Neurochem Res DOI 10.1007/s11064-013-1226-6

OVERVIEW

Probing the Orthosteric Binding Site of GABAA Receptors with Heterocyclic GABA Carboxylic Acid Bioisosteres Jette G. Petersen • Rikke Bergmann • Povl Krogsgaard-Larsen • Thomas Balle Bente Frølund

•

Received: 6 September 2013 / Revised: 9 December 2013 / Accepted: 11 December 2013 Ó Springer Science+Business Media New York 2013

Abstract The ionotropic GABAA receptors (GABAARs) are widely distributed in the central nervous system where they play essential roles in numerous physiological and pathological processes. A high degree of structural heterogeneity of the GABAAR has been revealed and extensive effort has been made to develop selective and potent GABAAR agonists. This review investigates the use of heterocyclic carboxylic acid bioisosteres within the GABAAR area. Several heterocycles including 3-hydroxyisoxazole, 3-hydroxyisoxazoline, 3-hydroxyisothiazole, and the 1- and 3-hydroxypyrazole rings have been employed in order to map the orthosteric binding site. The physicochemical properties of the heterocyclic moieties making them suitable for bioisosteric replacement of the carboxylic acid in the molecule of GABA are discussed. A variety of synthetic strategies for synthesis of the heterocyclic scaffolds are available. Likewise, methods for introduction of substituents into specific positions of the heterocyclic scaffolds facilitate the investigation of different regions in the orthosteric binding pocket in close vicinity of the core scaffolds of muscimol/GABA. The development of structural models, from pharmacophore models to receptor homology models, has provided more insight into the molecular basis for binding. Similar binding modes are proposed for the heterocyclic GABA

J. G. Petersen
R. Bergmann
P. Krogsgaard-Larsen
B. Frølund (&) Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark e-mail: [email protected] T. Balle Faculty of Pharmacy, The University of Sydney, Camperdown Campus, Sydney, NSW 2006, Australia

analogues covered in this review by use of ligand–receptor docking into the receptor homology model and the presented structure–activity relationships. A network of interactions between the analogues and the binding pocket is leaving no room for substituents and underline the limited space in the GABAAR orthosteric binding site when in the agonist conformation. Keywords GABA
Heterocyclic carboxylic acid bioisosteres
Structure–activity studies
Structural models

Introduction c-Aminobutyric acid (GABA) is an important neurotransmitter in the central nervous system (CNS) where it plays an inhibitory role dampening brain activity and balancing the effects of excitatory neurotransmitters such as glutamate and acetylcholine. Aberrant GABAergic function is known to be implicated in many disorders including anxiety, epilepsy, mood disorders, insomnia, cognitive disorders and schizophrenia [1, 2]. The effects of GABA are exerted through ionotropic GABAA receptors (GABAARs) and metabotropic GABAB receptors. The GABAARs belong to the Cys-loop receptor superfamily of ligand-gated ion channels also comprising the nicotinic acetylcholine receptors (nAChRs), 5-HT3 serotonin receptors, glycine receptors (GlyRs), and a zincactivated channel (ZAC) [3, 4]. The members of the Cysloop superfamily share a common structural arrangement with five subunits encircling a central ion conducting pore [5]. A schematic representation of the a1b2c2 GABAAR which is composed of two a1, two b2, and one c2 subunit is shown in Fig. 1. It contains an orthosteric binding site for

123

Neurochem Res

Fig. 1 Representation of the GABAA receptor. a View of the pentameric assembly of the a1b2c2 GABAAR. GABA and benzodiazepine (BZD) binding sites are indicated. b Side view of the GABAAR model, indicating the extracellular domain (ECD), the transmembrane domain (TMD), and the location of the GABA

Fig. 2 Structures of orthosteric GABAAR agonists: GABA, muscimol, thiomuscimol, THIP and 4-PIOL

GABA in each of the two b2:a1 interfaces and sites for allosteric modulators including the therapeutically important benzodiazepines in the a1:c2 interface [6]. From a pool of 19 different human subunits (a1–6, b1–3, c1–3, d, e, h, p, and q1–3), at least 26 native and mainly heteromeric GABAAR subtypes have been proposed, the a1b2c2, a2b3c2, a3b3c2 combinations being the predominant synaptic receptor subtypes [7]. GABAARs composed of q1-3 subunits assemble as homo- or pseudohomomers and are also known as the GABACRs [5]. The development of compounds capable of activating the GABAARs has decisively contributed to the understanding of the architecture and electrostatics of the orthosteric binding pocket. Conformational restrictions and bioisosteric replacements in the molecule of GABA have afforded a range of GABAAR agonists and antagonists with

123

binding site in the ECD. c A close-up of the GABA binding site at b2: a1 subunit interface. Regions containing important residues for ligand binding are indicated with colors and nominated according to the traditional nomenclature (Loops A–F)

different pharmacological profiles. Classical agonists include (Fig. 2): 5-(aminomethyl)isoxazol-3-ol (muscimol) [8], a naturally occurring compound that was isolated from the mushroom Amanita muscaria; thiomuscimol [9], the synthetic sulphur analogue of muscimol; 4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridine-3-ol (THIP) [10], and 5-(4piperidyl)-3-isoxazolol (4-PIOL) [11], a low efficacy partial agonist. Among the orthosteric agonists developed for the GABAAR there is a clear lack of subtype-selective compounds. Comprehensive attempts have been made to achieve subtype-selective agonists by manipulating potent agonists by means of introduction of substituents, conformational restriction and bioisosteric replacement. However, these efforts have been largely unsuccessful, although THIP interacts quite selectively with GABAARs containing d subunits [12, 13]. The GABAAR agonists reported to date are characterized by being relatively small and containing a basic amino group and an acidic moiety with strict limitations on their distance from each other [12, 13]. The considerable flexibility of GABA resulting from the free rotatability of single bonds has been restricted by introduction of compounds with rigid multiple bonds and ring systems. In particular, the conformational restriction of the structure of GABA by bioisosteric replacement of the carboxylic acid moiety with acidic heterocycles has been successful. This review will discuss heterocyclic GABAAR orthosteric agonists with emphasis on the synthetic possibilities, structure–activity relationships (SARs), and suggested binding modes.

Neurochem Res

Fig. 3 Compounds containing heterocyclic bioisosteres for the carboxylic acid group of GABA: muscimol (RS)-dihydromuscimol, thiomuscimol, azamuscimol, and 4-AHP

a

b

c

d

e Fig. 5 Synthesis routes for the formation of a muscimol from acetylenedicarboxylate; b (RS)-dihydromuscimol from a nitrile oxide; c thiomuscimol from 2-aminomaleamide; d azamuscimol from a xphthalimidoacetic acid ester; and e 4-AHP from pyrazole

Fig. 4 Electrostatic potential mapped on the surface of the molecular electron density for the anionic form of a acetic acid, b 3hydroxyisoxazole, c 3-hydroxyisoxazoline, d 3-hydroxythiazole, e 3-hydroxypyrazole and f 1-hydroxypyrazole. Calculations were performed using the B3LYP hybrid potential and cc-pydzs basis set. au: atomic units

Heterocyclic Carboxylic Acid GABAAR Orthosteric Agonists N- or C-hydroxy substituted heterocyclic rings used as bioisosteric replacement of the carboxylic acid group in GABA include (Fig. 3): the 3-hydroxyisoxazole ring of muscimol, the 3-hydroxyisoxazoline ring of (RS)-dihydromuscimol, the 3-hydroxyisothiazole ring of thiomuscimol, the 3-hydroxypyrazole ring of azamuscimol, and the 1-hydroxypyrazole ring of 4-(aminomethyl)-1-hydroxypyrazole (4-AHP). The acidic properties of these compounds are comparable to that of GABA with pKa values in the

range of 4.8–6.2 [14–17]. Also, the negative charge of the deprotonated hydroxyl groups, which are expected to interact with the GABAA receptor, can delocalise into the aromatic ring. As shown in Fig. 4, this gives an electrostatic profile resembling that of a carboxylate group. In addition to conformationally constraining the carboxylate functionality, the heterocyclic rings allow for introduction of substituents of different shape, size and electronic properties in well-defined positions useful for mapping the orthosteric binding site. The first compound containing a heterocyclic ring system to be investigated as a GABA analogue was the 3-hydroxyisoxazol analog, muscimol [8]. The structure of muscimol is restricted by the 3-hydroxyisoxazole ring and thus only has free rotation around the aminomethyl side chain. The pKa value of the acidic group of muscimol (4.8) is slightly higher than that of GABA (4.0) [14]. Furthermore, the 3-hydroxyisoxazole ring of muscimol allows for

123

Neurochem Res

a

b

Fig. 6 a Substituents in the 4-position of muscimol can be introduced in two steps from a protected muscimol scaffold. b Substituents in the 3and 5-positions of 4-AHP can be introduced by directed ortho-lithiation or using a regioselectively introduced iodine

the introduction of substituents in a fixed direction making systematic investigations of the binding pocket possible. Several ways of synthesizing muscimol have been reported [18–26] and a convenient gram-scale preparation is available starting from acetylenedicarboxylate [25] (Fig. 5). Muscimol is also commercially available. Substituents in the 4-position of the 3-hydroxyisoxazole ring can conveniently be introduced in the final stages of a synthetic protocol by lithiation of a protected muscimol scaffold as exemplified by the preparation of 4-methylmuscimol (Fig. 6) [17]. Earlier reported procedures required several synthetic steps subsequent to the introduction of substituents (methyl, ethyl, phenyl, 2-acetoxyethyl, 2-hydroxyethyl and 2-chloroethyl) in the 4-position of the 3-hydroxyisoxazole ring [15, 27–29]. Dihydromuscimol is a partially saturated analogue of muscimol and thus more flexible. Racemic dihydromuscimol can be synthesized from 3-hydroxy-4-aminobutyric acid [15] or by using a [1.3]-dipolar cycloaddition starting from bromonitrile oxide (Fig. 5) [30, 31]. The enantiomers of dihydromuscimol have also been obtained using this method [30, 32, 33]. The pKa of the acidic group of dihydromuscimol (5.8) [15] is higher than that of muscimol. The oxygen atom in the isoxazole ring of muscimol has been substituted for sulfur and nitrogen to give thiomuscimol [16] and azamuscimol [34], respectively. The acidities of the 3-hydroxyisothiazole and 3-hydroxypyrazole moieties of thiomuscimol (pKa 6.1) and azamuscimol (pKa 6.2) are weaker compared to muscimol. Thiomuscimol can be synthesized from 2-aminomaleamide [35] and is also commercially available. Azamuscimol can be synthesized from a x-phthalimidoacetic acid ester [20, 36]. Synthetically accessible substitutions on the 3-hydroxypyrazole ring include methylation of the heterocyclic nitrogens [37]. A new heterocyclic carboxylic acid bioisostere of GABA ideally setup to probe the binding site by easy introduction of substituents in two different positions was introduced recently, incorporating the 1-hydroxypyrazole ring [17]. The pKa values of the acidic groups of 4-AHP

123

and muscimol are comparable (5.4 [17] and 4.8 [16], respectively). This allowed for investigation of potentially unexplored areas of the receptor that could not be accessed using muscimol as template. 4-AHP was synthesized from pyrazole and substituents in the 3- and 5-positions could be conveniently introduced (Figs. 5, 6b). The 3 position was accessed via a regioselective iodination followed by a cross coupling procedure to introduce amine, alkyl, aryl and alkylaryl substituents. The 5-position was accessed directly using a directed ortho-lithiation strategy followed by treatment with an appropriate electrophile whereby halogens and alkyl substituents were introduced [17].

Structure–Activity Relationships The 3-hydroxyisoxazole, 3-hydroxyisothiazole, and the 1and 3-hydroxypyrazole ring-systems have formed the structural basis for the majority of orthosteric GABAAR agonists reported to date. Systematic modifications on the basic scaffolds have been made using the abovementioned chemical strategies and structure–activity studies performed to obtain more insight into the structural requirements for binding and activation of the GABAARs. In the absence of detailed structural information of the binding site, the design of the ligands described in this review has been based on classical medicinal chemistry. Given the complexity of GABAAR pharmacology arising from the presence of many different receptor subunits that can assemble in multiple combinations, the detailed insight into pharmacology and selectivity profiles of GABAAR agonists is still relatively sparse. Typically, the compounds have been characterized in radioligand binding assays using native tissues, and thus their binding affinities to native a1b2c2S GABAARs, which are dominant, have primarily been determined. The binding affinities of the majority of the compounds mentioned in this review have been measured by displacement of [3H] muscimol or [3H] GABA in rat brain preparations. Functional characterization of selected compounds has been carried out at the

Neurochem Res

human a1b2c2 GABAAR either using the FLIPR membrane potential (FMP) blue assay or whole-cell patch-clamp electrophysiology. Only few studies on orthosteric GABAAR ligands using different subunit combinations have been reported [9, 38, 39]. Muscimol, which has been widely used in the study of GABAARs, and the analogues (RS)-dihydromuscimol and thiomuscimol all display potencies for the GABAAR in the low nanomolar range (Table 1). The two enantiomers of dihydromuscimol display different pharmacological properties. The (S)-enantiomer shows high affinity for the GABAAR, whereas the inhibition of GABA uptake via GABA transporters displayed by dihydromuscimol proved to reside exclusively in the R-enantiomer [32]. Substituting a 1-hydroxypyrazole for the 3-isoxazolol ring to give 4-AHP, leads to binding affinity in the low micromolar range [17], whereas the isomeric 3-hydroxypyrazole analogue, azamuscimol, exhibited a 100-fold further reduction in binding affinity [16]. Relative agonist potencies at the GABAAR of the heterocylic muscimol analogues are (S)-dihydromTable 1 Pharmacological data for GABA, muscimol, and the heterocyclic analogues dihydromuscimol, thiomuscimol, azamuscimol, and 4-AHP Compound

[3H]muscimol or [3H]GABA binding at rat synaptic membranesa Ki or IC50 (lM)

[3H]muscimol binding at a1b2c2 GABAARa Ki (lM)

a1b2c2 GABAARb EC50 (lM)

GABA

0.033c

0.057f

1.8e (20)f

c

f

Muscimol

0.006

0.008

0.54e (5.0)f

(RS)dihydromuscimol

0.008c

nd

(5)g

(S)dihydromuscimol

0.004c

nd

nd

(R)dihydromuscimol

0.25c

nd

nd

Thiomuscimol

0.019d

0.042f

(5)f

Azamuscimol

16d

nd

nd

4-AHP

0.22e

nd

13e

nd not determined a

GABAA receptor binding affinity at rat synaptic membranes and at the human a1b2c2 GABAAR subtype

b

Functional characterization at the human a1b2c2 GABAAR subtype transiently expressed in tsA201 cells in the FLIPR membrane potential assay and/or in Xenopus laevis oocytes using voltage clamp technique c

From Ref. [32]

d

From Ref. [16]

e

From Ref. [17] From Ref. [9]

f g

From Ref. [38]

uscimol [ muscimol [ thiomuscimol [ 4-AHP  azamuscimol as shown in Table 1. The structure of muscimol has been extensively explored using classical medicinal chemistry (Fig. 7). Chemical variation of the aminomethyl chain has been pursued, changing the attachment point, extending the length and/or introduction of alkyl groups all leading to considerable loss of GABAAR binding affinity [40, 41]. Introduction of a methyl group into the 4-position of the 3-isoxazolol ring of muscimol severely inhibits the interaction with the GABAAR recognition site. Thus, 4-methylmuscimol is four orders of magnitude weaker than muscimol as an inhibitor of GABAAR binding, whereas 4-Et-muscimol as well as larger hydrophobic-substituted analogues (unpublished) are inactive [15]. However, connecting the 4-ethyl substituent and the aminomethyl chain into the fused bicyclic analogue THIP, did lead to moderately high GABAAR affinity [42]. Functionally, THIP has been shown to display partial or super-agonist activity depending on subunit composition [38, 43]. The corresponding non-fused THIP analogue 4-PIOL has been characterized as a low-efficacy GABAAR partial agonist [44]. Interestingly, introduction of substituents in the 4-position of the isoxazolol ring of 4-PIOL resulted in highly potent GABAAR antagonists in contrast to the observed SAR for muscimol, which indicates different binding modes of the isoxazolol rings of muscimol and 4-PIOL in the binding pocket [12, 13]. SAR studies on a series of substituted 4-AHP analogues were recently reported (Fig. 7; Table 2); [17]. Analogously to what was observed in the muscimol series, introduction of substituents such as methyl, bromo and chloro, in the 5-position of the 1-hydroxypyrazole ring (compounds 1–3), impaired binding affinity. However, the loss in binding affinity observed for the 5-substituted 4-AHP analogues was less pronounced than shown for the 4-substituted muscimol analogues. Substitution in the 3-position of the 1-hydroxypyrazole ring (compounds 4–6), addressing an additional region in the binding pocket, followed the trend that larger substituents generated lower binding affinities. The 3,5-dimethyl substituted analogue, 7, was inactive, implying that the tight-fitting binding pocket is unable to accommodate both substituents. Only few analogues of azamuscimol have been synthesized (Fig. 7). The pyrazole scaffold offers an additional position, N2, for exploring the binding pocket. However, introducing a methyl group into each of the nitrogen atoms in the heterocylic ring leads to loss of binding affinity [16]. Fused and non-fused bicyclic analogues of thiomuscimol, dihydromuscimol and azamuscimol, corresponding to THIP and 4-PIOL have been synthesized and pharmacologically characterized in the GABAAR system [45–47]. The GABAAR activities for these analogues follow the

123

Neurochem Res Fig. 7 Structures of muscimol, general structures of muscimol analogues, a series of 4-AHP analogues, azamuscimol analogues, and the two stereoisomers of dihydromuscimol

Table 2 Pharmacological data for 4-AHP and 3- or 5-substituted 4-AHP analogues R1

R2

[3H]muscimol binding rat synaptic membranesa Ki (lM)

a1b2c2 GABAARb EC50 (lM)

q1 GABAARb EC50 (lM)

4-AHP

H

H

0.22

13

3.1

1

Me

H

15

[500

[500

2

Br

H

2.8

nd

nd

3

Cl

H

0.63

29

[500

4

H

Me

68

nd

nd

5

H

Et

94

[500

[500

6

H

Bn

[100

nd

nd

7

Me

Me

[100

nd

nd

nd not determined a

GABAA receptor binding affinity at rat synaptic membranes [17]

b

Functional characterization at the human a1b2c2 GABAAR subtype transiently expressed in tsA201 cells in the FLIPR membrane potential assay [17]

trend seen for THIP and 4-PIOL, leading to lower or no affinity in general and partial agonist activity for the bicyclic non-fused analogues compared to the respective monocyclic compounds. As for 4-PIOL and muscimol, non-parallel SAR is observed for ring-substituted nonfused bicyclic analogues of thiomuscimol [45] and 4-AHP [46] relative to the parent compounds. Muscimol, thiomuscimol, and (RS)-dihydromuscimol have been characterized in recombinant human GABAARs expressed in Xenopus oocytes using patch-clamp technique by systematic variation of the subunit composition in order

123

to map the influence of different subunits on the efficacy and affinity [9, 38]. The compounds only displayed small differences in binding affinities and activity using recombinant GABAARs containing different a or b subunits. Analogously to muscimol, thiomuscimol and dihydromuscimol, a similar study performed on 4-AHP in GABAAR transiently expressed in tsA201 cells using the FLIPR membrane potential blue assay did not disclose any significant subtype selectivity [17]. As seen for the majority of GABAAR agonists, muscimol, thiomuscimol, and 4-AHP display some agonist

Neurochem Res

Fig. 8 The flipped arginine pharmacophore model explaining why distinct structure activity relationships are observed between muscimol and 4-piol derivatives [13]

activity at the GABACRs as well. Interestingly, the 5-substituted 4-AHP analogues showed some degree of selectivity for a1b2c2 over q1 GABAARs in the electrophysiological measurements as opposed to the observations for 4-methylmuscimol [17]. In general, the unsubstituted heterocyclic analogues of GABA show functional profiles comparable to that of GABA, thus validating the use of the different cores as true bioisosteric replacements for the carboxylic acid in GABA. However, despite a common general structure of the main heterocycles, differences in the GABAAR binding affinity and potency between the different classes of heterocyclic GABA analogues presented in this review are seen. All in all, the SARs performed underline the limited available space in the binding site and that small differences in structure are crucial for GABAAR activity and to some degree subtype selectivity.

Pharmacophores and Structural Models As an attempt to elucidate the similarities and differences between different classes of compounds, pharmacophore models based on muscimol, THIP and 4-PIOL [12, 13] have been useful. Most notably to explain the dramatic differences between the isoxazole moieties of muscimol and 4-PIOL for which the flexible ‘‘arginine model’’ shown in Fig. 8 offers an explanation [12, 13]. In this model, the nonparallel SAR of the isoxazole rings of muscimol and 4-PIOL is explained by different rotameric states of an arginine side-chain allowing the amines to be aligned while the isoxazole parts are not. A number of attempts to provide a more detailed insight into the molecular basis for binding and activation have been reported using homology models based on the homologous acetylcholine binding protein (AChBP) [48–52] and bacterial ion channels [53–60].

Fig. 9 Proposed GABA binding mode in the a1b2c2 GABAAR homology model in the agonist conformation. The b2 subunit is depicted in pink and the a1 subunit in purple (Color figure online)

More recently, co-crystallisation of glutamate with an eukaryotic glutamate gated chloride channel (GluCl) [61] has significantly improved the molecular basis for homology modelling of GABAARs. The GluCl structure shares sequence identities with a1, b2, and c2 GABAAR subunits of 30, 36 and 31 %, respectively. Locally, in the binding site region, the identity is as high as 48 % [62] and because GABA can be identified as a sub-structure of glutamate, details of ligand receptor interactions can be deduced from the structure [62]. In homology modelling, the GluCl structure has been used in conjunction with the structure of a bacterial ion channel from Erwinia chrysanthemi (ELIC) [63] resulting in an accumulated sequence identity of 57 % in the binding site region [62]. Based on this model, agonist binding modes for GABA, muscimol, and THIP were suggested. Compared to a previously reported antagonist model [60], the agonist model has a tighter binding site because the so-called loop-C caps it to support agonist binding as also observed in AChBP structures with agonists bound [64, 65]. In the proposed binding mode for GABA, important ligand–receptor interactions include (Fig. 9): (1) Salt bridges between the carboxylate group of GABA and a1Arg66 and between the ammonium group of GABA and b2Glu155 (2) hydrogen bonds between the carboxylate group of GABA and a1Thr129 as well as b2Thr202, and between the GABA ammonium group and backbone carbonyl of b2Ser156; and finally (3) a p–cation interaction between the GABA ammonium group and b2Tyr205 [62].

123

Neurochem Res Table 3 Binding affinities, pKa values and calculated conformational energy penalties at predicted bioactive conformation for muscimol, 4-AHP and thiomuscimol

Muscimol

[3H]muscimol binding rat synaptic membranesa Ki (lM)

pKa

Conformational energy penalty at 55°/60°a (kcal/mol)

0.006

4.8

0.3/0.6

4-AHP

0.22

5.5

2.0/1.4

Thiomuscimol

0.019

6.1

0.3/0.0

a

Torsion around aminomethyl sidechain (defined in Fig. 11)

In order for muscimol to assume a binding mode where the functional groups (amine and carboxylic acid bioisostere) are superimposable on the corresponding functional groups of GABA, muscimol has to adopt a high energy conformation, which seems unlikely for a high affinity compound. Instead, it was suggested that muscimol binds to the receptor in concert with a water molecule, which allows more extensive interactions with the receptor while maintaining a low energy conformation with a torsional angle of the aminomethyl side chain of *60°. Relative to the global energy minimum at 45° (Fig. 11), the suggested bioactive conformation is only associated with a conformational energy penalty of 0.6 kcal/mol. In this binding mode, the amino group of muscimol forms contacts to b2Glu155 and is within cation–p interaction distance to two aromatic residues, b2Tyr205 and b2Phe200 [62] in loop C (Fig. 1). A long withstanding puzzle in GABAAR medicinal chemistry has been to rationalize the relative affinities of the structurally very similar compounds muscimol, thiomuscimol, and 4-AHP, because affinities do not correlate with pKa values of the carboxylic acid bioisostere (Table 3). However, docking studies predict a binding mode for 4-AHP and thiomuscimol (not shown) similar to that of muscimol (Figs. 10, 11) with a dihedral angle of the aminomethyl side chain of 55°–60°. Interestingly, the conformational energy penalties associated with the predicted binding conformations differ significantly among the three compounds (Table 3). Muscimol and thiomuscimol pay small or no conformational energy penalties, while 4-AHP pays a conformational energy penalty in the range of 1.4–2.0 kcal/mol corresponding to an inherent less than tenfold reduction in binding affinity due to high internal strain in the molecule. When combined with a higher pKa value, it may explain the significantly lower affinity of 4-AHP relative to muscimol. Likewise, as the predicted bioactive conformation of thiomuscimol is in fact quite near its global energy minimum, it suggests that the loss of affinity of thiomuscimol relative to muscimol is directly related to the difference in pKa values between the two compounds. For azamuscimol, lower affinity compared to

123

Fig. 10 The GABAA binding site at the interface between the b2 subunit (pink) and the a1 subunit (purple). The proposedbinding modes of muscimol (cyan) and 4-AHP (green) are nearly identical interacting similarly in the binding site. A water molecule mediates additional contacts to the B-loop of the b2 subunit (Color figure online)

4-AHP may be due to existence of tautomeric forms and higher desolvation energy. The binding mode for THIP was suggested to resemble that of muscimol with the isoxazole ring in the same orientation and the amino moiety forming a salt bridge with b2Glu155 and a p–cation interaction with b2Tyr205 [62]. In this binding mode the water molecule is unable to fit the binding site, thus THIP is predicted to bind without a water molecule linking the ligand to the B-loop of the receptor. It’s evident from the model that space is too limited to favorably accommodate methyl substituted analogues such as 4-methylmuscimol and 5-methyl-4-AHP. Hence, the low affinity of these compounds relative to their un-substituted counterparts can be explained with a steric clash between the methyl groups and b2Tyr157. An additional rearrangement of the aminomethyl side chain resulting in a larger torsional angle would explain why methylation is less penalized for 4-AHP compared to muscimol. Figure 11 illustrates how a larger aminomethyl torsional angle is energetically favorable for 4-AHP and penalized for muscimol.

Outlook A major challenge in GABAAR drug discovery has been to obtain subtype-selective agonists. To a large extent, this strategy has failed and the most promising selectivity ratios obtained so far are between GABAA and GABAC receptors,

Neurochem Res Muscimol

Thio-muscimol

10

4-AHP

-

O

9

+

H3N O

Energy (kcal/mol)

8

N

7 6 5 4 3 2 1 0 0

30

60

90

120

150

180

Dihedral angle Fig. 11 Torsional energies of the aminomethyl substituent of muscimol (blue diamonds), thio-muscimol (grey squares) and 4-AHP (green triangles). The preferred dihedral angles of muscimol and thiomuscimol are below 90° due to intra-molecular attraction between the ring bound oxygen or sulphur and the positively charged amine. On the contrary the preferred dihedral angle of 4-AHP is exactly 90° explained by 4-AHP being a more symmetric molecule with carbon atoms flanking the aminomethyl side chain both in the 3- and 5-positions of the ring. Calculations were made using density functional theory (B3LYP/6-31G**) and the Poisson Boltzmann finite element solvation method (Color figure online)

which likely reflect that the individual GABAAR binding sites are too similar for agonists to bind selectively. To advance the field, a change in strategy is required. Inspired by benzodiazepines [66] and THIP, it seems that functional selectivity can be obtained for some receptor subtypes despite the lack of binding selectivity. The elucidation of the underlying mechanisms and structural determinants for this type of selectivity may along with better control of stoichiometries and subunit composition of recombinant receptors pave the way for future drug discovery projects. Along with a better understanding of the underlying mechanisms of receptor activation, the significant advances that have been made in synthesis of heterocyclic GABAAR agonists and in structural receptor modeling may enable future drug discovery in the field to be done on a more rational basis.

References 1. Foster AC, Kemp JA (2006) Glutamate- and GABA-based CNS therapeutics. Curr Opin Pharmacol 6(1):7–17

2. Johnston GAR (2005) GABAA receptor channel pharmacology. Curr Pharm Des 11(15):1867 3. Dougherty DA (2008) Cys-loop neuroreceptors: structure to the rescue? Chem Rev 108(5):1642–1653 4. Davies PA, Wang W, Hales TG, Kirkness EF (2003) A novel class of ligand-gated ion channel is activated by Zn2?. J Biol Chem 278(2):712–717 5. Sieghart W (2006) Structure, pharmacology, and function of GABAA receptor subtypes. In: Enna SJ (ed) Advances in pharmacology, vol 54. Academic Press, California, pp 231–263 6. Karlin A (2002) Emerging structure of the nicotinic acetylcholine receptors. Nat Rev Neurosci 3(2):102–114 7. Paul JW (2003) GABAA receptor subtypes in the brain: a paradigm for CNS drug discovery? Drug Discov Today 8(10):445–450 8. Johnston GAR, Curtis DR, de Groat WC, Duggan AW (1968) Central actions of ibotenic acid and muscimol. Biochem Pharmacol 17(12):2488–2489 9. Ebert B, Thompson SA, Saounatsou K, McKernan R, Krogsgaard-Larsen P, Wafford KA (1997) Differences in agonist/ antagonist binding affinity and receptor transduction using recombinant human c-aminobutyric acid type a receptors. Mol Pharmacol 52(6):1150–1156 10. Krogsgaard-Larsen P, Johnston GAR, Lodge D, Curtis DR (1977) A new class of GABA agonist. Nature 268(5615):5355 11. Byberg JR, Labouta IM, Falch E, Hjeds H, Krogsgaard-Larsen P (1987) Synthesis and biological activity of a GABAA agonist which has no effect on benzodiazepine binding and of structurally related glycine antagonists. Drug Des Deliv 1:261–274 12. Frølund B, Tagmose L, Liljefors T, Stensbøll TB, Engblom C, Kristiansen U, Krogsgaard-Larsen P (2000) A novel class of potent 3-isoxazolol GABAA antagonists: design, synthesis, and pharmacology. J Med Chem 43(26):4930–4933 13. Frølund B, Jørgensen AT, Tagmose L, Stensbøl TB, Vestergaard HT, Engblom C, Kristiansen U, Sanchez C, Krogsgaard-Larsen P, Liljefors T (2002) Novel Class of potent 4-arylalkyl substituted 3-isoxazolol GABAA antagonists: synthesis, pharmacology, and molecular modeling. J Med Chem 45(12):2454–2468 14. Jacobsen P, Labouta IM, Schaumburg K, Falch E, KrogsgaardLarsen P (1982) Hydroxy- and amino-substituted piperidinecarboxylic acids as c-aminobutyric acid agonists and uptake inhibitors. J Med Chem 25(10):1157–1162 15. Krogsgaard-Larsen P, Larsen ALN, Thyssen K (1978) GABA receptor agonists. synthesis of muscimol analogues including (R)and (S)-5-(1-aminoethyl)-3-isoxazolol and (RS)-5-aminomethyl2-isoxazolin-3-ol. Acta Chem Scand B32:469–477 16. Krogsgaard-Larsen P, Hjeds H, Curtis DR, Lodge D, Johnston GAR (1979) Dihydromuscimol, thiomuscimol, and related heterocyclic compounds as GABA analogues. J Neurochem 32(6):1717–1724 17. Petersen JG, Bergmann R, Møller HA, Jørgensen CG, Nielsen B, Kehler J, Frydenvang K, Kristensen J, Balle T, Jensen AA, Kristiansen U, Frølund B (2013) Synthesis and biological evaluation of 4-(aminomethyl)-1-hydroxypyrazole analogues of muscimol as c-aminobutyric acid(a) receptor agonists. J Med Chem 56(3):993–1006 18. Pevarello P, Varasi M (1992) An improved synthesis of muscimol. Synth Commun 22(13):1939–1948 19. Gagneux AR, Ha¨fliger F, Eugster CH, Good R (1965) Synthesis of pantherine (agarin). Tetrahedron Lett 6(25):2077–2079 20. Loev B, Wilson JW, Goodman MM (1970) Synthesis of compounds related to muscimol (pantherine, agarin). J Med Chem 13(4):738–741 21. McCarry BE, Savard M (1981) A facile synthesis of muscimol. Tetrahedron Lett 22(51):5153–5156

123

Neurochem Res 22. Ja¨ger V (1982) Frey M (1982) A short synthesis of muscimol. Liebigs Ann Chem 4:817–820 23. Welch WM (1982) A shorter synthesis of muscimol. Synth Commun 12(14):1089–1092 24. Oster TA, Harris TM (1983) Generation and reactions of the dianion of 3-hydroxy-5-methylisoxazole, a convenient b-keto amide synthon. total synthesis of muscimol. J Org Chem 48(23):4307–4311 25. Frey M, Ja¨ger V (1985) Synthesis of N-substituted muscimol derivatives including N-glycylmuscimol. Synthesis 12:1100–1104 26. Chiarino D, Napoletano M, Sala A (1986) A convenient synthesis of muscimol by a 1,3-dipolar cycloaddition reaction. Tetrahedron Lett 27(27):3181–3182 27. Nakamura N, Tajima Y, Sakai K (1982) Direct phenylation of isoxazoles using palladium catalysts. Synthesis of 4-phenylmuscimol. Heterocycles 17(1):235–245 28. Hjeds H, Christensen IT, Cornett C, Frølund B, Falch E, Pedersen JB, Krogsgaard-Larsen P (1992) 3-Hydroxyisoxazole bioisosteres of GABA. Synthesis of a series of 4-substituted muscimol analogues and identification of a bicyclic 2-isoxazoline rearrangement product. Acta Chem Scand 46:772–777 29. Bowden K, Crank G, Ross WJ (1968) The synthesis of pantherine and related compounds. J Chem Soc C 172–185 30. De Amici M, De Micheli C, Misani V (1990) Nitrile oxides in medicinal chemistry-2. Synthesis of the two enantiomers of dihydromuscimol. Tetrahedron 46(6):1975–1986 31. Caldirola P, De Amici M, De Micheli C (1986) An easy synthesis of dihydromuscimol. Tetrahedron Lett 27(38):4651–4652 32. Krogsgaard-Larsen P, Nielsen L, Falch E, Curtis DR (1985) GABA agonists. Resolution, absolute stereochemistry and enantioselectivity of (S)-(?)- and (R)-(-)-dihydromuscimol. J Med Chem 28(11):1612–1617 33. Castelhano AL, Billedeau R, Pliura DH, Bonaventura BJ, Krantz A (1988) Synthesis, chemistry, and absolute configuration of novel transglutaminase inhibitors containing a 3-halo-4, 5-dihydroisoxazole. Bioorg Chem 16(3):335–340 34. Krogsgaard-Larsen P, Johnston GAR, Curtis DR, Game CJA, McCulloch RM (1975) Structure and biological activity of a series of conformationally restricted analogues of GABA. J Neurochem 25(6):803–809 35. Lykkeberg J, Krogsgaard-Larsen P (1976) Structural analogues of GABA. Synthesis of 5-aminomethyl-3-isothiazolol (thiomuscimol). Acta Chem Scand Ser B 30:781–785 36. Lykkeberg J (1978) Synthesis of some pyrazol-5-ols related to muscimol. Acta Chem Scand Ser B 32:56–60 37. Hjeds H, Krogsgaard-Larsen P (1979) Muscimol analogues. synthesis of isomuscimol (3-aminomethyl-5-isoxazol) and some derivatives of azamuscimol (5-aminomethyl-2-pyrazolol). Acta Chem Scand Ser B 33:294–298 38. Ebert B, Wafford KA, Whiting PJ, Krogsgaard-Larsen P, Kemp JA (1994) Molecular pharmacology of gamma-aminobutyric acid type a receptor agonists and partial agonists in oocytes injected with different alpha, beta, and gamma receptor subunit combinations. Mol Pharmacol 46(5):957–963 39. Høstgaard-Jensen K, O’Connor R, Dalby N, Simonsen C, Finger B, Golubeva A, Hammer H, Bergmann M, Kristiansen U, KrogsgaardLarsen P, Bra¨uner-Osborne H, Ebert B, Frølund B, Cryan J, Jensen AA (2013) The orthosteric GABAA receptor ligand thio-4-PIOL displays distinctly different functional properties at synaptic and extrasynaptic receptors. Br J Pharmacol 170(4):919–932 40. Krogsgaard-Larsen P, Johnston GA (1978) Structure-activity studies on the inhibition of GABA binding to rat brain membranes by muscimol and related compounds. J Neurochem 30(6):1377–1382 41. Krogsgaard-Larsen P, Johnston GA, Curtis DR, Game CJ, McCulloch RM (1975) Structure and biological activity of a

123

42. 43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

series of conformationally restricted analogues of GABA. J Neurochem 25(6):803–809 Krogsgaard-Larsen P, Johnston GA, Lodge D, Curtis DR (1977) A new class of GABA agonist. Nature 268(5615):53–55 Brown N, Kerby J, Bonnert TP, Whiting PJ, Wafford KA (2002) Pharmacological characterization of a novel cell line expressing human a4b3d GABAA receptors. Br J Pharmacol 136(7):965–974 Byberg JR, Labouta IM, Falch E, Hjeds H, Krogsgaard-Larsen P, Curtis DR, Gynther BD (1987) Synthesis and biological activity of a GABAA agonist which has no effect on benzodiazepine binding and of structurally related glycine antagonists. Drug Des Deliv 1(4):261–274 Krehan D, ´ı Storustovu S, Liljefors T, Ebert B, Nielsen B, Krogsgaard-Larsen P, Frølund B (2006) Potent 4-arylalkyl-substituted 3-isothiazolol GABAA competitive/noncompetitive antagonists: synthesis and pharmacology. J Med Chem 49(4):1388–1396 Møller HA, Sander T, Kristensen JL, Nielsen B, Krall J, Bergmann ML, Christiansen B, Balle T, Jensen AA, Frølund B (2010) Novel 4-(piperidin-4-yl)-1-hydroxypyrazoles as c-aminobutyric acid(A) receptor ligands: synthesis, pharmacology, and structure– activity relationships. J Med Chem 53(8):3417–3421 Conti P, De Amici M, Pinto A, Tamborini L, Grazioso G, Frølund B, Nielsen B, Thomsen C, Ebert B (2006) De Micheli C (2006) synthesis of 3-hydroxy- and 3-carboxy-d2-isoxazoline amino acids and evaluation of their interaction with GABA receptors and transporters. Eur J Org Chem 24:5533–5542 Berezhnoy D, Gibbs TT, Farb DH (2009) Docking of 1,4-benzodiazepines in the a1/c2 GABAA receptor modulator site. Mol Pharmacol 76(2):440–450 Cromer BA, Morton CJ, Parker MW (2002) Anxiety over GABA(A) receptor structure relieved by AChBP. Trends Biochem Sci 27(6):280–287 Ci SQ, Ren TR, Ma CX, Su ZG (2007) Modeling of alphak/ gamma2 (k = 1, 2, 3 and 5) interface of GABA a receptor and docking studies with zolpidem: implications for selectivity. J Mol Graph Model 26(2):537–545 Sancar F, Ericksen SS, Kucken AM, Teissere JA, Czajkowski C (2007) Structural determinants for high-affinity zolpidem binding to GABA-A receptors. Mol Pharmacol 71(1):38–46 O’Mara M, Cromer B, Parker M, Chung S-H (2005) Homology model of the GABAA receptor examined using brownian dynamics. Biophys J 88(5):3286–3299 Trudell JR (2002) Unique assignment of inter-subunit association in GABAA a1b3c2 receptors determined by molecular modeling. Biochim Biophys Acta Biomembr 1565(1):91–96 Ernst M, Bruckner S, Boresch S, Sieghart W (2005) Comparative models of GABAA receptor extracellular and transmembrane domains: important insights in pharmacology and function. Mol Pharmacol 68(5):1291–1300 Padgett CL, Hanek AP, Lester HA, Dougherty DA, Lummis SCR (2007) Unnatural amino acid mutagenesis of the GABAA receptor binding site residues reveals a novel cation–p interaction between GABA and b2Tyr97. J Neurosci 27(4):886–892 Mokrab Y, Bavro VN, Mizuguchi K, Todorov NP, Martin IL, Dunn SMJ, Chan SL, Chau PL (2007) Exploring ligand recognition and ion flow in comparative models of the human GABA type A receptor. J Mol Graphics Modell 26(4):760–774 Chupakhin V, Palyulin V, Zefirov N (2006) Modeling the open and closed forms of GABAA receptor: analysis of ligand– receptor interactions for the GABA-binding site. Dokl Biochem Biophys 408(1):169–174 Jansen M, Rabe H, Strehle A, Dieler S, Debus F, Dannhardt G, Akabas MH, Lu¨ddens H (2008) Synthesis of GABAA receptor agonists and evaluation of their a-subunit selectivity and orientation in the GABA binding Site. J Med Chem 51(15):4430–4448

Neurochem Res 59. Law RJ, Lightstone FC (2009) Modeling neuronal nicotinic and GABA receptors: important interface salt-links and protein dynamics. Biophys J 97(6):1586–1594 60. Sander T, Frølund B, Bruun AT, Ivanov I, McCammon JA, Balle T (2011) New insights into the GABAA receptor structure and orthosteric ligand binding: receptor modeling guided by experimental data. Proteins 79(5):1458–1477 61. Hibbs RE, Gouaux E (2011) Principles of activation and permeation in an anion-selective Cys-loop receptor. Nature 474(7349):54–60 62. Bergmann R, Kongsbak K, Sørensen PL, Sander T, Balle T (2013) A Unified model of the GABAA receptor comprising agonist and benzodiazepine binding sites. PLoS ONE 8(1):e52323 63. Hilf RJC, Dutzler R (2008) X-ray structure of a prokaryotic pentameric ligand-gated ion channel. Nature 452(7185):375–379

64. Celie PH, van Rossum-Fikkert SE, van Dijk WJ, Brejc K, Smit AB, Sixma TK (2004) Nicotine and carbamylcholine binding to nicotinic acetylcholine receptors as studied in AChBP crystal structures. Neuron 41(6):907–914 65. Rohde LA, Ahring PK, Jensen ML, Nielsen EO, Peters D, Helgstrand C, Krintel C, Harpsoe K, Gajhede M, Kastrup JS, Balle T (2012) Intersubunit bridge formation governs agonist efficacy at nicotinic acetylcholine a4b2 receptors. J Biol Chem 287(6):4248–4259 66. Atack JR (2008) GABAA receptor subtype-selective efficacy: tpa023, an a2/a3 selective non-sedating anxiolytic and a5ia, an a5 selective cognition enhancer. CNS Neurosci Ther 14(1):25–35

123