3D Pharmacophore-Based Discovery of Novel KV10.1 Inhibitors with Antiproliferative Activity

cancers

Article 3D Pharmacophore-Based Discovery of Novel KV10.1 Inhibitors with Antiproliferative Activity

Žan Toplak 1, Louise Antonia Hendrickx 2, Špela Gubiˇc 1, Štefan Možina 1 , Bojana Žegura 3,4 , Alja Štern 3,4 , Matjaž Novak 3,4, Xiaoyi Shi 5, Steve Peigneur 2 , Jan Tytgat 2, Tihomir Tomašiˇc 1 , Luis A. Pardo 5,* and Lucija Peterlin Mašiˇc 1,*

1 Faculty of Pharmacy, University of Ljubljana, Aškerˇceva7, 1000 Ljubljana, Slovenia; [email protected] (Ž.T.); [email protected] (Š.G.); [email protected] (Š.M.); [email protected] (T.T.) 2 Toxicology and Pharmacology, Campus Gasthuisberg, University of Leuven, Onderwijs en Navorsing 2, Herestraat 49, P.O. Box 922, 3000 Leuven, Belgium; [email protected] (L.A.H.); [email protected] (S.P.); [email protected] (J.T.) 3 Department of Genetic Toxicology and Cancer Biology, National Institute of Biology, Veˇcnapot 111, 1000 Ljubljana, Slovenia; [email protected] (B.Ž.); [email protected] (A.Š.); [email protected] (M.N.) 4 University of Ljubljana, Kongresni trg 12, 1000 Ljubljana, Slovenia 5 AG Oncophysiology, Max-Planck Institute for Experimental Medicine, Hermann-Rein-Str. 3, 37075 Göttingen, Germany; [email protected] * Correspondence: [email protected] (L.A.P.); [email protected] (L.P.M.)

Simple Summary: A novel structural class of inhibitors of the voltage-gated potassium channel KV10.1 was discovered by a ligand-based drug design method using a 3D pharmacophore model. The virtual screening hit compound ZVS-08 inhibited the channel in a voltage-dependent manner Citation: Toplak, Ž.; Hendrickx, L.A.; consistent with the action of a gating modiﬁer. Structure–activity relationship studies revealed a Gubiˇc,Š.; Možina, Š.; Žegura, B.; Štern, A.; Novak, M.; Shi, X.; nanomolar KV10.1 inhibitor that is selective for some KV and NaV channels but exhibits signiﬁcant Peigneur, S.; Tytgat, J.; et al. 3D inhibition of the hERG channel. KV10.1 inhibitor 1 inhibited the growth of the MCF-7 cell line Pharmacophore-Based Discovery of expressing high levels of KV10.1 and low levels of hERG more potently than the Panc1 cell line (no

Novel KV10.1 Inhibitors with KV10.1 and high hERG expression). Moreover, the KV10.1 inhibitor 1 induced signiﬁcant apoptosis Antiproliferative Activity. Cancers in tumour spheroids of Colo-357 cells. This study may provide a basis for the use of computational 2021 13 , , 1244. https://doi.org/ drug design methods for the discovery of novel KV10.1 inhibitors as new promising anticancer drugs. 10.3390/cancers13061244

Abstract: (1) Background: The voltage-gated potassium channel KV10.1 (Eag1) is considered a Academic Editor: Jason Roszik near- universal tumour marker and represents a promising new target for the discovery of novel anticancer drugs. (2) Methods: We utilized the ligand-based drug discovery methodology using 3D Received: 29 January 2021 pharmacophore modelling and medicinal chemistry approaches to prepare a novel structural class of Accepted: 10 March 2021 K 10.1 inhibitors. Whole-cell patch clamp experiments were used to investigate potency, selectivity, Published: 12 March 2021 V kinetics and mode of inhibition. Anticancer activity was determined using 2D and 3D cell-based models. (3) Results: The virtual screening hit compound ZVS-08 discovered by 3D pharmacophore Publisher’s Note: MDPI stays neutral µ with regard to jurisdictional claims in modelling exhibited an IC50 value of 3.70 M against KV10.1 and inhibited the channel in a voltage- published maps and institutional affil- dependent manner consistent with the action of a gating modifier. Structural optimization resulted iations. in the most potent KV10.1 inhibitor of the series with an IC50 value of 740 nM, which was potent on the MCF-7 cell line expressing high KV10.1 levels and low hERG levels, induced significant apoptosis in tumour spheroids of Colo-357 cells and was not mutagenic. (4) Conclusions: Computational

ligand-based drug design methods can be successful in the discovery of new potent KV10.1 inhibitors. Copyright: © 2021 by the authors. The main problem in the ﬁeld of KV10.1 inhibitors remains selectivity against the hERG channel, Licensee MDPI, Basel, Switzerland. which needs to be addressed in the future also with target-based drug design methods. This article is an open access article distributed under the terms and Keywords: KV10.1; ion channels; hERG; pharmacophore modelling; virtual screening; antiprolifera- conditions of the Creative Commons tive activity Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Cancers 2021, 13, 1244. https://doi.org/10.3390/cancers13061244 https://www.mdpi.com/journal/cancers Cancers 2021, 13, 1244 2 of 24

1. Introduction

The voltage-gated potassium channel KV10.1 (Eag1) is a member of the ether-à-go-go + family, which also includes the erg (eag-related gene, KV11) and elk (eag-like K channel, KV12) subfamilies [1]. Like other KV channels, KV10.1 is a tetramer with a transmembrane region consisting of a voltage-sensor domain (VSD) with four alpha-helical segments (S1– S4) and segments S5 and S6 forming the ion-conducting pore (Figure1). Positively charged residues in the VSD respond to membrane depolarisation by a movement that translates into S5 and S6 helices opening the pore to K+ flux that drives the membrane voltage toward the negative K+ equilibrium potential [2]. In healthy tissues, KV10.1 is almost undetectable outside the central nervous system, although it is highly expressed in over 70% of human cancers, regardless of tumour type [3]. On this basis, KV10.1 is considered a nearly universal tumour marker and represents a promising new target for new anticancer drug discovery [4]. The channel is involved in cell cycle control and cell proliferation, migration, angiogenesis and resistance to hypoxia of cancerous cells [5]. The mechanisms of cancer progression are complex with changes in membrane potential control and non-canonical effects of overexpressed KV10.1 that impact the Ca2+ signalling and microtubule dynamics [6]. In vitro and in vivo animal experiments showed that inhibiting the KV10.1 channel leads to the reduction of tumour progression [7–11]. Therefore, inhibition of the KV10.1 could increase the survival of patients with the fibrosarcoma [12], ovary carcinoma [13], glioblastoma [14], acute lymphoid leukaemia [15], gastric [16], head and neck [17] and colon cancers [18], all entities where a correlation between KV10.1 and poorer outcome has been documented. Importantly, treatment with KV10.1 inhibitors of brain metastasis patients improved survival in a retro- spective study [14]. However, to date, no KV10.1-specific small-molecule inhibitors have been reported, as most of them also inhibit voltage-gated sodium channels or, most importantly, the cardiac hERG channel, thus posing the risk of cardiac side effects [4]. Recent studies have pointed to a particular structural difference between KV10.1 and hERG channels, suggesting the possibility of selective targeting of KV10.1 in cancer therapies [19]. This now opens exciting possibilities for further exploration using the published cryo-electron microscopy structures of KV10.1 and hERG [2,20]. To date, only a handful of different small molecules have been reported to inhibit KV10.1 (Figure1). The identification of K V10.1 inhibitors is mostly based on the already known inhibitory activity at the structurally similar hERG channel and/or on the anti- tumour properties of the small molecules. The anti-tumour activity of natural products has led to the discovery of their blocking of Kv10.1, but to our knowledge, no other ligand- or structure-based drug design approach has been used to discover new KV10.1 inhibitors. Cancers 2021, 13, 1244 3 of 24 Cancers 2021, 13, x FOR PEER REVIEW 3 of 25

Figure 1. Opposite two subunits of KV10.1 with voltage-sensor domain (VSD) in orange, pore domain and central cavity Figure 1. Opposite two subunits of KV10.1 with voltage-sensor domain (VSD) in orange, pore domain and central cavity in in yellow, selectivity filter in red box. Structures of known ligands binding to the VSD and the central cavity of the pore yellow, selectivity ﬁlter in red box. Structures of known ligands binding to the VSD and the central cavity of the pore domain. domain. The structure (PDB: 5k7l) in the figure represents Kv10.1 with the pore domain in the closed conformation due The structure (PDB: 5k7l) in the ﬁgure represents Kv10.1 with the pore domain in the closed conformation due to calmodulin to calmodulin binding and VSD in the upward/conductive conformation. TBA: tetrabutylammonium: TEA: tetrae- binding and VSD in the upward/conductive conformation. TBA: tetrabutylammonium: TEA: tetraethylammonium. The thylammonium. The structure of KV10.1 was prepared using PyMOL [21]. structure of KV10.1 was prepared using PyMOL [21].

Due to the high similarity between KV10.1 and hERG, particularly in the putative Due to the high similarity between K 10.1 and hERG, particularly in the putative drug binding sites in the central cavities of theV channels, achieving selective inhibition of drug binding sites in the central cavities of the channels, achieving selective inhibition of KV10.1 with pore-blocking small molecules presents a difficult hurdle. A VSD modulation K 10.1 with pore-blocking small molecules presents a difﬁcult hurdle. A VSD modulation approachV could potentially circumvent this problem [20]. To date, only four types of small approach could potentially circumvent this problem [20]. To date, only four types of molecules are known to bind to the VSD, namely, mibefradil, natural bromotyrosine pur- small molecules are known to bind to the VSD, namely, mibefradil, natural bromotyrosine purealidin I analogues, amiodarone and 20(S)-ginsenoside Rg3 (Figure 1). The exact bind- purpurealidin I analogues, amiodarone and 20(S)-ginsenoside Rg3 (Figure1). The exact ingbinding sites sitesof these of these inhibitors inhibitors are still are unknown, still unknown, but recently but recently published published cryo cryo-electron-electron microscopy structures of both KV10.1 and hERG offer the possibility of using the VSD as a microscopy structures of both KV10.1 and hERG offer the possibility of using the VSD as a bindingbinding region region for for new new mo modulatorsdulators [2,20,22,23] [2,20,22,23].. In In this this work, work, we we describe describe the the ligand ligand-based-based drug discovery of a new structural class of KV10.1 inhibitors using 3D pharmacophore drug discovery of a new structural class of KV10.1 inhibitors using 3D pharmacophore modelling,modelling, structural structural optimisation optimisation of of the the virtual virtual screening screening hit hit towards towards potent potent nanomolar nanomolar KV10.1 inhibitors, selectivity profiling against hERG and some selected ion channels, mu- KV10.1 inhibitors, selectivity proﬁling against hERG and some selected ion channels, tagenicitymutagenicity study study,, and and antiproliferative antiproliferative activity activity in 2D in 2Dand and 3D 3Dcell cellbased based models. models.

2.2. Materials Materials and and Methods Methods 2.1.2.1. Virtual Virtual Compound Compound Library Library Preparation Preparation ThreeThree compound compound libraries libraries were were prepar prepared:ed: (1) (1) a a selected selected set set of of purpurealidin purpurealidin ana- ana- logueslogues as as K KVV10.110.1 active active compounds, compounds, (2) (2) a a calculated calculated set set of of decoy decoy molecules, molecules, and and ( (3)3) a a mergedmerged set of commercially availableavailable compoundscompounds available available for for hit hit identiﬁcation identification via via virtual vir- tualscreening screening with with prioritised prioritised ligand-based ligand-based pharmacophore pharmacophore model. model. For For the the KKVV10.1 actives library,library, structures structures of of 11 11 purpurealidin purpurealidin I Ianalogues analogues were were retrieved retrieved from from a scientific a scientiﬁc publi- pub- cationlication of ofMoreelsMoreels et al et. al[22].[22. A]. second A second library library of 551 of decoy 551 decoy molecules molecules was generated was generated based onbased each on of each the 11 of theKV10.1 11 K inhibitorsV10.1 inhibitors by submitting by submitting their structures their structures to the DUDE to the DUDE decoy onlinedecoy generator online generator [24], which [24], resulted which resulted in 50 decoy in 50 molecules decoy molecules per compound per compound with similar with 1Dsimilar physicochemical 1D physicochemical properties properties but dissimilar but dissimilar 2D topology 2D topology in comparison in comparison to the 11 to active the 11 compounds.active compounds. AA third library library of of 556,000 556,000 compounds compounds from from commercial commercial providers providers was prepared was prepared based basedon the on diversity the diversity sets from sets from Asinex, Asinex, ChemBridge, ChemBridge, Enamine, Enamine, KeyOrganics, KeyOrganics, LifeChemicals, LifeChem- icals, Maybridge, and Vitas-M. Libraries were downloaded in SDF format, merged, and

Cancers 2021, 13, 1244 4 of 24

Maybridge, and Vitas-M. Libraries were downloaded in SDF format, merged, and duplicates removed using the LigandScout Database Merger and Duplicates Remover nodes as implemented in the Inte:Ligand Expert KNIME Extensions [25]. For each of the three libraries, a maximum of 200 conformations were generated for each molecule using LigandScout’s iCon algorithm [26,27] with the default “BEST” settings (Max. number of conformers per molecules: 200, Timeout (s): 600, RMS threshold: 0.8, energy window: 20.0, max. pool size: 4000, max. fragment build time: 30). Each library was saved in LDB (LigandScout database format) using LigandScout’s idbgen algorithm with default settings.

2.2. Ligand-Based Pharmacophore Modelling

Eleven purpurealidin-based KV10.1 inhibitors were used for creation of ten ligand- based pharmacophore models in LigandScout 4.3 Expert. The models were generated using the following ligand-based pharmacophore creation settings: Scoring function: pharmacophore ﬁt and atom overlap; Pharmacophore type: Merged feature pharmacophore; Number of omitted features for merged pharmacophore: 4; partially matching features optional, threshold (%): 10.0; Feature tolerance scale factor: 1.0; Maximum number of result pharmacophores: 10. Creation of an exclusion volumes coat around the alignment of the ligands was also enabled for each of the models. All ligands of the training set were automatically aligned to the generated pharmacophore models. The resulting ligand-based pharmacophore models were tested for their performance in the distinguishing the active and decoy molecules. The best performing model was selected for virtual screening to identify new KV10.1 inhibitors.

2.3. Virtual Screening Best ﬁnal ligand-based pharmacophore model was used for virtual screening of the library of 556,000 commercially available compounds. The following settings were used for virtual screening: Scoring function: Pharmacophore-Fit; Screening mode: Match all query features; Retrieval mode: Get best matching conformation; Max. number of omitted features: 0, Check exclusion volumes: Checked. The virtual screening identiﬁed 18 virtual screening hits that were then visually inspected and nine of them where selected and purchased for in vitro screening against Kv10.1 using whole-cell patch clamp.

2.4. Chemistry The synthesis of the virtual screening hit ZVS-08 and its analogues is shown in Figures2–6 . Synthesis of differently substituted 4-(phenylamino)phenols 15, 18, 21 and 26 was carried out using four different approaches. o-Nitro-p-triﬂuoromethyl-substituted phenol 15 was prepared in two steps (Figure2). First, the nitro group was introduced by nitration of 13 with concentrated nitric and sulphuric acid to give 14, followed by the addition of p-aminophenol at high temperature to yield 15 [28]. The nitro-substituted compound 18 was prepared by coupling p-aminophenol and 17 with the addition of base at reﬂux (Figure3). For the synthesis of compounds without the electron-withdrawing nitro group, palladium-catalysed amination with Pd(dba3)2 was used for compound 21 (Figure4 ), while CuO and Cu were used as catalysts for the synthesis of 26 (Figure6)[ 29]. The carboxylic acid 26 was converted to the methyl ester 27 using thionyl chloride in anhydrous MeOH (Figure6). Ethers 1 (Figure2) and 4 (Figure3) were formed from 15 and 18 using 3-dimethylamino-1-propyl chloride hydrochloride and Cs2CO3 as base. The preparation of epoxides 16, 19, 22, 24 and 28 was carried out using (±)-epichlorohydrin and K2CO3 as base [30]. The synthesis of 12 by Ritter reaction (Figure2) was carried out using acetonitrile in the presence of BF3OEt2 [31]. For epoxide opening, morpholine was used to give 2 and 8; aniline was used to yield compound 5 and dimethylamine for compounds ZVS-08, 3, 6, 7 and 9. Reduction of nitro group of ZVS-08 by catalytic hydrogenation gave compound 10 (Figure2). The methyl ester of 9 was hydrolysed with 1 M sodium hydroxide to give compound 11 (Figure6). Synthetic procedures, analytical data, chemicals, reagents, CancerCancerss 20212021, 13, 13, x, 1244FOR PEER REVIEW 5 5of of 25 24

Cancers 2021, 13, x FOR PEER REVIEW 5 of 25

Cancers 2021, 13, x FOR PEER REVIEW 5 of 25 ide to give compound 11 (Figure 6). Synthetic procedures, analytical data, chemicals, rea- and equipment used for organic synthesis and analysis are written in Supplementary gents, and equipment used for organic synthesis and analysis are written in Supplemen- Materialside to give under compound the Chemistry 11 (Figure section. 6). Synthetic procedures, analytical data, chemicals, rea- tary Materials under the Chemistry section. gentside to, giveand equipmentcompound used11 (Figure for organic 6). Synthetic synthesis procedures, and analysis analytical are written data, inchemicals, Supplemen- rea- tarygents Materials, and equipment under the used Chemistry for organic section. synthesis and analysis are written in Supplemen- tary Materials under the Chemistry section.

Figure 2. Reagents, solvents and conditions: (a) H2SO4, HNO3, 80 °C,◦ 1 week; (b) 4-aminophenol, isopropanol, 80 °C,◦ 40 h; Figure 2. Reagents, solvents and conditions: (a)H2SO4, HNO3, 80 C, 1 week; (b) 4-aminophenol, isopropanol, 80 C, 40 h; (b) 4-aminophenol, NaHCO3, EtOH, 100 °C,◦ 20 h; (c) 3-dimethylamino-1-propyl chloride hydrochloride, Cs2CO3, DMF, (Figureb) 4-aminophenol, 2. Reagents, NaHCOsolvents 3and, EtOH, conditions: 100 C, ( 20a) H h;2SO (c)4 3-dimethylamino-1-propyl, HNO3, 80 °C, 1 week; (b) 4 chloride-aminophenol, hydrochloride, isopropanol, Cs2 CO80 °C,3, DMF, 40 h; 100 °C,◦ 3 h; (d) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C,◦ 16 h; (e) aniline, MeOH, r.t., 40 h; (f) morpholine, MeOH, r.t., 100(b) 4C-aminophenol,, 3 h; (d)(±)-epichlorohydrin, NaHCO3, EtOH, MeCN,100 °C,K 20CO h; (,c 80) 3-dimethylaminoC, 16 h; (e) aniline,-1-propyl MeOH, chloride r.t., 40 hydrochloride, h; (f) morpholine, Cs2 MeOH,CO3, DMF, r.t., 36Figure–48 h; (2.g) Reagents, Me2NH × solventsHCl, MeOH, and conditions: 70 °C, 24 h; ( a(h2) )H 20%2SO3 4wt, HNO Pd/C,3, 80H 2°C,, MeOH, 1 week; r.t., (b 2) h;4- aminophenol,(i) BF3OEt2, MeCN, isopropanol, 80 °C, 16 80 h. °C, 40 h; ◦ 2 3 ◦ 36–48100(b) 4°C,-aminophenol, h; 3 ( gh;) Me(d)2 (±)NH-epichlorohydrin, ×NaHCOHCl, MeOH,3, EtOH, 70MeCN, 100C, °C, 24 K h;20CO ( hh;) , 20%( c80) 3°C,- wtdimethylamino 16 Pd/C, h; (e H) 2aniline,, MeOH,-1-propyl MeOH, r.t., 2chloride h;r.t., (i) 40 BF h;hydrochloride,3OEt (f) 2morpholine,, MeCN, 80Cs 2MeOH,C,CO 163, h.DMF, r.t., 2 2 3 2 31006–48 °C, h; 3 ( gh;) Me(d) NH(±)-epichlorohydrin, × HCl, MeOH, 70 MeCN, °C, 24 h;K 2(COh) 20%3, 80 wt°C, Pd/C, 16 h; H(e), aniline,MeOH, MeOH,r.t., 2 h; r.t.,(i) BF 40 OEth; (f,) MeCN,morpholine, 80 °C, MeOH, 16 h. r.t., 36–48 h; (g) Me2NH × HCl, MeOH, 70 °C, 24 h; (h) 20% wt Pd/C, H2, MeOH, r.t., 2 h; (i) BF3OEt2, MeCN, 80 °C, 16 h.

Figure 3. Reagents, solvents and conditions: (a) 4-aminophenol, NaHCO3, EtOH, 100 °C, 20 h; (b) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C, 16 h; (c) morpholine, MeOH, r.t., 36–48 h; (d) Me2NH × HCl, MeOH, 70 °C, 24 h; (e) 3-dimethylamino- Figure 3. Reagents, solvents and conditions: (a) 4-aminophenol, NaHCO3, EtOH, 100 ◦°C, 20 h; (b) (±)-epichlorohydrin, 1-Figurepropyl 3.chlorideReagents, hydrochloride, solvents and Cs conditions:2CO3, DMF, (a 100) 4-aminophenol, °C, 3 h. NaHCO3, EtOH, 100 C, 20 h; (b)(±)-epichlorohydrin, MeCN, K2CO3, 80 °C,◦ 16 h; (c) morpholine, MeOH, r.t., 36–48 h; (d) Me2NH × HCl, MeOH, 70 °C,◦ 24 h; (e) 3-dimethylamino- MeCN,Figure K3.2 COReagents,3, 80 C, solvents 16 h; (c) morpholine,and conditions: MeOH, (a) 4 r.t.,-aminophenol, 36–48 h; (d) MeNaHCO2NH ×3, HCl,EtOH, MeOH, 100 °C, 70 20C, h; 24 ( h;b) (e(±)) 3-dimethylamino--epichlorohydrin, 2 3 1MeCN,-propyl K chloride2CO3, 80 hydrochloride,°C, 16 h; (c) morpholine, Cs CO , DMF,MeOH, 100 r.t., ◦°C, 3 63– h.48 h; (d) Me2NH × HCl, MeOH, 70 °C, 24 h; (e) 3-dimethylamino- 1-propyl chloride hydrochloride, Cs2CO3, DMF, 100 C, 3 h. 1-propyl chloride hydrochloride, Cs2CO3, DMF, 100 °C, 3 h.

Figure 4. Reagents, solvents, and conditions: (a) 4-aminophenol, BrettPhos, Pd(dba3)2, NaOt-Bu, 90 °C, 12 h; (b) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C, 16 h; (c) Me2NH × HCl, MeOH, 70 °C, 24 h. Figure 4. Reagents, solvents, and conditions: (a) 4-aminophenol, BrettPhos, Pd(dba3)2, NaOt-Bu, 90 °C, 12 h; (b) (±)-epichlo- 2 3 2 rohydrin,Figure 4. Reagents,MeCN, K solventsCO , 80 ,°C, and 16 conditions: h; (c) Me NH(a) 4 ×-aminophenol, HCl, MeOH, 70BrettPhos, °C, 24 h. Pd(dba 3)2, NaOt-Bu, 90 °C, 12◦ h; (b) (±)-epichlo- Figure 4. Reagents, solvents, and conditions: (a) 4-aminophenol, BrettPhos, Pd(dba3)2, NaOt-Bu, 90 C, 12 h; (b)(±)- rohydrin, MeCN, K2CO3, 80 °C, 16 h;◦ (c) Me2NH × HCl, MeOH, 70 °C, 24 h.◦ epichlorohydrin, MeCN, K2CO3, 80 C, 16 h; (c) Me2NH × HCl, MeOH, 70 C, 24 h.

Cancers 2021, 13, 1244 6 of 24

Cancers 2021, 13, x FOR PEER REVIEW 6 of 25

Figure 5. Reagents, solvents, and conditions: (a) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C, 16◦ h; (b) Me2NH × HCl, MeOH, Figure 5. Reagents, solvents, and conditions: (a)(±)-epichlorohydrin, MeCN, K2CO3, 80 C, 16 h; (b) Me2NH × HCl, 70 °C, 24 h.◦ FigureMeOH, 5. 70Reagents,C, 24 h. solvents, and conditions: (a) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C, 16 h; (b) Me2NH × HCl, MeOH, 70 °C, 24 h.

Figure 6. Reagents, solvents, and conditions: (a) 4-aminophenol, K2CO3, CuO, Cu, 2-ethoxyethanol, 130 °C, 16 h; (b) thionyl chloride, anhydrous MeOH, 70 °C, overnight; (c) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C, 16 h; (d) NaOH, MeOH, water, Figure 6. Reagents, solvents, and conditions: (a) 4-aminophenol, K2CO3, CuO, Cu, 2-ethoxyethanol, 130 °C,◦ 16 h; (b) thionyl Figurer.t., 12 h. 6. Reagents, solvents, and conditions: (a) 4-aminophenol, K2CO3, CuO, Cu, 2-ethoxyethanol, 130 C, 16 h; (b) thionyl chloride, anhydrous MeOH, 70 °C,◦ overnight; (c) (±)-epichlorohydrin, MeCN, K2CO3, 80 °C,◦ 16 h; (d) NaOH, MeOH, water, chloride, anhydrous MeOH, 70 C, overnight; (c)(±)-epichlorohydrin, MeCN, K2CO3, 80 C, 16 h; (d) NaOH, MeOH, water, r.t., 12 h. r.t., 12 h. 2.5. Electrophysiological Recordings Stage V-VI oocytes were isolated by partial ovariectomy from Xenopus laevis frogs 2.5.2.5. Electrophysiological Electrophysiological Recordings Recordings (African clawed frog). Mature female frogs were purchased from CRB Xénopes (Rennes, StageStage V V-VI-VI oocytes oocytes were were isolated isolated by by partial partial ovariectomy ovariectomy from from XenopusXenopus laevis laevis frogsfrogs (AfricanFrance) andclawed were frog). housed Mature in the female Aquatic frogs Facility were (KUpurchased Leuven) from in compliance CRB Xénopes with (Rennes, the reg- ulations(African of clawed the European frog). Mature Union female (EU) concerning frogs were the purchased welfare fromof laboratory CRB Xénopes animals (Rennes, as de- France)France) and and were were housed housed in inthe the Aquatic Aquatic Facility Facility (KU (KU Leuven) Leuven) in compliance in compliance with withthe reg- the clared in Directive 2010/63/EU. The use of Xenopus laevis was approved by the Animal ulationsregulations of the of European the European Union Union (EU) (EU) concerning concerning the welfare the welfare of laboratory of laboratory animals animals as de- as Ethics Committee of the KU Leuven (Project nr. P186/2019). After anesthetising the frogs clareddeclared in Directive in Directive 2010/63/EU. 2010/63/EU. The The use use of ofXenopusXenopus laevis laevis waswas approved approved by by the the Animal Animal by a 15-min submersion in 0.1% tricaine methanesulphonate (pH 7.0), the oocytes were EthicsEthics Com Committeemittee of of the the KU KU Leuven Leuven (Project (Project nr. nr. P186/2019). After After anesthetising anesthetising the the frogs frogs collected. The isolated oocytes were then washed with a 1.5 mg/mL collagenase solution byby a a 15 15-min-min submersion submersion in in 0.1% 0.1% tricaine tricaine methanesulphonate methanesulphonate (pH (pH 7.0), 7.0), the the oocytes oocytes were were for 2 h to remove the follicle layer. collected.collected. The The isolated isolated oocytes oocytes were were then then washed washed with with a a 1.5 1.5 mg/mL collagenase collagenase solution solution Ion channels (KV1.x, KV2.1, KV4.2, NaV1.4, NaV1.5) were expressed in Xenopus laevis forfor 2 2 h h to to remove remove the the follicle follicle layer. layer. oocytes, by linearisation of the plasmids and subsequent in vitro transcription using a IonIon channels channels (K (KV1.x,1.x, K KV2.1,2.1, K KV4.2,4.2, Na NaV1.4,1.4, Na NaV1.5)1.5) were were expressed expressed in in XenopusXenopus laevis laevis commercial T7 or SP6V mMESSAGEV V mMACHINEV V transcription kit (Ambion, Carlsbad, oocytes,oocytes, by by linearisation linearisation of of the the plasmids plasmids and and subsequent subsequent inin vitro vitro transcriptiontranscription using using a a CA, USA). Defolliculated Xenopus oocytes were then injected with 20–50 nL of the cRNA commercialcommercial T7 or or SP6 SP6 mMESSAGE mMESSAGE mMACHINE mMACHINE transcription transcriptio kitn kit (Ambion, (Ambion, Carlsbad, Carlsbad, CA, at a concentration of 1 ng/nL using a micro-injector (Drummond Scientific1, Broomall, PA, CAUSA)., USA). Defolliculated DefolliculatedXenopus Xenopusoocytes oocytes were were then then injected injected with with 20–50 20–50 nL nL of of the the cRNA cRNA at atUSA).a a concentration concentration The oocytes of of 1 were1 ng/nL ng/nL incubated using a micro micro-injectorin a solution-injector (Drummondcontaini (Drummondng (in Scientific1, Scientiﬁc1, mM): NaCl, Broomall, Broomall, 96; KCl, PA PA, 2;, USA).CaClUSA).2 ,The 1.8; The MgCloocytes oocytes2, 2were wereand HEPES,incubated incubated 5 (pH in in a 7.4), a solution solution supplemented containi containingng with (in (in 50mM): mM): mg/L NaCl, NaCl, gentamycin 96; 96; KCl, KCl, sul- 2; 2; phate and 90 mg theophylline. After ex vivo translation, the ion channels are correctly CaClCaCl2,2 ,1.8; 1.8; MgCl MgCl2, 22, and 2 and HEPES, HEPES, 5 (pH 5 (pH 7.4), 7.4), supplemented supplemented with with 50 mg/L 50 mg/L gentamycin gentamycin sul- phateinsertedsulphate and in and 90 the mg 90 cell mg theophylline. membrane theophylline. of After the After oocytes. ex exvivo vivo translation, translation, the the ion ion channels channels are are correctly correctly insertedinsertedTwo in in- electhe the trodecell cell membrane membrane voltage-clamp of of the the recordings oocytes. oocytes. were performed at room temperature (18– ◦ 22 °C)TwoTwo-electrode using-elec atrode Geneclamp voltage-clampvoltage 500-clamp amplifier recordings recordings (Molecular were were performed Devices,performed atSan room at Jose, room temperature CA, temperature USA) controlled (18–22 (18C–) 22byusing °C) a pClamp using a Geneclamp a Geneclampdata acquisition 500 ampliﬁer 500 amplifier system (Molecular (Axon (Molecular Instruments1, Devices, Devices, San Jose,SanUnion Jose, CA, City, CA, USA) CA USA) controlled, USA). controlled Whole by a bycellpClamp a currentspClamp data datafrom acquisition acquisitionoocytes system were system recorded (Axon (Axon Instruments1, 1– Instruments1,7 days after Union injection. Union City, City, The CA, bathCA USA)., USA).solution Whole Whole com- cell cellpositioncurrents currents fromwas from either oocytes oocytes ND96 were were (in recorded mM): recorded 1–7NaCl, days 1– 96;7 afterdays KCl, injection.after 2; CaCl injection.2 The, 1.8; bath TheMgCl solution bath2, 2 solutionand composition HEPES, com- 5 (pH 7.5) or calcium- free ND96 supplemented with 10 mM BaCl2. Voltage and current positionwas either was ND96 either (in ND96 mM): (in NaCl, mM): 96; NaCl, KCl, 2; 96; CaCl KCl,2, 1.8;2; CaCl MgCl2, 21.8;, 2 andMgCl HEPES,2, 2 and 5 (pHHEPES, 7.5) or5 electrodes were filled with a 3 M solution of KCl in H2O. Resistances of both electrodes (pHcalcium- 7.5) or free calcium ND96- supplemented free ND96 supplemented with 10 mM BaClwith2 .10 Voltage mM BaCl and2 current. Voltage electrodes and current were were kept between 0.5 and 1.5 MΩ. electrodes were filled with a 3 M solution of KCl in H2O. Resistances of both electrodes were Thekept elicited between K v0.51.x, and Kv2.1 1.5, andMΩ. K v4.2 currents were filtered at 0.5 kHz and sampled at 2 kHz, NaV1.4, and Nav1.5 currents were filtered at 1 kHz and sampled at 20 kHz using a The elicited Kv1.x, Kv2.1, and Kv4.2 currents were filtered at 0.5 kHz and sampled at four-pole low-pass Bessel filter. Leak subtraction was performed using a -P/4 protocol. 2 kHz, NaV1.4, and Nav1.5 currents were filtered at 1 kHz and sampled at 20 kHz using a four-pole low-pass Bessel filter. Leak subtraction was performed using a -P/4 protocol.

Cancers 2021, 13, 1244 7 of 24

filled with a 3 M solution of KCl in H2O. Resistances of both electrodes were kept between 0.5 and 1.5 MΩ. The elicited Kv1.x, Kv2.1, and Kv4.2 currents were filtered at 0.5 kHz and sampled at 2 kHz, NaV1.4, and Nav1.5 currents were filtered at 1 kHz and sampled at 20 kHz using a four-pole low-pass Bessel filter. Leak subtraction was performed using a -P/4 protocol. For the electrophysiological analysis of the compounds, a number of protocols were applied from a holding potential of −90 mV. Currents for KV1.x, KV2.1, and KV4.2 were evoked by 0.5 s depolarising pulses to 0 mV, followed by 0.5 s repolarising pulses to −50 mV. Currents for NaV1.4 and NaV1.5 were evoked by 100 ms depolarising pulses to 0 mV. Whole-cell patch clamp experiments were performed on stably transfected HEK- 293 cells [32,33]. The cells were grown on fibronectin-coated glass coverslips in DMEM supplemented with 10% FCS and 300 µg/mL Zeocin. Experiments were performed at room temperature under continuous perfusion. The extracellular solution contained (mM) 160 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 10 Hepes/NaOH, pH 7.4, 8 glucose. For the experiments determining conductance–voltage relationships, tail currents were made visible by substituting 60 mM NaCl by KCl (162.5 mM KCl, 100 NaCl). Pipettes were pulled from #1 glass capillaries (WPI) to resistances of 1–3 MΩ when loaded with the intracellular solution (100 KCl, 45 N-methyl-D-glucamine, 10 1,2-bis(o-aminophenoxy)ethane-N,N,N0,N0- tetraacetic acid (BAPTA), 10 Hepes pH 7.35). Data acquisition was performed using a EPC10 Plus amplifier (HEKA Elektronik) and Patch Master software. Drug application was performed using an 8-channel fast solution exchanger (ALA). All compounds were dissolved in DMSO to a final concentration of 10 mM. 0.5% DMSO in extracellular solution served as control. KV10.1 currents were elicited by 500 ms depolarisations to +40 mV under continuous perfusion with the compound-containing or control solution. hERG current amplitudes were measured as the peak tail current during a 4 s repolarization to −40 mV after a depolarising pulse to +20 mV lasting for 2 s.

2.6. Statistical Analysis All electrophysiological data are presented as means ± SEM of n ≥ 3 independent experiments unless otherwise indicated. Oocyte data were analysed using pClamp Clampﬁt 10.4 (Molecular Devices, Downingtown, PA, USA) and OriginPro 9 (Originlab, Northamp- ton, MA, USA) or GraphPad Prism 8 software (GraphPad Software, Inc., San Diego, CA, USA). Patch-clamp data analysis was performed using FitMaster (HEKA Electronics (Lud- wigshafen, Germany)), IgorPro (Wavemetrics, Inc, Lake Oswego, OR, USA) and Prism v.8. The Dunnett test and one-way ANOVA were performed to calculate the signiﬁcance of the induced inhibition compared to the control (p < 0.05).

2.7. Cell Culture Cell lines MCF-7 (DSMZ ACC 115), PANC-1 (DSMZ ACC 783), and hTERT-RPE1 (ATCC CRL 4000) were grown in the media recommended by the cell provider in a hu- ◦ 4 midified atmosphere at 5% CO2 and 37 C. For cell growth experiments, 10 cells/well were seeded in 96-well flat bottom culture plates (Corning) and allowed to grow for 24 h. The compounds were then added in the corresponding growth medium, and the plate was imaged every hour for 72 h (two images per well and eight wells per condition in two independent experiments for compound 8 and five for ZVS-08 and compound 1) in an Incucyte system (Sartorius/Essen biosciences). Culture confluence was determined in the phase contrast images and is expressed as percent confluence per well. The growth was then normalised to the value in the absence of compounds. Spheroids were cultured in round bottom ultra-low attachment 96-well plates (Corn- ing) at a density of 5000 cells/well in 2% Matrigel (Corning) and centrifuged at 1000× g for 10 min. Spheroid formation was monitored continuously in the Incucyte system and treatments were added once spheroids were formed. As apoptosis sensitizer, 8.9 µM cycloheximide was added to controls and treated spheroids. Caspase 3/7 green reagent (Sartorius; 1:1000) was used as recommended by the manufacturer. Caspase 3/7 reagent Cancers 2021, 13, 1244 8 of 24

substrate crosses the cell membrane and its cleavage by activated caspase-3/7 results in the release of a DNA dye and ﬂuorescent staining of nuclear DNA.

2.8. Evaluation of Mutagenic Activity with Salmonella/Microsomal Reverse Mutation Assay Mutagenicity of compounds ZVS-08 and 8 was tested with the Salmonella/microsomal reverse mutation (Ames) assay on Salmonella typhimurium strains TA98 and TA100 without and with external metabolic activation according to the OECD guidelines (OECD TG 471) with minor modifications. Briefly, 100 µL of ZVS-08 and 8 (0.158, 0.5, 1.58, 5, 15.8, and 50 mg/mL corresponding to the final concentration of 0.0158, 0.05, 0.158, 0.5, 1.58, and 5 mg/plate, respectively), 100 mL bacterial culture of S. typhimurium (overnight culture at 37 ◦C), and 500 mL of phosphate buffer (for assays without metabolic activation) or 10% S9 mix (for assays with metabolic activation) were added to 2 mL of molten top agar (supplemented with histidine/biotin, final concentration 0.05 mM), gently mixed and poured onto minimal agar plates. The number of His+ revertants was counted after 48 h (TA100) and 72 h (TA98) of incubation at 37 ◦C. Three plates were used per treatment point. The mutagenic potential of the compounds was expressed as an induction factor (IF), where IF = (number of revertants in the presence of the compound)/(number of revertants in solvent control). Additionally, each plate was checked for possible toxic effects of tested compounds. The concentration is considered as cytotoxic when the microcolony lawn is reduced and is classified as moderately (distinguished by a marked thinning of the bacterial lawn that may result in a pronounced increase in the size of microcolonies compared to the solvent control plate), extremely reduced (distinguished by an extreme thinning of the bacterial background lawn that may result in an increase in the size of the microcolonies compared to the solvent control plate such that the microcolony lawn is visible to the unaided eye as isolated colonies) or absent (distinguished by a complete lack of any bacterial background lawn over greater than or equal to 90% of the plate). 4-nitroquinoline- N-oxide (4-NQNO; 0.25 µg/plate) for TA98 and sodium azide (NaN3; 0.25 µg/plate) for TA100 were used as positive controls without S9 metabolic activation, while benzo[a]pyrene (B[a]P; 2.5 µg/plate) was used as positive controls for both bacterial strains with S9 metabolic activation.

3. Results

3.1. Ligand-Based Identiﬁcation of Novel KV10.1 Inhibitors through in Silico Screening

The discovery of KV10.1 inhibitors has been limited to reports of known compounds or drugs that bind to this channel mostly as an off-target (Figure1). Even though the three- dimensional structure of KV10.1 has been determined by cryo-electron microscopy [2], the actual binding sites of small molecules in KV10.1 remain mostly unknown. There- fore, structure-based design and virtual screening are rarely used to identify new KV10.1 inhibitors. Limitations of structure-based in silico approaches in the design of KV10.1 inhibitors can be overcome by ligand-based methods, such as 3D similarity search and pharmacophore modelling. The latter method can be used to build ligand-based pharmacophore models (LBPM) based on a set of active ligands. The pharmacophore model consists of a set of interactions aligned in 3D space and is represented by a collection of pharmacophore features with deﬁned properties, such as aromatic ring, hydrogen bond donors and acceptors, hydrophobic, negatively and positively charged groups, and halogen bonds features. The known active compounds are deﬁned as a training set and aligned in 3D space, followed by the generation of a pharmacophore model. One of the most important requirements for generating ligand-based pharmacophore models is that the ligands have the same mode of action and bind in the same orientation in the binding site. Therefore, to generate and validate the LBPM, we used a set of 11 purpurealidin I analogues that showed KV10.1 inhibition [22]. This is the only series of analogues that was evaluated for inhibition of KV10.1. They also displayed antiproliferative effects in cancer cell lines with high KV10.1 expression. Based on competition experiments Cancers 2021, 13, 1244 9 of 24

with mibefradil, a suggested binding site for these compounds is in the VSD of KV10.1, although its exact location remains to be elucidated experimentally [22]. The LBPM method circumvents this limitation as the model is based solely on the alignment of the ligands in the training set. Due to the similarity between the ligands, we assumed that the binding site is shared among the active purpurealidin analogues, which is necessary for a valid LBPM generation. The higher diversity between KV10.1 and hERG channels in the VSD than in the pore of the channels highlights purpurealidin analogues as excellent starting points for identifying novel KV10.1 inhibitors [34]. In virtual screening using LBPM as a query, the ligands in the screening library are clustered into active and decoy (inactive) datasets, and then, their pharmacophore fit scores are calculated and ranked in the hitlist from highest to lowest score. Potentially active compounds have high pharmacophore fit scores, while decoys poorly fit the pharmacophore model and receive low scores. Screening results can be visualised by Receiver Operating Characteristic (ROC) plots, where the rate of true positives (active compounds) is on the y-axis and the rate of false positives (decoys) is on the x-axis. If the ROC plot curve follows the diagonal line, it means that the model is performing poorly and cannot distinguish between decoy and active compounds. A ROC curve plotted above the diagonal represents a pharmacophore model that can detect active compounds. In addition, the enrichment factor (EF) represents the number of active compounds identified by a query pharmacophore model, as opposed to the number of compounds hypothetically found to be active when randomly screened. Thus, an enrichment factor of 1 means that the ligands are randomly sorted. In contrast, high enrichment factors mean that only a small fraction of the hitlist needs to be screened in vitro to identify a high number of active compounds. To evaluate the performance of LBPM before running the real virtual screening campaign, it needs to be trained and validated. Ideally, the model will find a high number of active ligands from the training set and no ligands from the decoy set. Ligand-based pharmacophore modelling was performed in LigandScout 4.3 Expert [35,36]. Ten 3D LBPM with the best alignments were generated and scored with pharmacophore alignment scores. Exclusion volume features (Figure7A, grey spheres) were automatically added to each model based on the shape of the ensemble of aligned molecules. These features represent restricted space and are important to increase the selectivity of a model for virtual screening campaigns. An initial pharmacophore model was constructed based on the alignment of the 11 most potent purpurealidin-based KV10.1 inhibitors and is shown in Figure7A[ 22]. It consisted of 14 pharmacophore features, 12 of which were essential (four hydrophobic features, marked as yellow spheres; two hydrogen bond acceptors, marked as red spheres; one hydrogen bond donor, marked as a green arrow; two aromatic features, marked as blue discs; one positive ionisable feature, marked as a blue star; and two halogen bond features, marked as pink arrows), and two were marked as optional (dashed red circle for one hydrogen bond acceptor and dashed pink arrow for one halogen bond). For each active compound in the dataset, 50 decoys were generated using the DUD-E server [24]. The DUD-E database calculates decoys based on the given set of KV10.1 inhibitors, which means that some of the decoys could still inhibit KV10.1 as they are similar to the active molecules. However, this is highly unlikely. This initial LBPM was tested with a library of active and decoy molecules. The performance of this model was very good in terms of the enrichment factor of 51.1, finding 9 out of 11 active compounds from the dataset and no decoy ligands (Figure7B). However, it was too restrictive and performed poorly in virtual screening of commercially available compounds as no hits were identified. Therefore, our next step was to simplify the original LBPM while retaining its favourable screening properties, with the model able to distinguish between active and inactive compounds. Cancers 2021, 13, 1244 10 of 24

Cancers 2021, 13, x FOR PEER REVIEW 10 of 25

V FigureFigure 7. (A 7.) ( InitialA) Initial ligand-based ligand-based pharmacophore pharmacophore model model for KforV10.1 K 10 inhibitors.1 inhibitors based based on aon library a library of of purpurealidinpurpurealidin analogues. analogues. Common Common chemical chem featuresical features of the of aligned the aligned molecules molecules include include 4 hydrophobic 4 hydrophobic features (yellow spheres), 3 hydrogen bond acceptors (red spheres), 1 hydrogen bond do- features (yellow spheres), 3 hydrogen bond acceptors (red spheres), 1 hydrogen bond donor (a green nor (a green arrow), 2 aromatic features (blue discs), 1 positive ionisable feature (blue star), 3 halo- arrow), 2 aromatic features (blue discs), 1 positive ionisable feature (blue star), 3 halogen bond gen bond features (pink arrows), and exclusion volumes (grey spheres) that define restricted re- featuresgions (pink based arrows), on the shape and exclusion of the aligned volumes molecules. (grey spheres) Optional that features define restrictedare marked regions as dashed. based ( onC) theFinal shape ligand of the- alignedbased pharmacophore molecules. Optional model. features Common are chemical marked as features dashed. of ( Cthe) Final aligned ligand-based molecules pharmacophoreinclude 2 hydrogen model. bond Common acceptors chemical (red spheres), features of1 hydrogen the aligned bond molecules donor (a include green sphere), 2 hydrogen 2 aro- bondmatic acceptors features (red (blue spheres), discs), 1and hydrogen 1 positive bond ionisable donor feature (a green (blue sphere), star). 2Grey aromatic spheres features for exclusion (blue discs),volumes and 1 positiveare not shown. ionisable Resulting feature (blueROC star).plot (curve Grey spheres shown forin blue) exclusion for (B volumes) initial and are not(D)shown. final ResultingLBPM ROCfrom plotvirtually (curve screening shown in 562 blue) compounds for (B) initial (11 andKV10 (D.1 )active finalLBPM compounds from virtually and 551 screeninggenerated decoys) with the ligand-based pharmacophore model. TP = true positives; FP = false positives; 562 compounds (11 KV10.1 active compounds and 551 generated decoys) with the ligand-based AUC = area under the curve; EF = enrichment factor. pharmacophore model. TP = true positives; FP = false positives; AUC = area under the curve; EF = enrichment factor. An initial pharmacophore model was constructed based on the alignment of the 11 mostThrough potent several purpurealidin rounds- ofbased simplification, KV10.1 inhibitors we obtained and is shown our final in LBPM,Figure 7 whichA [22] was. It con- usedsisted for virtualof 14 pharmacophore screening. Specifically, features, 12 we of eliminated which were both essential optional (four pharmacophore hydrophobic fea- featurestures, and marked removed as yellow halogen spheres bond;and two hydrophobic hydrogen bond features. acceptors In addition,, marked we as converted red spheres; theone directional hydrogen hydrogen bond donor, bond marked donor featureas a green to a arrow non-vector; two aromatic feature. Ourfeatures, final marked model as (Figureblue7 discsC) consisted; one positive of six pharmacophoreionisable feature, features: marked twoas a aromatic,blue star; one and hydrogen two halogen bond bond donor,features, two hydrogen marked as bond pink acceptors, arrows), and two one positivewere marked ionisable as opt feature.ional (dashed The performance red circle for of theone model hydrogen in the bond validation acceptor screen and dashed was optimal pink arrow as it was for ableone tohalogen find all bond). active For ligands each ac- andtive no decoycompound compounds, in the dataset, resulting 50 decoys in a high were enrichment generated factor using of the 51.1 DUD (Figure-E server7D). When [24]. The this model was used in the virtual screening campaign, it was able to identify 18 potentially DUD-E database calculates decoys based on the given set of KV10.1 inhibitors, which active compounds from the library of 556,000 diverse commercially available compounds. means that some of the decoys could still inhibit KV10.1 as they are similar to the active The in silico hits were visually inspected and nine compounds (Table1) were selected and molecules. However, this is highly unlikely. This initial LBPM was tested with a library purchased from the commercial vendors. of active and decoy molecules. The performance of this model was very good in terms of the enrichment factor of 51.1, finding 9 out of 11 active compounds from the dataset and no decoy ligands (Figure 7B). However, it was too restrictive and performed poorly in

CancerCancerCancers s2021s 20212021, 13,, 1313, x,, xxFOR FORFOR PEER PEERPEER REVIEW REVIEWREVIEW 111111 of ofof 25 2525 Cancer Cancer s 2021s 2021, 13, 13, x, FORx FOR PEER PEER REVIEW REVIEW 1111 of of 25 25 Cancer Cancerss 20212021,, 1313,, xx FORFOR PEERPEER REVIEWREVIEW 1111 ofof 2525 Cancers 2021, 13, x FOR PEER REVIEW 11 of 25 Cancer s 2021, 13, x FOR PEER REVIEW 11 of 25

virtual screening of commercially available compounds as no hits were identified. There- virtualvirtualvirtualvirtual screening screening screeningscreening of of ofof commercially commercially commerciallycommercially available available availableavailable compounds compounds compoundscompounds as as asas no no nono hits hits hitshits were were werewere identified. identified. identified.identified. There- There- There-There- virtualvirtualfore, our screeningscreening next step ofof was commerciallycommercially to simplify availabletheavailable original compoundscompounds LBPM while asas noretainingno hitshits werewere its favourable identified.identified. screen- There-There- fore,fore,fore,virtualfore,virtual our our ourour screeningnext screeningnext nextnext step step stepstep was wasof ofwaswas commercially commerciallyto to toto simplify simplify simplifysimplify the the thetheavailable available original original originaloriginal compounds LBPM compoundsLBPM LBPMLBPM while while whilewhile as retainingas retaining retainingretainingno no hits hits wereits wereits itsits favourable favourable favourablefavourable identified. identified. screenscreen- screenscreen-There- There- fore,fore,ing properties, ourour nextnext stepstep with waswas the toto simplify simplify model theablethe originaloriginal to distinguish LBPMLBPM whilewhile between retainingretaining active itsits favourableandfavourable inactive screen-screen- com- ingingingfore,ingfore, properties, properties, properties, properties,our our next next step step with with with with was was the the the the to to model simplify model simplify model model able able able ablethe the to original tooriginal to to distinguish distinguish distinguish distinguish LBPM LBPM betweenwhile betweenwhile between between retaining retaining act act act activeiveiveive its its and and and andfavourable favourable inactive inactive inactive inactive screen- com-screen- com- com- com- ingingpounds. properties, properties, with with the the model model able able to to distinguish distinguish between between act activeive and and inactive inactive com- com- pounds.pounds.pounds.ingpounds.ing properties, properties, with with the the model model able able to to distinguish distinguish between between act activeive and and inactive inactive com- com- pounds.pounds.Through several rounds of simplification, we obtained our final LBPM, which was pounds.pounds.ThroughThroughThroughThrough several several severalseveral rounds rounds roundsrounds of of ofof simplification, simplification, simplification,simplification, we we wewe obtainedobtained obtainedobtained our our ourour finalfinal finalfinal LBPM, LBPM, LBPM,LBPM, which which whichwhich was was waswas usedThroughThrough for virtual severalseveral screening. roundsrounds Specifically, ofof simplification,simplification, we eliminated wewe obtainedobtained both optional ourour finalfinal pharmacophore LBPM,LBPM, whichwhich was wasfea- usedusedusedused for for Through forThroughfor virtual virtual virtualvirtual severalscreening. severalscreening. screening.screening. roundsrounds Specifically, Specifically, Specifically,Specifically, ofof simplification,simplification, we we wewe eliminated eliminated eliminatedeliminated wewe obtainedobtained both both bothboth optional optional optionaloptional ourour finalfinal pharmacophore pharmacophore pharmacophorepharmacophore LBPM,LBPM, whichwhich fea- fea- wasfea-fea-was usedusedtures for forand virtualvirtual removed screening.screening. halogen Specifically,Specifically, bond and wehydrophobicwe eliminatedeliminated features. bothboth optionaloptional In addition, pharmacophorepharmacophore we converted fea-fea- turesturesturesusedturesused andand for andandfor removedvirtual removed virtual removedremoved screening. screening. halogenhalogen halogenhalogen Specifically, bondSpecifically,bond bondbond andand andand hydrophobichydrophobic hydrophobicwehydrophobic we eliminated eliminated features.features. features.features. both both optional optional InIn InIn aa ddition,aaddition,ddition,ddition, pharmacophore pharmacophore wewe wewe convertedconverted convertedconverted fea- fea- turesturesthe directional andand removedremoved hydrogen halogenhalogen bond bondbond donor andand feature hydrophobichydrophobic to a non -features.features.vector fe ature.InIn aaddition,ddition, Our final wewe model convertedconverted (Fig- thethetheturesthetures directional directional directionaldirectional andand removedremoved hydrogen hydrogen hydrogenhydrogen halogenhalogen bond bond bondbond bonddonor bonddonor donordonor and featureand feature featurefeature hydrophobichydrophobic to to toto a a anona non nonnon--vector vector-- features.vectorvectorfeatures. fe fe fefeature.ature. ature.ature. InIn aa Ourddition, ddition,Our OurOur final final finalfinal we modelwe model modelmodel convertedconverted (Fig- (Fig- (Fig-(Fig- thetheure directional directional7C) consisted hydrogenhydrogen of six pharmacophore bondbond donordonor featurefeature features: toto aa nonnon two--vectorvector aromatic, fefeature.ature. one OurOur hydrogen finalfinal modelmodel bond (Fig-(Fig- do- ureureuretheurethe 7 7 C)directional 77directionalC)C)C) consisted consisted consistedconsisted hydrogen hydrogen of of ofof six six sixsix pharmacophore pharmacophore pharmacophore pharmacophorebond bond donor donor feature feature features: features: features:features: to to a a nontwo nontwo twotwo -aromatic,- vectoraromatic,vector aromatic,aromatic, fe feature.ature. one one oneone hydrogenOur hydrogen Ourhydrogenhydrogen final final model bond modelbond bondbond do-(Fig- do-(Fig- do-do- ureurenor, 77 twoC)C) consistedconsisted hydrogen ofof bond sixsix pharmacophorepharmacophore acceptors, and features:onefeatures: positive twotwo ionisable aromatic,aromatic, f eature.oneone hydrogenhydrogen The performance bondbond do-do- nor,nor,nor,urenor,ure two two 7 two7twoC)C) hydrogen consistedhydrogen consistedhydrogenhydrogen bond ofbond of bondbond sixsix acceptors, pharmacophoreacceptors, pharmacophoreacceptors,acceptors, and and andand one one oneone features:features: positive positive positivepositive twotwo ionisable ionisable ionisableionisable aromatic, aromatic, f f eature. feature.feature. eature.one one hydrogenhydrogenThe The TheThe performance performance performanceperformance bondbond do-do- nor,nor,of the twotwo model hydrogenhydrogen in the bondbondvalidation acceptors,acceptors, screen andand was oneone optimal positivepositive as itionisableionisable was able ffeature. eature.to find TheTheall active performanceperformance ligands ofofofnor,of nor,the the thethe twomodeltwomodel modelmodel hydrogenhydrogen in in inin the the thethe validation validation validation bondvalidationbond acceptors,acceptors, screen screen screenscreen was andwas and waswas optimal oneoptimal one optimaloptimal positivepositive as as asas it it ititwas ionisablewasionisable waswas able able ableable to fto f eature.totoeature. find find findfind all all TheallallThe active active activeactive performanceperformance ligands ligands ligandsligands ofofand thethe no modelmodel decoy inin compounds, thethe validationvalidation resulting screenscreen in waswas a high optimaloptimal enrichment asas itit waswas factor ableable of toto 51.1 findfind (Figure allall activeactive 7D). ligandsligands When andandandofandof theno theno nono decoy model decoy modeldecoydecoy compounds, compounds, in compounds,incompounds, the the validation validation resulting resulting resultingresulting screen screen in in inin a was a was ahigha high highhigh optimal optimal enr enr enrenrichmentichmentichmentichment as as it it wasfactor wasfactor factorfactor able able of of ofof 51.1to 51.1to 51.151.1 find find (Figure (Figure (Figure(Figure all all active active 7 7 D). 77D).D).D). Whenligands Whenligands WhenWhen andandthis nonomodel decoydecoy was compounds,compounds, used in the resultingresulting virtual screening inin aa highhigh enrenrcampaigichmentichmentn, itfactorfactor was ofableof 51.151.1 to (Figure (Figureidentify 77 D).D).18 poten- WhenWhen thisthisthisandthisand model model modelno modelno decoy decoy was was waswas compounds, used compounds,used usedused in in inin the the thethe virtual resulting virtual resulting virtualvirtual screening screening screeningscreening in in a a high high campaig campaig enrcampaig campaigenrichmentichmentn,n,n,n, it it it itwasfactor wasfactor waswas able able ableofable of 51.1 to 51.1to toto identify identify(Figure identifyidentify(Figure 187 18 7 18D).18D). poten- poten- poten- poten-When When Cancers 2021 13 thisthistially modelmodel active waswas compounds usedused inin thethe from virtualvirtual the library screeningscreening of 556,000 campaigcampaig diversen,n, itit waswas commercially ableable toto identifyidentify available 1818 poten-poten- com- , , 1244 tiallytiallytiallythistiallythis active modelactive model activeactive compounds wascompounds wascompoundscompounds used used in in from from the fromthefrom virtual virtualthe the thethe library library librarylibrary screening screening of of ofof 556,000 556,000 556,000556,000 campaig campaig diverse diverse diversediversen,n, it it commercially was commercially was commerciallycommercially ableable to to identify identify available available availableavailable 18 1811 com- poten-com- poten- com- ofcom- 24 tiallytiallypounds. activeactive The compoundscompounds in silico hits fromfrom were thethe visually librarylibrary ofinspectedof 556,000556,000 diverseanddiverse nine commerciallycommercially compounds available(availableTable 1) com-com-were pounds.pounds.pounds.tiallypounds.tially active active The The TheThe incompounds in compoundsinin silico silico silicosilico hits hits hitshits from werefromwere werewere the visuallythe visually visuallyvisually library library inspected inspected inspectedof inspectedof 556,000 556,000 and and andanddiverse diverse nine nine ninenine compoundscommercially compoundscommercially compoundscompounds ( (Table (Tableavailable (TableavailableTable 1) 1) 1)1) were were com- werewerecom- pounds.pounds.selected TheandThe purchasedinin silicosilico hitshits from werewere the visuallyvisually commercial inspectedinspected vendors andand. ninenine compoundscompounds ((TableTable 1)1) werewere selectedselectedselectedpounds.selectedpounds. and and andand TheThe purchased purchased purchasedpurchased inin silicosilico hitsfrom hitsfrom fromfrom were werethe the thethe commercial commercial visually commercialvisuallycommercial inspectedinspected vendors vendors vendorsvendors and.and . .. ninenine compoundscompounds ((TableTable 1)1) werewere selectedselectedselected and andand purchased purchasedpurchased from fromfrom the thethe commercial commercialcommercial vendors vendorsvendors.. . Tableselected 1. List and of purchased screening compoundsfrom the commercial with pharmacophore vendors. ﬁt scores and activity on KV10.1 at TableTableTable 1. 1.1. List ListList of ofof screening screeningscreening compounds compoundscompounds with withwith pharmacophore pharmacophorepharmacophore fit fitfit scores scoresscores and andand activity activityactivity on onon K KKV10.1VV10.110.1 at atat TableTable 1. 1. List List of of screening screening compounds compounds with with pharmacophore pharmacophore fit fit scores scores and and activity activity on on K VK10.1V10.1 at at compoundTableTable 1.1. ListList concentration ofof screeningscreening of compoundscompounds 50 µM. withwith pharmacophorepharmacophore fitfit scoresscores andand activityactivity onon KKVV10.110.1 atat compoundcompoundcompoundTable 1. List concentration concentrationconcentration of screening of of ofcompounds 50 5050 μM. μM.μM. with pharmacophore fit scores and activity on KV10.1 at compoundcompoundTable 1. List concentration concentration of screening of ofcompounds 50 50 μM. μM. with pharmacophore fit scores and activity on KV10.1 at compoundcompoundcompound concentration concentrationconcentration of ofof 50 5050 μM. μM.μM. compound concentration of 50 μM. Pharmacophore Fit Effect on Kv10.1 (% Compound ID Structure PharmacophorePharmacophorePharmacophore Fit FitFit EffectEffectEffect on onon Kv10.1 Kv10.1Kv10.1 (% (%(% CompoundCompoundCompound ID IDID StructureStructureStructure PharmacophorePharmacophorePharmacophoreScore Fit FitFit EffectEffectEffectInhibition) on onon Kv10.1 Kv10.1Kv10.1 (% (%(% CompoundCompound ID ID StructureStructure PharmacophoreScoreScoreScore Fit EffectInhibition)Inhibition)Inhibition) on Kv10.1 (% CompoundCompoundCompound ID IDID StructureStructureStructure PharmacophoreScoreScore Fit EffectInhibition)Inhibition) on Kv10.1 (% Compound ID Structure ScoreScoreScore Inhibition)Inhibition)Inhibition) ZVS-01ZVS-01 66.490066.4900Score −−1.13Inhibition)1.13± ±4.50 4.50 (3) (3) ZVSZVSZVSZVS--0101--0101 66.490066.490066.490066.4900 −−1−−1.1311.13.13.13 ± ± ±4.50± 4.50 4.504.50 (3) (3) (3)(3) ZVSZVSZVS---010101 66.490066.490066.4900 −−−111.13.13.13 ± ±± 4.50 4.504.50 (3) (3)(3) ZVS-01 66.4900 −1.13 ± 4.50 (3) ZVS-02 66.4745 5.03 ± 2.87 (3) ZVSZVSZVS-02ZVSZVS--0202--0202 66.474566.474566.474566.4745 5.03 5.035.035.03 ± ±± ±2.87± 2.87 2.872.87 (3) (3)(3) (3)(3) ZVSZVSZVS---020202 66.474566.474566.4745 5.035.035.03 ± ±± 2.87 2.872.87 (3) (3)(3) ZVS-02 66.4745 5.03 ± 2.87 (3) ZVS-03 66.4421 −1.14 ± 3.47 (5) ZVSZVSZVS-03ZVSZVS--0303--0303 66.442166.442166.442166.4421 −1−−1−1−11.14.14.14.14.14 ± ± ±3.47± 3.47 3.47 3.473.47 (5) (5) (5)(5) ZVSZVSZVS---030303 66.442166.442166.4421 −1−1−1.14.14.14 ± ±± 3.47 3.473.47 (5) (5)(5) ZVS-03 66.4421 −1.14 ± 3.47 (5)

ZVS-04 66.0596 −0.86 ± 0.75 (5) ZVSZVSZVS-04ZVSZVS--0404--0404 66.059666.059666.059666.0596 −0−−0−0−00.86.86.86.86.86 ± ± ±0.75± 0.75 0.75 0.750.75 (5) (5) (5)(5) ZVSZVSZVS---040404 66.059666.059666.0596 −0−0−0.86.86.86 ± ±± 0.75 0.750.75 (5) (5)(5) ZVS-04 66.0596 −0.86 ± 0.75 (5)

ZVS-05 65.9545 8.84 ± 4.92 (6) ZVSZVSZVS-05ZVSZVS--0505--0505 65.954565.954565.954565.9545 8.84 8.848.848.84 ± ±± ±4.92± 4.92 4.924.92 (6) (6)(6) (6)(6) ZVSZVS--0505 65.954565.9545 8.848.84 ±± 4.924.92 (6)(6) ZVSZVS--0505 65.954565.9545 8.848.84 ± ± 4.92 4.92 (6) (6)

ZVSZVSZVS-06--0606 65.850165.850165.8501 −7−7−7.50.50.50 ± ± ±4 4.654.65.65 (4) (4)(4) ZVSZVS-06ZVSZVS-06--0606 65.850165.850165.850165.8501 −7−−7−77.50.50.50.50 ±± ±4± 4.65.65 44.65.65 (4) (4) (4)(4) ZVSZVS--0606 65.850165.8501 −7−7.50.50 ± ± 4 4.65.65 (4) (4)

ZVS-07 65.8019 4.68 ± 1.63 (5) ZVSZVSZVSZVS--0707--0707 65.801965.801965.801965.8019 4.684.684.684.68 ± ± ±1.63± 1.63 1.631.63 (5) (5) (5)(5) ZVS-07ZVSZVS--0707 65.801965.801965.8019 4.684.684.68± ±±1.63 1.631.63 (5) (5)(5) ZVSZVS--0707 65.801965.8019 4.684.68 ± ± 1.63 1.63 (5) (5)

ZVS-08 65.7908 80.27 ± 0.63 (3) ZVSZVSZVSZVS--0808--0808 65.790865.790865.790865.7908 80.2780.2780.2780.27 ± ± ±0.63± 0.63 0.630.63 (3) (3) (3)(3) ZVS-08ZVSZVSZVS---080808 65.790865.790865.790865.7908 80.2780.2780.2780.27± ± ±±0.63 0.63 0.630.63 (3) (3) (3)(3) ZVS-08 65.7908 80.27 ± 0.63 (3)

ZVS-09 65.1035 0.15 ± 1.67 (4) ZVSZVSZVSZVS--0909--0909 65.103565.103565.103565.1035 0.150.150.150.15 ± ± ±1.67± 1.67 1.671.67 (4) (4) (4)(4) ZVSZVSZVS---090909 65.103565.103565.1035 0.150.150.15 ± ±± 1.67 1.671.67 (4) (4)(4) ZVS-09ZVS-09 65.103565.1035 0.150.15± ±1.67 1.67 (4) (4) CompoundCompoundCompound ID: ID:ID: screening screeningscreening name namename of ofof the thethe compound; compound;compound; Pha PhaPharmacophorermacophorermacophore fit fitfit score: score:score: scoring scoringscoring function functionfunction calcu- calcu-calcu- CompoundCompound ID: ID: screening screening name name of of the the compound; compound; Pha Pharmacophorermacophore fit fit score: score: scoring scoring function function calcu- calcu- latedCompoundlatedCompoundlatedCompound from fromfrom the thethe ID: ID:ID: alignment alignment alignment screening screeningscreening of of nameof name namethe thethe pharmacophore pharmacophoreof pharmacophoreofof the thethe compound; compound;compound; features featuresfeatures Pha PhaPharmacophorermacophorermacophore and andand the thethe feature featurefeature fit fitfit score: score:score: RMS RMSRMS scoring scoring scoringdeviation. deviation.deviation. function functionfunction Inhibition InhibitionInhibition calcu- calcu-calcu- latedCompoundlatedCompound from from the ID: the ID: alignment screening alignment screening name of of namethe ofthe thepharmacophore pharmacophoreof compound; the compound; Pharmacophore features features Pharmacophore and and ﬁt thescore: the feature feature scoring fit score: RMS RMS function scoringdeviation. deviation. calculated function Inhibition Inhibition from calcu- the datalateddatalateddatalated are are arefrom fromfrom means meansmeans the thethe ±alignment alignment alignment ±±standard standardstandard of oferrorof errorerror the thethe for pharmacophore pharmacophore pharmacophoreforfor the thethe number numbernumber of featuresof offeaturesfeatures independent independentindependent and andand the thethe experiments experiments experiments feature featurefeature RMS RMSRMS indicated indicatedindicated deviation. deviation.deviation. in inin paren- Inhibition Inhibition Inhibitionparen-paren- dataalignmentdatalated are are from means means of the ± pharmacophorealignment ±standard standard errorof error the features for pharmacophore for the the and number number the feature of offeatures independent RMSindependent deviation. and the experiments experiments Inhibitionfeature RMS data indicated indicated deviation. are means in in paren-± Inhibition paren-standard theses.datatheses.datatheses.data are are are means meansmeans ± ±± standard standardstandard error errorerror for forfor the thethe number numbernumber of ofof independent independentindependent experiments experimentsexperiments indicated indicatedindicated in inin paren- paren-parentheses.errortheses.data for are the means number ± standard of independent error experimentsfor the number indicated of independent in parentheses. experiments indicated in parentheses.theses.theses. theses.

3.2. Patch-Clamp Screening of Virtual Screening Hit Compounds

The nine selected in silico hits were tested by manual patch-clamp on a HEK293 cell line stably expressing KV10.1 [33] at a compound concentration of 50 µM (Table1). Compound activity was tested by determining the current amplitude elicited by repetitive

stimuli to +40 mV from a holding potential of −80 mV. The most potent inhibitor was hit

compound ZVS-08 and after its resynthesis an IC50 value of 3.70 ± 1.01 µM was determined against KV10.1. The hit compound ZVS-08 represents a new structural class of KV10.1 inhibitors with diarylamine structure and has a unique structure among the acquired virtual screening hits. The alignment of ZVS-08 in the pharmacophore model used in the virtual screening shows that all pharmacophore features of the model are fulﬁlled (Figure8). The most notable change from the purpurealidin analogues is the replacement of the hydrogen bond acceptor group of the amide bond with the oxygen atom of the nitro group of ZVS-08. Cancers 2021, 13, x FOR PEER REVIEW 12 of 25

3.2. Patch-Clamp Screening of Virtual Screening Hit Compounds The nine selected in silico hits were tested by manual patch-clamp on a HEK293 cell line stably expressing KV10.1 [33] at a compound concentration of 50 μM (Table 1). Com- pound activity was tested by determining the current amplitude elicited by repetitive stimuli to +40 mV from a holding potential of −80 mV. The most potent inhibitor was hit compound ZVS-08 and after its resynthesis an IC50 value of 3.70 ± 1.01 μM was determined against KV10.1. The hit compound ZVS-08 represents a new structural class of KV10.1 inhibitors with diarylamine structure and has a unique structure among the acquired virtual screening hits. The alignment of ZVS-08 in the pharmacophore model used in the virtual screening shows that all pharmacophore features of the model are fulfilled (Figure 8). The Cancers 2021, 13, 1244 12 of 24 most notable change from the purpurealidin analogues is the replacement of the hydrogen bond acceptor group of the amide bond with the oxygen atom of the nitro group of ZVS- 08.

FigureFigure 8.8. AlignmentAlignment ofof the the identiﬁed identified potent potent K KV10.1V10.1 hit hit compound compound ZVS-08 ZVS-08 with with the the ﬁnal final ligand-based ligand- pharmacophorebased pharmacophore model usedmodel for used virtual for screening virtual screening (middle), (middle), structures structures of the most of activethe most purpurealidin active analoguepurpurealidin (right) analogue and virtual (right) screening and virtual hit compound screening ZVS-08hit compound (left). ZVS-08 (left).

3.3.3.3. CharacterisationCharacterisation ofof thethe KKVV10.1 Inhibition by ZVS ZVS-08-08 TheThe inhibitioninhibition by by ZVS-08 ZVS-08 was was characterised characterised in in more more detail detail by by patch patch clamp. clamp. Most Most hERGhERG blockersblockers bindbind to aa hydrophobichydrophobic cavitycavity in the inner mouth of the channel pore [[20]20],, andand somesome well-characterisedwell-characterised compoundscompounds useuse thethe equivalentequivalent structurestructure inin KKVV10.110.1 (e.g.,(e.g., [[32]32])) toto interfereinterfere withwith ionion permeation.permeation. TheThe electrophysiologicalelectrophysiological behaviourbehaviour ofof ourour hithit compoundcompound ZVS-08ZVS-08 was was similar similar to tothat that of astemizole, of astemizole, a well-established a well-established open-channel open-channel blocker blocker of KV10.1 of withKV10.1 an with IC50 anin IC the50 in range the range of 200 of nM 200 [ 33nM]. [33] The. The inhibition inhibition was was concentration-dependent concentration-depend- (Figureent (Fig9ureA,B), 9A with,B), anwith apparent an apparent IC 50 ofIC 3.7050 of µ3.7M0 (TableµM (Table2). The 2) onset. The onset of inhibition of inhibition was slow was andslow required and required minutes minutes of continued of continued stimulation stimulation (Figure (Figure9C), which 9C), wouldwhich bewould compatible be com- withpatible an with intracellular an intracellular binding binding site. Moreover, site. Moreover, the inhibition the inhibition was not was homogeneous not homogeneous during theduring stimulus. the stimulus. Current Current amplitude amplitude decreased decreased along the along stimulus the stimulus in the presence in the presence of ZVS-08, of resultingZVS-08, resulting in apparent in inactivationapparent inactivation and smaller and amplitude smaller amplitude at the end ofat thethe pulseend of than the inpulse the beginningthan in the (Figure beginning9A,D). (Figure The effect 9A,D). was The not effect cumulative, was not cumulative, and the next and stimulus the next (2.3 stimulus s later) showed(2.3 s later) a similar showe peakd a similar current peak and current decay, indicatingand decay, that indicating the compound that the iscompound not trapped is not in thetrapped closed in channel. the closed The channel. apparent The inactivation apparent inactivation was more pronounced was more pronounced at more depolarised at more potentials (Figure9D). Additionally, to the changes in inactivation of K V10.1, ZVS-08 also depolarised potentials (Figure 9D). Additionally, to the changes in inactivation of KV10.1, noticeablyZVS-08 also slowed noticeably down slowed deactivation down dea ofctivation the channel of the (Figure channel9E), (Figure again indicating 9E), again thatindi- the compound needs to unbind before the channels can deactivate. The inhibition was cating that the compound needs to unbind before the channels can deactivate. The inhibi- slightly voltage dependent (Figure9C). In the presence of ZVS-08, the conductance–voltage tion was slightly voltage dependent (Figure 9C). In the presence of ZVS-08, the conduct- relationship was well described by a Boltzmann function with a blocking component ance–voltage relationship was well described by a Boltzmann function with a blocking (Figure9F). The activation was shifted by approximately −25 mV (semi-maximal activation component (Figure 9F). The activation was shifted by approximately −25 mV (semi-max- at −29.88 ± 1.8 mV in the presence and 4.39 ± 1.9 mV in the absence of ZVS-08), compatible imal activation at −29.88 ± 1.8 mV in the presence and 4.39 ± 1.9 mV in the absence of ZVS- with the action of a gating modiﬁer. 08), compatible with the action of a gating modifier. To clarify the mechanism of inhibition by ZVS-08, we compared the compound to two well characterised inhibitors: astemizole, which behaves as an open channel blocker binding to the promiscuous intracellular hydrophobic pocket in the channel family, and mibefradil, which acts from the extracellular side and modiﬁes channel gating. To test if ZVS-08 shares binding site with any of the two inhibitors, we performed competition experiments by determining if the presence of astemizole or mibefradil altered the effect of ZVS-08 on Kv10.1 current. The results are summarised in Figure 10. The degree of inhibition at the concentrations tested was independent of the presence of 500 nM astemizole or 1 µM mibefradil, indicating that the compounds do not compete for the same binding site.

3.4. Analogues of the Hit Compound ZVS-08 and Structure–Activity Relationships Analogues of the hit compound ZVS-08 were designed and synthesised to increase potency and selectivity and to investigate structure–activity relationships (SAR) important for KV10.1 inhibition. The hit compound ZVS-08 was modiﬁed at four different positions (Figure 11) to provide a focused library of compounds screened for KV10.1 (Table3) and selected compounds for hERG inhibition (Table2). Cancers 2021, 13, 1244 13 of 24 Cancers 2021, 13, x FOR PEER REVIEW 13 of 25

A B C 3×10-9

)

1.0 (

i l

p p -9

m 2×10

0.5 u

e t -9

v a 1×10

S 0.0 0 -8 -7 -6 -5 -4 0 1 2 3 Time (min) log (ZVS-08) (M)

D E F

1.0

d Control

i 0.5

o N ZVS-08

0.0 -100 -50 0 50 100 150 Vm (mV)

FigureFigure 9. 9. (A(A) )Inhibition Inhibition by by different concentrationsconcentrations of of ZVS-08 ZVS-08 hit hit compound. compound. (B )( Dose–responseB) Dose–response (data (data from from 3–10 independent3–10 inde- pendentdeterminations, determinations, mean ± meanSEM). ± (SEM).C,D) Current–voltage (C,D) Current– relationshipvoltage relationship in the absence in the (upperabsenc traces)e (upper or traces) presence or ofpresence 5 µM ZVS-08 of 5 µM ZVS-08 (lower traces) obtained by depolarisation from a holding potential of −80 mV to potentials ranging between (lower traces) obtained by depolarisation from a holding potential of −80 mV to potentials ranging between −60 and −60 and +100 mV. (E) The traces corresponding to depolarisation to +40 mV are superimposed for easier comparison. The +100 mV. (E) The traces corresponding to depolarisation to +40 mV are superimposed for easier comparison. The inset inset shows the normalised tail currents to highlight the slower deactivation in the presence of ZVS-08. (F) Conductance– voltageshows relationship the normalised in control tail currents (black to symbols) highlight and the in slower the presence deactivation (red symbols) in the presence of 5 µM of ZVS ZVS-08.-08, obtained (F) Conductance–voltage by biexponen- tialrelationship fit to the tail in current control (at (black −80 symbols)mV) and extrapolation and in the presence to time (red 0. The symbols) solid lines of 5 representµM ZVS-08, the obtainedresult of a by fit biexponential using a Boltz- ﬁt mannto the distribution tail current with (at − a80 block mV) at and positive extrapolation voltages.to time 0. The solid lines represent the result of a ﬁt using a Boltzmann distribution with a block at positive voltages. To clarify the mechanism of inhibition by ZVS-08, we compared the compound to two well characterised inhibitors: astemizole, which behaves as an open channel blocker Table 2. Potencies of compounds ZVS-08, 1 and 8 against K 10.1 and hERG. binding to the promiscuous intracellular hydrophobicV pocket in the channel family, and

Compound IDmibefradil, which acts from IC50 Kv10.1 the extracellular (µM) side and modifies ICchannel50 hERG gating. (µM) To test if ZVS-08ZVS-08 shares 3.70bindingµM (95% site CI:with 2.08 anyµM of to the 6.47 twoµM) inhibitors, 0.194 µ weM (95% performed CI: 0.047 competitionµM to 0.761 µex-M) 1periments by det 0.740erminingµM (95% if CI: the 490 presence nM to 1.17 of astemizoleµM) 0.207 orµ mibefradilM (95% CI: 0.113alteredµM the to 0.395effectµ M)of 8ZVS-08 on Kv10.1 1.01 current.µM (95% The CI: 796results nM toare 1.30 summarisedµM 0.156 in µFigureM (95% 10 CI:. The 0.070 degreeµM to of 0.373 inhibi-µM) tion at the concentrations tested was independent of the presence of 500 nM astemizole or 1 µM mibefradil, indicating that the compounds do not compete for the same binding site.

Cancers 2021, 13, x FOR PEER REVIEW 14 of 25

Cancers 2021, 13, 1244 14 of 24 Cancers 2021, 13, x FOR PEER REVIEW 14 of 25

Figure 10. Competition results of ZVS-08 (red circles) with astemizole (known pore blocker, black square) on KV10.1 (A,B) and mibefradil (gating modifier, black triangle) (C,D). (B,D) represent the same dataset as (A,C), but each series is plotted on a different axis to visualise that the dose–response is unchanged in the presence of the competitor. The fits resulted in the same IC50 for the presence and absence of the competitor (1.69 µM for A and B and 1.47 µM for (C,D)). Current amplitudes were measured in all cases at the end of a 500 ms depolarising pulse to +40 mV. FigureFigure 10. 10.Competition Competition results results ofof ZVS-08ZVS-08 (red circles) with with astemizole astemizole (known (known pore pore blocker, blocker, black black square) on KV10.1 (A,B) and mibefradil (gating modifier, black triangle) (C,D). (B,D) represent the square)3.4. Analogues on K 10.1 of (Athe,B Hit) and Compound mibefradil ZVS (gating-08 and modiﬁer, Structure black–Activity triangle) Relationships (C,D). (B,D) represent the same datasetV as (A,C), but each series is plotted on a different axis to visualise that the dose–re- A C samesponse datasetAnalogues is unchanged as ( , of), butthe in the eachhit presence com seriespound is of plotted the ZVS competitor. on-08 a differentwere The designed fits axis resulted to visualise and in thesynthesised that same the IC dose–response50 for to theincrease ispotency unchangedpresence and in absence the selectivity presence of the andco ofmpetitor the to competitor. investigate (1.69 µM The for structure ﬁtsA and resulted B and–activity in1.47 the µM same relationships for ( ICC,50D)for). Current the (SAR) presence am- im- andportantplitudes absence forwere of K the measuredV10.1 competitor inhibition. in all (1.69 cases µTheM at for thehit Aend compound and of Ba and500 1.47ms ZVS depolarisingµM-08 for was (C,D pulsemodified)). Current to +40 at amplitudesmV. four different were measuredpositions in (Figure all cases 11 at) theto provid end ofe a a 500 focused ms depolarising library of pulse compounds to +40 mV. screened for KV10.1 (Table 33) .4.and Analogues selected of compounds the Hit Compound for hERG ZVS -inhibition08 and Structure (Table– Activity2). Relationships Analogues of the hit compound ZVS-08 were designed and synthesised to increase potency and selectivity and to investigate structure–activity relationships (SAR) important for KV10.1 inhibition. The hit compound ZVS-08 was modified at four different positions (Figure 11) to provide a focused library of compounds screened for KV10.1 (Table 3) and selected compounds for hERG inhibition (Table 2).

Figure 11. Design of the hit compound ZVS-08 analogues. Figure 11. Design of the hit compound ZVS-08 analogues.

Figure 11. Design of the hit compound ZVS-08 analogues.

Cancers 2021, 13, x FOR PEER REVIEW 15 of 25 Cancer s 2021,, 13,, xx FORFOR PEERPEER REVIEWREVIEW 15 of 25 Cancers 2021, 13, x FOR PEER REVIEW 15 of 25 Cancers 2021, 13, x FOR PEER REVIEW 15 of 25 CancerCancer ss 20212021,, 1313,, xx FORFOR PEERPEER REVIEWREVIEW 1515 ofof 2525 Cancers 2021,, 13,, xx FORFOR PEERPEER REVIEWREVIEW 15 of 25 Cancer s 2021, 13, x FOR PEER REVIEW 15 of 25

Table 2. Potencies of compounds ZVS-08, 1 and 8 against KV10.1 and hERG. Table 2. Potencies of compounds ZVS-08, 1 and 8 against KV10.1 and hERG. V Table 2.. Potencies of compounds ZVS--08, 1 and 8 against KV10.1 and hERG. Table 2. Potencies of compounds ZVS-08, 1 and 8 against KV10.1 and hERG. Compound ID IC50 Kv10.1 (μM) V IC50 hERG (μM) TableTableCompound 22.. PotenciesPotencies ID ofof compoundscompounds ICZVSZVS50 --Kv10.108,08, 11 andand (μM) 88 aagainstgainst KKV10.110.1 andand hERG.hERG.IC50 hERG (μM) TableCompound 2.. Potencies ID of compounds ICZVS -Kv10.108, 1 and (μM) 8 against KV10.1 and hERG.IC hERG (μM) TableCompound 2. Potencies ID of compounds3.70 µM ICZVS (95%5050 -Kv10.108, CI: 1 and2.08 (μM) 8 µM against to 6.47 KV10.1 0 .194and hERG.µMIC 5050(95% hERG CI: (μM) 0.047 µM to CompoundZVS-08 ID 3.70 µMIC (95%50 Kv10.1 CI: 2.08 (μM) µM to 6.47 0.194 µMIC 50(95% hERG CI: (μM)0.047 µM to CompoundCompoundZVS--08 IDID 3.70 µMICIC (95%5050 Kv10.1Kv10.1 CI: 2.08 (μM)(μM) µM to 6.47 0.194 µMICIC 5050(95% hERGhERG CI: (μM) (μM)0.047 µM to CompoundZVS-08 ID 3.70 µMIC (95%50 Kv10.1 CI:µM) 2.08 (μM) µM to 6.47 0.194 µMIC 50(95%0.761 hERG CI:µM) (μM) 0.047 µM to CompoundZVS-08 ID 3.70 µMIC (95%50 Kv10.1 µM)CI: 2.08 (μM) µM to 6.47 0.194 µMIC 50(95%0.761 hERG CI:µM) (μM) 0.047 µM to ZVS-08 3.703.70 µMµM (95%(95% CI:CI:µM) 2.082.08 µMµM toto 6.476.47 00.194.194 µMµM (95%(95%0.761 CI:CI: µM) 0.0470.047 µMµM toto ZVS-08 0.7403.70 µM µM (95% (95% CI:µM) CI: 2.08 490 µMnM toto 6.471.17 0.194.194.207 µMµM (95%(95%0.761 CI:CI: µM) 0.0470.0470.113 µMµM toto ZVS1- 08 0.7403.70 µM µM (95% (95% µM)CI: CI: 2.08 490 µMnM toto 6.471.17 0.207.194 µM (95%0.761 CI:µM) 0.1130.047 µM to ZVS1 -08 0.740 µM (95%µM) CI: 490 nM to 1.17 0.207.207 µMµM (95%(95%0.7610.395 CI: CI:µM) 0.1130.113 µMµM toto Cancers 2021, 13, 1244 1 0.740 µM (95%µM) CI: 490 nM to 1.17 0.207 µM (95%0.3950.761 CI:µM) 0.113 15 µM of 24 to 1 0.7400.740 µMµM (95%(95%µM) CI:CI: 490490 nMnM toto 1.171.17 00.207.207 µMµM (95%(95%0.395 CI:CI: µM) 0.1130.113 µMµM toto 1 0.7401.01 µMµM (95%(95% µM) CI:CI: 7490 96 nMnM toto 1.301.17 0.207.207.156 µMµM (95%(95%0.395 CI:CI: µM) 0.1130.1130.070 µMµM toto 18 0.7401.01 µM µM (95% (95% µM)CI: CI: 7 49096 nM nM to to 1.30 1.17 0.156.207 µM (95%0.395 CI:µM) 0.0700.113 µM to 81 1.01 µM (95% µM)CI:µM 7 96 nM to 1.30 0.156.156 µMµM (95%(95%0.3950.373 CI: CI:µM) 0.0700.070 µMµM toto 8 1.01 µM (95% µM)µMCI: 7 96 nM to 1.30 0.156 µM (95%0.3730.395 CI:µM) 0.070 µM to 8 1.011.01 µMµM (95%(95% CI:CI:µM 77 9696 nMnM toto 1.301.30 00.156.156 µMµM (95%(95%0.373 CI:CI: µM) 0.0700.070 µMµM toto 8 1.01 µM (95% CI:µM 7 96 nM to 1.30 0.156.156 µMµM (95%(95%0.373 CI:CI: µM) 0.0700.070 µMµM toto Table 3. Structures8 and potencies1.01 µM of the (95% designed µMCI: 7 96 and nM synthesised to 1.30 0 ZVS.156 -08 µM analogues (95%0.373 CI:µM)1–12 0.070 . µM to Table 3. Structures8 and potencies of the designedµM and synthesised ZVS -080.373 analogues µM) 1–12. Table 3. Structures and potencies of the designedµM and synthesised ZVS -08 0.373analogues µM) 1 –12. Table 3.. Structures and potencies of the designed and synthesised ZVS --08 analogues 1–12.. TableCompoundCompound 3. Structures IDID and potencies of theStructure Structure designed and synthesised %% of of Kv10.1ZVS Kv10.1 -08 Inhibition analogues Inhibitionat 1– 10 12atµ. 10M μM TableTableCompound 33.. StructuresStructures ID andand potenciespotencies ofof thetheStructure designeddesigned andand synthesisedsynthesised% of ZVSZVS Kv10.1 --0808 analoguesanalogues InhibitionInhibition 11–– 12 12at.. 10 μM TableCompound 3.. Structures ID and potencies of theStructure designed and synthesised% of ZVS Kv10.1 -08 analogues Inhibition 1– 12at.. 10 μM TableCompound 3. Structures ID and potencies of theStructure designed and synthesised% of ZVS Kv10.1 -08 analogues Inhibition 1–12 at. 10 μM Compound ID Structure % of Kv10.1 Inhibition at 10 μM Compound11 ID Structure % of Kv10.187.3287.32 ±Inhibition 4.60± 4.60 (7) (7) at 10 μM Compound1 ID Structure % of Kv10.187.32 Inhibition ± 4.60 (7) at 10 μM 1 87.32 ± 4.60 (7) 1 87.32 ± 4.60 (7) 11 87.3287.32 ±± 4.604.60 (7)(7) 1 87.32 ± 4.60 (7) 12 23.9387.32± ±± 26.804.60 (7) (4) 2 23.9323.93 26.80± 26.80 (4) (4) 2 23.93 ± 26.80 (4) 2 23.93 ± 26.80 (4) 22 23.9323.93 ±± 26.8026.80 (4)(4) 2 23.93 ± 26.80 (4) 23 23.9322.08 ± 26.8033.52 (4) 3 22.0822.08± 33.52± 33.52 (4) (4) 3 22.08 ± 33.52 (4) 3 22.08 ± 33.52 (4) 33 22.0822.08 ±± 33.5233.52 (4)(4) 3 22.08 ± 33.52 (4) 34 22.0836.63 ± 33.5219.89 (4)(7) 4 36.6336.63± 19.89± 19.89 (7) (7) 4 36.63 ± 19.89 (7) 4 36.63 ± 19.89 (7) 44 36.6336.63 ±± 19.8919.89 (7)(7) 4 36.63 ± 19.89 (7) 45 36.632.53 ±± 11.7319.89 (3)(7) 5 2.53 ± 11.73 (3) 55 2.532.53± 11.73± 11.73 (3) (3) 5 2.53 ± 11.73 (3) 55 2.532.53 ±± 11.7311.73 (3)(3) 5 2.53 ± 11.73 (3) 56 57.482.53 ± ± 11.73 28.38 (3) (3) 6 57.48 ± 28.38 (3) 66 57.4857.48± 28.38± 28.38 (3) (3) 6 57.48 ± 28.38 (3) 66 57.4857.48 ±± 28.3828.38 (3)(3) 6 57.48 ± 28.38 (3) 67 57.4820.96 ± 28.3830.27 (3)(7) 7 20.96 ± 30.27 (7) 7 20.96 ± 30.27 (7) 7 20.96± ± 30.27 (7) 7 20.9620.96 30.27± 30.27 (7) (7) 7 20.96 ± 30.27 (7) 7 20.96 ± 30.27 (7) 8 81.20 ± 4.97 (3) 8 81.20 ± 4.97 (3) 8 81.20 ± 4.97 (3) 8 81.20 ± 4.97 (3) 8 81.2081.20± 4.97± 4.97 (3) (3) 8 81.20 ± 4.97 (3) 8 81.20 ± 4.97 (3) 9 2.00 ± 7.45 (4) 9 2.00 ± 7.45 (4) 9 2.00 ± 7.45 (4) 9 2.00 ± 7.45 (4) 99 2.002.002.00± 7.45±± 7.457.45 (4) (4)(4) 9 2.00 ± 7.45 (4) 109 57.052.00 ± 7.4521.71 (4) (2) 10 57.05 ± 21.71 (2) 10 57.05 ± 21.71 (2) 10 57.05 ± 21.71 (2) 1010 57.0557.05 ±± 21.7121.71 (2)(2) 10 57.0557.05± 21.71± 21.71 (2) (2) 1011 57.0537.67 ± ± 21.71 7.13 (2)(2) 11 37.67 ± 7.13 (2) 11 37.67 ± 7.13 (2) 11 37.67 ± 7.13 (2) 1111 37.6737.67 ±± 7.137.13 (2)(2) 11 37.6737.67± 7.13± 7.13 (2) (2) 1112 16.6037.67 ±± 38.367.13 (2) (2) 12 16.60 ± 38.36 (2) 12 16.60 ± 38.36 (2) 12 16.60 ± 38.36 (2) 12 16.60 ± 38.36 (2) Inhibition data12 are means ± standard error (or range) for the number of 16.60independent ± 38.36 experiments (2) Inhibition data12 are means ± standard error (or range) for the number of16.60 independent16.60± 38.36± 38.36 (2) experiments (2) Inhibitionindicated indata12 parentheses. are means ± standard error (or range) for the number of independent16.60 ± 38.36 experiments (2) indicatedInhibition in data parentheses. are means ± standard error (or range) for the number of independent experiments Inhibitionindicated indata parentheses. are means ± standard error (or range) for the number of independent experiments Inhibitionindicated indata parentheses. are means ± standard error (or range) for the number of independent experiments InhibitionindicatedInhibitionindicatedModifications data in indata are parentheses.parentheses. means are means ±instandard the ± standardaromatic error (or error portion range) (or forrange) of the ZVS number for- 08the were ofnumber independent performed of independent experiments on the experiments indicatedphenyl ring in indicatedModifications in parentheses. in the aromatic portion of ZVS--08 were performed on the phenyl ring parentheses.indicatedbearingModifications ortho in parentheses.-nitro inand the para aromatic-trifluoromethyl portion of substituents.ZVS-08 were One performed or both on groups the phenyl were ringsys- bearingModifications ortho--nitro andin the para aromatic--trifluoromethyltrifluoromethyl portion of substituents. substituents.ZVS-08 were OneOne performed oror bothboth on groupsgroups the phenyl werewere sys-ringsys- bearingModifications ortho-nitro inand the para aromatic-trifluoromethyl portion of substituents.ZVS-08 were Oneperformed or both on groups the phenyl were ringsys- tematicallybearingModifications ortho removed-nitro inand in the compoundspara aromatic-trifluoromethyl portion 3, 6 and of 7 substituents.ZVSto analyse-08 were the Oneperformed influence or both of on groupsboth the substituentsphenyl were ringsys- tematicallybearingModificationsModifications ortho removed-nitro in in theand in the aromatic compoundspara aromatic-trifluoromethyl portion portion 3, 6 and of ZVS-08of 7 substituents.toZVS analyse-08 were were the performed One performedinfluence or both on of on the groupsboth the phenyl substituentsphenyl were ring ringsys- bearingtematicallyonbearing the potency orthoortho removed--nitronitro of K andVand10.1 in compoundsparapara inhibition.--trifluoromethyltrifluoromethyl Th3, 6e andnitro 7 substituents. substituents.groupto analyse was thereplaced OneOne influence oror by bothboth a of methyl groups groupsboth substituents ester werewere group sys-sys- onbearingtematically the potency ortho removed-nitro of K VandV10.1 in compoundspara inhibition.-trifluoromethyl Th3, 6e andnitro 7 substituents.group to analyse was thereplaced One influence or by both a ofmethyl groups both substituents ester were group sys- bearingtematicallybearingon theortho potency ortho-nitro removed-nitro of and K Vand10.1 para-in compoundspara inhibition.trifluoromethyl-trifluoromethyl Th3, 6e andnitro substituents. 7 substituents.groupto analyse was One thereplaced One or influence both or by groupsboth a of methyl groupsboth were substituents ester system-were group sys- tematicallyon the potency removed of KV10.1 in compounds inhibition. Th3,, 6e andnitro 7 groupto analyse was thereplaced influence by a of methyl both substituents ester group tematicallyon the potency removed of KV10.1 in compounds inhibition.3 6 Th3, 76e andnitro 7 groupto analyse was thereplaced influence by a of methyl both substituents ester group aticallyonon thethe removed potencypotency in ofof compounds KKVV10.110.1 inhibition.inhibition., and ThThee tonitronitro analyse groupgroup the waswas influence replacedreplaced of byby both aa methylmethyl substituents esterester groupgroup on on the potency of KV10.1 inhibition. The nitro group was replaced by a methyl ester group theon potency the potency of KV 10.1of KV inhibition.10.1 inhibition. The nitro The nitro group group was replacedwas replaced by a methylby a methyl ester ester group group in compound 9, an amino group in compound 10, and a carboxylic acid group in compound

11, with the aim of removing the nitro group, which is sometimes associated with genotox-

icity [37]. The basic dimethylamino group was replaced by four different amines (R in 3 orange, Figure 11) to evaluate the importance of the cationic centre. Speciﬁcally, the basicity of the amine was reduced by introducing the aniline moiety (5) or removed by forming the acetamide (12). In addition, the aniline moiety in 5 was also introduced to assess whether the compound could be further extended in this part of the molecule and whether the Cancers 2021, 13, 1244 16 of 24

formation of additional cation–π or π–π interactions was possible. The morpholino moiety in 2 and 8 was incorporated to maintain the basic character of the parent compound ZVS-08, but to further increase the polarity of the compound. The aliphatic chain between the basic centre and the aromatic ring of ZVS-08 (R4 in green, Figure 11) was also modified. The hydroxy group was removed to generate compounds 1 and 4, which differ by the substitution of the aromatic ring. Based on the study of structure–activity relationships of ZVS-08 analogues, we identified structural parts important for KV10.1 inhibition of this new structural type of KV10.1 inhibitors (Table3). It is concluded that the two most potent K V10.1 inhibitors, 1 and 8, contain a combination of disubstituted phenyl ring with both ortho-nitro and para- trifluoromethyl groups and basic moieties such as dimethylamino group in compound Cancers 2021, 13, x FOR PEER REVIEW1 and morpholino group in compound 8. Both optimised compounds 1 and 8 exhibited17 of 25

improved potencies compared to the hit compound ZVS-08 with the IC50 values of 740 nM and 1.01 µM, respectively (Figure 12A, Table2).

Figure 12. Cont.

Figure 12. (A) Dose–response curves of compounds 1 and 8 for KV1.10 inhibition. The red trace has been included as reference and corresponds to ZVS-08 (from Figure 9B). The IC50 values from the fit were 740 nM (95% CI 490 nM to 1.17 µM, N = 7–9) for compound 1 and 1.01 µM (95% CI: 796 nM to 1.30 µM, N = 4–7) for compound 8. (B) Representative Kv10.1 current traces obtained in the

Cancers 2021, 13, x FOR PEER REVIEW 17 of 25

Cancers 2021, 13, 1244 17 of 24

Figure 12. (A) Dose–response curves of compounds 1 and 8 for KV1.10 inhibition. The red trace has been included as reference and corresponds to ZVS-08 (from Figure9B). The IC 50 values from the ﬁt Figure 12. (A) Dose–response curves of compounds 1 and 8 for KV1.10 inhibition. The red trace has were 740 nM (95% CI 490 nM to 1.17 µM, N = 7–9) for compound 1 and 1.01 µM (95% CI: 796 nM to been included as reference and corresponds to ZVS-08 (from Figure 9B). The IC50 values from the µ fit were1.30 740M, nM N (95% = 4–7) CI for 490 compound nM to 1.178 .(µM,B) RepresentativeN = 7–9) for compound Kv10.1 current 1 and traces1.01 µM obtained (95% CI: in 796 the presence nM toof 1.30 different µM, N concentrations= 4–7) for compound of Compound 8. (B) Representative1. Scale bars: Kv10.1 1 nA, current 50 ms. Thetraces cell obtained was depolarized in the to +40 mV for 500 ms. No leak subtraction was performed. (C) Representative traces obtained in the presence of different concentrations of Compound 8. Scale bars: 1 nA, 50 ms. (D) Dose–response curves of compounds ZVS-08, 1 and 8 for hERG inhibition. The IC50 values from the ﬁt were 0.194 µM (95% CI 0.047 µM to 0.761 µM, N = 2–7) for ZVS-08, 0.207 µM (95% CI 0.113 µM to 0.395 µM, N = 3–8) for compound 1 and 0.156 µM (95% CI: 0.070 µM to 0.373 µM, N = 3–8) for compound 8. (E) Representative traces of hERG currents obtained in the presence of different concentrations of Compound 1. Scale bars: 150 pA, 1 s. Only the end of the response at +20 mV and the tail current at −40 mV are represented. (F) Representative traces obtained in the presence of different concentrations of Compound 8. Scale bars: 150 pA, 1 s.

Disubstituted ortho-nitro and para-trifluoromethyl compounds with a weaker basic centre (5) or without it (12) were less potent. A similar effect of replacing the basic centre was observed with purpurealidin analogues [22]. The disubstituted compound in which the nitro group was replaced by the amino group in compound 10 was less potent. The unsubstituted compound 3 was almost inactive, indicating the importance of the disubstituted ortho-nitro and para-trifluoromethyl phenyl moiety in this structural type of KV10.1 inhibitors. Compounds 9 and 11, in which the nitro group was exchanged with the methyl ester and carboxylic acid groups, were also inactive. The analogue of ZVS-08 without the nitro group and retaining the para-trifluoromethyl group (6) was also less potent. All three ortho-nitro compounds (2, 4 and 7) without the para-trifluoromethyl group and with basic groups, showed almost the same inhibitory activity on KV10.1. Compounds 1 and 4 lack the hydroxyl group in the aliphatic part of the molecule, but this modification was not essential for the inhibitory potency against KV10.1. In general, the aliphatic amine plays an important role in binding. Compounds ZVS-08, 1 and 8 were tested for selectivity on hERG with the patch-clamp assay (Figure 12B, Table2). The selectivity of ZVS-08, 1 and 8 against the hERG channel remains an issue similar to the purpurealidin analogues [38]. However, with structural optimization, we increased the potency for KV10.1, while the potency for hERG remains the same for all three compounds ZVS-08, 1 and 8 (Table2). Cancers 2021, 13, x FOR PEER REVIEW 18 of 25

presence of different concentrations of Compound 1. Scale bars: 1 nA, 50 ms. The cell was depolarized to +40 mV for 500 ms. No leak subtraction was performed. (C) Representative traces obtained in the presence of different concentrations of Compound 8. Scale bars: 1 nA, 50 ms. (D) Dose–response curves of compounds ZVS-08, 1 and 8 for hERG inhibition. The IC50 values from the fit were 0.194 µM (95% CI 0.047 µM to 0.761 µM, N = 2–7) for ZVS-08, 0.207 µM (95% CI 0.113 µM to 0.395 µM, N = 3–8) for compound 1 and 0.156 µM (95% CI: 0.070 µM to 0.373 µM, N = 3–8) for compound 8. (E) Representative traces of hERG currents obtained in the presence of different con- Cancers 2021, 13, 1244 centrations of Compound 1. Scale bars: 150 pA, 1 s. Only the end of the response at +20 mV and18 of 24 the tail current at −40 mV are represented. (F) Representative traces obtained in the presence of different concentrations of Compound 8. Scale bars: 150 pA, 1 s. 3.5. Selectivity of Compounds ZVS-08 and 1 against Other Voltage-Gated Ion Channels 3.5. Selectivity of Compounds ZVS-08 and 1 against Other Voltage-Gated Ion Channels Selected potent KV10.1 inhibitors ZVS-08 and 1 were further evaluated for their se- Selected potent KV10.1 inhibitors ZVS-08 and 1 were further evaluated for their selectivity against other potential targets in a panel of voltage-gated ion channels using the lectivity against other potential targets in a panel of voltage-gated ion channels using the Xenopus laevis heterologous expression system, which allows efﬁcient horizontal pharma- Xenopus laevis heterologous expression system, which allows efficient horizontal pharmacological comparison. Several different voltage-gated ion channels relevant in cardiac cological comparison. Several different voltage-gated ion channels relevant in cardiac physiology, the immune system, and skeletal muscle were screened. Both KV10.1 inhibitors physiology, the immune system, and skeletal muscle were screened. Both KV10.1 inhibi- ZVS-08 and 1 showed no signiﬁcant inhibition and were thus considered selective for the tors ZVS-08 and 1 showed no significant inhibition and were thus considered selective for KV1.1, KV1.3, KV1.4, KV2.1, KV4.2, NaV1.4, and NaV1.5 ion channels at a concentration the KV1.1, KV1.3, KV1.4, KV2.1, KV4.2, NaV1.4, and NaV1.5 ion channels at a concentration of 10 µM (Figure 13, Table S1). However, the selectivity should be further investigated to of 10 µM (Figure 13, Table S1). However, the selectivity should be further investigated to check the activity of these compounds on a broader range of ion channels. check the activity of these compounds on a broader range of ion channels.

FigureFigure 13. 13. SelectivitySelectivity screening screening of ofZVS ZVS-08-08 and and 1 against 1 against voltage voltage-gated-gated potassium potassium and sodium and sodium channelschannels at at a a c concentrationoncentration of of 10 10 µM.µM. Data were were obtained with with the the two two-electrode-electrode voltage voltage clamp clamp on on Xenopus laevis oocytes. Data Data are presented as means ± SEMSEM of of nn ≥≥ 3 3independent independent experiments. experiments.

3.6.3.6. Effects Effects on on the the Growth Growth of of Cell Cell Lines Lines in 2D Cell Culture

KKV10.110.1 inhibitors with with aa favourable favourable pharmacological pharmacological profile profile are are promising promising candidates candi- datesfor cancer for cancer therapy. therapy. To investigate To investigate the activity the activity of the of new the new compounds compounds on cancer on cancer cells cells pro- proliferation,liferation, compounds compounds ZVS-08, ZVS-08,1 and1 and8 were8 were tested tested on on MCF-7, MCF-7, a a breastbreast cancercancer cell line showingshowing high high K V10.110.1 expression expression and and low low hERG hERG expression, expression, and and on Panc1, a pancreatic cancercancer cell cell line with undetectable K KV10.110.1 expression expression and and high high level level of of hERG hERG expression. expression. hTERThTERT-RPE1-RPE1 cells served as a non non-cancer-cancer cell line. hTERT hTERT-RPE1-RPE1 cells show transient physi-phys- iologicalological expressionexpression ofof Kv10.1Kv10.1 thatthat beginsbegins duringduring thethe G2 phase of the cell cycle and extends extends throughthrough mitosis mitosis as present in normal cells.cells. Different concentrations of of the compounds were added to the culture medium and the growth of the cell was monitored over 72 h ((FigureFigure 14)).. ZVS-08 and 1 inhibited the growth of MCF-7 cells to a very similar extent and also with similar potency, with IC50 in the range of 5 µM. Compound 8 was inactive on this cell line, and all three compounds were significantly less efficient on Panc1; only the highest concentrations used showed effects. hTERT-RPE1 cells, a non-transformed cell line, showed a similar sensitivity to ZVS-08 and compound 1 as Panc1. In summary, ZVS-08 and 1 were able to reduce the growth of cancer cells expressing KV10.1 but were much less active on cells expressing hERG or on normal cells, suggesting that their effect on KV10.1 serves more to inhibit growth in these cells. The lack of activity of compound 8 in this assay could be caused by a higher affinity for serum proteins, depleting the compound in the medium, as has been repeatedly observed with many compounds [39]. Cancers 2021, 13, x FOR PEER REVIEW 19 of 25

ZVS-08 and 1 inhibited the growth of MCF-7 cells to a very similar extent and also with similar potency, with IC50 in the range of 5 µM. Compound 8 was inactive on this cell line, and all three compounds were significantly less efficient on Panc1; only the highest concentrations used showed effects. hTERT-RPE1 cells, a non-transformed cell line, showed a similar sensitivity to ZVS-08 and compound 1 as Panc1. In summary, ZVS-08 and 1 were able to reduce the growth of cancer cells expressing KV10.1 but were much less active on cells expressing hERG or on normal cells, suggesting that their effect on KV10.1 Cancers 2021, 13, 1244 serves more to inhibit growth in these cells. The lack of activity of compound 8 19in ofthis 24 assay could be caused by a higher affinity for serum proteins, depleting the compound in the medium, as has been repeatedly observed with many compounds [39].

FigureFigure 14.14. Effects of ZVS-08ZVS-08 and compounds 11 andand 88 onon the the growth growth of of MCF MCF-7-7 (A (A) )and and Panc1 Panc1 cells cells

((BB).). ZVS-08ZVS-08 and compound 11 inhibitedinhibited the the growth growth of of MCF MCF-7-7 cells cells with with IC IC50 50of of6.8 6.8(95% (95% CI: CI:5.14 5.14–– 9.389.38 µµM)M) and 5.48 µMµM (95% CI: 4.5 4.51–6.761–6.76 µM).µM). Compound Compound 88 waswas ineffective ineffective at at the the concentrations concentrations tested.tested. None of of the the compounds compounds inhibited inhibited the the growth growth of of Panc1 Panc1 cells cells significantly significantly at atconcentration concentration below 10 µM, and an IC50 could only be determined for compound 1 (over 30 µM). below 10 µM, and an IC50 could only be determined for compound 1 (over 30 µM). 3.7.3.7. 3D3D CellCell BasedBased AssaysAssays ToTo testtest the the ability ability of of the the compounds compounds to to inhibit inhibit tumour tumour progression progression in ain more a more predictive predic- system,tive system, we performed we performed experiments experiments on tumour on tumour spheroids spheroids of the of pancreatic the pancreatic cancer cancer cell line cell Colo357,line Colo357, which which grows grows efficiently efficiently in this in culturethis culture modality. modality. The The cells cells were were seeded seeded in the in presencethe presence of Matrigel of Matrigel (see methods), (see methods), allowed allowed to form to spheroids, form spheroids, and the and compounds the compounds (50 µM) were(50 µM) added were to theadded growth to the medium. growth Themedium. control The contained control thecontained same amount the same of theamount vehicle of (0.5%the vehicle DMSO). (0.5% Apoptosis DMSO). was Apoptosis monitored was by monitored addition ofby aaddition fluorescent of a caspase fluorescent 2/7 activitycaspase reporter2/7 activity at the report sameer time at the as the same compounds. time as the Cleavage compounds. of the reporterCleavage by of caspase the reporter results byin greencaspase florescence, results in andgreen the florescence, intensity of and the fluorescencethe intensity wasof the recorded fluorescence over 72 was h. recorded overAs 72 h depicted. in Figure 15, ZVS-08 induced apoptosis efficiently starting very early, whileAs compound depicted1 inrequired Figure longer15, ZVS time,-08 induced but reached apoptosis similar efficiently efficacy. Compound starting very8 showed early, apoptosiswhile compound statistically 1 required significant longer over the time, control but starting reached only similar after efficacy. 24 h and Compounddid not reach 8 theshowed same apoptosis intensity statistically as the other significant compounds over in the the control time ofstarting the experiment. only after 24 This h and lower did efficacynot reach matches the same the intensity small effects as the detected other compounds in conventional in the 2D time culture of the with experiment. this compound. This lower efficacy matches the small effects detected in conventional 2D culture with this com- 3.8. Mutagenic Activity of ZVS-08 and 8 pound. The association of nitro groups with mutagenicity is well known in medicinal chemistry. For this reason, this is not the usual functional group among approved drugs, but recently, there have been advances in drug design incorporating the nitro group in antipara- sitic, antitubercular, and anticancer compounds. There is a need to understand bioactivation potential of compounds bearing aromatic nitro groups [37]. The potential mutagenic activity of compounds ZVS-08 and 8 was determined using the bacterial Salmonella/microsomal reverse mutation assay with TA98 and TA100 Salmonella typhimurium strains in the absence and presence of metabolic activation [40]. Compounds ZVS-08 and 8 were cytotoxic to both − ≥ bacterial strains ( /+ S9 metabolic activation) at concentrations 0.5 mg/plate. At non- cytotoxic concentrations (0.0158–0.158 mg/plate), no mutagenic response was observed in the absence (−S9) or presence (+S9) of metabolic activation (Tables S2 and S3). Cancers 2021, 13, 1244 20 of 24 Cancers 2021, 13, x FOR PEER REVIEW 20 of 25

i 7

s 8×10

) 7

e 6×10

m t 4×10

C ZVS-08

G Compound 8

( n 7

e 2×10 Compound 1

r Control

o 0 T 0 24 48 time (h)

B Control ZVS-08 Compound 1 Compound 8

FigureFigure 15. 15.(A ()A Apoptosis) Apoptosis induction induction by by ZVS-08 ZVS-08 and and compounds compounds1 and 1 and8 (50 8 (50µM) µM) on tumouron tumour spheroids sphe- formedroids formed by Colo by 357 Colo cells. 357 ZVS-08 cells. wasZVS markedly-08 was markedly faster and faster compound and compou8 wasnd slowest. 8 was Dataslowest. represent Data represent mean ± standard error (N = 3). (B) Representative images of spheroids 24 h after the ad- mean ± standard error (N = 3). (B) Representative images of spheroids 24 h after the addition of the dition of the indicated compounds (50 µM). Fluorescence intensity in the upper image corre- indicated compounds (50 µM). Fluorescence intensity in the upper image corresponds to cleavage of sponds to cleavage of the caspase 3/7 activity reporter; it has been merged with the phase contrast theimage caspase in the 3/7 lower activity raw. reporter; Scale bar: it has 400 been µm. merged with the phase contrast image in the lower raw. Scale bar: 400 µm.

4.3.8. Discussion Mutagenic Activity of ZVS-08 and 8 The association of nitro groups with mutagenicity is well known in medicinal chem- Inhibition of KV10.1 potassium channels with small molecules or antibodies reduces canceristry. cellFor growththis reason,in vitro thisand is notin vivo the .usual In a retrospectivefunctional group study among in brain approved metastases, drugs, treat- but recently, there have been advances in drug design incorporating the nitro group in an- ment with agents that can inhibit KV10.1 improved the outcome of patients who had moderatetiparasitic, expression antitubercular, of the channeland anticancer [27]. This, compounds. together There with the is a very need high to understand frequency ofbio- activation potential of compounds bearing aromatic nitro groups [37]. The potential mu- abnormal expression of the channel in human tumours, implies that KV10.1 inhibitors with appropriatetagenic activity potency of compounds and selectivity ZVS may-08 beand useful 8 was in determined the treatment using of many the bacterial tumour types. Salmo- nella/microsomal reverse mutation assay with TA98 and TA100 Salmonella typhimurium With this goal in mind, we set out to develop a new generation of KV10.1 inhibitors.

Cancers 2021, 13, 1244 21 of 24

The discovery of KV10.1-selective potassium channel inhibitors is challenging due to the high similarity with the hERG, especially the ion-conducting pore. Most of the known KV10.1 inhibitors (Figure1) also show hERG inhibition, which may lead to undesirable QT interval prolongations in patients [28]. Moreover, the exact location of the binding site for most of these inhibitors has not yet been experimentally determined. Structural information about the KV10.1 inhibitor complex is also not available, which limits the structure-based design of new selective KV10.1 inhibitors. Considering these limitations, we selected purpurealidine-based KV10.1 inhibitors that have been shown to bind to the VSD of the channel [8]. Given the greater structural dissimilarity between KV10.1 and hERG in the VSD than in the pore [20], our goal was to discover a new structural class of selective KV10.1 inhibitors by computer-assisted ligand-based virtual screening. The latter can overcome the limitation that the exact location of the binding site is unknown, as it only uses information about the ligand. However, in such a case, active and inactive ligands from the same structural class should be used in the modelling because they are expected to bind at the same binding site and in the same orientation. The novel structural class of diarylamine-based KV10.1 inhibitors presented here was identified by ligand-based virtual screening using a pharmacophore model. Phar- macophore modelling is a powerful method that aligns active ligands based on their pharmacophore features and then creates a ligand-based pharmacophore model consisting of pharmacophore features common to all active ligands. Our initial model performed well in the validation experiment as it was able to find a majority of the active ligands and no inactive ligands from the compound library. However, it was too complex and did not return any hits when used in the virtual screening of a library of 556,000 compounds. After a series of simplifications, we arrived at a model with only six features (compared to 14 features in the original model) that performed excellently, finding all active compounds and no inactive ones (Figure7D). By using this model in the virtual screen, we found 18 potential hit compounds, nine of which were purchased and tested for KV10.1 inhibition. Compound ZVS-08 proved to be a hit with an IC50 value of 3.70 µM. The hit compound ZVS-08 represents a new structural class of KV10.1 inhibitors with diarylamine structure. With the structure–activity relationship studies we identified structural parts important for KV10.1 inhibition of this new structural type of KV10.1 inhibitors. The most potent inhibitors from this series contain a combination of disubstituted phenyl ring with both ortho-nitro and para-trifluoromethyl groups and basic moieties such as dimethylamino or morpholino group. Among all virtual hits, the compound ZVS-08 was the only one potent against the KV10.1. Amine groups and aromatic rings are also present in all ZVS virtual hits, but it might be possible that the substitution of the aromatic ring plays an important role in compound binding to the active site. Hit compound ZVS-08 has the ring substituted with two electron withdrawing groups that are not present in any other virtual hit. Based on results, these two groups are important for KV10.1 inhibition, as evident also in the optimisation of ZVS-08. Structural optimization resulted in a nanomolar KV10.1 inhibitor 1 (IC50 value of 740 nM). With structural optimization, we improved only the potency for KV10.1 but not for hERG, which remains the same as for the hit compound ZVS-08. New solutions are needed for father optimization of the selectivity of KV10.1 inhibitors against hERG channel. The kinetics of inhibiting by ZVS-08 and its analogues resemble those of known com-pounds. The slow reduction in current amplitude along the stimulus is similar to inhibition by astemizole and is interpreted as inhibition of open channels. Nevertheless, this profile does not rule out an allosteric effect on gating, as mibefradil induces a kinetically similar effect. ZVS-08 induces a shift in the voltage dependence of KV10.1 currents in the hyperpolarizing direction reminiscent of that observed with mibefradil and with purpurealidine analogues, supporting the notion of an effect as a gating modifier rather than a blocker of open channels. The exact binding site for the new compounds is still unknown. It is assumed that the parent structures bind to the VSD at a different site than mibefradil [8]. This also seems to be Cancers 2021, 13, 1244 22 of 24

the case for the hit compound, as we observed no evidence of competition between ZVS-08 and mibefradil. We also tested competition with astemizol, due to the promiscuity of its binding site, but again we did not observe any competition. Therefore, we hypothesize that the new compounds bind to the VSD, probably from the inside. Mutagenesis experiments should help to identify the binding site [6]. The similarity between the behaviour of the ZVS-08 hit compound and astemizole would suggest that our hit shares the mechanism of inhibition and thus possibly the binding site in the channel pore and does not bind to the VSD like the purpurealidins from which our pharmacophore model was derived. We performed competition studies to test for a shared binding site. We determined the effect of different concentrations of ZVS-08 in the presence or absence of 500 nM astemizole (Figure 10). We observed no difference in affinity between the two conditions and conclude that ZVS-08 and astemizole do not share the same binding site. The main interest of selective KV10.1 inhibitors would be their use as anticancer agents. ZVS-08 and compound 1 inhibited the growth of non-tumour and tumour cells without KV10.1 expression in conventional 2D culture at concentrations higher than 10 µM. At lower concentrations, however, ZVS-08 and compound 1 derivatives were able to inhibit efficiently the growth of MCF-7 breast cancer cells, while leaving non-transformed cells unaffected. MCF-7 cells show significant expression levels of KV10.1, but undetectable expression of hERG, indicating that the effect might be related to inhibition of KV10.1. Importantly, the compounds did not affect the growth of Panc1 cells, which abundantly express hERG, but not KV10.1. There was therefore a correlation between KV10.1 expression and sensitivity to the compounds. In tumour spheroids, all compounds were able to induce significant levels of apoptosis. In the case of the hit compound ZVS-08, the effect was remarkably fast and intense. Compound 8, which was ineffective against tumour cells in monolayer cultures showed a slower onset in the induction of apoptosis but was able to induce cell death in longer times, which might indicate a more difficult access to the cells in the spheroid for this compound. These results do not exclude alternative mechanisms of action for the compounds, and further work will be required to determine to what extent Kv10.1 block is responsible for the effects observed.

5. Conclusions

To discover a novel structural class of KV10.1 inhibitors overexpressed in many different tumour types, we used a computational ligand-based approach. We identiﬁed the novel low micromolar KV10.1 inhibitor ZVS-08, which is based on a diarylamine scaffold. The inhibition proﬁle of ZVS-08 with the combination of data from the competition experiments suggests that the binding site is not shared with two classical KV10.1 inhibitors. By synthesising the analogues, we were able to increase the potency of KV10.1 inhibition. Analogue compound 1 inhibited the KV10.1 channel with an IC50 value of 740 nM and showed a slight increase in selectivity on the hERG channel. Cancer cell proliferation was inhibited at a concentration of 5.48 µM in KV10.1 expressing cell line, while the cancerous and non-cancerous cell lines without KV10.1 expression were not inhibited. With the broad window between mutagenic activity and inhibition of KV10.1, the new compounds show promising potential for further development for cancer therapy.

Supplementary Materials: The following are available online at https://www.mdpi.com/2072-6 694/13/6/1244/s1. Table S1: Selectivity screening of ZVS-08 and 1 against KV and NaV channels. Table S2: Mutagenic activity of compound ZVS-08. Table S3: Mutagenic activity of compound 8, chemistry: analytical procedures description and NMR spectrums. Author Contributions: Conceptualization, Ž.T., B.Ž., J.T., T.T., L.A.P., and L.P.M.; methodology, J.T., T.T., L.A.P., and L.P.M.; software, Ž.T. and T.T.; formal analysis, Ž.T., L.A.H., Š.G., Š.M., A.Š., M.N., S.P., and L.A.P.; investigation, Ž.T., L.A.H., A.Š., M.N., X.S., T.T., and L.A.P.; data curation, Ž.T., L.A.H., A.Š., M.N., X.S., S.P., T.T., and L.A.P.; writing—original draft preparation, Ž.T., B.Ž., J.T., T.T., L.A.P., and L.P.M.; writing—review and editing, J.T., T.T., L.A.P., and L.P.M.; supervision, B.Ž., J.T., T.T., L.A.P., and L.P.M.; project administration, J.T., T.T., L.A.P., and L.P.M.; funding acquisition, B.Ž., J.T., L.A.P., and L.P.M. All authors have read and agreed to the published version of the manuscript. Cancers 2021, 13, 1244 23 of 24

Funding: This research was funded by Slovenian Research Agency (ARRS) grant numbers J1-9192, N1-0098, P1-0245, and CELSA project. The work has also received funding from the Max-Planck Society and from the European Union through Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 813834-PHIONIC-H2020-MSCA-ITN-2018. J.T. was supported by grants G0E7120N, GOC2319N and GOA4919N (FWO-Vlaanderen), and grant CELSA/17/047 (KU Leuven). S.P. was supported by grant PDM/19/164 (KU Leuven). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are available on request from the corresponding authors. Acknowledgments: We thank OpenEye Scientiﬁc Software, Santa Fe, NM, USA, for free academic licenses for the use of their software. L.A.P. thanks the expert technical assistance of V. Díaz. Conﬂicts of Interest: The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

References 1. Lin, T.-F.; Jow, G.-M.; Fang, H.-Y.; Fu, S.-J.; Wu, H.-H.; Chiu, M.-M.; Jeng, C.-J. The Eag Domain Regulates the Voltage-Dependent Inactivation of Rat Eag1 K+ Channels. PLoS ONE 2014, 9, e110423. [CrossRef][PubMed] 2. Whicher, J.R.; MacKinnon, R. Structure of the Voltage-Gated K+ Channel Eag1 Reveals an Alternative Voltage Sensing Mechanism. Science 2016, 353, 664–669. [CrossRef][PubMed] 3. Ma, L.; Chin, Y.K.Y.; Dekan, Z.; Herzig, V.; Yuen Chow, C.; Heighway, J.; Wing Lam, S.; Guillemin, G.J.; Alewood, P.F.; King, G.F. Novel Venom-Derived Inhibitors of the Human EAG Channel, a Putative Antiepileptic Drug Target. Biochem. Pharmacol. 2018. [CrossRef] 4. Pardo, L.A.; Gomez-Varela, D.; Major, F.; Sansuk, K.; Leurs, R.; Downie, B.R.; Tietze, L.F.; Stuhmer, W. Approaches Targeting KV10.1 Open a Novel Window for Cancer Diagnosis and Therapy. Curr. Med. Chem. 2012, 19, 675–682. [CrossRef] 5. Pardo, L.A.; Stühmer, W. The Roles of K+ Channels in Cancer. Nat. Rev. Cancer 2014, 14, 39–48. [CrossRef][PubMed] 6. Movsisyan, N.; Pardo, L.A. Kv10.1 Regulates Microtubule Dynamics during Mitosis. Cancers 2020, 12, 2409. [CrossRef] 7. García-Quiroz, J.; García-Becerra, R.; Santos-Martínez, N.; Barrera, D.; Ordaz-Rosado, D.; Avila, E.; Halhali, A.; Villanueva, O.; Ibarra-Sánchez, M.J.; Esparza-López, J.; et al. In Vivo Dual Targeting of the Oncogenic Ether-à-Go-Go-1 Potassium Channel by Calcitriol and Astemizole Results in Enhanced Antineoplastic Effects in Breast Tumors. BMC Cancer 2014, 14, 745. [CrossRef] 8. Bai, Y.; Liao, H.; Liu, T.; Zeng, X.; Xiao, F.; Luo, L.; Guo, H.; Guo, L. MiR-296-3p Regulates Cell Growth and Multi-Drug Resistance of Human Glioblastoma by Targeting Ether-à-Go-Go (EAG1). Eur. J. Cancer 2013, 49, 710–724. [CrossRef] 9. Wu, J.; Wu, X.; Zhong, D.; Zhai, W.; Ding, Z.; Zhou, Y. Short Hairpin Rna (ShRNA) Ether à Go-Go 1 (Eag1) Inhibition of Human Osteosarcoma Angiogenesis via VEGF/PI3K/AKT Signaling. Int. J. Mol. Sci. 2012, 13, 12573–12583. [CrossRef] 10. Wang, X.; Chen, Y.; Li, J.; Guo, S.; Lin, X.; Zhang, H.; Zhan, Y.; An, H. Tetrandrine, a Novel Inhibitor of Ether-à-Go-Go-1 (Eag1), Targeted to Cervical Cancer Development. J. Cell. Physiol. 2018.[CrossRef][PubMed] 11. Gómez-Varela, D.; Zwick-Wallasch, E.; Knötgen, H.; Sánchez, A.; Hettmann, T.; Ossipov, D.; Weseloh, R.; Contreras-Jurado, C.; Rothe, M.; Stühmer, W.; et al. Monoclonal Antibody Blockade of the Human Eag1 Potassium Channel Function Exerts Antitumor Activity. Cancer Res. 2007, 67, 7343–7349. [CrossRef][PubMed] 12. de Queiroz, F.M.; Suarez-Kurtz, G.; Stühmer, W.; Pardo, L.A. Ether à Go-Go Potassium Channel Expression in Soft Tissue Sarcoma Patients. Mol. Cancer 2006, 5.[CrossRef] 13. Asher, V.; Khan, R.; Warren, A.; Shaw, R.; Schalkwyk, G.V.; Bali, A.; Sowter, H.M. The Eag Potassium Channel as a New Prognostic Marker in Ovarian Cancer. Diagn. Pathol. 2010, 5, 78. [CrossRef] 14. Martínez, R.; Stühmer, W.; Martin, S.; Schell, J.; Reichmann, A.; Rohde, V.; Pardo, L. Analysis of the Expression of Kv10.1 Potassium Channel in Patients with Brain Metastases and Glioblastoma Multiforme: Impact on Survival. BMC Cancer 2015, 15, 839. [CrossRef] 15. Agarwal, J.R.; Griesinger, F.; Stühmer, W.; Pardo, L.A. The Potassium Channel Ether à Go-Go Is a Novel Prognostic Factor with Functional Relevance in Acute Myeloid Leukemia. Mol. Cancer 2010, 9.[CrossRef] 16. Ding, X.-W.; Luo, H.-S.; Jin, X.; Yan, J.-J.; Ai, Y.-W. Aberrant Expression of Eag1 Potassium Channels in Gastric Cancer Patients and Cell Lines. Med. Oncol. 2007, 24, 345–350. [CrossRef] 17. Menéndez, S.T.; Villaronga, M.Á.; Rodrigo, J.P.; Álvarez-Teijeiro, S.; García-Carracedo, D.; Urdinguio, R.G.; Fraga, M.F.; Pardo, L.A.; Viloria, C.G.; Suárez, C.; et al. Frequent Aberrant Expression of the Human Ether à Go-Go (HEAG1) Potassium Channel in Head and Neck Cancer: Pathobiological Mechanisms and Clinical Implications. J. Mol. Med. 2012, 90, 1173–1184. [CrossRef] 18. Ousingsawat, J.; Spitzner, M.; Puntheeranurak, S.; Terracciano, L.; Tornillo, L.; Bubendorf, L.; Kunzelmann, K.; Schreiber, R. Expression of Voltage-Gated Potassium Channels in Human and Mouse Colonic Carcinoma. Clin. Cancer Res. 2007, 13, 824–831. [CrossRef] Cancers 2021, 13, 1244 24 of 24

19.S , terbuleac, D.; Maniu, C.L. An Antiarrhythmic Agent as a Promising Lead Compound for Targeting the HEAG1 Ion Channel in Cancer Therapy: Insights from Molecular Dynamics Simulations. Chem. Biol. Drug Des. 2016, 88, 683–689. [CrossRef] 20. Wang, W.; MacKinnon, R. Cryo-EM Structure of the Open Human Ether-à-Go-Go -Related K + Channel HERG. Cell 2017, 169, 422–430.e10. [CrossRef][PubMed] 21. The PyMOL Molecular Graphics System, version 2.3.0; Schrodinger, LLC: New York, NY, USA, 2019. 22. Moreels, L.; Bhat, C.; Voráˇcová, M.; Peigneur, S.; Goovaerts, H.; Mäki-Lohiluoma, E.; Zahed, F.; Pardo, L.A.; Yli-Kauhaluoma, J.; Kiuru, P.; et al. Synthesis of Novel Purpurealidin Analogs and Evaluation of Their Effect on the Cancer-Relevant Potassium Channel KV10.1. PLoS ONE 2017, 12, e0188811. [CrossRef][PubMed] 23. Gómez-Lagunas, F.; Carrillo, E.; Pardo, L.A.; Stühmer, W. Gating Modulation of the Tumor-Related Kv10.1 Channel by Mibefradil. J. Cell. Physiol. 2017, 232, 2019–2032. [CrossRef][PubMed] 24. Mysinger, M.M.; Carchia, M.; Irwin; John, J.; Shoichet, B.K. Directory of Useful Decoys, Enhanced (DUD-E): Better Ligands and Decoys for Better Benchmarking. J. Med. Chem. 2012, 55, 6582–6594. [CrossRef] 25. Inte:Ligand Expert KNIME Extensions Available from Inte:Ligand GmbH. Available online: http://www.Inteligand.com/knime- nodes (accessed on 10 January 2021). 26. Friedrich, N.-O.; de Bruyn Kops, C.; Flachsenberg, F.; Sommer, K.; Rarey, M.; Kirchmair, J. Benchmarking Commercial Conformer Ensemble Generators. J. Chem. Inf. Model. 2017, 57, 2719–2728. [CrossRef] 27. Poli, G.; Seidel, T.; Langer, T. Conformational Sampling of Small Molecules With ICon: Performance Assessment in Comparison With OMEGA. Front. Chem. 2018, 6, 229. [CrossRef] 28. Boydston, A.J.; Rice, J.D.; Sanderson, M.D.; Dykhno, O.L.; Bielawski, C.W. Synthesis and Study of Bidentate Benzimidazolylidene− Group 10 Metal Complexes and Related Main-Chain Organometallic Polymers. Organometallics 2006, 25, 6087–6098. [CrossRef] 29. Surry, D.S.; Buchwald, S.L. Dialkylbiaryl Phosphines in Pd-Catalyzed Amination: A User’s Guide. Chem. Sci. 2011, 2, 27–50. [CrossRef] 30. Marx, M.A.; Vaisburg, C.A.; Christensen, J.G.; Galemmo, J. LSD1 Inhibitors. U.S. Patent US 2017/0183308A1, 29 June 2017. 31. Sarkar, S.; Sonkar, R.; Bhatia, G.; Tadigoppula, N. Synthesis of New N-Acryl-1-Amino-2-Phenylethanol and N-Acyl-1-Amino-3- Aryloxypropanols and Evaluation of Their Antihyperlipidemic, LDL-Oxidation and Antioxidant Activity. Eur. J. Med. Chem. 2014, 80, 135–144. [CrossRef][PubMed] 32. Gómez-Varela, D.; Contreras-Jurado, C.; Furini, S.; García-Ferreiro, R.; Stühmer, W.; Pardo, L.A. Different Relevance of Inac- tivation and F468 Residue in the Mechanisms of HEag1 Channel Blockage by Astemizole, Imipramine and Dofetilide. FEBS Lett. 2006, 580, 5059–5066. [CrossRef][PubMed] 33. García-Ferreiro, R.E.; Kerschensteiner, D.; Major, F.; Monje, F.; Stühmer, W.; Pardo, L.A. Mechanism of Block of HEag1 K + Channels by Imipramine and Astemizole. J. Gen. Physiol. 2004, 124, 301–317. [CrossRef][PubMed] 34. Barros, F.; de la Peña, P.; Domínguez, P.; Sierra, L.M.; Pardo, L.A. The EAG Voltage-Dependent K+ Channel Subfamily: Similarities and Differences in Structural Organization and Gating. Front. Pharmacol. 2020, 11, 411. [CrossRef][PubMed] 35. Wolber, G.; Langer, T. LigandScout: 3-D Pharmacophores Derived from Protein-Bound Ligands and Their Use as Virtual Screening Filters. J. Chem. Inf. Model. 2005, 45, 160–169. [CrossRef][PubMed] 36. LigandScout v.4.3 Available from Inte:Ligand. Available online: https://www.Inteligand.com/ligandscout (accessed on 10 January 2021). 37. Nepali, K.; Lee, H.-Y.; Liou, J.-P. Nitro Group Containing Drugs. J. Med. Chem. 2019, 62, 2851–2893. [CrossRef][PubMed] 38. Moreels, L. The Search for Novel Inhibitors of the Cancer-Relevant Potassium Channel Kv10.1. Ph.D. Thesis, KU Leuven, Leuven, Belgium, 2018. 39. Fasano, M.; Curry, S.; Terreno, E.; Galliano, M.; Fanali, G.; Narciso, P.; Notari, S.; Ascenzi, P. The Extraordinary Ligand Binding Properties of Human Serum Albumin. IUBMB Life 2005, 57, 787–796. [CrossRef][PubMed] 40. Maron, D.M.; Ames, B.N. Revised Methods for the Salmonella Mutagenicity Test. Mutat. Res. Mutagen. Relat. Subj. 1983, 113, 173–215. [CrossRef]