3D-QSAR and 3D-QSSR models of negative allosteric modulators facilitate the design of a novel selective antagonist of human α4β2 neuronal nicotinic acetylcholine receptors

Abstract

Subtype selective molecules for α4β2 neuronal nicotinic acetylcholine receptors (nAChRs) have been sought as novel therapeutics for nicotine cessation. α4β2 nAChRs have been shown to be involved in mediating the addictive properties of nicotine while other subtypes (i.e., α3β4 and α7) are believed to mediate the undesired effects of potential CNS drugs. To obtain selective molecules, it is important to understand the physiochemical features of ligands that affect selectivity and potency on nAChR subtypes. Here we present novel QSAR/QSSR models for negative allosteric modulators of human α4β2 nAChRs and human α3β4 nAChRs. These models support previous homology model and site-directed mutagenesis studies that suggest a novel mechanism of antagonism. Additionally, information from the models presented in this work was used to synthesize novel molecules; which subsequently led to the discovery of a new selective antagonist of human α4β2 nAChRs.

Neuronal nicotinic acetylcholine receptors (nAChRs) are ligand gated ion channels and members of the cys-loop family of receptors. nAChRs are found both in the peripheral and central nervous systems and are implicated in many diseases and disorders such as: Alzheimer’s disease, epilepsy, autism, Parkinson’s disease, depression, anxiety, and nicotine addiction.^1,2 Worldwide, nicotine addiction is a significant problem. Smoking is the primary cause of preventable death worldwide and roughly 90% of the people who attempt to quit are unable to do so.³ It is now known that α4β2 nAChRs are primarily responsible for the addiction to tobacco related products.^4,5,6 Current FDA approved treatments for tobacco addiction are nicotine replacement, bupropion (Zyban®), and varenicline (Chantix®). Each of these therapies has a modest success of 20%–30% abstinence 1 year after quit date.^7,8 However, drugs such as varenicline have been associated with severe adverse cardiovascular effects.⁹ This combined with the low success rates of therapies warrant the need for novel small molecules that can be used in nicotine cessation. In an attempt to discover better therapeutics for nicotine cessation, some laboratories have proposed non-competitive antagonists that target nAChRs.^10,11 Mecamylamine, a non-selective non-competitive nAChR antagonist, was shown to promote 40% abstinence at the 1 year mark when used as an agonist-antagonist therapy in combination with the nicotine patch.¹⁰ In addition, Yoshimura et al. (2007)⁷ discovered a novel negative allosteric modulator (NAM) that was selective for neuronal nAChRs as opposed to the muscle nAChR which significantly blocked nicotine self-administration on fixed and progressive ratio schedules in rats. These data support the use of non-competitive antagonists and NAMs as nicotine cessation therapies; however, to produce new therapeutic molecules it is believed that nAChR subtype selectivity must be pursued.¹²

Our laboratory has previously published the synthesis and pharmacology of a novel class of NAMs.^{13,14,15,16,17,18,19} We have previously reported a novel NAM, KAB-18, which shows selectivity for human α4β2 (Hα4β2) nAChRs and through SAR have identified several chemical features important for its selectivity.¹⁹ One problem with the study of non-competitive and allosteric agents is the fact that most of these agents lack information concerning the site of interaction on their target receptor. To address this, we have constructed a homology model for the extracellular domain of the Hα4β2 nAChR and have identified the site in which these NAMs interact allosterically through blind docking and molecular dynamics (MD) simulations.¹⁹ In this study, three-dimensional qualitative structure-activity relationship (3D-QSAR) studies and three-dimensional qualitative structure-selectivity relationship (3D-QSSR) studies were completed to study the relationship between functional activity (e.g., IC₅₀ values) and selectivity of NAMs with their 3D structures. This study reports the construction and analysis of models that predict the detailed structural interactions of this novel class of NAMs^18,19 with their binding site on Hα4β2 nAChRs and Hα3β4 nAChRs. In addition to this, we propose a model which distinguishes the physiochemical features that are important for selectivity for Hα4β2 nAChRs versus Hα3β4 nAChRs that also agree with previously reported homology modeling, SAR, and site-directed mutagenesis studies.^19,26 Finally, these models were used in the generation of novel Hα4β2 nAChR antagonists.

To facilitate the presentation of data, four regions for the NAM scaffold have been defined (Figure 1B). These four regions were defined from a pharmacophore model that was generated previously by using KAB-18 and KAB-18 like molecules.¹⁹ This pharmacophore model featured four hydrophobic regions and one hydrogen bond acceptor region. Region 1 was defined as the substitution on the nitrogen moiety of the piperidine ring containing hydrophobic domain 1 (Figure 1B). Region 2 was defined as the ester acyl substitution containing the biphenyl (Figure 1B). Region 3 was the piperidine ring which has been defined in the pharmacophore as the fourth hydrophobic region (Figure 1B). Region 4 was the linkage between Region 2 and Region 3, containing an ester bond with a hydrogen bond accepting domain (Figure 1B). All of the NAMs presented in this manuscript contain one or more stereiogenic centers. In construction of the QSAR and QSSR models the selected conformation of compounds used in the alignment play a pivotal role in determining the position of the field contribution maps and validation of the model. The conformation of our NAMs was determined by selecting the stereoisomers that match those found to be most stable in previously conducted docking and MD simulations on the Hα4β2 nAChR homology model.¹⁹ Enantiomers of NAMs at position C3 (respective of the piperidine ring, Region 3, Figure 1B) have shown no functional difference in either Hα4β2 or Hα3β4 nAChRs.¹⁹ Choosing this set of conformations for NAMs in the QSAR and QSSR models allow the results to be compared to homology model data in addition to previous SAR data.¹⁹ The alignment of NAMs (Figure 1A) and the construction of QSAR and QSSR (Hα4β2/Hα3β4) models using CoMSIA are described in detail in the Methods section (refer to supplemental information). For every molecule used in the construction of the models, 3D structures were prepared and are also described in the Methods section. All functional data for training set molecules are detailed in Table 1 and the functional data for test set molecules are found in Table 2. Using SYBYL’s CoMSIA program, linear regression analyses were performed to correlate the experimentally derived pIC₅₀ values with the computationally derived pIC₅₀ values (Figure S1). 55 molecules were chosen for the Hα4β2 QSAR model training set and 42 molecules were chosen for the Hα4β2 QSAR and the QSSR models (Table 1). Correlation coefficients (r²) for the Hα4β2 QSAR, Hα3β4 QSAR and QSSR models were 0.766, 0.761, and 0.871 respectively for the training set (Table 3, Figure S1). 10 molecules were chosen for the test set for all molecules (Table 2). Correlation coefficients (r²) for the Hα4β2 QSAR, Hα3β4 QSAR and QSSR models were 0.620, 0.621, and 0.480 respectively for the test set (Table 3, Figure S1).

Figure 1. Alignment of the negative allosteric modulators.

(A) Alignment of molecules used in QSAR and QSSR models (Figures 2, 3, and 4). (B) Regions of the NAM scaffold displayed with scaffold molecule, KAB-18.

Table 1.

Observed and predicted IC₅₀ values of NAMs used in QSAR and QSSR models (Training Set Molecules).

		QSAR (Hα4β2)		QSAR (Hα3β4)		QSSR (Hα4β2/Hα3β4)
	Structure	Obs. pIC50a	Pred. pIC50	Obs. pIC50a	Pred. pIC50	Obs. [SI]a	Pred. [SI]
APB-1		4.99	4.96
APB-10		5.00	4.86
APB-4b		4.84	4.75	4.81	4.87	0.02	−0.08
APB-8		4.84	4.86	4.65	4.13	0.19	0.79
COB-1b		5.15	5.20	5.09	5.41	0.06	−0.19
COB-2b		5.16	4.93	5.32	5.21	−0.16	−0.07
COB-3b		5.62	5.67	5.19	4.82	0.43	0.73
COB-8		5.04	5.03	5.18	5.07	−0.14	−0.05
DDR-13b		5.22	5.20	3.48	3.43	1.74	1.78
DDR-15b		5.37	5.23	5.74	5.57	−0.38	−0.26
DDR-18b		5.19	5.32	3.41	3.36	1.78	1.97
DDR-19b		5.55	5.59	5.37	5.34	0.19	0.34
DDR-20		4.94	4.89	4.86	3.64	0.08	1.07
DDR-21		4.96	4.99	4.51	4.65	0.45	0.19
DDR-25		5.02	5.00
DDR-26		4.92	5.06	4.11	4.66	0.81	0.15
DDR-27		5.12	5.05
DDR-3b		4.78	4.80	4.84	5.14	−0.06	−0.54
DDR-4		4.68	4.64	4.39	4.55	0.29	−0.05
DDR-5b		4.70	4.66	1.00	0.68	3.70	4.00
FFB-1		5.04	4.2	3.75	0.73	1.29
IB-2b		4.98	5.15	4.96	4.84	0.02	0.31
IB-4b		5.05	5.15	5.02	4.84	0.02	0.31
JHB-7		5.25	5.36	4.13	4.13	1.12	1.18
JHB-9b		4.75	4.86	4.99	5.01	0.24	−0.35
JHB-11		5.15	5.24
JHB-12		5.27	4.99
JHB-13		4.40	4.53	4.07	4.09	0.33	0.24
KAB-9		4.18	4.39	4.92	4.60	−0.75	−0.32
KAB-10		4.61	5.01	5.38	4.94	−0.77	−0.08
KAB-11		4.53	4.60
KAB-13		4.95	4.80	4.73	4.97	0.22	−0.04
KAB-15		4.66	4.70	4.94	4.90	0.14	0.11
KAB-16		5.41	5.26	4.80	4.73	0.61	0.55
KAB-17		5.13	5.05	5.17	5.18	−0.04	−0.06
KAB-18b		4.87	5.03	1.00	1.73	3.87	3.30
KAB-19b		5.26	5.31	5.37	5.42	−0.11	0.04
KAB-20b		5.30	5.22	5.23	5.13	0.07	0.05
KAB-21b		5.20	5.14	4.73	4.79	0.47	0.43
KAB-22b		5.03	5.11	4.64	4.74	0.39	0.34
KAB-23b		5.10	5.26	4.92	4.93	0.19	0.34
KAB-24b		5.10	5.14	5.26	4.71	−0.16	0.37
KAB-28		5.14	5.24	4.57	4.78	0.57	0.44
KAB-30b		5.31	5.26	5.30	5.17	0.01	0.07
KAB-32		5.77	5.24
KAB-33		4.77	4.75
KAB-35		4.84	4.84
KAB-37		4.91	5.09	4.13	4.22	0.78	0.56
KAB-38		4.76	4.63	5.04	4.76	−0.28	−0.03
NEB-2		5.38	5.29
PPB-1		4.97	4.96	4.84	4.81	0.14	0.23
PPB-3		5.29	5.22	4.48	4.71	0.81	0.58
PPB-6		5.05	5.11
SMB-1		5.08	4.92	2.70	3.10	2.38	1.89
SMB-2		5.06	4.93

^a-Log of geometric mean, n = 4–10

^bPreviously reported (Henderson et al., 2010)

Table 2.

Observed and predicted IC₅₀ values of NAMs used in QSAR and QSSR models (Test Set Molecules).

		QSAR (Hα4β2)		QSAR (Hα3β4)		QSSR (Hα4β2/Hα3β4)
	Structure	Obs. pIC50a	Pred. pIC50	Obs. pIC50a	Pred. pIC50	Obs. [SI]a	Pred. [SI]
KAB-39		4.90	5.02	4.38	4.32	0.68	0.82
PPB-5		4.41	4.87	4.54	4.66	−0.13	−0.28
PPB-9		5.35	5.12	4.93	4.55	0.42	0.29
PPB-11		5.35	5.24	4.93	4.78	0.42	0.51
COB-5		5.29	5.35	5.43	4.99	−0.14	−0.44
APB-21		4.90	4.85	4.81	4.38	0.09	0.35
COB-4		5.09	5.00	4.98	5.05	0.11	0.65
DDR-17		5.54	5.17	5.14	5.10	0.39	0.13
KAB-8		4.84	5.03	4.97	4.75	−0.14	0.26
KAB-14		4.82	4.95	4.76	4.81	0.06	−0.01

^a-Log of geometric mean, n = 4–10

Table 3.

QSAR and QSSR analysis results.

	Data Set
Dependent Variable	QSAR(Hα4β2)	QSAR(Hα3β4)	QSSR
r2 (training set)	0.856	0.761	0.871
standard error	0.15	0.57	0.57
F value	26.13	15.1	14.3
Components	6	6	6
r2 (test set)	0.620	0.621	0.480

CoMSIA descriptors for steric (green/yellow field maps), electrostatic (blue/red field maps), hydrophobic (purple/grey field maps), hydrogen bond donor (cyan/magenta field maps), and hydrogen bond acceptors (orange/red field maps) were generated for the Hα4β2 QSAR, Hα3β4 QSAR, and QSSR models (Figure 2, 3, and 4, respectively). For detailed description of the model construction process, refer to the Methods section (see supplemental information). KAB-18 was used as the structural scaffold for all field contribution maps (Figures 2, 3, and 4) due to its selectivity for Hα4β2 nAChRs.¹⁹

Figure 2. CoMSIA models for Hα4β2 nAChRs.

Steric (A), electrostatic (B), hydrophobic (C), hydrogen bond donor (D), and hydrogen bond acceptor (E) field contribution maps are shown with KAB-18. Bottom right, color key to field contribution maps. Greater potency (lower IC₅₀ values) is correlated with: less bulk near yellow/grey, more bulk near green/purple, more hydrogen bond acceptors near blue/red, less hydrogen bond acceptors near red, More hydrogen bond donors near cyan and less hydrogen bond donors near magenta.

Figure 3. CoMSIA models for Hα3β4 nAChRs.

Steric (A), electrostatic (B), hydrophobic (C), hydrogen bond donor (D), and hydrogen bond acceptor (E) field contribution maps are shown with KAB-18. Bottom right, color key to field contribution maps. Greater potency (lower IC₅₀ values) is correlated with: less bulk near yellow/grey, more bulk near green/purple, more hydrogen bond acceptors near blue/red, less hydrogen bond acceptors near red, More hydrogen bond donors near cyan and less hydrogen bond donors near magenta.

Figure 4. 3D-QSSR CoMSIA models of NAMs.

Steric (A), electrostatic (B), hydrophobic (C), hydrogen bond donor (D), and hydrogen bond acceptor (E) field contribution maps are shown with KAB-18. Bottom right, color key to field contribution maps. Greater selectivity (more potent on Hα4β2 vs. Hα3β4) is correlated with: less bulk near yellow/grey, more bulk near green/purple, more hydrogen bond acceptors near blue/red, less hydrogen bond acceptors near red, More hydrogen bond donors near cyan and less hydrogen bond donors near magenta.

For the Hα4β2 nAChR QSAR model, steric contributions were important surrounding the phenyl rings of regions 1 and 2 as well as the piperidine of region 3 (Figure 2A). Important electrostatic interactions were predicted surrounding the anthranilic moiety (positive) and the phenyl (negative) of region 2 (Figure 2B). Hydrophobic contributions were shown to be important around the phenyl rings of regions 1 and 2 as well as the piperidine of region 3 (Figure 2C) Hydrophobic contributions were shown to be disfavored around the propyl chain of region 1 (Figure 2C). The protonated hydrogen of the piperidine nitrogen of region 3 is predicted to act as a hydrogen bond donor (Figure 2D) and the carbonyl of region 4 is predicted to act as a hydrogen bond acceptor (Figure 2E). These predictions support the binding mode of KAB-18 that has been recently presented by molecular dynamics (MD) simulation on the Hα4β2 nAChR homology model.¹⁹ This MD simulation predicted that the protonated hydrogen of the piperidine nitrogen is involved in a stabilizing hydrogen bond with the carbonyl group of Glu60 on the β2 subunit.¹⁹ This agrees with the favored hydrogen bond donor predicted to occur in Region 3 by the Hα4β2 QSAR where the cyan field surrounds the piperidine nitrogen (Figure 2D). The carbonyl of region 4 was also shown to be involved in another important hydrogen bond with Thr58 on the β2 subunit and this interaction was validated through site-directed mutagenesis.^19,26 This agrees with the favored hydrogen bond acceptor predicted in Region 4 of the Hα4β2 QSAR where the orange field surrounds the carbonyl of KAB-18 (Figure 2E). Additionally, the model predicts favorable interactions between negative electron densities of ligands in Region 1 (Figure 2B). This has been observed previously with an analog of KAB-18 that included a carbonyl in Region 1,¹⁹ which lead to a 3-fold increase in potency for Hα4β2 nAChRs.

The Hα3β4 nAChR QSAR model predicted that steric features in both region 1 and 2 had a positive effect on potency (Figure 3A). Electrostatic field maps predicted that negative electrostatics contribute to potency in region 1 and positive electrostatics contribute to potency in region 2 (Figure 3B). Hydrophobic features were shown to be disfavored in regions 1 and 2 (Figure 3C). Hydrogen bond donors were shown to be favorable with the piperidine of region 3 and the propyl chain in region 1 (Figure 3D). Hydrogen bond acceptors in region 4 were shown to be unfavorable around the carbonyl of region 4 and favorable with the phenyl of region 2 (Figure 3E). One feature that supports findings from previous SAR data,¹⁹ concerns aromatic-containing features in region 1 and 2 (e.g., phenylpropyl substitution in region 1) that contribute to a decrease of potency on Hα3β4 nAChRs. This agrees with the unfavored hydrophobics that are predicted by the Hα3β4 QSAR where the grey fields surround the phenylpropyl of KAB-18 (Figure 3C). One finding that differs significantly between the Hα4β2 and Hα3β4 QSAR is in the lack of a favored hydrogen bond acceptor in Region 3 of the Hα3β4 QSAR (Figure 3D). Also, the hydrophobics and hydrogen bond acceptor field maps are opposing when compared between the Hα4β2 and Hα3β4 QSAR models (Figure 2C, 2E and Figure 3C, 3E). Altogether, this suggests that this class of NAMs may have significantly different binding interactions between Hα4β2 and Hα3β4 nAChRs.

The QSSR model predicted that steric interactions are important for selectivity on Hα4β2 nAChRs surrounding the phenyl rings of Region 1, Region 2, and Region 3 (Figure 4). The field contribution maps also showed that sterics surrounding the propyl chain of Region 1 decrease selectivity (Figure 4A). Thus, the net effect of removing steric contributions in Region 1 may lead to an increase in selectivity. Electrostatic field maps predicted that positive electrostatics in Regions 1 (around the phenylpropyl of KAB-18) and 4 (ester region of KAB-18) contribute to selectivity for Hα4β2 nAChRs while negative electrostatics near Region 2 (biphenyl of KAB-18) decrease selectivity (Figure 4B). Hydrophobic maps predicted that hydrophobic features in Region 1, Region 2, and in Region 3 (piperidine region of KAB-18), specifically at the phenyl ring positions, contribute to selectivity for Hα4β2 nAChRs while hydrophobics in Region 4 decrease selectivity (Figure 4C). Donor contribution maps predicted that Hydrogen bond donors in Region 4 contribute to selectivity on Hα4β2 nAChRs (Figure 4D). Acceptor maps predicted that hydrogen bond acceptors are favored in Region 4 surrounding the ester carbonyl of (Figure 4E). Many of these findings predicted by the QSSR model support findings of our previous SAR data.¹⁹ As mentioned earlier, the presence of aromatic containing features in both Regions 1 and 2 were shown to be important for the selectivity on Hα4β2 nAChRs. The predictions of the steric and hydrophobic QSSR field contribution maps, which show both green and purple fields surrounding KAB-18’s phenylpropyl (Region 1) and phenyl group (Region 2), support this (Figure 4A and 4C). As mentioned earlier, SAR data supported the importance of the specific orientation of the carbonyl in Region 4. The acceptor field contribution maps support this finding as well as the favored HBA field (orange field) surrounds the carbonyl group of KAB-18. This suggests that the atom acceptor’s placement (flanking the biphenyl of Region 2) is important for selectivity (Figure 4E). Many of the features highlighted in this QSSR model correlate blind docking and MD simulation studies with an Hα4β2 nAChR homology model.¹⁹ The regions shown to be important in the QSSR model overlap with amino acid residues that have been proposed to interact with KAB-18 (i.e., Phe118 in Region 1 and Thr58 in Region 4 [both on β2 subunit]). These residues are not conserved in the Hα3β4 nAChR and may be important for mediating selectivity for Hα4β2 nAChRs. Previous SAR studies¹⁹ have identified 3 regions of importance for the selectivity of NAMs on Hα4β2 nAChRs: 1) The phenyl in Region 2; 2) the placement of the carbonyl group in Region 4; and 3) the presence of an aromatic feature in Region 1. When the aromatic feature is removed from Region 1, a loss of selectivity is observed; however, these molecules show an increase in potency up to 5-fold.¹⁹ This is supported by the Hα4β2 nAChR QSAR model (Figure 2C). The QSSR model also predicted that hydrophobic features were important for Hα4β2 selectivity in Regions 1 and 3 (Figure 4C) while steric features in Region 1 may decrease selectivity for Hα4β2 nAChRs (Figure 4A). Therefore, increasing the hydrophobics in Region 3 may improve selectivity for Hα4β2 nAChRs. We hypothesized that by combining these two findings (incorporating reduced sterics in Region 1 and increased hydrophobics in Region 3) we would obtain molecules that have improved potency on Hα4β2 nAChRs; but still preserve selectivity for Hα4β2 nAChRs. Novel molecules were synthesized (Scheme 1) and tested for inhibitory activity on both Hα4β2 and Hα3β4 nAChRs (Table 4).

Scheme 1.

Synthesis of novel biphenyl amides.

Table 4.

Inhibition of new antagonists on Hα4β2 and Hα3β4 nAChRs

		Hα4β2 nAChRs		Hα3β4 nAChRs
	R	IC50 Value (μM)a	nHb	IC50 Value (μM)a	nHb	Fmc
COB-170		7.5 (6.1–9.2)	−1.1	7.6 (6.4–9.0)	−1.8	1.0
COB-171		6.9 (5.8–8.2)	−1.3	10.7 (8.4–13.6)	−2.2	1.6
COB-172		13.9 (6.1–31.6)	−0.7	>100d	−0.4	>10
COB-173		13.6 (12.0–15.5)	−1.0	17.5 (10.7–28.4)	−0.9	1.3
COB-174		25.1 (19.9–31.7)	−1.5	39.6 (36.1–43.6)	−0.9	1.6
COB-175		10.7 (8.3–13.8)	−0.8	18.2 (9.9–33.5)	−1.2	1.8

^aValues represent geometric means (confidence limits), n = 3–7.

^bn_H, Hill coefficient.

^cFold difference in potency Hα3β4/Hα4β2

^dNo activity up to concentrations of 100 μM

Two molecules (COB-170 and COB-171) were found to be 2-fold more potent than lead molecule, KAB-18. The phenylmethyl analog (COB-170) had an IC₅₀ of 7.5 μM and 8.5 μM on Hα4β2 and Hα3β4 nAChRs, respectively (Table 4). The phenyl analog (COB-171) had an IC₅₀ of 6.9 μM and 10.7 μM on Hα4β2 and Hα3β4 nAChRs respectively (Table 4). The naphthylmethyl (COB-172) produced no change in potency on Hα4β2 nAChRs (IC₅₀ value, 13.9 μM, Table 4); but maintained selectivity for Hα4β2 nAChRs. The pyridinyl analog (COB-173) had an IC₅₀ value of 13.6 μM and 20.0 μM on Hα4β2 and Hα3β4 nAChRs, respectively (Table 4). The pyrimidyl analog (COB-174) had an IC₅₀ value of 25.1 μM and 39.6 μM on Hα4β2 and Hα3β4 nAChRs, respectively (Table 4). Finally the naphthyl analog (COB-175) had an IC₅₀ value of 10.7 μM and 18.2 μM on Hα4β2 and Hα3β4 nAChRs, respectively (Table 4). Concerning selectivity for Hα4β2 nAChRs over Hα3β4 nAChRs, COB-170, COB-171, COB-173, and COB-174 are all nonselective (Table 4). COB-175 showed a 2-fold preference for Hα4β2 nAChRs (Table 4). The methylnaphthyl analog (COB-172) is as potent and selective as lead molecule, KAB-18.

The Hα4β2 nAChR QSAR model shows that negative electronics contribute to potency near the piperidine moiety of Region 3 contribute to potency (Figure 2B). This is also supported by previous data that shows the addition of an amide group leads to an increase in potency.¹⁹ To determine the importance of this finding, several new molecules (IMB-132, IMB-133, IMB-134, IMB-135) containing amide groups in Region 1 were made from scaffolds that have been previously reported¹⁹ (KAB-18, DDR-14, KAB-24, COB-4) (Scheme 2).

Scheme 2.

Novel compounds containing amide functionality in Region 1 and previously reported scaffolds. Synthetic details have been previously described.^{13,16,20,21,22}

The new amide-containing molecules (IMB-132, IMB-133, IMB-134, IMB-135) were all more potent that their original scaffolds (DDR-14, KAB-18, KAB-24, COB-4) on Hα4β2 nAChRs (Table 5). IMB-132 resulted in a 1.5 fold increase in potency on Hα4β2 nAChRs (IC₅₀ value, 2.7 μM, Table 5). IMB-134 resulted in a 2.1 fold increase in potency on Hα4β2 nAChRs (IC₅₀ value, 3.9 μM, Table 5). IMB-135 resulted in a 1.4 increase in potency on Hα4β2 nAChRs (IC₅₀ value, 5.9 μM, Table 5). Finally, IMB-133 resulted in a 2.4 fold increase in potency on Hα4β2 nAChRs (IC₅₀ value, 5.7 μM, Table 5). IMB-132 showed a 2-fold decrease in potency on Hα3β4 nAChRs compared to its scaffold molecule, DDR-14 (IC₅₀ value 6.3 μM, Table 5). IMB-133, IMB-134, and IMB-135 showed no change in potency on Hα3β4 nAChRs compared to its scaffolds (KAB-18, KAB-24, COB-4, respectively).

Table 5.

Inhibition of new antagonists on Hα4β2 and Hα3β4 nAChRs

	Hα4β2 nAChRs		Hα3β4 nAChRs
	IC50 Value (μM)a	nH b	IC50 Value (μM)a	nH b	Fm c
IMB-132	4.3 (2.2–8.5)	−1.6	6.0 (4.3–8.3)	−1.6	1.5
DDR-14d	6.5 (4.1–10.4)	−1.6	2.7 (0.4–17.9)	−1.4
IMB-133	5.6 (3.6–9.0)	−0.9	>100e	~	2.4
KAB-18d	13.5 (9.7–18.5)	−1.4	>100e	~
IMB-134	3.9 (3.3–4.8)	−1.5	6.3 (5.4–7.3)	−2.1	2.1
KAB-24d	8.0 (4.2–15.3)	−0.8	5.5 (1.7–17.4)	−0.8
IMB-135	5.9 (5.2–6.7)	−1.2	9.4 (7.4–11.9)	−1.5	1.4
COB-4d	8.1 (2.1–30.7)	−0.8	10.5 (7.6–14.4)	−1.0

^aValues represent geometric means (confidence limits), n = 3–5.

^bn_H, Hill coefficient.

^cFold difference in potency Carbonyl/non-carbonyl

^dPreviously reported data¹⁹

^eNo activity up to concentrations of 100 μM

The QSAR and QSSR models presented here describe the physiochemical interactions that are important for potency on Hα4β2 and Hα3β4 nAChRs as well as selectivity for Hα4β2 nAChRs versus Hα3β4 nAChRs. The fact that these models agree with previously reported modeling and functional data support their strength and validity. With the models presented here combined with the information gathered from previous SAR, homology modeling, and site directed mutagenesis, there are many physiochemical features identified in this scaffold that mediate Hα4β2 nAChR selectivity. These include steric/hydrophobic features in both Region 1 and Region 2 as well as a hydrogen bond acceptor in Region 4. Previous SAR has highlighted the importance of aromatic rings in Region 1 and 2; but these models suggest that aromatics may not be necessary to preserve potency or selectivity on Hα4β2 nAChRs. If true, this allows for additional flexibility in the design of novel scaffolds. However, this will need to be confirmed by designing and synthesizing novel molecules containing distinct hydrophobic and steric features as opposed to aromatic features to determine how this will affect potency and selectivity. Previous SAR also points to the importance of the carbonyl group in Region 4. The QSAR and QSSR models suggest the importance of this position lies in the role as a hydrogen bond acceptor. This finding correlates with the docking studies that show a stable hydrogen bond between Thr58 of the β2 subunit and the carbonyl group in KAB-18’s Region 4.^19,26 There may be potential at this position for increasing selectivity by placing a feature that will enable stronger hydrogen bonding with Thr58 (β2 subunit). Most importantly, this finding suggests that maintaining an acceptor atom at this position is significant for both potency and selectivity and should be remain in the design of future molecules. The Hα4β2 nAChR QSAR model suggests that the following changes will increase potency on Hα4β2 nAChRs: 1) increasing electronegative character in Region 2’s anthranilic moiety, 2) increasing the electropositive character in the area surrounding Region 2’s phenyl ring, and 3) addition of a hydrogen bond donor in Region 2 and Region 4. This Hα4β2 QSAR model also suggests that a hydrogen bond donor in Region 3 is important for potency on Hα4β2 nAChRs. This agrees with previous modeling studies where the hydrogen of the piperidine nitrogen in Region 3 forms a stable hydrogen bond with Glu60 of the β2 subunit.^19,26

Using these QSAR and QSSR models, new molecules were synthesized. According to the QSAR and QSSR models, the new COB series were designed to: 1) enhance potency by reducing the unfavored sterics in Region 1 and 2) maintain selectivity by increasing favored hydrophobics in Region 3. The new IMB compounds were designed to increase potency by increasing the favorable electrostatics predicted in Region 1 and Region 3. Of the new COB molecules, two of the six showed a higher preference for Hα4β2 nAChRs over Hα3β4 nAChRs (Table 4). Two molecules (COB-170 and COB-171) were found to be 2-fold more potent than lead molecule, KAB-18. The naphthylmethyl (COB-172) maintained potency on Hα4β2 nAChRs and also maintained selectivity for Hα4β2 nAChRs (Table 4). This can be considered a significant improvement, as COB-172’s scaffold is more ‘drug like’ when compared to KAB-18 (Table S1); but is still selective for Hα4β2 nAChRs. The novel synthesis of the IMB compounds also shows strong evidence that inclusion of amide groups may improve potency to a small degree; but does so consistently. Altogether, this work presents novel QSAR and QSSR models for nAChRs. These models present important chemical features for a novel class of NAMS that promote both selectivity and potency on Hα4β2 nAChRs. Finally, these models have aided in the design and synthesis of potent, novel antagonists of Hα4β2 and Hα3β4 nAChRs as well as the discovery of a new, drug-like, selective antagonist of Hα4β2 nAChRs (COB-172).