Synthesis, structure activity relationship, radiolabeling and preclinical evaluation of high affinity ligands for the ion channel of the N-methyl-d-aspartate receptor as potential imaging probes for positron emission tomography

Article abstract of DOI:10.1016/j.bmc.2014.12.029

The N-methyl-d-aspartate receptor (NMDAr) is involved in many neurological and psychiatric disorders including Alzheimer's disease and schizophrenia. Currently, it is not possible to assess NMDAr availability in vivo. The purpose of this study was to develop a positron emission tomography (PET) ligand for the NMDAr ion channel. A series of di- and tri-N-substituted diarylguanidines was synthesized. In addition, in vitro binding affinity for the NMDAr ion channel in rat forebrain membrane fractions was assessed. Compounds 10, 11 and 32 were radiolabeled with either carbon-11 or fluorine-18. Ligands [¹¹C]10 and [¹⁸F]32 were evaluated ex vivo in B6C3 mice. Biodistribution studies showed higher uptake of [¹¹C]10 and [¹⁸F]32 in forebrain regions compared with cerebellum. In addition, for [¹¹C]10 54% and for [¹⁸F]32 70% of activity in the brain at 60 min was due to intact tracer. Pre-treatment with MK-801 (0.6 mg·kg^-1, ip) slightly decreased uptake in NMDAr-specific regions for [¹⁸F]32, but not for [¹¹C]10. As such [¹⁸F]32 has the best characteristics as a PET tracer for the ion channel of the NMDAr.

Full text of DOI:10.1016/j.bmc.2014.12.029

Accepted Manuscript
Synthesis, structure activity relationship, radiolabeling and preclinical evalua-
tion of high affinity ligands for the ion channel of the N-methyl-D-aspartate
receptor as potential imaging probes for positron emission tomography
Pieter J. Klein, Johannes A.M. Christiaans, Athanasios Metaxas, Robert C.
Schuit, Adriaan A. Lammertsma, Bart N.M. van Berckel, Albert D. Windhorst
PII:
S0968-0896(14)00877-3
DOI:
http://dx.doi.org/10.1016/j.bmc.2014.12.029
BMC 11961
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
23 July 2014
9 December 2014
11 December 2014
Please cite this article as: Klein, P.J., Christiaans, J.A.M., Metaxas, A., Schuit, R.C., Lammertsma, A.A., van
Berckel, B.N.M., Windhorst, A.D., Synthesis, structure activity relationship, radiolabeling and preclinical
evaluation of high affinity ligands for the ion channel of the N-methyl-D-aspartate receptor as potential imaging
probes for positron emission tomography, Bioorganic & Medicinal Chemistry (2014), doi: http://dx.doi.org/
1
0.1016/j.bmc.2014.12.029
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.

Synthesis, structure activity relationship, radiolabeling and preclinical evaluation of
high affinity ligands for the ion channel of the N-methyl-D-aspartate receptor as
potential imaging probes for positron emission tomography
a
Pieter J. Klein *, Johannes A.M. Christiaans , Athanasios Metaxas , Robert C. Schuit , Adriaan A.
a,b
a
a
a
Lammertsma , Bart N.M. van Berckel , Albert D. Windhorst
a
a
a
Department of Radiology & Nuclear Medicine, Neuroscience Campus Amsterdam, VU University
b
Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands. Current address: A*STAR-
NUS, Clinical Imaging Research Centre (CIRC), Centre for Translational Medicine, National
University of Singapore, 14 Medical Drive, Singapore, 117599
Corresponding Author
email: pj.klein@vumc.nl
tel: +31 30 4449101
fax: +31 30 4449121
Keywords
PET; NMDA; Ion channel; Radiolabeling; Uncompetitive antagonists; Guanidines
1

Abstract
The N-Methyl-D-Aspartate receptor (NMDAr) is involved in many neurological and
psychiatric disorders including Alzheimer’s disease and schizophrenia. Currently, it is not
possible to assess NMDAr availability in vivo. The purpose of this study was to develop a
positron emission tomography (PET) ligand for the NMDAr ion channel. A series of di- and
tri-N-substituted diarylguanidines was synthesized. In addition, in vitro binding affinity for
the NMDAr ion channel in rat forebrain membrane fractions was assessed. Compounds 10, 11
1
and 32 were radiolabeled with either carbon-11 or fluorine-18. Ligands [ C]10 and [ F]32
1
18
11
were evaluated ex vivo in B6C3 mice. Biodistribution studies showed higher uptake of [ C]10
1
and [ F]32 in forebrain regions compared with cerebellum. In addition, for [ C]10 54% and
8
11
1
for [ F]32 70% of activity in the brain at 60 min was due to intact tracer. Pre-treatment with
8
-1
18
MK-801 (0.6 mg kg , i.p.) slightly decreased uptake in NMDAr-specific regions for [ F]32,
1
but not for [ C]10. As such [ F]32 has the best characteristics as a PET tracer for the ion
1
18
channel of the NMDAr.
1
Introduction
Glutamate plays an important role in neurotransmission in the brain via metabotropic and
1
ionotropic receptors and it is the principal excitatory neurotransmitter. The N-methyl-D-
aspartate receptor (NMDAr) complex is one of the members of the ionotropic glutamate
2
,3
receptor family. The NMDAr is involved in many neurological and psychiatric disorders
including Alzheimer’s disease and schizophrenia. For instance, it has been proposed that
chronic activation of extrasynaptic NMDAr leads to a sustained neuronal amyloid-βrelease,
4
which could be involved in the pathogenesis of Alzheimer’s disease. At present, it is not
possible to assess NMDAr availability in vivo. Imaging of NMDAr using positron emission
2

tomography (PET) could make this possible but requires the development of a suitable PET
ligand.
5
NMDArs are heterotetrameric complexes of associated subunits. Each NMDA ion channel is
formed by the assembly of an obligatory glycine-binding NR-1 subunit, along with
combinations of distinct glutamate-binding NR-2 (A-D) and/or glycine binding NR-3
6
-8
subunits. Amongst the different binding sites on the NMDAr, those that reside inside the
pore of the ion channel provide the opportunity for quantification of active, ‘open’ receptors.
Non-competitive ion channel NMDAr antagonists, such as (+)-5-methyl-10,11-dihydro-5H-
dibenzo[a,d]cyclohepten-5,10-imine (MK-801) and ketamine, gain access to the NMDAr ion
channel binding site when the receptor is activated by its corresponding agonist glutamate and
9
,10
co-agonist glycine, i.e. these non-competitive antagonists are ‘use-dependent ligands’.
Therefore, their binding is proportional to the amount of activated, open receptor channels.
Thus, using radiolabeled non-competitive antagonists for the NMDAr ion channel and PET,
the status of regional NMDAr activation can be assessed in vivo.
The first study of N,N’-diarylguanidine affinity for the ion channel of the NMDAr complex
1
was reported by Keana. Since then, numerous compounds have been reported with NMDAr
1
ion channel affinity, some of which showed high affinity in the nanomolar range and
selectivity to the ion channel over sigma receptors (Figure 1), such as N-(1-naphthyl)-N’-3-
3
iodophenyl)-N’-methylguanidine (CNS 1261, 1, K: 4.21 nM against [ H]MK-801), which has
i
1
been labeled with either iodine-123 ([ I]1) or iodine-125 ([ I]1) , for single-photon
23
12
125
13
emission computed tomography (SPECT).
N-(2-chloro-5-(thiomethyl)phenyl)-N•-(3-(thiomethyl)phenyl)-N•-methylguanidine
(CNS
3
14,15
161, 2, K: 1.87 nM against [ H]MK-801) and N-(2-chloro-5-(thiomethyl)phenyl)-N•-(3-
5
i
3 16,17
methoxy)phenyl)-N•-methylguanidine (GMOM, 3, K: 5.20 nM against [ H]MK-801) both
(
i
3

labeled with carbon-11 are the most promising NMDAr ion channel pore tracers to date,
1
8
though clinical research applications have had limited success.
Figure 1. Structures of radiolabeled N,N’-diaryl-N-methylguanidines
123
In humans, the highest uptake of [ I]1 was found in thalamus followed by striatum, cortical
1
regions and the lowest in white matter, which correlates with the NMDAr distribution. A
9
123
challenge with ketamine showed that [ I]1 is a specific ligand for the ion channel of the
2
0, 21
123
However, [ I]1 revealed high nonspecific binding,
NMDAr and that it can be displaced.
limiting its potential for quantification as small changes in receptor availability are unlikely to
12
be detected.
Heterogeneous distribution with lowest uptake in cerebellum and highest in putamen and
11
thalamus was found in human using [ C]2. Specific binding, however, has not been
22,23
demonstrated and metabolism was rapid. Ahmed et al found no significant increase in the
11
15
uptake of [ C]2 in patients with Parkinson’s disease compared with control subjects.
Using regional brain distribution and blocking studies in conscious rats, Waterhouse et al
11
11
showed that [ C]3 had a relatively low degree of saturable binding in vivo. In addition, [ C]3
showed fast kinetics with high initial brain uptake followed by rapid washout in all regions
examined. Pretreatment with unlabelled 3 or MK-801 induced a moderate and uniform
4

11
decrease of [ C]3 in most regions. However, no significant differences in regional ratios were
found between blocked and control animals. The glycine-site agonist D-serine increased
11
11
C]3 uptake, while the NR2B modulator RO 25-6981 reduced the uptake of [ C]3 in all
[
regions, except for the striatum. PET studies in isoflurane anesthetized baboons provided a
fairly uniform regional brain distribution, and blocking with MK-801 did not significantly
1
7
alter regional values, indicating a lack of saturable binding.
Although a lot of effort has been put into the synthesis of N,N’diarylguanidines only two
aromatic unsubstituted compounds were synthesized with a methylene spacer between the
3
guanidine moiety and the aromatic ring, 1,3-dibenzylguanidine (IC : 6800 nM against [ H]-N-
50
3 24
l-(2-thienyl)cyclohexyl]piperidine ([ H]TCP)) and 3-benzyl-1-methyl-1-(naphthalen-1-
[
3
yl)guanidine (IC : 2900 nM against [ H]MK-801).
25
50
The aim of this study was to synthesize and evaluate analogues of 3 as potential PET ligands
for the NMDAr ion channel. Previous structure activity relationship (SAR) studies showed
that a meta substituted aryl group (R ) (Figure 2) with small alkyl groups is favorable.
1
Furthermore, on the other aryl group a 2,5-disubstitution pattern (R , R ) indicated
4
5
enhancement of potency, 2-bromo or 2-chloro combined with a 5-thiomethyl group showed
1
1,14,16,25
the highest binding affinity.
Therefore, in this study a series of diarylguanidines was
synthesized according to the structure in Figure 2 with phenyl or 3-methoxyphenyl as R and
1
n = 0-2 methylene groups. A second series of compounds consisted of the first series with a
methyl moiety substituted at the R position. The third library of compounds consisted of
2
analogues of 3 substituted at the 3-methoxy moiety with short fluoroalkoxy chains, in addition
2
compounds of this series were methylated at the R position instead of R . Synthesis,
3 2
binding affinity to the ion channel of the NMDAr, radiolabeling of the high affinity
compounds, LogD value and biodistribution in mice were investigated.
7
.4
5

Figure 2. General structure of N,N’-di- and tri-substituted arylguanidines
2
2
Results
.1 Chemistry
The disubstituted- and trisubstituted guanidines 3-22 (Scheme 1) were obtained via a method
14
adapted from Hu et al by reacting the appropriate cyanamides 24a-j and 25a-j with an amine
hydrochloric salt in chlorobenzene. Products were collected as free base, and fumarate or
hydrochloric acid salts.
The cyanamides 24a-j were prepared from commercially available primary amines 23a-j by
reaction with cyanogen bromide in diethyl ether (29-48% yield). Subsequent alkylation with
methyl iodide in presence of potassium carbonate in dimethylformamide yielded the N-
methylcyanamides 25a-j (62-99%). N-(3-hydroxyphenyl)-N-methylcyanamide (25g),
however, was synthesized using 1 equivalent of both potassium carbonate and methyl iodide
2
6
instead of 1.1 and 2.0 equivalents to prevent O-methylation (65%).
a
Scheme 1 . General synthesis scheme of guanidines
6

a
Reagents and conditions: (a) CNBr, Et O, 0ºC – rt, (29-48%); (b) K CO , CH I, DMF (62-
2
2
3
3
9
9%; (c) Ar’NH ·HCl, chlorobenzene, 140ºC, (25-84%); (d) Ar’NH(CH )·HCl,
2
3
chlorobenzene, 140ºC, (43-59%).
2-Chloro-N-methyl-5-(methylthio)aniline hydrochloride salt (81%) (29) (Scheme 2) was
synthesized from 2-chloro-5-(methylthio)aniline (28), which was prepared in 2 steps by a
2
modified Curtius rearrangement of 2-chloro-(5-thiomethyl)-benzoic acid (26) (72%), with n-
7
28
butyllithium as base and methyl iodide in THF. The t-butylcarbamate cyanamide (30) was
prepared from 27 in 70% yield. The trifluoroacetyl protected aniline (31) was obtained
quantitatively from aniline 28.
a
Scheme 2 . Synthesis of 2-chloro-5-(methylthio)aniline derivates.
7

a
Reagents and conditions: (a) Et N, diphenyl phosphorazidate, t-BuOH, reflux; (b) HCl,
3
water, THF, reflux (72%, 2 steps, from 26); (c) n-BuLi 1.6M in hexanes, CH I, THF, 0ºC - rt
3
(
81%); (d) CNBr, NaH, THF, 0ºC – rt, (70%); (e) TFAA, Et N, CH Cl , 0ºC – rt, (94%).
3 2 2
The N,N’-disubstituted guanidines 13-20 (Scheme 1) were obtained by reaction of cyanamide
4a-j with 28 (25-84%). The N,N’,N’-trisubstituted guanidines 21-22 (Scheme 1) were
2
obtained from the reaction of the N-methylcyanamide 24h,i with 29 (43-59%). Finally the
reaction between N-methylcyanamide 25a-j and 28 provided the N,N,N’-trisubstituted
guanidines 3-12 (36-78%) (Scheme 1). N,N’,N’-Trisubstituted guanidines 32-34 were
prepared by alkylation of 9 with appropriate bromo-fluoroalkane in 61-73% yield (Scheme 3).
a
Scheme 3 . Synthesis of fluoroalkoxy analogues
8

a
Reagents and conditions: (a) 32: BrCH F 2M in DMF, NaH, NaI, dimethoxyethane, rt (68%);
2
3
3: Br(CH ) F, K CO , KI, DMF, 75ºC (61%); 34: Br(CH ) F, K CO , KI, DMF, 75ºC (73%).
2 2 2 3 2 3 2 3
2.2 In vitro binding studies
The binding affinity of the ligands (3-22, 32-34) for the NMDAr channel was determined by
measuring the ability of various concentrations of unlabeled ligand to inhibit specific binding
3
of [ H]MK-801 to rat forebrain membranes (P2 fraction). Results are shown in Table 1, with
K (mean ± SD) values of 2-6 independent determinations, each conducted in triplicate.
i
3
Table 1. Binding affinity to the ion channel of the NMDAr against [ H]MK-801
R = H, R = H
R = CH , R = H
R = H, R = CH
2 3 3
2
3
2
3
3
a
a
a
R1
H
H
n
0
1
no. K (nM)
no. K (nM)
no. K (nM)
i
i
i
b
b
13
722 ± 166
> 10 µM
4
307 ± 130
> 10 µM
c
c
5
14
9

c
b
d
c
c
b
d
c
c
H
2
0
1
2
0
0
0
0
0
0
0
15
16
17
18
> 10 mM
136 ± 2.95
> 1 µM
6
3
7
8
9
> 10 mM
21.7 ± 7.52
> 1 µM
OCH3
OCH3
OCH3
OH
> 10 µM
> 10 µM
551 ± 100
18.4 ± 8.11
10.2 ± 3.08
11.7 ± 4.16
155 ± 59.0
179 ± 66.9
56.7 ± 10.8
b
b
c
OCH2F
OCHF2
OCF3
O(CH ) F
32
10
11
33
34
12
c
c
b
c
19
20
147 ± 25.3
235 ± 96.2
21
22
322 ± 33.7
650 ± 174
b
b
b
2
2
O(CH ) F
2
3
OCF CHF
2
2
a
3
K measured against 5 nM [ H]MK-801 presented as mean ± SD of 2-6 independent
i
b
c
d
determinations, each conducted in triplicate, fumaric acid salt, free base, hydrochloric acid
salt.
2.3 Radiochemistry
11
11
The N,N-[ C]methyl-,N’-tri substituted guanidine [ C]10 (Scheme 4) was obtained through
1
alkylation of N,N’-diarylguanidine 19 with [ C]CH I although the unwanted side product
1
3
1
N,N’-[ C]methyl-,N’-tri substituted guanidine ([ C]21) was formed as well. Several base
1
11
11
solvent combinations were explored (Table 2). Finally, [ C]10 was synthesized starting from
the precursor (19) in the presence of 5M sodium hydroxide in N,N-dimethylformamide (Entry
2, Table 2). After preparative HPLC purification (Figure S1, Supp. Info.), and formulating
11
into a sterile solution, [ C]10 was obtained in 26 ± 5% (n=4) radiochemical yield, (decay
1
1
corrected (DC), calculated from [ C]CH I) and a radiochemical purity higher than 99.7%.
3
-1
Specific activity was 170 ± 73.4 GBq·µmol at end of synthesis (EOS) and total synthesis
time, from end of bombardment (EOB), 38 min.
10

a
Scheme 4 . Radiosynthesis of [ C]10 and [ C]11 using [ C]CH3I
11
11
11
a
11
Reagents and conditions: (a) [ C]CH I, 5M NaOH, DMF, 80ºC.
3
1
Compound [ C]11 was synthesized in the same manner as [ C]10 (Table 2, entry 10) starting
1
11
1
1
from 20 in 15 ± 8% (n=2) radiochemical yield, (DC, calculated from [ C]CH I) and a
3
radiochemical purity higher than 98.7% (System B). The specific activity was 226 ± 141
-1
GBq·µmol (EOS).
Table 2. Radiochemical yields of carbon-11 labeled compounds under various conditions at
11
8
0 ºC (calculated from [ C]CH I))
3
a
b
c
d
d
e
Entry Prec.
Base
µmol base Solvent
RCY
RCY
Ratio
1
1
11
[ C]21
[
C]10
1
2
3
4
19
19
19
19
NaH
50
25
5
MeCN
DMF
48.3 ± 3.6 15.2 ± 0.1 3.2
49.2 ± 0.6 23.8 ± 0.0 2.1
NaOH
TBAOH
KOH
DMF
9.4 ± 1.6
1.0 ± 0.4 9.4
178
DMSO
54.8 ± 1.0 25.0 ± 3.0 2.2
11

5
6
7
8
19
19
19
19
TBAOH 25
TBAOH 10
DMF
3.0 ± 0.2
3.1 ± 1.3
0.8
0.4 ± 0.1 7.5
0.1 ± 0.1 31.0
DMF
f
f
TBAOH
NaOH
5
DMSO
DMSO
0.0
2.5
-
25
28.6
11.4
1
1
11
[ C]22
[
C]11
9
1
1
20
20
20
NaH
50
MeCN
DMF
37.9 ± 1.3 16.0 ± 1.3 2.4
42.0 ± 0.4 18.7 ± 1.0 2.2
47.4 ± 2.6 15.3 ± 1.5 3.1
0
NaOH
25
1
KOH
178
DMSO
a
b
Precursor, (1.4 µmol), NaH (60% dispersion in mineral oil), NaOH (5M), TBAOH (55%
c
d
solution), KOH (powdered), 300 µL, Radiochemical Yield (RCY) determined by HPLC
e
using system B in % ± SD (n=2), ratio = product divided by side product, n=1
f
1
1
11
11
To prevent formation of the [ C]methylated side products ([ C]21 and [ C]22), an attempt
was made to synthesize the N,N’-protected-,N’ precursors 35, 36 (Scheme 5, Table S1), but
this approach was unsuccessful.
a
Scheme 5 . Attempted synthesis of protected precursors 35 and 36
a
Reagents and conditions: (a) See Table S1 (Supporting information)
12

18
29
F]Fluoromethylbromide was prepared via the method of Iwata et al and subsequently
[
reacted with 9 in the presence of sodium hydride and potassium iodide in N,N-
dimethylformamide (Scheme 6). After HPLC purification (Figure S2, Supp. Info., System E)
18
and formulation into a sterile solution, [ F]32 was obtained in a maximum of 22%
18
-
radiochemical yield (DC, calculated from [ F]F ) and radiochemical purity higher than 98%
-1
(Figure S2, Supp. Info., System D). Specific activity was 52.7 ± 34.1 GBq·µmol (EOS) and
18
total synthesis time, from EOB, 92 min. Radiochemical yield of [ F]32 via the intermediate
18
[
F]fluoromethyltriflate was 22% (DC).
a 18
Scheme 6 . Radiosynthesis of [ F]32
a
18 18
Reagents and conditions: (a) [ F]FCH Br, NaH, KI, DMF, 100 ºC or [ F]FCH OTf, NaH,
2
2
DMF, 100 ºC.
2.4 Determination of lipophilicity expressed as LogDoct,7.4 of 3, 10, 11 and 32
1
The distribution coefficients of 3, 10, 11 and 32 were measured using [ C]3, [ C]10, [ C]11
1
11
11
1
and [ F]32. The LogDoct,7.4 values are shown in table 3.
8
30
Table 3. Measured LogD
13

a
Compound LogDoct,7.4
11
11
11
18
[
[
[
[
C]3
1.72 ± 0.01
1.76 ± 0.01
2.76 ± 0.01
1.45 ± 0.02
C]10
C]11
F]32
a
n=3, presented as mean ± SD.
2.5 Metabolite analysis
1
1
18
Metabolite analysis of [ C]10 and [ F]32 were determined in B6C3 mice at the following
time points: 5, 15, 30, 60 min using 3 or 4 animals per time point. Table 4 shows the
distribution of radioactivity amounts between polar and non-polar fraction. The non-polar
fraction is divided into three fractions, parent compound, metabolites and remaining activity
on the solid phase extraction (SPE) column.
Table 4. Metabolite analysis in % (SD)
a
Compound
Time
n
Polar
fraction
Non-polar fraction
b
Parent
Metabolite SPE
11
[
C]10
Brain
5
3
3
4
4
4
4
4
3 (2)
74 (28) 20 (21)
73 (31) 21 (23)
63 (20) 24 (14)
54 (19) 29 (10)
39 (20) 43 (15)
4 (5)
3 (3)
6 (9)
6 (6)
3 (2)
2 (1)
2 (1)
1
3
6
5
1
3
5
0
0
4 (4)
7 (3)
11 (5)
17 (4)
Plasma
5
0
29 (11) 17 (3)
25 (9) 14 (7)
53 (13)
61 (6)
14

6
0
4
34 (7)
7 (2)
59 (6)
8 (9)
2 (0)
18
[
F]32
Brain
5
1
3
6
5
1
3
6
3
4
3
4
4
3
4
3
3 (2)
2 (1)
3 (3)
4 (1)
7 (6)
20 (2)
86 (9)
2 (2)
4 (4)
3 (1)
1 (1)
7 (8)
2 (1)
5 (6)
1 (0)
5
0
0
72 (17) 24 (20)
76 (15) 18 (12)
70 (16) 25 (14)
Plasma
38 (6)
25 (6)
48 (11)
54 (8)
5
0
16 (13) 21 (14) 59 (14)
29 (0) 9 (4) 61 (3)
0
a
b
time in min. The remaining measured radio activity on the solid phase extraction (SPE)
column.
2.6 Ex vivo biodistribution
1
Biodistribution and regional brain uptake of [ C]10 and [ F]32 were determined in
1
18
anaesthetized B6C3 mice at the following time points: 5, 15, 30, 60 min.
1
Brain uptake of [ C]10 was highest at 5 min when compared with all other time points
1
(Figure 3; p<0.001, Least Significant Difference (LSD) posttests). Standardized uptake values
(SUV) at 5 min were 0.70 ± 0.09 for total brain and 1.41 ± 0.46, 0.83 ± 0.11, 0.86 ± 0.13,
0
.96 ± 0.16 and 0.55 ± 0.07, for hippocampus, cerebral cortex, striatum, prefrontal cortex and
cerebellum, respectively. Total brain uptake was 0.32 ± 0.11, 0.27 ± 0.11 and 0.14 ± 0.07 at
5, 30 and 60 min. The highest ratio of radioactivity uptake between forebrain regions and
1
cerebellum was observed at 5 min post injection. Radioactivity was cleared rapidly from the
brain and no regional differences in uptake were observed 15 min after tracer injection. Two
way Analysis of Variance (ANOVA) confirmed significant main effects of region [F(4,15)=4.8,
11
p<0.01] and time [F(3,45)=11.6, p<0.001] on brain uptake of [ C]10.
15

1
Organ uptake of [ C]10 was higher in lungs and kidney than in other organs (Figure 3;
1
1
p<0.001, LSD posttests). The highest [ C]10 uptake was measured in the lungs, 5 min
1
following tracer injection (p<0.001; LSD posttests), whereas the lowest levels of radioactivity
11
were observed in blood. Organ uptake of [ C]10 was higher at 5 min post injection than at the
11
other time points measured (p<0.001; LSD posttests). Significantly higher levels of [ C]10
uptake were observed in the kidney 60 min post-injection when compared with all other
organs (p<0.01; LSD posttests). Two way ANOVA confirmed significant main effects of
1
organ [F(4,12)=130.1, p<0.001] and time [F(3,36)=24.7, p<0.001] on uptake of [ C]10, as well as
1
significant organ x time interaction effects [F(12,365)=8.3, p<0.001].
1
Figure 3. Biodistribution of [ C]10 in the CNS (left) and selected organs (right).
1
18
Brain uptake of [ F]32 was highest at 5 min when compared with all other time points (Figure
4
0
; p<0.01, LSD posttests). SUV values at 5 min were 0.84 ± 0.12 for total brain and 1.05 ±
.09, 1.17 ± 0.07, 1.19 ± 0.15, 1.19 ± 0.10 and 0.84 ± 0.09 for hippocampus, cerebral cortex,
striatum, prefrontal cortex and cerebellum, respectively. Total brain uptake was 0.59 ± 0.04,
18
0.53 ± 0.07 and 0.23 ± 0.03 at 15, 30 and 60 min. [ F]32 uptake was lower in cerebellum than
16

in all other brain areas analyzed (p<0.01, LSD posttests). The highest ratio of radioactivity
uptake between forebrain regions and cerebellum was observed at 15 min post injection.
Radioactivity cleared rapidly from the brain and no regional differences in uptake were
observed 60 min after tracer injection. Two way ANOVA confirmed significant main effects
18
of region [F(4,14)=4.3, p<0.01] and time [F(3,42)=49.9, p<0.001] on brain uptake of [ F]32.
1
Organ uptake of [ F]32 was higher in lungs and kidney than in other organs (Figure 4;
8
18
p<0.001, LSD posttests). The highest [ F]32 uptake was measured in the lungs, 5 min
following tracer injection (p<0.001; LSD posttests), whereas the lowest levels of radioactivity
1
8
were observed in blood. Organ uptake of [ F]32 was higher at 5 min post injection than at
other time points (p<0.001; LSD posttests). Activity was rapidly cleared from the lungs, and
1
8
only the kidneys showed significantly higher levels of [ F]32 60 min post injection than other
organs (p<0.01; LSD posttests). Two way ANOVA confirmed significant main effects of
organ [F(4,15)=51.1, p<0.001] and time [F(3,45)=17.1, p<0.001], as well as significant organ x time
1
interaction effects [F(12,45)=12.1, p<0.001] on uptake of [ F]32.
8
1
Figure 4. Biodistribution of [ F]32 in the CNS (left) and selected organs (right).
8
17

2.7 Ex vivo autoradiography
Non specific binding was defined as activity that remained following pretreatment with MK-
-1
11
01 (0.6 mg·kg , intraperitoneal injection). For [ C]10, pretreatment resulted in an increased
8
1
1
t-test, t(8)=6.3, P<0.001) uptake of [ C]10 in striatum, frontal cortex, hippocampus, cerebral
(
18
cortex and cerebellum (Figures 5 & S3, Supp. Info.). For [ F]32, pretreatment resulted in a
1
decreased (t-test, t(8)=4.7, P<0.01) uptake of [ F]32 in hippocampus, striatum, cerebral
8
cortex, cerebellum and frontal cortex (Figures 5 & S4, Supp. Info.).
Figure 5. Relative optical density (ROD) values of several brain areas in control and after
-1
11
pretreatment with 0.6 mg·kg MK-801. On the left uptake of [ C]10, on the right uptake of
18
11 18
F]32, Data represent mean ± SEM of 2 ([ C]10) or 3 ([ F]32) determinations.
[
3
Discussion
In this study binding affinities to the NMDAr of a series of compounds, based on lead
compound 3, were investigated. Compounds with high affinity were labeled with a positron
18

emitter and subsequently evaluated preclinically as NMDAr uncompetitive antagonists.
Waterhouse et al reported previously, for 3, a binding affinity for the NMDAr of 5.20 ± 0.3
3
nM , which is different from the present value of 21.7 ± 7.52 nM. This difference in K values
1
i
between the two studies could be due to differences in animals used (Sprague-Dawley rats vs.
Wistar rats), differences in membrane preparation procedures (total membrane fraction vs. P2
fraction used here), different binding properties of NMDArs in the different brain regions
used for obtaining membranes (frontal cortex vs. total forebrain used here) or even different
assay conditions (i.e. time of incubation). The most striking difference between the two
studies, however, is the amount of glutamate and glycine used in the assays. In the present
study 1µM of glutamate and glycine was used to supplement basal binding conditions, and to
31
obtain a known and constant concentration of these modulators. Waterhouse et al used
0mM of glutamate and glycine, which represents supra-saturating conditions. As the binding
site within the channel of the NMDAr is only accessible when the receptor is activated by its
1
9,10
corresponding agonists, glutamate and glycine , the difference in K values between the two
i
studies may be due to the amount of open channel receptors, induced by the use of different
concentrations of glutamate and glycine in the two assays.
To confirm the in vitro binding result, 3 was screened commercially at Cerep (www.cerep.fr).
The result obtained in the presence of 100 µM glutamate (no glycine) was 15 nM, and ties in
with the value presented in this study.
3.1 Chemistry
The guanidines (3-22) were synthesized by reaction of the corresponding N-cyanamides (24a-
j) or N-methylcyanamides (25a-j) with either the hydrochloric acid salt of 2-chloro-5-
(methylthio)aniline (28) or 2-chloro-N-methyl-5-(methylthio)aniline (29) in yields
19

14,24,25
comparable with those reported in literature.
Starting cyanamides were prepared by the
reaction of the appropriate primary amines with cyanogenbromide in diethyl ether in yields of
up to 48%. The short fluoroalkyl guanidines (32-34) were synthesized from the phenolic
guanidine 9 by alkylation with F(CH ) Br (n=1-3) with potassium carbonate and potassium
2
n
iodide in DMF. Compounds 33 and 34 were obtained in a yield of 61 and 73%, respectively.
For the synthesis of the fluoromethoxy analogue (32), however, this method was not
applicable. Using sodium hydride and sodium iodide in dimethoxyethane gave 68% yield.
3.2 Structure Activity Relationships
Since only two compounds with a methylene spacer between the aromatic group (Figure 2)
2
4,25
and the guanidine moiety have been described , little is known about SARs within this class
of compounds. Introduction of one (14, 17, 5, 7) or two methylene groups (15, 18, 6, 8),
reduced the binding affinity for the ion channel of the NMDAr dramatically with an order of
2
0 to more than 10 compared with 13, 16, 4, 3, indicating that elongation of the chain length
6
1
is not tolerated for NMDAr binding.
Although the N-methyl group in the guanidine structure is present in all high affinity
14,25
compounds reported , one cannot conclude from SAR studies that this moiety is responsible
for the higher affinity for the ion channel. Nevertheless, a better selectivity over the sigma
14,25
receptor compared with non-methylated guanidines has been reported.
Introducing a N-methyl group in compounds 13, 16, 19 and 20 increased binding affinities 2,
6, 14 and 20 fold for 4, 3, 10 and 11, respectively. A combination of the 3-methoxy moiety
and the N-methyl group led to an increase in affinity to 21.7 ± 7.52 nM for 3 compared with
13 (722 ± 166 nM). In contrast, when methylated on the N’-guanidine position next to the 2-
20

chloro-5-(methylthio)phenyl ring, the affinity of compounds 21 and 22 are decreased 2 to 3
fold compared with 19 and 20, respectively. This indicates that N-methylation leads towards
improved affinity but N’-methylation is not accepted.
11
25
The need of a small alkyl group at the 3-position as described by Keana et al , Reddy et al ,
1
Hu et al and Dumont et al is supported by the results of R substitution of the aryl group
4
16
1
with 3-OH or 3-methoxy. The binding affinity of the phenol 9 decreased approximately 2-fold
compared with aryl 4 (307 ± 130 nM), while the methoxy analogue 3 showed a 14 fold
increase of potency.
Substitution of hydrogen by fluorine may have a positive effect on physicochemical
32
properties of ligands, such as binding affinity and CNS penetration. In addition, the presence
of a fluorine atom gives the opportunity of radiolabeling with fluorine-18 for use with PET, as
exchange of fluorine-19 by fluorine-18 would not affect the physicochemical properties.
However, this is only possible if the fluorine position is amenable to radiolabeling with
fluorine-18.
Introduction of a monofluoromethoxy group in 32 (18.4 nM) was tolerated as the affinity
towards the NMDAr was not significantly altered compared with 2 (21.7). Additionally,
substitution of one or two further hydrogen atoms by fluorine led to an increase to 10.2 nM
for the difluoromethoxy guanidine (10) and 11.7 nM for the trifluoromethoxy guanidine (11),
roughly a 2 fold increase in affinity. Substitution of the fluoromethoxy moiety of ligand 32
with 1-fluoroethoxy or 1-fluoropropoxy reduced binding affinity 8 fold for compound 33 (155
±
59.0 nM) and 10-fold for 34 (179 ± 66.9 nM). Given the improved affinity of the
trifluoromethoxy compared with the monofluoromethoxy compound, the 1,1,2,2-
tetrafluoroethoxy analogue (12) was also synthesized. Compared with the 1-fluoroethoxy
ligand (33) the affinity increased almost 3-fold to 56.7 ± 10.8 nM for 12. These results
21

suggest that substitution of the alkoxy moiety with one or more fluorine atoms is beneficial
for the binding affinity towards the NMDAr.
1
1
11
.3 Synthesis of [ C]10 and [ C]11
3
1
The carbon-11 labeling agent [ C]CH I was prepared by the reaction of cyclotron produced
1
3
11
[
C]CO with lithiumaluminiumhydride in THF, which subsequently was reacted with
2
11
hydroiodic acid in radiochemical yields of up to 90% (DC). After [ C]methylation of 19, two
1
1
11
products were obtained, the desired [ C]10 as the main product and the side product [ C]21.
Several base and solvent combinations were explored (Table 2, Entries 1-4). Three
11
combinations resulted in radiolabeling yields between 48 and 55% for [ C]10 and a ratio
between main product and side product of 2.1 to 3.2. The use of t-butylammonium hydroxide
(TBAOH) in combination with DMF lowered the yield (9%), although a highly increased
ratio of 9.8 was observed. Given this ratio and the highest radiolabeling yields obtained with
DMF and DMSO as solvents, the combination with TBAOH in different concentrations
(Table 2, Entries 5-7) was carried out. Unfortunately, yields dropped to 3% or less, although
even with a better ratio. Using DMSO and potassium hydroxide the highest yield (55%) was
obtained, while using sodium hydroxide as base resulted in a yield of 29%. Modifications of
reaction temperature and time were not investigated, as the yield of the radiolabeling was
sufficient for preclinical evaluation.
11
11
Due to the isomeric products [ C]10 and [ C]21, semi-preparative HPLC purification is
delicate. However, using a Platinum C18-EPS semi-preparative HPLC column, both isomers
could easily be separated with a retention time of 6 min for the desired product and 21 min for
the side product (Figure S1, Supp. Info.).
22

1
During optimization of the synthesis of [ C]11 starting from 20, the same pattern of labeling
1
11
11
yield and formation of side product [ C]22 was observed. [ C]11 was obtained in yields of
38-47% (Table 2, Entries 9-11).
1
1
11
To prevent the formation of [ C]21 and [ C]22, an attempt was made to synthesize the N’-
protected precursors 35 and 36 (Scheme 5). First, synthesis of the tert-butyloxycarbonyl (t-
boc) protected precursor (35) was attempted, but this failed due to the acidic conditions during
the condensation reaction of the guanidine. HCl in the reaction mixture eliminates the t-boc
protective group (Table S1, Entries 1-2). Therefore, subsequent reactions were carried out
with the free base. Guanidine formation in chlorobenzene had no success, using hexa-fluoro
3
3,34
isopropanol (HFIP) 35 was obtained in 1% yield (Table S1, Entry 3). Also, the reaction of
t-boc (27) or trifluoroacetyl protected (31) aniline with cyanamide 24h in either
chlorobenzene or HFIP did not yield the desired N,N’-protected-,N’ precursors 35 and 36
11
11
Table S1, Entries 4-6). Therefore, the synthesis of [ C]10 and [ C]11 was carried out with
(
the unprotected precursors 19 and 20.
1
8
3
.4 Synthesis of [ F]32
1
The synthesis of [ F]32 was carried via a two-step synthesis starting from [ F]fluorine. First,
8
18
18
[
F]fluoromethylbromide was prepared, distilled and purified according to the method of
29
Iwata et al in comparable yields of 37-74% (DC). Secondly, it was reacted with the phenol
precursor 9 in DMF with potassium carbonate and potassium iodide as catalysts, in yields up
1
to 8% (DC) overall of [ F]32. Changing the base to sodium hydride, however, improved the
8
18
yield of [ F]32 to 22% (DC) overall. Iwata et al also reported an improvement of alkylation
23

18
when the intermediate [ F]fluoromethylbromide was converted online to
18
29
F]fluoromethyltriflate via a heated AgOTf column. This method was also investigated, but
[
1
yields were not higher than those using [ F]fluoromethylbromide.
8
3.5 Distribution coefficients
Measured LogDoct,7.4 values of 3, 10, 11 and 32 are shown in Table 3. According to
35,36
Waterhouse and Pike only 11 (2.76) is within the optimum range of 2.0 - 3.5. However,
1
both [ C]10 and [ F]32 enter the brain (Figures 3 & 4), with LogD values of 1.76 and 1.45,
1
18
respectively.
3.6 Biological evaluation
Given the comparable affinity of 10 and 11 (10.2 and 11.7 nM respectively), only the carbon-
11
11 labeled compound [ C]10 was evaluated in B6C3 mice with respect to metabolite
1
formation and biodistribution in addition to the fluorine-18 labeled ligand [ F]32.
8
1
1
18
.7 Metabolites of [ C]10 and [ F]32
3
For PET it is important to consider metabolism of a radiolabeled compound in vivo. The
presence of radiolabeled metabolites may hamper quantification of PET images in humans
since PET cannot distinguish between signals from parent radiolabeled ligand and its
3
7
radiolabeled metabolites.
After 60 minutes, 54 ± 19% of the measured radioactivity in the brain was due to parent
11
compound ([ C]10), whilst the remaining radioactivity accounted for polar (11 ± 5%) and
24

non-polar (29 ± 10%) metabolites. In plasma, however, only a parent fraction of 7 ± 2% was
measured, 34 ± 7% was due to polar metabolites and 59 ± 6% to non-polar metabolites.
HPLC analysis revealed three radiolabeled non-polar metabolites in the brain, which were
more polar than the parent compound. Most likely, these metabolites are the result of cleavage
11
11
of the S-methyl or CHF moiety. Cleavage of the [ C]N-methyl moiety of [ C]10 could lead
2
to polar metabolites, most likely representing single carbon molecules such as
11
11 11
C]formaldehyde, [ C]formic acid or [ C]carbon dioxide and an unlabelled metabolite of 10.
[
18
The parent compound [ F]32 accounted for 70 ± 6% of the measured radioactivity in the
brain after 60 minutes, with 25 ± 14% non-polar and 4 ± 1% polar metabolites, respectively.
On the other hand, the parent fraction in plasma was only 9 ± 4%, whereas 61 ± 3% and 29 ±
0% was due to non-polar and polar metabolites, respectively.
18
The three radiolabeled non-polar metabolites of [ F]32, measured using HPLC, were more
1
polar than the parent compound [ F]32 and could be the result of cleavage of the N-methyl or
8
18
S-methyl moieties. Polar metabolites are probably due to cleavage of the [ F]CH F moiety.. It
2
has been reported that substitution of the fluoromethoxy moiety by a fluorodideuteromethoxy
38
group can prevent cleavage of this moiety. This needs to be investigated in future studies.
18
Defluorination, giving [ F]fluoride ion which binds rapidly to bone, is not observed as the
activity in the skull remains the same over time.
For quantitative PET imaging the amount of intact parent compound is the most important
factor. Radiolabeled metabolites are nearly always present in plasma after tracer
administration, and their presence can be accounted for by obtaining plasma samples and
quantifying the parent fraction. Radiolabeled metabolites present in the brain, on the other
hand, may prevent the possibility to model the tracer and to derive quantitative parameters.
25

If the rate and profile of metabolite formation remains identical between rodents and humans,
then our results in mice might preclude the possibility to successfully model the uptake of
18
[
F]32 in subsequent human PET studies. It should be noted, however, that extrapolating
profiles of metabolite formation across species is likely to be tenuous, and that there are well-
documented differences, both in the rate and in the species of metabolite formation between
36,39
rodents and humans . Hence, experiments in higher species are required to unequivocally
1
clarify whether the metabolic profile of [ F]32 will preclude the use of the radiotracer for
8
quantifying NMDAr activation using PET.
The range of standard deviations in these types of experiments can vary a lot. This can be
caused by a combination of factors such as in vivo variation, work-up procedures and time
pressure due to the short half-life time of carbon-11 (20 min) or fluorine-18 (110 min). The
remaining measured radio activity on the solid phase extraction (SPE) column is typically
assumed to be parent compound as it is the most apolar compound.
3.8 Biodistribution and blocking
For both radiolabeled ligands, regional uptake patterns are in line with the distribution of
40,41
NMDAr in the brain. The highest uptake was observed in hippocampus and the lowest in
cerebellum, with moderate levels in cerebral cortex, striatum and frontal cortex.
1
1
18
Ex vivo autoradiography was performed at 5 minutes for [ C]10, and at 15 minutes for [ F]32
post injection as the biodistribution showed the highest uptake in the hippocampus for both
compounds (Figure S3 & S4). Data from rodent studies indicate that 50% occupancy of the
43,44
NMDAr ion-channel by MK-801 is obtained at a dose of 0.18 mg/kg, in vivo.
In this study
26

-1
we used 0.6 mg·kg MK-801, which is at least three times the in vivo EC to cover the
5
0
majority of NMDArs and give us a specific signal.
11
It showed that uptake of [ C]10 was increased by pretreatment of MK-801. The reason for
this is unclear, and we have not further investigated it.
1
8
In contrast, ex vivo autoradiography suggests that the uptake of [ F]32 is selective for the
NMDAr, as uptake decreased following pretreatment with a pharmacological dose of MK-
11
18
01. Though the structural difference between [ C]10 and [ F]32 is only one fluorine atom, it
8
seems that this can have strong effects on the binding characteristics of a ligand.
1
8
11
One-to-one comparison of [ F]32 with [ C]3 in non-human primates would be the next step.
However, as reviewed by Waterhouse et al one should take care of using non-human primates
4
for studying the NMDA receptor as it is vulnerable for anesthetics. In that view PET
5
imaging in human should be the next step of development of this new tracer.
Conclusion
Structural modification of 3 led to three new ligands (10, 11 and 32) with increased affinity
11
18
for the ion channel of the NMDA receptor. Ligands [ C]10 and [ F]32 were evaluated by ex
vivo biodistribution in mice and showed rapid metabolism in plasma, however, in the brain
much less metabolites were observed. After pretreatment with a relatively low dose of the
1
selective NMDAr antagonist MK-801, [ F]32 showed slightly reduced brain uptake. These
8
1
results suggest that [ F]32 has promise as a PET tracer for the NMDA receptor.
8
27

Experimentals
Chemistry
Reagents and solvents were supplied by Aldrich Chemical Co., Fluka, ABCR, Acros or
Biosolve BV, and used as received unless stated otherwise. Nuclear magnetic resonance
1
NMR) spectra ( H, C, F NMR) were recorded on a Bruker Avance 250 (250.13, 62.90,
13
19
(
235.36 MHz, respectively). Chemical shifts of the NMR spectra are reported in parts per
million (ppm) relative to the solvent residual peak. Description of signals: s = singlet, bs =
broad singlet, d = doublet, t = triplet, q = quartet, m = multiplet, dd = double doublet, ddd =
doublet of dd, dt = doublet of triplets, tt = triplet of triplets, dq = double quartet. Thin-layer
chromatography (TLC) was performed on Merck DC-alufolien, silica gel 60, F254. Flash
column chromatography was performed on silica gel 60•, 23 00 – 400 Mesh. The HPLC
analysis system consisted of a Jasco PU-2089 HPLC pump, a Rheodyne injector with a 20 µL
loop, a Jasco UV-2075 Plus UV detector set at a wavelength of 254 nm and a Raytest Na(I)
radioactivity detector. HPLC data were collected and integrated using the software package
GINA 5.01. HPLC: System A: Alltech Platinum EPS, C18 100A 5µ, 250 x 4.6 mm column
using 50 mM NH H PO pH=2.5 / methanol (30/70, v/v) as eluent and a flow rate of 1
4
2
4
-1
mL•min ; System B: Waters µBondapak, C18 10µm 125A, 250 x 4.6 mm column using 10
-1
mM NH HCO pH=7.4 / acetonitrile (30/70, v/v) as eluent and a flow rate of 1.5 mL•min ;
4
3
System C: Alltech Platinum EPS, C18 100A 5µ 150 x 10.0 mm column using 10 mM
-1
NH HCO pH=7.4 / acetonitrile (10/90, v/v) as eluent and a flow rate of 5 mL•m ii n ; System
4
3
D: Alltech GraceSmart C18, 5µ, 250 x 4.6 mm column using 10 mM NH H PO pH=2.5 /
4
2
4
-1
acetonitrile (75/25, v/v) as eluent and a flow rate of 1.5 mL•min ; System E: Alltech Alltima
C18, 5µ, 250 x 10.0 mm column using 25 mM NH H PO pH=2.5 / acetonitrile (72/28, v/v) as
4
2
4
-1
eluent and a flow rate of 5.0 mL•min ; System F: Dionex Ultimate 3000 HPLC system,
28

equipped with a Phenomenex Gemini C18 5µ 250 x 10.0 mm column using a gradient of 50
mM NH H PO / acetonitrile (80/20 to 30/70, v/v) in 12.5 minutes at a flow rate of 3.0
4
2
4
-1
mL•mi nn . For high-resolution mass spectrometry (HRMS), a Finnigan MAT 90 was used.
11
Mass spectra (MS) were measured on an AB Sciex Qtrap 5500. [ C]CO was produced by the
2
14
11
18
−
18
18
N(p,α) C nuclear reaction and [ F]F was produced by the O(p,n) F nuclear reaction using
an IBA Cyclone 18/9 cyclotron. Radioactivity levels were measured using a Veenstra VDC-
405 dose calibrator. Radiochemistry was carried out in homemade, remotely controlled
4
2
devices.
Animals
For in vivo studies, B6C3F1/J mice were obtained from the Jackson Laboratory (USA) and
were housed in groups of four to six per cage until treatment. Animals were kept at a constant
temperature of 21 °C and under a 12-h light/dark cycle, in which lights were switched on at
8:00 a.m. Animals had unrestricted access to food (Teklad Global 16% Protein Rodent Diet,
Harlan, Madison, WI, USA) and water. All animal experiments were performed in
compliance with Dutch laws on animal experimentation and after approval by the local animal
ethics committee, study number DEC_PET12-01.
General Procedure for N-Cyanation (A). A solution of cyanic bromide [Caution, highly
toxic] (2 equiv.) in diethyl ether (10 mL) was added dropwise to a solution of the
appropriately substituted aniline (1 equiv.) in diethyl ether (10 mL) at 0 ºC . After complete
addition the mixture was warmed to ambient temperature and stirred 1-20 hours, which was
followed by TLC. Solids were filtrated and washed with ether. The filtrate was washed with
29