European Journal of Pharmaceutical Sciences 47 (2012) 813–823

Contents lists available at SciVerse ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage: www.elsevier .com/ locate/e jps

Analgesic effects mediated by neuronal nicotinic acetylcholine receptor agonists:Correlation with desensitization of a4b2� receptors

Jiahui Zhang, Yun-De Xiao, Kristen G. Jordan, Phil S. Hammond, Katherine M. Van Dyke,Anatoly A. Mazurov, Jason D. Speake, Patrick M. Lippiello, John W. James, Sharon R. Letchworth,Merouane Bencherif, Terry A. Hauser ⇑Targacept Inc., 200 East First Street, Winston-Salem, NC 27101, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 13 February 2012Received in revised form 17 August 2012Accepted 14 September 2012Available online 2 October 2012

Keywords:a4b2� Neuronal nicotinicacetylcholine receptorsReceptor desensitizationAnalgesic effectFormalin assay

0928-0987/$ - see front matter � 2012 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.ejps.2012.09.014

Abbreviations: ACh, acetylcholine; a4b2a5, (a4producing 50% of maximal desensitization response;tization response; EC50, concentration producingresponse; Emax, the maximal activation response; HEHS-a4b2, high sensitivity (a4b2)2b2; LS-a4b2, lowmultiple linear regression; PSA, the molecular polar snicotinic acetylcholine receptor.⇑ Corresponding author. Tel.: +1 336 480 2248; fax

E-mail address: terry.hauser@targacept.com (T.A. H

Nicotinic a4b2� agonists are known to be effective in a variety of preclinical pain models, but the underlyingmechanisms of analgesic action are not well-understood. In the present study, we characterized activationand desensitization properties for a set of seventeen novel a4b2�-selective agonists that display druggablephysical and pharmacokinetic attributes, and correlated the in vitro pharmacology results to efficaciesobserved in a mouse formalin model of analgesia. ABT-894 and Sazetidine-A, two compounds known tobe effective in the formalin assay, were included for comparison. The set of compounds displayed a rangeof activities at human (a4b2)2b2 (HS-a4b2), (a4b2)2a5 (a4b2a5) and (a4b2)2a4 (LS-a4b2) receptors.

We report the novel finding that desensitization of a4b2� receptors may drive part of the antinociceptiveoutcome. Our molecular modeling approaches revealed that when receptor desensitization rather thanactivation activities at a4b2� receptors are considered, there is a better correlation between analgesiascores and combined in vitro properties.

Our results suggest that although all three a4b2 subtypes assessed are involved, it is desensitization ofa4b2a5 receptors that plays a more prominent role in the antinociceptive action of nicotinic compounds.For modulation of Phase I responses, correlations are significantly improved from an r2 value of 0.53 to 0.67and 0.66 when HS- and LS-a4b2 DC50 values are considered, respectively. More profoundly, considering theDC50 at a4b2a5 takes the r2 from 0.53 to 0.70. For Phase II analgesia scores, adding HS- or LS-a4b2 desen-sitization potencies did not improve the correlations significantly. Considering the a4b2a5 DC50 valuesignificantly increased the r2 from 0.70 to 0.79 for Phase II, and strongly suggested a more prominent rolefor a4b2a5 nAChRs in the modulation of pain in the formalin assay.

The present studies demonstrate that compounds which are more potent at desensitization of a4b2�

receptors display better analgesia scores in the formalin test. Consideration of desensitization properties ata4b2� receptors, especially at a4b2a5, in multiple linear regression analyses significantly improves corre-lations with efficacies of analgesia. Thus, a4b2� nicotinic acetylcholine receptor desensitization may con-tribute to efficacy in the mediation of pain, and represent a mechanism for analgesic effects mediated bynicotinic agonists.

� 2012 Elsevier B.V. All rights reserved.

1. Introduction potential presence of additional subunits) are known to be analge-

Agonists selective for certain neuronal nicotinic acetylcholinereceptors (nAChRs) such as a4b2� (the asterisk indicates the

ll rights reserved.

b2)2a5; DC50, concentrationDmax, the maximal desensi-

50% of maximal activationK, human embryonic kidney;sensitivity (a4b2)2a4; MLR,

urface area; nAChR, neuronal

: +1 336 480 2113.auser).

sic in various pre-clinical pain models. One of the earliest demon-strations of this effect was reported in 1932, when nicotine wasshown to be efficacious in a cat model of visceral pain (Flores,2000). More recent studies have extended these findings to includea number of additional nicotinic compounds. Specifically, epibati-dine demonstrates powerful analgesic properties which are 200-fold more potent than those of morphine (Yogeeswari et al.,2006). Sazetidine-A is a potent analgesic in a rat model of forma-lin-induced pain (Cucchiaro et al., 2008). In addition, ABT-594 isnot only potently effective in animal models of acute, inflamma-tory and neuropathic pain (Bannon et al., 1998), but also showedanalgesic effects in patients with diabetic peripheral neuropathic

814 J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823

pain (Rowbotham et al., 2009). These analgesic compounds are allpotent agonists at a4b2� receptors (Bannon et al., 1998; Cucchiaroet al., 2008; Flores, 2000; Yogeeswari et al., 2006). Further, knock-out mice lacking either a4 or b2 genes show reduced sensitivity tothe analgesic effects of nicotinic compounds in acute pain tests(Marubio et al., 1999). Conversely, gain-of-function mice withknock-in a4 subunits are more sensitive to nicotine-induced anal-gesia (Damaj et al., 2007).

Although there is a general consensus about the involvement ofa4 and b2 subunits in analgesia, full understanding of the distincta4b2� subtypes and their required pharmacological manipulationfor the effect remain elusive. The existence of two subtypes ofa4b2�, high- and low-sensitivity (HS- and LS-) receptors, has beendemonstrated both in vitro and in vivo (Benwell et al., 1988;Moroni et al., 2006; Nguyen et al., 2003; Zwart and Vijverberg,1998). The HS- and LS-a4b2 receptors differ significantly not onlyin their stoichiometries, with (a4b2)2b2 for HS- and (a4b2)2a4 forLS-a4b2, but also in their sensitivity to activation by agonists,desensitization kinetics and regulation by chronic exposure toagonists (Moroni et al., 2006; Zwart and Vijverberg, 1998). Hence,the separate expression of HS- and LS-a4b2 receptors and charac-terization of the functional properties of these specific receptorsubtypes should be a logical approach for delineating a4b2�

receptor-targeted compounds and their analgesic effects.There has also been some controversy as to whether activation

of a4b2� receptors is sufficient to produce analgesia, in spite of awealth of supporting evidence. Recently, Gao and colleagues pro-filed a series of nicotinic agonists, both selective and nonselectivefor a4b2� receptors, in various in vitro and in vivo assessments.They found that some highly selective a4b2� agonists did not ap-pear to be analgesic, and concluded that activation of a4b2� recep-tors was necessary but not sufficient to produce analgesia (Gao etal., 2010). Additionally, the authors suggested that activities ofother receptor subtype(s) could produce an increase in the overalllevel of inhibitory synaptic transmission in the spinal cord beyondthat produced by a4b2� agonism alone.

One potential target for optimization of nicotinic modulation ofanalgesia is the activity at a5-containing nicotinic receptors (a5�).The role for a5� nAChRs in analgesia is supported by a number ofstudies using various experimental approaches. Specifically, spinalexpression of a5 was increased in response to spinal nerve ligation(Vincler and Eisenach, 2004). Additionally, it was demonstratedthat antinociception induced by nicotinic agonists in wild-typeanimals was lost when tested in a5 knockout mice (Jackson etal., 2010). Interestingly, it has been shown that a5 subunits inthe brain are exclusively associated with a4b2� receptors (Mao etal., 2008).

Taken together, these data raise questions regarding what as-pect of functional activity at a4b2� nAChRs is relevant to analgesia,and whether and how incorporation of a5 subunits into a4b2�mayaffect activity in the mediation of pain. In the present studies, wecotransfected concatenated human b2-a4 dimers with a4, b2 anda5 monomers in order to express LS-, HS-a4b2, and (a4b2)2a5(a4b2a5) nAChRs, respectively, in HEK293F cells. This approachhas been demonstrated previously to produce a homogeneous pop-ulation of receptors, the stoichiometry of which was determined bythe monomer added Papke et al., 2010; Zhou et al., 2003. Usingthese cells, we then conducted side-by-side examinations of acti-vation and desensitization properties for a set of 17 a4b2� agoniststhat demonstrated selectivity for a4b2� receptors, displayed a widerange of agonist activities, and had adequate physical and pharma-cokinetic profiles. In order to compare our results to those of oth-ers, we examined two reference compounds in the same in vitroand in vivo systems. These were ABT-894 and Sazetidine-A, bothknown to be a4b2-selective (Xiao et al., 2006; Ji et al., 2007) andeffective in a variety of pre-clinical pain models including the

formalin assay (Cucchiaro et al., 2008; Jain, 2004). Functional activ-ities observed at the three a4b2� receptor subtypes, along with var-ious parameters that may play a role in the activity profile, werethen examined for correlations with analgesic efficacies observedin a mouse formalin model. We further developed mathematicalmodels to quantitatively predict the analgesia efficacy as a functionof the in vitro pharmacological and physical profile of thecompounds.

2. Materials and methods

2.1. Chemicals

Acetylcholine (ACh) and atropine were obtained from Sigma (St.Louis, MO). Novel nicotinic compounds, ABT-894 and Sazetidine-Awere synthesized at Targacept Inc. (Winston-Salem, NC). Dul-becco’s modified Eagle’s medium (DMEM), fetal bovine serum, heatinactivated horse serum, L-glutamine, sodium pyruvate, zeocin,and Hank’s Balanced Salt Solution were obtained from Invitrogen(Carlsbad, CA). Geneticin and hygromycin B were obtained fromCalbiochem (San Diego, CA).

2.2. Transient transfection

Human a4, b2 and a5 cDNA monomers and b2-a4 dimers wereprovided by Dr. Isabel Bermudez (Oxford Brookes University, Ox-ford, UK). Single a4 and b2 subunits were concatenated by syn-thetic linkers into dimeric b2-a4 constructs using the T4 ligaseenzyme as specified by the manufacturer (NEB Biolabs, UK). Bothdimers and monomers of a4, a5 and b2 were subcloned into thepCI plasmid vector (Promega, UK), with modifications of the vectorby oligo-hybridization to contain an AscI and EcoRV restriction site,and a SwaI site downstream from the SV40 late region.

The FreeStyle 293 Expression System with HEK293F cells (Invit-rogen, Carlsbad, CA) was used for the transient transfection accord-ing to the manufacturer’s instructions. To control the expressionlevel, conditions for transfection were optimized by the numberof cells transfected and the amount and ratio of plasmid DNA used.Receptor protein expression was optimized by harvesting cellsduring a specific post-transfection period and plating cells at a spe-cific density. Briefly, concatenated dimers of b2-a4 were co-trans-fected with monomeric a4 for the expression of LS-a4b2, b2 for theexpression of HS-a4b2, or a5 for the expression of a4b2a5. ThecDNA ratio was 4:1 for b2-a4 dimer to a4, b2 or a5 monomer,and the total amount of cDNA was 20 lg for the transfection of4 � 107 cells. Transfected cells were maintained in FreeStyle 293Expression Medium for 48–72 h and then subcultured into poly-D-lysine coated 96-well plates at a density of 0.8–1 � 106 cells/ml. Cells were further incubated at 29 �C for 48 h before functionalactivities were examined.

2.3. Cell cultures

The SH-EP1 cells stably transfected with human (h) consensusa4b2 nAChRs and SH-SY5Y human neuroblastoma cells were pro-vided by Dr. Ronald J. Lukas (Barrow Neurological Institute, Phoe-nix, AZ). HEK293 cells stably transfected with ha7/RIC3 wereprovided by Dr. John Lindstrom (University of Pennsylvania, Phila-delphia, PA). The rat pheochromocytoma PC-12 cell line, Shootersubclone, was obtained from Dr. Eric Shooter (Stanford University,Stanford, CA).

The SH-EP1ha4b2, SH-SY5Y and PC-12 cells were all main-tained in DMEM with 10% horse serum, 5% fetal bovine serum,1 mM sodium pyruvate and 4 mM L-Glutamine. Zeocin (0.25 mg/ml) and hygromycin B (0.13 mg/ml) were used for a4 and b2

J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823 815

selection, respectively, for SH-EP1 ha4b2 cells. HEK ha7/RIC3 cellswere maintained in DMEM with 10% fetal bovine serum and 4 mML-glutamine, and zeocin (0.5 mg/ml) and geneticin (0.6 mg/ml)were used for subunit expression selection. All cells were main-tained in a humidified atmosphere containing 5% CO2 in air at37 �C.

2.4. Membrane potential assay

HEK293F cells transfected with ha4b2� receptors were incu-bated at 29 �C with a fluorescent membrane potential dye (Molec-ular Devices, Union City, CA) in the presence of 0.5 lM atropine inHank’s Balanced Salt Solution with 20 mM Tris-HEPES (pH 7.4).Cells were then challenged with test compounds, and membranepotential changes were detected using FLIPRTETRA (Molecular De-vices, Union City, CA), with excitation at 510–545 nm and emis-sion at 565–625 nm. The desensitization properties ofcompounds were obtained by treating cells with various concen-trations of compounds for 30 min (min), followed by challengingcells with 10 lM ACh while measuring membrane potentialresponses.

Under the conditions used, ACh activated HS-, LS-a4b2 anda4b2a5 receptors with respective EC50 values of 163 ± 1.2, 456 ±1.4 and 297 ± 1.4 nM. The responses of ACh (10 lM) were alwaysmeasured in parallel in order to normalize the activation or desen-sitization responses. Data were expressed as the Mean ± SEM fromindependent experiments. Values for concentrations producing50% of the maximal activation or desensitization response (EC50 orDC50), the maximal activation or desensitization response (Emax orDmax) were generated by fitting concentration–response curveswith nonlinear regression using GraphPad Prism (La Jolla, CA).

2.5. Calcium flux assay

Two days before experiments, SH-EP1ha4b2 cells were culturedin 96-well plates at a density of 6–8 � 104 cells/well and incubatedat 37 �C overnight. The next day, cells were transferred to an incu-bator set at 29 �C for 24 h. Cells were loaded with a calcium (Ca)-sensitive fluorescent dye Calcium 4 (Molecular Devices, Union City,CA) in modified Earle’s balanced salt solution composed of the fol-lowing: 4 mM CaCl2, 0.8 mM MgSO4, 20 mM NaCl, 5 mM KCl, 5 mMD-glucose, 20 mM Tris-HEPES, and 120 mM N-methyl-D-glucamine,pH 7.4. The Ca mobilization was detected using FLIPRTETRA withexcitation of 470–495 nm and emission of 515–575 nm. The Caflux data were normalized as the percentage of the Ca response in-duced by 10 lM nicotine, and expressed as mean ± SEM.

2.6. Competitive binding assays

All membrane preparation procedures were carried out at 4 �C,and centrifugation was carried out at 40,000g for 20 min. Sourcesfor different nAChRs are: rat (r) cortex for ra4b2, SH-EP1ha4b2cells for ha4b2, HEK ha7/RIC3 cells for ha7, SH-SY5Y for ha3b4and PC-12 for ra3b4 nAChRs. Rat cortices (Analytical BiologicalServices, Wilmington, Delaware) were isolated from adult Spra-gue–Dawley rats. Cells or tissues were homogenized in ice-coldphosphate buffered saline (PBS) and centrifuged. After centrifuga-tion, pellets were then washed, suspended in PBS, aliquoted andfrozen at �20 �C if not immediately used.

The [3H]-epibatidine and [3H]-nicotine were obtained fromPerkinElmer (Waltham, MA). Protein concentrations were deter-mined by a Pierce Coomassie Plus Assay Kit (Pierce Chemical Com-pany, Rockford, IL). Membranes were incubated in PBS in thepresence of radioligands and various concentrations of compoundsfor 2 h at 25 �C. Non-specific binding was defined in the presenceof 10 lM epibatidine. Incubation was terminated by rapid filtration

on a multi-manifold tissue harvester (Brandel, Gaithersburg, MD)using GF/B filters presoaked in 0.33% polyethyleneimine. Radioac-tivity was determined by liquid scintillation counting using a Per-kin–Elmer Trilux Microbeta (Waltham, MA).

Binding data were expressed as the average percentage of totalspecific binding, and plotted against the log concentration of thecompound. The IC50 was determined by least squares non-linearregression. Binding affinity Ki was calculated using the Cheng–Prusoff equation of Ki = IC50/(1 + N/Kd), where N is the concentra-tion of the radioligand and Kd is the affinity of [3H]-epibatidineor [3H]-nicotine defined specifically for the assay under the condi-tions used.

2.7. Animal care

Adult male CD-1 (ICR) mice weighing approximately 20–25 gwere supplied by Charles River Laboratories (Raleigh, NC). Animalswere housed and cared for in accordance with the ‘‘Guide for theCare and Use of Laboratory Animals’’ as published by the NationalResearch Council (1996). All protocols were approved by the Insti-tutional Animal Care and Use Committee at Targacept Inc.

2.8. Rotarod study

The effects of compounds on locomotor activity were quantita-tively evaluated with the rotarod test in mice (Karl et al., 2003). Forthis study, adult male mice were trained to perform on the accel-erating rotarod (0–40 RPM over 120 s) apparatus. Training con-sisted of three trials per day over the course of 3 days. On the4th day animals (N = 6 per group) were administered test drug orvehicle subcutaneously, placed on the 3.5 cm rotating rod andthe time/latency required for the mouse to fall from the rod (witha maximum of 120 s) was recorded. Doses expressed as the basecompound that resulted in a significant decrease in latency com-pared with vehicle treated animals were considered motor-impair-ing, and the minimum effective dose (MED) was reported for eachcompound.

2.9. Formalin study

The formalin test in mice was performed as described by Dub-uisson and Dennis (1977). The doses of each compound testedwere chosen based on the rotarod study to exclude the possibilityof side effects affecting motor responses. Briefly, after an acclima-tion period, test compound, morphine or saline were administeredsubcutaneously 20 min before an intra-dermal injection of 25 ll of2.5% formalin solution into the right hind paw. Once injected, micewere immediately returned to a clear Plexiglas™ observationchamber. Phase I of the pain response was defined as 0–9 min,and Phase II was 10–40 min following formalin injection(Malmberg and Bannon, 2007). Sessions were videotaped forviewing and scoring at a later time by raters blinded to the exper-imental conditions.

Each subject was observed for 1 min at 5-min intervals over a40-min session, and the time spent licking the affected paw was re-corded. Several doses of each compound were tested in differentgroups of animals (N = 10 per treatment group) and relative anal-gesia scores on a scale from 0 to 100 were obtained by normalizinganalgesic effects to that of saline and morphine. In every formalinassay, the effect of 5 mg/kg morphine was tested in parallel as apositive control with an assigned analgesia score of 100, and the ef-fect of saline as a vehicle control with a score of 0. Analgesia scoresof each compound tested at a base dose of 3 mg/kg, a common dosetested for all compounds, were reported and used for correlationanalyses. The analgesic effect of a compound was compared to

816 J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823

the saline control using one-way ANOVA followed by Dunn’s posthoc, with P < 0.05 considered significant.

2.10. In vivo concentration evaluations

Plasma and brain concentration in mice coincident with tim-ing of the behavioral evaluations in the formalin study were as-sessed using a cassette dosing method similar to that describedby Berman and colleagues (Berman et al., 1997), where up to fivecompounds were administered subcutaneously (N = 6 per group)in combination at 1.5 lmol/kg each. All other parameters, includ-ing timing of manipulations and administration of formalin, werein line with the formalin test procedures described above. Follow-ing drug administration, formalin was injected at 20 min, plasmaand brain samples were collected from half of the subjects at25 min (corresponding to midpoint of Phase I) and from theremaining subjects at 45 min (corresponding to midpoint ofPhase II). Quantitative analyses of the plasma and brain sampleswere performed by HPLC-MS/MS with a sensitivity range up to1000 nM. Results fell out of the sensitivity range are reportedas >1000 nM.

2.11. Physical properties of novel nicotinic compounds

Based on the compound structure, Pipeline Pilot (Accelerys, Inc.,San Diego, CA) was used to determine the molecular properties foreach compound, including the number of hydrogen-bond (H-bond)donors, the partition coefficient for n-octanol/water CLog P, thedistribution coefficient Log D at pH of 7.4, the molecular polar sur-face area (PSA) and the molecular weight.

2.12. Principal components analysis

Principal component analysis (PCA) described by Jolliffe (1986)was used to compare the differences of chemical structures of nic-otinic compounds. PCAs were determined from 134 structural fea-tures, with all calculations carried out by Molecular OperatingEnvironment (Chemical Computing Group Inc., Montreal, QC,Canada). PCA reduces the dimensionality of molecular features intoa small number of uncorrelated variables (principal components),with the first principal component accounting for as much variabil-ity in the original data as possible, and each succeeding componentaccounting for as much of the remaining variance as possible. Thus,PCA can provide a lower-dimensional picture using two or threedimensional plots, to show the relative distribution of compoundsin high dimensional chemical space.

2.13. Correlation analysis and statistics

In vitro pharmacological and physical parameters includedbinding affinities at ra4b2� nAChRs, the number of H-bond donorsand receptor activation and desensitization potencies at ha4b2a5,LS- and HS-a4b2 subtypes. In all cases, Ki, EC50 and DC50 valueswere transformed by taking the log(10) of each value. Those num-bers were then used for the correlation analyses. Correlations be-tween analgesia scores and in vitro parameters were analyzed bymultiple linear regression (MLR), and coefficients of determination(r2) were calculated using Pipeline Pilot (Accelerys Inc., San Diego,CA). Significance tests of an individual coefficient in the regressionmodel, variance ratio (F) and probability factor related to F-ratio(P-value) were performed or determined using Microsoft Excel(Redmond, WA). Statistical qualities of MLR equations were judgedby parameters of coefficient of determination, variance ratio andprobability factor related to F-ratio.

3. Results

3.1. Structures of novel nicotinic compounds

The novel nicotinic compounds selected for this study werechosen based on their chemical structural diversity, their relativeselectivity for a4b2� receptors, and their broad range of agonistactivities at a4b2� receptors. The structures of five compoundsused in the present studies are shown in Fig. 1A. Others have notbeen revealed due to proprietary and patent-related issues. How-ever, the structural differences of the remaining compounds ascompared to these five novel a4b2�-selective compounds, as wellas ABT-894 and Sazetidine-A, are graphically depicted in a threedimensional picture using the method of principal componentanalysis (Fig. 1B). As can be seen, a relatively diverse group ofchemotypes was included in our studies.

3.2. The selectivity for a4b2� receptors by novel nicotinic compounds

To investigate the selectivity for a4b2� receptors, competitivebinding assays were carried out with several nAChR subtypes.The resulting Ki values are shown in Table 1. As demonstrated, Kivalues were similar between human and rat tissues, both fora4b2� and a3b4 receptors. All compounds tested demonstratedhigh affinity for a4b2� receptors, with Ki values less than 100 nMat both human and rat a4b2� receptors with the exception of Com-pound 15, which had a Ki value of 15 nM at ha4b2� but 330 nM atra4b2� receptors. Additionally, these compounds were more than20-fold selective for ha4b2� over ha7 receptors, with the exceptionof Compounds 9 and 11 which were only 7- and 3-fold selective,respectively. Moreover, all of the novel nicotinic compounds weremore than 44-fold selective for ha4b2� over ha3b4 receptors.

For ABT-894 and Sazetidine-A, Ki values were all in the sub-nanomolar range for both ha4b2� and ra4b2� receptors. Both com-pounds were more than 100-fold selective for ha4b2� over ha7 andha3b4 receptors. Although an exact comparison is impossible dueto the differences of tissues and radioligands used, our results areconsistent with the previous reports that ABT-894 and Sazeti-dine-A are potent and selective a4b2� agonists Xiao et al., 2006;Ji et al., 2007.

3.3. Activation and desensitization of a4b2� receptors by novelnicotinic compounds

Transient expression of ha4b2� nAChRs was achieved inHEK293F cells by co-transfecting concatenated b2-a4 dimers withmonomeric b2, a5 or a4 for the expression of HS-a4b2, a4b2a5 orLS-a4b2, respectively. To ensure that a nicotinic, a4b2�-mediatedmechanism of action is involved for each of these a4b2� receptortypes, the responses induced by 10 lM acetylcholine (ACh) in thepresence of various concentrations of dihydro-b-erythroidine andmecamylamine, two different nicotinic receptor antagonists, weredetermined. At each receptor subtype, full inhibition of ACh-evoked responses was observed for both antagonists. The IC50 val-ues were 135, 88 and 161 nM for dihydro-b-erythroidine, and1902, 249 and 236 nM for mecamylamine at HS-a4b2, a4b2a5and LS-a4b2 subtypes, respectively (data not shown).

Subsequently the activation and desensitization activities of 17novel nicotinic compounds were examined using these cells. Theactivation and desensitization curves for Compounds 3, 4, 5, 7,and 8 are shown in Fig. 2. The EC50 and DC50 values of all test com-pounds at each ha4b2� receptor subtype are listed in Table 2. Asshown in Fig. 2A and Table 2, activation efficacies for thesecompounds ranged from partial to full agonism, with some com-pounds showing no detectable activity (EC50 > 10,000 nM) over

Compound 7 Compound 8Compound 5Compound 3 Compound 4

(A)

(B)

Fig. 1. Structures and the structural differences of nicotinic compounds. (A) Structures of novel nicotinic compounds. (B) Principal component analysis of structural featuresof compounds. The distribution of 19 compounds in the space of the first three principal components determined from 134 structural features is shown from two differentviews of the 3-dimentional graph. The arrow indicates the rotation direction of the graph. The color of red, green, magenta, goldenrod, cyan, navy-blue and dark-grayrepresent Compounds 3, 4, 5, 7, 8, ABT-894 and Sazetidine-A, respectively. Variances on the PCA1, PCA2 or PCA3 coordinate suggest the structural differences of compounds.

Table 1The binding affinity of novel nicotinic compounds at different types of nAChRs.

Compound Ki ± SEM, nM

r_a4b2 h_a4b2 h_a7 r_a3b4 h_a3b4

ABT-894 0.9 ± 0.1 0.9 ± 0.2 750 ± 335 560 ± 114 180 ± 28Sazetidine-A 0.05 ± 0.0 0.26 ± 0.2 3300 ± 745 3000 ± 630 30 ± 2Compound 1 7 ± 1.4 2 ± 0.3 450 ± 60 330 ± 42 100 ± 19Compound 2 6 ± 0.6 5 ± 0.7 420 ± 52 580 ± 144 310 ± 31Compound 3 64 ± 11.5 16 ± 2.9 65,000 ± 13,622 9700 ± 1836 5100 ± 1086Compound 4 8 ± 1.2 9 ± 3.4 990 ± 385 720 ± 127 420 ± 166Compound 5 2 ± 1.1 1 ± 0.4 550 ± 224 1300 ± 367 800 ± 238Compound 6 63 ± 11.5 6 ± 2.0 520 ± 96 320 ± 59 420 ± 96Compound 7 1 ± 0.6 2 ± 0.4 76 ± 17 510 ± 159 790 ± 259Compound 8 11 ± 9.7 3 ± 1.2 2000 ± 371 2200 ± 307 3400 ± 1671Compound 9 12 ± 1.9 9 ± 2.1 70 ± 5 1100 ± 422 1800 ± 512Compound 10 7 ± 2.3 2 ± 0.5 87 ± 26 300 ± 44 530 ± 73Compound 11 110 ± 46 11 ± 2.9 29 ± 11 1700 ± 817 1300 ± 476Compound 12 5 ± 3 5 ± 3.2 1000 ± 338 10,000 ± 8891 750 ± 75Compound 13 20 ± 8.0 6 ± 2.4 59,000 ± 27,839 49,000 ± 21,222 9700 ± 1705Compound 14 77 ± 32 33 ± 9.3 720 ± 307 260 ± 127 2000 ± 810Compound 15 330 ± 98 15 ± 2.5 300 ± 54 1400 ± 174 2000 ± 890Compound 16 25 ± 7.7 14 ± 2.8 790 ± 73 330 ± 113 1100 ± 159Compound 17 38 ± 17 25 ± 13.1 1200 ± 464 670 ± 135 1100 ± 498

Competitive binding assays were carried out using [3H]-epibatidine and membranes made from rat cortical membranes, SH-EP1 ha4b2, HEK ha7/RIC3, PC-12 and SH-SY5Ycells. The Ki values listed are mean ± SEM of 3–8 experiments. A majority of the compounds demonstrated high affinity and relative selectivity for a4b2� receptors over a7 ora3b4 receptors.r: rat.h: human.

J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823 817

Fig. 2. Activation and desensitization of a4b2� receptors by novel nicotinic compounds. HEK293F cells separately expressing human a4b2a5, LS- and HS-a4b2 receptors weretreated with compounds along with (A) or 30 min before (B) 10 lM ACh challenge. Membrane potential data are mean ± SEM of more than three experiments. (A) Compoundstested displayed a range of agonist activities at a4b2� receptors. (B) Each compound produced full desensitization of the ACh-induced response with comparable DC50 valuesat all three a4b2� subtypes.

818 J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823

the concentration range tested (0.01–10,000 nM) and experimentalconditions used. Note that most of the compounds in this studyhaving no detectable agonist activity were not simply antagonists.Nearly all of the compounds demonstrated partial agonist re-sponses in a calcium (Ca) flux assay in SH-EP1 cells with mixedsubtypes of HS- and LS-a4b2 receptors, where sodium in the assaybuffer was replaced with N-methyl-D-glucamine to reveal any ago-nist activity (Table 2). Compounds 1 and 6, two compounds thatdid not show any agonist activity in the Ca flux assay, were equallypotent and efficacious at desensitization and antagonism of ha4b2�

receptors in SH-EP1 cells (data not shown).Regardless of the activation efficacies, almost all compounds

examined produced full desensitization at all three ha4b2� subtypes(Fig. 2B and Table 2). We observed that for any given compound, theDC50 value was usually lower than the EC50 value at the same recep-tor subtype. While DC50 values for a specific compound at the threedifferent a4b2� receptors might be similar, in several cases (includ-ing Compounds 7 and 8), their EC50 values differed greatly.

3.4. Activation and desensitization of a4b2� receptors by ABT-894 andSazetidine-A

We also examined the functional properties of ABT-894 and Saz-etidine-A, using the same HEK293F cells expressing human HS-a4b2, a4b2a5 and LS-a4b2 receptors. Concentration-dependent in-creases in responses were observed for both agonists (Fig. 3). WhileABT-894 partially activated HS-a4b2 and a4b2a5, it fully activatedLS-a4b2. In contrast, Sazetidine-A fully activated HS-a4b2 but onlypartially activated a4b2a5 and LS-a4b2 (Fig. 3A).

Unlike activation of a4b2� receptors by these two agonists, theirdesensitization was more potent, and full inhibition of the ACh-in-duced response was observed at all three receptor subtypes(Fig. 3B). For each agonist, DC50 values were comparable at all threea4b2� receptors (Table 2). The Ca flux results for ABT-894 are sim-ilar to those reported previously (Ji et al., 2007). The results for Saz-etidine-A are in line with previous reports that it is a full agonist atHS-a4b2, a poor partial agonist at LS-a4b2 (Carbone et al., 2009)and it is more potent and efficacious at desensitization than at acti-vation of the a4b2� receptors (Xiao et al., 2006).

3.5. Analgesic effects and rotarod study of novel nicotinic compounds

The formalin assay in mice was performed for the same set of 19compounds. The doses for the assay were selected based on results

from a mouse rotarod study, in which latency to fall in the acceler-ating rotarod was examined. The lowest doses which significantlyimpaired coordination in the rotarod study (MED, Table 3) or high-er were not included in the formalin assay, to exclude the possibil-ity of a non-analgesic effect of compounds resulting in impairmentof the animal’s ability to elicit a motor response.

A normalized analgesia score was obtained by comparing theanalgesic effect of each dose of compounds to that of saline (thevehicle control) and 5 mg/kg morphine (the positive control)tested within the same experiment. The saline and morphine con-trol in each phase of the assay were assigned a score of 0 and 100,respectively. The analgesia scores during Phase I and Phase II foreach compound (3 mg/kg, s.c.) are reported in Table 3. In general,an analgesia score of 40 or higher is reflective of a significant anal-gesic effect of a compound compared to the saline control. It is ofnote that both ABT-894 and Sazetidine-A demonstrated significantanalgesic effects in both Phases of the formalin test. These resultsalign with previously published data for Sazetidine-A in a rat for-malin model (Cucchiaro et al., 2008).

Along with ABT-894 and Sazetidine-A, Compounds 2 and 7–10displayed significant effects in Phase I of the formalin assay. Thecompounds that were effective in Phase I of the assay also showedsignificant analgesia in Phase II. As well, Compounds 4, 5, 12, 15and 16 were effective in Phase II of the assay.

3.6. Physical properties and in vivo concentrations of novel nicotiniccompounds

Physical properties of a compound are known to correlate within vivo bioavailability and the ability of a compound to cross theblood–brain barrier (Modi, 2003; Wang et al., 2007). As shown inTable 3, the number of H-bond donors for all test compoundswas less than or equal to 2, while their CLogP values varied be-tween�0.24 and 2.21 with most of them around 2. Whereas nearlyall of the compounds displayed values for Log D at pH 7.4 between�1 and 1, their PSA values are between 17 and 62. The molecularweight of these compounds range from 197.27 to 323.43 with amedian of 230.27. In addition, all 17 novel nicotinic compoundsas well as ABT-894 and Sazetidine-A showed no violations of Lipin-ski’s rules.

In separate studies, plasma and brain samples were collectedfrom mice treated with compounds for examination of exposurelevels. Compound concentrations, at time points of 25 and45 min (corresponding to midpoint of Phase I and Phase II of the

Tabl

e2

Act

ivat

ion

and

dese

nsit

izat

ion

ofhu

man

a4b

2�re

cept

ors

byno

vel

nico

tini

cco

mpo

unds

.

Com

pou

nd

HS-

a4b

2a

4b2a

5LS

-a4b

2H

S-a

4b2

a4b

2a5

LS-a

4b2

a4b

2�C

afl

ux

EC5

0,n

ME m

ax,%

AC

hEC

50,n

ME m

ax,%

AC

hEC

50,n

ME m

ax,%

AC

hD

C 50,n

MD

max

,%A

Ch

DC

50,n

MD

max

,%A

Ch

DC

50,n

MD

max

,%A

Ch

EC5

0,n

ME m

ax,%

Nic

AB

T-89

43

±2

50±

518

±2

66±

617

±1

120

±10

11±

110

7±

319

±1

99±

149

±1

102

±1

260

±93

99±

3Sa

zeti

din

e-A

0.98

±1

109

±2

0.02

±2

17±

20.

8±

421

±5

0.2

±1

106

±2

0.2

±1

102

±2

0.5

±1

101

±0.

327

±12

67±

11C

ompo

un

d1

>10,

000

0>1

0,00

00

>10,

000

052

2±

111

0±

925

4±

210

9±

944

30±

210

4±

2672

,000

±17

,180

1±

1C

ompo

un

d2

>10,

000

0>1

0,00

00

>10,

000

023

±1

73±

423

±2

55±

510

1±

177

±3

8900

±24

8779

±5

Com

pou

nd

359

9±

239

±4

>10,

000

014

73±

151

±3

364

±2

111

±12

518

±2

111

±12

1592

±1

92±

1034

00±

759

54±

12C

ompo

un

d4

112

±1

117

±1

81±

211

9±

831

±2

118

±11

36±

110

3±

572

±2

110

±8

80±

110

7±

371

±8

130

±17

Com

pou

nd

5>1

0,00

016

>10,

000

020

8±

142

±1

9±

197

±3

23±

110

2±

239

±1

94±

463

0±

179

80±

10C

ompo

un

d6

>10,

000

0>1

0,00

00

>10,

000

087

4±

211

0±

1085

4±

111

1±

732

,220

±2

73±

378

,000

±21

,896

1±

0C

ompo

un

d7

>10,

000

031

38±

283

±17

185

±2

103

±9

176

±1

98±

140

6±

110

6±

569

8±

194

±2

58±

711

0±

8C

ompo

un

d8

3±

423

±1

2076

±2

56±

627

4±

187

±4

10±

196

±3

15±

199

±1

44±

198

±3

130

±20

99±

3C

ompo

un

d9

>10,

000

0>1

0,00

01

>10,

000

061

±2

93±

839

8±

197

±4

367

±1

85±

367

00±

1370

53±

4C

ompo

un

d10

>10,

000

0>1

0,00

00

>10,

000

512

4±

287

±7

274

±2

93±

712

62±

195

±7

18,0

00±

8332

20±

7C

ompo

un

d11

>10,

000

0>1

0,00

00

>10,

000

411

1±

190

±4

231

±1

93±

677

5±

182

±6

6500

±11

814

±2

Com

pou

nd

12>1

0,00

00

>10,

000

013

12±

114

±0.

431

±2

94±

654

±2

93±

813

03±

295

±6

6300

±64

425

±13

Com

pou

nd

13>1

0,00

00

>10,

000

010

41±

132

±2

36±

110

0±

254

±1

100

±3

390

±2

96±

1243

00±

2061

60±

9C

ompo

un

d14

300

±2

19±

248

0±

320

±4

471

±1

62±

460

±1

103

±5

170

±2

114

±8

244

±1

107

±3

4500

±17

6278

±9

Com

pou

nd

15>1

0,00

012

>10,

000

4>1

0,00

021

291

±1

105

±5

344

±1

104

±6

1698

±1

108

±7

5200

±25

2946

±5

Com

pou

nd

1620

1±

325

±5

700

±1

35±

215

1±

274

±8

39±

110

3±

443

±1

104

±4

225

±2

104

±8

4100

±15

6068

±9

Com

pou

nd

1725

9±

421

±4

>10,

000

027

6±

348

±9

96±

110

4±

311

3±

110

6±

641

2±

210

7±

919

00±

888

44±

14

Fun

ctio

nal

acti

viti

esw

ere

assa

yed

inH

EK29

3Fce

lls

sepa

rate

lyex

pres

sin

gh

um

ana

4b2�

rece

ptor

su

sin

gm

embr

ane

pote

nti

alm

eth

odu

nle

ssot

her

wis

ein

dica

ted.

The

Ca

flu

xas

say

was

perf

orm

edin

SH-E

P1ha

4b2

cell

s.D

ata

are

mea

n±

SEM

ofm

inim

alof

two

expe

rim

ents

assa

yed

indu

plic

ates

.>1

0,00

0n

M:

no

acti

vati

onw

asde

tect

able

un

der

the

con

diti

ons

use

d.N

ic:

nic

otin

e.

J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823 819

formalin test, respectively) after compound administration, arelisted in Table 3. All compounds tested demonstrated brain pene-tration, with detectable levels in the brain at both time points. Amajority of the compounds showed good in vivo plasma and brainconcentrations. Considering that doses used in this cassette dosingstudy were approximately 10-fold lower than those in the formalinstudy, animals in the formalin assay would have been exposed tonanomolar to micromolar concentrations of each compound bothperipherally and in the CNS.

3.7. Correlation between analgesia scores and a4b2� receptor activitiesof novel nicotinic compounds

To investigate the role that a4b2� nAChR activation or desensi-tization plays in analgesia, multiple linear regression (MLR) wasused to analyze correlations between analgesic effects of the com-pounds and their HS-, LS-a4b2 and a4b2a5 functional activities,alone or in combination with other in vitro profiles (i.e. a4b2�

binding affinities, physical properties including number of H-bonddonors, CLog P, PSA or Log D at pH 7.4), or in vivo plasma and brainconcentrations. In the MLR analyses, the Ki values obtained usingrat cortical membranes were used for the analyses. Previous stud-ies have shown that binding affinity of competitive inhibitors ofnicotinic binding sites is very similar between rat and mouse brain(Marks et al., 1986). The results of these MLR analyses along withthe variance ratio (F) and P-values are shown in Table 4. All F val-ues shown in the table indicate that at least one of the coefficientsin each model was significant.

For all in vitro and in vivo parameters examined individually,ra4b2� binding affinities showed the most highly significant corre-lation with analgesia scores for both Phase I and Phase II. Althoughinclusion of the number of H-bond donors did not achieve signifi-cance for Phase I, it did result in an increase in correlation for PhaseII with an improved coefficient of determination (r2) from 0.41 to0.70 (P = 0.001). Inclusion of in vivo concentration data neithermade a significant contribution nor improved the correlation (datanot shown).

We next explored the addition of a third term to the MLRequations that include both the Ki and the number of H-bond do-nors of the compound. For correlations with analgesic efficaciesof both Phase I and Phase II, adding activation potencies for allthree a4b2� subtypes maintained but did not significantly im-prove the relationships. The r2 values were 0.62, 0.62 or 0.53 ascompared to 0.53 for Phase I, and 0.72, 0.74 or 0.71 as comparedto 0.70 for Phase II after adding EC50 values for HS-a4b2, a4b2a5or LS-a4b2, respectively. No statistical significance was observedfor any of these changes in the r2 values. In contrast, addingdesensitization potencies as a third term in the MLR equationsprovided a significant improvement in correlations for Phase Ianalgesia scores at all three subtypes of a4b2� receptors. The r2

values were 0.67, 0.70 or 0.66 as compared to 0.53 for Phase Iafter addition of DC50 values at HS-a4b2, a4b2a5 or LS-a4b2,respectively, with respective P-values of 0.02, 0.01 or 0.03. Inaddition, the r2 value significantly increased from 0.70 to 0.79(P = 0.0001) for Phase II when the DC50 values at a4b2a5 wereadded. Interestingly, a significant effect on the r2 for Phase IIwas not seen when the HS- or LS-a4b2 DC50 values were consid-ered. Given the sample size of 19 for this study, MLR analyseswith more than three terms were not performed due to the po-tential for over-fit.

Thus, MLR analysis demonstrated that better analgesia scoreswere obtained for compounds showing higher affinities and great-er desensitization potencies at a4b2� receptors, suggesting theinvolvement of a4b2� desensitization, especially of a4b2a5, in painmodulation.

Fig. 3. Activation and desensitization of a4b2� receptors by ABT-894 and Sazetidine-A. HEK293F cells separately expressing human a4b2a5, LS- and HS-a4b2 nAChRs weretreated with ABT-894 or Sazetidine-A along with (A) or 30 min before (B) 10 lM ACh challenge. Membrane potential data are mean ± SEM of 3 experiments. (A) ABT-894partially activated HS-a4b2 and a4b2a5 but fully activated LS-a4b2 receptors. Sazetidine-A (Saz) was a potent full agonist at HS-a4b2 and a partial agonist at both a4b2a5and LS-a4b2 subtypes. (B) Both agonists fully inhibited the ACh-induced response at all three a4b2� receptors, whereas Sazetidine-A was more potent compared to ABT-894.

Table 3Analgesia scores, rotarod study, physical and pharmacokinetic properties of novel nicotinic compounds.

Compound Analgesia score RotaRod Physical Property In vivo concentrations

C25 ± SEM, nM C45 ± SEM, nM

Phase I Phase II MED, mg/kg H_Donor CLogP LogD pH 7.4 PSA MW Plasma Brain Plasma Brain

ABT-894 44* 79* >30 1 1.98 �0.01 28 244.12 165 ± 12 426 ± 60 72 ± 53 446 ± 67Sazetidine-A 125* 100* NR 2 1.12 �0.79 54 260.33 58 ± 5 7 ± 0 20 ± 5 6 ± 0Compound 1 5 22 56 1 2.21 �0.49 28 231.34 >1000 280 ± 6 >1000 290 ± 4Compound 2 42* 47* >30 1 0.36 �0.99 42 203.28 >1000 >1000 >1000 >1000Compound 3 4 4 >100 1 2.13 �1.00 32 236.35 489 ± 171 41 ± 4 85 ± 40 36 ± 2Compound 4 19 42* 30 1 �0.24 �2.07 32 198.24 257 ± 85 8 ± 2 512 ± 238 6 ± 0Compound 5 15 48* >100 1 0.58 �0.35 47 235.28 362 ± 94 102 ± 4 367 ± 182 109 ± 7Compound 6 2 30 30 1 0.8 �0.92 28 229.32 >1000 405 ± 25 >1000 414 ± 33Compound 7 40* 61* >30 1 0.39 �0.12 37 223.25 277 ± 74 208 ± 10 279 ± 147 207 ± 16Compound 8 54* 44* >30 1 �0.21 �0.31 61 230.27 740 ± 224 64 ± 8 161 ± 71 54 ± 2Compound 9 62* 73* >100 1 1.86 �0.15 17 215.31 >1000 >1000 >1000 >1000Compound 10 84* 116* >100 1 1.86 �0.15 17 215.31 >1000 >1000 >1000 >1000Compound 11 0 0 >100 1 1.69 0.88 25 212.29 481 ± 27 18 ± 1 269 ± 70 13 ± 0Compound 12 19 57* 100 1 1.69 0.88 25 212.29 505 ± 79 12 ± 0 308 ± 66 7 ± 0Compound 13 7 15 >100 1 1.77 �1.29 62 323.43 177 ± 22 6 ± 0 84 ± 1 4 ± 0Compound 14 0 0 >30 1 0.87 �1.37 48 300.40 >1000 166 ± 9 >1000 81 ± 4Compound 15 7 48 10 0 2.01 0.87 30 197.27 >1000 4 ± 2 25 ± 15 6 ± 2Compound 16 28 48* >10 1 1.22 �1.56 31 286.42 34 ± 16 12 ± 1 12 ± 4 10 ± 0Compound 17 0 0 >30 1 1.77 �1.03 31 286.42 >1000 >1000 >1000 >1000

Analgesia scores, rotarod study, physical and pharmacokinetic properties of novel nicotinic compounds.The analgesia scores of each compound were obtained in Phase I and Phase II of the mouse formalin assay (N = 10), and the minimum effective dose (MED) that impaired thelocomotor ability were from mouse rotarod study (N = 6). Physical properties were determined based on compound structure. The plasma and brain concentrations wereobtained at 25 and 45 min after injection of compounds in mice following a similar procedure of the formalin assay (N = 6). ABT-894 and Sazetidine-A both demonstratedsignificant analgesic effects.MED: the minimum effective dose.H_Donor: number of H-bond donors.PSA: the molecular polar surface area.MW: molecular weight.NR: not reported.* Significant analgesic activity compared to vehicle control.

820 J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823

Table 4Correlation of analgesia scores with a4b2� receptor activities and physical properties of novel nicotinic compounds.

In vitro parameters analyzed Coefficient of determination (r2) Variance ratio (F) P-value

Score_Phase I Score_Phase II Score_Phase I Score_Phase II Score_Phase I Score_Phase II

Ki_a4b2 0.47* 0.41* 15.0 11.7 0.001 0.003Ki_a4b2, H_Donor 0.53 0.70* 8.9 19.0 0.001, 0.18 <0.0001, 0.001Ki_a4b2, H_Donor, EC50_HS-a4b2 0.62 0.72 8.1 12.6 0.002, 0.04, 0.08 <0.0001, 0.003, 0.44Ki_a4b2, H_Donor, EC50_a4b2a5 0.62 0.74 8.2 14.4 0.02, 0.04, 0.07 0.001, 0.001, 0.16Ki_a4b2, H_Donor, EC50_LS-a4b2 0.53 0.71 5.6 12.2 0.002, 0.36, 0.73 <0.0001, 0.01, 0.59Ki_a4b2, H_Donor, DC50_HS-a4b2 0.67* 0.76 10.4 15.5 0.02, 0.03, 0.02 0.001, 0.0003, 0.09Ki_a4b2, H_Donor, DC50_a4b2a5 0.70* 0.79* 11.4 18.6 0.01, 0.01, 0.01 0.0003, 0.03, 0.0001Ki_a4b2, H_Donor, DC50_LS-a4b2 0.66* 0.73 9.6 13.4 0.01, 0.04, 0.03 0.0003, 0.001, 0.27

Correlation of analgesia scores with a4b2� receptor activities and physical properties of novel nicotinic compounds.Parameters analyzed were obtained as described in the Methods. MLR was used to analyze the correlations. For analgesic scores in Phase I and Phase II of the mouse formalinassay, the coefficient of determination (r2) and the variance ratio (F) for each set of parameters analyzed are listed separately, while the P-values for each individual parameteranalyzed are listed correspondingly under each conditions.H_Donor: number of H-bond donors.Bold numbers are with significance.* Significant contribution by the last term added.

J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823 821

3.8. Modeling of analgesic effects of novel nicotinic compounds usingdesensitization potencies at a4b2a5 receptors

Considering ligand affinities at ra4b2� receptors, the number ofH-bond donors (H_Donor) and DC50 values at ha4b2a5 receptors,equations were generated to quantitatively predict analgesiascores separately for Phase I and Phase II of the formalin assayusing MLR models. The model with the highest coefficient of deter-mination value was chosen from all the models generated. Themathematical equations to predict analgesic efficacy of Phase I(Score I) and Phase II (Score II) were:

Score I ¼ 159:9� 58:8�H Donor� 26:6� LogðKiÞ � 17:8

� LogðDC50Þ

Score II ¼ 134:6� 40:2�H Donor� 19:5� LogðKiÞ � 25:6

� LogðDC50Þ

Subsequently, the predicted analgesia scores with the demon-strated analgesic efficacies obtained in the formalin test for eachcompound were compared. Plots of predicted vs. observed analge-sic efficacy scores for novel nicotinic compounds are shown inFig. 4A and B for Phase I and Phase II, respectively. Correlation coef-ficients of determination of 0.70 and 0.79 (F values of 11.4 and18.6, respectively) were obtained for Phase I and Phase II betweenpredicted and observed analgesia scores for the set of compoundstested, indicating that these models were suitable to predict anal-gesic efficacy in the mouse formalin test. Additionally, these mod-els suggest that the lower the number of H-bond donors, the Ki andthe DC50 values are, the higher the analgesia scores will be for thecompound, again consistent with the observations.

4. Discussion

The novel and important finding of the current studies is the di-rect correlation between desensitization of a4b2� nAChRs and theobserved analgesic efficacy in both acute (Phase I) and persistent(Phase II) pain in a mouse formalin model, suggesting that desen-sitization of a4b2� receptors may play a crucial role in the analge-sic effects of nicotinic compounds. Although the role that nAChRsplay in antinociception has mainly focused on a4 and b2 subunitsand their activation (Damaj et al., 2007; Gao et al., 2010; Marubioet al., 1999), the specific a4b2� subtypes involved, their functionalrequirements and the underlying mechanisms have not been fullyelucidated. By determining the in vitro pharmacological profiles of

a group of novel nicotinic compounds at several prevalent sub-types of a4b2� receptors and investigating their in vivo propertiesin a controlled and normalized assessment of nociception, we be-gin to get a clearer picture of how these receptors may be function-ing in the mediation of pain.

In agreement with previous reports (Cucchiaro et al., 2008; Da-maj et al., 2007; Gao et al., 2010; Marubio et al., 1999), the presentstudies demonstrate that a4b2� nAChRs play a role in the media-tion of analgesic effects, based on the observation that there is agood correlation between the analgesic efficacy and the a4b2�

affinity alone. These findings are consistent with our suggestedrole for desensitization since the affinity of agonists for nAChRshas been shown to influence both the onset of, and recovery fromdesensitization (Wang and Sun, 2005). The additional positive cor-relation between analgesic effects and the number of H-bond do-nors may be the consequence of the greater blood–brain barrierpenetration (Modi, 2003).

As an extension of previous findings noting the involvement ofagonism at a4b2� receptors in analgesia, we report herein the no-vel finding that desensitization of a4b2� receptors may drive partof the antinociceptive outcome. Using a formalin assay in mice,we observed that Sazetidine-A, which is known as a desensitizerinstead of activator of a4b2� receptors (Cucchiaro et al., 2008; Xiaoet al., 2006), displays robust analgesic effects both for Phase I andPhase II. Likewise, Compound 10 demonstrates good analgesic ef-fects, activates a4b2� receptors poorly, but desensitizes them po-tently. Our molecular modeling approaches revealed that whenreceptor desensitization rather than activation activities at a4b2�

receptors are considered, there is a better correlation betweenanalgesia scores and combined in vitro properties. Exceptions tothis correlation are Compounds 11, 12, 13 and 17 which all po-tently desensitize a4b2� receptors but are ineffective in the forma-lin model. Table 3 shows that the brain levels of Compounds 11–13are low at both time points examined. Due to poor brain perme-ability, these compounds may not have reached critical levels nec-essary for activity at the central site of action. The brainconcentration of Compound 17 is at the other end of the spectrum,reaching micromolar levels. At the 3 mg/kg, s.c. dose tested in theformalin model, the exposure levels of Compound 17 may havebeen too high, exceeding the levels needed for activity in the assay.Unfortunately the 3 mg/kg dose was the lowest tested in the modelfor this compound, not allowing us to determine if a lower dose,possibly one achieving nanomolar levels in the brain approximat-ing its DC50, may have been efficacious.

Our results suggest that although all three a4b2a5, HS- andLS-a4b2 subtypes assessed are involved, it is desensitization of

Score I = 159.9 − 58.8H_Donor–26.6Log (Ki)–17.8 Log (DC50)

r²=0.79

0

20

40

60

80

100

0 20 40 60 80 100 120 140Analgesia Score_Observed

Phase IIr²=0.70

-20

0

20

40

60

80

100

0 20 40 60 80 100 120 140

Ana

lges

ia S

core

_Pre

dict

ed

Analgesia Score_Observed

Phase I

Score II = 134.6 − 40.2 H_Donor–19.5Log (Ki) –25.6 Log (DC50)

(A) (B)

Fig. 4. Modeling of analgesic effects of novel nicotinic compounds using desensitization potencies at a4b2a5 receptors. Considering the Ki obtained with rat brain tissue, thenumber of H-bond donors (H_Donor) and the DC50 at ha4b2a5 nAChRs, models were generated to quantitatively predict analgesia scores separately for Phase I (A) and PhaseII (B) of the mouse formalin test. Mathematical equations and coefficients of determination (r2) are listed for each Phase. Significant correlations were observed betweenpredicted and observed analgesia scores for both Phases of the formalin assay.

822 J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823

a4b2a5 receptors that seems to play a more prominent role in theantinociceptive action of nicotinic compounds. Correlations forPhase I are significantly improved from an r2 value of 0.53 to0.67 and 0.66 when HS- and LS-a4b2 DC50 values are considered,respectively. More profoundly, considering the DC50 at a4b2a5takes the r2 from 0.53 to 0.70, placing a focus on the involvementof the a5-containing receptor. For Phase II analgesia scores, addingHS- or LS-a4b2 desensitization potencies did not improve the cor-relations significantly. Considering the a4b2a5 DC50 value signif-icantly increased the r2 from 0.70 to 0.79 for Phase II, andstrongly suggested a more prominent role for a4b2a5 nAChRs inthe modulation of pain in the formalin assay.

Phase I responses in the formalin assay are generally attributedto activation of peripheral nociceptors and Phase II responses areprimarily due to a central loci of action (Alsharari et al., 2012).With this understanding and the significant effects observed whenconsidering desensitization at each of the a4b2 subtypes, one maysuspect a role for all three receptor subtypes in the periphery inrelation to Phase I formalin responses, but a more prominent rolefor a4b2a5 receptors centrally during Phase II responses.

The significance of the desensitization results with the a4b2a5subtype are supported by previous reports showing that activationof a4b2� receptors is necessary but not sufficient to produce analge-sia (Gao et al., 2010). Additionally, a5� receptors have been shownto be expressed in the spinal cord (Vincler and Eisenach, 2005), akey site in pain pathways, and have been previously implicated inanalgesia (Jackson et al., 2010; Vincler and Eisenach, 2004, 2005).

How might desensitization be promoting analgesia? Is desensi-tization the mechanism or is it a marker for other nicotinic recep-tor activity? During prolonged exposure to a4b2�-selectiveagonists, changes in receptor composition may be induced (Govindet al., 2012; Lester et al., 2009; Fenster et al., 1999.). Altered incor-poration of a5 subunits into a4b2-containing receptors, or modu-lation of a4b2a5 receptor levels or ratios may be realized (Mao etal., 2008). With such alteration of receptors, changes in functionmay occur since a4b2a5 subtypes are more sensitive to desensiti-zation than to activation (Brown et al., 2007; Gerzanich et al.,1998; Kuryatov et al., 2008). Desensitization of a4b2� receptorsin return may lead to increased opioid release, decreased serotoninrelease, changes in other neurotransmitters or second messengersthat can produce analgesic effects at the supraspinal, spinal or pri-mary afferents level (Cucchiaro et al., 2008; Curzon et al., 1998;Tassonyi et al., 2002). Recent reports have also described a co-ago-nist character of nicotinic agonists (Prickaerts et al., 2012; Zwart

and Vijverberg, 2000). This property has been described at concen-trations below those needed for intrinsic agonism; levels similar tothose at which desensitization is observed (Fenster et al., 1999;Marks et al., 2010). Desensitization may be a surrogate measurefor compounds that may act as a co-agonist in vivo, enhancingthe activity of the endogenous ligand acetylcholine. At this pointin our understanding of nicotinic receptors in pain, one shouldnot discount the potential that nicotinic ligands may utilize multi-ple mechanisms in analgesia, including desensitization, receptorupregulation, co-agonism and intrinsic agonism.

We have described a role for desensitization in the modulationof Phase I and Phase II responses in the formalin model. Futurework should extend the analyses to additional pain models tounderstand the breadth of the role desensitization of nicotinicreceptors plays, especially that of a4b2a5. As ligands with differentnicotinic pharmacologic profiles may be effective in different painmodels, we may find that nicotinic receptor desensitization maycorrelate with effectiveness against some types of pain and notothers.

In summary, we have demonstrated in the present studies thatcompounds which are more potent at desensitization of a4b2�

nAChRs display better analgesia scores in the formalin test. Consis-tent with these observations, consideration of desensitizationproperties at a4b2� nAChRs, especially at a4b2a5 receptors, inMLR significantly improve correlations with efficacies of analgesia.Thus, a4b2� nAChR desensitization may contribute to efficacy inthe mediation of pain, and represent an additional mechanismfor analgesic effects mediated by nicotinic agonists. Accordingly,consideration of a4b2a5 receptor desensitization during the dis-covery process should aid in the identification of nicotinic ligandsthat have potential as pain therapeutics.

Acknowledgements

We thank M. Kiser and L.A. Guerra for generating binding and Caflux data, respectively; J.M. Stukes, S.P. Wene and RC Pritchard forcollecting in vivo concentration data; and Drs. J.P. Strachan, S. Akir-eddy, B. Bhatti, S. Breining, M. Melvin and L. Miao, and R. Heemstra,R. Whitaker and T. Showalter for their efforts to synthesize novelnicotinic compounds. All of them are employees at Targacept Inc.We also thank Dr. M.I. Damaj, Professor of Pharmacology at VirginiaCommonwealth University, for his constructive comments andsuggestions.

J. Zhang et al. / European Journal of Pharmaceutical Sciences 47 (2012) 813–823 823

References

AlSharari, S.D., Carroll, F.I., McIntosh, J.M., Damaj, M.I., 2012. The antinociceptiveeffects of nicotinic partial agonists varenicline and sazetidine-A in murine acuteand tonic pain models. J. Pharmacol. Exp. Ther. 342, 742–749.

Bannon, A.W., Decker, M.W., Curzon, P., Buckley, M.J., Kim, D.J., Radek, R.J., Lynch,J.K., Wasicak, J.T., Lin, N.H., Arnold, W.H., Holladay, M.W., Williams, M., Arneric,S.P., 1998. ABT-594 [(R)-5-(2 azetidinylmethoxy)-2-chloropyridine]: a novel,orally effective antinociceptive agent acting via neuronal nicotinic acetylcholinereceptors: II. In vivo characterization. J. Pharmacol. Exp. Ther. 285, 787–794.

Benwell, M.E., Balfour, D.J., Anderson, J.M., 1988. Evidence that tobacco smokingincreases the density of (�)-[3H]nicotine binding sites in human brain. J.Neurochem. 50, 1243–1247.

Berman, J., Halm, K., Adkison, K., Shaffer, J., 1997. Simultaneous pharmacokineticscreening of a mixture of compounds in the dog using API LC/MS/MS analysisfor increased throughput. J. Med. Chem. 40, 827–829.

Brown, R.W., Collins, A.C., Lindstrom, J.M., Whiteaker, P., 2007. Nicotinic alpha5subunit deletion locally reduces high-affinity agonist activation withoutaltering nicotinic receptor numbers. J. Neurochem. 103, 204–215.

Carbone, A.L., Moroni, M., Groot-Kormelink, P.J., Bermudez, I., 2009. Pentamericconcatenated (alpha4)(2)(beta2)(3) and (alpha4)(3)(beta2)(2) nicotinicacetylcholine receptors: subunit arrangement determines functionalexpression. Br. J. Pharmacol. 156, 970–981.

Cucchiaro, G., Xiao, Y., Gonzalez-Sulser, A., Kellar, K.J., 2008. Analgesic effects ofSazetidine-A, a new nicotinic cholinergic drug. Anesthesiology 109, 512–519.

Curzon, P., Nikkel, A.L., Bannon, A.W., Arneric, S.P., Decker, M.W., 1998. Differencesbetween the antinociceptive effects of the cholinergic channel activators A-85380 and (±)-epibatidine in rats. J. Pharmacol. Exp. Ther. 287, 847–853.

Damaj, M.I., Fonck, C., Marks, M.J., Deshpande, P., Labarca, C., Lester, H.A., Collins,A.C., Martin, B.R., 2007. Genetic approaches identify differential roles foralpha4beta2� nicotinic receptors in acute models of antinociception in mice. J.Pharmacol. Exp. Ther. 321, 1161–1169.

Dubuisson, D., Dennis, S.G., 1977. The formalin test: a quantitative study of theanalgesic effects of morphine, meperidine and brain stem stimulation in ratsand cats. Pain 4, 161–174.

Fenster, C.P., Whitworth, T.L., Sheffield, E.B., Quick, M.W., Lester, R.A., 1999.Upregulation of surface alpha4beta2 nicotinic receptors is initiated by receptordesensitization after chronic exposure to nicotine. J. Neurosci. 19, 4804–4814.

Flores, C.M., 2000. The promise and pitfalls of a nicotinic cholinergic approach topain management. Pain 88, 1–6.

Gao, B., Hierl, M., Clarkin, K., Juan, T., Nguyen, H., Valk, M., Deng, H., Guo, W., Lehto,S.G., Matson, D., McDermott, J.S., Knop, J., Gaida, K., Cao, L., Waldon, D., Albrecht,B.K., Boezio, A.A., Copeland, K.W., Harmange, J.C., Springer, S.K., Malmberg, A.B.,McDonough, S.I., 2010. Pharmacological effects of nonselective and subtype-selective nicotinic acetylcholine receptor agonists in animal models ofpersistent pain. Pain 149, 33–49.

Gerzanich, V., Wang, F., Kuryatov, A., Lindstrom, J., 1998. Alpha 5 Subunit altersdesensitization, pharmacology, Ca++ permeability and Ca++ modulation ofhuman neuronal alpha 3 nicotinic receptors. J. Pharmacol. Exp. Ther. 286, 311–320.

Govind, A.P., Walsh, H., Green, W.N., 2012. Nicotine-induced upregulation of nativeneuronal nicotinic receptors is caused by multiple mechanisms. J. Neurosci. 32,2227–2238.

Jackson, K.J., Marks, M.J., Vann, R.E., Chen, X., Gamage, T.F., Warner, J.A., Damaj, M.I.,2010. Role of alpha5 nicotinic acetylcholine receptors in pharmacological andbehavioral effects of nicotine in mice. J. Pharmacol. Exp. Ther. 334, 137–146.

Jain, K.K., 2004. Modulators of nicotinic acetylcholine receptors as analgesics. Curr.Opin. Investig. Drugs 5, 76–81.

Ji, J., Schrimpf, M.R., Sippy, K.B., Bunnelle, W.H., Li, T., Anderson, D.J., Faltynek, C.,Surowy, C.S., Dyhring, T., Ahring, P.K., Meyer, M.D., 2007. Synthesis andstructure-activity relationship studies of 3,6-diazabicyclo[3.2.0]heptanes asnovel alpha4beta2 nicotinic acetylcholine receptor selective agonists. J. Med.Chem. 50, 5493–5508.

Jolliffe, I.T., 1986. Principal Component Analysis. Springer-Verlag New York, Inc..Karl, T., Pabst, R., von Hörsten, S., 2003. Behavioral phenotyping of mice in

pharmacological and toxicological research. Exp. Toxicol. Pathol. 55, 69–83.Kuryatov, A., Onksen, J., Lindstrom, J., 2008. Roles of accessory subunits in

alpha4beta2(�) nicotinic receptors. Mol. Pharmacol. 74, 132–143.Lester, H.A., Xiao, c., Srinivasan, R., Son, C.D., Miwa, J., Pantoja, R., Banghart, M.R.,

Dougherty, D.A., Goate, A.M., Wang, J.C., 2009. Nicotine is a selective

pharmacological chaperone of acetylcholine receptor number andstoichiometry. Implications for drug discovery. AAPS J. 11, 167–177.

Malmberg, A.B., Bannon, A.W., 2007. Models of nociception: hot-plate, tail-flick andformalin tests in rodents. In: Crawley, J.N., Gerfen, C.R., Rogawski, M.A., Sibley,D.R., Skolnick, P., Wray, S. (Eds.), Current Protocols in Neuroscience. John Wiley& Sons, pp. 8.9.1–8.9.16.

Mao, D., Perry, D.C., Yasuda, R.P., Wolfe, B.B., Kellar, K.J., 2008. Thealpha4beta2alpha5 nicotinic cholinergic receptor in rat brain is resistant toup-regulation by nicotine in vivo. J. Neurochem. 104, 446–456.

Marks, M.J., Stitzel, J.A., Romm, E., Wehner, J.M., Collins, A.C., 1986. Nicotinic bindingsites in rat and mouse brain: comparison of acetylcholine, nicotine, and alpha-bungarotoxin. Mol. Pharmacol. 30, 427–436.

Marks, M.J., Meinerz, N.M., Brown, R.W., Collins, A.C., 2010. 86Rb+ efflux mediatedby alpha4beta2�-nicotinic acetylcholine receptors with high and low-sensitivityto stimulation by acetylcholine display similar agonist-induced desensitization.Biochem. Pharmacol. 80, 1238–1251.

Marubio, L.M., del Mar Arroyo-Jimenez, M., Cordero-Erausquin, M., Lena, C., LeNovere, N., de Kerchove d’Exaerde, A., Huchet, M., Damaj, M.I., Changeux, J.P.,1999. Reduced antinociception in mice lacking neuronal nicotinic receptorsubunits. Nature 398, 805–810.

Modi, S., 2003. Computational approaches to the understanding of ADMETproperties and problems. Drug Discov. Today 8, 621–623.

Moroni, M., Zwart, R., Sher, E., Cassels, B.K., Bermudez, I., 2006. Alpha4beta2nicotinic receptors with high and low acetylcholine sensitivity: pharmacology,stoichiometry, and sensitivity to long-term exposure to nicotine. Mol.Pharmacol. 70, 755–768.

Nguyen, H.N., Rasmussen, B.A., Perry, D.C., 2003. Subtype-selective up-regulation bychronic nicotine of high-affinity nicotinic receptors in rat brain demonstratedby receptor autoradiography. J. Pharmacol. Exp. Ther. 307, 1090–1097.

Papke, R.L., Wecker, L., Stitzel, J.A., 2010. Activation and inhibition of mouse muscleand neuronal nicotinic acetylcholine receptors expressed in Xenopus oocytes. J.Pharmacol. Exp. Ther. 333, 501–518.

Prickaerts, J., van Goethem, N.P., Chesworth, R., Shapiro, G., Boess, F.G., Methfessel,C., Reneerkens, O.A., Flood, D.G., Hilt, D., Gawryl, M., Bertrand, S., Bertand, D.,Konig, G., 2012. EVP-6124, a novel and selective a7 nicotinic acetylcholinereceptor partial agonist, improves memory performance by potentiating theacetylcholine response of a7 nicotinic acetylcholine receptors.Neuropharmacology 62, 1099–1110.

Rowbotham, M.C., Duan, W.R., Thomas, J., Nothaft, W., Backonja, M.M., 2009. Arandomized, double-blind, placebo-controlled trial evaluating the efficacy andsafety of ABT-594 in patients with diabetic peripheral neuropathic pain. Pain146, 245–252.

Tassonyi, E., Charpantier, E., Muller, D., Dumont, L., Bertrand, D., 2002. The role ofnicotinic acetylcholine receptors in the mechanisms of anesthesia. Brain Res.Bull. 57, 133–150.

Vincler, M., Eisenach, J.C., 2004. Plasticity of spinal nicotinic acetylcholine receptorsfollowing spinal nerve ligation. Neurosci. Res. 48, 139–145.

Vincler, M.A., Eisenach, J.C., 2005. Knock down of the alpha 5 nicotinic acetylcholinereceptor in spinal nerve-ligated rats alleviates mechanical allodynia. Pharmacol.Biochem. Behav. 80, 135–143.

Wang, H., Sun, X., 2005. Desensitized nicotinic receptors in brain. Brain Res. BrainRes. Rev. 48, 420–437.

Wang, Y., Sun, J., Liu, H., He, Z., 2007. Rapidly profiling blood–brain barrierpenetration with liposome EKC. Electrophoresis 28, 2391–2395.

Xiao, Y., Fan, H., Musachio, J.L., Wei, Z.L., Chellappan, S.K., Kozikowski, A.P., Kellar,K.J., 2006. Sazetidine-A, a novel ligand that desensitizes alpha4beta2 nicotinicacetylcholine receptors without activating them. Mol. Pharmacol. 70, 1454–1460.

Yogeeswari, P., Sriram, D., Ratan Bal, T., Thirumurugan, R., 2006. Epibatidine and itsanalogues as nicotinic acetylcholine receptor agonist: an update. Nat. Prod. Res.20, 497–505.

Zhou, Y., Nelson, M.E., Kuryatov, A., Choi, C., Cooper, J., Lindstrom, J., 2003. Humanalpha4beta2 acetylcholine receptors formed from linked subunits. J. Neurosci.23, 9004–9015.

Zwart, R., Vijverberg, H.P., 1998. Four pharmacologically distinct subtypes ofalpha4beta2 nicotinic acetylcholine receptor expressed in Xenopus laevisoocytes. Mol. Pharmacol. 54, 1124–1131.

Zwart, R., Vijverberg, H.P., 2000. Potentiation and inhibition of neuronalalpha4beta4 nicotinic acetylcholine receptors by choline. Eur. J. Pharmacol.393, 209–214.