Hit-to-lead optimization of a latency-associated nuclear antigen inhibitor against Kaposi's sarcoma-associated herpesvirus infections.

Research paper Hit-to-lead optimization of a latency-associated nuclear antigen inhibitor against Kaposi’s sarcoma-associated herpesvirus infections Philine Kirsch a , b , c , Saskia C. Stein c , d , Aylin Berwanger a , b , c , Julia Rinkes a , b , c , Valentin Jakob a , b , c , Thomas F. Schulz c , d , Martin Empting a , b , c , * a Department of Drug Design and Optimization (DDOP), Helmholtz-Institute for Pharmaceutical Research Saarland (HIPS) - Helmholtz Centre for Infection Research (HZI), Campus E8.1, 66123, Saarbrücken, Germany b Department of Pharmacy, Saarland University, Campus E8.1, 66123, Saarbrücken, Germany c German Centre for Infection Research (DZIF), Partner Site Hannover-Braunschweig, 66123, Saarbrücken, Germany d Institute of Virology, Hannover Medical School, Carl-Neuberg-Strasse 1, 30625, Hannover, Germany article info Article history: Received 11 February 2020 Received in revised form 25 May 2020 Accepted 30 May 2020 Available online 28 June 2020 Keywords: Hit-to-lead optimization Latency-associated nuclear antigen (LANA) Kaposi’s sarcoma herpesvirus (KSHV) Fluorescence polarization (FP)-Based interaction inhibition assay Electrophoretic mobility shift assay (EMSA) CuAAC STD-NMR abstract The Latency-associated nuclear antigen (LANA) plays a central role for the latent persistence of the Kaposi’s Sarcoma Herpesvirus (KSHV) in the human host and helps to establish lifelong infections. Herein, we report our efforts towards hit-to-lead generation starting from a previously discovered LANADNA inhibitor. By tethering the viral genome to the host nucleosomes, LANA ensures the segregation and persistence of the viral DNA during mitosis. LANA is also required for the replication of the latent viral episome during the S phase of the cell cycle. We aim to inhibit the interaction between LANA and the viral genome to prevent the latent persistence of KSHV in the host organism. Medicinal chemistry-driven optimization studies and structure-activity-relationship investigation led to the discovery of an improved LANA inhibitor. The functional activity of our compounds was evaluated using a fluorescence polarization (FP)-based interaction inhibition assay and electrophoretic mobility shift assay (EMSA). Even though a crystal structure of the ligand protein complex was not available, we successfully conducted hit optimization toward a low micromolar protein-nucleic acid-interaction inhibitor. Additionally, we applied STD-NMR studies to corroborate target binding and to gain insights into the binding orientation of our most potent inhibitor, providing opportunities for further rational design of more efficient LANAtargeting anti KSHV agents in future studies. ©2020 The Author(s). Published by Elsevier Masson SAS. This is an open access article under the CC BYNC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Kaposi’s Sarcoma Herpesvirus (KSHV) is a human gamma herpesvirus and establishes a lifelong latent infection in B-cells and endothelial cells [1,2]. The virus was identified as the etiological agent of Kaposi’s Sarcoma (KS) and is involved in two other neoplastic diseases, multicentric Castleman’s disease and pleural effusion lymphoma [1,3]. In healthy individuals, KSHV-associated diseases are rare. However, in immunosuppressed patients, e.g., transplant recipients or patients with the acquired immunodeficiency syndrome (AIDS), KSHV is highly oncogenic [4,5]. However, classic KS mainly can also occur in elderly men especially from KSHV-endemic areas and endemic KS in East and Central Africa [6]. The main key player for the establishment and maintenance of the latent infection is the latency-associated nuclear antigen (LANA) [7e9]. It is an origin-binding protein, whose C-terminal domain binds to the viral genome and whose N-terminal region interacts simultaneously with host nucleosomes [10e12]. This allows the segregation of latent viral episomes during mitosis and their partitioning to daughter cells [13]. LANA has also additional functions like latent viral replication, transcriptional control and survival in the host cell [14e16]. The C-terminal DNA-binding domain (DBD) of LANA binds the viral genome in a sequence-specific manner [17]. Located on the terminal repeats (TRs) are three specific LANA binding sites (LBS), LBS1, LBS2 and LBS3. LBS1 has a hundred fold higher affinity to LANA compared to LBS2 and LBS3 [17]. In the majority of KSHV-associated cancer cells the viral genome is present and LANA is expressed [18]. It has been shown, that the *Corresponding author. Department of Drug Design and Optimization (DDOP), Helmholtz-Institute for Pharmaceutical Research Saarland (HIPS) - Helmholtz Centre for Infection Research (HZI), Campus E8.1, 66123, Saarbrücken, Germany.. E-mail address: [email protected] (M. Empting). Contents lists available at ScienceDirect European Journal of Medicinal Chemistry journal homepage: http://www.elsevier.com/locate/ejmech https://doi.org/10.1016/j.ejmech.2020.112525 0223-5234/©2020 The Author(s). Published by Elsevier Masson SAS. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). European Journal of Medicinal Chemistry 202 (2020) 112525 persistence of viral DNA is affected by disturbing or influencing LANA [15]. The inhibition of the interaction between LANA and viral DNA could lead to a reduction or loss of viral genomes in the infected cells. Today’s treatment of KSHV and KSHV-associated diseases is difficult and still limited [19,20]. It is clear, that there is an urgent need for specific drugs, which interfere with novel steps in the KSHV lifecycle. In view of its central role during latent viral persistence, LANA is considered to be a very promising target for the development of specific antiviral therapeutics against KSHV. In a study previously published by us in 2019, we described the discovery of first inhibitors, which interfere with the LANA-DNA interaction [11]. Further inhibitor scaffolds have been identified using a functional screen and an in-house compound library [21]. Starting with a fragment-based drug discovery approach, we successfully developed a fragment-sized inhibitor Icapable to compete with the viral DNA (Fig. 1). For the evaluation of functional activity of our compounds, we used a fluorescence-polarization (FP)-based assay and electrophoretic mobility shift assay (EMSA) experiments. For our most promising fragment-sized inhibitor I, we observed an IC 50 value of 17 ±1 m M in our FP-assay using a LANA DNA binding domain (DBD) mutant and 435 ±6 m M in the EMSA studies using the wild-type LANA C-terminal domain (CTD). We confirmed target binding using microscale thermophoresis (MST) and saturation transfer difference (STD)-NMR experiments. Additionally, the STD-NMR experiments and molecular docking studies provided important information on the putative orientation of Inhibitor Iwhen bound to LANA. Based on the STD-NMR studies and docking results we suggested that the nitrogen at the pyridine core acts as a hydrogen bond acceptor and protons 2, 3 and 4 are not in direct interaction with the protein surface, hence these positions should be further investigated as potential growth vectors. Furthermore, two glutamines are presumably involved in hydrogenbond interactions with the carboxyl group. However, it was not clear whether the carboxylic acid function is necessary for binding [11]. Based on these findings, we embarked on structure-activity relationship (SAR) studies and further medicinal chemistry optimization to improve the potency of our hit compounds. Herein, we report our recent advances in improving our LANA-DNAinteraction inhibitors using compound Ias a starting point. Unfortunately, our efforts in solving a co-crystal structure of inhibitor I in complex with LANA have not been successful to date. This renders unambiguous experiment-supported structure-based optimization unfeasible. Therefore, we systematically investigated the LANA-DNA-interaction inhibition of new synthesized compounds using FP-based competition assay and EMSA experiments as the SAR drivers. 2. Design concept Based on the previously applied STD-NMR and docking studies we modified Inhibitor Iin a step-by-step manner. Inhibitor Iwas divided in two regions, the benzoic acid part Aand the pyridine core B(Fig. 1). The triazole core was not yet modified in order to exploit the robust and facile Copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) click chemistry. First, region Awas modified and variations of the carboxylic acid were introduced. As a second step, we have modified the pyridine moiety, region B. From our previous results we assumed that the nitrogen at the pyridine motif is essential for binding and functions as hydrogen bond acceptor. STD-NMR data revealed that Proton 1 interacts tightly with LANA while proton 2 is also in close proximity to the protein surface. However, our docking studies suggested that the latter might be at least partially solvent exposed. In contrast, protons at position 3 and 4 did not show direct contact with the LANA surface according to their weak STD-NMR effects. These observations inspired us to investigate positions 2, 3 and 4 as potential growth vectors in the presented study. Abbreviations AIDS acquired immune deficiency syndrome CTD C-terminal domain DCM dichloromethane DMSO dimethylsulfoxide DBD DNA binding domain DMF dimethylformamide DIPEA diisopropylethylamine EE ethyl acetate EtOH ethanol EMSA electrophoretic mobility shift assay FA formic acid FP fluorescence polarization HPLC high pressure liquid chromatography HHV-8 human herpesvirus 8 KS Kaposi Sarcoma KSHV Kaposi’s sarcoma-associated herpesvirus LANA latency-associated nuclear antigen LBS LANA binding site LCMS liquid chromatography mass spectrometer MeCN acetonitrile MeOH methanol MST microscale thermophoresis PBS phosphate-buffered saline PE petroleum benzene STD NMR saturation transfer difference nuclear magnetic resonance SPR surface plasmon resonance TR terminal repeat wt wild-type Fig. 1. Previously described LANA-DNA interaction inhibitor Iand its predicted binding mode which provides the basis for structural optimization by rational design and growth vector exploration. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 1125252 3. Results and discussion 3.1. Chemistry 3.1.1. Modifications of region A 3-azidopyridine 2was generated by a standard azidation method using 3-aminopyridine 1,NaNO 2 and NaN 3 in a mixture of EtOAc and 6M HCl [11,22]. In a second step, as depicted in Scheme 1, various commercially available ethynylbenzene derivatives were used in a standard copper-catalyzed CuAAC reaction with 3azidopyridine to provide the triazoles 3e5, 8, 10, 11, and 13 [11]. The ethyl ester 6and amide 7analogue were generated from the carboxylic acid 3by thionyl chloride-mediated activation and subsequent treatment with ethanol or aq. ammonia solution. The hydrolysis of 3-chloro-4-methylester intermediate 8with NaOH in ethanol produced the corresponding acid 9. The N-acetyl analogue 12 was synthesized from amine 11 with acetyl chloride under basic conditions. 3.1.2. Modifications of region B Compound 16 bearing an additional CH 2 -linker between triazole and pyridine core was synthesized starting from (bromomethyl)benzene 14, which was converted to the azide 15 using NaN 3 in DMSO [23], followed by a click reaction with 4ethynylbenzoic acid. Different arylazides 17a-n and 20 decorated with various substitutions were generated by reaction of the corresponding commercially available amines 18a-n and 21 with NaNO 2 and NaN 3 in 6 M HCl and EtOAc (Scheme 2). The subsequent CuAAC click reaction with 4-ethynylbenzoic acid provided the target molecules 19a-n and 22. The hydroxypyridine 19o analogue was generated from the methoxypyridine 19k by treating with 48% aqueous HBr solution at 80  C. As depicted in Scheme 3, the syntheses of target compounds via Suzuki coupling was achieved using two different synthetic routes. In route 1, Suzuki coupling with different commercially available boronic acids and halogenated pyridine-3-amines 23, 27 and 28 in presence of Pd(P(Ph 3 ) 4 achieved phenyl-substituted pyridine amines 24a-b and 29a-b in the first step. Subsequently, the amines were converted to the corresponding azides 25a-b and 30a-b followed by a CuAAc click reaction with 4ethynylbenzoic acid to obtain the target compounds 26a-b and 31a-b. In parallel, the alternative route 2 was established for late stage modifications via Suzuki coupling. First, halogenated pyridine-3-amines 32a-b were converted to the corresponding azide 33a-b, followed by click reaction with 4-ethynylbenzoate to obtain the corresponding triazole intermediates 34a-b.Subsequently, phenyl-substituted compounds 35a-k were achieved via Suzuki coupling using corresponding boronic acids and Pd(P(Ph 3 ) 4 . Finally, the hydrolysis of the esters with NaOH in methanol produced the target carboxylic acid compounds 36a-k. As depicted in Scheme 4, for the synthesis of the series of pyridine-phenoxy target compounds 40a-e, cupper-catalyzed Ullmann reaction was used in the first step using 6-bromo-4-methylpyridin-3-amine 37, the corresponding phenol derivative or thiophenol, Cs 2 CO 3 and CuI to obtain the aminopyridine-phenoxy intermediates 38a-e. The obtained amines were transformed into the corresponding azides 39a-e as described above. Last step was a CuAAC reaction of azides with 4-ethynylbenzoic acid to obtain the target compounds 40a-e. The isoquinoline 43 and quinoline 46 analogue were synthesized starting from isoquinoline-4-amin 41 and quinoline-3amin 44 by standard azidation to 42 and 45, followed by CuAAC click reaction with 4-ethynylbenzoic acid. Further isoquinoline derivatives 50a-c were synthesized in a 3 step procedure (Scheme 5). A direct transformation of bromo isoquinolines 47a-c into the corresponding azides using NaN 3 , Cu(I) and Na 2 CO 3 at 85  C over night as described in literature was not efficient [24]. LCMS-guided reaction monitoring showed the formation the primary amine and other side products. For this reason, we extended the reaction time until we detected full conversion into the corresponding primary amine 48a-c with the aim to subsequently transform these intermediates into the corresponding azides. Indeed, we achieved successful azidation (intermediates 49a-c) and CuAAC coupling, respectively, using amines 48a-c and the conditions described above yielding the desired isoquinoline products 50a-c. 3.2. Functional evaluation using LANA-DNA interaction inhibition assays and SAR studies The target compounds were tested for functional activity in the FP-based LANA-DNA interaction inhibition assay using LBS2 as the probe and an oligomerization-deficient LANA DBD mutant [11]. This oligomerization-deficient C-terminal LANA mutant (aa10081146) has nine amino acid point mutations: K1055E, K1138S, K1140D, K1141D, R1039Q, R1040Q, A1121E, K1109A, and D1110A. For this mutant, also in presence of oligonucleotides, which represent the viral LANA-binding sites LBS1, LBS2 or LBS3, a high water solubility was shown [10,11,17,25]. All compounds showing an IC 50 values less than 250 m M were further tested in an orthogonal LANA-DNA interaction inhibition assay employing EMSA methodology, the same LANA DBD mutant and LBS1 as probe. As described above, the latter oligo has a higher affinity to the target rendering the EMSA experiment a more stringent read out for compound efficacy. For the first series of compounds, we investigated the significance of the carboxylic acid in the Western part of the molecule (region A) by varying its position, attaching additional groups or substituting it by other polar functional groups capable of Scheme 1. Modification of region A. a . a Reagents and conditions: a) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, yield 50%, 2 h; b) corresponding ethynylbenzene, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 60e82%; c) 1. SOCl 2 , DMF, 60 C, 1 h, 2. EtOH, DIPEA, rt, 16 h, yield 53%; d) 1. SOCl 2 , DMF, 60 C, 1 h, 2. NH 4 OH, rt, 16 h, yield 27%; e) 2 M NaOH, MeOH, rt, 16 h, yield 66%; f) Actyl chloride, Et 3 N, DCM, DMF, rt, 16 h, yield 14%. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 112525 3 participating in hydrogen bonding. The results are shown in Table 1. Moving the carboxylic acid from para (inhibitor I)tometa position (4) decreases the activity significantly. Also an additional chlorine atom attached in meta position (9) lead to a complete loss of activity. The replacement of the carboxylic acid by a methyl ester (5), ethyl ester (6) or amide (7) was also detrimental. Furthermore, moving from the carboxylic acid to the methyl alcohol (10), amine (11), acetamide (12) or nitrile (13) also resulted in inactive compounds. These results indicate that the carboxylic acid in para position in region A is essential for inhibitory activity. Therefore, we kept the p-carboxylic acid in region A fixed for further optimization studies and focused on the modifications at the pyridine core in region B. First, we examined the effect on inserting a short linker between the triazole and the pyridine core (16). This, however, resulted in loss of activity. As described before, from previous STD-NMR and molecular docking experiments we expected, that growing the fragment-sized Inhibitor Iin different positions at the pyridine core (region B) would potentially increase potency. To explore the influence of larger structural motifs at the pyridine core in position 4 we introduced a variety of residues. As listed in Table 2, growing in this position is accepted and resulted in moderate to potent inhibitory effects in FP assay ranging from IC 50 values of 86 ±6 m M(19a)to18±4 m M(19c). The size of the introduced residue seems to play an important role. While a small methyl group is not favorable, but accepted (19a,IC 50 86 ±6 m M), further increasing the size from chlorine (19b) to phenyl (19c) improves IC 50 values to 29 ±1 m M m M and 18 ±4 m M, respectively. This observation might hint at a steric ortho effect. The additional bulky phenyl ring strongly hinders the rotation of the bond between triazole and pyridine and, therefore, might fix the nitrogen in the pyridine core in a more favorable orientation. In EMSA experiments, 4-substituted compounds 19b (EMSA: 94% inhibition @ 500 m M) and 19c (EMSA: 100% inhibition @ 500 m M) showed a higher efficiency compared to Inhibitor I(EMSA: 83% inhibition @ 500 m M) [11]. Additionally, we shifted the nitrogen of the pyridine core from meta (19a) to para position (22) which resulted in an inactive compound. The improvements in the EMSA assay for compounds 19b-c over our initial hit compound Iwere not perfectly mirrored by the FP IC 50 values, which presumably is rooted in the usage of different DNA probes (LBS1 vs LBS2, respectively). Nevertheless, the results for compounds 19b and 19c were a major step towards achieving LANA inhibitors suitable for cellular assays and encouraged us to explore the potential of growing the hit scaffold in this direction even further. In the next series of compounds, Inhibitor Iwas grown in position 3 at the pyridine core by introducing a variety of aromatic rings. As listed in Table 3, a small methyl residue in position 3 (19d, IC 50 of 45 ±5 m M; EMSA: 78% inhibition @ 500 m M) is tolerated, but the fluorinated analogue 19j and most of the phenyl substituted compounds 36a-d,36f-i, and 31a showed a complete loss or only moderate activity. However, compounds with an additional hydroxyl function attached to the phenyl ring (36e, IC 50 of 153 ±7 m M, EMSA: 20% inhibition @ 500 m M) showed moderate activity. Moving from a phenyl 36a to a smaller and more polar furanyl residue 36j the potency was restored (IC 50 of 19 ±2 m M). Unfortunately, in EMSA experiments we observed only a weak effect (34% inhibition @ 500 m M) for this compound. Interestingly, by attaching an additional chlorine atom in position 4 and having a phenyl in position 3 (31b) resulted in a highly potent compound with IC 50 of Scheme 2. Azide synthesis and CuAAC click reaction. a . a Reagents and conditions: a) NaN 3 ,Et 3 N, DMSO, rt, 16 h, yield 78%; b) 4-ethynylbenzoic acid, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 20e90%; c) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, 2 h, yield 4e98%; d) 48% aq. HBr, 80 C, 12 h, yield 86%. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 1125254 38 ±3 m M and full inhibition in FP and EMSA assays, respectively. These results further corroborate the notion of a beneficial ortho effect. To explore the influence of growing inhibitor Iat the pyridine core in position 2, a set of different target compounds was synthesized (Tables 3 and 4). The direct attachment of a nitrile group to the pyridine was tolerated (19e:IC 50 52 ±37 m M, EMSA: n. i.). Moving to chlorine, hydroxy or methoxy group we observed a significant loss in activity (19f:IC 50 >250 m M; 19o: IC 50 214 ±24 m M and 19k: IC 50 218 ±192 m M). Also in EMSA experiments 19j (75% inhibition @ 500 m M) and 19k (11% inhibition @ 500 m M) did not show a significant effect. An increase in activity was observed by introducing bulkier substituents and an additional methyl group for R 1 . In detail, an unpolar bulky phenyl or pScheme 3. Synthesis of target compounds via Suzuki coupling using two different routes. a . a Reagents and conditions: a) corresponding boronic acid, Na 2 CO 3 , Pd(P(Ph 3 ) 4 , 1,4-dioxan, H 2 O, 90 C, 16 h, yield 21e75%; b) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, 2 h, yield 86e99%; c) 4-ethynyl benzoic acid, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 36e76%; d) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, 2 h, yield 34e80%; e) 4-ethynyl benzoate, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 63e80%; f) corresponding boronic acid, Na 2 CO 3 , Pd(P(Ph 3 ) 4 , 1,4-dioxan, H 2 O, 90 C, 16 h, yield 20e90%; g) 2 M NaOH, MeOH, rt, 16 h, yield 16e83%. Scheme 4. Synthesis of target compounds via Ullmann Reaction. a . a Reagents and conditions: a) corresponding phenol or thiophenol, Cs 2 CO 3 , CuI, DMF, 130 C, 16 h, yield 28e94%; b) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, 2 h, yield 70e99%; c) 4ethynyl benzoic acid, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 45e93%. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 112525 5 chlorophenyl was accepted in position 2 and we observed IC 50 values of 36 ±5 m M for 36k and 58 ±7 m M for 36l and moderate inhibition in EMSA. Analogues 26a and 26b with polar hydroxyl groups attached at the phenyl showed good potency with IC 50 Scheme 5. Synthesis of isoquinoline derivatives. a . a Reagents and conditions: a) NaNO 2 , NaN 3 , EtOAc, 6M HCl, 0 C/rt, 2 h, yield 68e91%; b) 4-ethynyl benzoic acid, CuSO 4 5H 2 0, Na-Ascorbate, DIPEA, MeOH, H 2 O, rt, 16 h, yield 25e90%; c) NaN 3 ,Na 2 CO 3, CuSO 4 5H 2 0, Na-Ascorbate, L-proline, DMF, H 2 O, 85 C, 24 h, yield 88e97%. Table 1 Inhibition activities of compounds with modification in region A. Cpd R FP Assay (LBS2) a Cpd R FP Assay (LBS2) IC 50 IC 50 Inhibitor I 17 ±1 m M9n.i. 4>250 m M10 n.i. 5n.i. b 11 >250 m M 6n.i. 12 n.i. 7n.i. 13 >250 m M a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation. b No inhibition at 500 m M. Table 2 Inhibitory activities of analogues modified in position 4 eobserving higher efficiency for 4-substituted compounds. Cpd R FP Assay (LBS2) a EMSA (LBS1) b IC 50 inhibition @ 500 m M Inhibitor I 17 ±1 m M 82% 16 >250 m M n.d. d 19a 86 ±6 m M 39% 19b 29 ±1 m M 94% 19c 18 ±4 m M 100% 22 n.i. c n.d. a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation. b Electrophoretic mobility shift assay using LBS1 as probe. c No inhibition at 500 m M. d Not determined. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 1125256 values of 21 ±3 m M and 25 ±1 m M, respectively. Furthermore, the efficiency of these two analogues in our EMSA studies was high with a full inhibition @ 500 m M. By attaching methylamine (19g), isopropylamine (19h) and anilino (19i) at position 2 we observed an increase in activity from small to bigger size, whereby the methylamine compound 19g was completely inactive and the anilino analogue 19i showed a moderate activity of IC 50 of 110 m M and 29% inhibition in EMSA. Additionally, a series of compounds was synthesized with a more flexible and bulky phenoxy group in position 2 (Table 5). The phenoxy analogue 19l, similar to the aminophenyl compound 19i, was inactive, indicating that an amino linker between pyridine and phenyl is more suitable for activity compared to the oxygen linker. By attaching an additional methylgroup in position 4 (R 1 ) at the pyridine core 40a an increase in activity compared to 19l was observed leading to a moderate IC 50 of 198 ±8 m M. The fluorinated analogues 40b-d also showed moderate activities like compound 40a while the o-fluoro analogue 40b possessed the best IC 50 of 64 ±2 m M. For the m-(40c) and p-fluoro (40d) derivatives IC 50 sof 122 ±3 m M and 134 ±2 m M were observed, respectively. Unfortunately, all these compounds showed no effect in EMSA experiments. Exchanging the oxygen linker by a sulfur (40e,IC 50 175 ±10 m M) was tolerated (compare with 40a,IC 50 198 ±8 m M). As expected, by removing the nitrogen in the pyridine core resulted in an inactive compound (19m). Astonishingly, moving the nitrogen to the phenoxy residue (19n) yielded a highly potent compound with an IC 50 of 19 ±1 m M showing also full inhibition in the EMSA experiments. Intrigued by the notion that fragment growing in position 3 was possible in combination with ortho-substituents, we focused our efforts on further exploring these two positions by installing a connected structural motif. To this end, we designed and synthesized isochinoline analogues (Table 6). In general, isoquinoline analogues were pleasingly effective. The unsubstituted isochinoline Table 3 Inhibitory activities of analogues modified in position 3. Most derivatives substituted in this position (R 2 ) showed a significant decrease in activity. cpd R 1 R 2 FP Assay (LBS2) EMSA (LBS1) IC 50 Inhibition @ 500 m M Inhibitor I HH 17±1 m M 82% 19d HMe 45±5 m M 78% 19j H F n. i. n. d. 36a H>250 m Mn.d. 36f Hn.i. n. d. 36g H>250 m Mn.d. 36b Hn. i. n. d. 36c H110 ±32 m Mn.i. 36d Hn. i. n. d. 36i Hn. i. n. .d 36e H153 ±7 m M 20% 31a H>250 m Mn.d. 36h H>250 m Mn.d. 36j H19 ±2 m M 34% 31b Cl 38 ±3 m M 100% a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation. b Electrophoretic mobility shift assay using LBS1 as probe. c No inhibition at 500 m M. d Not determined. Table 4 Inhibitory activities of analogues modified in position 2. Attaching polar hydroxyl benzene groups increases inhibitory activity. cpd R 1 R 2 FP Assay (LBS2) EMSA (LBS1) IC 50 Inhibition @ 500 m M Inhibitor I HH 17±1 m M 82% 19e HCN 52±37 m M n.i. 19f Me Cl >250 m M n.d. 19o H OH 214 ±24 m M 75% 19k H OMe 218 ±192 m M 11% 36k Me 36 ±5 m M 36% 36l Me 58 ±7 m M 30% 26a Me 21 ±3 m M 100% 26b Me 25 ±1 m M 100% 19g Hn.i. n.d. 19h H>250 m M n.d. 19i H110 ±20 m M 29% a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation. b Electrophoretic mobility shift assay using LBS1 as probe. c No inhibition at 500 m M. d Not determined. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 112525 7 43 showed an FP IC 50 value of 33 ±1 m M and 96% inhibition in EMSA experiments. Moving from isoquinoline to quinoline 46 resulted in a slight loss in activity compared to 43 (46, IC 50 70 ±34 m M, 61% inhibition in EMSA). Attaching an additional methyl (50a) or chlorine (50b) at the isoquinoline motif was beneficial for the inhibitory effect. Noteworthy, 50a showed the lowest FP IC 50 of 8 ±1 m M reported to date, while 50b also possessed a decent IC 50 of 17 ±1 m M. Furthermore, in EMSA experiments 100% inhibition was detected for both compounds at 500 m M. Finally, an isoquinoline methylester analogue 50c was inactive, however. 3.3. Further characterization and EMSA studies using wild-type LANA For further evaluation and characterization the most promising compounds were selected. On the bases of our results, we chose compounds 19c,31b,26a-b,19n,50a, and 50b, which possessed the best IC 50 values in the FP-based assay and showed strong inhibitory effects at 500 m M in EMSA using the oligomerizationdefizient LANA DBD mutant (aa1008-1146). First, these compounds were initially tested for inhibition at 250 m M using LANA DBD mutant (Fig. 2, A) to see if they are also able to disturb the LANA-DNA interaction at a lower concentration in EMSA and compared these results also with inhibitor I. Inhibitor I and the compounds 19c,31b, and 26b showed no inhibitory effect on the LANA mut -DNA interaction at 250 m M. However, strong inhibitory effects were observed at this concentration for compounds 26a,19n,50a, and 50b as observed by the disappearance of the bands for the LANA mut -DNA complex (upper band, Fig. 2A). We also determined the inhibitory activity of our best compounds against the interaction between wild-type LANA CTD (aa934-1162) and viral LBS1 (Fig. 2, B) in EMSA. Our wild-type LANA CTD construct is longer compared to the LANA DBD mutant and has no mutations and still shows a sufficient solubility in aqueous medium also in presence of viral LBS1. The compounds were also tested at 250 m M. Unfortunately, no inhibitory effect was observed for compounds I,19c,31b,26a-b,19n, and 50b. However, Compound 50a showed a significant effect and was able to inhibit the interaction between wild-type LANA CTD and LBS1. Furthermore, we titrated the compounds showing an effective inhibition in EMSA using the LANA DBD mutant (Fig. 2, A), in EMSA experiments using the LANA DBD mutant and LBS1 as a probe to determine the IC 50 values. The results are listed in Table 7 and detailed information can be found in the supporting information. As reported earlier by us, inhibitor Ishowed an IC 50 in FP assay (LBS2) of 17 ±1 m MandanIC 50 in EMSA of 426 ±2 m M using LANA DBD mutant [11]. The observed IC 50 values using LBS2 for the most promising inhibitors were basically in the same range. Additionally, we also tested the most promising inhibitors Table 5 Inhibitory activities of Phenoxy analogues. Shifting nitrogen to the phenoxy residue improves inhibitory efficiency. cpd R 2 FP Assay (LBS2) EMSA (LBS1) IC 50 inhibition @ 500 m M Inhibitor I 17 ±1 m M 82% 19l n. i. n. d. 40a 198 ±8 m Mn.i. 40b 64 ±2 m Mn.i. 40c 122 ±3 m Mn.i. 40d 134 ±2 m Mn.i. 40e 175 ±10 m Mn.i. 19m n.i. n. d. 19n 19 ±1 m M 100% a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation. b Electrophoretic mobility shift assay using LBS1 as probe. c No inhibition at 500 m M. d Not determined. Table 6 Inhibitory activities of Isoquinoline derivatives. Adding an annulated ring structure in direction of identified growth vector results in the most efficient inhibitors to date. Cpd R FP Assay (LBS2) EMSA (LBS1) IC 50 inhibition @ 500 m M Inhibitor I 17 ±1 m M 82% 43 33 ±1 m M 96% 46 70 ±34 m M 61% 50a 8±1 m M 100% 50b 17 ±1 m M 100% 50c >250 m Mn.d. a Fluorescence-polarization assay using LBS2 as probe, data representing average of duplicates ±standard deviation; b Electrophoretic mobility shift assay using LBS1 as probe; c No inhibition at 500 m M; d Not determined. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 1125258 in FP assay using LBS1 and LBS3, respectively. Compound 50a showed a 2-fold better IC 50 of 8 9 m MagainstallLBScompared to I. Furthermore, we could increase the inhibitory activity in EMSA experiments using LBS1 by 7-fold. As a consequence compounds 19n and 50a are the most potent LANA-LBS1inhibitors reported so far (IC 50 values of 64 ±12 m Mand 53 ±3 m M). Interestingly, compounds showing increased IC 50 values of 50e60 m M against LBS1 and LBS3 in FP assay were also not effective in EMSA at 250 m M. Excepted is however inhibitor I, which showed also IC 50 values around 20 m M against LBS1 and LBS3, but no effect at 250 m MinEMSA. As described above, only compound 50a showed an inhibitory effect at a concentration of 250 m M in EMSA using wild-type LANA CTD (Fig. 3 A). A dose-response EMSA experiment with wild-type LANA CTD yielded an IC 50 value of 60 ±4 m M(Fig. 3 B). These results indicate that compound 50a is equally potent against wild-type LANA CTD and the oligomerization-deficient LANA DBD mutant. In comparison to inhibitor I(IC 50 of Fig. 2. EMSA gels with inhibitor I,19c,31b,26a-b,19n,50a-b. Compounds were tested at a final concentration of 250 m M and LBS1 was used as probe. (A) Using an oligomerizationdeficient LANA DBD mutant, a strong inhibitory effect (disappearance of LANA-DNA complex band) was observed for compounds 26a,19n and 50a-b (B) Using wild-type LANA CTD, a significant inhibitory effect for compound 50a was observed. Table 7 Comparison of most efficient LANA-DNA inhibitors. Cpd Structure FP-Assay a IC 50 (LBS2) (LANA mut ) FP-Assay IC 50 (LBS1) (LANA mut ) FP-Assay IC 50 (LBS3) (LANA mut )EMSA b IC 50 (LBS1) (LANA mut ) Inhibitor I 17 ±1 m M20±3 m M19±3 m M 426 ±2 m M 19c 18 ±4 m M52±2 m M42±3 m M n.i. at 250 m M c 31b 38 ±3 m M55±7 m M45±4 m M n.i. at 250 m M 26a 21 ±3 m M30±2 m M34±3 m M 156 ±27 m M 26b 25 ±1 m M64±1 m M63±8 m M n.i. at 250 m M 19n 19 ±1 m M15±1 m M25±1 m M64±12 m M 50a 8±1 m M9±2 m M8±1 m M53±43 m M 50b 17 ±1 m M14±1 m M15±1 m M93±8 m M a Fluorescence-polarization assay using LBS1, LBS2 and LBS3 as probe, data representing average of duplicates ±standard deviation. b Electrophoretic mobility shift assay using LBS1 as probe. c No inhibition at 250 m M. P. Kirsch et al. / European Journal of Medicinal Chemistry 202 (2020) 112525 9