Structurally Similar Allosteric Modulators of α7 Nicotinic Acetylcholine Receptors Exhibit Five Distinct Pharmacological Effects*

Background: Nicotinic receptors are activated by acetylcholine and have been implicated in several neurological disorders. Results: Allosteric modulators, sharing close chemical similarity, exhibit five distinct pharmacological effects on α7 nicotinic receptors. Conclusion: Small changes in chemical structure have profound effects on the pharmacological properties of allosteric modulators. Significance: These findings may provide opportunities for novel approaches to therapeutic drug discovery.

neurotransmitter-gated ion channels (1,2). In common with other Cys loop receptors, nAChRs are pentameric complexes in which five transmembrane subunits co-assemble to form a central ion channel (2,3). Although all nAChRs share a similar three-dimensional structure (4,5), there is considerable diversity in their subunit composition. For example, in mammalian species, there are 16 nAChR subunits (␣1-␣7, ␣9, ␣10, ␤1-␤4, ␥, ␦, and ⑀) that can co-assemble to generate a diverse family of nAChR subtypes (6,7). In addition, some nAChR subunits (such as ␣7) form functional homomeric nAChRs, containing five copies of the same subunit (8). The human ␣7 nAChR has been identified as a potential target for therapeutic drug discovery and has been implicated in a number of neurological and psychiatric disorders (9 -13). The ␣7 nAChR is somewhat atypical of this receptor family, in that it undergoes very rapid desensitization in response to activation by its endogenous neurotransmitter acetylcholine (8). However, allosteric modulation can facilitate activation of ␣7 nAChRs with only low levels of desensitization (14 -17). Activation of nAChRs and the opening of the cation-selective pore is associated with the binding of agonists such as acetylcholine to an orthosteric binding site located in the extracellular domain of the receptor at the interface between two adjacent subunits (18,19). In addition, a range of compounds have been identified that can modulate nAChR activation by binding to distinct allosteric sites (20) and may have potential as novel approaches to therapeutic drug discovery (21). An area that has attracted considerable interest concerns compounds that lack intrinsic agonist activity on nAChRs but are able to potentiate agonist-evoked responses by binding to a distinct allosteric site. Such compounds have been described as positive allosteric modulators (PAMs) and include compounds such as TQS, a PAM that displays selectivity for ␣7 nAChRs (15). In the case of rapidly desensitizing nAChRs such as ␣7, PAMs have been classified as being either type I or type II, depending on whether they cause little or no effect on the rate of agonistinduced desensitization (type I PAMs) or cause a reduction in desensitization (type II PAMs) (20,22,23). In addition, there is evidence that the binding of ligands to allosteric sites on nAChRs can result in efficient activation of nAChRs in the absence of orthosteric agonists (17,24). In the case of ␣7 nAChRs, there is evidence for PAMs and allosteric agonists binding to an intrasubunit cavity located within the transmembrane domain (17,25,26).
We have shown previously that minor changes in the chemical structure of nAChR allosteric modulators can result in dramatic differences in pharmacological properties (17,24). For example, replacing a fluorine atom with a chlorine atom converts an ␣7-selective PAM into an allosteric agonist (24). The present study extends these findings, with the aim of identifying the influence of changes in chemical structure on the pharmacological properties of nAChR allosteric modulators. A series of compounds, differing only in methyl substitution of a single aromatic ring, have been examined (see Fig. 1). This series includes an ␣7-selective allosteric agonist, 4MP-TQS, a compound that has been examined previously (24), together with all other possible combinations of methyl substitutions at a single phenyl ring (18 additional compounds). The influence of changes in chemical structure has been examined on ␣7 nAChRs by means of two-electrode voltage-clamp recording and radioligand binding. Whereas previous studies of compounds with close chemical similarity to TQS have identified only allosteric agonists or type II PAMs (15,17,24), studies conducted with this series of 19 methyl-substituted compounds have revealed five distinct pharmacological effects on ␣7 nAChRs. In summary, the 19 methyl-substituted compounds examined in this study can be classified in one of five categories: allosteric agonists, type I PAMs, type II PAMs, negative allosteric modulators (NAMs), or silent allosteric modulators (SAMs).

EXPERIMENTAL PROCEDURES
Chemical Synthesis-Nineteen compounds were synthesized that differed only in the pattern of methyl substitution of an aromatic substituent located at position 2 of a tetrahydroisoquinoline ring (see Fig. 1). Compounds were prepared by InCl 3catalyzed reaction of sulfanilamide, cyclopentadiene, and the corresponding substituted benzaldehyde, according to methods described previously (27). The substituted benzaldehydes were either purchased commercially or prepared according to literature procedures (28). In most cases (13 of the 19 compounds), the cis-cis-diastereoisomer was obtained as the major product from the InCl 3 -catalyzed multicomponent reaction (see Fig. 1). However, in some cases (6 of the 19 compounds), the cis-trans-diastereoisomer was obtained as the major product (see Fig. 1). For all cis-trans-isomers and for two of the cis-cis-isomers, only a single diastereoisomer was detected by 1 H NMR. Further details concerning the synthesis of these compounds and the major diastereoisomer obtained for each compound, together with the ratio of isomers present in the purified sample, are provided in the supplemental materials.
Xenopus laevis Oocyte Electrophysiology-Oocytes were isolated from female X. laevis and defolliculated as described previously (33). Heterologous expression was achieved by injection of either cRNA (6 -12 ng) into oocyte cytoplasm in the case of wild-type and mutated ␣7 or plasmid cDNA constructs (10 -30 ng) into oocyte nuclei in the case of 5-HT 3A . In vitro transcription of cRNA was carried out using mMESSAGE mMACHINE SP6 transcription kit (Ambion, Huntington, UK). Oocytes were injected in a volume of 32.2 nl using a Drummond variable volume microinjector. Two electrode voltage-clamp recordings were performed (with the oocyte membrane potential held at Ϫ60 mV), as described previously (33) using a Warner Instruments OC-725C amplifier (Harvard Apparatus, Edenbridge, UK), PowerLab 8SP, and Chart 5 software (AD Instruments, Oxford, UK). Methyl-TQS compounds were dissolved in DMSO to generate 100 mM stock solutions. Compounds were applied to oocytes using a BPS-8 solenoid valve solution exchange system (ALA Scientific Inc., Westbury, NY), con-trolled by Chart software. For multiple comparisons of agonist activation rates, statistical significance was determined with a one-way analysis of variance (ANOVA). Statistical significance of desensitization rates was determined by paired Student's t tests. A p value of Ͻ0.05 was considered significant. The activation and desensitization phases of current responses were best fitted by a single exponential function.
Radioligand Binding-Radioligand binding to transiently transfected tsA201 cells was performed as described previously (30,34) with Radioligand binding was assayed by filtration onto Whatman GF/A filters (presoaked in 0.5% polyethylenimine), followed by rapid washing with phosphate-buffered saline (Oxoid) using a Brandel cell harvester. Bound radioligand was determined by scintillation counting.

RESULTS
Nineteen compounds were synthesized that share close chemical similarity to one another but form a series containing all possible combinations of methyl substitution on a single aromatic ring ( Fig. 1). The majority of these compounds were obtained as the cis-cis-diastereoisomer but with varying degrees of diastereoselectivity ( Fig. 1). However, those compounds containing an aromatic ring bearing two ortho methyl groups were obtained as the cistrans-isomer with very high selectivity (Fig. 1). The pharmacological properties of all 19 compounds were examined on human ␣7 nAChRs expressed in Xenopus oocytes.
Allosteric Agonist Activation of ␣7 nAChRs-Of the 19 methylsubstituted compounds examined, seven of these (2,3MP-TQS, 2,3,4MP-TQS, 2,3,4,5MP-TQS, 2,4MP-TQS, 3,4MP-TQS, 3,4,5MP-TQS, and 4MP-TQS) were found to have properties typical of ␣7 nAChR allosteric agonists (Fig. 2). In contrast to the rapidly desensitizing responses observed with orthosteric agonists such as acetylcholine, all seven of these methyl-substituted compounds activated ␣7 nAChRs with very much reduced levels of desensitization (Fig. 2B). Activation by the methyl-substituted compounds resulted in responses that were similar to responses with allosteric agonists of ␣7 nAChRs examined previously (17,24,35). Responses had a slow onset, were slow to reach a plateau, and had very different kinetics to activation by acetylcholine, which causes rapid desensitization (Fig. 2B). Agonist activation rates were analyzed from the start of agonist application to the peak response ( Fig. 2A and Table  1). The rate of activation by all of the allosteric agonists examined was significantly slower compared with activation by acetylcholine (p Ͻ 0.01; Table 1). In addition, two of the allosteric agonists (3,4MP-TQS and 4MP-TQS) had significantly slower activation rates than the other five allosteric agonists (p Ͻ 0.05; Table 1).
None of the PAMs examined had a significant influence on the rate of receptor activation compared with acetylcholine alone. Although the activation rates did not differ, differences were observed in the rate of desensitization of ␣7 nAChRs with these PAMs, as has been described previously with other PAMs (24). Two of the compounds (2,3,4,6MP-TQS and PentaMP-TQS) displayed properties characteristic of type I PAMs, having no significant difference on the rate of receptor desensitization compared with acetylcholine (p Ͼ 0.05; Fig. 3A and Table 2). The other six compounds (2MP-TQS, 2,3,5MP-TQS, 2,4,5MP-TQS, 2,5MP-TQS, 3MP-TQS, and 3,5MP-TQS) caused significant slowing in the rate of receptor desensitization compared with acetylcholine (p Ͻ 0.05; Fig. 4A and Table 2), a property that is characteristic of type II PAMs.
Influence of the Transmembrane L247T Mutation-Mutations of amino acid Leu-247 (such as L247T) in ␣7 nAChRs have been found to have dramatic effects on the functional properties of the receptor (36). The L247T mutation causes some competitive antagonists to act as agonists (37), and it also causes some PAMs, such as TQS, to act as agonists (17). The influence of the L247T mutation was examined on those compounds that displayed no agonist activity and neither potentiated nor inhibited acetylcholine-evoked responses (2,3,5,6MP-TQS and 2,4,6MP-TQS). Interestingly, it was found that the L247T mutation converts both 2,3,5,6MP-TQS and 2,4,6MP-TQS, which are silent allosteric modulators on wild-type ␣7 nAChRs, into weak nondesensitizing agonists, a property that is characteristic of allosteric agonists acting on ␣7 nAChRs (Fig.  6D). This supports the conclusion that 2,3,5,6MP-TQS and 2,4,6MP-TQS are binding at the previously identified transmembrane allosteric site (17,25) and should be classified as SAMs.
Influence of the Transmembrane M253L Mutation-Recent studies have proposed that ␣7-selective PAMs and allosteric agonists can act via a transmembrane binding site (17,25). One of the lines of evidence supporting this proposal is that potentiation by TQS and also activation by 4BP-TQS is not observed with ␣7 receptors containing the transmembrane mutation M253L (17). In contrast, M253L has been shown to have no significant effect on activation by the conventional orthosteric agonist acetylcholine (17,25).
The effect of the M253L mutation was also examined on those compounds that act as antagonists of acetylcholine (2,3, 6MP-TQS and 2,6MP-TQS) (Fig. 5D). Responses evoked by an EC 50 concentration of acetylcholine (100 M) were inhibited by 64 Ϯ 4.0% with 2,3,6MP-TQS and by 73 Ϯ 1.7% with 2,6MP-TQS. These levels of inhibition are not significantly different to that observed with wild-type ␣7 nAChRs (Student's t test, p Ͼ  0.05). We note, however, that the M253L mutation was generated so as to introduce into the nAChR ␣7 subunit the amino acid occurring at the analogous position in the mouse 5-HT3A subunit (17,25). For this reason, the effect of 2,3,6MP-TQS and 2,6MP-TQS was examined on the mouse 5-HT3A subunit expressed in oocytes. A maximal concentration of either 2,3,6MP-TQS or 2,6MP-TQS was preapplied and then co-applied with the 5-HT 3 receptor (5-HT 3 R) agonist 1-(3-chlorophenyl)biguanide hydrochloride (CPBG) (1 M). Both TQS compounds resulted in inhibition of responses evoked by CPBG (Fig. 5D). The finding that 2,3,6MP-TQS and 2,6MP-TQS act as inhibitors of both ␣7 nAChRs and 5-HT 3 R may explain why the M253L mutation in ␣7 has no significant effect on allosteric modulation by 2,3,6MP-TQS or 2,6MP-TQS but has a dramatic effect on modulation by other chemically similar allosteric modulators. Radioligand Binding-The dramatic effect of the M253L mutation on those methyl-substituted compounds displaying agonist or PAM activity helps to provide support for the conclusion that these compounds may act at a previously proposed transmembrane allosteric site (17,25). In contrast, the common inhibitory effect of 2,3,6MP-TQS and 2,6MP-TQS on ␣7 nAChRs and on 5-HT 3A receptors means that no conclusions can be made from studies with the M253L mutation concerning the two methyl-substituted compounds that display antagonist effects on responses evoked by acetylcholine. Our working hypothesis is that all of the structurally similar methyl-substituted compounds examined in this study are likely to interact with a broadly similar allosteric site on the ␣7 nAChR, as discussed previously for other related compounds (17,24). A consequence of this hypothesis would be that the antagonism of acetylcholine-evoked responses that has been observed with 2,3,6MP-TQS and 2,6MP-TQS is due to a noncompetitive mechanism of action. With the aim of testing this hypothesis, we examined the ability of methyl-substituted compounds to displace the binding of an orthosteric radioligand ([ 3 H]␣-bun-  Fig. 7. This suggests that none of the methyl-substituted compounds bind competitively at the orthosteric nicotinic ligand-binding site. In contrast, as has been reported previously (38), the competitive antagonist MLA causes complete displacement of [ 3 H]␣-bungarotoxin from ␣7 nAChRs (IC 50 ϭ 37 Ϯ 4.8 nM; Fig. 7). This provides support for our conclusion that the antagonism of acetylcholine-evoked responses by 2,3,6MP-TQS and 2,6MP-TQS is by a noncompetitive (allosteric) mechanism and that all of the methyl-substituted compounds examined in this study act at a site distinct from the conventional orthosteric binding site. It is also consistent with our evidence that the antagonism of acetylcholineevoked responses by 2,3,6MP-TQS and 2,6MP-TQS is not surmountable (Fig. 5C).
Several of the compounds examined in the present study lack allosteric agonist activity but cause dramatic potentiation of responses evoked by acetylcholine (PAM activity). Based on their effect on the rate of agonist-evoked desensitization, compounds were identified that fulfill the criteria of both type I PAMs (which have no effect on receptor desensitization) and type II PAMs (which cause a slowing of desensitization were seen within the series). Two of the compounds examined (2,3,4,6MP-TQS and PentaMP-TQS) resembled type I PAMs,   whereas six of the compounds (2MP-TQS, 2,3,5MP-TQS, 2,4,5MP-TQS, 2,5MP-TQS, 3MP-TQS, and 3,5MP-TQS) resembled type II PAMs. Compounds belonging to both groups of PAM were effective in shifting the potency of acetylcholine to the left and increasing the maximum efficacies, a phenomenon that is characteristic of nAChR PAMs (15, 39 -41).
Work described here and elsewhere supports the conclusion that allosteric agonists and PAMs of ␣7 nAChRs can bind to a common transmembrane site (25,26). This is supported by evidence that the effects of both allosteric agonists and of PAMs can be blocked completely by a single point mutation (M253L) located in the transmembrane region, a mutation that has no significant effect on agonist activation by the orthosteric agonist acetylcholine (17,25).
Two of the compounds examined in this series (2,3,6MP-TQS and 2,6MP-TQS) lacked allosteric agonist activity but caused dramatic inhibition of responses evoked by acetylcholine and also of responses evoked by an allosteric agonist (2,4MP-TQS). When these compounds were examined on the ␣7 M253L nAChR, the mutation was found to have no significant effect on the level of inhibition by these compounds. However, the M253L mutation was generated so as to introduce into the nAChR ␣7 subunit the amino acid occurring at the analogous position in the mouse 5-HT3A subunit (17,25). Our finding that responses evoked by the orthosteric agonist CPBG on 5-HT 3 Rs were inhibited by 2,3,6MP-TQS and 2,6MP-TQS may explain why the M253L mutation had no effect on the antagonist activity of these compounds on ␣7 nAChRs. This is because the M253L mutation was designed to change the existing methionine in ␣7 to the corresponding amino acid (leucine) at the analogous position in the mouse 5-HT3A subunit (25).
Evidence to support the conclusion that these antagonists bind noncompetitively with respect to acetylcholine is provided by data demonstrating that the antagonism by 2,3,6MP-TQS and 2,6MP-TQS is not surmountable (Fig. 5C). In addition, competition radioligand binding data provide further support for the conclusion that both 2,3,6MP-TQS and 2,6MP-TQS bind at a site other than the conventional orthosteric binding site. The structural similarity between 2,3,6MP-TQS and 2,6MP-TQS with that of other compounds differing only in methyl substitution of a phenyl ring (including allosteric agonists and PAMs) suggests that the most likely explanation for the inhibition of acetylcholine responses is that these compounds are binding noncompetitively with respect to acetylcholine and at a common or overlapping transmembrane site to that of the allosteric agonists and PAMs examined here.
As others have done previously (42,43), we have used the term "silent allosteric modulator" (SAM) to denote compounds that interact with an allosteric site but that do not exert a modulatory effect on responses to the orthosteric agonist (i.e. neither a positive or negative allosteric effect) and that do have modulatory effects on compounds that interact at the same allosteric site. Two of the compounds examined in this study (2,3,5,6MP-TQS and 2,4,6MP-TQS) exhibited no allosteric agonist activity and neither potentiated nor inhibited responses evoked by acetylcholine. They did, however, act as antagonists of allosteric agonists. The simplest explanation for this observation is that 2,3,5,6MP-TQS and 2,4,6MP-TQS interact with the same allosteric site as allosteric agonists such as 4MP-TQS (17,24) and can therefore be considered as being SAMs. The fact that neither compound caused displacement of [ 3 H]␣-bungarotoxin from its orthosteric binding site supports this conclusion. In this respect, they can be considered as acting in a manner that is analogous to previously described SAMs of G protein-coupled receptors (42,43). Similarly, it has been shown previously that ␣7 nAChR PAMs such as TQS can act as antagonists of responses evoked by allosteric agonists such as 4BP-TQS (24,35), presumably because they are binding competitively to a common site. The two compounds that acted as SAMs on wild-type ␣7 nAChRs (2,3,5,6MP-TQS and 2,4,6MP-TQS) both displayed agonist activity on mutant ␣7 L247T nAChRs. The L247T mutation is known to have a higher frequency of spontaneous openings (44), so it is possible that binding of 2,3,5,6MP-TQS or 2,4,6MP-TQS stabilizes the open conformation of the receptor, resulting in agonist activity.
Synthesis of the methyl-substituted compounds in the InCl 3catalyzed reaction can potentially result in of the formation of up to four different diastereoisomers. Typically, however, only two isomers are observed, classified as cis-cis or cis-trans on the basis of the relative stereochemistry of the three stereocenters on the isoquinoline ring (Fig. 1). The stereochemistry of the major isomer of each of the methyl-substituted compounds described in this study was assigned on the basis of the coupling constants observed in the 1 H NMR (see supplemental materials for details). In most cases, the cis-cis-isomer was obtained as the major product, but reactions with more hindered benzaldehydes, containing two methyl groups ortho to the aldehyde group, gave the cis-trans-isomer as the major product (Fig. 1). It is likely that the presence of these two methyl groups leads the corresponding imine to adopt a very different conformation to avoid steric clash between the imine and the nearby methyl groups. This in turn is likely to raise the energy of the transition state, leading to the formation of the relatively hindered cis-cisisomer during the cyclization reaction. As a consequence, the formation of the cis-trans-isomer becomes preferred to the extent that it is the sole product observed with these more hindered imines. Interestingly, all of the compounds that have been classified as allosteric agonists or as type II PAMs were found to be cis-cis-diastereoisomers, whereas all of the NAMs, SAMs, and type I PAMs were cis-trans-diastereoisomers.
Whereas we have previously obtained evidence that the size of the group attached to position 4 of the phenyl ring of TQS compounds can influence allosteric properties (24), our present findings indicate the role of groups attached at positions 2 and 6. These conclusions about the importance of positions 2 and 6 are also consistent with data obtained from all of the compounds similar to TQS that we have examined previously, all of which are either allosteric agonists or type II PAMs (17). All of these compound are cis-cis-diastereoisomers, and none contain substituents at both positions 2 and 6 (17). Our findings are also consistent with recent studies demonstrating that the cis-cis-(ϩ)-enantiomer of 4BP-TQS (GAT107) is active as an allosteric agonist (45,46).
In conclusion, we have synthesized a series of 19 compounds, containing all possible variations of methyl substitution at a single aromatic ring. Whereas previous studies of compounds with close chemical similarity to TQS have identified only allosteric agonists or type II PAMs (15,17,24), studies conducted with this series of 19 methyl-substituted compounds have revealed five distinct pharmacological effects on ␣7 nAChRs. In addition to allosteric agonists and type II PAMs, we have identified type I PAMs and also compounds that reduce agonist-evoked responses and can be considered as NAMs (are, alternatively, as noncompetitive antagonists). Finally, compounds have been identified that have no significant effect on orthosteric agonist-evoked responses but block responses to allosteric agonists (classified as SAMs). In summary, the 19 methyl-substituted compounds examined in this study (Fig. 1) can be classified in one of five categories: allosteric agonists (7 of the 19 compounds), type I PAMs (2 compounds), type II PAMs (6 compounds), NAMs (2 compounds), or SAMs (2 compounds). The data we have obtained are consistent with all of these compounds interacting with a common transmembrane allosteric site, as has been proposed previously for other allosteric modulators of ␣7 nAChRs (17, 24 -26).