LP99: Discovery and Synthesis of the First Selective BRD7/9 Bromodomain Inhibitor

The bromodomain-containing proteins BRD9 and BRD7 are part of the human SWI/SNF chromatin-remodeling complexes BAF and PBAF. To date, no selective inhibitor for BRD7/9 has been reported despite its potential value as a biological tool or as a lead for future therapeutics. The quinolone-fused lactam LP99 is now reported as the first potent and selective inhibitor of the BRD7 and BRD9 bromodomains. Development of LP99 from a fragment hit was expedited through balancing structure-based inhibitor design and biophysical characterization against tractable chemical synthesis: Complexity-building nitro-Mannich/lactamization cascade processes allowed for early structure–activity relationship studies whereas an enantioselective organocatalytic nitro-Mannich reaction enabled the synthesis of the lead scaffold in enantioenriched form and on scale. This epigenetic probe was shown to inhibit the association of BRD7 and BRD9 to acetylated histones in vitro and in cells. Moreover, LP99 was used to demonstrate that BRD7/9 plays a role in regulating pro-inflammatory cytokine secretion.

Abstract: The bromodomain-containing proteins BRD9 and BRD7 are part of the human SWI/SNF chromatin-remodeling complexes BAFand PBAF.T od ate,n os elective inhibitor for BRD7/9 has been reported despite its potential value as ab iological tool or as al ead for future therapeutics.T he quinolone-fused lactam LP99 is now reported as the first potent and selective inhibitor of the BRD7 and BRD9 bromodomains.D evelopment of LP99 from af ragment hit was expedited through balancing structure-based inhibitor design and biophysical characterization against tractable chemical synthesis:C omplexity-building nitro-Mannich/lactamization cascade processes allowed for early structureactivity relationship studies whereas an enantioselective organocatalytic nitro-Mannichreaction enabled the synthesis of the lead scaffold in enantioenriched form and on scale.T his epigenetic probe was shown to inhibit the association of BRD7 and BRD9 to acetylated histones in vitro and in cells.M oreover, LP99 was used to demonstrate that BRD7/9 plays arole in regulating pro-inflammatory cytokine secretion.
Bromodomains (BRDs) are protein interaction modules that selectively recognize e-N-lysine acetylation motifs,akey event in reading the posttranslational modifications that constitute the epigenetic code.B RD-containing protein 7 (BRD7), which is frequently down-regulated in cancer, [1] has aproposed tumor suppression function through regulation of p53 [2] and PI3K. [3] Furthermore,BRD7 has been shown to be required for BRCA1-dependent transcription, [4] and BRD7 polymorphism has been linked to an increased risk of pancreatic cancer. [5] In contrast, BRD9 is often overexpressed in cancer owing to ag ain of the short arm of chromosome 5( 5p), the most frequent karyotypic change in cervical cancer. [6] Theclosely related BRDs BRD7 and BRD9 are part of the SWI/SNF nucleosome-remodeling complex, which plays ac rucial role regulating gene expression programs,i ncluding the expression of inflammatory genes. Although the functions of the catalytic subunits of this complex, BRM and BRG1, in immune phenomena and inflammatory responses have been described, ar ole for BRD7 and BRD9 in inflammatory processes has not yet been demonstrated. [7] Owing to the complexity of BRD7/9-mediated interactions in chromatin, selective,p otent inhibitors of these bromodomains would constitute valuable biological tools,enabling functional studies on these essential chromatin interaction domains and potentially allowing for exploitation in small-molecule therapies for various diseases.T od ate,n o potent and selective inhibitors have been reported. [8] Our aim was to design and develop aprobe for the BRD7 and BRD9 bromodomains,a chieving potencyand selectivity with as uitably decorated fragment [9] rich in sp 3 -hybridized carbon atoms that, guided by biophysical assays,was designed to maximize specific binding interactions whilst retaining synthetic tractability.T hrough the use of reaction cascades, which bring together simple starting materials to quickly generate structurally complex products in an efficient one-pot process, [9,10] we hoped to quickly generate structure-activity relationships (SARs). Subsequently,e nantioselective catalysis would allow the scaled synthesis of known enantiomers of late-stage intermediates for advanced SARs,a nd, ultimately, the scaled-up synthesis of ap robe compound.
Thedevelopment of the first potent and selective BRD7/9 BRD inhibitor began with the simple fragment 1-methylqui-nolone (1), which was shown to be an orthosteric ligand of the BRD of BRD9-related ATAD2 ( Figure 1A). [11] Compound 1 was also shown to bind BRD9 by isothermal titration calorimetry (ITC;F igure 1B;s ee also the Supporting Information, Figure S1). The N-methyl amide moiety is an acetyl lysine mimetic,f orming similar hydrogen bonds to ac onserved asparagine (N1064) and aw ater molecule as seen in acetyl-lysine recognition. Although BRDs share similar acetyl-lysine recognition motifs,t here are significant differences in distal parts of the binding pocket. Owing to ashift in the ZA loop in BRD9, this region is much larger in BRD9, with residues A46, F47, P48, T50, and I53 forming al arge hydrophobic cavity ( Figure 1B). TheC7position was viewed as an ideal attachment point for as tructurally complex heterocyclic appendage that could exploit this cavity for selective inhibition of BRD9. Furthermore,a nalysis of the BRD9/1 model suggested that aC 4m ethyl group on the quinolone would occupy as hallow hydrophobic pocket, thereby increasing potency.
To identify the best core scaffold for asymmetric elaboration, as eries of quinolones bearing various N-heterocycles were synthesized (Scheme 1). [12] Compound 4,r eadily accessible by known methods, [13] was used in palladium-catalyzed Buchwald-Hartwig couplings to give various cyclic amides, carbamates,and ureas (5-10). All compounds were tested for BRD9 BRD binding using adifferential scanning fluorimetry (DSF) assay (Table 1). Changes in the melting temperatures (DT m )confirmed that nearly all of the elaborated quinolones were tolerated, and valerolactam 6 was selected as al ead structure for further characterization and optimization. The valerolactam series was selected in preference to the urea series mainly because of the wealth of direct and reliable methods available for their asymmetric synthesis. Co-crystallization with BRD9 showed the conserved Hbonding of acetyl-lysine recognition:C ompound 6 exhibited Hbonds to N100 and to aconserved water molecule.T he valerolactam moiety extends into the desired hydrophobic region between F44 and I53 whilst positioning the amide carbonyl near, but not in Hbond contact with, Y106. Furthermore,t he introduced methyl group at the C4 position demonstrated additional hydrophobic interactions with A54 and Y106. ITC confirmed the activity determined by the DSF assay: 6 bound to BRD9 with ad issociation constant (K D )o f612 nm ( Figure 2B).

Angewandte
Communications C4/C5/C6-substituted compounds (21-24), which were tested against BRD9 for potencya nd BRD4(1) for selectivity by DSF (see Table S1). All of the compounds showed selectivity for BRD9 over BRD4(1);h owever,l actam 24 (R 1 = H, R 2 = Ph) was considered to be the most promising compound for further development owing to its potencyand the opportunity for further optimization. To investigate any stereochemical preferences, 24 was resolved into its component enantiomers by preparative HPLC on ac hiral stationary phase.T he importance of the absolute stereochemical configuration of the inhibitors was confirmed by the fact that (À)-24 showed am odest increase in potencyc ompared to 6 (ITC: K D = 493 nm vs.610 nm)whereas (+ +)-24 was an order of magnitude weaker (ITC: K D = 4.3 mm ;s ee the Supporting Information). Co-crystallization of the active enantiomer,which was shown to have a2 R ,3S absolute configuration, with BRD9 (Figure S9) revealed binding consistent with that of 6,b ut with additional Hbond interactions observed between the NH motif of the Boc-protected amino group to the backbone carbonyl group of G43, and between the lactam carbonyl group to Y106. Thec o-crystal structure suggested that potencyc ould be further boosted by additional substitution of the newly installed aryl ring and alternative derivatization of the amine to optimize Hbonding to G43 and hydrophobic interactions with F47. Halogenated, methylated, and methoxylated benzaldehydes were used in the nitro-Mannich/lactamization cascade process (25-32)a nd carried through to the N-Boc-protected lactams 33-40 as before.A s the C6-aryl-substituted lactams gave exceedingly low yields in the Buchwald-Hartwig coupling (2 %), aCu-mediated Goldberg coupling was used instead to furnish the coupled products 41-48 in acceptable yields. [15] All substituents, except a para-methoxy group,w ere well tolerated, with 48 (R 1 = H, R 2 = p-Cl-C 6 H 4 )b eing the strongest binder according to DSF analysis (DT m = 4.4 AE 0.72 8 8C; Table S1).
Va riations to the substituent on the amino group were more productive in improving binding to BRD9. The carbamate protecting group was removed with HCl/dioxane, and the resulting amine was derivatized through reactions with ar ange of acyl chlorides,c hloroformates,i socyanates, and sulfonyl chlorides to give various amides,c arbamates, ureas,and sulfonamides for testing (49-65;see Table S1). No single chemical class dominated binding;t he four best compounds, 48, 55 (R 3 = Bz), 60 (R 3 = SO 2 iBu), and 64 (R 3 = CONHPh), all possessed different functional groups.
Towards the asymmetric synthesis of the most active compounds,various organocatalysts were trialed in the nitro-Mannich reaction of 11 with imine 66 to obtain the desired enantiomer of 67 in aselective fashion (Scheme 3). Of these, recently developed bifunctional cinchona-alkaloid-derived phase-transfer catalysts were found to impart the highest enantiofacial selectivity. [16] Following optimization, quinidinederived catalyst 68 furnished the desired product 67 on gram scale as a7:1 mixture of diastereomers,both in 90 % ee.N-Boc deprotection and concomitant cyclization with TFA, followed by epimerization with DBU,a fforded lactam 69 as as ingle diastereomer.T he synthesis was then completed as before to give compounds 48, 55, 60,a nd 64 in 90 % ee,w ith further

Angewandte
Chemie purification by preparative HPLC on achiral stationary phase providing these compounds in > 98 % ee.I TC analysis of these compounds revealed (2R,3S)-60,h ereafter referred to as LP99,t ob et he most potent compound synthesized, with a K D value of 99 nm against BRD9 (Table 2). This binding was entirely driven by enthalpic interactions (DH = À11 kcal mol À1 ), with an et loss in entropy upon binding (TDS = À2.0 kcal mol À1 ), which is consistent with anumber of specific interactions and may offer an advantage for wider selectivity. [17] Thei mportance of chirality and configuration in this work is highlighted by the fact that the enantiomer of LP99 showed no detectable binding to BRD9 by ITC ( Figure S8).
Inhibitor LP99 was further assessed in an umber of biological assays.T his compound was profiled broadly for BRD selectivity by DSF (Figure 3) against all expressible BRDs (48 of 61 in the human genome), showing exquisite selectivity with < 1 8 8Cs tabilization of all BRDs,i ncluding other members of sub-family IV,e xcept BRD7/9 ( Figure 3).
Inhibition of BRD7/9-histone interactions in cell assays was also examined. Thec ellular efficacyo fLP99 on BRD9 was investigated using af luorescence recovery after photobleaching (FRAP) assay ( Figure S10): [18] LP99 was found to disrupt BRD9 interactions with chromatin at aconcentration of 0.8 mm.T om easure this further, ab ioluminescence reso-nance energy transfer (BRET) assay was performed. BRD7and BRD9-NanoLuc luciferase fusion proteins and fluorescently labelled histone H3.3-and H4-HaloTag fusions were expressed in HEK293 cells. [19] Thea ddition of LP99 decreased BRET for both BRD7 and BRD9 in both the H3.3 and H4 systems in ad ose-dependent manner,w ith cellular IC 50 values in the low micromolar range for both histones ( Figure 4A;s ee also Figure S11 and Table S3). Taken together,t hese cellular assays demonstrate that the BRD7/9 inhibitor LP99 is able to disrupt the binding of BRD7 and BRD9 to chromatin in cells.F urthermore, cytotoxicity tests in U2OS cells for 24 and 72 hours showed the inhibitor to be non-toxic at concentrations of < 33 mm ( Figure S12).
To investigate if BRD7/9 could influence the expression of pro-inflammatory cytokines,ahuman THP-1 monocyticc ell line was stimulated with lipopolysaccharide (LPS), and the influence of LP99 on the secretion of interleukin 6(IL-6) was measured by an enzyme-linked immunosorbent assay (ELISA;F igure 4B). LP99 inhibited IL-6 secretion from THP-1 cells in adose-dependent manner,demonstrating that BRD7/9 BRDs are potential targets for anti-inflammatory treatment. Thee ffect of LP99 on IL-6 expression demonstrates for the first time that asmall-molecule BRD7/9 inhibitor may have as imilar function and utility to IL-6 neutralizing antibodies,s uch as tocilizumab,i nt he treatment of rheumatoid arthritis. [20] In conclusion, by using astructure-based design approach, the simple BRD binding fragment 1 has been developed into LP99,apotent and selective inhibitor of the closely related BRDs of BRD7 and BRD9. Incorporating tractable chemical synthesis, through anitro-Mannich/lactamizationcascade and   . . the use of ab ifunctional cinchona-alkaloid-derived phasetransfer catalyst, allowed for rapid establishment of structure-activity relationships and access to the lead enantioenriched scaffold on scale. Theu se of ligand-protein co-crystallography was crucial to determine the preferred absolute configuration of the ligands and in the design and synthesis of analogues with increased potency. Thes electivity of the most potent analogue, LP99,w as extensively characterized, and the compound was shown to inhibit only two of the 48 BRDs. Furthermore,t his compound was shown to disrupt the association of tagged BRD7 and BRD9 constructs from both bulk chromatin (by aF RAP assay) and individual histone proteins (by aNanoBRET assay). We have shown for the first time that aside from the BRM and BRG catalytic subunits of the SWI/SNF complex, BRD7 and BRD9 also play ar ole in the regulation of inflammatory cytokines and are potential novel targets for anti-inflammatory treatment. As the first potent, selective,a nd cell-active inhibitor of BRD7/9, LP99 will serve as av aluable tool in further deciphering the biological roles of these important BRDcontaining proteins and serve as as tarting point in the discovery of an ew class of epigenetic therapeutics.