Removal of Human Ether-Ã -go-go Related Gene (hERG) K+ Channel

Article pubs.acs.org/jmc

Removal of Human Ether-à-go-go Related Gene (hERG) K+ Channel Affinity through Rigidity: A Case of Clofilium Analogues Julien Louvel,* Joaõ F. S. Carvalho, Zhiyi Yu, Marjolein Soethoudt, Eelke B. Lenselink, Elisabeth Klaasse, Johannes Brussee, and Adriaan P. IJzerman Division of Medicinal Chemistry, Leiden Academic Centre for Drug Research, Leiden University, P.O. Box 9502, 2300 RA Leiden, The Netherlands S Supporting Information *

ABSTRACT: Cardiotoxicity is a side effect that plagues modern drug design and is very often due to the off-target blockade of the human ether-à-go-go related gene (hERG) potassium channel. To better understand the structural determinants of this blockade, we designed and synthesized a series of 40 derivatives of clofilium, a class III antiarrhythmic agent. These were evaluated in radioligand binding and patch-clamp assays to establish structure−affinity relationships (SAR) for this potassium channel. Efforts were especially focused on studying the influence of the structural rigidity and the nature of the linkers composing the clofilium scaffold. It was shown that introducing triple bonds and oxygen atoms in the n-butyl linker of the molecule greatly reduced affinity without significantly modifying the pKa of the essential basic nitrogen. These findings could prove useful in the first stages of drug discovery as a systematic way of reducing the risk of hERG K+ channel blockade-induced cardiotoxicity.

■

INTRODUCTION

zwitterions (introduction of carboxylic acids), or discrete structural modifications.9 In this study, we aimed at gaining knowledge on compounds specifically targeting the hERG channel, namely class III antiarrhythmic agents. We based our approach on the implicit assumption that better knowledge of the SAR data for such compounds can ease the discovery of chemical alerts for hERG potassium channel-related toxicity. Clofilium 1,10 a potent class III antiarrhythmic agent, was selected for this study, and 40 derivatives thereof were designed and synthesized and evaluated in a [3H]astemizole radioligand binding assay, using HEK293 membranes expressing the hERG potassium channel. A selection of these compounds was then also tested in a manual patch-clamp assay. In this study, we focused especially on the influence of the structural rigidity and the nature of the linkers composing the clofilium scaffold. Up to now, reports of rigidity as a way to alleviate hERG affinity have been scarce and limited to the use of certain constrained or bulky cycles.11−16 We envisioned bringing rigidity to the scaffold with multiple bonds as well. The results we obtained during our investigations enabled us to suggest features that can be implemented into new chemical entities in order to decrease or avoid cardiotoxicity as a side effect.

Class III antiarrhythmic agents owe their effect to their interaction and blockade of the rapidly acting delayed rectifier potassium current channel (IKr), which is encoded by the human ether-à-go-go related gene (hERG). By delaying the outward potassium flow during phase three of the cardiac membrane repolarization, these drugs prolong the QT interval in the electrocardiogram. While this effect is useful in patients suffering from short QT syndrome, hERG K+ channel blockade can also result from off-target interaction with other nonantiarrhythmic drugs. It then leads to the same prolongation of the QT interval, thereby increasing the risk of ventricular arrhythmia and fibrillation that may lead to torsades de pointes and sudden death.1−3 This issue has become a major safety concern for the pharmaceutical industry,4 leading to the black labeling of numerous drugs or their withdrawal from the market (such as astemizole, cisapride, sertindole, and terfenadine)5,6 and the requirement for every drug candidate to show a negligible interaction with the hERG K+ channel. Screening of lead compounds in the industry is usually carried out through two assays, namely (automated) patch-clamp assays and radioligand binding techniques.7,8 Even though there is no perfect predictive model for hERG affinity, and especially how to avoid it, some strategies are usually followed. They mainly consist of reducing lipophilicity, lowering the pKa of basic nitrogen atoms (to prevent protonation), forming © 2013 American Chemical Society

Received: May 1, 2013 Published: November 13, 2013 9427

dx.doi.org/10.1021/jm4010434 | J. Med. Chem. 2013, 56, 9427−9440

Journal of Medicinal Chemistry

Article

Scheme 1. Clofilium Pharmacophore

Scheme 2a

a Reagents and conditions: (a) CuI, PdCl2(PPh3)2, Et3N, THF, rt; (b) TsCl, KOH, Et2O, 0 °C to rt; (c) N-ethylheptylamine, K2CO3, DMF, rt; (d) N-heptyldiethylamine, CH3CN, reflux.

■

first synthesized in two steps by (i) copper-catalyzed addition of the Grignard reagent derived from trimethylsilylacetylene,22 followed by (ii) cleavage of the silyl group by K2CO3 in methanol.23 Compounds containing a two-carbon spacer were obtained from the corresponding phenylpropionic acids by conversion to the corresponding ester,24 followed by a two-step, one-pot procedure25 consisting of a reduction to the corresponding aldehyde in the presence of DIBAL-H at −78 °C followed by methanolysis and treatment by the Ohira−Bestmann reagent in the presence of K2CO3.26 Finally, a copper-catalyzed Mannich reaction between the acetylenic derivatives 16−19 and 31−33, paraformaldehyde, and a secondary ethylamine afforded the corresponding propargylic amines 20−26 and 34−39 (Scheme 3).21 Derivatives 42 and 43 containing a 1,3-diyne were synthesized by copper-catalyzed Glazer−Hay coupling in air between 4-ethynyl substituted benzenes and N-propargyl-N-ethylbutylamine.27 This latter compound was obtained by reaction of N-ethylbutylamine with propargyl bromide (Scheme 4). Compounds bearing an ether linker were also synthesized, from 4-chlorophenol 44, 4chlorobenzylalcohol 48, and biphenyl methanol 49. These alcohols were treated with propargyl bromide in the presence of K2CO3 or NaH to afford alkynes 45 and 50−51.28 A similar Mannich reaction as previously used provided the corresponding propargylamines 46−47 and 52−55 (Scheme 5). Compounds bearing an (E)-alkene were synthesized from the corresponding allylic acetates via a palladium-catalyzed allylic amination.29

RESULTS AND DISCUSSION Design and Synthesis. The derivatives that were synthesized were built along the clofilium/N-ethyl-N-[4-(4-nitrophenyl)butyl]heptan-1-amine (LY97241)17 pharmacophore (Scheme 1). It consists of a tertiary ethylamine (or quaternary ammonium) substituted on one side by an alkyl chain and on the other by a para-substituted alkylphenyl moiety. The modifications brought to this scaffold were the introduction of unsaturations in the linker A between the nitrogen atom and the phenyl group, in the form of aliphatic cycles including the nitrogen atom and double and triple bonds. The introduction of a heteroatom (oxygen) and an additional substituent (diphenylmethylene instead of benzyl) were also carried out. The synthetic approach that was adopted furnished tertiary ethylamines, and a few of them were further ethylated to yield the corresponding diethylammonium salts. Compounds bearing a triple bond were synthesized using two different methods. 4-Substituted phenylacetylene-bearing compound 6 was accessed as follows: Sonogashira coupling of 4-chloroiodobenzene 2 and propargylic alcohol,18 tosylation of the hydroxyl group,19 and then substitution by a N-ethylheptylamine.20 The tosylate was also aminated with N-heptyldiethylamine, affording the corresponding tertiary ammonium tosylate salt 7 (Scheme 2).21 For substrates having one methylene group between the aromatic ring and the triple bond, terminal alkynes 16−19 were 9428



Article

Scheme 3a

Reagents and conditions: (a) BrMg-C≡C-SiMe3, CuBr, THF, reflux; (b) K2CO3, MeOH, rt; (c) Cu(OAc)2·H2O, Et(H)NCH2CH2R′, (CH2O)n, 1,4-dioxane, 90 °C; (d) (MeO)2CO, H2SO4, 80 °C; (e) (i) DIBAL-H, CH2Cl2, −78 °C, (ii) Ohira−Bestmann reagent, K2CO3, MeOH, rt.

a

Scheme 4a

a

Reagents and conditions: (a) HC≡C−CH2Br, K2CO3, acetone, reflux; (b) CuCl2, 4-R-C6H5−C≡CH, O2, CH2Cl2, rt.

Scheme 5a

Reagents and conditions: (a) K2CO3, HC≡C−CH2Br, acetone, reflux; (b) Cu(OAc)2·H2O, Et(H)NCH2CH2R′, (CH2O)n, 1,4-dioxane, 90 °C; (c) NaH, HC≡C−CH2Br, THF, 0 °C to rt.

a

proved to be ineffective this time, whereas the addition of triphenylphosphine as a ligand afforded the desired allylamines 73−78 (Scheme 7).29 Propargylamine 21 was converted to the corresponding (Z) allylamine 79 by hydrogenation over Lindlar’s catalyst (Scheme 8).36 Allylamine 78 was converted to the corresponding ethylammonium iodide 80 by alkylation with ethyl iodide (Scheme 9). We intended to do a hydrogenation of 78 on Pd/C in order to obtain the alkyl analogue, but the sole product of this reaction was the dechlorinated amine 81 (Scheme 10). A Friedel−Crafts acylation of chlorobenzene 82 with ethyl-(4-chloroformyl)butyrate 83 using AlCl3 afforded ester 84,37 and a Wolff−Kishner reduction of 84 and acid 85 afforded acids 86−87.38 Those were in turn esterified, and the obtained esters 88−89 were reduced to the corresponding aldehydes in the presence of DIBAL-H and a reductive amination with sodium triacetoxyborohydride and N-ethylbutylamine or

4-Chlorobenzaldehyde 56 was converted into the corresponding cinnamyl alcohol by a Horner−Wadsworth−Emmons reaction followed by a reduction of the obtained ester 57 with DIBAL-H.30 The resulting alcohol 58 was then acetylated.31 Treatment of acetate 59 with secondary ethylamines in the presence of a catalytic amount of palladium(II) chloride and one equivalent of tertbutylammonium bromide (TBAB) afforded the desired cinnamylamines 60−61 (Scheme 6).32 Conversely, 4-chlorophenethylalcohol 63 was oxidized to the corresponding aldehyde using the Dess−Martin reagent.33 3-(4Chlorophenyl)-propionaldehyde 66 was obtained by Heck coupling of 4-chloroiodobenzene 2 and allylic alcohol 62.34 The resulting aldehydes 65 and 66 (as well as 64 (R = H, n = 1)) were treated by an excess of vinylmagnesium bromide, affording the corresponding branched allylic alcohols 67−69,35 which were in turn acetylated.31 The catalytic system used previously 9429



Article

Scheme 6a

Reagents and conditions: (a) DBU, LiCl, (MeO)2P(O)CH2C(O)OMe, CH3CN, rt; (b) DIBAL-H, CH2Cl2, −78 to −40 °C; (c) Ac2O, DMAP, CH2Cl2, 0 °C to rt; (d) Et(H)NCH2CH2R′, PdCl2, TBAB, K2CO3, toluene, 85 °C.

a

Scheme 7a

Reagents and conditions: (a) DMP, CH2Cl2, 0 °C to rt; (b) Pd(OAc)2, NaHCO3, TBAC, DMF, 25 °C; (c) H2CC(H)-MgBr, THF, −10 °C; (d) Ac2O, DMAP, CH2Cl2, 0 °C to rt; (e) Pd(OAc)2, PPh3, EtN(H)CH2CH2R′, K2CO3, THF, 65 °C.

a

N-ethylheptylamine afforded amines 90−92, respectively (Scheme 11). Finally, six-membered aliphatic rings were used as a source of conformational constraint. Benzyl- and benzoylpiperidines 95 and 96 were obtained by alkylation of piperidines 93 and 94, respectively (Scheme 12).39 Biology. Displacement of [3H]astemizole with increasing concentrations of cold ligand afforded concentration−effect curves that showed two types of curves, depending on the compound tested. While most of the curves showed a one-site profile, some compounds gave a curve with two affinity sites. This was most notable for the quaternary amine derivatives, such as 1, 7, and 81, as there is a large separation between the high- and low-affinity sites. This biphasic behavior in radioligand binding studies has been demonstrated by us before20 and may have to do with the different conformational states the channel can attain. In that same study, it was shown that the high affinity site corresponds well to data derived from patch-clamp studies. It was therefore decided to test all other newly synthesized analogues at 12 concentrations at least, allowing us to reliably determine the high-affinity site IC50 values for each compound displaying one.

The patch-clamp measurements were carried out in a manual assay with at least three different concentrations of the selected compounds, and a dose−response curve was fitted to determine the IC50 value. Structure−Activity Relationships. Influence of Charge. First of all, the influence of the presence of a quaternary ammonium was studied. The affinity observed for compound 90 was lower than for clofilium 1 (43 vs 2 nM, Table 6 and Table 1). The same trend was observed when comparing compounds 6/7 (4080 vs 40 nM, Table 1) and 74/80 (49 vs 1.5 nM, Table 5). In an earlier study on derivatives of E-4031, we found no such significant difference between tertiary amines and quaternary ammonium derivatives, provided the additional substituent did not increase the lipophilicity of the ammonium too much.40 According to the binding mode proposed for clofilium by Mitcheson et al.,41,42 a possible explanation for this divergence could lie in the presence of an n-heptyl chain (instead of arylethyl in the E-4031 derivatives) on the nitrogen atom. This tail probably does not interact with the Phe656 residues at the bottom of the channel as specifically as aromatic residues do (hydrophobic

Scheme 8a

a

Reagents and conditions: (a) Lindlar’s catalyst, quinoline, H2, EtOAc. 9430



Article

Scheme 9a

a

Reagents and conditions: (a) ethyl iodide, acetone, 20 °C.

Scheme 10a

a

Reagents and conditions: (a) H2, Pd/C, MeOH.

Scheme 11a

Reagents and conditions: (a) AlCl3, DCE; (b) KOH, N2H4, 150−170 °C; (c) (MeO)2CO, H2SO4, 80 °C; (d) (i) DIBAL-H, CH2Cl2, −78 °C, (ii) EtN(H)CH2CH2R, NaBH(OAc)3, DCE, rt. a

Scheme 12a

a

Table 2. Binding Affinities of Propargylamines 20−26

compd 20 21 22 23 24 25 26

Reagents and conditions: (a) n-bromoheptane, NaI, K2CO3, acetone, reflux.

Table 1. Binding Affinities of Clofilium 1, Propargylamine 6, and Propargylammonium Derivative 7 compd

IC50 [nM]a

1 6 7

2±1 4080 ± 1520 40 ± 21

R H H Cl Cl Me Me H

R′ Et n-pent Et n-pent Et n-pent n-pent

R″ H H H H H H Ph

IC50 [nM]a 7820 ± 2280 3543 ± 224 2654 ± 468 865 ± 61 3470 ± 710 5090 ± 1430 238 ± 67

IC50 ± SEM (n = 3) of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel.

a

a High-affinity IC50 ± SEM (n = 3) of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel.

charge on the nitrogen atom might prevent or at least reduce folding of the tail by repulsive interaction, which increases the lipophilic interactions with Phe656 and thus leads to a better affinity for the hERG K+ channel.

interaction vs π−π stacking) and has thus the possibility to coil. In the case of an ammonium compound, the presence of a permanent 9431



Article

Table 3. Binding Affinities of Propargylamines 34−39, 42−43, and 46−47

compd

R

R′

A

IC50 [nM]a or displacement [%]b

34 35 36 37 38 39 42 43 46 47

H H Cl Cl Me Me Cl Me Cl Cl

Et n-pent Et n-pent Et n-pent Et Et Et n-pent

−CH2−CH2− −CH2−CH2− −CH2−CH2− −CH2−CH2− −CH2−CH2− −CH2−CH2− −CC− −CC− −O−CH2− −O−CH2−

48% 4394 ± 1073 6007 ± 2481 765 ± 33 5590 ± 470 1890 ± 250 9929 ± 1701 47% 14443 ± 1923 2473 ± 439

Table 6. Binding Affinities of Amines 81 and 90−92

R″

IC50 [nM]a

52 53 54 55

Cl Cl H H

Et n-pent Et n-pent

H H Ph Ph

7372 ± 859 834 ± 95 4810 ± 460 970 ± 110


a

Table 5. Binding Affinities of Allylamines 60−61 and 73−80

compd

R

R′

n

IC50 [nM]a or displacement [%]b

60 61 73 74 75 76 77 78 79c 80d

Cl Cl Cl Cl H H Cl Cl H Cl

Et n-pent Et n-Pent Et n-pent Et n-pent n-pent n-pent

0 0 1 1 1 1 2 2 1 1

1050 ± 150 243 ± 160 1127 ± 500 49 ± 26 49% 176 ± 32 1868 ± 307 238 ± 84 1178 ± 237 1.5 ± 0.3

n

IC50 [nM]a

81 90 91 92

H Cl Cl Cl

n-pent n-pent Et n-pent

1 1 2 2

506 ± 83 43 ± 3 2500 ± 113 123 ± 25

the OH group of Ser624. Thus, replacing the 4-chlorophenyl moiety with a nonsubstituted phenyl ring led to an expected significant decrease in affinity (506 nM for 81 vs 43 nM for 90, Table 6). This phenomenon was verified in all cases for all the different derivatives having the same scaffold (76 vs 74, 23 vs 21, 37 vs 35). In the case where the chloro substituent was replaced by a methyl group, the decrease in affinity was less outspoken, albeit significant (765 nM for 37 (Cl), 1890 nM for 39 (Me), and 4394 nM for 35 (H), Table 3). The results were also consistent with previous reports on general SAR for hERG K+ channel blockers.9 Influence of the Alkyl Chain. When the n-heptyl chain was replaced by a n-butyl group, a great decrease in binding affinity was observed in all cases (20 vs 21, 22 vs 23, and 24 vs 25 (Table 2), 34 vs 35, 36 vs 37, 38 vs 39, and 46 vs 47 (Table 3), 52 vs 53 and 54 vs 55 (Table 4), 60 vs 61, 73 vs 74, 75 vs 76, and 77 vs 78 (Table 5), and 91 vs 92 (Table 6)). These data seem to be in agreement with the available crystal structures of potassium channels in the open state showing there is a large aperture at the bottom of the cavity.43,44 Also, the first studies conducted by Armstrong et al. using quaternary amines to probe the structure of potassium channels showed that an increase in the bulkiness of the substituent of the ammonium led to a better binding affinity.45 Conversely, and as observed, a decrease of its size leads to a decrease in affinity, as was also observed during optimization of a series of ligands for the σ1 receptor.46 Influence of the Rigidity of the Linker. The influence of rigidity was studied for compounds bearing the same number of carbon units between the aromatic ring and the nitrogen atom. Starting from the tertiary amine analogue of clofilium 90, the introduction of an (E)-alkene in the butyl chain between the phenyl ring and the nitrogen atom did not change binding affinity (43 nM for 90 vs 49 nM for 74). However, when an alkyne was introduced, the affinity was 20 times lower (865 nM for 23). These results suggest that clofilium derivative 90 binds to the channel with a nonbent conformation of its butyl linker, similar to the one of 74, because there is no variation between the alkyl linker and the (E)-alkene linker. On the other hand, when an alkyne is introduced, this conformation cannot be adopted any longer and the affinity decreases. The same trend was observed for derivatives without the chloro substituent (81, 76, and 21). Interestingly, replacing the (E)-alkene with a (Z)-alkene led to a 7-fold decrease in affinity (1178 nM for 79 vs 176 nM for 76). This seems to further consolidate the argument of a linear, nonbent conformation of the butyl linker because the geometry of a (Z)-alkene differs very much from the one of an (E)-alkene. Rigidity was also introduced through the incorporation of cycles and carbonyl groups. In this case, the nitrogen atom was included within a piperidine ring, either in a benzylpiperidine

Table 4. Binding Affinities of Propargylamines 52−55

R′

R′


IC50 ± SEM (n = 3) of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel. b Displacement of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel at a concentration of 10 μM.

R

R

a

a

compd

compd

IC50 ± SEM (n = 3) of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel. b Displacement of specific [3H]astemizole binding to membranes of HEK293 cells stably expressing the hERG K+ channel at a concentration of 10 μM. c(Z)-Alkene. dQuaternary ethylammonium. a

Influence of Substitution of the Aromatic Ring. According to the general binding model for clofilium,41 the chloro substituent plays a fundamental role in interacting with the channel cavity via 9432



Article

Influence of the Nature of the Linker. When the methylene unit adjacent to the phenyl ring in 36 (3360 nM) is replaced by an oxygen atom, the affinity plummets further (46, 14443 nM). Because the rigidity and geometry of an ether function are similar to those of a methylene, the decrease in affinity might be of electronic origin, such as a withdrawal of electrons from the phenyl ring, thus making the polar and π-stacking interactions with Tyr652 and Ser624 less favorable. This strategy had already been used to decrease hERG affinity and recently in the optimization of NR2B NMDA antagonists.48 When further elongating 46 with an additional methylene unit between the oxygen atom and the phenyl ring, compound 52 showed a 2-fold increased affinity compared to 46 (7372 vs 14443 nM). This seems to correlate with the importance of placing the oxygen atom directly on the phenyl ring to reduce the binding affinity for the hERG potassium channel. When the linker is further rigidified by replacing the ethylene unit with an alkyne, the obtained diyne 42 displays a 3-fold lower affinity compared to 36 (9929 vs 3360 nM). Because 42 is much more rigid than 46, this result further accounts for an electronic effect of the oxygen atom. The same trend is observed in the methylsubstituted series (38 vs 43). The addition of a second aromatic ring on the benzylic carbon was also studied, leading to an increase in affinity (3543 nM for 21 vs 238 nM for 26). A possible explanation would be that the introduction of the second aromatic ring has two effects: (i) an increase in lipophilicity and (ii) increasing the possibilities of π−π stacking of the molecule with the channel residues. The latter has been reported previously and recently discussed in an article about “minimally structured” hERG blockers, in which such blockers contain a second, distal aromatic ring.47 The gain in affinity roughly compensates for the loss of a chloro substituent (834 nM for 53 vs 970 nM for 55 and 7372 nM for 52 vs 4810 nM for 54, Table 4). Patch-Clamp Study. Whole-cell voltage-clamp recordings were performed in a manual patch-clamp setup (see Experimental Section and Figure 1a,b), after we learned that many of our compounds could not be tested in an automatic setup. IC50 values obtained in the radioligand binding assay were converted to Ki values using the Cheng−Prusoff equation (Ki is more representative of the equilibrium of the ligand with the protein in the absence of a radioligand, which is the case in a patch-clamp assay). First of all, clofilium 1 was evaluated and an IC50 value of 50 nM was obtained, which is 125 times higher than the affinity value obtained in the radioligand binding assay. This difference can be explained by the fact that most blockers enter the channel from the cytoplasm.49 Because clofilium is a permanently charged entity, its crossing of the membrane can be greatly decreased if not compromised, thus yielding a lower apparent potency in an intact cell assay. Six other compounds were tested, all tertiary amines containing the 4-chlorophenyl motif of clofilium (Table 8). These compounds turned out to be approximately 2−34-fold more potent in this assay than in the radioligand binding assay. Compounds 74 and 90, which were the most potent in the radioligand binding assay, remained the most potent in the patch-clamp assay as well (in the case of 74, the IC50 could not be determined precisely for lack of any steady state at the concentrations tested). Similarly, compounds 47 and 42 remained the least potent. Inexplicably, compounds 23 and 37 seemed to stand out of the general trend (although in the case of 23 no steady state was reached below 100 nM) (Table 8). Similar results in terms of increase in potency were actually observed in

Table 7. Binding Affinities of Amines 95−96

compd

X

IC50 [nM]a

95 96

H,H O

58 ± 6 124 ± 13


a

(95) or in a benzoylpiperidine scaffold (96) (Table 7). Both compounds showed a higher affinity for the hERG potassium channel than 81 and its (E)-alkene derivative 76 (58 nM for 95, 124 nM for 96). This effect could result from the increased exposure of the protonated nitrogen atom when it is included in a ring compared to a noncyclic amine (which also explains the higher basicity of cyclic amines compared to their acyclic analogues), thus favoring the cation (ammonium)−π interactions with Tyr652.41 To our knowledge, previous studies using cycles as a way to alleviate hERG affinity reported strategies of either adding another cycle (bridged or fused) to the one containing the nitrogen atom,11,12 or adding a cycle in the linker between the aromatic moiety and the nitrogen atom.16 Our approach thus shows that including the nitrogen atom within a cycle (as opposed to it being contained within an acyclic structure) rather increases hERG affinity. Influence of the Linker Length. Recent studies have evidenced key structural features needed to obtain the most simple hERG blockers, among which is the distance between the nitrogen atom and the aromatic ring(s), which is supposed to range from 5−9 Å, for flexible compounds.47 We determined the distance between the center of the aromatic ring and the nitrogen atom for compounds 6, 23, 37, 61, 74, 78, 90, and 92 using ChemBio3D Ultra 13 (after MM2 energy minimization), and all distances were in the 6.1−9.0 Å range, which falls in the reported range. We then moved on to studying the influence of rigidity coupled to linker length for each type of linker: alkane, alkene, alkyne. In the alkane series, the addition of one methylene unit, thus leading to an n-pentyl linker, led to a slight decrease of the affinity (43 nM for 90 vs 123 nM for 92, Table 6). Although the length does not appear optimal any longer, the chain is apparently flexible enough in order to establish the binding interactions with the residues of the cavity. In the alkene series, an elongation to a 2-pentenyl linker of the chain length leads to a 5-fold decrease in the binding affinity (49 nM for 74 vs 238 nM for 78). Because of the rigidity, the linker apparently has less conformational flexibility to ensure the crucial interactions with the channel. In the same way, a shortening to a 2-propenyl linker gives a similar decreased affinity (243 nM for 61). In that case, the compound will probably either make a compromise between the two essential interactions (with Tyr652 and Ser624)41 or favor one over the other. In the alkyne series, an elongation to a 2-pentyne linker did not change the affinity much (865 nM for 23 vs 765 nM for 37) because the already diminished affinity due to rigidity is probably compensated by a gain in flexibility. On the other hand, a shortening to a 2-propyne linker led to an affinity 40 times lower (4080 nM for 6). This latter compound is so rigid that it most probably cannot accommodate more than one of the interactions with the cavity at a time, accounting for its low affinity. 9433



Article

Table 8. Comparison of Inhibitory Activity of Selected Compounds in Radioligand Binding Assay and Patch-Clamp Assay compd

Ki [nM]a

IC50 [nM] patch-clampb

1 23 37 42 47 74 90

0.40 ± 0.19 167 ± 12 148 ± 6.4 1921 ± 329 478 ± 85 9.5 ± 5.0 8.3 ± 0.58

50 4.9 53 473 58

Removal of Human Ether-Ã -go-go Related Gene (hERG) K+ Channel

Recommend Documents