The twin drug approach for novel nicotinic acetylcholine receptor ligands

Article abstract of DOI:10.1016/j.bmc.2015.06.034

The association of two pharmacophoric entities generates so-called 'twin drugs' or dimer derivatives. We applied this approach for the design of a small compound library for the interaction with α4β2 nicotinic acetylcholine receptors (nAChRs). In this compound series, the nAChR ligand N,N-dimethyl-2-(pyridin-3-yloxy)ethan-1-amine 9 served as one pharmacological entity and it was initially kept constant as one part of the 'twin' compound. 'Twin' compounds with identical or non-identical entities using the 'no linker mode' or 'overlap' mode were synthesized and evaluated for their nAChR affinities. Compound 17a showed the highest affinity for the α4β2 nAChR subtype (K_i = 0.188 nM) and its (di)fluoro analogs could retain nanomolar affinities, when replacing pyridine as the hydrogen bond acceptor system by mono- or difluoro-phenyls. The 'twin drug' approach proved to provide compounds with high affinity and subtype selectivity for α4β2 nAChRs.

Full text of DOI:10.1016/j.bmc.2015.06.034

Accepted Manuscript
The twin drug approach for novel nicotinic acetylcholine receptor ligands
Isabelle Tomassoli, Daniela Gündisch
PII:
S0968-0896(15)00528-3
DOI:
http://dx.doi.org/10.1016/j.bmc.2015.06.034
BMC 12390
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
30 March 2015
30 May 2015
10 June 2015
Please cite this article as: Tomassoli, I., Gündisch, D., The twin drug approach for novel nicotinic acetylcholine
receptor ligands, Bioorganic & Medicinal Chemistry (2015), doi: http://dx.doi.org/10.1016/j.bmc.2015.06.034
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The twin drug approach for novel nicotinic acetylcholine receptor ligands
Isabelle Tomassoli and Daniela Gündisch*
Department of Pharmaceutical Sciences, The Daniel K. Inouye College of Pharmacy, University
of Hawai’i at Hilo, Hilo, HI, U.S.A.
ABSTRACT
The association of two pharmacophoric entities generates so-called "twin drugs" or dimer
derivatives. We applied this approach for the design of a small compound library for the
interaction with α4β2* nicotinic acetylcholine receptors (nAChRs). In this compound series, the
nAChR ligand N,N-dimethyl-2-(pyridin-3-yloxy)ethan-1-amine 9 served as one pharmacological
entity and it was initially kept constant as one part of the "twin" compound. "Twin" compounds
with identical or non-identical entities using the "no linker mode" or "overlap" mode were
synthesized and evaluated for their nAChR affinities. Compound 17a showed the highest affinity
for the α4β2* nAChR subtype (Ki = 0.188 nM) and its (di)fluoro analogs could retain nanomolar
affinities, when replacing pyridine as the hydrogen bond acceptor system by mono- or difluoro-
phenyls. The "twin drug" approach proved to provide compounds with high affinity and subtype
selectivity for α4β2* nAChRs.
Keywords:
Twin drugs
Nicotinic acetylcholine receptor
nAChR
Structure-activity relationship
3D QSAR pharmacophore
Abbreviations: ADME, absorption/distribution/metabolism/excretion; BBB, blood-brain barrier;
CD3OD, tetradeuteromethanol; CDCl3, deuterochloroform; CH2Cl2, dichloromethane; ClogD,
distribution coefficient; ClogP, calculated partition coefficient; CNS, central nervous system;
DAT, dopamine transporter; DCC, N,N’-dicyclohexylcarbodiimide; DIAD, diisopropyl

azodicarboxylate; DMAP, 4-(Dimethylamino)pyridine; DMF, N,N-dimethylformamide; Et2O,
diethyl ether; Et3N, triethylamine; EtOAc, ethyl acetate; HA, heavy atom; HBA, hydrogen bond
acceptor; HBD, hydrogen bond donor; HCHO, formaldehyde; HCOOH, formic acid; HCl,
hydrogen chloride; HEPES, 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid; HSS, HEPES-
buffered salt solution; K2CO3, potassium carbonate; KBr, potassium bromide; KOH, potassium
hydroxide; LE, ligand efficiency; LLE, lipophilic ligand efficiency; MeCN, acetonitrile; MeI,
iodomethane; MeOH, methanol; MgSO4, magnesium sulfate; MW microwave or molecular
weight; nAChR, nicotinic acetylcholine receptor; NaHCO3, sodium hydrogen carbonate; NaOH,
sodium hydroxide; NaOtBu, sodium tert-butoxide; nBu4NBr, tetra-n-butylammonium bromide;
NHA, number of hydrogen bond acceptors; NHD number of hydrogen bond donors; Pd/C,
palladium on activated charcoal; PEI, poly(ethyleneimine); PPh3, triphenylphosphine; RMS, root
mean square; Ro5, rule of five; SerT, serotonin transporter; tBuOH, tert-butanol; THF,
tetrahydrofuran; TPSA, topological polar surface area; TRIS, Tri(hydroxymethyl)-
aminomethane; VMAT2, vesicular monoamine transporter 2; ZnBr2, zinc bromide.
*Corresponding author. Tel.: +1-808-933-2943; fax: +1-808-933-2974. E-mail:
danielag@hawaii.edu

1. Introduction
"Twin drugs", which can be also defined as hybrids, can be found among natural products with
potent bioactivity.1-3 Combining two active compounds in a single molecule increases interaction
points with biological target(s) and can provide desired multiple or complementary modes of
action. Such compounds are new entities with their own pharmacokinetics and
pharmacodynamic properties.1-4 The combination of two (identical or non-identical) biologically
active compounds can be classified by the type of connection: “linker mode” – “no linker mode”
– “overlap mode” (Fig. 1).
A
B
A
B
A
B
No linker Mode
Linker Mode
Overlap Mode
Figure 1: Different connection types for twin drugs/hybrids. Compounds A and B can be
structurally identical or non-identical. They can interact with the same or with different
biological targets.
Some natural products active on nAChRs are displaying a “twin” structure (Fig. 2). There is
e.g. lobeline 1 found in Lobelia inflata (Indian tobacco).5 Lobeline 1 can be considered as a
non-identical twin composed of sedamine and its carbonyl analog overlapped at the “cation part”
(N-methyl piperidine). The “cation part” is an important pharmacophoric element for the
interaction with nAChRs.6,7 Lobeline 1 interacts with α4β2* (Ki = 4-30 nM) and α7 (Ki = 6.6
µM) nAChRs, VMAT2, DAT, SerT, and opioid receptors.8 Anatalline 2 and 3,5-bis-(1-methyl-
pyrrolidin-2-yl)-pyridine 3 are minor alkaloids found in tobacco roots.9 Anatalline 2 has its
overlap at the “cation part” (piperidine moiety) like lobeline and it can be considered as a twin of
tobacco alkaloid anabasine with 4-phenylpiperidine, a privileged scaffold of a variety of
bioactive compounds. Anatalline 2 has been identified in tobacco smoke, but no pharmacological
studies have been reported so far.10,11 Alkaloid 3 is a merged compound of two nicotine
molecules overlapped at the pyridyl part displaying a hydrogen acceptor system.12 A hydrogen
acceptor system at a defined distance to the “cation part” is the second important
pharmacophoric element for the interaction with nAChRs.6,7 Compound 3 shows micromolar
affinity for α4β2* (Ki = 1.18 µM).12 Further, curare alkaloids are prominent examples of twin
drugs which interact with muscle type nAChRs (Fig. 2; tubocurarine 4).13 A prototype of a

synthetically derived “twin” is suxamethonium (succinylcholine) 5, a duplication of the
neurotransmitter acetylcholine used as a depolarizing muscle relaxant.13 Also dimers of the
highly potent nAChR frog alkaloid epibatidine 6 and of the plant alkaloid cytisine, e.g. 1,2-bis-
N-cytisinylethane (CC4) 7, have been synthesized and pharmacologically evaluated in the past.14-
16 Both series combine the single molecules at their basic nitrogen motif. Recently, the interferon
inducer tilorone 8 developed in the 1970s, was discovered as a selective α7 nAChR agonist (IC50
= 110 nM).17 Two identical diethylaminoethanol moieties, which can be considered as choline-
like motifs, are attached symmetrically to a fluorenone core. In general, the twin drug or hybrid
approach has been rarely used in the development of CNS-active nAChR compounds.
NAChRs are ligand-gated cation channels assembled as homo- or heteropentamers and
permeable for Na+, K+ and Ca2+.18,19 They are of great interest as therapeutic targets for a variety
of central nervous system (CNS) diseases, e.g. β2-containing subtypes, which seem to play a
major role in mood disorders and addiction.20,21
Figure 2: Structures of some natural product “twins” (1-4) active on nAChRs or/and found in
tobacco, “twins” synthetically derived from natural products (5-7), and tilorone (8).
Recently, we applied the “twin drug” concept for the development of a novel nAChR ligand,
BPC 17a (Fig. 3A).22 BPC 17a is a “non-identical twin drug” using the overlap mode in its
design (Fig. 3) and has partial agonist activity for α4β2* nAChRs, improved selectivity profiles
over varenicline and cytisine, and efficacy in the mouse tail suspension test.22 The design

strategy for BPC 17a has not been developed previously for α4β2* nAChR ligands, especially
using the hydrogen bond acceptor (HBA) moiety for the overlap mode. Here, we report on the
design, synthesis, and in vitro evaluation of “twin” compounds based on the “overlap” mode as
pharmacological tools and as new lead compounds for β2-containing nAChRs.
Figure 3: Design of BPC (17a) and compound 10
2. Design
In our search for α4β2* nAChR ligands with a desirable selectivity and functionality profile,
we used the “twin drug” approach as an alternative strategy in the optimization process of
biologically active compounds.1-4 Like mentioned earlier, the combination of two active ligands
in a single molecule increases interaction points with its biological target(s), which can lead to a
compound with a better on-/off-target profile.1-4 In addition, we took into consideration, that
substitution at position 5 of the pyridine moiety is mostly tolerated for nAChR activity including
the possibility of enhancing α4β2* subtype selectivity in contrast to non-substituted analogs.7
The design for the compound series reported here is based on N,N-dimethyl-2-(pyridin-3-
yloxy)ethan-1-amine 9 displaying a choline-like motif attached to a pyridyl moiety. Compound 9
is a non-azacyclic pyridyl ether nAChR ligand prototype developed by Abbott Laboratories and
has a Ki value of 32.6 nM for α4β2* nAChRs.23 Compound 9 was also described by Simsek et
al. as active on α4β2* nAChRs (Ki = 24 2.4 nM) and an ED50 value of 15 mg/kg in the mice
tail-flick assay.24 Firstly, we attached two molecules of N,N-dimethyl-2-(pyridin-3-yloxy)ethan-
1-amine 9 via “no linker mode” to each other applying the simplest “twin drug” design apparent

in suxamethonium’s (5) structure to obtain the “twin” compound 10. We were interested to
investigate, if the interaction with α4β2* could be retained (Fig. 3B). To enhance affinity for
α4β2* nAChR and to reduce the number of rotatable bonds, we designed a small “twin drug”
library based on two non-identical structural entities (Fig. 4 and Fig. 5). We chose the “overlap
mode” to keep compounds’ physicochemical parameters in CNS drug-like ranges.25 N,N-
dimethyl-2-(pyridin-3-yloxy)ethan-1-amine 9 and diverse diaza(bi)cyclic pyridines (11, 12a/b,
13a/b) (Fig. 4; Table 1) were overlapped at their pyridyl moieties to obtain the “twins” (Fig. 5A).
The diaza(bi)cyclic pyridines (11, 12a/b, 13a/b) used in this design were developed by
Romanelli et al., NeuroSearch A/S and Abbott Laboratories (Fig. 4; Table 1).26-32 Since
compound 3 has reduced affinity for α4β2* comparing with nicotine, we did not use the
“identical twin drug” approach in the overlap mode for the diaza(bi)cyclic pyridine series. In
addition, we were interested by introducing several azacyclic systems at position 5 at the pyridyl
moiety of compound 9 to investigate the influence of such moieties on nAChRs’ affinities (Fig.
5B). Here, the dimethylamine part of 9 could function as the “cation part” (Fig. 5B). In the
diaza(bi)cyclic pyridine series, we kept the acyclic “cation part” (choline like substructure)
constant (Fig. 5A). In addition, we replaced the pyridine by mono- or di-fluoro phenyl to keep
HBA properties (Fig. 5A). The fluorine substitution will enhance lipophilicity properties and
favorise BBB penetration, but will alter the hydrogen bond acceptor property.
Figure 4: Structures of nAChR ligands (9, 11, 12a/b, 13a/b) used as pharmacological entities for
the twin drug approach.
Diaza(bi)cyclic scaffolds for the twin drug approach
"cyclic cation part"
HBA systems
O
R
NH
A
N
R
N
N
N
acyclic "cation part"
N
N
Y
X
R: H, CH3
X/Y: N/CH
X/Y: CF/CH
X/Y: CF/CF
N-containing scaffolds for further SAR
HBA system
O
HO
O
B
N
N
N
N
acyclic "cation part"
Y
X
N
N
X/Y: N/CH

Figure 5: Design of nAChR ligands (17a-26). A: “Twin drugs”; B: N-containing scaffolds as
substituents at position 5 at the pyridyl moiety.
Table 1. Affinities, in vivo activity and calculated physicochemical parameters of known nAChR
ligands (“single entities”)
α4β2*
α7*
Cpd #
Structure
In vivo activity
Mr
ClogPd
TPSAd
logBBd
Ki [nM]a
Ki [nM]b
ED50 (mg/kg) =
15
(mouse tail-flick
test) [24]
32.6 [23],
24 2.4 [24]
9
165.21
0.93
25.36
0.09
MEDc = 1.9
µmol/kg
(mouse hot plate
test) [32]
11
0.02 [32]
202.27
176.23
0.97
0.65
28.16
28.16
-0.09
0.04
1.9 (IC50)
[28,31]
190 (IC50)
[31]
12a
150 (IC50)
[31]
4,600 (IC50)
[31]
12b
13a
13b
190.26
162.21
176.23
1.19
0.19
19.37
28.16
19.37
0.03
-0.03
0.07
90 [26]
90 [26]
0.88
d
a[3H]cytisine; [ H]alpha-bungarotoxin; MED: minimally effective dose; ClogP, TPSA, and
b 3
c
logBB values were calculated using ACD/ADME Suite 5.0.
3. Results
3.1. Chemistry
The twin compound 10 was prepared in one step (Scheme 1). Various methods can be used to
form a biaryl system. A general method is the Ullmann reaction which requires the use of
copper salts.33,34 To avoid possible complex formation of copper (II) ions with compound 16a,
alternative synthetic pathways were considered. Other methods to form biaryl systems use
aryboronic acid, arylzinc and arylstannane derivatives.34 In our case, we adapted an effective
catalytic alternative of the Ullmann reaction.35,36 This reaction uses Pd(OAc)2 as the catalyst,
tert-butylammonium bromide as base in DMF/H2O. To reduce the reaction time, the synthesis

was conducted in a Q-tubeTM. Compound 10 was obtained in a moderate yield and 96 % purity
after purification by preparative HPLC.
Scheme 1. i) Pd(OAc)2, nBu4NBr, K2CO3, DMF/H2O, Q-tubeTM, 115°C, 4h.
The “twin compounds” (17a-21c) and compound 9 analogs (22-26) were synthesized using the
procedure described in Scheme 2. The key intermediates 15a-c were synthesized in two steps
with 23-54% overall yield. The first step is a Mitsunobu reaction of N-tboc protected
aminoethanol and the appropriate bromo-substituted hydroxypyridine or fluorophenol (14a-c).37
Scheme 2. i) PPh3, DIAD, THF, rt, 5-6 h. ii) HCl/1,4dioxane, rt, 20h. iii) HCHO, HCOOH,
reflux 6h, rt 20h. iv) Pd(OAc)2, R,S-BINAP, NaOtBu, tBuOH 1% in toluene, Q-tube, 160°C, 2h
[N2] or MW, 140°C, 1h20. v) N-deprotection for mono N-tboc protected piperazine,
homopiperazine and bispidine analogs: HCl/1,4dioxane, rt, 20h.

The reaction has been optimized by evaluating different reagents and conditions.38-41 The
highest yields were obtained by using triphenylphosphine. Product formation was not observed
with tri-n-butylphosphine. 1,1’-(Azodicarbonyl)-dipiperidine and diethyl azodicarboxylate gave
low yield whereas di-isopropyl azodicarboxylate (DIAD) produced the highest yields.
Conducting the reaction under nitrogen atmosphere had no influence on the product yields. The
optimal reaction time was 5-6 h, when using triphenylphosphine and DIAD. For 3-bromo-5-
hydroxypyridine 14a and 3-bromo-5-fluorophenol 14b, the highest yields were obtained when
reagents are introduced in the following sequence: DIAD and N-tboc ethanolamine were
dissolved in THF; then, the 3-bromo-5-hydroxypyridine 14a or 3-bromo-5-fluorophenol 14b and
triphenylphosphine were added. This sequence was changed for the reaction with 5-bromo-2,3-
difluorophenol 14c due to a lower yield first obtained. DIAD, triphenylphosphine and 5-bromo-
2,3-difluorophenol 14c were dissolved in THF and stirred at room temperature for 1 h, then N-
tBoc ethanolamine was added. Since there was no benefit in an intermediate purification step,
the obtained crude products were directly used for the N-deprotection step. The N-tboc group
was cleaved under mild conditions using HCl in 1,4 dioxane. This reaction is achieved in good
yield (95 %) and gives pure products (> 90 %) by precipitation. The N-methylation of the
primary amine was carried out in presence of formaldehyde and formic acid (Eschweiler-Clarke
reaction) and gave 16a-c in low to moderate yields (23% - 62%).42 Purification was done by
flash chromatography using CH2Cl2/MeOH as eluent. The Buchwald-Hartwig coupling reaction
was applied to attach diverse (di)aza(bi)cyclic scaffolds to compounds 16a-c (Scheme 2).43
Various ligands, bases, catalysts and solvents are described for this coupling in the literature.44-49
For our reactions, highest yields were achieved using R,S-BINAP and XPhos. PPh3 and 1,1'-
bis(diphenylphosphino) ferrocene were ineffective. Among the bases used (NaOtBu, Cs2CO3
and K2CO3) NaOtBu produced the best results. Furthermore, we evaluated different solvents like
toluene, toluene/tert-butanol (5/1) and DMF/H2O for our reactions. Toluene and toluene/tert-
butanol gave the best results. The mixture toluene/tert-butanol was used for complete
dissolution of reagents when necessary. Finally, a number of catalysts have been tested such as
Pd(OAC)2, Pd2(dba)3, and Pd(PPh3)4. The best product formation was observed using Pd(OAc)2.
The reaction can be achieved either by microwave irradiation or in a pressure reactor (Q-tubeTM).
For microwave assisted reactions palladium acetate, R,S-BINAP or XPhos and sodium tert-

butoxide were used. The reaction was heated 1 h 20 min at 140°C. Using the Q-tubeTM,
temperature and reaction time had to be adapted, so, 2 h at 160°C gave equal results to the
microwave alternative. In addition, one protecting group is necessary to perform the coupling of
diaza(bi)cyclic systems to avoid bis-coupling product formation. The diazabicyclic system
bispidine was prepared using our published method.50,51 Subsequent removal of the N-tboc-
protecting group by HCl in dioxane led to the desired products. The highest purity (95-99%)
was obtained when purified on preparative HPLC using a gradient of MeOH/H2O.
3.2. In vitro characterization and structure-activity relationship (SAR) studies
The binding affinities of our synthesized compounds were determined for α4β2*, α3β4*, α7*,
and (α1)2β1γδ (“muscle type”) nAChRs in radioligand binding assays previously described.51,52
[3H]Epibatidine and [3H]MLA were used as radioligands. Crude membrane fractions were
prepared from Sprague-Dawley rat forebrains, pig adrenals or Torpedo californica electroplax to
serve as receptor sources. The Ki values are summarized in Table 2.
Compound 10: The “twin” compound 10 based on two molecules of N,N-dimethyl-2-(pyridin-
3-yloxy)ethan-1-amine 9 possesses a Ki value of 290 nM for the α4β2* subtype, therefore an
approximately nine-fold decreased affinity in comparison with the single entity 9. The second
molecule of 9 was attached using the no-linker mode at position 5 of the pyridyl moiety where
larger substituents are normally tolerated best.23 There was no interaction with the α3β4* or α7*
nAChR subtypes, whereas a Ki of 905 nM was observed for the muscle type which could be
related to a more favorable distance between both basic nitrogen atoms similar to
suxamethonium 5. An ideal distance of 10-12 carbon atoms between the two basic nitrogen
atoms has been discussed for muscle relaxants in the past.13 Since we aimed for this project to
develop CNS based drugs, we focused on our non-identical twin compounds using the overlap
mode which shortens the distance between the two basic nitrogen atoms to avoid the interaction
with the muscle subtype.
Non-identical “twin” compounds: The SAR for the non-identical “twin” compounds 17a-21c
can be discussed in the context of the type of diaza(bi)cyclic scaffolds attached and the influence
of the HBA system used here. As mentioned earlier the dimethylaminoethoxy group of 9 was
kept constant. The dimethylaminoethoxy group shows a choline like motif and an acyclic
“cation” part. Compounds (17a, 18a, 19a, 20a, 21a) with a pyridyl moiety serving as the HBA

system displayed higher affinity for α4β2* nAChR along with high subtype selectivity (Tab. 2).
Larger diaza(bi)cyclic scaffolds increased the affinity for the α4β2* subtype in the pyridyl series
which has been also found in non-twin drug based nAChR ligands.7 The “twin” compound 21a
shows higher affinity than 13b for α4β2*, whereas the non-methylated analogs (20a versus 13a)
have the opposite trend (Tab. 1 and 2). The homopiperazine based “twin” compounds (18a, 19a)
have slightly reduced affinities for α4β2* nAChR, but improved subtype selectivity comparing
with compounds 12a and 12b which interact with the α7* subtype. The “twin” compound 17a
(BPC) with the highest affinity in this series is bearing the bispidine (3,7-
diazabicyclo[3.3.1]nonane) scaffold. The Ki value obtained is 0.188 nM for α4β2* nAChR with
high subtype selectivity. The affinity for the single entity (11) developed by Abbott Laboratories
is higher (Ki = 0.02 nM) and its minimally effective dose in the mouse hot plate paradigm for
antinociceptive effect is 1.9 µmol/kg.32 There are no further data on subtype selectivity available
for 11. Our “twin” compound 17a (BPC) shows superior subtype selectivity and functionality in
comparison with the smoking cessation drugs cytisine and varenicline, and the efficacy in the
mouse tail suspension test underlines its potential use in mood disorders.22 Previous studies have
shown that larger diaza(bi)cyclic systems providing the protonated nitrogen have higher affinity
for α4β2* nAChR due to lipophilic interactions.7,28 The piperazine derivative 20a displays the
lowest affinity among the three compounds, which is about 770-fold lower in comparison with
compound 17a. Its piperazine scaffold is less bulky than the homopiperazine one in 18a and
especially the diazabicylic system bispidine in 17a.
N-methylated analogs of compounds 18a and 20a: In case of the piperazines, introduction of a
methyl group (21a) results in a higher affinity for α4β2* (Ki = 66.8 nM vs Ki = 145.2 nM for
20a). In contrast, N-methylation of 18a decreases the affinity (Ki = 212 nM vs Ki = 6.9 nM for
19a). Taken together, the bulkiness of diaza(bi)cyclic cation part along with its orientation in
space are important for high affinity for the α4β2* subtype.
HBA system: Among the “twin” drugs, compound 17a (BPC) has the highest affinity for
α4β2* nAChR (Ki = 0.188 nM). To improve lipophilicity and blood brain barrier penetration, the
pyridine ring (HBA system) was replaced by a monofluoro-phenyl (17b) or a difluoro-phenyl
moiety (17c).54,55 The difluoro-phenyl derivative 17c possesses a Ki value of 65 nM, and the
determined Ki value for the monofluoro-phenyl analog 17b is 186 nM. The HBA strength of the
fluorophenyl moieties is lower in comparison with the pyridyl moiety explaining the reduced

affinities obtained. Since the electron density has an impact of the HBA strength, partial charges
of the aromatic system were calculated using the Gasteiger-Marsilli method (Discovery Studio
3.1). Table S2 summarizes the partial charges for each atom within the aromatic system. The
total charge distribution is 0.666 for the pyridine ring. This value is lower for the fluoro- and
difluoro-phenyl moieties resulting in a decrease in HBA strength and subsequently in reduced
affinity (17b, 17c). There is a complete loss of affinity for non-bispidine based compounds in
this series.
Compound 9 analogs: Since bulkier basic nitrogen containing scaffolds fit better into the
nAChR binding pocket, and we assume that the diaza(bi)cyclic moieties in our “twin” drugs are
responsible for the cation-π interaction instead of the dimethylamine group in the parent
compound 9, we studied the influence of some small monoazacyclic systems in position 5 for
compound 9 (Fig. 5B). Few derivatives of 9 have been synthesized by Abbott Laboratories
presenting different substituent size in position 5, but non-aromatic cyclic systems were not
among them.23 Introduction of different e.g. halogen atoms, trifluoromethane or a nitro group
decreased the affinity for α4β2* in comparison with the parent compound 9.23 When we
introduced small monoazacyclic systems, only the azetidine (22, Ki = 784 nM) and the piperidin-
4-ol (26, Ki = 613 nM) moieties were tolerated, but with reduced affinities for α4β2*, and no
activity on other nAChR subtypes tested. A pyrrolidine (23), piperidine (24) or morpholine (25)
substituent resulted in a complete loss of affinity for all nAChR subtypes. Steric
aspects/bulkiness can be considered here, when there is not an additional nitrogen atom present
which has higher basicity or hydrogen bond donor effects. This means that an azetidine
substituent is just tolerated, whereas bigger azacyclic systems are only allowed, when a second
nitrogen atom is present (e.g. 20a) capable for strong cation-π interaction. An HBD system or a
motif which provides also hydrogen bond donor interaction (OH group in 26) seems to recover
affinity for the α4β2* subtype. Other monoazacylic substituents with shifted positions of the
nitrogen atom need to be studied in the future along with implemented HBD systems to explore
this chemical space further.

Table 2. Affinities (Ki values
SEM) and calculated physicochemical parameters (ClogP,
TPSA, logBB) of “twin compounds”, their mono- or difluoro analogs (10, 17a-21c), and
compound 9 analogs (22-26).
α4β2*
Ki [nM]
α3β4*
Ki [nM]
α7*
Ki [nM]
(α1)2β1γδ
Ki [nM]
Cpd #
Structure
Mr
ClogPa TPSAa
logBBa
0.04
“Twin” compounds and their fluoro analogs
10
290 20
> 1,000
> 1,000
> 1,000
> 1,000
905 13
> 1,000
330.42
290.40
1.46
1.16
50.72
40.63
17a
(BPC)
0.188 0.02
0.11
17b
186 11
> 1,000
> 1,000
> 1,000
307.40
2.05
27.74
0.22
17c
18a
65 4.5
6.9 1.2
> 1,000
> 1,000
> 1,000
> 1,000
> 1,000
> 1,000
325.39
264.36
1.99
0.84
27.74
40.63
0.19
0.08
18b
> 1,000
> 1,000
n.d.
n.d.
281.36
1.73
27.74
0.29
18c
19a
19b
> 1,000
212 17
> 1,000
> 1,000
> 1,000
> 1,000
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
299.35
278.39
295.39
1.68
1.38
2.27
27.74
31.84
18.95
0.30
0.19
0.52
19c
20a
> 1,000
> 1,000
> 1,000
n.d.
n.d.
313.38
250.33
2.22
0.38
18.95
40.63
0.55
0.01
145.2 15.6
> 1,000
> 1,000

20b
> 1,000
> 1,000
n.d.
n.d.
267.34
1.27
27.74
0.16
HN
N
O
F
N
20c
21a
21b
> 1,000
66.8 3.5
> 1,000
> 1,000
> 1,000
> 1,000
n.d.
> 1,000
n.d.
n.d.
> 1,000
n.d.
285.33
264.36
281.36
1.22
1.07
1.96
27.74
31.84
18.95
0.17
0.14
0.39
F
21c
> 1,000
n.d.
n.d.
n.d.
299.35
1.91
18.95
0.40
Compound 9 analogs
22
23
784 45
> 1,000
> 1,000
n.d.
n.d.
221.29
235.32
1.12
1.68
28.60
28.60
0.12
0.25
> 1,000
> 1,000
> 1,000
N
O
N
24
25
> 1,000
> 1,000
> 1,000
> 1,000
> 1,000
n.d.
n.d.
n.d.
249.35
251.32
2.25
0.63
28.60
37.83
0.31
N
-0.03
26
613 13
> 1,000
> 1,000
n.d.
265.35
0.72
48.83
0.02
Values are generated from 2-10 independent experiments; n. d. = not determined; aClogP, TPSA,
and logBB values were calculated using ACD/ADME Suite 5.0.
Pharmacophore: A detailed crystal structure of an entire nAChR is still not available, but 3D
QSAR pharmacophore models have been created.27,56-61 Those models display two essential
pharmacophoric elements: a cationic center and a hydrogen bond acceptor (HBA). The distance
between these two elements can vary from 4-7Å. To illustrate and identify pharmacophoric

features (HBA, HBD, positive ionizable, aromatic ring) present in the “twin” compounds and its
derivatives, a 3D QSAR pharmacophore model was developed based on compounds 17a, 18a,
19a, 20a, 21a, 22-26 using Discovery Studio 3.1. The full ligand set was sorted by a random
index fixed at 60. All the ligands are considered to be active on the same binding site. A set of
10 pharmacophores was generated from six training set compounds (17a, 18a, 20a, 23, 24, 26,)
and validated by the test set of four compounds (21a, 19a, 22, 25). Furthermore, a cross-
validation was used based on the Fisher randomization test method with a significance of 90% to
evaluate the statistical relevance of the pharmacophore model. Five features were considered in
all models (HBA, HBD, ring aromatic, hydrophobic and positive ionizable). One pharmacophore
model was eliminated because the significance value was only 80%. The selection for the best
pharmacophore model was made by the best fit and lowest error values on the predicted activity
for the most active compounds (Tab. 3, e.g. 17a and 18a). The pharmacophore model #9 was
considered as the best one, and it displayed a q^2 of 0.733, a RMS error of 0.965 and a mean
absolute error of 0.862. The model obtained displays two positive ionizable features (Red), one
HBA (Green) and an aromatic feature (Orange) (Fig. 6). The two different positive ionizable
features will be given an arbitrary name for a better understanding when discussing the mapping
of the “twin” compounds with pharmacophore #9. This denomination will be based on their
position related to the mapping of compound 17a with pharmacophore #9 (Fig. 6). Positive
ionizable feature #1 (PI-1) is the positive ionizable feature that matches the N-dimethyl group of
the choline-like motif, and positive ionizable feature #2 (PI-2) is the positive ionizable feature
fitting the azacyclic system. The distances between the pharmacophoric features are displayed in
Figure S1.
Figure 6: Mapping of compound 17a onto pharmacophore #9. Positive charges, hydrogen bond
acceptor (HBA) and aromatic ring features are shown in Red, Green and Orange, respectively.

PI-1 is the positive ionizable feature #1 that fits the N-dimethyl group of the choline-like motif,
PI-2 (the positive ionizable feature #2) fits the azacyclic system. The attribution is based on the
fitting of 17a with the best pharmacophore generated.
Compound 17a has the best fit value for the generated model (Tab. 3). The secondary amine
of the bispidine moiety fits one of the positive ionizable features (PI-2, Fig. 6). The second
positive ionizable feature (PI-1, Fig. 6) matches the N-dimethyl group of the choline-like motif.
The centroid of the pyridine ring maps onto the aromatic ring feature (shown in Orange in Fig.
6). The ether oxygen of the dimethylaminoethoxy moiety fits with an HBA feature (shown in
Green in Fig. 6).
The mapping of the “twin” compounds (18a, 19a, 20a, 21a) and compounds 22-26 with
pharmacophore #9 are shown in Figures S2, S3 and S4 (supporting material). Table 3 displays
the fit values of the compounds with pharmacophore #9. Compounds with the highest affinity
(e.g. 17a and 18a) have the highest fit value to pharmacophore #9 (Table 3). However, one
inversion is observed: Compound 20a displays a higher fit value (9.97) to the pharmacophore
than 21a (9.60) whereas 21a has higher affinity than 20a (Tab. 2 and Tab. 3). Figure S2 shows
the mapping of compounds 17a, 18a and 20a. These three compounds map all features of
pharmacophore #9 with the highest fit value (11.55, 10.20 and 9.97, respectively) and display
high affinity (0.19, 6.9 and 145.2 nM, respectively). The N-dimethyl nitrogen atom fits the
positive ionizable feature (PI-1) for compounds 19a, 21a, 22 and 26 (Fig. S3). The centroid of
the pyridine ring does not perfectly fit the aromatic ring feature of compounds 22 and 26, which
could explain the decrease in affinity when comparing with 19a and 21a. A greater difference is
observed for the second positive ionizable feature (PI-2, Fig. S3). The piperidin-4-ol part of
compound 26 only partially fits the positive ionizable feature (PI-2). There is no match of the
azetidine ring with the ionizable feature (PI-2) for compound 22, which has greatly reduced
affinity.
Mapping of the non-active compounds 23, 24, and 25 is shown in Figure S4. These three
compounds fit only one positive ionizable feature (PI-1), the HBA feature and the aromatic ring
feature. A second positive charge (PI-2) seems to be necessary to keep the affinity related to this
model. Compound 9 analogs (22-26) which are not fulfilling the criteria of a “twin” compound
are non-active (23-25) or have greatly reduced affinity (22, 26).

The estimated affinities, error values and fit values are listed in Table 3. A “ligand
pharmacophore mapping” protocol was used to obtain predicted affinity values for the test set
(see Experimental Part). The error value was defined as the ratio of predicted affinity to
measured affinity when the error value is > 1 (or as the ratio of measured affinity to predicted
affinity, when the error value is negative). Error values between actual and predicted affinity
were < 5, only one compound (17a) had an error value larger than 5, but lower than 10, which is
still within an acceptable error range. The most active compound (17a) has a predicted affinity of
1.3 nM whereas its measured value is 0.188 nM. The error values demonstrate a model with
good prediction. All compounds are classified by their affinities as highly active (Ki < 150 nM),
moderately active (150 < Ki < 1,000 nM) and inactive (Ki > 1,000 nM). Most of the active
compounds were correctly predicted. Only one moderately active compound (19a) was predicted
to be highly active and one moderately active compound (22) was predicted to be inactive.
Herein, we have generated a good theoretical model for our “twin” compounds that correlates
the experimental data measured: a) The predicted affinity values obtained from pharmacophore
#9 are in correlation with the Ki values obtained (e.g. 17a, 18a, 19a, 20a, 21a); b) The mapping
of the “twin” compounds with the pharmacophore #9 can explain the differences observed for
the activities (e.g. 22-26).
Table 3. Experimental and estimated Ki values of training set compounds based on the best
model generated.
“Error” Valueb
Fit Valuec
Cpds
Measured Ki for
α4β2*
Predicted Ki for
α4β2a
Measured
activity scaled
Predicted
activity scaled
17a
18a
0.19
6.9
1.3
6.73
4.01
-1.43
-3.08
1.65
3.65
-3.63
-3.63
-3.63
1.54
11.55
10.20
9.48
9.97
9.60
8.15
8.21
8.21
8.21
8.66
+++
+++
++
+++
+++
++
+
+++
+++
+++
+++
+++
+
27.7
19a
20a
21a
22
212
145.2
147.9
47.2
66.8
110.3
3,159
2,748
2,748
2,748
972.8
864.9
> 5,000#
> 5,000#
> 1,000#
632.2
23
+
24
+
+
25
+
+
26
++
++
#A Ki value of 10,000 was used for generating the pharmacophore model. aThe value was
predicted by the best pharmacophore generated. bThe error value is a ratio between the predicted
c
and measured value or the inverse. The fit value is a measure of how well a ligand fits the

pharmacophore; the higher the value, the better the fitting. d Activity scale: highly active +++, Ki
< 150 nM; moderately active ++, 150 nM < Ki < 1,000 nM; inactive +, Ki > 1,000 nM.
3.3. Physicochemical Properties and ADME Calculation
The design of the new ligands was guided by in silico ADME calculations.62,63 ADME related
properties like BBB penetration were calculated using Discovery Studio 3.1. The blood-brain
penetration model was derived from over 800 compounds classified as CNS therapeutics. The
plot (Fig. 7) displays the drug-like properties with logP and polar surface area (PSA) values
(absorption and BBB penetration). Compounds 17a-26 have promising properties for absorption
and blood brain barrier penetration, since their values appear within the “Absorption” and
“BBB” ellipses (Fig. 7).
17c
18c
24
19c
19b
18b
2 1c
21b
20c
20b
17b
19a
21a
23
22
17a
18a
26
25
20a
Figure 7: Plot of PSA (Polar Surface Area) vs. LogP for “twin” compounds and compound 9
analogs showing confidence limit ellipses (95 % and 99 %) corresponding to the BBB (Blood
Brain Barrier) and Intestinal Absorption models. The plot was generated using Discovery Studio
3.1 (Accelrys, San Diego, CA).
In addition, other physicochemical properties can be considered to be important for oral
absorption and BBB.64-66 E.g. the CNS multiparameter optimization (MPO) is based on six
fundamental physicochemical properties (a) lipophilicity (ClogP); (b) distribution coefficient
(ClogD); (c) molecular weight (MW); (d) topological polar surface area (TPSA); (e) number of
hydrogen bond donors (NHD); (f) most basic center (pKa). Based on the drug set, median values

were found to be ClogP = 2.8, ClogD = 1.7, MW = 305.3 Da, TPSA = 44.8 Å2, NHD = 1, and
pKa = 8.4. Ghose et al. compared physicochemical property of CNS and non-CNS drugs.25 This
study provides the following guideline for CNS drug design, TPSA < 76 Å2 (25-60 Å2), more
than one nitrogen (one or two, including one aliphatic amine), fewer than seven (two to four)
linear chains outside of rings, fewer than three (zero or one) polar hydrogen atoms, volume of
740-970 Å3, solvent accessible surface area of 460-580 Å2.
We calculated important parameters (pKa, MolRefrac, NHA, NHD, Rings, Aromatic rings,
logBB) for all synthesized compounds using ACD/ADME Suite 5.0 (ACD/Labs) software. Data
are listed in Table S1 (supporting information). Values in Green are within the range of 50 % of
marketed CNS drugs. Values in Black are in “acceptable” ranges, but would need further
profiling. Most of our compounds are in the range for CNS drug-likeness parameter values. In
general, the twin drug compounds and analogs of compound 9 (10, 17a-26) present a better CNS
drug-likeness profile than the single entities (9, 11-13b). Most active compounds showed
“good” or “acceptable” values (e.g. 17a, 18a and 21a). No correlations were identified between
affinity and physicochemical properties. However a trend can be highlighted: one hydrogen bond
donor (NHD) and five hydrogen bond acceptors (NHA) seems a favorable combination to
maintain good affinity (e.g. 17a and 18a). The lack of one of them (NHD = 0 and/or NHA = 4)
results in a decrease or loss of activity. A few exceptions can be pointed out: compound 26
possesses NHD = 1 and NHA = 5, but only displays a moderate activity (Ki = 613 nM). Also,
compound 10 shows a moderate activity (Ki = 290 nM) with NHD = 0 and NHA = 6.
The concept of binding efficiency is used for any drug development process. It correlates
potency, size and lipophilicity. Ligand Efficiency (LE) represents the efficiency of binding with
respect to size as measured by the number of heavy atoms. It has a minimum of 0.3 for hit
compounds. Lipophilic Ligand Efficiency (LLE) focuses on maximizing lipophilicity efficiency.
LLE mean values for lead compounds and marketed drugs are respectively 3.8 and 5-6.2.67-69
In order to identify lead compounds, ligand efficiency index (LE) and lipophilic ligand
efficiency (LLE) were calculated for our compound set and the single entities (Tab. S3). LE and
LLE values were obtained by using the following equations: LE = (1.37 x pKi)/HA where HA is
the number of heavy atoms, and LLE = pKi-clogP.70,71 The LE minimum value for hit
compounds is 0.3.68 The LLE mean value for lead compounds is 3.8, and it is in the range of 5-
6.2 for marketed drugs.68 LE and LLE values for all compounds (9, 10, 17a-c, 18a, 19a, 20a,

21a, 22, 26) are above the mean values. Single entity compounds 9, 11-13b show good LE (0.67-
0.98) and LLE (5.63-9.73) properties to be considered as lead compounds. Compounds 11 and
12a have the highest LE and LLE values. Among the “twin” compounds (10, 17a, 18a, 19a,
20a, 21a), two compounds have shown high LE and LLE, compounds 17a (LE = 0.63; LLE =
8.6) and 18a (LE = 0.59; LLE = 9.7). 17a is the combination of 9 and 11. 18a is the
combination of 9 and 12a. Among compounds 9, 11-13b, compounds 9, 11 and 12a are the most
active compounds with the best LE and LLE confirming their role of lead compounds. Taken
together, the “twin” compounds 17a and 18a can be considered as lead compounds as well.
4. Conclusion
In summary, the “twin drug” approach proves to be a successful strategy for the development
of α4β2* nAChR ligands with improved subtype selectivity and desired functionality profile
recently described especially for our lead compound BPC (17a). Using the “overlap mode” in the
design strategy provides compounds with a “good” or “acceptable” drug-likeness profile. The
compounds were evaluated for their affinities at four different nAChRs. Active compounds
displayed α4β2* subtype selectivity. The bispidine scaffold containing compound 17a (BPC)
showed the highest affinity among the “twin” compounds and even kept high affinities for
α4β2* nAChRs when the strength of its HBA system was reduced (17b, 17c) in contrast to the
piperazine and homopiperazine analogs. Not following the “twin” design and introducing small
monoazacyclic substituents at position 5 of compound 9 caused a dramatically drop or complete
loss in affinity for nAChRs.
5. Experimental Section
All reagents and solvents were obtained from various suppliers (Acros, Aldrich, Alfa Aesar,
Fisher Scientific, Frontier Scientific Merck, Oakwood or Sigma) and used without further
purification unless otherwise noted. Tetrahydrofuran (THF), 1,4-dioxane and toluene were dried
over sodium. Water was taken from a water purification system Direct-QTM 5 (Millipore).
Reactions were monitored by thin layer chromatography (TLC) using aluminium sheets coated
with silica gel 60 F254 (Merck). Compounds were visualized using UV light (254 nm) and/or
using a KMnO4 solution (1%). Column chromatography was carried out on silica gel 60 (0.063-
0.200 mm) using gradient mixtures of CH2Cl2 with MeOH (20:1, 9:1 or 8:2) as mobile phases.

1H NMR spectra (400 MHz) and 13C NMR spectra (100 MHz) were recorded on an Avance 400
spectrometer (Bruker). All NMR spectra were recorded at rt. Chemical shifts (δ) are given in
parts per million (ppm) relative to the remaining protons of the deuterated solvent used as
internal standard. Coupling constants J are given in Hertz (Hz) and spin multiplicities are given
as s (singlet), d (doublet), dd (doublet of doublets), dt (doublet of triplets), t (triplet), q (quartet),
m (multiplet) and br (broad). Mass spectra were recorded on a Varian 500-MS mass
spectrometer with an electron spray ionization source (ESI-MS). HRMS runs were performed on
Agilent Technologies 6530 Accurate Mass Q-TOF/LCMS. The purity of the compounds was
determined by a Shimadzu Prominence HPLC system at an appropriate wavelength. All
compounds proved to possess ≥ 95% purity.
5.1. General procedure A: Mitsunobu reaction
In a round bottom flask, DIAD (1-1.2 eq) and N-tboc-ethanolamine (1.3 eq) were dissolved in
THF (15 mL). Then the 3-bromo-5-hydroxypyridine (1 eq) and the triphenylphosphine (1.2-1.5
eq) were added. The reaction mixture was stirred at room temperature for 5 h. The solvent was
removed under reduced pressure and the residue was used without any further purification for
deprotection.
5.2. General procedure B: Mitsunobu reaction
In a round bottom flask, DIAD (1 eq), triphenylphosphine (1.1 eq) and the appropriate
nucleophile (1 eq) were dissolved in THF (15 mL). The reaction mixture was stirred at room
temperature for 1 h. Then N-tboc-ethanolamine was added (1.2 eq) The reaction mixture was
stirred at room temperature for an additional 5 h. The solvent was removed under reduced
pressure and the residue was used without any further purification for deprotection.
5.3. General procedure C: Cleavage of the N-tboc protecting group from N-tboc protected
ethanolamine derivatives
HCl in 1,4-dioxane (4 N, 5-6 mL) was added to a stirred solution of the N-tboc protected
compound (2.5-4 mmol), dissolved in 5-6 mL of 1,4-dioxane, and the mixture was allowed to stir
at rt for 2-15 h. The white solid was filtered and washed with cyclohexane and diethylether.

5.4. 2-[(5-bromopyridin-3-yl)oxy]ethanamine (15a)
The Mitsunobu reaction was performed according to general procedure A with 3-bromo-5-
hydroxypyridine 14a (500 mg, 2.87 mmol), N-tboc-ethanolamine (620 mg, 3.85 mmol), DIAD
(576 mg, 2.85 mmol), triphenylphosphine (1.13 g, 4.31 mmol) and THF. The reaction time was 5
h. Then the general procedure C was used for the N-tboc deprotection. Final product 15a was
obtained as a white solid (336 mg, 54%). 1H NMR (MeOD, d4, 400 MHz) δ 8.75 (s, 1H), 8.70 (d,
J = 2.4 Hz, 1H), 8.49 (t, J = 1.6 Hz, 1H), 4.53 (t, J = 4.4 Hz, 2H), 3.47 (t, J = 4.4 Hz, 2H) ; 13C
NMR (MeOD, d4, 100 MHz) δ 158.11, 138.61, 134.41, 132.04, 123.65, 67.84, 39.82.
5.5. 2-(3-bromo-5-fluorophenoxy)ethanamine (15b)
The Mitsunobu reaction was performed according to general procedure A with 3-bromo-5-
fluorophenol 14b (565 mg, 2.96 mmol), N-tboc-ethanolamine (607 mg, 3.77 mmol), DIAD (698
mg, 3.45 mmol), triphenylphosphine (920 mg, 3.51 mmol) and THF. The reaction time was 5 h.
Then the general procedure C was used for the N-tboc deprotection. Final product 15b was
obtained as a white solid (162 mg, 23%). 1H NMR (MeOD, d4, 400 MHz) δ 7.00 (s, br, 1H), 6.98
(dt, J = 1.2-3.6 Hz, 1H), 6.83 (dt, J = 2-10.4 Hz, 1H), 4.25 (t, J = 4 Hz, 2H), 3.37 (t, J = 4.8 Hz,
2H) ; 13C NMR (MeOD, d4, 100 MHz) δ 161.26, 132.07, 123.84, 115.43, 113.07, 102.65, 66.08,
40.05.
5.6. 2-(5-bromo-2,3-difluorophenoxy)ethanamine (15c)
The Mitsunobu reaction was performed according to general procedure B with 5-bromo-2,3-
difluorofluorophenol 14c (814 mg, 3.90 mmol), N-tboc-ethanolamine (635 mg, 3.94 mmol),
DIAD (714 mg, 3.53 mmol), triphenylphosphine (1.32 g, 5.03 mmol) and THF. The total
reaction time was 6 h. Then the general procedure C was used for the N-tboc deprotection. Final
1
product 15c was obtained as a white solid (403 mg, 36%). H NMR (MeOD, d4, 400 MHz) δ
7.22 (d, J = 3.2 Hz, 1H), 7.19 (dt, J = 2.4-3.2 Hz, 1H), 4.35 (t, J = 4.8 Hz, 2H), 3.40 (t, J = 4.8
Hz, 2H); 13C NMR (MeOD, d4, 100 MHz) δ 154.26, 149.24, 143.50, 115.53, 114.83, 67.76,
40.07.
5.7. General procedure D: Eschweiler-Clarke methylation

The appropriate ethanamine (15a, 15b or 15c) (1eq) was taken into aqueous formaldehyde
solution (3-6 mL) and formic acid (5-6 mL) was added. The resulting mixture was refluxed for 6
h. After cooling down, the solution was basified to pH = 9-10 with NaOH 10 N. The aqueous
solution was extracted with 4 x 20 mL of CH2Cl2. The combined organic layers were dried over
MgSO4, filtered and the solvent removed under reduced pressure to give a yellowish/brownish
oil. The crude material was purified using silicagel chromatography with a gradient of
CH2Cl2/MeOH 10% to give the desired product.
5.8. 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine (16a)
The methylation reaction was achieved according to general procedure D with 2-[(5-
bromopyridin-3-yl)oxy]ethanamine 15a (750 mg, 3.46 mmol), aqueous formaldehyde (4 mL)
solution and formic acid (4 mL) The resulting mixture was refluxed for 6h The crude compound
was purified on flash chromatography using a gradient of CH2Cl2/MeOH. Final product 16a was
1
obtained as a yellow oil (522 mg, 62%). H NMR (CDCl3, 400 MHz): 8.25 (d, J=1.2 Hz, 1H),
8.23 (d, J=2.4 Hz, 1H), 7.36 (t, J= 2.4 Hz, 1H), 4.07 (t, J=5.6 Hz, 2H), 2.71 (t, J=5.2 Hz, 2H),
2.31 (s, 6H) ; 13C NMR (CDCl3, 100 MHz) : 155.3, 142.9, 136.4, 123.9, 120.3, 66.7, 57.9, 45.8.
5.9. 2-(3-bromo-5-fluorophenoxy)-N,N-dimethylethanamine (16b)
The methylation reaction was achieved according to general procedure D with 2-(3-bromo-5-
fluorophenoxy)ethanamine 15b (656 mg, 2.80 mmol), aqueous formaldehyde (3 mL) solution
and formic acid (3 mL) The resulting mixture was refluxed for 6h The crude compound was
purified on flash chromatography using a gradient of CH2Cl2/MeOH. Final product 16b was
1
obtained as a white oil (197 mg, 27%). H NMR (CDCl3, 400 MHz): 6.97 (s, br, 1H), 6.90 (dt,
J=2.2-8.2 Hz, 1H,), 6.74 (dt, J= 2.2-10.8 Hz, 1H), 4.09 (t, J=5.6 Hz, 2H), 2.76 (t, J=5.2 Hz, 2H),
2.33 (s, 6H) ; 13C NMR (CDCl3, 100 MHz) : 166.09, 162.21, 123.8, 115.29, 112.33, 102.49,
67.35, 58.78, 47.77.
5.10. 2-(5-bromo-2,3-difluorophenoxy)-N,N-dimethylethanamine (16c)
The methylation reaction was achieved according to general procedure D with 2-(5-bromo-
2,3-difluorophenoxy)ethanamine 15c (430 mg, 1.71 mmol), aqueous formaldehyde (3 mL)
solution and formic acid (3 mL) The resulting mixture was refluxed for 6h The crude compound

was purified on flash chromatography using a gradient of CH2Cl2/MeOH. Final product 16c was
1
obtained as a white oil (111 mg, 23%). H NMR (CDCl3, 400 MHz): 7.13 ( d, J=10.9 Hz), 7.08
13
(dt, J=2.2-8.5 Hz, 1H), 4.19 (t, J=5.3 Hz, 2H), 2.82 (t, J=5.4 Hz, 2H), 2.36 (s, 6H) ; C NMR
(CDCl3, 100 MHz): 151.27, 150.34, 143.19, 116.53, 114.97, 113.81, 69.11, 58.70, 45.92.
5.11. General Procedure E: Buchwald-Hartwig cross coupling reaction / microwave procedure
In a microwave vessel, palladium acetate (5-10 mol%), (R,S)- BINAP or XPhos (10-30
mol%), sodium tert-butoxide (1.5-3 eq), the appropriate amine (1-1.3 eq), 2-[(5-bromopyridin-3-
yl)oxy]-N,N-dimethylethanamine 16a (1 eq) were added and dissolved in a mixture of toluene
with tert-butanol (5:1) (1 mL). The reaction vessel was sealed with a septum and placed into the
microwave cavity. Microwave irradiation of 150 W was used and the temperature ramped from
rt to 140°C. Once 140°C was reached the reaction mixture was held for 80 min. After cooling
down, the reaction vessel was opened and the reaction mixture was filtered on a C-18 SPE
column and eluted with methanol. The solvent was removed under reduced pressure. The crude
compound was purified by preparative HPLC system using RP C-18 silicagel and appropriate
MeOH/H2O gradients. The chromatograms were scanned at 254 and 220 nm. The appropriate
fractions were collected. The fractions containing the desired product were concentrated under
reduced pressure.
5.12. General Procedure F: Buchwald-Hartwig cross coupling / Q-tube procedure
A Q-tube was loaded with palladium acetate (5-10 mol%), (R,S)-BINAP or XPhos (10-30
mol%), sodium tert-butoxide (1.5-3 eq), the appropriate amine (1.2-1.3 eq), the appropriate
bromo-containing compounds 16a, 16b, or 16c (1 eq) and then dissolved in a mixture of toluene
with tert-butanol (5:1) (2 mL). The Q-tube was flushed with nitrogen and the reaction mixture
was heated at 160°C for 2 h. The pressure was released and the reaction vessel was opened. The
reaction mixture was filtered on a C-18 SPE column and eluted with cyclohexane. The solvent
was removed under reduced pressure. The crude compound was purified by preparative HPLC
system using RP C-18 silicagel and appropriate MeOH/H2O gradients. The chromatograms were
scanned at 254 and 220 nm. The appropriate fractions were collected. The fractions containing
the desired product were concentrated under reduced pressure.

5.13. General procedure G: Cleavage of the N-tboc protecting group
HCl in 1,4-dioxane (4 N, 1-2 mL) was added to a stirred solution of the N-tboc protected
compound (0.15-0.55 mmol), dissolved in 1-2 mL of 1,4-dioxane, and the mixture was allowed
to stir at rt for 15 h. The solvent was removed. The residue was taken into KOH 0.2 N (5 mL)
and extracted with 5x10 mL of CH2Cl2. The organic layers were collected and dried over
MgSO4. The solution was filtered and the solvent removed to give the desired compound.
5.14. 2,2'-([3,3'-bipyridine]-5,5'-diylbis(oxy))bis(N,N-dimethylethanamine) (10)
A Q-tube was loaded with 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine 16a (93
mg, 038 mmol), Pd(OAc)2 (6 mg, 0.027 mmol), tetrabutylammonium bromide (71 mg, 0.22
mmol), K2CO3 (98 mg, 0.71 mmol) dissolved in 2 mL DMF/H2O (1/1). The reaction mixture
was heated at 115°C for 4 h. After cooling down, water (20 mL) and Et2O (30 mL) were added
to the reaction mixture. The organic layer was extracted and washed with 3x15mL of water, then
it was dried over MgSO4, filtererd and the solvent removed. The product was purified by
preparative HPLC method 60-100-60/10mL x min-1/30min (H2O/MeOH) yield (33 mg, 20%).
HRMS (ESI) m/z calcd for C18H26N4O2 (M+H)+: 331.2129, found: 331.2126.; 1H NMR (CDCl3,
400 MHz) : δ 8.32 (d, J = 2 Hz, 2H), 8.20 (t, J = 2.4 Hz, 2H), 7.21 (d, J = 1.2 Hz, 2H), 4.10 (t, J
13
= 5.6 Hz, 4H), 2.74 (t, J = 5.2 Hz, 4H), 2.33 (s, 6H); C NMR (CDCl3, 100 MHz) : δ 155.3,
140.65, 137.40, 134.04, 120.11, 66.63, 58.25, 45.99.
5.15. 2-((5-(3,7-diazabicyclo[3.3.1]nonan-3-yl)pyridin-3-yl)oxy)-N,N-dimethylethanamine (17a)
Compound 17a was prepared according to the general procedure E for Buchwald-Hartwig
cross coupling with 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine 16a (91 mg, 0.37
mmol), Ntboc-bispidine (145 mg, 0.64mmol), palladium acetate (6 mg, 0.027 mmol), BINAP
(16 mg, 0.026 mmol), sodium tert-butoxide (58 mg, 0.60 mmol). The crude product was purified
by preparative HPLC: 60-100-60/10 mL x min-1/30 min (H2O/MeOH). Then the N-tBoc
protective group was removed using the general procedure G with 1,4-dioxane (1.5 mL) and HCl
in 1,4-dioxane (1.5 mL) for 0.15 mmol of material to obtain (18 mg, 42%). HRMS (ESI) m/z
calcd for C16H26N4O (M+H)+: 291.2177, found: 291.2176.; 1H NMR (CDCl3, 400 MHz) : δ 7.95
(d, J = 1.6 Hz, 1H), 7.78 (d, J = 2.4 Hz, 1H), 6.74 (t, J = 2.4 Hz, 1H), 4.09 (t, J = 6 Hz, 2H), 3.73
(d, J = 11.6 Hz, 2H), 3.14-3.00 (m, 6H), 2.71 (t, J = 6 Hz, 2H), 2.66 (s, broad, 3H), 2.31 (s, 6H),

13
1.96-1.81 (m, 4H) ; C NMR (CDCl3, 100 MHz) : δ 155.44, 147.74, 130.59, 126.97, 108.27,
66.28, 58.26, 53.53, 52.17, 45.83, 31.01, 28.93.
5.16. 2-(3-(3,7-diazabicyclo[3.3.1]nonan-3-yl)-5-fluorophenoxy)-N,N-dimethylethanamine (17b)
Compound 17b was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(3-bromo-5-fluorophenoxy)-N,N-dimethylethanamine 16b (136 mg, 0.52
mmol), N-tboc-bispidine (143 mg, 0.63mmol), palladium acetate (5 mg, 0.022 mmol), BINAP
(25 mg, 0.040 mmol), sodium tert-butoxide (99 mg, 1.03 mmol). The product was purified by
preparative HPLC method 60-100-60/10mL x min-1/30min (H2O/MeOH). Then the N-tboc
protective group was removed using the general procedure G with 1,4-dioxane (1 mL) and HCl
in 1,4-dioxane (1 mL) for 0.055 mmol of material to yield (10 mg, 6%). HRMS (ESI) m/z calcd
for C17H26FN3O (M+H)+: 308.2133, found: 308.2134.; 1H NMR (CDCl3, 400 MHz) : δ 6.27 (d, J
= 2 Hz, 21H), 6.12 (d, J = 10.4 Hz, 1H), 4.03 (t, J = 5.6 Hz, 2 H), 3.71 (d, J = 8 Hz, 3H), 3.15-
3.10 (m, 6H), 2.71 (t, J = 5.6 Hz, 2H), 2.33 (s, 6H), 1.96-1.81 (m, 4H) ; 13C NMR (CDCl3, 100
MHz) : δ 165.77, 163.38, 160.78, 92.97, 95.14, 92.34, 66.22, 58.35, 53.93, 52.32, 46.01, 31.24,
29.15.
5.17. 2-(5-(3,7-diazabicyclo[3.3.1]nonan-3-yl)-2,3-difluorophenoxy)-N,N-dimethylethanamine
(17c)
Compound 17c was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(5-bromo-2,3-difluorophenoxy)-N,N-dimethylethanamine 16c (186 mg,
0.66 mmol), N-tboc-bispidine (145 mg, 0.64 mmol), palladium acetate (9 mg, 0.040 mmol),
BINAP (47 mg, 0.075 mmol), sodium tert-butoxide (138 mg, 1.43 mmol). The product was
purified by preparative HPLC method 60-100-60/10mL x min-1/30min (H2O/MeOH). Then the
N-tboc protective group was removed using the general procedure G with 1,4-dioxane (1 mL)
and HCl in 1,4-dioxane (1 mL) for 0.015 mmol of material to yield 14 mg, (6%). HRMS (ESI)
m/z calcd for C17H25F2N3O (M+H)+: 326.2038, found : 326.2047.; 1H NMR (CDCl3, 400 MHz) :
δ 6.50 (s, 1H), 6.35 (d, J = 11.2 Hz, 1H), 4.17 (d, J = 12.8 Hz, 2H), 3.62 (d, J = 11.6 Hz, 2H),
3.32 (d, J = 12.8 Hz, 2H), 3.12 (d, J = 13.2 Hz, 2H), 3.01 (d, J = 11.6 Hz, 2H), 2.76 (t, J = 5.6
Hz, 2H), 2.35 (s, 6H) 1.98-1.81 (m, 4H); 13C NMR (CDCl3, 100 MHz) : δ 152.74, 150.31,
148.55, 147.28, 100.15, 97.46, 68.51, 58.25, 55.26, 51.08, 46.10, 30.84, 28.58.

5.18. 2-((5-(1,4-diazepan-1-yl)pyridin-3-yl)oxy)-N,N-dimethylethanamine (18a)
Compound 18a was prepared according to general procedure F for Buchwald-Hartwig cross
coupling with 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine 16a (134 mg, 0.55
mmol), homopiperazine (200 mg, 1.00 mmol), palladium acetate (11 mg, 0.049 mmol), BINAP
(52 mg, 0.083 mmol), sodium tert-butoxide (180 mg, 1.87 mmol). The product was purified by
preparative HPLC: 60-100-60/10mL x min-1/30min (H2O/MeOH). Then the N-tboc protective
group was removed using the general procedure G with 1,4-dioxane (1 mL) and HCl in 1,4-
dioxane (1 mL) for 0.081 mmol of crude material to yield (9 mg, 6%). HRMS (ESI) m/z calcd
for C14H24N4O (M+H)+ 265.2023, found: 265.2028; 1H NMR (CDCl3, 400 MHz) : δ 7.78 (d, J =
2.4 Hz, 1H), 7.65 (d, J = 2Hz, 1H), 6.52 (t, J = 2 Hz, 1H), 4.09 (t, J = 5.6 Hz, 2H), 3.57-3.51 (m,
4H,),3.47 (s, 1H), 3.02 (t, J = 5.2 Hz, 2H), 2.83 (t, J = 5.6 Hz, 2H), 2.72 (t, J = 5.6 Hz, 2H), 2.33
(s, 6H), 1.91 (q, J = 6 Hz, 2H) ; 13C NMR (CDCl3, 100 MHz) : δ155.99, 145.55, 128.06, 124.17,
105.04, 66.28, 58.49, 51.86, 48.22, 48.16, 47.97, 46.00, 29.35.
5.19. 2-(3-(1,4-diazepan-1-yl)-5-fluorophenoxy)-N,N-dimethylethan-1-amine (18b)
Compound 18b was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(3-bromo-5-fluorophenoxy)-N,N-dimethylethanamine 16b (113 mg, 0.43
mmol), N-tboc-homopiperazine (144 mg, 0.72 mmol), palladium acetate (9 mg, 0.038 mmol),
BINAP (60 mg, 0.096 mmol), sodium tert-butoxide (104 mg, 1.08 mmol). The product was
purified by preparative HPLC method 80-100-80/10mL x min-1/30min (H2O/MeOH). Then the
N-tboc protective group was removed using the general procedure G with 1,4-dioxane (1 mL)
and HCl in 1,4-dioxane (1 mL) for 0.080 mmol of material to yield 18 mg (10%). HRMS (ESI)
m/z calcd for C15H24FN3O (M+H)+: 282.1976, found: 282.1995.; 1H NMR (MeOD d4, 400 MHz)
: δ 6.09, (t, J = 8-4 Hz, 2H), 5.99 (dt, J = 8-4 Hz, 1H), 4.05 (t, J = 4 Hz, 2H), 3.54 (q, J = 4 Hz,
4H), 2.99 (t, J = 4 Hz, 2H), 2.81 (t, J = 4 Hz, 2H), 2.74 (t, J = 4 Hz, 2H), 2.33 (s, 6H), 1.94 (J = 4
Hz, 2H); 13C NMR (MeOD d4, 100 MHz) : δ 167.61, 165.24, 162.54, 151.84, 95.30, 92.24,
90.53, 66.63, 59.10, 51.95, 45.82, 29.71.
5.20. 2-(5-(1,4-diazepan-1-yl)-2,3-difluorophenoxy)-N,N-dimethylethan-1-amine (18c)

Compound 18c was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(5-bromo-2,3-fluorophenoxy)-N,N-dimethylethanamine 16c (100 mg, 0.36
mmol), N-tboc-homopiperazine (184 mg, 0.92mmol), palladium acetate (8 mg, 0.036 mmol),
BINAP (33 mg, 0.052 mmol), sodium tert-butoxide (43 mg, 0.44 mmol). The product was
purified by preparative HPLC method 80-100-80/10mL x min-1/30min (H2O/MeOH). Then the
N-tboc protective group was removed using the general procedure G with 1,4-dioxane (0.5 mL)
and HCl in 1,4-dioxane (0.4 mL) for 0.032 mmol of material to yield to yield 8 mg (8%). HRMS
(ESI) m/z calcd for C15H23F2N3O (M+H)+: 300.1882, found: 300.1888.; 1H NMR (MeOD d4, 400
MHz) : δ 6.20-6.13 (m, 2H), 4.16 (t, J = 4 Hz, 2H), 3.67 (dq, J = 4 Hz, 1H), 3.58 (q, J = 4 Hz,
1H), 3.53 (q J = 4 Hz, 4H), 2.95 (t, J = 4 Hz, 1H), 2.78 (q, J = 4 Hz, 3H), 2.35 (s, 6H), 1.96-1.89
(m, 2H); 13C NMR (MeOD d4, 100 MHz) : δ 162.30, 161.41, 151.76, 149.89, 95.22, 93.40,
93.17, 68.86, 59.04, 52.56, 52.24, 45.96, 30.13.
5.21. N,N-dimethyl-2-((5-(4-methyl-1,4-diazepan-1-yl)pyridin-3-yl)oxy)ethanamine (19a)
Compound 19a was prepared according to general procedure F for Buchwald-Hartwig cross
coupling with 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine 16a (95 mg, 0.38 mmol),
N-methyl homopiperazine (73 mg, 0.64 mmol), palladium acetate (9 mg, 0.040 mmol), BINAP
(25 mg, 0.040 mmol), sodium tert-butoxide (64 mg, 0.66 mmol). The product was purified by
preparative HPLC: 60-100-60/10mL x min-1/30min (H2O/MeOH) yield (55 mg, 51%). HRMS
1
(ESI) m/z calcd for C15H26N4O (M+H)+: 279.2179, found: 279.2181; H NMR (CDCl3, 400
MHz) : δ 7.76 (d, J = 2.4 Hz, 1H), 7.64 (d, J = 2 Hz, 1H), 6.50 (t, J = 2.4 Hz, 1H), 4.08 (t, J = 5.6
Hz, 2H), 3.54 (t, J = 4.4 Hz, 2H,),3.46 (q, J = 3.6 Hz, 2H), 2.72-2.67 (m, 4H), 2.54 (t, J = 5.2 Hz,
2H), 2.36 (s, 3H), 2.32 (s, 6H), 2.03 (q, J = 6.4 Hz, 2H) ; 13C NMR (CDCl3, 100 MHz) : δ
155.82, 146.15, 127.98, 124.06, 104.96, 66.20, 58.43, 57.83, 57.17, 48.47, 48.15, 46.75, 45.95,
27.57.
5.22. 2-(3-fluoro-5-(4-methyl-1,4-diazepan-1-yl)phenoxy)-N,N-dimethylethan-1-amine (19b)
Compound 19b was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(3-bromo-5-fluorophenoxy)-N,N-dimethylethanamine 16b (144 mg, 0.55
mmol), N-methylhomopiperazine (88 mg, 0.77mmol), palladium acetate (11 mg, 0.051 mmol),
BINAP (37 mg, 0.059 mmol), sodium tert-butoxide (140 mg, 1.45 mmol). The product was

purified by preparative HPLC method 80-100-80/10mL x min-1/35min (H2O/MeOH) yield (22.4
1
mg, 14%). HRMS (ESI) m/z calcd for C16H26FN3O (M+H)+: 296.2133, found: 296.2129; H
NMR (CDCl3, 400 MHz) : δ 5.99 (d, J = 14 Hz, 2H), 5.96 (d, J = 8.8 Hz, 1H), 3.99 (t, J = 5.6 Hz,
2H), 3.49 (t, J = 4.8 Hz, 2H), 3.41 (t, J = 6.4 Hz, 2H), 2.70-2.64 (m, 4H), 2.53 (t, J = 5.2 Hz, 2H),
13
2.34 (s, 3H), 1.99 (m, 2H) ; C NMR (CDCl3, 100 MHz) : δ 163.73, 161.05, 151.16, 94.48,
92.08, 89.48, 65.98, 58.32, 57.96, 57.09, 48.68, 48.40, 46.66, 45.92, 27.63.
5.23. 2-(2,3-difluoro-5-(4-methyl-1,4-diazepan-1-yl)phenoxy)-N,N-dimethylethan-1-amine (19c)
Compound 19c was prepared according to the general procedure F for Buchwald-Hartwig
cross coupling with 2-(5-bromo-2,3-difluorophenoxy)-N,N-dimethylethanamine 16c (130 mg,
0.47 mmol), N-methylhomopiperazine (97 mg, 0.85mmol), palladium acetate (16 mg, 0.070
mmol), BINAP (44 mg, 0.071 mmol), sodium tert-butoxide (123 mg, 1.28 mmol). The product
was purified by preparative HPLC method 60-100-60/10mL x min-1/30min (H2O/MeOH) yield
(12 mg, 8%). HRMS (ESI) m/z calcd for C16H25F2N3O (M+H)+: 314.2038, found: 314.2031; 1H
NMR (CDCl3, 400 MHz) : δ 6.06 (t, J = 3 Hz, 1H), 6.02 (t, J = 3Hz, 1H), 4.12 (t, J = 6 Hz, 2H),
3.48 (t, J = 3 Hz, 2H), 3.39 (t, J = 3 Hz, 2H), 2.75 (t, J = 6 Hz, 2H), 2.66 (t, J = 3 Hz, 2H), 2.55
(t, J = 3 Hz, 2H), 2.38 (s, 3H), 2.35 (s, 6H), 1.98 (q, J = 3 Hz, 2H) ; 13C NMR (CDCl3, 100 MHz)
: δ 153.17, 150.67, 148.60, 145.43, 128.80, 94.55, 92.58, 68.44, 58.24, 57.88, 57.13, 48.94,
48.49, 46.71, 46.03, 27.67.
5.24. N,N-dimethyl-2-{[5-(piperazin-1-yl)pyridin-3-yl]oxy}ethanamine (20a)
Compound 20a was prepared according to the general procedure E for Buchwald-Hartwig
cross coupling with 2-[(5-bromopyridin-3-yl)oxy]-N,N-dimethylethanamine 16a (92 mg, 0.37
mmol), Ntboc-piperazine (114 mg, 0.61 mmol), palladium acetate (6 mg, 0.027 mmol), XPhos
(23 mg, 0.048 mmol), sodium tert-butoxide (52 mg, 0.54 mmol). Then the N-tboc protective
group was removed using the general procedure G with 1,4-dioxane (6 mL) and HCl in 1,4-
dioxane (6 mL) for 0.5 mmol of crude material. The product was purified by preparative HPLC:
60-100-60/10mL x min-1/30min (H2O/MeOH) yield (53 mg, 43%). HRMS (ESI) m/z calcd for
1
C13H22N4O (M+H)+: 251.1865, found: 251.1866. H NMR (CDCl3, 400 MHz) : δ 7.65 (d, J = 2
Hz, 1H), 7.53 (d, J = 2 Hz, 1H), 6.64 (t, J = 2 Hz, 1H), 3.93 (t, J = 5.6 Hz, 2H), 3.04 (t, J = 5.2
Hz, 4H), 2.87 (t, J = 5.2 Hz, 4H), 2.61 (t, J = 2 Hz, 2H), 2.17 (s, 6H).