogonal methods included microscale thermophoresis (MST) and a qPCR-based assay (performed by DiscoveRx). DSF, ITC and MST assays were performed with highly pure, crystallization-grade proteins. The first bromodomains of BRD4 (BRD4-1) and BRDT (BRDT-1) served to validate the different methods applied (Fig. 1, Table 1).

Figure 1: Binding potential of kinase inhibitors for BRD4-1 and BRDT-1.

(A) Chemical structures of the studied dihydropteridinone and pyrimidodiazepinone containing kinase inhibitors. (B-D) Correlation of thermostability and dissociation constants as determined by DSF, ITC, MST and qPCR assay. (E-G) Correlation of dissociation constants with each other. For statistical analysis, r is the respective Pearson’s correlation coefficient; asterisks indicate the significance of the two-tailed P value (α = 0.05). For graphical purposes, data were fit to semilogarithmic (B-D) or log-log lines (E-G). The dotted line indicates an idealized 1:1 relationship between the respective dissociation constants.

Table 1:

Binding potential of kinase inhibitors for BRD4-1 and BRDT-1 and cell growth inhibitory activity

	Binding potential								Cell growth inhibitory potential
Compound	BRD4-1				BRDT-1				MM1.S	HEK293T
	DSFa ΔTm (°C)	MSTb Kd (nM)	qPCRc Kd (nM)c	ITCd Kd (nM)d	DSFa ΔTm (°C)a	MSTb Kd (nM)b	qPCRc Kd (nM)c	ITCd Kd (nM)d	IC50 (nM)e
BI2536	10.5 ± 0.08	57 ± 11	24 ± 0.1	112 ± 5.0	6.3 ± 0.24	240 ± 40	105 ± 35	806 ± 31	1.2 ± 0.15	0.6 ± 0.14
Volasertib	8.6 ± 0.06	169 ± 37	77 ± 23	317 ± 17	5.0 ± 0.09	1,030 ± 203	113 ± 38	1700 ± 160	4.5 ± 0.25	1.1 ± 0.08
Ro3280	8.2 ± 0.06	260 ± 38	120 ± 14	267 ± 24	4.0 ± 0.07	1,900 ± 630	390 ± 28	1800 ± 70	5.7 ± 0.24	2.6 ± 0.14
LRRK2-IN-1	8.5 ± 0.07	69 ± 13	66 ± 30	194 ± 12	3.8 ± 0.09	700 ± 350	390 ± 28	1700 ± 90	530 ± 77	2,800 ± 1,400
ERK5-IN-1	7.3 ± 0.07	205 ± 45	69 ± 18	213 ± 11	2.3 ± 0.15	1,600 ± 290	500 ± 28	2100 ± 100	280 ± 81	> 10,000
XMD8-92	6.4 ± 0.06	800 ±150	205 ± 7.1	735 ± 38	2.9 ± 0.12	3,500 ± 960	675 ± 21	4200 ± 180	2,100 ± 1,800	> 10,000
SG3-179	13.5 ± 0.2	---	12 ± 2.9	54 ± 4.9	11.9 ± 0.15	---	---	59 ± 4.6	26 ± 3.4	140 ± 12
JQ1	11.2 ± 0.06	35 ± 11	8.2 ± 2.5	71 ± 6.6	7.2 ± 0.06	260 ± 77	22 ± 0.7	188 ± 7.0	33 ± 4.9	107 ± 11

^aSingle experiment in quadruplicate (± standard error of the mean, SEM);

^bAverage of three experiments (± standard deviation, SD);

^cAverage of two experiments (± SD);

^dSingle experiment (± SEM);

^eSingle experiment in hexaplicate (± SEM)

The thermal shift data obtained by DSF showed significant correlation with the dissociation constants (K_d) determined by ITC, MST and qPCR, confirming that DSF is a robust method to quickly assess the relative binding potential for a series of compounds (Fig. 1B–D). All compounds showed higher binding affinity for BRD4-1 over BRDT-1, BI2536 being the most potent inhibitor followed by LRRK2-IN-1, ERK5-IN-1, Ro3280 and XMD8-92. A more potent dual BET-kinase inhibitor, SG3-179, which specifically targets the tyrosine kinases JAK2, FLT3 and RET¹², served as a positive control along with the BET inhibitor JQ1 (Supplementary Fig. S2). The K_d values determined by qPCR assay were generally smaller than those obtained by ITC or MST (Fig. 1E–F), while ITC and MST data followed a near 1:1 relationship (Fig. 1G).

The discrepancies between the K_d values obtained by qPCR assay vs. direct binding studies may be due to differences in protein preparation and assay conditions. Combined, the results provided confidence in our binding studies using native, label-free protein to determine the relationship of dissociation constants, thermal shifts and crystal structure information.

Differential binding affinity of kinase inhibitors across BET bromodomains

Next, all individual bromodomains of the BET family (BETs) were subjected to binding studies by DSF and ITC (Fig. 2A–B, Supplementary Tables S1–S2). Both data sets showed differential binding potential of inhibitors across BETs, with BRDT-1 and BRD4-1 being the least and most sensitive bromodomains, respectively (Fig. 2A–B, left panel). Except for BRDT-2, the mean DSF and ITC values followed the same trend. BI2536, ERK5-IN-1 and LRRK2-IN-1 were the most potent BET ligands, XMD8-92 the least potent (Fig. 2A–B, right panel). JQ1 and SG3-179 showed significantly higher binding potential across BETs than any of the kinase inhibitors tested. Pearson’s correlation analysis revealed a statistically significant correlation between ITC and DSF values for the first (BD1) and second (BD2) bromodomains, BD1 being more sensitive for kinase inhibitors than BD2 (Fig. 2C). Most bromodomain-inhibitor interactions were predominantly enthalpy-driven with small changes in entropy (Supplementary Figs. S3–S4). However, significant unfavorable entropy contributions were observed for BRDT-1 and BRD2-1, while favorable entropy contributed in part to inhibitor interactions with BRD4-2 and BRD2-2. Using the mean values of binding potential as assessed by DSF and ITC, the resulting fold-change in affinity revealed distinct BET selectivity profiles for each inhibitor (Fig. 2D). Except for ERK5-IN-1, which was moderately selective for BD2, all inhibitors appeared to preferentially target BD1. High-affinity control compounds JQ1 and SG3-179 lacked selectivity for BD1 or BD2.

Figure 2: Binding affinity and selectivity of kinase inhibitors against BET bromodomains.

(A) Thermal shifts (ΔT_m) from DSF experiments plotted against bromodomains (left panel) and inhibitors (right panel). The black bars indicate the arithmetic means. (B) Same as (A) for K_d values from ITC experiments. The black bars indicate the geometric means. Tabulated data are shown in Supplementary Tables S1–S2 and ITC thermograms in Supplementary Fig. S4. (C) Pearson’s correlation of ΔT_m and K_d values for the first (red) and second (blue) bromodomains. (D) Fold change of ΔT_m (blue) and K_d (red) values relative to the mean values from (A) and (B).

Cellular activity of BET-kinase inhibitors

Cellular activity of inhibitors was assessed in multiple myeloma cell line MM1.S, which is highly sensitive to inhibition of BRD4, and HEK293T cells (Fig. 3). Compounds inhibiting both PLK1 and BRD4 (BI2536, volasertib and Ro3280) showed strongest inhibition of cell growth in both cell lines with IC50 values between 0.6 and 5.7 nM (Fig. 3A–B, Table 1). Treatment of cells with GSK461364, a PLK1 inhibitor with no BRD4 activity, resulted in similar low nanomolar IC₅₀ values, suggesting that inhibition of PLK1 is the major reason for the strong cell growth inhibitory activity by dual BET-PLK1 inhibitors. To confirm PLK1 inhibitory activity, phosphorylation of the PLK1 substrate TCTP (translationally controlled tumor protein) at S46 was monitored by immunoblotting (Fig. 3C). As expected, BI2536, volasertib, Ro3280 and GSK461364, but not JQ1 and LRRK2-IN-1, significantly reduced the levels of phosphorylated TCTP at 30 – 100 nM drug concentration. LRRK2-IN-1, ERK5-IN-1 and XMD8-92 showed moderate to weak growth inhibitory activity with IC₅₀ values between 280 and 2,100 nM against MM1.S cells and 2,800 to >10,000 nM against HEK293T cells. For MM1.S cells, immunoblotting showed a dose-dependent decrease of c-Myc accompanied by an increase of p21^Cip1 (cyclin-dependent kinase inhibitor 1) levels for all BET-kinase inhibitors, BI2536 and volasertib having the strongest effect, followed by Ro3280, ERK5-IN-1, LRRK2-IN-1 and XMD8-92 (Fig. 3A, 3C). In HEK293T cells, levels of c-Myc remained unaltered while p21^Cip1 levels increased significantly for BI2536, followed by volasertib, LRRK2-IN-1 and Ro3280 (Fig. 3B). ERK5-IN-1 and XMD8-92 lacked the potential to increase p21^Cip1 levels, reflecting their weak activity on HEK293T cell growth.

Figure 3: Cellular activity of dual BET-kinase inhibitors.

(A) Growth inhibitory activity of compounds against MM1.S cells (left panel, 72 h drug exposure, cell-titer blue assay) and effect on c-Myc and p21 levels (right panel, 6 h drug exposure, immunoblotting). (B) Same as (A) in HEK293T cells. (C) Levels of phospho-TCTP (PLK1i biomarker) were assessed in parallel with BRD4i biomarkers to evaluate dual BRD4-PLK1 inhibitory activity in MM1.S cells (6 h drug exposure, immunoblotting).

Structural insights into BRD4 interaction with kinase inhibitors

All crystal structures of BET bromodomains liganded with kinase inhibitors were obtained by co-crystallization to allow unobstructed formation of the protein-inhibitor complex in solution prior to the buildup of a crystal lattice. Differences in the chemical structure of the inhibitor, size and length of solvent exposed moieties and binding affinity required establishment of new crystallization conditions for most BET-inhibitors complexes. As a result, most crystals were of different space group and/or unit cell dimensions (Supplementary Tables S3–S4). Crystal structures were determined for all inhibitors with BRD4-1, showing canonical interaction of the dihydropteridinone, pyrimidodiazepinone and benzopyrimidodiazepinone carbonyl oxygen with the side chain of N140 (Fig. 4A).

Figure 4: Cocrystal structures of kinase inhibitors with BRD4-1.

(A) Binding pattern of kinase inhibitors in the KAc site of BRD4-1 (PDB ID 4O74, 5V67, 5VBP, 5VBO, 7K6G, and 7K6H, respectively). (B) Superposition of inhibitors illustrating the conformational freedom and constraints upon interaction with the KAc site. Color code is BI2536 (yellow), volasertib (pink), Ro3280 (purple), LRRK2-IN-1 (orange), ERK5-IN-1 (green), XMD8-92 (blue). Black dotted lines indicate H-bonds (d = 2.2 – 3.5 Å). Displayed residues are in hydrophobic van-der-Waals distance (d = 3.3 – 4.0 Å). Structured water molecules are shown as cyan spheres.

All inhibitors directly interacted with two structured water molecules, bridging residues Y97 and C136 with the carbonyl oxygen, and residues P82 and Q85 with the pyrimidineamine moiety at the exit of the KAc site (Supplementary Fig. S5). Structure alignment revealed an almost identical positioning of the respective warheads to interact with N140 (Fig. 4B). However, each inhibitor assumed a different conformation in the binding site, which affects hydrophobic van-der-Waals (VDW) interactions between the benzamide moiety (phenyl in XMD8-92) and the WPF shelf on one side, and L92 on the opposite side. Notably, all inhibitors established T-shaped pi-pi stacking interactions between the benzamide ring and the side chain of W81 of the WPF shelf.

In kinase ATP sites, the benzodiazepinone moieties of LRRK2-IN-1 and ERK5-IN-1 exhibit an atropisomeric conformation different from that in the KAc site of bromodomains (Supplementary Fig. S6), a phenomenon previously reported for a series of analogs of ERK5-IN-1 in BRD4-1.¹⁷ The conformation of the dihydropteridinone and diazepinone warheads of volasertib and Ro3280 is almost identical. However, the long solubilizing tail of volasertib rotates ~180° between the two binding sites.

Structural basis of BET bromodomain inhibition by kinase inhibitors

Next, we attempted co-crystal structure determination of other BET bromodomains liganded with kinase inhibitors. Except for BRD4-2, high-resolution structures of fully occupied BET-inhibitor complexes were determined for BRDT-1 (5), BRD2-1 (4), BRD2-2 (3), BRD3-2 (3), BRDT-2 (2) and BRD3-1 (1) (Fig. 5). As expected from the structures of BRD4-1/inhibitor complexes, all inhibitors showed the same principal binding mode and canonical interaction of the warhead across BET bromodomains (Supplementary Fig. S7). Apart from Ro3280, moieties beyond the warhead adopted significantly different conformations in the respective KAc binding sites (Fig. 6). All BET bromodomains established pi-pi stacking interactions between the indole moiety of the WPF shelf and the inhibitor, except for BRDT-1 which lacked such interactions with BI2536, volasertib and LRRK2-IN-1 (Supplementary Fig. S8). While this explains the reduced potency of kinase inhibitors against BRDT, the structural basis of differential activity across the other BET bromodomains is less obvious. The observed conformational changes of each inhibitor occur in moieties adjacent to the warhead, not involved in H-bonding interactions with KAc site residues. The different inhibitor shapes are accompanied by altered VDW distances between inhibitor, the WPF shelf and a conserved leucine ((L⁹² in BRD4-1) on the opposite site (Fig. 6, Supplementary Fig. 7). Except for LRRK2-IN-1, the conformation of the WPF shelf is almost identical across the BET-inhibitor complexes. However, the leucine residue along with the preceding lysine (in BD1) or alanine in (BD2) assumes adopts different conformations, possibly influencing the positioning of inhibitor in the KAc site.

Figure 5: Kinase inhibitors adopt different binding poses across BET bromodomains.

Crystal structures of BET bromodomain/kinase inhibitor complexes were obtained by co-crystallization. All structures were aligned through the BRD4-1/BI2536 complex. Inhibitor is shown in yellow, BET bromodomains in pink (BRD2-1), purple (BRD3-1), grey (BRD4-1), green (BRDT-1), blue (BRD2-2), beige (BRD3-2), cyan (BRDT-2). The PDB identification codes are indicated for each structure. Detailed interactions of each BET-inhibitor complex are shown in Supplementary Figs. S7–S8, data collection and refinement statistics in Supplementary Table S4, and electron density maps in Supplementary Fig. S10.

Figure 6: Conformational states of kinase inhibitors appear to be influenced by VDW interactions with conserved hydrophobic residues.

The left panel shows a representative binding pose of inhibitor in a certain BET bromodomain along with VDW distances between inhibitor and conserved hydrophobic residues. The middle panel shows the superposition of the KL or AL flanks and the WPF shelf of the indicated BET bromodomains, the right panel the resulting inhibitor conformations. (A) BI2536, (B) volasertib, (C) Ro3280, (D) LRRK2-IN-1, (E) ERK5-IN-1. The color code is the same as in Fig. 6.

To evaluate whether the inhibitor binding patterns in crystal structures reflect the binding affinities in solution, we employed the Molecular Mechanics/Generalized Born Surface Area (MM-GBSA) method^{19, 20} of the Schrödinger suite to predict the binding energy of each BET-inhibitor complex from the structure coordinate sets. The cocrystal structures contain 1 – 4 fully occupied BET-inhibitor complexes in the asymmetric unit, depending on space group and unit cell dimensions. The total number of BET-inhibitor complexes with bound BI2536, volasertib, Ro3280, LRRK2-IN-1 or ERK5-IN-1 was 37. This culmination of structural data allowed for a statistical analysis of predicted vs. experimentally determined binding affinity data (Fig. 7). The predicted binding energy from MM-GBSA calculation of each protein-inhibitor interaction, ΔG_bind, showed a modest, yet statistically significant correlation with the binding potential determined by DSF (Fig. 7A) and ITC (Fig. 7B). All thermodynamic parameters derived from ITC experiments, ΔG, ΔH and -TΔS, showed similar trends with the predicted binding energies (Fig. 7C). While the modest statistical significance provides confidence in a direct relationship between the structural and biochemical data, it should be noted that the crystalline state of any protein-ligand complex is influenced by the crystal lattice and resulting protein-protein contacts. Although all structures were obtained by co-crystallization, allowing each BET-inhibitor complex to form a crystal lattice rather than adapting to an existing lattice by in-diffusion, we cannot rule out conformational constraints imposed by crystal forces. A larger set of data would be needed to substantiate a structure-binding affinity relationship.

Fig. 7: Binding energies calculated from cocrystal structures correlate with the binding affinity of BET-kinase inhibitors.

Each BET-inhibitor complex was subjected to energy calculations using Prime MM-GBSA of the Schrodinger suite, i.e., receptor grid generation, “score in place only” by Glide, followed by MM-GBSA calculation. The resulting binding energy of protein-inhibitor interaction, ΔG_bind = E_complex − E_receptor − E_ligand, was analyzed against the binding affinity data from DSF (A) and ITC (B) assays. (C) Correlation of thermodynamic parameters from ITC experiments and ΔG_bind values. Shown are the Pearson’s correlation coefficient (r), statistical significance value (P, two-tailed, α = 0.05) and sample size (N).

CHEMISTRY

The synthetic route to BET-kinase inhibitor SG3-179 is shown in Scheme 1. Boc protection of 4-chloro-3-nitroaniline followed by reduction of the nitro group of 1 gave the aniline 2. The A-ring aniline 4, bearing a tert-butylsulfonamide group was prepared by tert-butylsulfinylation of 2, oxidation of the intermediate tert-butylsulfinamide 3 and removal of the Boc group, using standard methods.²¹ Addition of aniline 4 to 2,4-dichloro-5-methylpyrimidine gave the 4-anilinopyrimidine 5. The B-ring aniline 6 was prepared TBTU coupling of 4-amino-2-fluorobenzoic acid and 1-methylpiperidin-4-amine. Finally, acidcatalyzed reaction of the aniline 6 with 4-anilinopyrimidine 5 provided the 2,4-dianilinopyrimidine SG3-179 (> 95% purity, Supplementary Figs. S11–14).

Scheme 1. Synthesis of BET inhibitor SG3-179^a.

^aReagents and conditions: (a) di-tert-butyl dicarbonate, THF, reflux, 78%; (b) FeCl₃·(H₂O)₆, NH₂NH₂·H₂O, MeOH, reflux, 86%; (c) tert-butylsulfinyl chloride, DCM, pyridine, 0 °C, 81%; (d) m-CPBA, DCM, 18 h; TFA; NaOH (aq.) 86%; (e) 2,4-dichloro-5-methylpyrimidine, 120 °C, 1.5 days, 52%; (f) 6, HCl_aq., 160 °C, 26%; (g) 4-amino-2-fluorobenzoic acid, DIPEA, TBTU, r.t.; 1-methylpiperidin-4-amine; 85%.

DISCUSSION

Although several kinase inhibitors have been reported to also inhibit BET bromodomains, structure-activity relationship studies have been limited to BRD4. A detailed understanding of intra-BET selectivity, however, may inform further development and practicality of such drugs in a clinical setting. Here, we focused on dihydropteridinone and pyrimidodiazepinone kinase inhibitors targeting PLK1, ERK5 or LRRK2. They were less potent than dual BET-JAK2 inhibitor SG3-179 and BET inhibitor JQ1 but showed differential binding potential across all BET bromodomains (Fig. 2). Both bromodomains of BRD4 and BRD3 were more sensitive for kinase inhibitors than BRDT and BRD2. While most kinase inhibitors preferentially interacted with the first bromodomains, ERK5-IN-1 appeared to be more selective for the second bromodomains. The differences in binding affinity and intra-BET selectivity may reflect the differential activity observed in cell growth inhibition and signaling (Fig. 3). Dual BRD4-PLK1 inhibitors were extremely potent in MM1.S and HEK293T cells, while dual BRD4-ERK5 and -LRRK2 inhibitors showed moderate inhibitory and signaling activities (Fig. 3). Potent and simultaneous inhibition of essential proteins such as PLK1 and BRD4 is likely to exert unwanted effects in normal cells, rendering BI2536, volasertib and Ro3280 less suitable as viable drugs. The attractiveness of dual BET-kinase inhibitors may instead lie in those compounds that potently inhibit a kinase less crucial for normal cells but driving diseased cells, along with moderate inhibition of BET proteins to increase drug efficacy. ERK5-IN-1 and LRRK2-IN-1 may be more promising candidates for further development, as they lack inhibition of PLK1 and were the weakest among the inhibitors tested in cell-based assays. However, a comprehensive analysis across a large panel of cell lines is certainly needed to test this hypothesis.

Previously reported cocrystal structures of dihydropteridinone and pyrimidodiazepinone kinase inhibitors with bromodomains were limited to BRD4-1 liganded with BI2536, LRRK2-IN-1, XMD8-92 and XMD17-26, a close analog of ERK5-IN-1 (PDB 4OGI, 5WA5, 5LRQ, 6CD5).^{10, 17, 22} During this work a total of 23 cocrystal structures across all BET bromodomains were determined along with dissociation constants by direct binding studies. Additionally, new cocrystal structures were determined for diaminopyrimidine JAK2 inhibitor SG3-179 with BRD3-1 and BRD4-2, and four new structures of pan-bromodomain inhibitor bromosporine with BRD2-1, BRD2-2, BRD3-2 and BRDT-2 (Supplementary Fig. S9), both of which served as positive controls in binding and crystallization studies. The data provide first structural insights into the interaction of kinase inhibitors with BET bromodomains beyond BRD4.

Collectively, the observed conformational states that each kinase inhibitor adopts upon interaction with BET bromodomains reflect differences in shape complementarity with the respective KAc site. This is particularly evident for ERK5-IN-1, LRRK2-IN-1 and volasertib but less pronounced for BI2536 and Ro3280 (Fig. 6, Supplementary Fig. S7). Except for the BRD4-1/LRRK2-IN-1 complex, the KAc site residues assumed similar conformations across all BET-inhibitor complexes. Thus, the differential inhibitory activity cannot be explained by a conspicuous loss or gain of inhibitor-residue interaction or steric hindrance. Lack of pi-pi stacking between the WPF shelf of BRDT-1 and inhibitors was the only noticeable difference in hydrophobic interactions (Supplementary Fig. S8). Sandwiched between the WPF shelf and a conserved leucine residue, optimal positioning of the inhibitor is likely influenced by the dynamics and flexibility of bromodomains.²³ It appears that the conglomerate of small changes in VDW distances between hydrophobic residues and inhibitor atoms beyond the warhead determines the overall shape that each inhibitor preferably adopts upon KAc site occupation (Fig.6). Depending on the length and number of rotatable bonds of the solubilizing tail, the resulting conformational state of each inhibitor likely differs in energy and entropy contributions towards complex formation, consequently impacting binding affinity. The modest but stastically significant correlation between the predicted binding energy from BET-inhibitor structures and the experimentally determined binding affinity data seems to support this notion (Fig. 7). While the integration of structural and binding affinity data offers novel insights into the action of kinase inhibitors across all BET bromodomains, our findings also underscore that a rational structure-based design of monovalent intra-BET selective kinase inhibitors is difficult. However, the observed changes in exit vectors suitable for the generation of bivalent inhibitors may offer alternative paths towards achieving intra-BET selectivity. For example, it is well established that PROTACs often provide a substantial increase in selectivity for highly similar bromodomains and kinases, despite carrying a nonselective warhead.^24–27 The new structural information provided herein may therefore inspire new directions towards the generation of intra-BET selective kinase inhibitors.

EXPERIMENTAL SECTION

Compounds and reagents:

Reagents for biochemical and crystallographic experiments were from Fisher Scientific and Hampton Research unless otherwise indicated. JQ1 (S7110), BI2536 (S1109), Volasertib (S2235), Ro3280 (S7248), LRRK2-IN-1 (S7584), ERK5-IN-1 (S7334) and XMD8-92 (S7525) were from Selleck Chemicals with purities of >95% for all compounds according to the manufacturer. Data correlation and nonlinear regression analyses were performed with Prism 9.1 (GraphPad). The concentration of purified proteins were determined by A280 molar absorbance using a Nanodrop ND-2000c spectrophotometer (Nanodrop Technologies).

Synthesis of SG3-179.

The details of analytical instrumentation and general chemical information are the same as recently reported.²⁸

1,1-Dimethylethyl 4-chloro-3-nitrophenylcarbamate (1):

This was prepared using the reported procedure of Sloss et al.²⁹ Sequentially 4-chloro-3-nitroaniline (5.00 g, 28.97 mmol) and di-tert-butyl dicarbonate (10.12 g, 46.36 mmol) were dissolved in dry THF (20 mL). The mixture was stirred and heated at reflux for 20 h. The solvent was removed and the yellow oil was triturated using EtOAc/hexanes to give the title compound²⁹ 1 as a yellow solid (6.18 g, 78%). Mp: 103 °C (dec). ¹H NMR (400 MHz, CDCl₃) δ: 8.06 (d, J = 2.3 Hz, 1H), 7.46 (dd, J = 8.7, 2.3 Hz, 1H), 7.42 (d, J = 8.7 Hz, 1H), 6.64 (s, 1H), 1.52 (s, 9H). HPLC-MS (ESI+): m/z 297.1 [30%, (M³⁷Cl+H)⁺], 295.1 [100%, (M³⁵Cl+H)⁺].

tert-Butyl 3-amino-4-chlorophenylcarbamate (2):

This was prepared using the reported procedure of Sloss et al.²⁹ The nitroarene 1 (6.18 g, 22.66 mmol), iron(III) chloride hexahydrate (0.183 g, 0.679 mmol), and activated carbon (1 g) were combined in MeOH (50 mL) and stirred at reflux for 10 min. Hydrazine hydrate (6.78 mL, 90.65 mmol) was added slowly and the mixture was stirred at reflux for 1 h. cooling to room temperature, the solution was filtered over a bed of Celite and concentrated under reduced pressure. The resulting residue was dissolved in EtOAc (80 mL) and washed with water (80 mL) and brine (80 mL). The organic layer was dried (Na₂SO₄) and concentrated under reduced pressure. The resulting solid was triturated using EtOAc/hexanes to give the title compound²⁹ 2 as a white solid (4.74 g, 86%). Mp: 139–140 °C. ¹H NMR (400 MHz, CDCl₃) δ: 7.15 (s, 1H), 7.12 (d, J = 8.6 Hz, 1H), 6.57 (dd, J = 8.6, 2.5 Hz, 1H), 6.42 (s, 1H, disappeared on D₂O shake), 1.50 (s, 9H). HPLC-MS (ESI+): m/z 509.2 [10%, (M³⁷Cl+M³⁵Cl+H)⁺], 507.2 [15%, (2M³⁵Cl+H)⁺], 265.1 [10%, (M³⁵Cl+H)⁺], 189.2 [100%, (M³⁷Cl-tBu+H)²⁺], 187.2 [100%, (M³⁵Cl-tBu+H)²⁺].

1,1-Dimethylethyl 4-chloro-3-(1,1-dimethylethylsulfinamido)phenylcarbamate (3):

To a solution of 2 (2.43 g, 10 mmol) and pyridine (2.42 mL, 30 mmol) in DCM (5 mL) was added a solution of t-butylsulfinyl chloride (1.23 mL, 10 mmol) in DCM (5 mL) dropwise at 0 °C under Argon. The mixture was stirred at 0 °C for 2 h, then warmed to room temperature and further stirred for 5 h. The reaction mixture was diluted with EtOAc (100 mL) and washed with hydrochloric acid (50 mL of 1 M aq. solution), water (50 mL), and brine (50 mL). The organic layer was dried (Na₂SO₄) and concentrated under reduced pressure. The resulting residue was purified via column chromatography (SiO₂) eluting with hexanes/EtOAc (0:10 to 3:7 v/v) to give the title compound 3 as a light-yellow foam (2.81 g, 81%). Mp: 124 °C (dec). ¹H NMR (400 MHz, CDCl₃) δ: 7.41 (d, J = 2.4 Hz, 1H), 7.21 (d, J = 8.8 Hz, 1H), 6.93 (brd, J = 8.8 Hz, 1H), 6.50 (brs, 1H), 6.06 (brs, 1H), 1.51 (s, 9H), 1.35 (s, 9H). HPLC-MS (ESI+): m/z 717.2 [80%, (M³⁷Cl+M³⁵Cl+H)⁺], 715.2 [70%, (2M³⁵Cl+H)⁺], 371.2 [50%, (M³⁷Cl+H)⁺], 369.2 [100%, (M³⁵Cl+H)⁺].

N-(5-Amino-2-chlorophenyl)-2-methylpropane-2-sulfonamide (4):

To a solution of 3 (2.80 g, 8.07 mmol) in DCM (30 mL) was added m-CPBA (65%, 2.143 g, 8.07 mmol) under Argon. The mixture was stirred overnight at room temperature. The reaction mixture was diluted with DCM (200 mL) and washed with saturated NaHCO₃ (2 × 200 mL) and brine (1 × 200 mL). The organic layer was dried (Na₂SO₄) and concentrated under reduced pressure. The resulting solid was triturated using EtOAc/hexanes to give the N-Boc-sulfonamide as a light yellow solid (2.59 g, 88%). A portion of this (2.45 g) was treated with a mixture of TFA/DCM (20 mL, 1:1 v:v) at room temperature for 2 h. The mixture was concentrated and treated with sodium hydroxide (1M aq.) until pH 12 and extracted with ethyl acetate (2 × 200 mL). The combined organic layers were washed with saturated water (100 mL) and brine (200 mL). The organic layer was dried (Na₂SO₄) and concentrated under reduced pressure to give the aniline 4 (1.74 g, 98%). Mp: 124–125 °C. ¹H NMR (400 MHz, CDCl₃) δ: 7.23 (d, J = 2.4 Hz, 1H), 7.11 (d, J = 8.6 Hz, 1H), 6.50 (s, 1H), 6.42 (dd, J = 8.6, 2.4 Hz, 1H), 1.41 (s, 9H). HPLC-MS (ESI+): m/z 527.2 [30%, (2M³⁷Cl+H)⁺], 525.2 [40%, (2M³⁵Cl+H)⁺], 265.2 [40%, (M³⁷Cl+H)⁺], 263.2 [100%, (M³⁵Cl+H)⁺].

2-Chloro-N⁴-{4-chloro-[3-(1,1-dimethylethylsulfonamido)]phenyl}-5-methylpyrimidin-4-amine (5):

A mixture of 2,4-dichloro-5-methylpyrimidine (0.082 g), aniline 4 (0.131 g), and DIPEA (0.210 mL) in isopropanol (5 mL) was stirred and heated in a sealed pressure vessel at 120 °C for 1.5 days. Water (5 mL) was added and the precipitate filtered, washed with water (2 × 5 mL), hexanes (2 × 5 mL), and dried to give the title compound 5 as a light brown solid (0.101 g, 52%). Mp: 274 °C (dec). ¹H NMR (400 MHz, DMSO-d₆): δ 9.33 (s, 1H, disappeared on D₂O shake), 8.99 (s, 1H, disappeared on D₂O shake), 8.07 (s, 1H), 7.92 (d, J = 2.4 Hz, 1H), 7.56 (brd, J = 8.7 Hz, 1H), 7.44 (d, J = 8.7 Hz, 1H), 2.16 (s, 3H), 1.32 (s, 9H). HPLC-MS (ESI+): m/z 391.1 [70%, (M³⁵Cl³⁷Cl+H)⁺], 389.1 [100%, (M³⁵Cl³⁵Cl+H)⁺].

4-Amino-2-fluoro-N-(1-methylpiperidin-4-yl)benzamide (6):

This was prepared the previously reported method.³⁰ Diisopropylethylamine (1.57 mL, 2.0 equiv.) was added to a mixture of 4-amino-2-fluorobenzoic acid (0.700 g, 4.51 mmol), TBTU (1.59 g, 4.95 mmol), in DCM (18 mL). The mixture was stirred at room temperature for 30 min followed by the addition of 1-methylpiperidin-4-amine (0.68 mL, 5.42 mmol). The mixture was stirred for 18 h at room temperature. The solvent was removed, and the resulting oil was stirred in NaOH (1 M aq. solution, 50 mL) at room temperature for 30 min. The aqueous layer was extracted with DCM (50 mL) and EtOAc (50 mL). The organic layers were combined, dried (Na₂SO₄), and concentrated under reduced pressure. The resulting solid was recrystallized from EtOH/hexanes to provide the title compound 6 (0.967 g, 85%) as an off-white solid. Mp: 202 °C (dec). ¹H NMR (400 MHz, DMSO-d₆) δ: 7.41–7.36 (m, 1H), 7.34 (t, J = 8.8 Hz, 1H), 6.35 (dd, J = 8.5, 2.0 Hz, 1H), 6.25 (dd, J = 14.2, 2.0 Hz, 1H), 5.87 (s, 2H, disappeared on D₂O shake), 3.71–3.57 (m, 1H), 2.67 (d, J = 11.6 Hz, 2H), 2.12 (s, 3H), 1.92 (t, J = 10.6 Hz, 2H), 1.70 (d, J = 10.6 Hz, 2H), 1.49 (qd, J = 11.6, 3.5 Hz, 2H). ¹⁹F NMR (376 MHz, DMSO-d₆): δ −112.88. HPLC-MS (ESI+): m/z 252.3 [100%, (M+H)⁺].

5-Methyl-N⁴-{4-chloro-[3-(1,1-dimethylethylsulfonamido)]phenyl}-N²-[3-fluoro-4-(1-methylpiperidin-4-ylcarbamoyl)phenyl]pyrimidine-2,4-diamine (SG3-179):

A mixture of the chloropyrimidine 5 (50 mg, 1.0 equiv.), and the aniline 6 (32 mg) (1.0 equiv.), hydrochoric acid (2 drops of a 4M aq. solution), and EtOH (1 mL) was heated in a microwave reactor at 160 °C for 15 minutes. The reaction mixture was diluted with EtOAc (20 mL) and washed with saturated NaHCO₃ (20 mL). The aqueous layer was then re-extracted with EtOAc (20 mL). The combined organic layers were dried over Na₂SO₄ and concentrated under reduced pressure. The resulting residue was further purified via preparative TLC using DCM/MeOH 15% and afforded the title product SG3-179 as an off-white solid (20 mg, 26%). Mp: 252 °C (dec). HPLC: 98% [t_R = 9.7 min, 40% MeOH, 60% water (with 0.1% TFA), 20 min]. ¹H NMR (400 MHz, DMSO-d₆): δ 9.44 (s, 1H, disappeared on D₂O shake), 8.63 (s, 1H, disappeared on D₂O shake), 7.97 (s, 1H), 7.82–7.71 (m, 4H), 7.41 (t, J = 8.6 Hz, 2H), 7.39 (d, J = 8.8 Hz, 2H), 7.32 (dd, J = 8.6, 2.0 Hz, 1H), 3.77–3.65 (brs, 1H), 2.85–2.73 (brs, 2H), 2.22 (s, 3H), 2.15–2.07 (brs, 2H), 2.11 (s, 3H), 1.82–1.72 (brd, J = 12.5 Hz, 2H), 1.61–1.47 (m, 2H), 1.30 (s, 9H). ¹⁹F NMR (376 MHz, DMSO-d₆): δ −112.33. HPLC-MS (ESI+): m/z 604.3 [25%, (M³⁵Cl+H)⁺], 303.3 [50%, (M³⁷Cl+2H)²⁺], 302.8 [100%, (M³⁵Cl+2H)²⁺]. LC-MS (ESI+): 606.2 [40%, (M³⁷Cl+H)⁺]. 604.2 [100%, (M³⁵Cl+H)⁺]. HRMS (ESI+): m/z calcd for C₂₈H₃₅ClFN₇O₃S (M+H)⁺ 604.2267, found 604.2251.

Protein expression and purification:

Expression plasmids for individual human BET bromodomains were from Addgene or the DNA sequences were custom synthesized by GeneArt (Thermo Fisher Scientific) and cloned in-frame of a modified pET15b or pET28a vector providing an N-terminal hexa-histidine tag followed by a Tobacco Etch Virus (TEV) cleavage site (Supplementary Table S5). Plasmids were transformed into E. coli BL21 (DE3) cells and grown at 37 °C in LB medium (Fisher Scientific) containing carbenicillin (0.1 mg/mL). At OD600 of 0.6, the culture was cooled down to 18 °C and induced with 0.1 mM IPTG. After 18 h growth, cells were harvested by centrifugation at 6,000 × g for 25 min and stored at −80 °C. Harvested cell pellets were re-suspended in 50 mM Na/K phosphate buffer (pH 7.4) containing 150 mM NaCl, 40 mM imidazole, 0.01% w/v lysozyme and 0.01% v/v Triton X-100 at 4 °C for 1 h, subjected to sonication and the lysate was clarified by centrifugation (30,000 × g for 45 min at 4 °C). Proteins were purified by FPLC at 4 °C using columns and chromatography materials from GE Healthcare. The lysate was subjected to an immobilized Ni2+ affinity chromatography column equilibrated with 50 mM Na/K phosphate buffer (pH 7.4) containing 150 mM NaCl and 40 mM imidazole using a gradient from 40 to 500 mM of imidazole. Fractions containing the target protein were combined and incubated for 2 – 16 h with TEV protease at 4 °C, and the cleaved His6-tag was removed by a second Ni2+ affinity column. BRDs were purified to homogeneity by size exclusion chromatography using Superdex 75 and were of crystallization grade quality (> 95% purity as judged by SDS-PAGE). BRD-containing fractions were combined, concentrated and aliquots were flash-frozen in liquid N₂ and stored at −80 °C.

Differential scanning fluorimetry (DSF):

DSF experiments were performed with a Applied Biosystem StepOnePlus (thermal shift determination) real-time PCR system (Thermo Fisher Scientific) using 96-well format plates, assayed in quadruplicate. To obtain robust fluorescence signals, the assay was optimized regarding concentration of protein (4.5 μM for thermal shift determination) and the fluorescence dye SYPRO Orange (Invitrogen, Thermo Fisher Scientific) (5X concentration). For thermal shift determination, dilutions of compound in assay buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 2 mM DTT, 20% DMSO) were prepared. Protein in assay buffer including fluorescence dye, 100 μM compound and 2% DMSO was assayed in 20 μL reaction volumes. Reaction mixtures were heated from 25 °C to 95 °C at 1 °C/min with fluorescence readings every 0.5 °C at 610 nm. The observed thermal shift (ΔT_m) was recorded as the difference between the T_m of sample and DMSO reference wells.

Isothermal titration calorimetry (ITC):

ITC experiments were performed using an ITC200 microcalorimeter from Malvern Panalytical (Spectris PLC). BRDs were buffer exchanged in ITC buffer (50 mM HEPES pH 7.4, 150 mM NaCl) using PD10 columns (GE life sciences) and concentrated to ~ 10 mg/ml. Experiments were carried out in ITC buffer while stirring at 750 rpm using reverse titration. The microsyringe (40 μL load volume) was loaded with 150–350 μM protein in ITC buffer and was inserted into the calorimetric cell (0.2 ml cell volume) filled with 20–25 μM compound in ITC buffer. All titrations were conducted using an initial control injection of 0.3 μl followed by 25 identical injections (1.52 μl per injection) with a duration of 3.04 sec (per injection) and a spacing of 150 sec between injections. The ratio of titrants in the titration experiments were optimized to ensure complete saturation of the titrant (protein or compounds) before the final injection thus facilitating the estimation of baseline for each injection. Experimental data were corrected by subtracting the heats of dilution determined from independent titrations. The collected data were analyzed using MicroCal^™ Origin software provided with the ITC instrument to determine the enthalpies of binding (ΔH) and binding constants (KB) as previously described.³¹ Thermodynamic parameters were calculated using the basic equation of thermodynamics (ΔG = ΔH - TΔS = -RTlnKB, where ΔG, ΔH and ΔS are the changes in free energy, enthalpy and entropy of binding, respectively). In all cases a single binding site model was used.

Microscale thermophoresis (MST).

MST experiments were performed using a Monolith NT.115 Pico instrument (NanoTemper Technologies). His-tagged BRD4-1 and BRDT-1 were labeled with RED-Tris-NTA (Nanotemper) according to the manufacturer’s instructions, except that ¼ of the NT-647-His labeling dye was used. Measurements were performed with a constant concentration of labeled protein (20 nM) and increasing concentrations of compound using a 16-point 2x serial dilution series (0.5 nM to 10 μM for high affinity and 2.5 nM to 50 μM for low affinity ligands). The experiments were performed in 50 mM Tris pH 7.4, 150 mM NaCl, 10 mM MgCl2, 0.8mg/ml BSA buffer supplemented with 0.05% Tween and 0.05% DMSO at 20% MST power and between 20–30% LED power at 24 °C with standard capillaries. The MST traces were recorded using standard parameters (5 s MST power off, 30 s MST power on and 5 s MST power off). Measurements were taken at −1 – 0 s (cold region) and 14–15 s (hot region). Data from 3 independent replicate experiments were merged and analyzed using the NTAnalysis software (NanoTemper Technologies). The macroscopic dissociation constants (K_d) were determined using equation [BL]/[B0] = {([L0]+[B0] + K_d) − √{([L0]+[B0] + K_d)² − 4[L0][B0]}}/2[B0], where [B0] corresponds to the total concentration of target binding sites, [L0] to the concentration of titrated ligand, and [BL] to the concentration of formed complex between ligand and target binding sites.³²

Crystallization and X-ray crystallography:

All crystallization experiments were performed at 18 °C. Aliquots of purified BRDs were set up for crystallization screening using a Mosquito crystallization robot (TP Labtech). Initially, coarse screens were set up using Greiner 3-well plates at three different ratios of precipitant to protein (200+400 nl, 300+300 nl, 400+200 nl) per condition. Conditions producing crystals were further optimized and scaled up for manual set up of 2 μl drops. For co-crystallization, compound was either pre-mixed with protein on ice or added to protein-reservoir solution drops to achieve final concentrations of 0.5–2.5 mM compound and 5–10 % DMSO. Crystals were cryoprotected using the well solution supplemented with ethylene glycol (15–30%) and flash frozen in liquid nitrogen. X-ray diffraction data were collected at −180 °C at beamlines 22-ID and 22-BM (SER-CAT), beamline 23-ID (GM/CA-CAT) of the Advanced Photon Source (Argonne National Laboratories), and in the Moffitt Chemical Biology Core using a Rigaku Micro-Max 007-HF X-ray generator equipped with a CCD Saturn 944 system. Data were reduced and scaled with XDS³³ or DIALS³⁴ and Aimless.³⁵ Molecular replacement and structure refinement was performed with PHENIX³⁶ and model building with Coot.³⁷ Initial models for the small molecule ligands were generated using MarvinSketch (ChemAxon, Cambridge, MA) with ligand restraints from eLBOW of the PHENIX suite. All structures were validated by MolProbity. Figures were prepared using PyMOL (Schrödinger). The coordinates and structure factors were deposited with the PDB. Crystallization conditions, crystallographic data collection and refinement statistics are shown in Supplementary Tables S3–S4.

Cell proliferation and signaling:

MM1.S and HEK293T cells (ATCC) were maintained in RPMI1640 and Dulbecco modified Eagle medium, repectively, with 10% fetal bovine serum. For cell growth inhibition studies, cells were seeded in 96-well plates at approximately 2,500 cells per well (0.1 mL). Adherent cells were allowed to settle and attach to the plate surface for 20 hours and cell media was removed for replenishing with fresh media containing desired drug concentration before dosing. A three-fold 9-point dilution series of compound was prepared starting at 10 μM concentration in culture media supplemented with 0.1% DMSO. Cells were incubated with compound for 72 hours in hexaplicates at which time 15 μL of CellTiter Blue reagent (Promega) was added to each well, followed by orbital shaking for 5 minutes and incubation for 3 hours at 37 °C. Plate fluorescence was recorded in a Wallac EnVision 2103 Multilabel Reader (PerkinElmer) using excitation and emission filters of 570 and 615 nm, respectively.

For western blotting, cells were treated with compound for 6 h, lysed in lysis buffer (50 mM Tris-HCl pH 8.0, 5 mM EDTA, 150 mM NaCl, 0.5% NP40, 1 × protease inhibitor cocktail), centrifuged for 10 min at 14,000 × g and the insoluble debris discarded. Cell lysate (10–50 μg of protein) was fractionated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred to Immobilion P membranes (Millipore). The membranes were blocked for 1 h with phosphate-buffered saline containing 5% nonfat dry milk and 0.1% Tween20, incubated with primary and secondary antibodies, and developed using the Supersignal reagent (Thermo Scientific). The following antibodies were purchased from commercial sources: Vinculin (Sigma, V9131), c-Myc (Cell Signaling, 5605), p21 (Cell Signaling, 2946), TCTP (Cell Signaling, 5128), and pTCTP (Cell Signaling, 5251).

PDB ID codes

BRD4-1

ERK5-IN-1 (7K6G), LRRK2-IN-1 (5VBO), Ro3280 (5VBP), volasertib (5V67), XMD8-92 (7K6H)

BRD4-2

SG3-179 (RKO0)

BRD2-1

BI2536 (7LAI), ERK5-IN-1 (7LAU), Ro3280 (7LAJ), volasertib (7LAK), bromosporine (7LAH)

BRD2-2

BI2536 (7L9G), LRRK2-IN-1 (7L9K), Ro3280 (7L9J), bromosporine (7L6D)

BRD3-1

ERK5-IN-1 (7LAZ), SG3-179 (7LAY)

BRD3-2

BI2536 (7L9L), ERK5-IN-1 (7LBT), Ro3280 (7L72), bromosporine (7LB4)

BRDT-1

BI2536 (5VBQ), ERK5-IN-1 (7L73), LRRK2-IN-1 (7LEM), Ro3280 (7BJY), volasertib (5VBR)

BRDT-2

ERK5-IN-1 (7LEK), volasertib (7LEJ), bromosporine (7LEL)

ABBREVIATIONS

BET

bromodomain and extraterminal

BRD

bromodomain

DIPEA

N,N-Diisopropylethylamine

DSF

differential scanning fluorimetry

ERK5

Extracellular signal-regulated kinase 5

ITC

isothermal titration calorimetry

JAK2

Janus kinase 2

LRRK2

Leucine-rich repeat serine/threonine-protein kinase 2

m-CPBA

meta-chloroperbenzoic acid

MST

microscale thermophoresis

PLK1

Polo-like kinase 1

TBTU

2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethylaminium tetrafluoroborate

TCTP

translationally controlled tumor protein

TFA

trifluoroacetic acid