Enhanced arecoline derivatives as muscarinic acetylcholine receptor M1 ligands for potential application as PET radiotracers

Article abstract of DOI:10.1016/j.ejmech.2020.112623

Supported by their involvement in many neurodegenerative disorders, muscarinic acetylcholine receptors (mAChRs) are an interesting target for PET imaging. Nevertheless, no radiotracer is established in clinical routine. Within this work we aim to develop novel PET tracers based on the structure of arecoline. Fifteen novel arecoline derivatives were synthesized, characterized and tested for their affinity to the mAChRs M1-M5 and the conceivable off-target acetylcholinesterase. Five arecoline derivatives and arecoline were labeled with carbon-11 in good yields. Arecaidine diphenylmethyl ester (3b), arecaidine bis(4-fluorophenyl)methyl ester (3c) and arecaidine (4-bromophenyl)(4-fluorophenyl)methyl ester (3e) showed a tremendous gain in mAChR affinity compared to arecoline and a pronounced subtype selectivity for M1. Metabolic stability and serum protein binding of [¹¹C]3b and [¹¹C]3c were in line with properties of established brain tracers. Nonspecific binding of [¹¹C]3c was prevalent in kinetic and endpoint experiment on living cells as well as in autoradiography on native mouse brain sections, which motivates us to decrease the lipophilicity of this substance class prior to in vivo experiments.

Full text of DOI:10.1016/j.ejmech.2020.112623

Journal Pre-proof
Enhanced arecoline derivatives as muscarinic acetylcholine receptor M1 ligands for
potential application as PET radiotracers
Marius Ozenil, Katharina Pacher, Theresa Balber, Chrysoula Vraka, Alexander Roller,
Wolfgang Holzer, Helmut Spreitzer, Markus Mitterhauser, Wolfgang Wadsak, Marcus
Hacker, Verena Pichler
PII:
S0223-5234(20)30595-X
DOI:
https://doi.org/10.1016/j.ejmech.2020.112623
EJMECH 112623
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 28 May 2020
Revised Date: 17 June 2020
Accepted Date: 18 June 2020
Please cite this article as: M. Ozenil, K. Pacher, T. Balber, C. Vraka, A. Roller, W. Holzer, H. Spreitzer,
M. Mitterhauser, W. Wadsak, M. Hacker, V. Pichler, Enhanced arecoline derivatives as muscarinic
acetylcholine receptor M1 ligands for potential application as PET radiotracers, European Journal of
Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112623.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
© 2020 The Author(s). Published by Elsevier Masson SAS.

1
2
Enhanced arecoline derivatives as muscarinic acetylcholine receptor M1
ligands for potential application as PET radiotracers
3
4
5
Marius Ozenil,1 Katharina Pacher,1 Theresa Balber,1,2 Chrysoula Vraka,1 Alexander Roller,3 Wolfgang
Holzer,4 Helmut Spreitzer,4 Markus Mitterhauser,1,2 Wolfgang Wadsak,1,5 Marcus Hacker,1 Verena
Pichler1,*
6
7
1 Department of Biomedical Imaging and Image-guided Therapy, Division of Nuclear Medicine,
Medical University of Vienna
8
2 Ludwig Boltzmann Institute Applied Diagnostics, Vienna
3 X-ray Structure Analysis Centre, Faculty of Chemistry, University of Vienna
4 Department of Pharmaceutical Chemistry, Faculty of Life Sciences, University of Vienna
5 CBmed GmbH - Center for Biomarker Research in Medicine, Graz
*corresponding author
9
10
11
12
13
14
Abstract
15
16
17
18
19
20
21
22
23
24
Supported by their involvement in many neurodegenerative disorders, muscarinic acetylcholine
receptors (mAChRs) are an interesting target for PET imaging. Nevertheless, no radiotracer is
established in clinical routine. Within this work we aim to develop novel PET tracers based on the
structure of arecoline. Fifteen novel arecoline derivatives were synthesized, characterized and tested
for their affinity to the mAChRs M1-M5 and the conceivable off-target acetylcholinesterase. Five
arecoline derivatives and arecoline were labeled with carbon-11 in good yields. Arecaidine
diphenylmethyl ester (3b), arecaidine bis(4-fluorophenyl)methyl ester (3c) and arecaidine (4-
bromophenyl)(4-fluorophenyl)methyl ester (3e) showed a tremendous gain in mAChR affinity
compared to arecoline and a pronounced subtype selectivity for M1. Metabolic stability and serum
protein binding of [11C]3b and [11C]3c were in line with properties of established brain tracers.
1

25
26
27
Nonspecific binding of [11C]3c was prevalent in kinetic and endpoint experiment on living cells as well
as in autoradiography on native mouse brain sections, which motivates us to decrease the
lipophilicity of this substance class prior to in vivo experiments.
28
29
Keywords: muscarinic acetylcholine receptors, PET, carbon-11, neuroimaging
2

30
Introduction
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Muscarinic acetylcholine receptors (mAChRs) are G-protein coupled receptors, which bind
acetylcholine as endogenous ligand. Naming of the mAChRs is based on the additional activation by
the fungal toxin muscarine. Their involvement in neurotransmission of the central nervous system as
well as in the regulation of heart rate, muscle contraction and glandular secretion assigns mAChRs a
pivotal role in physiology.[1] Conversely, alterations in mAChR signal transduction are involved in
neurological disorders such as Schizophrenia, Alzheimer’s or Parkinson’s disease.[2] mAChRs are
divided into five subtypes (M1-M5), all of which are expressed in the human brain. M1, M4 and M5
are expressed only in certain brain areas, whereas M2 and M3 are abundant throughout the brain.[3]
The subtypes are classified by their activation of G proteins from the Gq/11 (M1, M3 and M5) or Gi/o
family (M2 and M4).[1] Comparison of the crystal structures of M1-M4 proteins showed high
similarity, which leads to difficulties in the development of subtype selective mAChR ligands.[4] A
closer relation of M1 to M4[5] was found compared to the structural relation of M1 to M3, which
potentially explains that the clinically relevant antagonist pirenzepine displays a higher affinity to M1
and M4 than M2 and M3.[6] The neurotransmitter acetylcholine binds to the orthosteric binding site,
which is highly conserved and sterically demanding. As a consequence, a huge amount of ligands
described in literature for mAChR are positive allosteric modulators (PAMs).[7] However, most of
them suffer from limited affinity.[1]
48
49
50
51
52
53
54
55
The naturally occurring alkaloid arecoline exhibits orthosteric partial agonist properties toward
mAChRs and is found in the betel nut. Its stimulating effects in the central nervous system underline
its ability to cross the blood brain barrier.[8] Although arecoline is also known to have activity on
selected nicotinic acetylcholine receptors[8], derivatization with typical muscarinic moieties as
applied within this work is strongly expected to deliver muscarinic selectivity. The inhibitor constants
(Ki) of arecoline toward mAChRs were determined to be in the range of 15-77 µM,[9] which is
certainly insufficient for mAChRs positron emission tomography (PET) imaging. However, arecoline
might act as a suitable lead structure for the development of high affinity mAChR ligands.[10] A
3

56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
typical pharmacophore in the orthosteric mAChR M1 binding motif displays a ‘cationic head’ which
interacts with the Asp105A residue.[11] In arecoline this feature is represented by the protonated
tertiary amine. Derivatives of arecoline should not contain bulky residues on the amine in order to
maintain the critical cationic head. In fact, studies on previously synthesized arecoline derivatives
confirmed that replacement of the N-methyl group of arecoline resulted in reduction or a complete
loss of agonistic activity.[12] Instead it appears feasible to use the N-methyl position unchanged,
leaving the O-methyl position open for chemical derivatization. The O-methyl position is moreover
the attack point for esterases leading to arecaidine as metabolite.[13] Moreover, the introduction of
carbon-11 at the N-methyl position does not alter the investigated structure leading to similar
behaviour as the unlabeled preclinically investigated compound and open up the possibility for
application in PET. PET is a non-invasive imaging modality and can be used to study receptor
distribution and occupancy in brain or the peripheral nervous system. Considering the mAChR
density in brain, it can be estimated that the affinity constant for a mAChR tracer should be around
3-50 nM for in vivo imaging.[14,15] Numerous animal and human imaging studies of mAChRs have
been undertaken in healthy and diseased brain. Till now, no protocol has been established in clinical
routine. Considering that acetylcholine was the first neurotransmitter to be discovered, even further
presses the question why mAChR-imaging has not yet found its way into clinics. On the one hand,
this can be attributed to the current lack of impact on clinical decisions derived from information
obtained from mAChRs imaging.[10] On the other hand, the ongoing research effort put on novel
radiotracers for mAChRs (Figure 1) emphasizes that imaging properties of current radiotracers are
still to be improved.
4

11CH3
N
O
N
H311C N
O
HN
O
OH
OH
F
N
O
O
O
O
11CH3
I
II
III
NH2
S
11CH3
N
O
N
O
N
HN
N
N
O
N
S
N
F
O
11CH3
O11CH3
VI
V
IV
77
78
79
Figure 1: Structures of carbon-11 radioligands investigated as mAChR PET tracers.
80
81
82
83
84
85
86
87
88
89
90
91
92
93
In the early days of mAChR tracer development, ligands with excessive affinity (e.g.
[11C]scopolamine[16] (I), [11C]benztropine[17]) heavily caused flow-dependent accumulation. More
recent mAChR tracer developments, [11C]GSK1034702[18] (II) and (S,R)-1-methylpiperidin-3-yl)2-
cyclopentyl-2-hydroxy-2-phenylacetate[19] (III), suffer from low specific in vivo binding and the
agonist [11C]AF150(S)[20] (IV) was deemed challenging for in vivo applications because of its rapid
metabolism and limited binding affinity. Furthermore, one of the main challenges in current mAChR
tracer development remains subtype selectivity[21], since for the majority of mAChR ligands subtype
selectivity and affinity have been contradictory properties.[1] The high affinity M4 PAM
[11C]VU0467485 (V) and two of its congeners overcome this discrepancy, yet limited specific binding
and insufficient BBB permeability stopped their further development.[22] [11C]LSN3172176 (VI) can
currently be considered as the most promising candidate for clinical mAChR M1 imaging and was just
advanced to human studies for further evaluation.[23] A BBB penetrant selective mAChR M1 PET
tracer would not only allow to investigate the molecular pathology of brain diseases, but also
facilitate the discovery of drugs for therapeutic applications.[23]
5

94
95
96
97
98
99
In this study, we aim to significantly enhance the potency of arecoline derivatives by synthesis of a
set of orthosteric compounds targeting the mAChR M1. We investigated their physico-chemical
behaviour and affinity toward all five muscarinic acetylcholine receptors. The most successful
candidates of this preclinical screening showed affinity in the low nanomolar range, and were
radiolabeled with carbon-11 for further pharmacological characterization toward radiotracer
development for PET.
100
101
102
Results and Discussion
Synthesis of arecoline-based low molecular weight compounds
103
104
105
106
107
108
109
110
111
112
113
114
115
Starting from arecaidine (1-methyl-3,6-dihydro-2H-pyridine-5-carboxylic acid) and 4-
methylmorpholine-2-carbonitrile a set of fifteen compounds was synthesized and six precursors
required for potential radiolabeling with carbon-11 were prepared from N-protected guvacine. Figure
2 and Figure 3 outline the synthetic routes for arecoline derivatives (1-3). 1a was prepared by
methylation of N-Fmoc guvacine in presence of Na2CO3. Esters 1b-c were prepared by reacting N-
Fmoc guvacine with alcohols activated as trichloroacetimidate. 1j was prepared via the acid chloride
route. Activation of N-protected or unprotected arecaidine with N,N’-dicyclohexylcarbodiimide and
subsequent spontaneous migration of the acyl residue gave the N-substituted urea derivatives 1k
and 3k. Fmoc deprotection by diethylamine afforded compounds 2a-c and 2j-k. Precursor 2i was
obtained by Boc removal in presence of trifluoroacetic acid. 1i and 3b-j were synthesized via
carbonyldiimidazol-mediated coupling. 3l-m were prepared by alkylation with an α-chloro amide.
Preparative chromatography using an alpha-1-acid glycoprotein stationary phase allowed for chiral
resolution of 3e.
116
117
118
The protected precursor compounds (1) were synthesized in good to excellent yields, whereas the
synthesis of the non-radioactive reference compounds (3) was more demanding leading to only
moderate yields. This difference in yields can be explained by two factors, resulting from the
6

119
120
121
122
zwitterionic characteristics of arecaidine compared to N-Fmoc guvacine. Firstly, the limited solubility
of arecaidine in most organic solvents necessitated DMF-compatible esterification procedures.
Secondly, arecaidine’s interionic interactions can reduce the access of the activating reagent to the
carboxylic group.
123
124
125
126
127
128
Figure 2: Synthesis overview of arecoline derivatives (3b-l) and their N-desmethyl congeners (2a,b,c,i,j,k).* 3b-g: 1. CDI, DMF
2. Na+R-, 3-25 %; 3h-j: 1. CDI, DMF, 2. RH, 3-18 %; 3k: DCC, DMF, 69 %; 3l: 1. NaH in DMF, 2. NaI, 11-(chloroacetyl)-5,11-
dihydro-6H-pyrido[2,3-b][1,4]benzodiazepin-6-one, 13 %. ** 1a: PG = Fmoc, Na2CO3, CH3I, DMF, 96 %; 1b-c: PG = Fmoc,
RC(=NH)CCl3, DCM, 87-93 %; 1i: PG = Boc, CDI, DMF, 2. RH, 12 %; 1j: PG = Fmoc, 1. SOCl2, 2. RH, 93 %; 1k: PG = Fmoc, DCC,
DMF, 48 %. *** 2a,b,c,j,k: DEA, ACN, 63-93 %; 2i: TFA, DCM, 79 %.
129
130
Figure 3: Synthesis of the arecoline derivative 3m by nucleophilic substitution.
131
132
133
Derivatives of 4-methylmorpholine-2-carboxylic acid (6-7) were prepared as congeners with
decreased lipophilicity compared to 3b and 3i (Figure 4). 6 was prepared by Ritter reaction in formic
acid[24] from 4. 8 was prepared by Grignard reaction with benzylmagnesium bromide from 4.
7

134
135
136
137
Carbonyldiimidazol-mediated coupling of 5 with sodium diphenylmethanolate afforded 7.
Surprisingly, compounds 4 and 5 were not literature reported. Synthesis of 4 could be realized by
addition of N-methylethanolamine to 2-chloro acrylonitrile and subsequent base cyclization, using an
adopted procedure of Kopach et al.[25] Acidic hydrolysis of the nitrile 4 afforded compound 5.
138
139
140
Figure 4: Synthesis of 4-methylmorpholine-2-carbonitrile (4) and its related carboxylic acid (5). 4 and 5 were used for the
synthesis of potential mAChR ligands 6-8.
141
142
143
144
145
146
147
148
149
150
Precursors (2) and non-radioactive reference compounds (3, 6-8) were fully characterized by 2D-
NMR spectroscopy, HRMS and UV-HPLC. Regarding benzhydryl derived compounds, the shift and
multiplicity of the benzylic proton signal is influenced by the type of the neighboring element and
hence the type of carbonyl compound. Esters (3b-g, 7) show a singlet at 6.86-7.00 ppm, whereas the
amides 3i and 6 exhibited a high field shifted signal at 6.28 ppm and 6.25 ppm respectively.
Furthermore, the type of carbonyl compound influences the shielding of the respective vinylic proton
of the arecaidine backbone. For the ester derivatives (3b-h, 3j, 3l) a multiplet at 6.87-7.32 ppm was
observed, compared to a high field shifted mutliplet at 6.15-6.56 ppm for the amide derivatives (3i,
k). In HRMS measurements all analyzed compounds could be ionized in positive mode yielding
[M+H]+ and/or [M+Na]+ cations of expected masses.
8

151
152
153
154
Compound 3k and 5 were isolated as monocrystals and analyzed by X-ray single crystal diffraction
analysis. The compounds crystalized in an orthorhombic crystal system (space group Pna21) and in a
monoclinic crystal system (space group P21/n), respectively. A detailed description is shown in the
Supporting Information.
155
156
Subtype selectivity analysis toward muscarinic acetylcholine receptors
157
158
159
160
161
Inhibition constants (Ki) against mAChR subtypes of arecoline, 3b-m and 6-8 were determined with a
competitive radioligand binding assay applying the common mAChR ligand N-methyl-
[3H]scopolamine ([3H]NMS). Arecoline, 3d, 3j-m and 8 feature inhibition constants in the micromolar
range or higher against all mAChR subtypes, whereas 3b, 3c, 3e-i, 6 and 7 show a drastically
increased affinity with values in the submicromolar to low nanomolar range (Table 1).
162
Table 1: Inhibition constants (Ki) given in nM, n≥3, determined by [3H]NMS competition binding.
M1
20,300±2,400
3.1±0.5
M2
M3
16,700±5,900
56±22
M4
4,700±410
16±4
M5
7,000±570
14±1
3a
3b
3,800±890
110±39
710±110
>4,500
3c
5.0±1.6
77±28
19±5
32±5
3d
>4,700
>2,200
>3,300
64±11
>2,600
42±23
(-)-3e
(+)-3e
3f
15.6±2.1
560±110
670±70
1480±40
3300±500
>4,500
56±10
7.8±1.7
52±6
42±2
27±5
850±240
690±170
1240±210
1,500±290
>2,200
720±420
1200±500
400±74
>3,300
>3,300
690±70
800±190
580±130
>2,600
>2,600
3h
270±130
3i
153±60
3j
148,000±35,000
204,000±53,000
>4,500
3k
>4,500
>2,200
9

3l
3m
6
>4,700
>4,700
610±50
710±90
>4,700
>4,500
>4,500
>4,500
>4,500
>4,500
>2,200
>2,200
>2,200
>2,200
>2,200
>3,300
>3,300
>3,300
>3,300
>3,300
>2,600
>2,600
>2,600
>2,600
>2,600
7
8
163
164
165
166
167
No specific competitive binding could be observed on membranes prepared from wild type CHO cells
(negative control), proving the high specificity of these ligands in the competitive radioligand binding
assay. The determined inhibition constants for arecoline (3a) match well with previously reported
data.[9]
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
3l-m are conjugates of the arecoline backbone with the clinically applied M1-selective antagonist
pirenzepine and were designed to promote subtype M1 selectivity. However, their low affinity
excluded them from further development. Exchange of the O-methyl group of arecoline with a
diphenylmethyl (3b), bis(4-fluorophenyl)methyl (3c) or (4-bromophenyl)(4-fluorophenyl)methyl (3e)
moiety led to a tremendous increase of affinity toward all subtypes to levels suitable for PET tracers.
Structures with fluorine substituents were investigated to enable potential use of fluorine-18 for
radiolabeling. The affinity dropped strongly when arecaidine was substituted with 4-
methylmorpholine-2-carboxylic acid (6, 7). Enantiomers of 3e show a distinct affinity profile, with a
twofold difference in M1 affinity being the most notable difference (p = .007). Interestingly, 3b, 3c
and 3e show a preference for subtype M1, whereas arecoline (3a) shows a higher affinity for subtype
M2 compared to M1. The derivatization of arecoline with diphenylmethyl moieties affected the
subtype selectivity to such an extent that also in the overall subtype selectivity profile the difference
in affinity is the highest between M1 and M2 for 3b (35-fold), 3c (140-fold), (-)-3e (36-fold) and (+)-3e
(86-fold). From a thermodynamical perspective, the similar target densities of the subtypes in human
brain[26] paired with the strong subtype selectivity for M1 over M2 should lead to an equally strong
subtype selectivity in in vivo imaging. In 3b and 3c the subtype selectivity of M1 over M3-M5 is less
10

184
185
186
187
188
189
190
pronounced (between 2.7-fold and 18-fold), which might appear insufficient for subtype-selective in
vivo imaging. However, [18F]FP-TZTP, which is a known in vivo M2 selective tracer, features only 3.4-
fold selectivity of the M1 over the M2 receptor in vitro.[27] Ravasi et al.[28] proposed, that in vivo
M2 selectivity of [18F]FP-TZTP is caused by the slower koff from M2 compared to the other subtypes.
This highlights the importance of ligand-receptor kinetics for mAChR tracer development and
motivated us to study the in vitro kinetics of [11C]3c (see section Interactions of [11C]3c with CHO-M1
cells).
191
Specificity analysis of competing electric eel acetylcholinesterase (EeAChE) affinity and metabolism
192
193
194
195
196
197
198
199
200
201
202
203
204
Compounds 3a-m and 6-8 could unintentionally act as ligand or substrate for AChE
(acetylcholinesterase), which would cause specificity problems. AChE is a certainly conceivable off-
target in the development of orthosteric ligands for mAChRs and especially of interest for arecoline-
derived compounds when considering the structural similarity to the known AChE-inhibitors
neostigmine and pyridostigmine.[29] Inhibition constants (Ki) against EeAChE (acetylcholinesterase
from Electrophorus electricus) of 3a-m and 6-8 and tacrine as positive control were determined by
Ellman’s assay. The obtained Ki of the positive control tacrine was 30 ± 10 nM, which corresponds
well to the range of literature reported values[30] (for graphs see Supporting Information, Figure S7).
For 3a-m and 6-8 no inhibition compared to the blank could be perceived in the cATI/(sec/OD)
diagram at the highest concentration tested (16.5 µM). Graphs are shown in the Supporting
Information, Figure S6. Furthermore, a two-way ANOVA followed by a Dunnett’s multiple
comparisons test revealed that the slopes are not significantly different (α > 0.05), which allows to
conclude that the Ki against EeAChE of 3a-m and 6-8 is higher than 16.5 µM.
205
206
207
208
Interestingly, no significant EeAChE metabolism could be observed for arecoline derivatives 3a-m and
6-8 under these conditions, not even after 60 min of incubation (α > 0.05). Overall, it was ruled out
that 3a-m and 6-8 are EeAChE substrates under the used assay conditions. Additionally, 3a-m and 6-8
did not show any EeAChE inhibitory effects at concentrations up to 25 µM.
11

209
210
Concluding, the investigated set of mAChR ligands exhibits high specificity against the target of
choice as they were neither inhibitors nor substrates to the off-target EeAChE.
211
HPLC-logPowpH7.4 determination and structure-activity relationship
212
213
214
215
216
Although the predictive factor of HPLC-logPowpH7.4 for prediction of blood brain barrier penetration is
critically discussed in literature,[31–33] still this value is broadly determined in drug development.
pH7.4
The HPLC-logPow
as a measure for the affinity of compounds to a lipophilic stationary phase
allows estimation of nonspecific binding in fatty tissue, other lipophilic tissues or plasma protein
binding and was therefore analyzed.
pH7.4
217
218
219
220
The HPLC-logPow
value was determined chromatographically and tPSA was calculated (Table 2).
All target compounds (3a-m and 6-8) exhibited HPLC-logP values in the range of established BBB
permeable brain tracers[31] and 3a-k and 6-8 exhibit a tPSA (total polar surface area) < 60 Å2,
supporting BBB penetration.[34]
221
Table 2: Overview of substance parameters. tPSA was calculated using ChemBioDraw 13.0.2.3021.
pH7.4
tPSA [Å2]*
29.54
29.54
29.54
48.00
29.54
54.26
38.77
46.61
32.34
57.23
HPLC-logPow
-1.62±0.62
3.32±0.04
3.62±0.08
3.07±0.04
4.12±0.15
1.91±0.17
3.66±0.08
2.94±0.05
2.25±0.12
-0.70±0.54
3a
3b
3c
3d
3e
3f
3g
3h
3i
3j
12

3k
3l
3m
6
52.65
91.31
91.31
41.57
38.77
29.54
2.24±0.13
0.57±0.37
0.96±0.31
2.30±0.12
2.79±0.05
1.00±0.30
7
8
222
223
224
225
226
227
228
229
230
231
For identification of the most promising compounds in terms of subtype selectivity independent from
the absolute value, the ratio values of the Ki of M4/M1 was correlated with the ones for M3/M1.
Three main groups could be recognized showing low subtype selectivity (Figure 5A, blue),
intermediate preference for M1 (Figure 6A, red) and high preference for M1 subtype (Figure 5A,
green). The three compounds with the highest preference for the M1 subtype were as follows in
increasing order: 3i < 3c < 3b. Obviously, the diphenylmethyl moiety was crucial to achieve highly
improved subtype selectivity against the mAChR-M1 subtype compared to arecoline. Notable is also
the enantiomeric selectivity of (+)-3e over (-)-3e for mAChR-M1 giving the opportunity to fine-tune
the affinities via stereochemistry.
232
233
234
Figure 5: (A) Evaluation of the subtype selectivity by comparison of the ratio values of M4/M1 and M3/M1. The analysis
identified a group of compounds with low subtype selectivity (blue), compounds with intermediate (red) and high preference
13

235
236
for M1 (green). (B) Correlation of the pKi toward M1 and the HPLC-logPow (pH 7.4). A logarithmic trend was observed
showing that higher lipophilicity leads to higher affinities.
237
238
239
240
241
242
Correlation of the pKi toward M1 and the logPowpH7.4 values showed a logarithmic trend of affinity and
lipophilicity (R2 = 0,56). However, as the Ki of compounds 3d, 3l, 3m and 8 toward M1 was higher
than the concentration range of the experiment, only a trend analysis was possible. Especially, the
analysis of the enantiomers (+)-3e and (-)-3e exemplarily shows, that an increase in affinity is
possible without an alteration in lipophilicity and underlines the importance of enantiomeric
selectivity for the binding to mAChR M1.
243
244
245
Overall, the most promising compounds in terms of subtype selectivity, absolute affinity against
mAChR M1 and tPSA were 3b and 3c, which were further evaluated for radiolabeling with carbon-11.
The isostere 3i was additionally used for comparison purposes in metabolic studies.
246
Docking studies
247
248
Docking of protonated 3c in the active center of mAChR M1 (PDB: 5CXV)[5] supports that it adopts a
similar binding pose compared to tiotropium (Figure 6A).
14

249
250
251
252
Figure 6: A: Lowest binding energy docking poses of 3c (salmon, -20.0 kcal/mol) and tiotropium (light green, -22.4 kcal/mol)
in the orthosteric binding site of human M1 mAChR. Ser109A is covered by the docked ligands. B,C: Pharmacophore-derived
interactions of tiotropium (B) and 3c (C) with the binding pocket environment calculated and visualized with LigandScout.
253
254
255
256
257
258
259
260
In this pose tiotropium as well as protonated 3c feature electrostatic interaction of the quaternary
ammonium cation with ASP105A and hydrophobic interactions of the aromatics to several apolar
amino acid side chains. In the docked binding pose of tiotropium ASN382A acts as hydrogen bond
donor and acceptor for the hydroxyl and carbonyl moiety, whereas in protonated 3c the carbonyl
acts as hydrogen bond acceptor for CYS407A. Additionally, protonated 3c features hydrogen bonds
to SER109A and THR189A (Figure 6C). Hydrogen bonds to organic fluorine have low binding energies,
yet they are known to affect chemical behavior[35] and can hence potentially influence ligand
binding.
261
262
Furthermore, we performed molecular docking of protonated 3b and 3i in order to get a molecular
insight into the reason for the approx. 50-fold higher mAChR M1 binding affinity of 3b.
15

263
264
265
266
267
268
269
Unexpectedly, in these docking experiments 3i (lowest estimated binding energy: -20.3 kcal/mol;
lowest binding affinity score: -31.4) performed even slightly better than 3b (-20.2 kcal/mol; -26.9).
Consequently, we assume that the experimentally observed loss of affinity of 3i compared to 3b is
not based on the lack of a suitable conformer inside the binding pocket caused by the more rigid
amide bond, but rather results from reduced flexibility during its way to the active center.
Furthermore, also electronical differences of the neighboring moieties between 3b and 3i, as
observed by NMR spectroscopy, could affect the capability of the ligands to reach the binding pocket.
270
Radiolabeling of the most promising compounds with carbon-11
271
272
273
274
275
276
The most promising compounds 3b and 3c regarding subtype selectivity, receptor specificity and
physico-chemical parameters were chosen for radiolabeling. The radiolabelling of 3i was performed
for the analysis of metabolites obtained by liver microsomes. All other compounds were radiolabeled
to show the substrate scope of the chosen reaction conditions. First small scale reaction were
performed to investigate the optimal labeling conditions. Subsequently, fully-automated
radiosyntheses were set up to study metabolism, plasma protein binding and cell binding.
277
278
279
280
281
282
Small scale reactions of compounds 2a, 2b and 2j with [11C]CH3I yielded [11C]3a, [11C]3b and [11C]3j
(Figure 7) in good yields in both solvents, ACN (Table 3) and DMSO (Table 4) as determined by RP-
HPLC without further purification. When using ACN, radiochemical yields[36] significantly increased
with reaction time and temperature. In DMSO, labeling was almost quantitative at both tested
reaction times and temperatures. Thus, DMSO at 20 °C for 2 min was initially chosen for
implementation of the automated radiosynthesis.
283
284
285
Table 3: Non-isolated radiochemical yields [%] in ACN as determined by radio-HPLC (ND = not determined, n = 1). ACN was
not investigated as solvent for labeling of [11C]3j, because an almost quantitative non-isolated radiochemical yield was
observed using DMSO (Table 4).
[11C]3a
[11C]3b
[11C]3j
20 °C
23
70 °C
55
20 °C
37
70 °C
58
20 °C
ND
70 °C
ND
2 min
5 min
33
81
45
68
ND
ND
16

286
287
Table 4: Non-isolated radiochemical yields [%] in DMSO as determined by radio-HPLC (n = 1).
[11C]3a
[11C]3b
[11C]3j
20 °C
92
120 °C
91
20 °C
79
120 °C
87
20 °C
98
120 °C
95
2 min
5 min
93
91
87
85
96
94
288
289
290
291
292
293
294
295
The fully automated syntheses comprised of the preparation of [11C]CH3I, reaction with the
respective precursor, semi-preparative purification as well as formulation to physiological conditions
was set up for compounds 3a, 3b, 3c, 3i, 3j and 3k. Although the small-scale reaction showed
complete conversion at room temperature in DMSO, for the fully-automated synthesis the reaction
temperature was increased to 80 °C, because unreacted [11C]CH3I was observed during semi-
preparative HPLC of the crude reaction mixture at 20 °C for all compounds. Precursor 2k features two
potential nucleophilic sites that could react with [11C]CH3I (see Supporting Information).
296
297
Figure 7: Reaction scheme of the radiochemical labeling.
298
299
300
301
302
303
In total 32 large-scale batches of carbon-11 compounds were produced (Table 5). Due to different
semi-preparative HPLC runtimes the overall synthesis time from end of bombardment varies slightly
among the products and ranges between 40-45 min, which is approximately two half-lives of carbon-
11 (t1/2 = 20 min). Precursors (2) were employed in a roughly 100-fold excess compared to [11C]CH3I.
The radiochemical purity as determined by HPLC exceeded 96 % in all batches and pH and osmolality
were determined to be in the physiological range. All radiolabeled compounds showed high purity
17

304
305
and physico-chemical parameters as determined in the quality control were in accordance with the
guidelines of the European Pharmacopoeia for radiopharmaceutical preparations.[37]
306
307
Table 5: Production parameters of carbon-11-labeled compounds. Yield is related to the estimated starting activity of
[11C]CO2.
[11C]3a
3
[11C]3b
5
[11C]3c
19
[11C]3i
2
[11C]3j
1
[11C]3k
2
n
yield (GBq)
RCY (%)
5.9±1.0
18.7±2.7
5.3±1.6
17.4±6.8
3.8±1.3
12.9±3.8
2.7±2.0
8.6±4.8
7.2
4.3±1.0
13.2±2.4
21.1
Am at EOS
(GBq/µmol)
58±17
93±80
122±115
30±6
43
132±36
308
309
310
311
We showed that 3a-c and 3i-k can be carbon-11 labelled straightforwardly in high yields,
independent of the carbonyl substituent (R). The high yields of the radiosynthesis of all tested
compounds enable a potential application for at least two patients per synthesis.
312
Physico-chemical parameters and pharmacological behavior of radiolabeled compounds
313
314
315
316
317
318
319
320
321
322
Human liver microsome metabolism was studied with compounds [11C]3b, [11C]3c and [11C]3i. For all
of them the main radiometabolite was identified as [11C]arecaidine and the rate of its formation was
found out to be heavily dependent on the structure of the parent compound (Figure 8). [11C]3i was
almost unaffected by human liver microsomes, whereas 30 % [11C]3b and 53 % [11C]3c were
unmetabolized after 1 h. Furthermore, a second radiometabolite with a retention time between
arecaidine and the parent compound was observed for all three compounds (Supporting
Information, Table S2, Figure S8). This metabolite never represented more than 10 % of the total
activity and was identified as arecaidine ethyl ester (C9H15NO2), based on HPLC-HRMS measurements
(Supporting Information) of nonradioactive incubations of 3b with human liver microsomes.
Formation of the ethyl ester can be explained by the presence of EtOH in the tracer formulation. In
18

323
324
325
326
327
vivo experiments are expected to be unaffected by [11C]arecaidine ethyl ester formation, rationalized
by the concomitant stronger dilution of EtOH. Considering the expected low transport of
[11C]arecaidine across the blood brain barrier[38] and the low affinity to all mAChRs, we are
confident that ester cleavage in the liver will not interfere with neuroimaging but can potentially
affect applications in the peripheral nervous system.
328
329
330
Figure 8: Metabolic stability of compounds [11C]3b, [11C]3c and [11C]3i against human liver microsomes. Error bars represent
standard deviation of n=3 ([11C]3b and [11C]3c) or n=2 ([11C]3i) incubations of individual batches.
331
332
333
334
335
Serum protein binding of [11C]3b, [11C]3c and [11C]3i was studied by ultrafiltration, demonstrating
5.8±1.0 %, 9.0±1.2 % and 18.2±0.4 % (n≥3, ±sd) free fraction, respectively. The measured serum
protein binding is comparable to established brain tracers[39], substantiating the possible utilization
of [11C]3b, [11C]3c and [11C]3i for in vivo imaging. Plasma stability of [11C]3c was 94±2% after 60 min
(n=3).
336
337
338
339
340
To reduce tracer metabolism, we exchanged the ester for an amide bond (3i), which resulted in
drastic improvement of metabolic stability. However, the mAChR affinity of 3i is much lower
compared to its bioisostere 3b. To sum it up, [11C]3b and [11C]3c represent the most promising
substances of this series regarding mAChR PET tracer development. Comparing these two, [11C]3c
stands out, because of its higher in vitro metabolic stability and M1 subtype selectivity.
341
Interactions of [11C]3c with CHO-M1 cells
342
343
The kinetics of the interaction of [11C]3c with living CHO-M1 cells was studied using LigandTracer. On
the first view promising binding curves, reaching equilibrium in 8 to 15 min, were obtained. However,
19

344
345
346
347
348
349
350
the observed binding did not reach saturation at a concentration 100-fold above Ki but showed a
rather linear equilibrium signal increase with concentration (maximum concentration of 10× Ki shown
in Figure S9A for the sake of readability). Furthermore, CHO-M1 cells preblocked with 10 µM
scopolamine showed essentially the same binding kinetics as unblocked CHO-M1 cells Figure S9A).
Additionally, the binding curvature appeared independent on [11C]3c concentration (Figure S9B).
Considering these observations, we concluded that the binding observed in the LigandTracer
experiments cannot exclusively be explained by the interaction of 3c with mAChR M1.
351
352
353
354
355
356
357
358
359
To study this further, we performed a direct radioligand binding assay using [11C]3c and investigated
the binding behavior of blocked CHO-M1 cells versus unblocked CHO-M1 cells at three different
concentrations in six-well plates (Figure 9). At a radioligand concentration of 0.3 nM blocking with
10 µM scopolamine and 10 µM pirenzepine significantly reduced the binding of [11C]3c to CHO-M1
cells. However, CHO cells lacking the M1 receptor (blank) show an overall lower radioligand binding
compared to the blocked CHO-M1 cells. Hence, the observed [11C]3c binding is not fully blockable
with the used concentrations and reference ligands. With increasing concentrations of 3c blocking
with scopolamine and pirenzepine becomes less effective, leading to no significant difference in
radioligand binding between unblocked and blocked CHO-M1 cells at 3 nM and 30 nM of 3c.
360
*
1.5
1.0
0.5
0.0
1.5
1.0
0.5
0.0
c(3c) = 3 nM
c(3c) = 30 nM
ns
ns
361
362
363
Figure 9:[11C]3c binding to CHO cells. Data are displayed normalized to the binding on the non-blocked CHO-M1 cells. Groups
were compared using an ordinary one-way ANOVA and Tukey’s multiple comparisons test, with a single pooled variance (*:
20

364
365
P ≤ 0.05, ***: P ≤ 0.001, ****: P ≤ 0.0001, ns: not significant, n ≥ 3). At all tested [11C]3c concentrations the normalized cell-
bound activity of the CHO-K1 control cells was not significantly different from the blank.
366
367
368
369
As control, the similar assay was performed at ≤ 4 °C to investigate the ligand binding without active
process contributions (Figure 9).[40] At CHO-M1 cells ≤ 4 °C [11C]3c binding is reduced to the same
extent as in blocked CHO-M1 cells, indicating that active cellular processes are involved in the
binding of [11C]3c to living cells.
370
371
372
373
The extent of specific binding is limited to the amount of expressed target, whereas nonspecific
binding increases linearly with the radioligand concentration and is not saturable. Thus, the specific
binding was only visible at a low radioligand concentration (0.3 nM), while at higher concentrations
(3 nM, 30 nM) mostly nonspecific binding was detected.
374
In vitro autoradiography
375
376
377
378
379
380
Total binding of [11C]pirenzepine (15 nM) was reduced to background signal in the presence of 10 µM
scopolamine or atropine confirming that the brain sections and the used procedure allows the
visualization of specific binding to mAChRs. However, neither of the blocking agents significantly
reduced the radioactive signal derived from total [11C]3c (30 nM) binding in a total of 30 brain
sections derived from different brain regions of two different subjects, demonstrating only
nonspecific binding (Figure 10).
21

381
382
383
384
385
Figure 10: Autoradiography of mice brain sections. Blocking was performed by co-incubation with 10 µM scopolamine. The
signal at 15 nM [11C]pirenzepine blocked is below limit of detection and the position of the invisible sections is therefore
marked with a blue ellipse. Radioactivity is displayed as a linear color gradient. Sections separated in boxes are derived from
different experiments, allowing only semi-quantitative analysis.
386
387
388
Also at a concentration close to the KD [11C]3c (3 nM) did not show a difference between baseline and
blocking, whereas [11C]pirenzepine at a concentration close to the KD (15 nM) clearly showed specific
binding.
389
Conclusion
390
391
392
393
394
395
396
397
Despite the pivotal role of mAChRs in human physiology and disease, non-invasive external imaging
thereof has not yet found its way into clinical routine. Constrained imaging properties of the
currently available mAChR PET tracers demand improvement in order to correspond to the clinical
needs. In search of an improved PET tracer for mAChRs fifteen novel arecoline-derived compounds
were synthesized, characterized and tested for their affinity toward mAChRs and EeAChE.
Conjugation of arecaidine to diphenylmethyl-containing structures afforded Ki values in the low
nanomolar range, which is suitable for the application as mAChR PET probe. Besides, the difference
in affinity of (+)-3e and (-)-3e underline the applicability of stereochemistry to impact mAChR M1
22

398
399
400
401
402
403
404
405
406
407
ligand development. Moreover, all tested compounds were shown to neither act as orthosteric
ligand nor substrate for the off-target AChE. Fully-automated carbon-11-labeling procedures for
[11C]3a-c and [11C]3i-k were developed and the most promising candidates underwent further studies
to assess their metabolic stability and serum protein binding, where [11C]3b and [11C]3c showed
suitable properties for future tracer development. Real-time and endpoint radioligand binding
studies using [11C]3c on living cells as well as in vitro autoradiography on mouse brain sections
revealed pronounced nonspecific binding, which motivates us to reduce the lipophilicity for future
tracer candidates of similar structure prior to in vivo studies. Although the observed nonspecific
binding of [11C]3c restrains this compound from acting as imaging agent, it can still be considered as
mAChR M1 ligand for purposes where nonspecific binding is less of a problem, e.g. therapeutics.[41]
408
409
Experimental Section
Materials
410
411
412
413
414
415
416
417
418
419
420
421
422
Diphenylmethyl trichloroacetimidate[42], furan-2-yl(phenyl)methanol[43], arecaidine[44] and N-Boc-
guvacine[44] were synthesized as published previously. N-Fmoc-guvacine (≥ 98 %, Santa Cruz
Biotechnology), arecoline hydrobromide (XA BC-Biotech), scopolamine hydrobromide (Sigma-
Aldrich), pirenzepine dihydrochloride (≥ 98 %, Sigma-Aldrich), diethylamine (DEA, > 99 %, Fisher
Scientific), thionylchloride (≥ 99 %, Sigma-Aldrich), triethylene glycol monomethyl ether (95 %,
Aldrich), diphenylmethanol (99 %, Aldrich), 4,4′-dimethoxybenzhydrol (98+%, Alfa Aesar), 4,4’-
difluorobenzhydrol (98+%, Alfa Aesar), (4-bromophenyl)(4-fluorophenyl)methanol (fluorochem),
benzhydrylamine (97 %, Alfa Aesar), N,N’-dicyclohexycarbodiimide (DCC, 99 %, Alfa Aesar), 1,1’-
carbonyldiimidazole (CDI, reagent grade, Sigma Aldrich), iodomethane (CH3I, ≥ 99 %, Sigma-Aldrich),
trichloroacetonitrile (98 %, Alfa Aesar), 1,8-diazabicyclo[5,4,0]undec-7-en (> 97 %, Merck-
Schuchardt),
11-(chloroacetyl)-5,11-dihydro-6H-pyrido[2,3-b][1,4]benzodiazepine-6-one
(abcr),
trifluoroacetic acid (99 %, Alfa Aesar), aq. HBr (48 %, Sigma Aldrich), formic acid (≥ 95 %, Merck), 2-
chloroacrylonitrile (98 %, Sigma Aldrich), 2-(methylamino)ethanol (99 %, Alfa Aesar), KOtBu (≥ 98 %,
23

423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
Sigma Aldrich), 5,5'-dithiobis-(2-nitrobenzoic acid) (DTNB, ≥98 %, Carl Roth), acetylthiocholine iodide
(ATI, 98 %, Sigma Aldrich), tacrine hydrochloride (≥98 %, Santa Cruz Biotechnology),
≥
acetylcholinesterase from Electrophorus electricus (1323 units/mg, Sigma Aldrich), sodium hydride
(NaH, 60 % dispersion in mineral oil, Aldrich), Mg turnings (≥ 99 %, Sigma Aldrich), benzyl chloride (>
99 %, Merck), 5-bromo-2-fluoropyridine (> 98 %, TCI), ethyl formate (97 %, Sigma Aldrich),
dibromoethane (> 98 %, Merck), sodium carbonate (Na2CO3, anhydrous for analysis, Merck), sodium
bicarbonate (NaHCO3, ≥ 98 %, Fluka), sodium iodide (NaI, 99.5 %, Merck), poly(ethyleneimine) (PEI,
50 % in H2O, Fluka), N,N-dimethylformamide (DMF, 99.8 % extradry over molecular sieve, Acros),
acetonitrile (ACN, Sigma Aldrich), dichloromethane (DCM, EMPROVE® ESSENTIAL, Merck), n-hexane
(PE, Hi-PerSolv CHROMANORM, VWR), ethyl acetate (EE, 99.9 %, VWR), diethyl ether (Et2O, 100 %,
VWR), ethanol (EMROVE® exp, Merck), dimethylsulfoxide (DMSO, ≥ 99.9 %, Sigma Aldrich) and C18
Sep-Pak (WAT020515, Waters) cartridges were purchased and used as received. Solvents indicated as
“dry” were stored over molecular sieves for a minimum of 2 days. Reaction progress was monitored
by pre-coated TLC sheets ALUGRAM® Xtra SIL G/UV254 (0.20 mm silica gel 60 with fluorescent
indicator). Silica gel on TLC plates (0.25 mm, 20x20 cm, 99571 Sigma Aldrich) was used for
preparative separation unless stated otherwise. In indicated cases aluminiumoxide 60 F254 neutral
(Merck) was used for preparative TLC. [N-methyl-3H]scopolamine methyl chloride (2.964 TBq/µmol,
37 MBq, in 1 mL ethanol) was purchased from PerkinElmer.
441
442
Instrumentation
443
444
445
446
447
NMR samples were measured in chloroform-d (CDCl3, ≥ 99.8 %, stabilized with silver foil, Sigma
Aldrich) at 25 °C. The center of the (residual) solvent signal was used as an internal standard which
was related to TMS with δ 7.26 ppm (1H in CDCl3) and δ 77.0 ppm (13C in CDCl3). 19F NMR spectra
were referenced via Ξ ratio and gsHMBC spectra of 15N were referenced against neat, external
CH3NO2. Unless stated otherwise full and unambiguous assignment of NMR signals was achieved by
24

448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
combination of standard NMR techniques, such as COSY, NOESY, APT, HSQC and HMBC experiments.
NMR spectra of intermediates were recorded with a Bruker Avance III 200 spectrometer (200 MHz
1
for H). NMR spectra of final compounds were recorded with a Bruker Avance III 400 spectrometer
1
(400 MHz for H, 100 MHz for 13C, 40 MHz for 15N and 376 MHz for 19F) or a Bruker Avance III 600
spectrometer (600 MHz for 1H, 150 MHz for 13C and 565 MHz for 19F). Mass spectra were obtained on
a Bruker maXis 4G instrument (ESI-TOF, HRMS). All newly synthesized or tested compounds (1a-8)
passed a pan assay interference compounds filter.[45] An Agilent 1260 Infinity HPLC System with a
flow of 1 mL/min and XBridge Shield RP18 2.5 µm, 50 x 3.0 mm column was used for determination
of the radiochemical yield in the small scale reactions, quality control of the radiotracers as well as
for the assessment of the tracer stability in formulation (further parameters shown in Table 6). LogP
measurements were performed and interpreted as published recently[31] on an Agilent HPLC system
consisting of an autosampler (series 1100), pump (series 1200), diode array detector (series 1100)
and a radiodetector (Ramona, Elysia-Raytest) using an apHERA (5 µm, 10 x 6 mm) stationary phase.
All tested compounds (3a-m, 6-8) were shown to have a purity exceeding 95 % as analyzed by the
logP HPLC method without addition of toluene and triphenylene. The same HPLC System was used
for the analysis of human liver microsomes metabolites with an XSelectTM (HSS T3, 3.5 µm, 100 x
4.6 mm) stationary phase, a flow of 1 mL/min and a gradient program (0.0-3.0 min: 95 % pH 9.3, 5 %
ACN; 5.0-14.0 min: X % pH 9.3, 100-X % ACN; 15.0-16.0 min: 95 % pH 9.3, 5 % ACN; with X being 50,
45 and 63 for compounds 3b, 3c and 3i, respectively). For the analysis of EeAChE metabolism a
similar gradient was used: 0.0-3.0 min: 98 % pH 9.3, 2 % ACN; 5.0-14.0 min: 40 % pH 9.3, 60 % ACN;
15.0-16.0 min: 98 % pH 9.3, 2 % ACN. The used eluent buffer is a 25 mM NH4H2PO4 solution set to pH
9.3 with NaOH. The LOQ given in Table 6 was estimated based on the standard deviation of the
response and the slope.[46]
471
472
Table 6: Separation conditions for analytical HPLC as part of radiotracer quality control and for the concentration
determination of the 3e enantiomers.
Compound
Eluent
kprec.
kprod.
LOQ [µg/mL]
25

3a
3b
3c
3e
3i
20% ACN, 80% pH 9.3
60% ACN, 40% pH 9.3
60% ACN, 40% pH 9.3
75% ACN, 25% pH 9.3
40% ACN, 60% pH 9.3
20% ACN, 80% pH 9.3
50% ACN, 50% pH 9.3
1.20
2.00
1.93
-
2.67
3.53
3.73
2.35
4.60
3.67
2.67
0.88
1.05
1.07
1.20
1.83
1.99
0.76
2.53
1.80
1.53
3j
3k
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
CHO-K1 cells stably transfected with human muscarinic receptors M1-M5 were obtained from
Missouri University of Science and Technology cDNA Resource Center (Cell Catalog#: CEM1000000,
CEM2000000, CEM3000000, CEM4000000, CEM5000000) and cultivated in Ham’s F12 Nutrient
Mixture (Gibco, Life Technologies Limited) containing 10 % FBS (Gibco, Life Technologies Limited),
250 µg/mL Geneticin® (G418, Thermo Fisher) at 37 °C and 5 % CO2 in a cell incubator. GibcoTM
Trypsin-EDTA (0.05%) was used for passaging cells. Untransfected CHO-K1 cells were obtained from
ATCC and cultivated like described above but in media free of Geneticin®. Protease Inhibitor Cocktail
powder (P2714-1BTL, Sigma-Aldrich) was dissolved in 10 mL water and used as such. The protein
concentration of membrane suspensions was determined with Pierce™ BCA Protein Assay Kit
(Thermo Scientific) on a BioTek Synergy HTX multi-mode reader. A M-36 tygon tubed Cell Harvester
(Brandel®) and Whatman™ GF/B filters were used for the filtration of radioligand binding assays.
Filter disks were counted with 2 mL Ultima Gold™ (high flashpoint LSC cocktail, PerkinElmer) using a
300 SL Automatic TDCR liquid Scintillation Counter (HIDEX) in CPM mode. LigandTracer grey and
yellow (ridgeview instruments ab, Sweden) with LigandTracer Control V2.0.1 and TraceDrawer V1.8.1
were used for measuring real-time kinetic on living cells.
489
490
491
A GE PETtrace 860 cyclotron (General Electric Medical System, Sweden) was used for the production
of [11C]CO2 by a 14N(p, α)11C reaction in gas target filled with N2 + 1% O2 (Topline, Messer) at a beam
current of 65 µA. Transformation to [11C]CH3I, radiolabeling of the precursor and tracer purification
26

492
493
494
were carried out in a TRACERlab™ FX C Pro synthesizer using the built in semi-preparative HPLC
system featuring a Linear Instruments Model 200 UV/Vis detector and a LaPrep HPLC pump
(VWR).[47]
495
Table 7: Separation conditions for semi-preparative HPLC during radiosynthesis.
Flow
Compound
Column
Eluent
kprec.
kprod.
[mL/min]
5.0
3a
3j
SupelcosilTM LC-ABZb, 5 μm, 15% ACN, 85% pH 9.3
3.25
1.70
2.63
3.12
4.88
2.37
4.60
4.60
250 × 10 mm
20% ACN, 80% pH 9.3
5.0
3b
3c
SupelcosilTM LC-ABZb, 5 μm, 50% ACN, 50% pH 9.3
5.0
250 × 10 mm
Nucleodur PolarTec, 5 µm,
250 x 10 mm
50% ACN, 50% pH 9.3
45% ACN, 55% pH 9.3
7.5
3i
5.0
5.0
2.60
1.75
3.92
3.33
ProntoSIL 120 C18 ace-EPS,
5 µm, 250 x 10 mm
3k
60% ACN, 40% pH 6.8
496
497
498
499
500
501
502
Metabolic stability was studied with Corning® Gentest™ HLM, ~20-Donor Pool, Mixed Gender
microsomes and Corning® Gentest™ NADPH Regenerating System, Solution A and B. Pooled Human
Serum (Innovative Research, Inc.), Centrifree®, Ultracel®PL Regenerated Cellulose Centrifugal filters
(Merck Millipore), Microcon®-10 Ultracel®PL-10 Regenerated Cellulose 10,000 NMWL Centrifugal
Filters and a 2480 Wizard2 Automatic Gamma Counter (Perkin Elmer) were used for determination of
serum protein binding.
503
504
Synthesis
505
506
1-(9H-fluoren-9-yl)methyl 3-methyl 1,2,5,6-tetrahydropyridine-1,3-dicarboxylate (N-Fmoc-
guvacoline) (1a). N-Fmoc-guvacine (50.0 mg, 0.143 mmol) and Na2CO3 (18.2 mg, 0.172 mmol) were
27

507
508
509
510
suspended in dry DMF (5 mL) for 30 min at 20 °C. CH3I (100 µL, 1.61 mmol) was added dropwise and
the reaction mixture was stirred o/n. The reaction mixture was poured on H2O (60 mL) and extracted
with Et2O (3x 30 mL). The combined organic layers were washed with H2O (2x 15 mL), dried with
Na2SO4, filtered and evaporated to give 1a (49.8 mg, 96 %) as a highly viscous oil.
511
512
513
1H-NMR (200MHz, CDCl3) δ 7.77 (d, J=7.3Hz, 2H), 7.59 (d, J=7.3Hz, 2H), 7.40 (t, J=7.3Hz, 2H), 7.31 (t,
J=7.3Hz, 2H), 7.08 (s, 1H), 4.46 (d, J=6.8Hz, 2H), 4.27 (t, J=6.8Hz, 1H), 4.22-4.17 (m, 2H), 3.79 (s, 3H),
3.52 (brs, 2H), 2.29 (brs, 2H).
514
515
516
517
518
519
methyl 1,2,5,6-tetrahydropyridine-3-carboxylate (guvacoline) (2a). 1a (49.8 mg, 0.137 mmol) was
dissolved in ACN (3 mL), diethylamine (300 µL, 2.91 mmol) was added and the reaction mixture was
stirred for 2.5 h at 20 °C. The reaction mixture was evaporated to about 1 mL and the product was
isolated by preparative TLC (PE:EE = 1:6 + 1% DEA; Rf = 0.13-0.33) and appeared as yellowish oil
(17.9 mg, 93 %).
520
521
522
523
1H-NMR (400MHz, CDCl3) δ 7.04 (m, 1H, H-4), 3.73 (s, 3H, O-CH3), 3.55 (m, 2H), 2.93 (t, J=5.7Hz, 2H,
H-6), 2.28-2.21 (m, 2H, H-5).* 13C-NMR (100MHz, CDCl3) δ 166.5 (C=O), 138.3 (C-4), 130.3 (C-3), 51.5
(O-CH3), 44.1 (C-2), 41.9 (C-6), 26.2 (C-5). HRMS (ESI) calcd for C7H12NO2 (M + H+) 142.0863, found
142.0876. *The guvacoline 1H NH signal could not be found.
524
525
526
527
528
3-benzhydryl 1-(9H-fluoren-9-yl)methyl 1,2,5,6-tetrahydropyridine-1,3-dicarboxylate (1b). N-Fmoc-
guvacine (50.0 mg, 0.143 mmol) and diphenylmethyl trichloroacetimidate (61.1 mg, 0.186 mmol)
were dissolved in dry DCM (2 mL) and stirred at 20 °C under argon for 16 h. The product was isolated
via preparative TLC (PE:EE = 3:1; Rf = 0.25-0.56) and formed white crystals (64.3 mg, 87 %).
529
530
1H-NMR (200MHz, CDCl3) δ 7.74 (brs, 2H), 7.58 (brs, 2H), 7.46-7.20 (m, 14H), 6.99 (brs, 1H), 4.52-4.05
(m, 5H), 3.54 (brs, 2H), 2.32 (brs, 2H), 2.06 (s, 1H).
28