Design, Synthesis, and Pharmacological Evaluation of Novel β2/3 Subunit-Selective γ-Aminobutyric Acid Type A (GABAA) Receptor Modulators

Article abstract of DOI:10.1021/acs.jmedchem.8b00859

Subunit-selective modulation of γ-aminobutyric acid type A receptors (GABA_AR) is considered to exert fewer side effects compared to unselective clinically used drugs. Here, the β2/3 subunit-selective GABA_AR modulators valerenic acid (VA) and loreclezole (LOR) guided the synthesis of novel subunit-selective ligands with simplified structures. We studied their effects on GABA_ARs expressed in Xenopus laevis oocytes using two-microelectrode voltage clamp technique. Five compounds showed significantly more efficacious modulation of GABA-evoked currents than VA and LOR with retained potency and selectivity. Compound 18 [(E)-2-Cyano-3-(2,4-dichlorophenyl)but-2-enamide] induced the highest maximal modulation of GABA-induced chloride currents (E_max: 3114 ± 242%), while 12 [(Z)-3-(2,4-dichlorophenyl)but-2-enenitrile] displayed the highest potency (EC₅₀: 13 ± 2 μM). Furthermore, in hippocampal neurons 12 facilitated phasic and tonic GABAergic inhibition, and in vivo studies revealed significantly more potent protection against pentylenetetrazole (PTZ)-induced seizures compared to VA and LOR. Collectively, compound 12 constitutes a novel, simplified, and subunit-selective GABA_AR modulator with low-dose anticonvulsant activity.

Full text of DOI:10.1021/acs.jmedchem.8b00859

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE
Article
Design, synthesis and pharmacological evaluation of novel #2/3 subunit-
selective #-Aminobutyric Acid Type A (GABAA) receptor modulators
Marco Stadler, Serena Monticelli, Thomas Seidel, Denise Luger, Isabella Salzer, Stefan Boehm, Wolfgang
Holzer, Christoph Schwarzer, Ernst Urban, Sophia Khom, Thierry Langer, Vittorio Pace, and Steffen Hering
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00859 • Publication Date (Web): 05 Oct 2018
Downloaded from http://pubs.acs.org on October 6, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Design, synthesis and pharmacological evaluation of novel β2/3 subunit-selective γ-
Aminobutyric Acid Type A (GABA ) receptor modulators
A
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
2
1
3
Marco Stadler* , Serena Monticelli , Thomas Seidel , Denise Luger , Isabella Salzer , Stefan
3
2
4
2
1,5
Boehm , Wolfgang Holzer , Christoph Schwarzer , Ernst Urban , Sophia Khom , Thierry
2
2
1
Langer , Vittorio Pace* , Steffen Hering .
1Department of Pharmacology and Toxicology, University of Vienna, Althanstraße 14, 1090
Vienna, Austria
2Department of Pharmaceutical Chemistry, University of Vienna, Althanstraße 14, 1090 Vienna,
Austria
3Department of Neurophysiology and Neuropharmacology, Medical University Vienna,
Schwarzspanierstraße 17, 1090 Vienna, Austria
4Department of Pharmacology, Medical University Innsbruck, Peter-Mayr-Straße 1a, 6020
Innsbruck, Austria
5Department of Neuroscience, The Scripps Research Institute, 10550 N Torrey Pines Rd, La Jolla,
CA 92037, USA
Keywords
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
GABA, GABA receptor, valerenic acid, loreclezole, pharmacophore model, subunit-selectivity,
A
pentylenetetrazole, anticonvulsant, hippocampal neurons, miniature inhibitory postsynaptic
currents, tonic inhibition
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Abstract
Subunit-selective modulation of γ-aminobutyric acid type-A-receptors (GABA R) is considered to
A
exert fewer side effects compared to unselective clinically used drugs. Here, the β2/3 subunit-
selective GABA R modulators valerenic acid (VA) and loreclezole (LOR) guided the synthesis of
A
novel subunit-selective ligands with simplified structures. We studied their effects on GABA Rs
A
expressed in Xenopus laevis oocytes using two-microelectrode voltage clamp technique. Five
compounds showed significantly more efficacious modulation of GABA-evoked currents than VA
and LOR with retained potency and selectivity. Compound 18 [(E)-2-cyano-3-(2,4-
dichlorophenyl)but-2-enamide] induced the highest maximal modulation of GABA-induced
chloride currents (Emax: 3114 ± 242%), while 12 [(Z)-3-(2,4-dichlorophenyl)but-2-enenitrile]
displayed the highest potency (EC : 13 ± 2 µM). Furthermore, in hippocampal neurons 12
5
0
facilitated phasic and tonic GABAergic inhibition and in vivo studies revealed significantly more
potent protection against pentylenetetrazole (PTZ)-induced seizures compared to VA and LOR.
Collectively, compound 12 constitutes a novel, simplified and subunit-selective GABA R
A
modulator with low-dose anticonvulsant activity.
2
ACS Paragon Plus Environment

Page 3 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
γ-Aminobutyric acid (GABA) type A receptors (GABA R) represent the major inhibitory
A
neurotransmitter receptors in the mammalian brain.1–3 GABA Rs belong to the Cys-loop
A
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
superfamily of ligand-gated ion channels and are assembled of five subunits forming a central
chloride-conducting pore.4–6 Nineteen genes encoding different subunits have been discovered in
the human genome, comprising α1−6, β1−3, γ1−3, δ, ε, θ, π, and ρ1−3.7,8 The assembly of
GABA Rs as heteropentamers produces complex subtype heterogeneity in structure, which is the
A
major determinant of their pharmacological properties including agonist sensitivity as well as its
sensitivity for drugs.1,7 Drugs, such as benzodiazepines, enhancing chloride currents through
GABA Rs play an important role in the treatment of various pathological conditions such as
A
general anxiety, panic disorders, sleep disturbances and epilepsy.9–12 However, due to their
relatively unselective binding severe side effects including dependence, unwanted sedation, and
amnesia complicate their long-term use.13–15 Hence, there is unmet medical need for subunit
selective GABA R modulators lacking these unwanted side effects.11,16,17
A
Besides clinically used drugs such as benzodiazepines, barbiturates, neurosteroids or anaesthetics,
e.g.1,2,11,18–24, GABA R are modulated by numerous natural products offering an unexploited
A
source for potential lead structures for drug development.25–27 Valerenic acid (VA), a
sesquiterpenoid and major constituent of common Valerian was recently shown to selectively
modulate β2/3 subunit-comprising GABA Rs whereas β1-containing receptors are only negligibly
A
affected.28,29 Moreover, in vivo, VA induces anxiolytic29–31 and anticonvulsive31,32 effects, but does
3
0
not significantly reduce locomotor activity even at high doses. These findings combined with
3
3
interesting pharmacokinetic properties make VA a promising drug candidate. However, VA as
starting material for the synthesis of novel drug scaffolds with fewer side effects is unfavourable
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
since it is high-priced and, moreover, the chemical structure is intricate34,35 raising interest for
simplified and easily accessible alternatives. Recently, we provided evidence that amino acids
β3N265, β3R269, β3M286 and β3F289 encompass VA’s binding pocket in the upper TM domain
3
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
at the β+/α- interface on α1β3γ2S receptors. Thus, VA is considered to share the binding site of
other β2/3 subunit-selective ligands such as etomidate,19,37,38 propofol38,39 or loreclezole (LOR).40–
4
2
In the past pharmacophore models have been successfully applied to study structural
requirements for subtype-selective targeting of GABA Rs.43–48Assuming that ligands sharing the
A
same binding site also exhibit the same interaction pattern with the residues forming the binding
site, we selected VA and LOR as training compounds to generate a ligand-based pharmacophore
model. With this model in hand, the design of novel, simplified β2/3 subunit-selective GABA R
A
modulators was guided. GABA R modulation was assessed in Xenopus laevis oocytes expressing
A
4
9
GABA Rs via a semi-automated two-microelectrode voltage clamp screening system. In iterative
A
cycles of design, synthesis and pharmacological evaluation the chemical scaffold was optimised in
a straightforward manner within the multidimensional physicochemical property space. The most
potent compounds 12 and 15 enhanced neuronal GABAergic transmission in whole-cell patch-
clamp recordings on cultured hippocampal neurons and elevated pentylenetetrazole (PTZ)-induced
seizure
threshold
in
mice.
4
ACS Paragon Plus Environment

Page 5 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Results
A pharmacophore model of VA and LOR reveals common structural features
In general, ligand-based pharmacophore modelling aims to perceive relevant common
stereoelectronic features of two or more ligands that are active towards a target of interest. VA and
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
LOR are both efficacious modulators and putatively interact with the same site on GABA Rs.36
A
With the aim to obtain compounds sharing the promising pharmacological properties of VA29–32
combined with a synthetically easily accessible structure like that of LOR we generated a ligand-
based pharmacophore model using the software LigandScout50 (for detailed procedure see
Experimental Section). Interestingly, superimposing the ligands revealed common structural
features indicating structural requirements for efficacious, subunit-selective GABA R modulation
A
(Fig. 1A). The aligned conformations of VA and LOR together with the resulting pharmacophore
model are shown in Fig. 1B (left panel). The final pharmacophore model comprises three
hydrophobic features and two H-bond acceptors (HBA) which correspond to the chloride/methyl-
groups located at the ring structures and the carboxylic acid/triazole-group, respectively (Fig. 1A).
This pharmacophore model was then used as a search filter for a virtual screening of different
compound vendor databases. From the obtained hit-lists a single, from a synthetical viewpoint
easily accessible compound (compound 0, for structure see Suppl. Fig. 1) was picked and further
simplified to remove structural parts that were not required to match the query pharmacophore
model. This resulted in compound 30 representing the first synthesized and tested hit compound of
this study. For all compounds a pharmacophore-fit score (a quantitative metric of the
pharmacophore-modelling software LigandScout) was calculated (for details see Suppl. Fig. 2).
For compound 30 one of the highest scores could be calculated making it a compound of particular
importance. An alignment showing the good match of compound 30 and the pharmacophore model
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
is depicted in Fig. 1B, middle panel. All further compounds were derived from compound 30,
guided by the pharmacophore model and differ in aspects like the type of HBA moiety, the type of
phenyl ring substituents and changes in the side-chain (see Scheme 1).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Highly efficacious modulation of α1β3γ2S GABA Rs by pharmacophore-derived ligands
A
Compound 30 can be structurally divided into three entities: a ring-structure with hydrophobic
substituents, a short linker, and a hydrogen bond acceptor moiety. Based on compound 30, we
generated compounds subdivided into four scaffolds: a) phenylethene (1-20), b) pyridinoneethene
(21-23), c) phenylethane (24-27) and d) phenylether derivatives (28-39) (see Scheme 1).
In total, 39 compounds were synthesized either according to database suggestions or by continuous
modification in iterative cycles after pharmacological screening. 30 compounds elicited significant
IGABA potentiation on α1β3γ2S channels (GABA EC ) at 30 µM and 35 compounds at a
3-7
concentration of 100 µM, respectively (statistically significant difference from zero calculated
using unpaired student’s t-test, p<0.05). Notably, the phenylethene derivatives (compounds 1-20)
were more active at 30 and 100 µM compared to phenylethane, pyridinoneethene and phenylether
derivatives (Fig. 1C). In addition – as illustrated in Fig. 1C - 5 compounds (12, 15, 16, 17 and 18)
also induced significantly stronger IGABA modulation than VA at 30 µM, and 12 compounds (4, 5,
10, 11, 12, 15, 16, 17, 18, 24, 25 and 26) stronger modulation at 100 µM (one-way ANOVA, post-
hoc Dunnett; for details see Table 1). Compared to LOR, two compounds (12 and 15) modulated
GABA Rs more efficaciously at 30 µM and seven (10, 11, 12, 15, 16, 18, 26) at 100 µM. The
A
strongest IGABA modulatory effect at 30 µM was evoked by 15 (1488 ± 140%; n = 5), while
compound 16 was most efficacious at 100 µM (2181 ± 93%; n = 4), both comprising a phenylethene
scaffold.
6
ACS Paragon Plus Environment

Page 7 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Synthesis
Synthesis of β-monosubstituted acrylonitriles and β-ether-type propionitrile/carboxylic acids
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compounds 1, 3, 5, 6, 9, 10 and 19 were prepared through a Suzuki-Miyaura cross-coupling
starting from the corresponding phenylboronic acid and the commercially available mixture of E/Z-
3
-bromo-2-methylacrylonitrile. Silica gelchromatographic purification yielded compounds 5, 6, 9
and 10 as single geometric isomers. Compounds 1 and 19 showed very low activity, therefore,
these compounds were used as E/Z-mixtures and no effort was invested into their resolution.
Isomers 5 and 6 were obtained through resolution of compound mixture 3. Scaffold 22 was formed
upon the base mediated nucleophilic displacement of 3,5-dimethylpyridine-2-pyridone on E/Z-3-
bromo-2-methylacrylonitrile. A secondary alkylbromide substituted functionality served as
excellent placeholder for directly forming nitrile-featuring aromatic ethers (28, 29, 31 and 33)
under basic conditions. The protocol was applicable also to ester-analogues, which upon alkaline
hydrolysis provided the carboxylic acids 35 and 36 (Scheme 2).
Synthesis of α,β-disubstituted acrylonitriles and amides
β,β-disubstituted acrylonitriles (4, 7, 8, 11, 12 and 20) were rapidly assembled through Wittig-
Horner-Emmons-type reactions starting from diethyl cyanomethylphosphonate and aromatic
5
1
ketones in the presence of a lithium amide base (LiHDMS). In some instances (7, 11, 12, 20) the
chromatographic purification succeeded in yielding geometrically pure materials. An activated
methylene substrate such as malononitrile was employed under Knoevenagel conditions (HDMS
5
2
in AcOH) for accessing the alkenyl dinitrile 15 in high chemical yield. The copper (II)-catalyzed
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
hydration of nitriles smoothly yielded the corresponding amides eventually separable by
5
3
chromatography (Scheme 3).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Synthesis of saturated nitrile derivatives
54
The Corey’s Mg-mediated reduction of acrylonitrile-type compounds (3, 10, 11) directly gave
the saturated analogues 24-26 in very high yields. Notably, compound 27 was synthesized via the
5
5
56
In-catalyzed silane-mediated deoxygenation of the readily accessible cyanohydrine (Scheme
4
).
Synthesis of 1H-tetrazoles
Sedelmeier’s Et AlN -mediated cycloaddition of nitriles yielded 1H-5-substituted tetrazoles in one
2
3
chemical step starting from the above prepared nitriles (1, 22, 28, 31, 29, 33)57,58 (Scheme 5).
Synthesis of enantiopure aryloxynitriles
Upon treatment of the primary amide with oxalyl chloride, the corresponding enantiopure
compound 38 was obtained in excellent yield. Alternatively, analogue 39 resulted from the direct
reaction of the ester with dimethyl aluminium amide (Scheme 6).59
Structure-activity relationship of α1β3γ2S GABA R modulation
A
The influence of a rigid olefinic linker and the effect of E/Z-configuration of variously substituted
ring scaffolds on IGABA modulation was studied by means of a small, focused library comprising
8
ACS Paragon Plus Environment

Page 9 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
0 phenylethene derivatives (1-20). Most notably, the Z-configured compound 12 elicited
markedly increased (~4-fold) IGABA potentiation at 30 µM compared to the E-isomer 11 (12: 1040
± 122%, n = 7 vs. 11: 278 ± 84%, n = 6). This finding was surprising since E-configuration
represents a much better match to the pharmacophore model. Fig. 1B, right panel, shows the
aligned conformations of 11 and 12 together with the resulting pharmacophore model favouring
the E-isomer 11. However, most of the E-configured olefinic nitriles showed IGABA enhancement
comparable to their reduced counterparts at both 30 and 100 µM (see compounds 5, 10, 11 and 24,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
5, 26, 27). Furthermore, we observed that omitting β-methyl groups did not significantly alter the
IGABA modulatory effect in E-configured ligands (see compounds 5, 10, 11), but resulted in a
dramatic activity drop in Z-configured nitriles (6, 9) (see Table 1).
The pharmacophore model obtained from VA and LOR also indicated a common HBA moiety
(Fig. 1A and B) potentially critical for IGABA modulation. This prompted us to further study
variations of the HBA moieties of differently configured derivatives. We and others previously
reported that substitution of the carboxylic acid residue of VA to the corresponding amide as well
as nitrile and tetrazole analogs30,60 induced more potent and/or efficacious GABA R modulation.
A
Interestingly, converting the nitriles 11 and 12 to the corresponding amides 13 and 14 resulted in
significantly decreased activity of both compounds (13: 175 ± 15% and 603 ± 39%; 14: 110 ± 6%
and 397 ± 10%, n = 6).
Since our efforts yielded in stereoisomeric compounds where both E and Z-configured isomers
exerted strong IGABA potentiating activity, we speculated that introduction of a second HBA in α-
position might cause additional interactions with pocket-forming residues and thus could enhance
activity. For this purpose, we synthesized an α-dinitrile-variant of compound 12 (15), E and Z-
cyano-amides (16 and 17 respectively) as well as the isomer-mix (18). Compound 15 induced
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
significantly stronger IGABA modulation at 30 µM compared to the Z-mono-nitrile 12 (15: 30 µM:
1
488 ± 140%, n = 5 vs. 12: 30 µM: 1040 ± 122%, n = 7; p<0.05, student’s t-test), while their effects
at 100 µM were comparable (15 100 µM: 1474 ± 153%, n = 5 vs. 12 100 µM: 1223 ± 140%, n =
). Thus, compound 15 exerted the strongest modulation of IGABA at 30 µM in this study. Cyano-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
7
amide derivatives 16 and 17 were both less efficacious at 30 µM (2-3-fold), however, at 100 µM
the E-α-cyano-nitrile 16 induced the most efficacious IGABA modulation of this set of derivatives
(100 µM: 2181 ± 93%, n = 4), while 17 was about 50% less effective (100 µM: 1085 ± 102%, n =
4
). The role of the ring substitution pattern was not assessed in a systematic way since only 2,4-
disubstituted compounds fitted the pharmacophore. Both 3,5-dimethyl compounds 1 (41 ± 7% and
82 ± 21%, n = 3) and 2 (48 ± 3% and 138 ± 6%, n = 3) induced only moderate IGABA potentiation
either due to the nature of the methyl-groups or the unfavourable substitution pattern.
The strategic decision to switch to a phenylethyl-linker combined with the introduction of small
hydrophobic residues in α/β-position resulted in 4 compounds (24-27, all racemic mixtures)
displaying substantially increased IGABA modulatory activity. Among all phenylethane derivatives,
compound 26 induced the strongest IGABA enhancement at both 30 and 100 µM (26: 383 ± 48%
and 1055 ± 98%, n = 5). Interestingly, replacing the chlorine by a methyl-group in para-position
strongly reduced the IGABA potentiating effect (27: 188 ± 22% and 545 ± 57%, n = 6) compared to
parent compound 26. Furthermore, swapping β-methyl to α-position (25) also significantly reduced
IGABA modulation by more than 40% at 30 µM (220 ± 34%, n = 5), however, at 100 µM enhanced
IGABA comparable to 26 (25: 840 ± 69%, n = 5). In contrast to the β-methylated ligands 26 and 27,
exchanging p-chloride to methyl among the α-methylated variants 24 and 25 did not affect IGABA
potentiation (24: 262 ± 22% and 931 ± 54%, n = 6 vs. 25: 220 ± 34% and 840 ± 69%, n = 5).
Notably, 3 compounds (24, 25, 26) induced similar IGABA potentiation at 30 µM, but were more
efficacious than VA at 100 µM (see Table 1).
10
ACS Paragon Plus Environment

Page 11 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Based on the starting compound 30 [5-(1-(4-chloro-2-methylphenoxy)ethyl)-1H-tetrazole] a set of
twelve phenylether derivatives was generated. Like compound 30, the other 1H-tetrazole
derivatives (32, 34, 37) and carboxylic acid derivatives (35, 36) displayed an IGABA (GABA EC3-7)
enhancement of <100% on α1β3γ2S receptors at a concentration of 100 µM. Replacing the 1H-
tetrazole by a carboxylic acid moiety did not increase their IGABA modulatory activity. In contrast,
ligands containing nitrile-groups (28, 29, 31, 33, 38, 39) were more efficacious with compounds
31 and 33 showing the highest IGABA potentiation at 30 and 100 µM (31: 110 ± 12% and 516 ± 58%
vs. 33: 125 ± 16% and 460 ± 63%; n = 3) (see Fig. 1B and Table 1 for detailed results). The set of
phenylether compounds tested here did not exceed the IGABA potentiating effect of VA and LOR,
however, IGABA potentiation of compounds 29, 31, 33, 38 at 100 µM was not statistically different
as compared to VA (one-way ANOVA, post-hoc Dunnett).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Other variants of the 3,5-substituted ring scaffold, pyridinoneethene compounds, comprising either
nitrile or tetrazole (21-23) were inactive. Attempts to enhance activity using 2,4-di-fluoro
substituted compounds (7, 8) as well as 1-napthyl-scaffolds (19, 20) were not successful suggesting
a critical role of the electron-density of the ring-scaffold (see Table 1).
Pharmacophore-derived ligands show potent and highly efficacious IGABA enhancement on
α1β3γ2S receptors
Compounds 4, 5, 10, 11, 12, 15, 16, 17, 18 and 26 were classified as highly active (based on their
IGABA potentiation at 30 µM) and chosen for further pharmacological characterization. Compounds
11, 12, 15, 16 and 18 displayed significantly more efficacious IGABA modulation than VA and
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
amongst those, 2 compounds (16 and 18) were also significantly more efficacious than LOR (Fig.
2
). All compounds enhanced IGABA with similar potency as VA and LOR (see Table 2) except
compound 26 which was significantly less potent.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Potency of IGABA enhancement of the structurally simplified compounds 12 and 15 was comparable
to that of VA and LOR, thus, they constitute the most potent IGABA modulators combined with high
efficacy on α1β3γ2S receptors in this study (12: EC : 13 ± 2 µM, Emax: 1896 ± 114%, n = 7 and
5
0
1
5: EC : 13 ± 4 µM, Emax: 1931 ± 324%, n = 7). In addition, we compared IGABA potentiation of
50
the 10 ligands at submaximal concentrations (1, 3 and 10 µM). This analysis revealed that also
compounds 12 and 15 enhanced IGABA more efficaciously at submaximal concentrations than VA
(p<0.0001, one-way ANOVA post-hoc Dunnett). Similarly, a highly significant difference
compared to both VA and LOR could be observed at all three concentrations (p<0.0001, see Suppl.
Table 1).
Pharmacophore-derived ligands retain β2/3 subunit-selectivity
GABA R modulators with higher selectivity for β2/3- over β1-containing channels are predicted
A
to exert fewer side effects such as sedation or ataxia30,61,62 and thus represent important scaffolds
for drug development. We have, therefore, studied whether the most promising compounds of the
pharmacophore-based compound library exhibited similar subunit-selectivity like VA and LOR.
We selected compounds 4, 12, 15, 16 and 17 and studied their effects on IGABA through α1β1γ2S
and α1β2γ2S receptors and compared them to their effects on β3-comprising receptors.
Compounds 4, 12 and 15 showed significantly higher potency on β2/3- compared to β1-containing
receptors (see Table 3). For compounds 16 and 17 EC and E at β1-containing receptors could
5
0
max
12
ACS Paragon Plus Environment

Page 13 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
even be not determined due to lack of saturation at tested concentrations up to 300 µM suggesting
a pronounced left-shift of the β1 concentration-response curve. To obtain additional statistical
evidence of β2/3-selectivity by 16 and 17 we analysed the differences in IGABA modulation at
concentrations 10, 30 and 100 µM. We found highly significant differences for both compound 16
and 17 on β2/3- compared to β1-containing receptors showing β2/3 subunit-selectivity (see Suppl.
Table 2). Our data provide evidence that these novel, simplified compounds derived from the
ligand-based pharmacophore model based on VA and LOR are not only highly efficacious and
potent, but in addition also retain β2/3 subunit-selectivity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compounds 12 and 15 might share the binding pocket of VA and LOR
Recently, we suggested that VA interacts with residues β3N265, β3R269, β3M286 and β3F289
3
6
located at the β+/α− interface in the upper half of the TMD on α1β3γ2S receptors. Mutations of
β3N26528,29,36 and β3F28936 resulted in an almost complete loss of efficacy and potency. In
addition, M235 located within the α1 subunit is another critical determinant of efficacious IGABA
modulation by VA. β2/3N265 was also shown to be critical for the IGABA modulatory effect of
LOR40–42 supporting a common binding pocket at the β+/α− interface.
It was therefore interesting whether IGABA enhancement by compounds 12 and 15, the most potent
compounds from our pharmacophore-driven approach, also requires residues α1M235, β3N265
and β3F289, respectively, and thus might interact with the same pocket. As illustrated in Fig. 4,
mutation of β3N265 to the corresponding serine of the β1 subunit induced a right-shift of the
concentration-response-curve of 12. Furthermore, besides this loss of potency a reduction of
efficacy was observed (EC : 103 ± 16 µM; Emax: 1001 ± 75%, n = 5, see Suppl. Table 3). A
5
0
similar observation was made for β1-comprising wildtype receptors (see Fig. 3A). For compound
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
5, a similar reduction in efficacy, but not in potency could be observed (EC : 24 ± 3 µM; Emax:
5
0
8
42 ± 39%, n = 4). With the prospect of imitating the binding pocket of β3-comprising receptors
we introduced an asparagine residue specific for β2/3 subunits into the homologous position of β1
receptors (β1S290N). For both 12 and 15, this mutation induced potency and efficacy comparable
to β3 wildtype receptors (12: EC : 10 ± 1 µM; E : 1775 ± 102%, n = 4 vs. 15: EC : 11 ± 2 µM;
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
max
50
Emax: 1745 ± 113%, n = 4). The β3F289S mutation in TM3 dramatically reduced efficacy (5-6-
fold) of 12 (Emax: 357 ± 52%, n = 7) and 15 (Emax: 292 ± 38%, n = 5) and induced a significant loss
of potency of compound 12 (171 ± 35 µM, n = 7), while for 15 a strong trend towards reduced
sensitivity could be observed (EC : 52 ± 17 µM, n = 5). Moreover, the α1 subunit mutation
5
0
M235W substantially reduced efficacy of 12 (Emax: 585 ± 51%, n = 7) and 15 (Emax: 541 ± 101%,
n = 4, for detailed results see Suppl. Table 3).
Furthermore, compounds 12 and 15 failed to evoke currents when applied to uninjected control
oocytes that did not express GABA Rs. Effects of 12 and 15 were fully reversible on wash-out and
A
potently inhibited by the GABA R pore blocker picrotoxin (Suppl. Fig. 3).
A
Taken together, these findings suggest a crucial involvement of residues α1M235, β3N265 and
β3F289 in either binding of 12 and 15 or transduction of the IGABA modulatory effect and support
a common binding pocket with VA.
Compound 12 and 15 potentiate tonic currents and 12 prolongs decay of mIPSCs in
cultured hippocampal neurons
GABA Rs can be categorized according to their subunit assembly and distribution within neurons
A
as synaptic and extrasynaptic receptors mediating phasic and tonic inhibition, respectively.63,64 To
14
ACS Paragon Plus Environment

Page 15 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
reveal whether compound 12 and 15 affect synaptic, extrasynaptic or both types of GABA R-
A
mediated inhibition, effects on miniature inhibitory postsynaptic currents (mIPSCs) and tonic
currents were determined in cultured hippocampal neurons. Neither compound 12 nor 15 (10 µM)
did alter mean frequencies (12: 78 ± 14% of control, n = 5 vs. solvent: 95 ± 44% of control, n = 5)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(15: 110 ± 74% of control, n = 6 vs. solvent: 112 ± 149% of control, n = 6) or amplitudes (12: 97
±
20% of control, n = 5 vs. solvent: 100 ± 26% of control, n = 5) (15: 110 ± 11%, n = 6 vs. solvent:
100 ± 26%, n = 6) of mIPSCs. 12 significantly prolonged current decay times (12: 213 ± 41% of
control, n = 5 vs. 100 ± 23% of control, n = 5; p<0.001), while 15 displayed a strong trend towards
prolonged mIPSC decay, which, however, was not significant (15: 133 ± 16% of control, n = 6 vs.
solvent: 100 ± 51%, n = 6), p = 0.163) as depicted in Fig. 5 (B and C).
However, both compound 12 and 15 significantly increased holding currents (Fig. 5A). To reveal
whether this change in holding currents was related to an action on tonic GABAergic conductance,
3
0 µM of the GABA R antagonist bicuculline methiodide (BMI) was applied in the continuous
A
presence of either 10 µM compound 12, 15 or solvent (Fig. 5D). BMI’s effect (i.e. decreasing
holding currents) was significantly larger for compound 12 and 15 than for solvent (12: -38 ± 7
pA, n = 6 vs. solvent: -22 ± 5 pA, n = 6, p<0.05) (15: -58 ± 10 pA, n = 6 vs. solvent: -37 ± 7 pA, n
=
6, p<0.05) and a similar result was obtained when 30 µM VA was compared with solvent (VA:
-115 ± 29 pA, n = 6 vs. solvent: -63 ± 24 pA, n = 6, p<0.05) (Fig. 5E). This proves that both
compound 12 and 15 enhance tonic GABA R-mediated currents.
A
Compounds 12 and 15 elevate PTZ-induced seizure threshold
Enhancing GABA R-mediated neurotransmission is a major strategy for the development of anti-
A
seizure drugs.65–68 Both LOR41,42,67,69 and VA (and derivatives)31,32 display anticonvulsive effects
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
in vivo. Thus, we tested the anticonvulsive potential of compounds 12 and 15 in the PTZ-induced
seizure threshold model using male C57BL/6N mice. As depicted in Fig. 6, compound 12
significantly elevated PTZ-induced seizure threshold at doses 0.1 and 0.3mg/kg (control: 39.5 ±
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.8 mg/kg PTZ). The strongest anticonvulsive effect was observed at 0.3 mg/kg (52.6 ± 2.1 mg/kg
PTZ, n = 4, p<0.0001; compare VA at 10 mg/kg: 49.0 ± 1.8 mg/kg PTZ, n = 4). Compound 15 also
significantly elevated seizure threshold at 0.3 mg/kg (15: 50.2 ± 1.9 mg/kg PTZ, n = 4, p<0.0001),
but no effects were observed at other tested doses. LOR was previously reported to provide
42
significant protection from PTZ-induced seizures only at doses >30 mg/kg. These results are in
line with our observations where LOR did not show anticonvulsive effects at doses <3 mg/kg (3
mg/kg: 46.5 ± 2.8 mg/kg PTZ, n = 4 vs. control: 39.5 ± 0.8 mg/kg, n = 6, p<0.01). Thus, compared
to VA and LOR, 30-fold lower doses of 12 are needed to achieve comparably pronounced
anticonvulsive activity (0.1 mg/kg 12 vs. 3 mg/kg VA/LOR; see Fig. 6, data for seizure threshold
3
1
32
elevation by VA are taken from Hintersteiner et al. and Khom et al. ). However, we observed a
sudden drop in activity of 12 and 15 to baseline at 1 mg/kg similar to VA and other previously
tested VA derivatives.31,32
Taken together, compound 12 (at ≥0.03 mg/kg) and 15 (at 0.3 mg/kg) displayed protection against
PTZ-induced seizures by elevating the seizure-threshold at significantly lower doses than VA and
LOR reflecting more potent anticonvulsive activity.
Activity of compounds 12 and 15 on other receptors of the central nervous system
The dose response curves of 12 and 15 in the PTZ test were bell-shaped suggesting loss of
anticonvulsant activity at higher concentrations. To address the question whether high
concentrations activate other receptors that might potentially counteract the anticonvulsive activity,
16
ACS Paragon Plus Environment

Page 17 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
we studied the effects of 12 and 15 on four major CNS receptors: the neuronal α4β2 nicotinic
acetylcholine receptor (nAChR), AMPA-type glutamate receptor (GluR2), N-methyl-D-aspartate
receptor (NMDAR) and the transient receptor potential vanilloid 1 receptor (TRPV1).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
At 10 µM (corresponding to EC of I
modulation) and at 100 µM (a concentration close to
50
GABA
saturation of IGABA modulation, Fig. 2A) neither 12 nor 15 affected agonist-induced currents
through NMDARs and neither compound activated TRPV1 (Fig. 7A, B). At 10 µM neither 12 nor
1
5 affected currents through nAChRs or GluR2 receptors. However, co-application of 100 µM
significantly inhibited inward currents induced by 100 µM acetylcholine (ACh) or glutamate (Glu)
nAChR - 12: 100 µM: 70 ± 6% of control, p<0.01; nAChR - 15: 100 µM: 58 ± 7% of control,
(
p<0.01 and GluR2 - 12: 100 µM: 73 ± 5% of control, p<0.01; GluR2 - 15: 100 µM: 79 ± 2% of
control, p<0.01), respectively (Fig. 7C, D).
Discussion
In this study, we present the design and synthesis of a library of novel and easily accessible
compounds derived from the β2/3 subunit-selective GABA R modulators VA and LOR. We
A
focused on β2/3-selective GABA R modulators because sparing modulation of β1-containing
A
6
1
receptors is associated with reduced sedation, ethanol potentiation , reduced abuse potential and
liability to cognitive decline in rodents.62
The β2/3-selective ligands VA and LOR - two seemingly structurally unrelated compounds -
require the presence of β2/3N265 for potent and efficacious GABA R modulation, therefore likely
A
sharing a common binding pocket.28,36 Thus, these ligands served to generate a pharmacophore
5
0
model that was obtained via the pharmacophore modelling software LIGANDSCOUT 4.04. The
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
pharmacophore is relatively small and simple – consisting only of hydrophobic interactions
surrounding a ring motive and a hydrogen-bond acceptor attached via a short linker. Our
pharmacophore model-supported design effort resulted in compounds with more efficacious and
equally potent IGABA modulation of α1β3γ2S receptors expressed in Xenopus laevis oocytes and
anticonvulsive effects of the most potent compounds. By exploring structure-activity relationships
of 39 compounds within the library, we identified structural features required for potent and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
efficacious GABA R modulation. In addition, the top 5 compounds (4, 12, 15, 16, 17) retained
A
β2/3 subunit-selectivity or at least showed a strong trend towards subunit-selective IGABA
modulation. Intriguingly, for these compounds subunit-selectivity did not seem to be stereoisomer-
specific as single isomers (E: 16; Z: 12, 17), the non-chiral α-di-substituted compound (15) and the
isomer mixture 4 showed β2/3 subunit-selective IGABA modulation. Surprisingly, the Z-isomer 12
modulated α1β3γ2S receptors more potently compared to the E-isomer 11 even though the E-
configuration is clearly preferred by the pharmacophore model derived from VA and LOR, two E-
configured ligands. As compounds 12 and 15 represented the most potent ligands in this study they
were analysed in more detail.
Our data suggest that compounds 12 and 15 designed based on the pharmacophore model might
also interact with the previously suggested binding pocket for VA. Like for VA, potent and
efficacious IGABA modulation by compounds 12 and 15 is highly dependent on the presence of
amino acids β3N265 and β3F389. This is shown by the substantial loss of activity on β3N265S
receptors and a gain in activity on β1S290N comprising receptors similar to β3 wildtype.
Compounds 12 and 15 might exert their β2/3 subunit-selective effects by specific H-bond
interactions between the nitrile and the amide group of β3N265. Since the introduction of a second
nitrile group in α-position (15) did not further increase the IGABA modulatory effect, we speculate
that activity is mainly determined by the cis-nitrile-group and that no additional H-bond
18
ACS Paragon Plus Environment

Page 19 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
interactions exist potentially promoting IGABA modulation. Moreover, the β3F289S mutation also
strongly affected IGABA enhancement by compounds 12 and 15 conceivably due to the resulting
loss of essential π-π-interactions of the phenylalanine- and 12/15-benzene-ring. This is consistent
with the loss of activity of the 3,5-diflourinated compounds 7 and 8 due to the electron withdrawing
effect of the highly electronegative fluoride atom. Efficacy of 12 and 15 was also decreased about
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
-fold on the α1M235W mutant, which is also in line with the data obtained for VA, presumably
occluding the entrance or reducing the volume of the pocket.36
Currents evoked by 12 and 15 were fully reversible on wash out and the activated channels were
inhibited by the known potent GABA R blocker picrotoxin. No current activation was observed in
A
uninjected control oocytes that did not express GABA Rs. Taken together this suggests, that the
A
observed effects are specifically evoked through GABA Rs and that the possibility of irreversible
A
binding of 12 and 15 can be excluded.
Furthermore, we studied the effect of 12 and 15 on spontaneous GABA transmission (mIPSC) in
cultured rat hippocampal neurons. Compound 12 at 10 µM significantly increased the decay time
of mIPSCs indicative of synaptic GABA R enhancement, while not affecting frequencies or
A
amplitudes of mIPSCs. In addition, compound 12 and 15 evoked a significant shift in holding
currents towards more negative values, and these currents were blocked by the GABA R
A
antagonist bicuculline suggesting enhanced tonic neuronal inhibition most likely mediated by
extrasynaptic GABA Rs. Thus, compound 12 facilitates both phasic as well as tonic GABAergic
A
inhibition. At present, we can only speculate which GABA R subunit combination might mediate
A
the synaptic or extrasynaptic effects. In hippocampal neurons, several subunit compositions have
been reported for synaptic and extrasynaptic GABA Rs: amongst synaptic receptors, the
A
combination α1β2γ2 is found most frequently, whereas α1βδ, α4βδ and α5βγ‐containing pentamers
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
are most likely expressed extrasynaptically.63,64,70 Addressing the question of subunit composition
of extrasynaptic GABA Rs targeted by compound 12 and 15 warrants further studies.
A
Previous studies demonstrated elevation of PTZ-induced seizure threshold by VA31,32 and
LOR.41,42,67,69 We were therefore interested if 12 or 15 display similar activity. Compound 12
significantly elevated the PTZ-induced seizure threshold at 0.1 and 0.3 mg/kg (compared to 3
mg/kg VA and LOR) while 15 elevated the seizure threshold only at 0.3 mg/kg (Fig. 6). Thus, with
respect to an effective dose 12 and 15 displayed stronger anticonvulsive activity than VA and LOR.
However, anticonvulsive effects of 12 and 15 are lost at higher doses (i.e. 1 mg/kg) (see also
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
31
32
Hintersteiner et al. and Khom et al. for similar effects of VA and derivatives). This might result
either from inhibition of GABA Rs due to an open-channel block at higher drug concentrations
A
2
8
71,72
(as shown for VA ) and/or from excitatory stimuli due to binding to other receptors.
To address
this latter question, we studied the effects of 12 and 15 on four principal excitatory CNS channels:
NMDA, TRPV1, α4β2 nACh and GluR2 receptors. NMDARs and TRPV1 receptors were not
affected by 10 or 100 µM of 12 and 15. Co-application of 10 µM 12 or 15 with either Glu or ACh
did not affect α4β2 nAChRs and GluR2 receptors suggesting preferential interaction of these
compounds with GABA Rs. However, at a concentration of 100 µM both compounds caused
A
significant inhibition of agonist-induced inward currents through neuronal α4β2 nACh and GluR2
receptors. While inhibition of GluR2 receptors will contribute to anticonvulsant activity,73,74 the
situation is more ambiguous with blockers of α4β2 receptors. On the one hand, the anticonvulsant
activity of carbamazepine, at least in Autosomal Dominant Nocturnal Frontal Lobe Epilepsy
7
5
(ADNFLE), is believed to involve antagonism at α4β2. On the other hand, activation of α4β2
nAChRs promotes GABA release in various brain regions including the hippocampus,76 and
inhibition of these receptors must be expected to counteract the anticonvulsant effects of 12 and
20
ACS Paragon Plus Environment

Page 21 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
5. We speculate that the drop in anticonvulsant activity in the PTZ test is at least partially caused
by inhibition of nAChRs. Interactions with other receptors can, however, not be excluded.
In summary, we present a straightforward approach for identification of novel β2/3 subunit-
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
selective GABA receptor ligands and obtained important new insights into the molecular base of
A
subunit selectivity. Guided by a pharmacophore-based compound design strategy we generated a
set of highly active and structurally simple, small molecules that may serve as scaffolds for the
development of GABA R modulators with anticonvulsive activity. Future efforts to improve the
A
properties of compounds 12 and 15 will have to focus particularly on optimising the dose-
dependency of their anticonvulsant activity.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Experimental Section
Ligand-Based Pharmacophore Modeling
50
The ligand-based pharmacophore model was derived using the software LigandScout by the
following procedure: First, conformational ensembles were generated for VA and LOR (max. 500
conformations) to account for the flexibility of their side chains. Then, using VA as the reference
ligand, a flexible alignment of VA and LOR was carried out to find the best matching conformation
pair of both ligands. From this best matching conformation pair a shared feature pharmacophore
was derived which resulted in a set of features common to VA and LOR that match both in feature
type and inter feature distances. The pharmacophore was then used for the virtual screening of
various compound vendor databases.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Interference compounds
All compounds of this study were examined for known classes of assay interference compounds
by using publicly available filter software (http://zinc15.docking.org/patterns/home,
http://www.cbligand.org/PAINS/,
http://fafdrugs4.mti.univ-paris-diderot.fr/,
7
7
http://advisor.docking.org). While compounds containing the α,β-unsaturated nitrile moiety were
flagged as potential covalent inhibitors, only compound 15 resulted as a potential interference
compound from this analysis (http://fafdrugs4.mti.univ-paris-diderot.fr/). To exclude false positive
effects due to unspecific or covalent binding, additional experiments on effects of hit compounds
1
2 and 15 on α1β3γ2S receptors were performed. As depicted in Suppl. Fig. 3, both compounds
2 and 15 showed no effects on uninjected oocytes and fully reversible currents specifically
1
induced through α1β3γ2S GABA Rs dispelling concern about covalent inhibition or effects due to
A
unselective binding.
22
ACS Paragon Plus Environment

Page 23 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Animals and animal welfare
All experiments involving animals were approved by the Austrian Animal Experimentation Ethics
Board in compliance with the European convention for the protection of vertebrate animals used
for experimental and other scientific purposes ETS no.: 123, which is in line with the EU Directive
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
010/63/EU (GZ 66.011/0123-II/3b/2015 and 66.006/0029-WF/V/3b/2014). Every possible effort
was taken to minimize the number of animals used.
Female Xenopus laevis frogs were purchased from NASCO (Fort Atkinson, USA) and kept in
groups in temperature-controlled, continuous-flow water tanks (20 ± 1°C). Male mice (C57BL/6N)
were obtained from Charles River Laboratories (Sulzfeld, Germany). Mice were group-housed
(maximum 5 mice per type II L cage) with ad libitum access to food and water. The temperature
in the holding (for mice and frogs) and testing facilities was fixed to 22 ± 2°C; the humidity ranged
between 40-60%; a 12h light-dark cycle was in operation (lights on from 07.00 to 19.00).
Ion channel expression in Xenopus laevis oocytes and two-microelectrode voltage-clamp
experiments
Preparation of stage V–VI oocytes from Xenopus laevis and synthesis of capped runoff poly(A)
cRNA transcripts from linearized cDNA templates (pCMV vector) was performed as described
7
8
elsewhere. Female Xenopus laevis frogs were anesthetized by 15 min incubation in a 0.2% MS-
222 (methane sulfonate salt of 3-aminobenzoic acid ethyl ester; Sigma Aldrich, Vienna, Austria)
solution before removal of parts of the ovaries. Follicle membranes from isolated oocytes were
enzymatically digested with 2mg/mL collagenase (Type 1A, Sigma-Aldrich, Vienna, Austria).
GABA R mutations α1M235W, β1S290N, β3N265S and β3F289S were introduced by site-
A
directed mutagenesis using the QuikChange mutagenesis kit (Agilent Technologies, Vienna,
Austria). The coding regions of plasmids were sequenced before experimental use. After cDNA
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
linearization, capped cRNA transcripts were produced using the mMESSAGE mMACHINE® T7
transcription kit (Life Technologies). Selected oocytes were injected with 10–50nL of DEPC–
treated water (diethyl pyrocarbonate, Sigma, Vienna, Austria) containing the different cRNAs at a
concentration ranging between 200–3000 pg/nL/subunit. To ensure expression of the γ2S subunit
in the case of α1β2/3γ2S, α1M235Wβ3γ2S, α1β1S290Nγ2S, α1β3F289Sγ2S receptors, cRNAs
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
7
9
were mixed in a ratio of 1:1:10. For expression of receptors composed of α1β3N265Sγ2S and
36
α1β1γ2S cRNAS were mixed in a ratio of 3:1:10. α4β2 nAChR were a generous gift of Petra
Scholze, Brain Research Center, Vienna.
The amount of cRNAs was determined by means of a NanoDrop ND-1000 (Kisker-Biotech,
8
0
Steinfurt, Germany). Oocytes were stored at +18°C in ND96 solution, all from Sigma-Aldrich,
Vienna, Austria). Chloride currents through GABA Rs (IGABA) were measured at room
A
temperature (20-24°C) by means of the two-microelectrode voltage clamp technique making use
of a TURBO TEC-05X amplifier (npi electronic, Tamm, Germany). Iagonist were elicited at a
holding potential of -70 mV. Data acquisition was carried out by means of an Axon Digidata 1440A
interface using pCLAMP v.11 (Molecular Devices, Sunnyvale, CA, USA). ND96 solution was
used as bath solution. External solution for recordings on NMDA receptors contained 96 mM NaCl,
2 mM KCl, 5 mM HEPES, 1.8 mM CaCl2*2H2O and 10 µM glycine. Microelectrodes were filled
with 3M KCl and had resistances between 1 and 3 MΩ.
Fast Perfusion System
Agonists (GABA, ACh, Glu, NMDA and piperine) and compounds were applied by means of a
49
fast perfusion system as described previously ; drug or control solutions were applied by means
of a TECAN Miniprep 60 enabling automation of the experiments (ScreeningTool, npi electronic,
Tamm Germany). To elicit Iagonist, the chamber was perfused with 150 μL of agonist- or compound-
24
ACS Paragon Plus Environment

Page 25 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
81
containing solutions, respectively, at a volume rate of 200 μL/s. To account for possible slow
recovery of GABA Rs from increasing levels of desensitization in the presence of high compound
A
concentrations, the duration of washout periods was extended stepwise, i.e. 1min (GABA EC )
3
-7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
to 1.5min (co-application of GABA EC3-7 in the presence of 0.1-1 µM compound;) to 2.5 min (co-
application of GABA EC3-7 in the presence of 3-10 µM compound) to 5 min (co-application of
GABA EC3-7 in the presence of 30-100 µM compound) to 15 min (GABA EC3-7 in the presence of
300 µM compound). For experiments in Xenopus oocyte assays all compounds were dissolved in
DMSO to a concentration of 100 mM. Concentrations >300 µM were not tested to keep DMSO
concentrations as low as possible (max. 0.3%). Oocytes with maximal current amplitudes >5 µA
were discarded due to exclude voltage-clamp errors.
Modulation of GABA , nAChR, GluR2 and NMDAR
A
Enhancement of chloride currents by modulators of the GABA R was measured at a GABA
A
concentration eliciting between 3 and 7% of the maximal current amplitude (EC ). The GABA
3
–7
EC3–7 was determined for each oocyte individually. Enhancement of the chloride current was
defined as (I(GABA+compound)/IGABA) – 1, where I(GABA+compound) is the current response in the presence
of a given compound and IGABA is the control GABA current. Emax reflects the maximal IGABA
enhancement. Concentration–response curves were generated, and the data were fitted by nonlinear
regression analysis using Origin Software (OriginLab Corporation, USA). Data were fitted to the
xnH
nH
nH
following
equation:
y
=
min
+
(max-min)
*
/
(k +x ).
In this equation, k corresponds to the EC value, x-values represent concentrations, and n stands
5
0
H
for the Hill coefficient. Each data point represents the mean ± SEM from at least 3 oocytes and ≥2
oocyte batches.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Enhancement of currents through nAChR, GluR2 and NMDAR was measured at 100 µM of ACh,
Glu and NMDA, respectively. Effects of compounds on agonist responses were defined as
(I(agonist+compound)/Iagonist)*100, where I(agonist+compound) is the current response in the presence of a given
compound and Iagonist the control current.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Cell culture
8
2
Primary cultures of rat hippocampal neurons were prepared as described before. Neonatal (P0 -
P1) Sprague-Dawley rat pups were killed by decapitation, immediately thereafter hippocampi were
collected, cut into small pieces and digested in papain (1 mg/ml, Worthington, Lakewood, NJ,
USA, in Leibovitz Medium, Sigma-Aldrich, Vienna, Austria, supplemented with 1 mM kynureinic
acid, Sigma-Aldrich) for 20 min at 37°C. Afterwards, the tissue was dissociated by trituration in
Dulbeco’s Modified Eagle’s Medium (DMEM, Sigma-Aldrich) containing 10% heat-inactivated
Fetal Bovine Serum (Biochrom, MB Stricker, Tutzing, Germany), 5 µg/ml insulin, 5 µg/ml
transferrin, 5 ng/ml Na-selenite, 19 µg/ml putrescine, 10 nM progesterone, 2 mM MgCl , 25 000
2
IU/l penicillin and 25 mg/l streptomycin. Subsequently, approximately 50 000 cells were seeded
into a glass ring placed on a 35 mm cell culture dish (Nuclon, Bartelt, Graz, Austria) coated with
poly-D-lysine (0.1 mg/ml, Sigma-Aldrich) and containing the culture medium mentioned above.
After 2-3 hours, the glass rings were removed and on the next day, the medium was exchanged to
3
ml penicillin - streptomycin-free medium. After 3-5 days, 1 µM cytosine arabinoside (Sigma-
Aldrich) was added to the culture medium to prevent proliferation of non-neuronal cells. The cells
were kept at 37°C in a humidified 5% CO atmosphere for 3 weeks before recordings were made.
2
Patch-clamp electrophysiology
Patch-clamp recordings on primary neurons were performed at room temperature (20-24 °C) using
an Axopatch 200B amplifier, a Digidata 1440 interface and the Clampfit 10.4 software (Molecular
26
ACS Paragon Plus Environment

Page 27 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Devices, Sunnyvale, California, USA). Miniature inhibitory post synaptic currents (mIPSC) were
recorded in the whole-cell configuration of the voltage-clamp mode at -70 mV. mIPSC recordings
were low-pass filtered at 2 kHz and digitized at 20 kHz and analysed offline using Clampex 10.4
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(Molecular Devices) and miniAnalysis (Synaptosoft, Fort Lee, NJ, USA) software. The patch-
pipettes were pulled from borosilicate glass capillaries (GB150-8B, Science products, Hoffenheim,
Germany) using a Sutter P-97 puller (Sutter Instruments, Novato, California, USA). Pipette
resistances ranged from 2.0 to 2.5 M. The pipettes were front and back filled with internal
solution and the recordings were started 5 min after breaking through the membrane by applying
negative pressure, to allow for diffusion of the internal solution into the patched cell.
The internal solution contained (in mM): CsCl (130), tetraethylammonium chloride (20), CaCl2
(0.24), EGTA (5.0), Glucose (10), HEPES (10), CsOH (15) achieving a pH of 7.4. The external
solution contained (in mM): NaCl (134), Glucose (20), HEPES (10), KCl (5.8), CaCl (2.5), MgCl
2
2
(2.0), KOH (2.9), NaOH (2.1) resulting in a pH of 7.4, additionally the solution contained 10 µM
CNQX (Biotrend, Cologne, Germany) and 0.5 µM TTX (Latoxan, Valence, France). The
combination of these solutions does not result in the formation of a liquid junction potential. Drugs
were applied using a DAD-12 drug application device (ALA Scientific Instruments, Westbury,
New York, USA), which permits complete solution exchange surrounding the recorded cell within
5
0 to 150 ms. A 100 mM stock of the test substance was dissolved in DMSO and then further
diluted in external solution, hence the remaining external solution was supplemented with 0.1%
DMSO (solvent). In mIPSC recordings, Test compounds were applied for 300 s subsequent to and
followed by 300 s application of solvent (compound free external control solution), respectively.
Thereafter 30 µM bicuculline methiodide (BMI, Sigma-Aldrich) was applied for 10 s to verify the
recording of mIPSCs. The first 60 s period of either solvent or test substance application were
omitted from the analysis of mIPSCs.
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 89
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
For recordings of tonic GABA currents, the same external and internal solution was used. After
establishing a stable whole-cell configuration, the cells were equilibrated with solvent (0.1%
DMSO) for 120 s before the start of the recording. The steady-state current at -70 mV was recorded
for 120 s in presence of solvent, followed by a 10 s application of BMI (30 µM). Thereafter the
test-substances (valerenic acid, VA, 30 µM; 12 or 15, 10 µM) were applied for 300 s, followed by
co-application of either test-substance with BMI, respectively. The average current amplitude of a
3 s-period of steady-state current at the end of the 10 s application of BMI was subtracted from a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
s-period immediately before the application of BMI (pA).
In vivo PTZ tests
Only male C57BL/6N mice (age 3-6 months) were used in the test described below. Intraperitoneal
(i.p.) injection of control or compound-containing solutions was done 30min before the test.
Indicated doses in the results and discussion section represent mg/kg bodyweight. Seizure threshold
was determined by pentylenetetrazole (PTZ) tail-vein infusion on freely moving animals at a rate
of 100 µl/min (10 mg/ml PTZ in saline, pH = 7.4). Infusion was stopped when animals displayed
generalized clonic seizures. Animals were immediately killed by cervical displacement after onset
of seizures. The seizure threshold dose was calculated from the infused dose in relation to body
weight (mg/kg).
Statistical analysis
Statistical significance for oocyte and in vivo experiments was calculated using either one-way
ANOVA followed by a post-hoc mean comparison with Dunnet or Tukey (OriginLab Corporation,
USA or GraphPad, La Jolla, USA) or student’s t-test. P-values of <0.05 were accepted as
statistically significant. All data are given as mean ± SEM. In patch-clamp experiments, n values
represent the number of recorded cells. For mIPSC recordings, data are presented as weighted
28
ACS Paragon Plus Environment

Page 29 of 89
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
arithmetic mean ± weighted SD. Statistically significant differences were calculated using a
student’s t-test. Statistically significant differences in the effects on mIPSC frequency were tested
using a Friedman test, followed by Dunn’s multiple comparisons test. In tonic GABA current
recordings, all values are presented as average ± SEM. Statistically significant differences between
solvent-treated and substance-treated BMI-sensitive currents were calculated using the Wilcoxon
matched-pairs signed rank test.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Chemistry: General Information and Synthetic Procedures
Melting points were determined on a Reichert–Kofler hot-stage microscope and are uncorrected.
1
13
19
HRMS spectra were recorded on a Bruker maXis 4G instrument (ESI-TOF, APCI). H, C, F
1
NMR spectra were recorded with a Bruker Avance III 400 spectrometer (400 MHz for H, 100
1
3
19
1
MHz for C, 377 MHz for F), with a Bruker Avance III 500 spectrometer (500 MHz for H, 125
13
1
13
MHz for C) and with a Bruker DRX 200 spectrometer (200 MHz for H, 50 MHz for C) at 297
K. The center of the (residual) solvent signal was used as an internal standard which was related to
1
1
1
TMS with δ 7.26 ppm ( H in CDCl ), δ 5.32 ppm ( H in CD D ), δ 4.78 ppm ( H in CD OD), δ
3
2
2
3
1
3
13
13
77.00 ppm ( C in CDCl ), δ 54.00 ppm ( C in CD D ), and δ 49.15 ppm ( C in CD OD). Absolute
3
2
2
3
19
referencing via Ξ ratio was used for the F NMR spectra. Spin-spin coupling constants (J) are
given in Hz. As far as possible, full and unambiguous assignment of all resonances was performed
by combined application of standard NMR techniques, such as APT, HSQC, HMBC, COSY and
NOESY experiments. THF was distilled over Na/benzophenone. Light petroleum refers to the
fraction of bp 40-60 °C. Chemicals were purchased from Sigma-Aldrich, Acros, Alfa Aesar,
Fluorochem and TCI Europe, otherwise specified. Solutions were evaporated under reduced
pressure with a rotary evaporator. TLC was carried out on aluminium sheets precoated with silica
gel 60F254 (Macherey-Nagel, Merck); the spots were visualized under UV light (λ = 254 nm)
29
ACS Paragon Plus Environment