Molecular Determinants of Subtype-selective Efficacies of Cytisine and the Novel Compound NS3861 at Heteromeric Nicotinic Acetylcholine Receptors

Background: The contribution of β-subunits to agonist efficacy in nicotinic receptors is incompletely understood.

Results: Two nicotinic agonists displayed opposing efficacy profiles at receptors containing β2- or β4-subunits and maximal efficacy was determined by both ligand-binding and transmembrane β-subunit domains.

Conclusion: The β-subunit is an important determinant of agonist efficacy.

Significance: These findings provide support to structure-guided nicotinic receptor drug design.

Abstract

Deciphering which specific agonist-receptor interactions affect efficacy levels is of high importance, because this will ultimately aid in designing selective drugs. The novel compound NS3861 and cytisine are agonists of nicotinic acetylcholine receptors (nAChRs) and both bind with high affinity to heteromeric α3β4 and α4β2 nAChRs. However, initial data revealed that the activation patterns of the two compounds show very distinct maximal efficacy readouts at various heteromeric nAChRs. To investigate the molecular determinants behind these observations, we performed in-depth patch clamp electrophysiological measurements of efficacy levels at heteromeric combinations of α3- and α4-, with β2- and β4-subunits, and various chimeric constructs thereof. Compared with cytisine, which selectively activates receptors containing β4- but not β2-subunits, NS3861 displays the opposite β-subunit preference and a complete lack of activation at α4-containing receptors. The maximal efficacy of NS3861 appeared solely dependent on the nature of the ligand-binding domain, whereas efficacy of cytisine was additionally affected by the nature of the β-subunit transmembrane domain. Molecular docking to nAChR subtype homology models suggests agonist specific interactions to two different residues on the complementary subunits as responsible for the β-subunit preference of both compounds. Furthermore, a principal subunit serine to threonine substitution may explain the lack of NS3861 activation at α4-containing receptors. In conclusion, our results are consistent with a hypothesis where agonist interactions with the principal subunit (α) primarily determine binding affinity, whereas interactions with key amino acids at the complementary subunit (β) affect agonist efficacy.

Introduction

Activation of the neuronal nicotinic acetylcholine receptors (nAChRs)³ is known to involve agonist binding in a site containing a set of highly conserved aromatic amino acid residues in the extracellular ligand-binding domain (LBD) (1). In heteromeric neuronal nAChRs, this agonist binding site has been shown to be located in the interface between an α- and a β-subunit, described as the principal and complementary binding site components, respectively (2). Consistent with these findings, functional studies have identified segments of the LBD of both α- and β-subunits as important determinants of agonist affinity (3, 4). A significant part of the binding site is made up by the flexible C-loop that changes spatial location upon agonist binding (5). This movement is subsequently translated into the pore opening by a not yet fully elucidated mechanism thought to involve rigid body rotation of the inner β-sheets, movements of loops in the interface between the LBD and the transmembrane domain (TMD), and a tilt of helix M2 (6).

Agonists can display strong subtype selectivity, primarily in terms of efficacy, for heteromeric neuronal nAChRs containing particular α- or β-subunits (7, 8), despite a very high sequence identity of the residues involved in agonist binding (9). By analogy to the findings that agonist efficacy in glutamate receptors appears tightly linked to domain closure (10), C-loop closure has been suggested to be directly coupled to agonist efficacy in Cys-loop receptors (11–13). However, in a recent study, it was demonstrated that a structurally related series of agonists with similar binding affinities but divergent efficacies all resulted in similar C-loop closure in co-crystallization experiments with acetylcholine-binding protein (AChBP) (14). Hence, it appears that neither the binding determinants nor the degree of C-loop closure are sufficient to explain how efficiently an agonist gates a receptor.

Despite having a large range of efficacies, the above mentioned series of partial agonists all contained the same core scaffold, which explains the similar binding affinities (14). The major differences were observed in the interactions with the complementary subunit. Based on this, it was hypothesized that agonist efficacy is dependent on the strength with which the agonist can mediate interactions between the principal and complementary subunits. From the current knowledge it is likely that this inter-subunit interaction is an integral first step in triggering the movements or stabilizing the conformation that eventually lead to channel gating, however, it also prompts new questions. The first question is whether the principal subunit only delivers the framework to which an agonist binds, whereas determinants important for ligand efficacy are located on the complementary subunit? Furthermore, irrespective of the exact mechanism, translation of binding into ion-channel gating involves transmission of conformational changes in the LBD to the pore-forming TMD. The next question is then, whether the LBD fully determines agonist efficacy and the TMD consequently acts purely as a mechanical channel-gating switch, i.e. would any TMD that interlinks well with a given LBD result in similar efficacy levels of specific agonists?

As part of a drug discovery program, novel nAChR agonists of various selectivity profiles were designed. One such agonist, NS3861, was found to selectively activate α3- but not α4-containing nAChRs and it displays higher efficacy at the α3β2 receptor compared with the α3β4 receptor (present report). To the best of our knowledge, such a selectivity pattern has not previously been reported. Interestingly, this selectivity profile reciprocates that of the nAChR agonist cytisine, which is known to selectively activate β4- but not β2-containing nAChRs (8, 15). These two compounds were therefore used as pharmacological tools to explore the relative contributions to maximum agonist efficacy of the principal (α) and complementary (β) subunits as well as the LBDs versus the TMDs. The studies entailed patch clamp electrophysiological measurements of heteromeric combinations of α3- and α4- with β2- and β4-subunits, and various chimeric constructs thereof. Receptor homology models and molecular docking were used to interpret the experimental data in terms of agonist binding modes and interactions to the binding site residues of the principal and complementary subunits.

In essence, we found the identity of the α-subunit to contribute to agonist efficacy in an “either/or” fashion. However, the determinants responsible for “fine tuning” agonist efficacy were particularly associated with the β-subunit LBD, although the β-subunit TMD was also observed to have some impact. Our findings from the molecular docking approach show that the observed subunit selectivities of NS3861 and cytisine can be attributed to a few residues in the binding sites that differ between α3- and α4- as well as β2- and β4-subunits.

EXPERIMENTAL PROCEDURES

Materials

(+/−)-NS3861 was synthesized at NeuroSearch A/S. (−)-Cytisine (C2899), (−)-nicotine (N5260), acetylcholine (ACh) (A9101), and dihydro-β-erythroidine hydrobromide (D149) were purchased from Sigma. All other chemicals were purchased from Sigma or Merck and were analytical grade. Molecular biology kits were from Qiagen (QIAquick columns, Maxiprep DNA, and Gel Extraction kits). T4 DNA ligase, T7 DNA polymerase, SalI, NotI, and NruI enzymes were from New England Biolabs. Oligonucleotides as well as sequencing services were from MWG Biotech. Escherichia coli strain RZ1032 was from Quantum Biotechnologies and strain XL1-Blue from Stratagene. The rat pituitary carcinoma cell line GH4Cl (CCL-82.2) and the human embryonic kidney 293 cell line HEK293 (CRL-1573) were from the American Type Culture Collection. Cell culture flasks were from Nunc, Dulbecco's modified Eagle's medium (DMEM) from Lonza (BE12–604/U1), Ham's F-10 medium from Invitrogen (31550-023), fetal bovine serum (FBS) from Invitrogen (10270-106), horse serum from Invitrogen (26050-088), poly-d-lysine from Sigma (P7405), Lipofectamine Plus^TM and trypsin/EDTA from Invitrogen, Geneticin G418 from Sigma (A1720), and Zeocin from Invitrogen (450430). The calcium indicator fluo-4/AM and [³H]epibatidine (55 Ci/mmol) were from Invitrogen and PerkinElmer Life Sciences, respectively. Borosilicate capillary tubes were from Vitrex (155710).

Molecular Biology

Cloning of the human nAChRs α3-, α4-, β2-, and β4-subunits was as described previously (16). nAChR α-subunits and α-chimeras were in the pNS3n vector, whereas nAChR β-subunits and β-chimeras were in the pNS3z vector. pNS3n and pNS3z are custom designed vectors derived from pcDNA3 (Invitrogen) where “n” and “z” denotes selection using the NeoR or ZeoR genes, respectively. Both vectors use the CMV promoter to drive expression of the insert in mammalian cells. To generate α3/α4-, α4/α3-, β2/β4-, and β4/β2-chimeras a unique SalI restriction site was introduced close to the beginning of TM1 in each subunit by site-directed mutagenesis as described by Slilaty et al. (17). Briefly, uracilated plasmids containing nAChR subunits were isolated from E. coli strain RZ1032, linearized with NruI, and used as template in mutagenesis reactions using a mutagenesis oligo, a closing oligo, T7 DNA polymerase, and T4 DNA ligase. Sequences of the mutagenesis oligonucleotides were: α3-SalI, CTCACTGTACATCCGTCGACTGCCCTTGTTCTACAC; α4-SalI, GCCTTCGTCATCCGTCGACTGCCGCTCTTCTACAC; β2-SalI, GACTTCA-TCATTCGTCGACGGCCGCTCTTCTACAC; and β4-SalI, GACTTCATCATCCGTCGACGGCCTCTGTTCTACAC. Correct introduction of SalI sites were verified by restriction enzyme digestion and further sequencing. Chimeric nAChRs were made by restriction digestion of the plasmids with SalI and NotI (polylinker site), gel electrophoresis purification of fragments, and re-ligation of relevant nAChR combinations. The resulting chimeric constructs were verified by restriction enzyme digestion and sequencing. To obtain an α4-subunit with a Thr¹⁸³ to Ser mutation as well as a β2-subunit with Val¹³⁶ to Ile and Phe¹⁴⁴ to Leu mutations, α4- and β2- LBDs, including the mutations, were purchased from GenScript and ligated with the respective TMDs using the SalI site.

Cell Culture and Transfection

HEK293 cells were propagated at 37 °C in culture flasks in a humidified atmosphere containing 5% CO₂. The growth medium consisted of DMEM supplemented with 10% FBS. For generation of clones stably expressing α3β2, α3β4, α4β2, and α4β4 nAChRs, HEK293 cells were seeded in T12.5 culture flasks, cultured to 50–70% confluence, and transfected with a total of 1 μg of expression plasmids using Lipofectamine Plus^TM according to the manufacturer's protocol. Twenty-four hours post-transfection, the cells were detached using trypsin/EDTA and seeded in T75 culture flasks with a tear-off lid. To select stably expressing cells the culture medium was supplemented with 0.5 mg/ml of G418 and 0.125 mg/ml of Zeocin. Following selection, single clones were picked and propagated in selection media until sufficient cells for freezing were available. Thereafter the cells were propagated in regular culture media.

GH4Cl cells were cultured in tissue culture flasks at 37 °C in a humidified atmosphere containing 5% CO₂. The culture medium consisted of Ham's F-10 medium supplemented with 15% horse serum and 2.5% FBS. On the day before transfection, cells were seeded in 35-mm Petri dishes containing poly-d-lysine-coated round glass coverslips (Ø 3.5 mm, custom made at VWR International). The GH4Cl cells were then transfected using Lipofectamine Plus^TM according to manufacturer's protocol. Each Petri dish was transfected with a mixture of plasmids containing cDNAs coding for relevant nAChRs and green fluorescent protein (Clontech); the latter to identify transfected cells in the patch clamp set-up. After transfection, the culture medium was supplemented with 50 mm KCl, as elevated K⁺ has previously been shown to facilitate expression of nAChRs in the GH4Cl cell line (18).

Xenopus laevis oocyte preparation and electrophysiological experiments were performed as described previously (19). Briefly, to obtain isolated oocytes lobes from ovaries of female adult X. laevis were removed and defolliculated using collagenase. Oocytes were injected with a total of ∼25 ng of cRNA encoding human wild-type or concatenated subunits and incubated for 2–7 days at 15–18 °C in modified Barth's solution (90 mm NaCl, 1.0 mm KCl, 0.66 mm NaNO₃, 2.4 mm NaHCO₃, 10 mm HEPES, 2.5 mm sodium pyruvate, 0.74 mm CaCl₂, 0.82 mm MgCl₂, 100 μg/ml of gentamycin, and pH adjusted to 7.5).

Radioligand Binding Experiments

HEK293 cells stably expressing human α3β2, α3β4, α4β2, and α4β4 receptors were harvested, washed once with 50 mm Tris-HCl (pH 7.4), and stored at −80 °C until the day of experiment. Thawed membrane pellets were re-suspended in 15 ml of ice-cold Tris-HCl buffer and centrifuged for 10 min (27,000 × g) at 4 °C. The final pellets were re-suspended in Tris-HCl buffer and used for binding experiments. The conditions for receptor binding assays were as described previously for [³H]epibatidine binding to rat brain (20). Briefly, binding to α3β4 receptors was performed using 0.3 nm [³H]epibatidine at 14 to 27 μg of protein/assay, whereas α3β2, α4β2, and α4β4 receptors were labeled with 0.03 nm [³H]epibatidine at 32 to 300 μg of protein/assay. The samples were incubated in a final volume of 1050 μl (for α3β4) and 8400 μl for 1 and 4 h, respectively, at room temperature. Nonspecific binding was determined in the presence of 30 μm (−)-nicotine, and binding was terminated by rapid filtration. Radioactivity was determined by conventional liquid scintillation counting. Compounds were tested at 8 concentrations ranging from 0.03 nm to 30 μm. Estimates of IC₅₀ values in binding experiments were analyzed using the nonlinear curve-fitting program GraphPad Prism, and K_i values were calculated from IC₅₀ values using the Cheng and Prusoff equation: K_i = IC₅₀/(1/(liter/K_d)). The K_d values for each receptor type are shown in Table 3.

TABLE 3.

Radioligand displacement by cytisine, NS3861, and nicotine at α3β2, α3β4, α4β2, and α4β4 nAChRs

Radioligand binding was determined as the potency for displacement of [³H]epibatidine-binding in membrane preparations from HEK293 cell lines stably expressing human nAChRs. Saturation data were fitted using a hyperbolic function assuming a single binding site for [³H]epibatidine (K_d values). K_i values are reported as mean ± S.E.M. in nm, n = 3–4 determinations, each conducted in triplicate.

Compound	Ki
	α3β2	α3β4	α4β2	α4β4
Epibatidine Kd	0.025	0.082	0.012	0.022
Cytisine	8.7 ± 1.3	196 ± 19	0.43 ± 0.09	0.77 ± 0.09
NS3861	25 ± 4	0.62 ± 0.09	55 ± 11	7.8 ± 0.7
Nicotine	23 ± 2	250 ± 12	1.7 ± 0.2	13 ± 1

Patch Clamp Electrophysiology

Electrophysiological measurements in GH4Cl cells were performed in voltage clamp using conventional whole cell patch clamp techniques. All data were obtained with an EPC-9 amplifier (HEKA). The holding potential was −60 mV in all experiments. Coverslips with cells were placed in a diamond-shaped polycarbonate recording chamber (Warner Instruments) fixed at the stage of an inverted microscope (Olympus). Throughout the experiment, the cells were perfused with extracellular buffer (140 mm NaCl, 4 mm KCl, 2 mm CaCl₂, 1 mm MgCl₂, and 10 mm HEPES, pH adjusted to 7.4 using NaOH). Micropipettes were made from borosilicate capillary tubes by means of a horizontal micropipette puller (Zeitz Instruments). The pipettes were filled with intracellular buffer (120 mm potassium gluconate, 6 mm KCl, 5 mm NaCl, 2 mm MgCl₂, 10 mm HEPES, 0.5 mm EGTA, 2 mm ATP, and 0.2 mm GTP (ATP and GTP were added immediately before use), pH adjusted to 7.4 using KOH). Initial pipette resistance was ∼2 MΩ. Agonist/antagonists were delivered with an ultra-fast application system using a double barreled application pipette (the so-called θ-tube) controlled via a piezo-ceramic device (Burleigh Instruments) as described previously (16). Data were sampled at 20 kHz, low-pass filtered at 6.7 kHz and only accepted if the series resistance was <10 MΩ. Series resistance was compensated by 80%. Agonist application pulses lasted 1 s and the stimulation frequency was 1 pulse per 30 s, to ensure full recovery of the nAChRs from agonist-induced desensitization between pulses. The responses of the patch clamp electrophysiological experiments were quantified by measuring peak current amplitude and relating this to 1 mm ACh control responses. Concentration-response curves were analyzed using GraphPad Prism. Figures displaying current traces were made using SigmaPlot.

X. laevis Oocyte Electrophysiology

Oocytes were subjected to two-electrode voltage-clamp electrophysiological testing using a custom-built system as described previously (19). Drugs were dissolved in OR2 (90 mm NaCl, 2.5 mm KCl, 2.5 mm CaCl₂, 1.0 mm MgCl₂, 5.0 mm HEPES, and pH adjusted to 7.5), and solutions were applied directly to the oocytes via a glass capillary tube placed in the vicinity of the cell to ensure rapid solution exchange (few seconds). Each application lasted ∼1 min and peak current amplitudes were measured. To ensure full recovery between applications at high agonist concentrations, the protocol included dynamic adjustments of washing periods.

Homology Modeling and Induced-fit Docking

Homology models of the α3β2, α3β4, and α4β4 dimer interfaces were built as previously described for our α4β2 homology model (21) using the same template structures. The amino acid sequences of the other three subunit combinations were aligned to the templates by use of the alignment for the α4β2 homology model and substitution of the α4 sequence with that of α3 and/or the β2 sequence with that of β4 using T-Coffee, version 9.01 (22). The amino acid sequences of the LBDs of α3, α4, β2, and β4 (Protein Knowledgebase entries P32297, P43681, P17787, and P30926 from www.uniprot.org (23)) and residue numbers are listed accordingly. The rotamer library in PyMOL (24) was used to select side chain conformations of Val¹³⁶, Phe¹⁴⁴, and Leu¹⁴⁶ in β2 as well as of Ile¹³⁴, Leu¹⁴², and Leu¹⁴⁴ in β4, which point away from the agonist binding site to allow maximal available space for ligand docking. Subsequently, the protein structures were prepared for docking with the Protein Preparation Wizard workflow (25) using default settings. (+/−)NS3861 and (−)-cytisine were built in Maestro (26) and energy minimized using MacroModel (27) with default settings. Both compounds were used as input for induced fit docking (28) in each of the four homology models using the centroid of (−)-nicotine included in the models to define the box center. Ligand and receptor van der Waals scaling was set to 0.8 and 1.0, respectively, maximum number of poses to 10, only the residues for which we previously altered conformations were optimized during refinement and otherwise default settings were used. The best scoring of each receptor-ligand combination according to IFDScore were selected for analysis.

RESULTS

Activation patterns of cytisine and the novel compound NS3861 (Fig. 1) were investigated to address the relative contribution(s) to maximal agonist efficacy of the LBD and the TMD in heteromeric neuronal nAChRs. The compounds were tested at various combinations of wild-type subunits and subunit chimeras, which were generated by fusing the LBD of one subunit to the TMD of another subunit (denoted “LDB/TMD”, i.e. α3/α4, α4/α3, β2/β4, and β4/β2). Because agonist properties are known to vary between heterologous expression systems (29), it was chosen to address agonist efficacy using an ultra-fast solution exchange methodology in whole cell patch clamp analysis of receptors transiently expressed in the rat GH4C1 cell line.

FIGURE 1.

Chemical structures of (−)-cytisine and (+/−)-NS3861.

Current densities for cells expressing chimeric subunits were, on average, less than for the comparable wild-type subunit combinations when the LBD contained α3, whereas receptors with α4 expressed well (Table 1). Lower expression levels of chimeric constructs versus wild-type can possibly reflect reduced efficiency of chimeric subunit assembly. However, in all the receptor combinations used here, current densities were more than sufficient to ensure reliable observations of partial agonists.

TABLE 1.

Time constants and current densities of wild-type and chimeric nAChRs

Patch clamp data were obtained as described in the legend to Fig. 2. Time constants for current-decay of 1 mm ACh-evoked responses were calculated using a one-phase exponential decay nonlinear regression curve fitting routine. Data used for fitting were from the peak recorded value and 200 ms forward in time. The current densities are calculated as peak current of 1 mm ACh (pA) divided by the cell capacitance (pF) for n = 10 experiments and values are presented ± S.E.M.

Receptor	Tau	Current density
	ms	pA/pF
α3 + β2	35 ± 2	584 ± 220
α3 + β4	273 ± 15	761 ± 278
α3 + β2/β4	106 ± 7	163 ± 59
α3 + β4/β2	102 ± 7	191 ± 37
α3/α4 + β2	51 ± 3	111 ± 39
α3/α4 + β4	568 ± 28	429 ± 103
α3/α4 + β2/β4	90 ± 5	125 ± 20
α3/α4 + β4/β2	93 ± 6	70 ± 15
α4 + β2	68 ± 4	1236 ± 372
α4 + β4	213 ± 12	345 ± 76
α4 + β2/β4	92 ± 7	407 ± 133
α4 + β4/β2	71 ± 4	183 ± 46
α4/α3 + β2	56 ± 2	1350 ± 429
α4/α3 + β4	60 ± 3	565 ± 138
α4/α3 + β2/β4	71 ± 3	689 ± 212
α4/α3 + β4/β2	87 ± 5	673 ± 213

A prerequisite for using LBD/TMD chimeric constructs for evaluating agonist efficacy is conservation of basic biophysical properties such as expression level, kinetic properties, and agonist or antagonist sensitivity relative to wild-type receptor counterparts. To validate that the constructs used in this study adhere to this premise, two “fully” chimeric subunit combinations (α3/α4 + β4/β2 and α4/α3 + β2/β4) were thoroughly evaluated initially and compared with wild-type heteromeric receptors.

Characterization of Wild-type and Chimeric nAChR Constructs

The potency of ACh for activating wild-type α3β4 and α4β2 was in the range of ∼80 μm (Fig. 2), which is comparable with previous studies (15, 30–32). This shows that the α4β2 nAChRs predominantly assembled into the (α4)₃(β2)₂ stoichiometry in these experiments (21). Expression of the fully chimeric subunit combinations α4/α3 + β2/β4 and α3/α4 + β4/β2 yielded functional receptors with potencies of ACh in the same range as for the wild-type receptors (Fig. 2), indicating normal responsiveness of chimeric receptors to ACh.

FIGURE 2.

CRRs for ACh and DHβE at α3β4 and α4β2 nAChRs as well as chimeras thereof. Functional receptor activation was measured in whole cell patch clamp experiments using transiently transfected GH4C1 cells in a setup equipped with an ultra-fast application system. Agonist was applied for 1 s and peak current amplitudes were baseline subtracted and normalized to a 1 mm ACh response recorded in the same cell. A, ACh CRRs obtained in patch clamp experiments at α3β4, α4β2, α3/α4 + β4/β2, and α4/α3 + β2/β4 nAChRs (symbol identification as in panel C). Data points are plotted as mean ± S.D. of n = 4–12 experiments as a function of the ACh concentration and fitted to the Hill equation by nonlinear regression. The data point for 3160 μm ACh at α3β4 was omitted from the fit as this yielded a low submaximal response. B, DHβE CRRs obtained in patch clamp experiments at α3β4, α4β2, α3/α4 + β4/β2, and α4/α3 + β2/β4 nAChRs (symbol identification as in panel C). Experiments were largely conducted as described above except cells were preincubated with DHβE for 30 s before a co-application of 1 mm ACh and DHβE. Data points are plotted as mean ± S.D. of n = 5–9 experiments as a function of the DHβE concentration and fitted to the Hill equation by nonlinear regression. C, table of ACh agonist potencies (EC₅₀ values) and DHβE inhibitor potencies (IC₅₀ values) at the four receptor subtypes. All EC₅₀ values have overlapping 95% confidence intervals, whereas pairwise overlap was observed for IC₅₀ values.

When evaluating efficacies of agonists at heteromeric nAChRs, a further complication relates to the fact that these can assemble in different stoichiometries of α- and β-subunits (30, 31, 33). Because recent results have shown that acetylcholine and other agonists can activate an additional binding site in the α4α4 interface (21), it is always necessary to consider whether agonists have stoichiometry dependent activation patterns.

ACh concentration-response relationships (CRR) for the fully chimeric and wild-type receptors were therefore evaluated in X. laevis oocytes. Under conditions favoring the 3α:2β stoichiometry both wild-type and chimeric receptors with α4 in the LBD displayed ACh potencies in the same range as obtained in patch clamp, whereas conditions favoring 2α:3β resulted in lower EC₅₀ values, in particular for α4β2 (Table 2). Receptors containing the α3-LBD in 3α:2β stoichiometries appeared slightly (∼3-fold) less potent compared with the patch clamp experiments, but all data for wild-type receptors are in good agreement with previously reported results (15, 30–32).

TABLE 2.

Concentration-response relationships (CRRs) for ACh, cytisine, and NS3861 recorded in X. laevis oocytes

Oocytes were injected with nAChR subunits in 1:4 or 4:1 ratios as indicated to obtain receptors in (α*)₂(β*)₃ or (α*)₃(β*)₂ stoichiometries, respectively. CRRs for cytisine and NS3861 were recorded paired with an ACh-CRR, i.e. two 5-point CRRs for each oocyte, the first ACh and the second cytisine or NS3861. Peak current amplitudes were baseline subtracted and normalized to the maximal value obtained by fitting ACh-CRRs from each oocyte to the Hill equation by non-linear regression. Normalized values were then fitted to the Hill equation to establish maximal efficacy (E_max) and concentration of half maximal activation (EC₅₀). Data are presented as E_max in %, and EC₅₀ in μm from n = 6–14 experiments.

Receptor	α:β ratio	ACh		Cytisine		NS3861
		Emax	EC50	Emax	EC50	Emax	EC50
		%	μm	%	μm	%	μm
α3β2	4:1	100	290	23	280	160	1.6
α3β4	4:1	97	330	48	180	65	1
α3/α4 + β4/β2	4:1	100	350	35	150	71	1.4
α4β2	1:4	100	1.2	NEa		NE
α4β2	4:1	98	93	NE		NE
α4β4	1:4	100	20	37	0.12	17	1.9
α4β4	4:1	99	71	22	0.72	6.3	1.2
α4/α3 + β2/β4	4:1	98	100	3.8	13	9.0	0.64

^a NE, no effect.

Kinetic properties of ACh-activated currents at wild-type receptors containing the β2-subunit clearly displayed faster desensitization kinetics (Tau: ∼50 ms) relative to those containing β4-subunits (Tau: ∼250 ms) (Fig. 3 and Table 1). A more slowly desensitizing/steady-state current, particularly in β2-containing receptors often followed initial rapid desensitization. These findings are in good support of earlier studies on nAChR kinetic properties (7, 34–36). With respect to desensitization kinetics for receptors containing chimeric subunits, all turned out to have desensitization time constants intermediate between those of wild-type receptors, except α3/α4 + β4 for which the value was slightly (∼2-fold) higher.

FIGURE 3.

Representative traces of ACh-, cytisine-, and NS3861-evoked currents in patch-clamp experiments at the indicated nAChRs. Data were obtained as described in the legend to Fig. 2 and agonist application is shown as a solid bar above all current traces. The cytisine or NS3861 concentration giving rise to maximal efficacy at each receptor combination was identified by testing a range of concentrations. Traces shown are from compound concentrations that yielded maximal responses. Note that in many cases, the maximal response was not obtained with the highest concentration tested (see e.g. 100 μm NS3861 at α3β2 and 10 μm NS3861 at α3β4 in Fig. 4B). There is no obvious technical reason for this but from previous similar observations it was speculated that it could be due to nonspecific ion-channel blockage at high ligand concentrations (46). A–P, cytisine-evoked currents shown alongside 1 mm ACh-evoked currents recorded in the same cells. Cytisine concentrations were 100 μm in traces F, J, L, N, and P, 316 μm in trace B, and 1 mm in the remaining traces (from 5,000 to 100,000 times the highest observed binding K_i value). Q-AA, NS3861-evoked currents shown alongside 1 mm ACh-evoked currents recorded in the same cells. NS3861 concentrations were 3.16 μm in traces R and S, 10 μm in traces Q, T, U, V, W, and X, and 100 μm in the remaining traces (from 2,000 to 5,000 times the highest observed binding K_i value).

Finally, conservation of basal pharmacological properties of the orthosteric site in the fully chimeric receptors was tested by determining the ability of the β2-selective competitive antagonist DHβE to block wild-type and fully chimeric receptors. The amino acid determinants responsible for binding of DHβE are known to be distributed across both α- and β-subunits (37–39). The potency of DHβE for inhibition of α4β2 exceeded that for inhibition of α3β4 by more than 3 orders of magnitude (Fig. 2). Replacing the TMD of α4β2 with that of α3β4 (i.e. α4/α3 + β2/β4) did not affect high affinity DHβE inhibition, whereas substitution of the LBD of α4β2 with α3β4 (i.e. α3/α4 + β4/β2) reduced DHβE potency to the level of the wild-type α3β4 nAChR (Fig. 2). This shows that the antagonist potency of DHβE is fully defined by the LBD and that the structural integrity of the LBD is preserved in the chimeric receptors.

Based on these data, it is highly likely that all combinations of wild-type and chimeric constructs result in receptors, which express reliably in GH4C1 cells. Comparisons of ACh EC₅₀ values, desensitization time constants, and DHβE IC₅₀ values for the fully chimeric receptors show that basic biophysical and ligand-binding properties were well preserved and generally match those of wild-type receptors.

Cytisine

Initially, cytisine was characterized with respect to its binding affinity for heteromeric nAChRs in radioligand displacement experiments, using membrane preparations from HEK293 cells stably expressing human α3β2, α3β4, α4β2, or α4β4 nAChRs. Cytisine was observed to displace [³H]epibatidine from α4β2 and α4β4 nAChRs with subnanomolar affinity (Table 3). Approximately 20-fold lower affinity was observed at the α3β2 receptors and a further ∼20-fold affinity reduction was seen at α3β4 receptors.

Using patch clamp electrophysiology, the CRR of cytisine was subsequently determined at wild-type α3β4 and α4β2 nAChRs as well as fully chimeric receptors having the same LBD interfaces. Cytisine behaved as a low potency full agonist at the wild-type α3β4 nAChR (Figs. 3B and 4) but produced no significant activation of the α4β2 nAChR (E_max < 10%) at concentrations up to 1 mm (Figs. 3I and 4). At the fully chimeric receptors, cytisine had low efficacy at α3/α4 + β4/β2 with potency roughly matching that at α3β4 receptors (Figs. 3H and 4), whereas no appreciable efficacy was seen at α4/α3 + β2/β4 receptors (Figs. 3O and 4). Judging from the appearance of current traces for α3β4 and α4β4 (Fig. 3, B and J), desensitization kinetics of cytisine resembled that of ACh. These results corroborate previous reports suggesting that cytisine binds potently to α4β2 but primarily activates β4-containing receptors (7, 8, 15, 40).

FIGURE 4.

CRRs for cytisine and NS3861 at wild-type and chimeric nAChRs. Patch clamp data were obtained as described in the legend to Fig. 2. A, cytisine is a full agonist at α3β4 and a partial agonist at α3/α4 + β4/β2 nAChRs (symbol identification as in panel C). Data points are plotted as mean ± S.D. of n = 3–11 experiments as a function of the cytisine concentration and fitted to the Hill equation by nonlinear regression. B, NS3861 is a full agonist at α3β2 and a partial agonist at α3β4 and α3/α4 + β4/β2 nAChRs (symbol identification as in panel C). Data points are plotted as mean ± S.D. of n = 3–10 cells as a function of the NS3861 concentration and fitted to the Hill equation by nonlinear regression. The data points for 100 μm NS3861 on α3β2 and 10 μm NS3861 on α3β4 were omitted from the fitting routine as these concentrations yielded low submaximal responses. C, table of fitted maximal efficacy (E_max) and half-maximal activation concentration (EC₅₀) at the five receptor subtypes. − indicates E_max < 10% set as minimum efficacy for reliable observations and fitting. EC₅₀ values for the two cytisine as well as the three NS3861 fitted graphs have overlapping 95% confidence intervals, respectively.

To further probe the importance of different LBD and TMD combinations, maximal efficacy of cytisine was determined at the remaining permutations of wild-type and chimeric constructs (Fig. 3, A, C–G, K–N, and P). Combinations between any α-subunit (chimeric or wild-type) with wild-type β4 yielded high maximum cytisine efficacy (E_max = 76–113%), whereas all combinations with a wild-type β2-subunit resulted in an almost complete loss (E_max < 10%) of efficacy (Fig. 5A). Replacing the TMD of β4 with that of β2 gave a substantial, but not complete, loss of efficacy (E_max ∼13–30% for α3, α4, α3/α4, or α4/α3 in combination with β4/β2). On the other hand, irrespective of the α-subunit, replacing the TMD of β2 with that of β4 did not increase the efficacy levels above that observed for receptors containing the wild-type β2-subunit (E_max < 10%).

FIGURE 5.

Maximal efficacies of cytisine and NS3861 at wild-type and chimeric nAChRs. Patch clamp data for cytisine and NS3861 were obtained as described in the legends to Figs. 2–4 and maximal compound efficacy values are reported from the most efficacious concentrations as described in the legend to Fig. 3. The horizontal dotted lines indicate the 10% level set as minimum for reliable efficacy detection. A, cytisine is a full agonist at receptors containing full-length β4-subunits, a partial agonist with β4/β2-subunits, and has no efficacy at receptors containing a β2-LBD. B, NS3861 is a full agonist at receptors containing the LBD of α3β2, a partial agonist with the LBD of α3/β4, and has no efficacy at α4-containing receptors. Values are expressed as mean ± S.D. of n = 3–14 cells for each subunit combination.

Finally, to investigate whether subunit stoichiometry of α4-containing receptors could affect the observed maximal efficacies of cytisine, CRRs were recorded on nAChRs expressed in oocytes (Table 2). These experiments qualitatively confirm the subtype-selective efficacy of cytisine for β4-containing receptors observed from patch clamp studies in combinations with α4. Independent of the subunit ratio, cytisine did not activate α4β2 receptors but activated α4β4 receptors with submicromolar potencies (Table 2).

NS3861

Characterization of NS3861 with respect to [³H]epibatidine displacement binding showed this compound to bind potently at all tested nAChR subtypes (Table 3). NS3861 was ∼10-fold affinity selective for α3β4, relative to the α4β4 nAChR and in both cases a ∼10–40-fold affinity reduction was observed by substituting β4 with β2. Thus, in sharp contrast to cytisine and nicotine, NS3861 has higher binding affinities at β4- versus β2-containing receptors.

By analogy to the experiments reported for cytisine, the CRR of NS3861 at α3β2, α3β4, and α4β2, and two fully chimeric receptors was investigated using patch clamp electrophysiology. NS3861 was found to activate wild-type α3β4 nAChRs (Figs. 3R and 4) but did not activate wild-type α4β2 (Figs. 3Y and 4). Interestingly, whereas NS3861 was a partial agonist at α3β4, it proved to be a full agonist at α3β2 albeit with ∼10-fold lower potency consistent with the binding affinities (Figs. 3Q and 4). Efficacy at the α3/α4 + β4/β2 receptor was almost identical to that of wild-type α3β4 (Figs. 3X and 4), whereas no efficacy could be seen at α4/α3 + β2/β4 (Figs. 3AA and 4). The activation and desensitization properties of currents evoked by NS3861 were generally reminiscent of the currents evoked by ACh, as judged from the appearance of current traces (Fig. 3, Q and R).

Maximal efficacy of NS3861 was next investigated in experiments involving the remaining permutations of wild-type and chimeric constructs (Fig. 3, S–W and Z). Unlike the efficacy profile of cytisine, which was independent of the α-subunit, NS3861 was only capable of activating subunit combinations containing the LBD of α3, as no appreciable efficacy (E_max < 10%) was detected in subunit combinations involving the LBD of α4 (Fig. 5B). However, substituting wild-type α3 for the α3/α4 chimera resulted in indistinguishable maximum efficacies. It is further evident, that in combination with an α3-LBD the identity of the β-subunit is highly important for efficacy. In these experiments, NS3861 acted as a full agonist whenever a β2-LBD was present (E_max = 97–129%), whereas it was only a partial agonist when a β4-LBD was present (E_max = 30–55%). By analogy to the findings made for the α-subunit, the β-subunit influence on efficacy was also confined to the extracellular LBD.

The subtype-selective efficacy of NS3861 was qualitatively reproduced in oocytes (Table 2). Independent of subunit ratios, NS3861 displayed no efficacy at α4β2 receptors. α3β2 and α3β4 receptors expressed in 3α:2β stoichiometries were, however, activated with maximal efficacies significantly above or partial relative to ACh, respectively. Because both combinations contain an α3α3 interface the observed β2 subtype-selective efficacy appears independent of this interface.

Binding Modes of Cytisine and NS3861 in nAChR Homology Models

To obtain possible binding modes for cytisine and NS3861 the compounds were docked into homology models of α3β2, α3β4, α4β2, and α4β4 using an induced-fit protocol. Thus, side chain conformations of selected residues are sampled during the docking calculation with the ligand present in the binding site and different ligand-receptor complexes are given as output. The best scoring complexes of cytisine in each receptor model display the same binding mode matching the interactions of classical nicotinic agonists (5, 41), with hydrogen bonds to the backbone carbonyl of a tryptophan and a water molecule (Fig. 6B). Ile¹³⁴ and Leu¹⁴² in β4 adopt side chain conformations, which display better van der Waals interactions to cytisine compared with the corresponding Val¹³⁶ and Phe¹⁴⁴ of β2; the only residues in the binding site that differ between the β2- and β4-subunits (Fig. 6A).

FIGURE 6.

Comparison of binding site residues and binding modes of cytisine and NS3861 as obtained from induced-fit docking in the nAChR homology models. A, sequence alignment of binding site residues (on red background) in α3-, α4-, β2-, and β4-subunits as defined by residues within 5 Å of the agonist included in the models and with side chains pointing toward it. B, comparison of obtained binding modes of cytisine in α4β2 (green carbons) and α4β4 (cyan carbons). C, comparison of obtained binding modes of NS3861 in α3β2 (purple carbons) and α3β4 (magenta carbons). D, comparison of the obtained binding modes of NS3861 in α3β2 (purple carbons) and α4β2 (orange carbons). The overall structure of the homology models and residues that are identical in the compared models are represented as a white cartoon and sticks, respectively. Yellow dashed lines indicate hydrogen bonds and red dashed lines indicate steric clashes. The water molecule is represented as a small red sphere.

Contrary to cytisine, the obtained binding modes of NS3861 in all four homology models only display the hydrogen bond to the tryptophan and has the thiophene ring located perpendicular to the pyridine of, for example, nicotine (41), with the bromine atom located near Phe¹⁴⁴ and Leu¹⁴² of β2 and β4, respectively (Fig. 6C). Furthermore, because of Thr¹⁸³ in α4 compared with Ser¹⁸¹ in α3 the binding modes of NS3861 obtained in α4-containing receptors display a ∼1 Å shift of the thiophene ring toward the β-subunit as measured on the sulfur and bromine atoms. This brings the bromine atom within 3.4 Å of one of the sulfur atoms in the disulfide bond of the C-loop and within 3.1 Å of one of the methyl groups of Leu^144/146 in the β-subunits (Fig. 6D).

Agonistic Behavior of NS3861 and Cytisine on Point Mutated Receptors

To investigate if Ile¹³⁴ and Leu¹⁴² of β4 were responsible for the selective cytisine activation of β4- versus β2-containing receptors, a mutant β2-subunit was constructed with two point mutations substituting the corresponding Val¹³⁶ and Phe¹⁴⁴ with Ile and Leu. Cytisine was then tested at α4β2(V136I/F144L) receptors expressed in oocytes. Here it displayed clear agonistic behavior resulting in activation of the mutant receptor with an efficacy level (E_max = 18%; n = 9; 95% confidence interval, 16–19%) comparable with what is observed for α4β4 receptors in this system (Table 2). Likewise, NS3861 was applied to α4(T183S)β2 mutant receptors in oocytes to test if Thr¹⁸³ was responsible for the lack of activation of α4-containing receptors. Indeed, NS3861 activates the mutant receptor acting as a partial agonist (E_max = 36%; n = 9; 95% confidence interval, 33–40%). Hence, the oocyte experiments qualitatively demonstrate the importance of Ile¹³⁴ and Leu¹⁴² in β4 as well as Ser¹⁸¹ in α3 for agonistic behavior of cytisine and NS3861, respectively.

DISCUSSION

Early studies have shown that both α- and β-subunits in nAChRs take part in defining the pharmacology of agonists at the macroscopic level (7, 8). Recently, it was shown that at the microscopic level very specific agonist-receptor interactions at the complementary subunit determine efficacy of a series of compounds (14). The present report extends this knowledge to functional studies of two agonists (cytisine and NS3861) with particularly pronounced profiles of efficacy selectivity (in the following the term “efficacy selectivity” is used to describe agonist subtype-selective efficacies). Furthermore, it was investigated whether efficacy is solely determined by the LBD or whether the TMD also plays a role.

Despite a high binding selectivity for β2- versus β4-containing nAChRs, our patch clamp data show that cytisine is a full agonist on β4-containing receptors and at best a very weak partial agonist when β2 is present, in line with previous reports (7, 8, 15). Furthermore, even though binding to α4β4 occurs with ∼250-fold higher affinity compared with α3β4, the efficacy selectivity of cytisine is independent of the identity of the α-subunit. Importantly, cytisine displayed partial efficacy resulting in peak currents up to ∼30% at receptors containing the β4/β2-chimera clearly indicating a role of the TMD in determining maximal efficacy of this compound.

Maximal efficacy of agonists at α4β2 receptors have previously been shown to be dependent on receptor stoichiometry (21). However, concentration-response data obtained in oocytes expressing uniform populations of either (α4)₃(β2)₂ or (α4)₂(β2)₃ receptors lead to the conclusion that the presence of an α4α4 interface has no or very limited effect on cytisine receptor activation (Table 2).

In agreement with our previous observations on determinants of efficacy at nAChRs (14), induced-fit docking of cytisine to homology models of the α4β2 and α4β4 interfaces suggests a larger van der Waals contact area and thus better interactions to the β4- versus β2-subunit as an explanation for the β4-selective activation. The only differences in the agonist binding site between β2- and β4-subunits are Val¹³⁶ and Phe¹⁴⁴ versus Ile¹³⁴ and Leu¹⁴², respectively (Fig. 6A). The longer side chain of Ile versus Val and the nonplanar nature of Leu versus Phe allow a larger van der Waals contact surface between the complementary subunit and cytisine in β4- versus β2-containing receptors (Fig. 6B). Because cytisine is relatively short compared with many other agonists, e.g. NS3861 and epibatidine, it may be imperative specifically for cytisine that the binding site residues of the complementary subunit are long, flexible and hydrophobic to make intimate van der Waals contacts and result in receptor activation. This hypothesis is supported by oocyte data in a receptor containing a mutant β2-subunit with Val¹³⁶ and Phe¹⁴⁴ point mutated to Ile and Leu, respectively. In this expression system, cytisine efficacy approaches the level observed for α4β4 and although not comparable with the full activation observed in patch clamp, it is in sharp contrast to the lack of activation of wild-type α4β2 in oocytes.

The novel nAChR agonist NS3861 presents a selectivity profile distinct from that of cytisine. Our patch clamp experiments show the nature of the α-subunit to be crucial for NS3861 activity. In addition, efficacy levels were determined by the nature of the β-subunit in a manner reciprocal to that of cytisine (i.e. β2 over β4). Full activation was seen at α3β2 receptors, partial efficacy at α3β4 receptors, and only insignificant peak currents at α4β2 receptors. As NS3861 displays a binding affinity in the nanomolar range on α4β2 receptors, the lack of activation at α4-containing receptors clearly does not reflect lack of binding. Furthermore, it is also independent of the presence of an α4α4 interface. Irrespective of α- and β-subunit stoichiometry, NS3861 shows no efficacy at α4β2 receptors and only marginal efficacy at α4β4 receptors expressed in oocytes. Efficacy selectivity for α3β2 versus α3β4 was also qualitatively confirmed in oocytes under conditions where both populations contain an α3α3 interface, which indicates that the maximal efficacy is independent of this interface. Additionally, contrary to cytisine the results from chimeric receptor combinations show that the efficacy of NS3861 was independent of TMD identity in these studies.

Structurally, the observed α3-selectivity of NS3861 might appear surprising as the only difference in the agonist binding site between the α3- and α4-subunit is Ser¹⁸¹ versus Thr¹⁸³, respectively (Fig. 6A). However, in our obtained binding modes the orientation of the thiophene ring of NS3861 is almost perpendicular to the pyridine ring of, for example, NS3531 or nicotine in co-crystal structures with a homologous AChBP (14, 41), presumably due to the presence of the bromo substituent. This puts the agonist under the influence of Thr¹⁸³ in α4 as evident by the ∼1 Å shift in binding mode in α4β2 compared with α3β2 as measured on the thiophene sulfur and the bromine atoms (Fig. 6D). Because the docking protocol used can only account for side chain flexibility, the full consequences in terms of backbone conformational changes in the receptor are uncertain. However, the shift in binding mode of NS3861, caused by Thr¹⁸³ of α4, brings the bromine within 3.4 Å of one of the sulfur atoms in the disulfide bond of the C-loop and within 3.1 Å of one of the methyl groups of Leu¹⁴⁶ and Leu¹⁴⁴ in the β2- and β4-subunits, respectively (Fig. 6D). Such short distances correspond to steric clashes between agonist and receptor. This could indicate that the shifted binding mode of NS3861 in α4-containing receptors opposes channel activation by hindering a conformational change of the interface into the activated state, by preventing either a necessary agonist-mediated intimate contact between the α4- and the β-subunit and/or preventing full closure of the C-loop. This is backed by the observation that, whereas NS3861 was unable to activate wild-type α4β2 receptors it displays agonistic behavior at α4(T183S)β2 mutant receptors in oocytes. Although the obtained efficacy level in these experiments did not match that of wild-type α3β2 it still demonstrates that Thr¹⁸³ is, at least partly, responsible for the lack of NS3861 efficacy on α4-containing receptors.

Consistent with the poor hydrogen bond acceptor ability of the thiophene sulfur atom we do not observe an interaction to the water molecule, as we do for cytisine (Fig. 6B) and, which has been reported for other agonists (12, 14, 41). The fact that NS3861 is a full agonist at α3β2 indicates that, in contrast to α4β2 agonists (42), a hydrogen bond acceptor is not essential for full activation of α3β2 receptors. This seems to be consistent with the situation in α7 nAChRs where a co-crystal structure of AChBP with an α7-selective agonist displays a binding mode, which does not involve interaction with the water molecule (12). This is also consistent with superior binding affinity at α3- versus α4-containing receptors.

As was the case for cytisine, the β-subunit plays an important role for NS3861 efficacy but we observe almost the reverse β-subunit selectivity profile with preference for β2. To the best of our knowledge, NS3861 represents the first agonist reported to possess efficacy selectivity for the α3β2 nAChR subtype. Based on the obtained binding modes at the α3β2 and α3β4 interfaces we do not observe a specific interaction as an obvious reason for the β2-efficacy selectivity of NS3861. Given that binding affinity at β2-containing receptors is ∼10-fold lower than that at β4-containing receptors, this is perhaps not surprising. From the proximity of the bromine atom of NS3861 and Phe¹⁴⁴ of β2 (Fig. 6C) we propose an electrostatic interaction between the negative electrostatic potential around the circumference of the bromine and the positive electrostatic potential at the edge of the phenyl group in the phenylalanine (43). Because polarization of aromatic rings is not explicitly handled in forcefield-based methods, such an interaction cannot be expected as part of the output from the induced-fit docking. However, our binding model of NS3861 in α3β2 requires only a small conformational change of Phe¹⁴⁴ to establish this electrostatic interaction, which is not possible with Leu¹⁴² in the β4-subunit and could be the reason for the selective activation.

The results obtained with cytisine and NS3861 corroborate previous reports that both the α- and β-subunits take part in determining the binding affinity in heteromeric receptors (3, 4). The data are also consistent with previous observations that both subunits can affect agonist efficacy (7, 8). However, by evaluating specific binding site determinants that make contacts to these agonists, a more detailed picture emerges. The data presented here for cytisine are perfectly consistent with a hypothesis stating that the α-subunit mostly plays a requisite role for agonist efficacy, in the sense that it provides a set of key determinants essential for agonist binding. Positioning of the positive charge of the ligand inside the aromatic nest delivered mostly by the α-subunit provides an excessive binding potential, which is necessary for high affinity binding, but not sufficient for gating. Ligand interactions with residues on the β-subunit are much weaker and thus play a less important role in agonist binding. However, because inter-subunit contacts are suggested to determine agonist efficacy (14) these weaker β-subunit interactions have a decisive role in fine-tuning agonist efficacy.

As illustrated by our homology models and binding modes derived from docking, both α3- and α4-subunits are capable of binding cytisine in a manner permissive for gating. However, despite a large difference in binding affinity, maximal efficacy ends up being fully determined by the β-subunit. Compared with other nicotine-like agonists, such as epibatidine and NS3920 (14), cytisine is ∼2 and ∼3 Å shorter, respectively, measured from the cationic nitrogen to the most distal atom, and only the presence of Ile and Leu in β4 allows for sufficient van der Waals contact surface when it is docked in the model. Likewise, the α3-subunit is capable of binding NS3861 in a “correct” manner, but partial versus full agonism is determined by the β-subunit possibly through interactions with Phe¹⁴⁴ in β2.

The fact that NS3861 does not activate α4-containing receptors may seem contradictory to the hypothesis of a requisite role for the α-subunit. However, the fact that Thr¹⁸³ in α4 prevents activation can be explained by the shift in NS3861 binding mode observed from our docking results, which may prevent NS3861 from binding in a manner permissive for gating. Thus, despite high affinity binding through the key α-determinants, no β-subunit can adapt sufficiently to give efficacy.

Recent publications have claimed that partial agonist efficacy is due to prevention of full C-loop closure (11–13) and thus originates from interactions with the principal component of the binding site, i.e. the α-subunit. Although the α3 efficacy selectivity of NS3861 could potentially also be explained by changes in C-loop closure, the remaining results obtained in this work are inconsistent with this hypothesis. Cytisine fully activates both α3β4 and α4β4 receptors but results in virtually no efficacy at α3β2 and α4β2 receptors (Fig. 5). As seen, this can be rationalized by specific interactions between cytisine and residues from the β-subunit. In contrast, the C-loop closure hypothesis would imply that the specific combination of cytisine and β2 should prevent closure of the α3- and α4- C-loops. However, docking of cytisine to both α3β2 and α4β2 homology models show that in both cases a fully closed C-loop can accommodate this relatively small agonist without significant steric clashes. Furthermore, a very recent co-crystal structure of a AChBP with cytisine confirms our obtained binding mode and shows a fully closed C-loop (44). In fact, two additional crystal structures of AChBPs in complex with partial agonists, lobeline (13) and varenicline (44), also display full C-loop closure. Thus, the present data and the mentioned three AChBP crystal structures support the study by Rohde et al. (14) in which it is concluded that C-loop closure is not enough to explain differences in maximum efficacies and that agonist interactions with key binding site residues at the β-subunit plays a decisive role in controlling efficacy levels.

Another inference from the present data is that maximal efficacy can be affected by the nature of the β-subunit TMD. Although that may seem trivial, the influence is clearly highly agonist dependent. All the data obtained here were normalized to maximal ACh-evoked currents. Given that the maximal efficacy levels of NS3861 at receptors with the same α3- and β-LBD (e.g. α3β2 and α3 + β2/β4) were virtually identical, maximal efficacy of ACh and NS3861 do not differ relative to each other. ACh and NS3861 thus have either identical or no β-subunit TMD dependence. Cytisine on the other hand clearly went from a full to a partial agonist when a wild-type β4-subunit was replaced with a β4/β2-subunit. Although this underscores the importance of the β-subunit for determining efficacy it also suggests that β2-TMD containing receptors may be more difficult to gate and require particularly strong interactions to the complementary subunit, which cytisine may not be capable of even with a β4-LBD. The observations that β2-containing receptors generally desensitize more rapidly than β4-containing receptors is also consistent with such basic gating differences. No effects of LBD/TMD chimeras were seen for the α-subunits as long as the LBD remained constant, which in fact support the hypothesis of the requisite role of these subunits.

The gating mechanism of nAChRs is still not fully elucidated, but the data presented here points toward an important role of the β-subunit, which potentially involves global changes of β-subunit conformation. Such changes may be linked to the F-loop movement in the complementary subunit previously suggested as part of the activation mechanism (45).