Structure-selectivity investigations of D2-like receptor ligands by CoMFA and CoMSIA guiding the discovery of D3 selective PET radioligands

Article abstract of DOI:10.1021/jm0611152

Elucidation of the physiological role of the D₃ receptor and its distribution in the brain using positron emission tomography (PET) is hampered by the lack of bioavailable subtype selective tracer ligands. To develop appropriate D₃ r

Full text of DOI:10.1021/jm0611152

J. Med. Chem. 2007, 50, 489-500
489
Structure-Selectivity Investigations of D2-Like Receptor Ligands by CoMFA and CoMSIA
Guiding the Discovery of D3 Selective PET Radioligands
Ismail Salama,† Carsten Hocke,‡ Wolfgang Utz,† Olaf Prante,‡ Frank Boeckler,† Harald Hu¨bner,† Torsten Kuwert,‡ and
Peter Gmeiner*,†
Department of Medicinal Chemistry, Emil Fischer Center, Friedrich Alexander UniVersity, Schuhstrasse 19, D-91052 Erlangen, Germany, and
Department of Nuclear Medicine, Krankenhausstrasse 12, D-91054 Erlangen, Germany
ReceiVed September 26, 2006
Elucidation of the physiological role of the D3 receptor and its distribution in the brain using positron emission
tomography (PET) is hampered by the lack of bioavailable subtype selective tracer ligands. To develop
appropriate D3 radioligands, we designed an integrative procedure involving the elucidation of structural
features determining D3 selectivity over both congeners D2 and D4 by comparative molecular analysis. Thus,
we have successfully generated CoMFA and CoMSIA models based on the affinitiy differences of a series
of 79 ligands representing a broad range of selectivities. These models yielded highly significant cross-
validations (q2cv(D3/D2) ) 0.86; q2cv(D3/D4) ) 0.92) and excellent predictions of a 16-ligand test set (r2
pred
) 0.79-0.93). Exploiting this information, synthesis and receptor binding studies directed us to the fluorinated
lead compounds 78 and 79, featuring subnanomolar D3 affinities and considerable selectivities over D2 and
D4 and, subsequently, to the subtype selective PET tracers [18F]78 and [18F]79.
Introduction
and disappointing binding characteristics in preliminary auto-
radiography experiments. Thus, the discovery of highly selective
radioligands as PET tracers is still an important aim announcing
new insights into the role of the D3 receptor subtype in the
pathophysiology of numerous diseases.
Positron emission tomography (PETa) is a high-performance
molecular imaging technology that allows for functional and
quantitative information on receptor densities in the CNS.1 PET
has already gained growing importance for the diagnosis of
Morbus Parkinson by in ViVo imaging of alterations of the
dopaminergic signaling pathways and disturbances of D2-like
receptor densities using the radiopharmaceuticals 6-[18F]fluoro-
L-dopa2 and [11C]raclopride,3 respectively. More recently, the
availability of the D2 and D3 binding radioligand [18F]fallypride
enabled PET studies on human beings for the exploration of
low D2/D3 receptor densities in the cortex besides higher levels
in the putamen.4
The lack of bioavailable D3-subtype selective PET ligands
excluding crosstalk with the strongly related subtypes D2 and
D4 still hampers a noninvasive investigation of the physiological
role of D3. Because the D3 receptor is an interesting therapeutic
target for the treatment of Parkinson’s disease and schizophre-
nia,5 this method would probably guide us to a better under-
standing of the genesis and the therapy of these disorders.
To efficiently approach suitable D3 radioligands, we expected
an integrative procedure involving the elucidation of structural
features determining D3 selectivity by comparative molecular
analysis to be highly beneficial. Extending our recent work on
the 3D-QSAR (3D-quantitative structure-activity relationship)
analyses of dopaminergic agents,11-13 we herein describe the
generation of selectivity contour maps by CoMFA and com-
parative molecular similarity indices analysis (CoMSIA)14-23
displaying the molecular origins for a subtype specific recogni-
tion of D3 over both congeners D2 and D4.24-35 Exploiting this
information, the synthesis and receptor binding studies of
fluorinated lead compounds led us to the selective PET tracers
[18F]78 and [18F]79.
Results and Discussions
Assembly of the Training Set. Taking advantage of our in-
house data on D2-like ligands,10,36-42 we created a training set
of 63 structurally diverse compounds all belonging to the
privileged structure of 1,4-disubstituted arylpiperidines and
-piperazines. The training set was assembled from class A, B,
and C compounds representing D3- and D4-selective ligands and
D2 preferential ligands, respectively (Chart 1). A vital part of
our modeling strategy was to enrich our training set of our in-
house dopaminergics 2-46 and 50-63 by a number of external
reference ligands, including 1 (BP 897),43 47 (L-745,870),44 48
(haloperidol), and 49 (aripiprazole), thus enhancing the ligands’
structural diversity. To ensure that the experimentally obtained
D3/D2 and D3/D4 affinity differences are the result of consistent
assay conditions, all biological data have been measured in
our lab.
In the past, various attempts have been made to develop a
suitable PET ligand for the dopamine D3 receptor including the
preparation of the 11C-labeled imidazo[2,1-b]thiazolylpiperazine
derivative RGH-1756 and the substituted aminotetraline
[11C]GR218231.6-8 Using the benzothiophene derivate 7 (FAUC
365) as a lead compound,9 we synthesized fluorine-18-labeled
derivatives as putative PET imaging agents.10 However, the
above-mentioned tracers showed a low signal for specific
binding to the D3 receptor or rapid efflux from the brain in ViVo
* To whom correspondence should be addressed. Tel.: +49
(9131)-8529383. Fax:
+49(9131)-8522585. E-mail:
gmeiner@
pharmazie.uni-erlangen.de.
† Department of Medicinal Chemistry.
‡ Department of Nuclear Medicine.
a Abbreviations: PET, positron emission tomography; QSAR, quantita-
tive structure-activity relationship; CoMSIA, comparative molecular
similarity indices analysis; QSSR, quantitative structure-selectivity relation-
ship; PLS, partial least-squares; APS, all placement search; AOS, all
orientation search.
Alignment. Finding a suitable alignment is probably the most
important step to establish a 3D-QSAR model successfully. This
was even more crucial for our bidirectional quantitative
structure-selectivity relationship (QSSR) model that was
10.1021/jm0611152 CCC: $37.00 © 2007 American Chemical Society
Published on Web 01/06/2007

490 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
Chart 1
expected to precisely predict D3 selectivity over both congeners
D2 and D4. After compiling representative conformations and
deducing the shape of all other ligands thereof, we decided to
employ one common alignment for both QSSR models when
the D3-selective antagonist 7 was selected as a template. The
alignment was performed by comparison of steric overlap and
molecular electrostatic potentials using the module ASP,45 as
implemented in the program package TSAR46
for steric/electrostatic energies, did not improve the model. The
step by step modification of these parameters was done manually
(we did not use a respective algorithm).
For the grid spacing, we used a 1 Å lattice although this
increased the computation time by a factor of 8 (compared to
the default value of 2.0 Å) trying to ensure that no important
contributions to the correlation analysis were lost due to the
arbitrary cutoff values.19
CoMFA. The CoMFA method was employed for deriving a
3D-QSSR model consisting of a training set of 63 ligands (Chart
1, Table 1). The leave-one-out partial least-squares (PLS)
analysis of the obtained models yielded excellent cross-validated
q2-values of 0.85 (four components) and 0.92 (five components)
for D3/D2 and D3/D4 selectivity, respectively. These correlation
coefficients suggest that our model is reliable and accurate.
Subsequently, non-cross-validated PLS regressions were com-
puted using the previously obtained optimum number of
components giving regression coefficients r2 of 0.95 (D3/D2)
and 0.98 (D3/D4) (Table 2). To quantify the translational
dependence of q2, we applied the all placement search (APS)
procedure published by Wang et al.,47 which systematically
translates the aligned dataset in space. Starting from the default
region, the upper and lower margins are systematically increased
by increments of 0.1 Å in the direction of the x, y, and z axes
until the grid spacing distance of 1.0 Å is reached. By this
translation process, we obtained 10 × 10 × 10 ) 1000 different
placements, followed by PLS analysis after every retranslation.
Figure 1 shows the frequency distribution of the obtained q2
values among all the placements after the translation (1 Å
lattice). The range of the values 0.84-0.88 (D3/D2) and 0.91-
0.93 (D3/D4) is quite narrow, indicating that our CoMFA model
is rather stable toward different grid placements. The highest
CoMSIA. CoMSIA computes the steric and electrostatic
fields (as in CoMFA), but it also calculates additional hydro-
phobic, hydrogen-bond donor, and hydrogen-bond acceptor
fields. The resulting contour maps are easier to interpret than
those in CoMFA because a Gaussian function is used to
determine the distance-dependence. Therefore, the similarity
indices can also be calculated at the grid points inside the
molecules, not just outside, as it is in CoMFA.15
Several CoMSIA models were generated using the combina-
tions of different fields (Table 2). The purpose of using several
combinations of different fields is not only to increase the
significance and predictive power of the 3D-QSSR models, but
also to partition the various properties into spatial locations
where they play a decisive role in determining the selectivity.
CoMSIA, in most instances, performs similarly to CoMFA in
terms of predictive ability. The results of the initial CoMSIA
models for different combinations are summarized in Table 2.
For D3/D2 selectivity CoMSIA runs, q2 values of over 0.86 were
obtained, indicating that the obtained models should be more
accurate than the corresponding CoMFA models in predicting
the D3/D2 selectivity values, while for D3/D4 selectivity, the
CoMFA model performs significantly better than CoMSIA
models based on the obtained q2 and r2 values. For D3/D2
selectivity, a combined use of all the five descriptors resulted
in the best model (q2 ) 0.86 and r2 ) 0.94 with four
components). For D3/D4 selectivity, a combined use of the steric,
electrostatic, hydrophobic, and hydrogen-bond donor descriptors
produced the best model (q2 ) 0.91 and r2 ) 0.97 with four
components).
q2 for (D3/D2) 0.88 was found for 10 components with sc/v
)
0.50, while for (D3/D4), the highest q2 was 0.93 for 10
components with sc/v ) 0.63. However, it should be noted that,
frequently, an increase of q2 values by less than 5% for the
cv
use of an additional component is considered inappropriate due
to the “parsimony principle”.48 Thus, we used the original
models, which are in line with this rule, to derive the graphical
CoMFA maps. During the optimization of our CoMFA model,
variations of parameters, such as grid spacing and cutoff values
Validation of the QSSR Models. Employing a test set of
14 ligands, we intended to verify the excellent statistical
parameters that we observed and to investigate whether fluorine-
substituted derivatives can be predicted well. As can be seen in

Structure-SelectiVity InVestigations
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3 491
Table 1. Data Set of 63 Compounds Used in the Training Set and Their Experimental and Predicted Selectivitya
e
e
∆D3/D2
∆D3/D4
D3/D2
(exp)
D3/D4
(exp)
cpd
pos
n
X
Y
Z
R1
OMe
Cl
Cl
Cl
Cl
OMe
Cl
OMe
OMe
OMe
OMe
Cl
OMe
Cl
OMe
Cl
OMe
OMe
OMe
Cl
R2
R3
R4
CoMFA
CoMSIA
CoMFA
CoMSIA
1
2
3
H
H
H
H
H
H
Cl
H
H
H
H
Cl
H
Cl
H
Cl
H
H
H
Cl
Cl
-(CH)4-
-(CH)4-
2.17
2.66
2.44
1.96
2.66
2.57
3.86
3.02
2.17
2.72
1.99
2.43
2.14
3.46
1.48
3.67
2.39
2.65
2.80
3.49
2.98
0.32
0.11
1.86
0.14
-0.10
-0.33
1.13
1.83
0.72
-0.14
1.03
0.45
0.12
-0.15
0.10
0.28
-0.37
-0.51
0.21
0.66
0.75
-0.40
0.46
-0.03
-0.59
-0.19
0.53
-0.78
0.48
0.10
0.29
0.54
0.14
-0.01
0.05
0.12
-0.29
-0.02
0.50
1.07
1.01
0.10
0.74
-0.08
-0.60
-0.31
0.40
-0.63
0.30
-0.17
-0.22
0.44
-0.22
-0.37
0.66
0.72
0.30
0.48
-0.57
0.17
-0.55
0.33
-1.07
-0.23
0.14
1.44
2.30
2.22
1.74
2.93
1.81
2.83
1.90
1.99
2.60
1.43
1.89
1.39
2.95
1.13
3.41
1.62
1.81
2.20
2.91
3.25
-0.92
-1.20
1.48
0.18
-2.73
0.02
0.64
1.38
-0.51
-3.23
-2.12
-2.73
-3.14
-3.57
-3.38
-1.79
-2.22
-2.15
-2.82
-2.72
-1.98
-2.56
-1.79
-1.75
-1.46
-4.70
-0.14
1.11
1.82
1.16
1.47
1.00
1.96
1.44
0.99
0.93
0.49
1.00
0.45
0.44
1.34
0.73
-0.13
-0.07
0.79
-0.56
-0.17
0.32
-0.05
0.12
0.13
0.93
-0.34
-0.50
0.61
-0.44
-0.15
0.61
0.19
0.36
0.52
1.10
H
Cl
OMe
Ph
4
OMe
H
5
6b
S
S
S
S
H
H
7
8
9
5CCH
5-CN
6CCH
H
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31c
32
33
34
35
36d
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
S
O
O
O
O
O
N
N
N
N
N
N
0.20
0.25
H
-0.62
-0.36
0.38
-0.40
0.55
-0.13
0.02
0.39
-0.18
0.01
1.27
0.28
0.04
-0.41
-1.99
0.40
0.06
0.19
-0.65
-0.51
0.45
-0.27
0.44
-0.12
-0.12
0.39
-0.46
0.16
1.15
0.19
0.02
-0.63
-2.28
0.42
-0.27
0.46
-1.16
0.33
0.21
1.01
-0.44
0.00
-0.57
0.40
-0.05
0.03
-1.03
0.00
0.08
-0.45
0.54
5-Br
5-Br
5-CN
H
5-CN
5-Br
6-CN
5-Br
6-CN
Cl
-0.29
0.25
0.31
-0.39
-0.07
-0.71
0.07
2
2
2
2
3
3
4
5
7
3
3
3
3
3
2
1
2
3
4
1
4
3
3
3
0.06
-0.15
-0.58
-0.40
0.88
-1.07
0.82
H
6-CH2OH
7-I
7-CH3
7-CCH
H
H
2-OMe
H
4-Cl
4-F
2-F
H
-0.41
1.03
-0.34
-0.28
-1.46
0.65
-0.35
0.53
-0.95
0.11
0.31
0.19
0.05
-0.33
0.53
-1.47
-0.28
-0.16
0.36
-0.19
0.71
0.60
0.58
0.23
0.13
-0.21
-0.74
-0.10
-0.55
-0.30
0.39
0.34
0.75
-0.38
-0.42
-0.23
0.63
-0.15
-0.24
0.11
0.06
0.00
-0.69
0.21
-0.25
0.47
0.08
-1.71
-1.94
0.01
-0.11
0.82
0.21
-0.11
0.34
0.00
-0.38
-0.84
0.10
-0.31
-0.33
0.41
-0.20
-0.21
0.12
-0.09
-0.19
0.33
-0.26
0.00
N
N
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
N
N
N
N
N
N
N
N
N
C
C
H
H
H
H
H
H
H
H
H
H
H
H
H
0.33
-0.25
-0.05
-0.07
0.04
-0.21
-0.94
0.44
0.18
-0.08
-0.03
-0.71
0.29
0.06
0.37
CH3
H
H
H
H
CH3
H
H
H
0.03
0.35
-0.36
0.31
-1.39
0.66
H
0.24
-0.14
-1.13
-1.17
-0.46
-0.15
0.45
0.32
-0.36
0.39
-0.48
-0.47
-0.42
-0.10
-0.54
-0.36
0.37
-0.87
-0.39
0.96
0.53
0.07
-0.23
0.77
0.72
0.28
-0.57
-0.71
0.52
0.28
-0.49
0.80
1.02
4
4
5
5
4
4
4
5
5
4
4
5
4
5
2,3-diCl
2-OCH3
2,3-diCl
2-OMe
2,3-diCl
2-OMe
2-OMe
2,3-diCl
2-OMe
2-OMe
2-OMe
2-OMe
2-OMe
2-OMe
CH3
CH3
CH3
CH3
n-propyl
n-propyl
CH3
CH3
CH3
n-propyl
CH3
CH3
n-propyl
n-propyl
H
H
H
H
H
H
-0.29
-0.04
0.65
0.90
0.34
0.60
-0.28
0.48
-0.23
-0.15
-0.76
-0.32
0.50
4-OMe
4-OMe
4-OMe
4-OMe
2-Br
2-Br
2-Br
2-Br
-0.15
0.17
0.28
-0.04
-0.25
a Calculated as -log(Ki(D3)/Ki(D2)) and -log(Ki(D3)/Ki(D4)), respectively. b FAUC 346.39 c FAUC 113.36 d FAUC 21337. e ∆(D3/D2)/(D3/D4) is the error
of fitted selectivities ) [(D3/D2)/(D3/D4)experimental - (D3/D2)/(D3/D4)fitted].
Table 2, the trained CoMFA and CoMSIA models can reproduce
the experimental data very well for both D3/D2 and D3/D4
selectivity values of all the compounds included in the training
set. The ultimate test for the predictability of a QSAR or a QSSR
analysis in the drug design process is to predict the biological
activity/selectivity of new compounds that have not been
included in the training set. The 3D-QSSR models obtained in
this study were challenged with a test set consisting of 14
randomly selected compounds (64-77, Table 3) from the
original dataset of 77 ligands. The predicted versus the
experimental selectivity values for the training set, the test set,
and two newly synthesized compounds (78 and 79) are depicted

492 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
Table 2. Summary of the 3D-QSSR Modelsa
D3/D2 model
D3/D4 model
CoMSIA
CoMFA
SE
CoMSIA
CoMFA
SE
PLS statistic
SEHDA
SEHD
SED
SEHDA
SEHD
SED
q2
0.85
0.53
0.95
0.29
300.0
4
0.86
0.51
0.94
0.33
227.9
4
0.86
0.51
0.94
0.33
232.5
4
0.86
0.51
0.93
0.35
203.8
4
0.92
0.62
0.98
0.30
565.2
5
0.90
0.62
0.97
0.39
428.7
4
0.91
0.64
0.97
0.39
426.6
4
0.90
0.66
0.96
0.40
391.0
4
cv
sPRESS
r2
S
F
components
descriptors
field
3544
6581
5609
3525
3544
6581
5609
3525
contribution
steric
electrostatic
hydrophobic
donor
0.639
0.361
0.062
0.184
0.283
0.195
0.276
0.094
0.251
0.388
0.267
0.166
0.434
0.666
0.334
0.121
0.202
0.271
0.215
0.190
0.148
0.255
0.341
0.256
0.222
0.449
0.400
0.329
acceptor
a S, steric; E, electrostatic; H, hydrophobic; D, donor; A, acceptor.
3D Contour Maps. To visualize the information content of
the derived 3D-QSSR models, CoMFA and CoMSIA contour
maps were generated. The field energies at each lattice point
were calculated as the scalar results of the coefficient and the
standard deviation associated with a particular column of the
data table (“stdev*coeff”), which was always plotted as the
percentage of the contribution to the CoMFA or CoMSIA
equation. In the figures discussed below, the isocontour diagrams
of the field contributions (“stdev*coeff”) for different properties
calculated by the CoMFA and CoMSIA analysis are illustrated
with exemplary ligands. Selectivity fields depict the change in
binding preference occurring upon the change in molecular fields
around ligands. The contour plots may help to identify important
regions where any change may affect the binding preference.
Furthermore, they may be helpful in identifying important
features contributing to interactions between the ligand and the
active site of a receptor.
Models for D3/D2 Selectivity. In Figure 3, panels A and B,
the steric properties derived from the D3/D2 selectivity data are
displayed for CoMSIA. Areas indicated by green contours
correspond to regions where steric occupancy with bulky groups
should increase selectivity of D3 against D2. Areas encompassed
by yellow isopleths should be sterically avoided; otherwise,
reduced D3/D2 selectivity can be expected. As depicted, the
cyanoindole derivative 21, a very D3-selective ligand, orients
its heterocyclic ring to a green area, while the very low D3/D2
-selectivity of the 3-substituted azaindole 27 could be due to
the orientation of its pyrazolopyridine system to a yellow region.
Accordingly, the methyl groups of 50, 51, 57, and 60 and the
propyl group of 55, 59, and 62 are pointing toward the lower
yellow area. The triazole moieties of 50-63 are directed to the
upper, unfavored yellow isopleth. In addition, the yellow map
near the ortho-position of the phenyl group indicates that bulky
substituents there may decrease the D3/D2 selectivity. This
observation is confirmed by the fact that ortho-methoxyphenyl
containing ligands always display lower selectivity compared
to the dichlorophenyl derivatives. In panels C and D, the
electrostatic property maps include 16 and 57 as examples for
high and low D3-selectivity ligands, respectively. The blue and
red areas above and below the carboxamide moiety of the
indolyl-2-carboxamide 16 reflect a common placement of this
function within highly selective ligands. The other blue and red
areas are positioned in the vicinity of the protonated piperazine
nitrogen (N1) and the electron-rich, tertiary nitrogen (N2) of
highly selective ligands. Consequently, these two isopleths
determine the preferred position of the piperazine ring. In
Figure 1. Histogram showing the distribution of q2 values calculated
by SAMPLS leave-one-out cross-validation of D3/D2 selectivity CoMFA
(A) and D3/D4 selectivity CoMFA (B) after systematic translation of
aligned molecules within the lattice by an APS.
in Figure 2. In addition to these test set ligands, we also included
78 and 79 in the calculation of the r2pred value. The calculation
was performed according to the formula of Cramer et al.14 and
gave better results for CoMFA, with r2
) 0.84 than for the
pred
CoMSIA model (r2pred ) 0.78) in the case of D3/D2 selectivity.
The r2
of the D3/D4 selectivity showed values of 0.90 and
pred
0.93 for CoMFA and CoMSIA models, respectively. Thus, our
models display very high predictivity both in regular cross-
validation and in the prediction of a test set. For a rational design
of [18F]-labeled PET ligands, it was of special interest for us to
see that the calculated D3/D2 and D3/D4 selectivities of the
fluorinated test compounds 74-77 proved to be in good
accordance to the experimentally derived affinity differences.

Structure-SelectiVity InVestigations
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3 493
Table 3. Data set of 16 Compounds Used in the Test Set and Their Experimental and Predicted Selectivitiesa
a Calculated as -log(Ki(D3)/Ki(D2)) and -log(Ki(D3)/Ki(D4)), respectively. b Reference 39. c Reference 42. d Reference 10. e ∆(D3/D2)/(D3/D4) is the error
of predicted selectivities ) [(D3/D2)/(D3/D4)experimental - (D3/D2)/(D3/D4)predicted].
contrast, the red area for the benzyltriazole 57 is completely
buried in the transparent blue electrostatic potential at the N1,
while the transparent red electrostatic potential at the N2 is
completely embedded in the blue mesh. The hydrophobic effect
on the selectivity can be drawn from panels E and F, suggesting
that occupation of the 2 or 2 and 3 positions of the phenyl ring
by a hydrophobic group is crucial for a highly selective D3/D2
ligand, as illustrated by compounds 7, 12, 14, 16, 20, 21, and
50 and 2, 3, 4, and 5, respectively, while the yellow mesh on
the left is due to the Br of 14 and 20 and the Cl of 3. The
N-benzyltriazole carboxamides, which have very reduced
D3/D2 selectivity, have 2-chloro substituents in the small
orange isopleth on the left side, which is exemplified by ligands
60-63.
Models for D3/D4 Selectivity. Steric and electrostatic fields
of CoMFA contour maps are shown in Figure 4. The presence

494 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
Figure 2. Plots of the predicted versus observed selectivity values. A and C represent CoMFA results, whereas B and D refer to CoMSIA results.
Training set compounds (1-63) are indicated by open circles and test set compounds are indicated by filled circles (64-77) or filled rhombi
(78-79).
Figure 3. Contour maps of a D3/D2 selectivity CoMSIA model illustrating steric (A and B), electrostatic (C and D), and hydrophobic (E and F)
properties. For all features, one ligand with high (A, 21; C, 16; E, 20) and another with low D3/D2 selectivity (B, 27; D, 57; F, 60) are shown in
comparison. The mesh fields represent the stdev*coeff plots, whereas the transparent surfaces indicate the fields of the particular ligand, thus
facilitating the recognition of matching or mismatching features. In A and B, green isopleths enclose areas where steric bulk will enhance the
selectivity. Yellow contours highlight areas that should be kept unoccupied, otherwise the selectivity will decrease. In C and D, red isopleths
enclose areas where an increase of negative charge will enhance selectivity, whereas in blue-contoured areas, an increase of positive charge is
favorable for selectivity. In E and F, yellow isopleths encompass regions favorable for hydrophobic groups, and in orange-contoured areas, more
hydrophilic groups are favorable for selectivity.
of a substituent near a green region, the absence of a substituent
near a yellow region, the increase of a negative charge near a
red region, or a positive charge near a blue region shifts the
binding preference toward the D3 receptor, while the presence
of a substituent near a yellow region, the absence of a substituent
near a green region, the increase of a negative charge near a
blue region, or a positive charge near a red region shifts the
binding preference in the opposite direction toward the D4

Structure-SelectiVity InVestigations
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3 495
Figure 4. Contour maps of D3/D4 selectivity models as obtained by CoMFA (A, B, C, and D) and CoMSIA (E and F) analyses. These maps are
demonstrated by the highly selective compounds (A, 14; C, 20; E, 20) and less-selective compounds (B, 38; D, 26; F, 4). In A and B, green
isopleths enclose areas where steric bulk will enhance the selectivity. Yellow contours highlight areas that should be kept unoccupied, otherwise
the selectivity will decrease. In C and D, red isopleths enclose areas where an increase of negative charge will enhance selectivity, whereas in
blue-contoured areas an increase of positive charge is favorable for selectivity. In E and F, isopleths in cyan represent regions of hydrogen-bond
acceptors on the receptor site.
receptor. Such fields could, in principle, suggest the way a
leading structure should be modified for gaining higher selectiv-
ity. Panels A and B display the benzofurane 14 as an example
for a highly selective D3 ligand, and the pyrazole 38 as an
example for a ligand having very low selectivity. The green
areas near the benzofurane positions 5, 6, and 7 in 14 indicate
that sterically demanding substituents in these positions favor
D3/D4 selectivity. This is also the case with 8-10, 13-15, and
17-21. The yellow maps positioned on the right side are due
to all the biaryl- or heteroarylmethylene derivatives that have
very low D3/D4 selectivity and a bent orientation determining
this V shape of the yellow areas. In panels C and D, the
carboxamides 20 and 26 are depicted as examples for highly
D3-selective and low-selective ligands, respectively. The two
blue and red electrostatic isocontours on the left arise from the
appropriately positioned carboxamide function of all the highly
D3-selective ligands, while the blue area above the piperazine
ring corresponds to the protonated N1 of those ligands. Because
of the lower spacer length of 26, the piperazine moiety is shifted
left placing the protonated N1 and the N2 in the red and blue
isopleth, respectively. Furthermore, the two pyrazolopyridine
nitrogens as well as the o-methoxy substituent are directed to
unfavorable large blue areas. So the selectivity for the D3
receptor is larger for the four-carbon spacer carboxamides, and
it is dramatically decreased by decreasing or increasing the
number of the C-spacer or removing the carboxamide moiety.
The graphical interpretation of the field contributions of the
H-bond donor (from CoMSIA) is shown in panels E and F.
Cyan isopleth contour maps define where a H-bond donor group
within a ligand will be advantageous for the binding preference
toward the D3 receptor, while purple isopleths represent H-bond
donors that shift the binding preference in the opposite direction
toward the D4 receptor. In principle, they should highlight the
areas beyond the ligands where putative partners in the receptor
can form a hydrogen bond that will influence the binding affinity
and selectivity significantly. For the D3-selective ligand 20, the
cyan transparent area of the protonated N1 is positioned on the
cyan mesh, which corresponds to the Asp3.32 in the D3 receptor.
However, the cyan transparent near N2 of the D4-selective
azaindole 47 is aligned on the purple mesh.
Prediction of D3-Selective PET Tracers. Based on our initial
approach toward the development of D3 radioligands for PET
involving the test compounds [18F]-74-77,10 this work focused
on the rational prediction of novel [18F]-labeled PET ligands
with improved D3 receptor selectivity. Consequently, we chose
the highly potent biphenylcarboxamide 5 (Ki ) 0.28 nM) as an
interesting lead compound for the structural design of potential
PET tracers. To decrease the lipophilicity of the target com-
pounds causing accumulation in membranes and, thus, very high
unspecific binding (5: ClogP)5.94) and simultaneously gaining
the possibility to perform nucleophilic [18F]-substitution reac-
tions using ortho-(hetero)aryl halides as labeling precursors, we
examined further substitution patterns by replacing phenyl by
an ortho-fluoro-substituted pyridine nucleus, and also o-chlorine
by an o-methoxy substituent, to obtain the azabiphenyl deriva-
tives 78 and 79, which can be described as structural hybrids
of the lead compound 5 and the fluoropyridylcarboxamides 76/
77. Structurally related biphenyl carboxamides, such as GR
103691 and NGB 2904, have already been described as selective
antagonists for D3 receptors.5,49,50
When the target compounds 78 and 79 were predicted
employing the CoMFA and CoMSIA model, the 2-fluoropyri-
din-6-yl derivative 78 was suggested to have a D3/D2 selectivity
of 2.29/1.87 and a D3/D4 selectivity of 1.86/2.02 as predicted

496 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
Table 4. Radiosynthesis and Radiochemical Yields (RCY; [%]) of the
PET Ligands [18F]78 and [18F]79a
Scheme 1. Synthetic Pathway of the Predicted D3 Ligands (79
and 78) as References of Their 18F-Labeled Analogs and the
Respective Precursor Compounds 83 and 84
labeling
precursor
position
leaving group
RCY [%]
C log Pb
-
83
84
3
2
NO2
[18F]79: 86 ( 5
79: 3.36
78: 3.57
Br-
[18F]78: 73 ( 2
a 500 µL DMF, 8 µmol precursor (84 or 83), 140 °C, n.c.a. [18F]fluoride
(20 MBq), kryptofix 2.2.2, K2CO3, and t ) 20 min. b Calculated value using
the program C log P; Log P of the reference 5 was 5.35
for [18F]78 by using bromide as a leaving group (Table 4). Both
radioligands [18F]79 and [18F]78 were synthesized in very high
radiochemical yields and in high radiochemical purity under
the same reaction conditions, thus providing evidence for the
accessibility of these two 18F-labeled PET ligands in sufficient
quality.
Biological Testing. Receptor binding experiments were
established to evaluate the binding profile of the fluorinated
ligands 78 and 79 (Table 5). D1 receptor affinities were
determined utilizing porcine striatal membranes and the D1
selective radioligand [3H]SCH 23390.52 D2long, D2short, D3, and
from CoMFA/CoMSIA, respectively. The 2-fluoropyridin-5-yl
derivative 79 was supposed to give a D3/D2 selectivity of 2.46/
1.84 and a D3/D4 selectivity of 2.25/1.95 by CoMFA/CoMSIA
prediction, respectively. It is also worthy to note that the
calculation of log P values indicated a significant loss of
lipophilicity due to the C,N exchange (5, C log P ) 5.35; 78,
C log P ) 3.57; 79, C log P ) 3.36).
D4 receptor affinities were investigated employing the cloned
53
human dopamine receptor subtypes D2long, D2short
,
D3,54 and
55
D4.4 stably expressed in Chinese hamster ovary cells (CHO)
and the radioligand [3H]spiperone.50 For the investigation of
affinity to the related serotonergic receptor subtypes 5-HT1A
and 5-HT2, porcine cortex membranes were used together with
the selective radioligands [3H]WAY 100635 and [3H]ketanserin
for 5-HT1A and 5-HT2 receptors, respectively.56
At this point, we were ready to precede with the synthesis of
the two novel ligands 78 and 79 to verify our aforementioned
predictions by in vitro biological testing and to investigate the
radiochemical accessibility of the [18F]-labeled PET ligands.
Chemistry and Radiolabeling. After verifying that D3
selectivities of fluorine-substituted test compounds can be
predicted well, we tried to prove the effectiveness of the new
QSSR models in the design and synthesis of novel D3-selective
PET ligands. In our previous work, we successfully demon-
strated the feasibility of a direct nucleophilic 18F-substitution
on ortho-halogen-substituted pyridines, following the pioneering
study of Dolci et al.10,51 The syntheses of the putatively D3-
selective ligands 78 and 79 and the corresponding labeling
precursors 84 and 83 were accomplished as illustrated in Scheme
1. Suzuki coupling of 4-carboxybenzene boronic acid with the
corresponding bromo-substituted pyridine derivatives 80a-d
was carried out in acetonitrile/water to afford the biarylcar-
boxylic acids 81a-d in 43-67% yield. The primary amine 82
was obtained starting from the commercially available 2-meth-
oxyphenylpiperazine by N-alkylation with 4-bromobutylphthal-
imide and subsequent hydrazinolysis.39 The fluoro-substituted
lead compounds 78 and 79 and the labeling precursors 83 and
84 were synthesized by DCC-promoted coupling of the car-
boxylic acids 81a-d with the aminobutyl-substituted arylpip-
erazine 82. The carboxamides 78 and 79 were subjected to in
vitro biological testing and served as authentic reference
compounds in analytical radio-HPLC. Compounds 83 and 84
The data of the radioligand binding studies showed only poor
recognition of the D1 and 5-HT2 receptor (Table 5). Although
both test compounds 78 and 79 revealed high affinities to the
subtypes of the D2 receptor family as well as to 5-HT1A when
binding data between 12 nM and 34 nM were measured, the
desired D3 selectivity of potential PET ligands is indicated by
Ki values of 0.37 nM and 0.45 nM, as determined for 78 and
79, respectively. Thus, D3 affinities were excellent and the
selectivities over D2long, D2short, D4, 5-HT1A, and 5-HT2 for the
fluorinated carboxamide 78 and its regioisomer 79 proved to
be >40 and >25, respectively.
Conclusion
The 3D-QSSR analysis presented herein makes it possible
to relate chemical structures of ligands with their binding
selectivity with respect to different subtypes of a target receptor
when using the CoMFA or CoMSIA technique. The derived
CoMFA models showed high cross-validation correlation coef-
ficient q2 values of 0.85 and 0.92 for D3/D2 and D3/D4
selectivity, respectively, while for the CoMSIA models, the q2
values were 0.86 and 0.91. The high q2 values, along with further
testing, indicate that the obtained QSSR models should be
valuable in predicting the selectivity of new ligands. Using a
test set of 14 compounds, we evaluated the predictive perfor-
mance of the models on external ligands, demonstrating their
applicability by a mean predictive r2 of 0.84 (CoMFA) and 0.78
(CoMSIA) for the D3/D2 selectivity model and 0.90 (CoMFA)
and 0.93 (CoMSIA) for the D3/D4 selectivity model, respec-
represent precursors for the nucleophilic, aromatic 18F-for-NO2
-
and 18F-for-Br substitution in DMF, providing access to the
radiolabeled ligands [18F]79 and [18F]78, respectively. Both
leaving groups are suited very well for a nucleophilic, aromatic
substitution, as shown by 86% decay-corrected radiochemical
yield for [18F]79 applying 18F-for-NO2 substitution and 73%

Structure-SelectiVity InVestigations
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3 497
Table 5. Binding Affinities of the Fluorinated Target Compounds 78 and 79 to the Human Dopamine Receptor Subtypes D2long, D2short, D3, and D4, the
a
Porcine D1 Receptor, and the Porcine 5-HT1A and 5-HT2
Ki values ( SD in [nM]
[3H]SCH 23390
D1
[3H]spiperone
[3H]8-OH-DPAT
5-HT1A
[3H]ketanserin
5-HT2
cpd
D2long
D2short
D3
D4.4
78
79
870 ( 14
1300 ( 210
27 ( 7.1
15 ( 7.1
16 ( 1.4
12 ( 0.71
0.37 ( 0.064
0.45 ( 0.0070
28 ( 3.5
34 ( 1.4
16 ( 1.4
21 ( 6.4
450 ( 28
760 ( 120
a Ki values in nM are based on the means of two experiments, each done in triplicate.
cause a variation in the q2 value and, hence, the predictivity of the
model. A Sybyl programming language (SPL) script was written
and published by Wang et al.47 to perform the all orientation search
(AOS)/APS routine automatically. We applied the APS routine to
address the phenomena of the translational dependence of the
CoMFA described before. The five similarity indices in CoMSIA
(steric, electrostatic, hydrophobic, H-bond donor, and H-bond
acceptor descriptors) were calculated18 using a probe atom with a
radius of 1 Å and a charge of +1.0 placed at the lattice points of
the same region box as was used for the conventional CoMFA
calculations. A Gaussian-type distance dependence was used
between the grid points and each atom of the molecule. The default
value of 0.3 was chosen for the attenuation factor (R). Here, steric
indices are related to the third power of the atomic radii, electrostatic
descriptors are derived from atomic partial charges, hydrophobic
fields are derived from atom-based parameters,66 and H-bond donor
and acceptor indices are obtained by a rule-based method based
on experimental results.67
Partial Least-Squares (PLS) Analysis. The conventional
CoMFA and CoMSIA descriptors derived above were utilized as
explanatory variables, and the differences ∆pKi ) pKi(D3) -
pKi(D2/4) represented the target variable in PLS14,68 regression
analyses to derive 3D-QSAR models using the implementation in
the SYBYL package. The leave-one-out (LOO) algorithm was
selected in the crossvalidation run to obtain the optimal number of
components, the lowest standard error of prediction and the
corresponding q2 coefficient. Only a subset of CoMFA field sample
points showing a standard deviation of g2.0 kcal/mol (σmin value
for column filtering) were taken to perform PLS regression analysis.
The cross-validated coefficient, q2, was calculated using the
following equation
tively. This was extended to the successful application of our
analyses in guiding the synthesis of novel PET tracers for D3
receptors, namely, [18F]78 and its regioisomer [18F]79, both
revealing high subtype selectivity and good binding affinity.
Experimental Section
Structure Generation, Conformational Analysis, and Align-
ment. Structure building and refinement of the structures 1-79
(Chart 1 and Tables 1 and 3) has been accomplished using SYBYL
6.9 molecular modeling package57 running on indigo2 and octane2
Workstations (MIPS R10000 and R12000).
Because the used molecules show high structural diversity, we
selected several templates to compile representative conformations
for each group: 7, 22, 23, and 25 for heteroarylcarboxamides, 31
for fused heteroaryl- and biarylmethylenes, and 45 for phenyltet-
rahydropyridines.
Their geometry was optimized by a grid search (Tripos force
field,58 Gasteiger-Marsili charges),59,60 followed by hierarchical
clustering to select the most reasonable low-energy conformer. The
shape of all the other ligands was deduced from the corresponding
templates. The reference structures haloperidol (48) and aripiprazole
(49) were constructed, and their conformational space was evaluated
as described for the template structures. Finally all compounds were
optimized with MOPAC using the AM1-Hamiltonian61,62 (MMOK
was used to fix amide bonds). Subsequently, we calculated VESPA
charges63 (implemented in VAMP64) for all ligands, which then
were aligned to template 7 (showing high selectivity for D3/D2 and
D3/D4) using the module ASP45 as implemented in the QSAR
package TSAR.46 One goal of this study was to test the predictability
of the analyses and to design and synthesize new PET imaging
agents of improved selectivity toward the D3 receptor. Therefore,
we divided the compounds into a training set containing 63
compounds and a test set of 14 compounds to assess the predictive
power of the model. These sets contained compounds from all
structural families and represented a balanced number of both the
more-selective and the less-selective compounds.
2
(Ypred - Yactual
)
∑
∑
q2 ) 1 -
2
(Yactual - Ymean
)
CoMFA and CoMSIA. CoMFA was performed using the QSAR
module in Sybyl 6.9. The steric and electrostatic potential fields
for CoMFA were calculated at each lattice intersection of a regularly
spaced grid of 1.0 Å, while the other SYBYL default parameters
were used. The grid was setup with boundaries extending 4 Å
beyond the van der Waals envelopes of all molecules including a
distance-dependent dielectric constant. A sp3 carbon atom with a
charge of +1.0 served as the probe atom to calculate steric and
electrostatic fields. These contributions were truncated at 30 kcal/
mol and scaled by the CoMFA standard option. As first reported
by Cho et. al,65 the cross-validated r2 (q2) value, which serves as a
quantitative measure of the predictability, fluctuates with the
absolute orientation of the aligned molecular aggregate toward the
grid. The reason for this fluctuation is the fact that conventional
CoMFA samples the continuous molecular field at discrete lattice
points and calculates the steric and electrostatic field energies on
each lattice point with distance-sensitive functions, such as the
Lennard-Jones 6-12 potential. When the molecular aggregate
rotates/translates, so does the molecular field surrounding the
aggregate. The lattice box in CoMFA, however, is always axis-
aligned and does not rotate/translate along with them. Thus, different
points in the same molecular field will be mapped onto the lattice
points, resulting in different field energy values. These values, when
processed subsequently by PLS to produce the final model, will
where Ypred, Yactual, and Ymean are the predicted, actual, and mean
values of the target property, respectively. Σ(Ypred - Yactual)2 results
in the predictive sum of squares (PRESS). The number of principal
components resulting in the lowest PRESS value (the optimal
number of components (ONC)) was included in the generation of
the final non-crossvalidated PLS model,69 which yielded the final
correlation coefficient r2 and its standard error s. CoMFA and
CoMSIA coefficient maps were generated by interpolation of the
pairwise products between the PLS coefficients and the standard
deviations of the corresponding CoMFA or CoMSIA descriptor
values.
General Procedure for the Preparation of Pyridinyl Benzoic
Acids 81a-d via Suzuki Reaction.70 To a solution of p-
carboxybenzene boronic acid (1 mmol) and bromopyridine (1
mmol) in 0.4 M sodium carbonate solution (5 mL) and acetonitrile
(5 mL) was added Pd(PPh3)4 (60 mg, 0.05 mmol). The mixture
was stirred at 90 °C for 3 h. After filtration of the hot suspension,
the filtrate was concentrated under reduced pressure to about half
of the volume. The concentrate was washed with CH2Cl2, and the
separated aqueous layer was acidified with HCl (32%). The
precipitate was collected, washed with water, and dried on air
without purification.
4-(6-Bromopyridin-2-yl)-benzoic Acid (81c). Starting from
4-carboxybenzene boronic acid (166 mg, 1 mmol), 80c (237 mg, 1

498 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
mmol), Na2CO3 solution (5 mL, 0.4 M), acetonitrile (5 mL), and
Pd(PPh3)4 (60 mg, 0.05 mmol), 81c (175 mg, 67%) was obtained
as a white solid: 1H NMR (360 MHz, DMSO-d6) δ 7.7 (d, 1H),
7.85 (t, 1H), 8.08 (d, 2H), 8.11 (d, 1H), 8.18 (d, 2H), 13.1
(m, 1H).
ments with the human D2long, D2short, D3 and D4.4 receptors were
run with preparations of membranes from CHO cells expressing
the corresponding receptor and [3H]spiperone at a final concentra-
tion of 0.1 nM (for D4 at 0.2 nM). The assays were carried out
with a protein concentration of 1.5-8 µg/assay tube and Kd values
of 0.05 nM for D2long, 0.03 nM for D2short, 0.08 nM for D3, and
0.16 nM for D4.4. The investigation of serotonin 5-HT1A and 5-HT2
binding was performed as described in the literature.56 In brief,
porcine cortical membranes were subjected to the binding assay at
a concentration of 55 µg/assay tube and 100 µg/assay tube for
determination of 5-HT1A and 5-HT2 binding utilizing [3H]WAY
100-635 and [3H]ketanserin each at a final concentration of 0.1
nM and 0.5 nM, respectively, and with KD values of 0.07 nM (for
5-HT1A) and 1.5 nM (for 5-HT2). Protein concentration was
established by the method of Lowry using bovine serum albumin
as standard.71 Data analysis of the resulting competition curves was
accomplished by nonlinear regression analysis using the algorithms
in PRISM (GraphPad software, San Diego, CA). Ki values were
derived from the corresponding EC50 data utilizing the equation of
Cheng and Prusoff.72
General Procedure for the Carboxamide Synthesis via DCC
Coupling. 4-(4-(2-Methoxyphenyl)piperazin-1-yl)butylamine (82)
was prepared in two steps from 1-(2-methoxyphenyl)piperazine and
bromobutylphthalimide, followed by cleaving the phthalimide with
hydrazine.39 To a solution of the amine 82 (1 equiv) in dry CH2Cl2
(15 mL) was added DMAP (1.25 equiv) and the corresponding
carboxylic acids 81a-d (1 equiv) under an argon atmosphere. To
this solution DCC (1.25 equiv) was added. The mixture was stirred
at room temperature overnight. The precipitate was removed, and
the solution was washed with water (2 × 20 mL) and brine (20
mL). After separation, the organic layer was dried with Na2SO4
and concentrated under reduced pressure. The residue was purified
by column chromatography over silica gel using CH2Cl2/MeOH
(9:1) to give 78, 79, 83, and 84 in chemical purity >95%.
N-[4-[4-(2-Methoxyphenyl)piperazin-1-yl]butyl]-4-(6-bromopy-
ridin-2-yl)benzamide (84). According to the general procedure for
the DCC coupling, 4-[4-(2-methoxyphenyl)-piperazin-1-yl]buty-
lamine (119 mg, 0.45 mmol), DMAP (74 mg, 0.6 mmol), and 81c
(125 mg, 0.45 mmol) were stirred in 15 mL CH2Cl2. After adding
dicyclohexylcarbodiimide (124 mg, 0.6 mmol), 84 (172 mg, 73%)
was obtained as a white solid: 1H NMR (360 MHz, CDCl3) δ 1.55
(m, 4H), 2.36 (t, 2H), 2.49 (m, 4H), 2.94 (s, 4H), 3.29 (m, 2H),
3.74 (s, 3H), 6.84 (m, 4H), 7.59 (d, 1H), 7.81 (t, 1H), 7.93 (d, 2H),
8.06 (m, 3H), 8.48 (t, 1H); APCI-MS m/z 524.46 (M+, 100).
N-[4-[4-(2-Methoxyphenyl)piperazin-1-yl]butyl]-4-(6-fluoro-
pyridin-2-yl)benzamide (78). According to the general procedure
for the DCC coupling 4-[4-(2-methoxyphenyl)-piperazin-1-yl]-
butylamine (105 mg, 0.4 mmol), DMAP (61.1 mg, 0.5 mmol), and
81d (87 mg, 0.4 mmol) were stirred in 15 mL CH2Cl2. After adding
dicyclohexylcarbodiimide (103 mg, 0.5 mmol), 78 (120 mg, 65%)
was obtained as a white solid: 1H NMR (360 MHz, CDCl3) δ 1.7
(m, 4H), 2.55 (t, 2H), 2.73 (s, 4H), 3.08 (m, 4H), 3.51 (q, 2H),
3.86 (s, 3H), 6.87 (m, 4H), 6.98 (m, 1H), 7.07 (dd, 1H), 7.64 (m,
1H), 7.88 (m, 3H), 8.06 (m, 2H); APCI-MS m/z 463.56 (M+, 100).
Anal. (C27H31FN4O2): C, H, N.
Acknowledgment. This work was supported by the DFG
(Pr 677/2). I. Salama wishes to thank the Egyptian government
for the long-term scholarship. The authors wish to thank Dr.
J.-C. Schwartz, Dr. P. Sokoloff (INSERM, Paris), and the
Institute of Psychiatry, Toronto, as well as J. Shine (The Garvan
Institute of Medical Research, Sydney) for providing dopamine
D4, D3, and D2 receptor expressing cell lines, respectively.
Supporting Information Available: Figures showing contour
maps of D3/D2 and D3/D4 selectivity models as obtained by CoMFA
and CoMSIA analyses, methods and materials, preparation/analyti-
cal data of 79, 81a,b,d, and 83, and a table of elemental analysis
data. This material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Massoud, T. F.; Gambhir, S. S. Molecular imaging in living
subjects: Seeing fundamental biological processes in a new light.
Genes DeV. 2003, 17, 545-580.
Radiosyntheses of N-[4-[4-(2-Methoxyphenyl)piperazin-1-yl]-
butyl]-4-(6-[18F]fluoropyridin-2-yl)benzamide ([18F]78) and N-[4-
[4-(2-Methoxyphenyl)piperazin-1-yl]butyl]-4-(6-[18F]fluoropyridin-
3-yl)benzamide ([18F]79). No-carrier-added [18F]fluoride was
produced by the 18O(p,n)18F reaction using a RDS 111 cyclotron
(CTI) at PET Net GmbH (Erlangen, Germany). A QMA-cartridge
with [18F]fluoride (typically 300 MBq) was eluted with a solution
of 15 mg Kryptofix 2.2.2/15 µL of 1 N potassium carbonate stock
solution in 1 mL acetonitrile/water (8:2). The solvent was evapo-
rated by azeotropic drying under a stream of nitrogen at 85 °C.
This procedure was repeated two times using 500 µL of acetonitrile.
The dry residue was resolubilized with a solution of 8 µmol of
precursor 84 or 83 in 500 µL of dry DMF. Samples of the solution
(25 µL) were isolated in periods of 5, 10, 20, and 30 min. These
samples were used for determination of radiochemical yields by
reversed-phase HPLC. The identification of radiofluorinated PET
ligands [18F]78 and [18F]79 was performed by reversed-phase radio
HPLC (RP 18 Select B5 column (250 × 4 mm) eluted with
acetonitrile/water (70:30, 0.1% TFA, 1 mL/min)) using the UV
absorbance at 254 nm of the standard compounds 78 (tR ) 14.6
min) and 79 (tR ) 12.5 min) as a reference signal. Analytical HPLC
was performed on the following system: HPLC Hewlett-Packard
(HP 1100) with a quarternary pump and variable wavelength
detector (HP 1100) connected to a radio-HPLC detector D505TR
(Canberra Packard). Computer analysis of the HPLC data was
performed using FLO-One software (Canberra Packard).
(2) Lemaire, C.; Damhaut, P.; Plenevaux, A.; Comar, D. Enantioselective
synthesis of 6-[fluorine-18]-fluoro-L-dopa from no-carrier-added
fluorine-18-fluoride. J. Nucl. Med. 1994, 35, 1996-2002.
(3) Langer, O.; Nagren, K.; Dolle´, F.; Lundkvist, C.; Sandell, J.; Swahn,
C.-G.; Vaufrey, F.; Crouzel, C.; Maziere, B.; Halldin, C. Precursor
synthesis and radiolabelling of the dopamine D-2 receptor ligand
[C-11]raclopride from [C-11]methyl triflate. J. Labelled Compd.
Radiopharm. 1999, 42, 1183-1193.
(4) Mukherjee, J.; Christian, B. T.; Dunigan, K. A.; Shi, B.; Narayanan,
T. K.; Satter, M.; Mantil, J. Brain imaging of 18F-fallypride in normal
volunteers: Blood analysis, distribution, test-retest studies, and
preliminary assessment of sensitivity to aging effects on dopamine
D-2/D-3 receptors. Synapse 2002, 46, 170-188.
(5) Boeckler, F.; Gmeiner, P. The structural evolution of dopamine D3
receptor ligands: Structure-activity relationships and selected neu-
ropharmacological aspects. Pharmacol. Ther. 2006, 112, 281-333.
(6) Langer, O.; Gulyas, B.; Sandell, J.; Laszlovszky, I.; Kiss, B.; Domany,
G.; Acs, T.; Farde, L.; Halldin, C. Radiochemical labelling of the
dopamine D-3 receptor ligand RGH-1756. J. Labelled Compd.
Radiopharm. 2000, 43, 1069-1074.
(7) Sovago, J.; Farde, L.; Halldin, C.; Langer, O.; Laszlovszky, I.; Kiss,
B.; Gulyas, B. Positron emission in tomographic evaluation of the
putative dopamine-D3 receptor ligand, [11C]RGH-1756 the monkey
brain. Neurochem. Int. 2004, 45, 609-617.
(8) De Vries, E. F.; Kortekaas, R.; van Waarde, A.; Dijkstra, D.; Elsinga,
P. H.; Vaalburg, W. Synthesis and evaluation of dopamine D3
receptor antagonist 11C-GR218231 as PET tracer for P-glycoprotein.
J. Nucl. Med. 2005, 46, 1384-1392.
(9) Seeman, P.; Wilson, A.; Gmeiner, P.; Kapur, S. Dopamine D2 and
D3 receptors in human putamen, caudate nucleus, and globus pallidus.
Synapse 2006, 60, 205-211.
(10) Hocke, C.; Prante, O.; Lo¨ber, S.; Hu¨bner, H.; Gmeiner, P.; Kuwert,
T. Synthesis and evaluation of 18F-labeled dopamine D3 receptor
ligands as potential PET imaging agents. Bioorg. Med. Chem. Lett.
2005, 15, 4819-4823.
Receptor Binding Experiments. Receptor binding studies were
carried out as described in the literature.52 In brief, the dopamine
D1 receptor assay was done with porcine striatal membranes at a
final protein concentration of 40 µg/assay tube and the radioligand
[3H]SCH 23390 at 0.3 nM (Kd ) 0.65 nM). Competition experi-

Structure-SelectiVity InVestigations
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3 499
(11) Lanig, H.; Utz, W.; Gmeiner, P. Comparative molecular field analysis
of dopamine D4 receptor antagonists including 3-[4-(4-chlorophenyl)-
piperazin-1-ylmethyl]pyrazolo[1,5-a]pyridine (FAUC 113), 3-[4-(4-
chlorophenyl)piperazin-1-ylmethyl]-1H-pyrrolo-[2,3-b]pyridine (L-
745,870), and clozapine. J. Med. Chem. 2001, 44, 1151-1157.
(12) Boeckler, F.; Ohnmacht, U.; Lehmann, T.; Utz, W.; Hu¨bner, H.;
Gmeiner, P. CoMFA and CoMSIA investigations revealing novel
insights into the binding modes of dopamine D3 receptor agonists.
J. Med. Chem. 2005, 48, 2493- 2508.
(13) Salama, I.; Schlotter, K.; Utz, W.; Hu¨bner, H.; Gmeiner, P.; Boeckler,
F. CoMFA and CoMSIA investigations of dopamine D3 receptor
ligands leading to the prediction, synthesis, and evaluation of rigidized
FAUC 365 analogues. Bioorg. Med. Chem. 2006, 14, 5898-5912.
(14) Cramer, R. D., III; Patterson, D. E.; Bunce, J. D. Comparative
molecular field analysis (CoMFA). 1. Effect of shape on binding of
steroids to carrier proteins. J. Am. Chem. Soc. 1988, 110, 5959- 5967.
(15) Kubinyi, H., Ed. 3D-QSAR in Drug Design. Theory, Methods and
Applications; ESCOM: Leiden, The Netherlands, 1993.
(16) Kubinyi, H.; Folkers, G.; Martin, Y. C., Eds. 3D-QSAR in Drug
Design. Volume 2. Ligand-Protein Interactions and Molecular
Similarity; Kluwer/ESCOM: Dordrecht, The Netherlands, 1997.
(17) Kubinyi, H.; Folkers, G.; Martin, Y. C., Eds. 3D-QSAR in Drug
Design. Volume 3. Recent AdVances; Kluwer/ ESCOM: Dordrecht,
The Netherlands, 1998.
(18) Klebe, G.; Abraham, U.; Mietzner, T. Molecular similarity indices
in a comparative analysis (CoMSIA) of drug molecules to correlate
and predict their biological activity. J. Med. Chem. 1994, 37, 4130-
4146.
(19) Bo¨hm, M.; Stu¨rzebecher, J.; Klebe, G. Three-dimensional quantitative
structure-activity relationship analyses using comparative molecular
field analysis and comparative molecular similarity indices analysis
to elucidate selectivity differences of inhibitors binding to trypsin,
thrombin, and factor Xa. J. Med. Chem. 1999, 42, 458-477.
(20) Baskin, I. I.; Tikhonova, I. G.; Palyulin, V. A.; Zefirov, N. S.
Selectivity fields: Comparative molecular field analysis (CoMFA)
of the glycine/NMDA and AMPA receptors. J. Med. Chem. 2003,
46, 4063-4069.
(31) Hackling, A.; Ghosh, R.; Perachon, S.; Mann, A.; Ho¨ltje, H. D.;
Wermuth, C. G.; Schwartz, J. C.; Sippl, W.; Sokoloff, P.; Stark, H.
N-(ω-(4-(2-Methoxyphenyl)piperazin-1-yl)alkyl)carboxamides as
dopamine D2 and D3 receptor ligands. J. Med. Chem. 2003, 46,
3883-3899.
(32) De Paulis, T.; El Tayar, N.; Carrupt, P. A.; Testa, B.; Van de
Waterbeemd, H. Quantitative structure-affinity relationships of
dopamine D2 receptor antagonists: A comparison between orthop-
ramides and 6-methoxysalicylamides. HelV. Chim. Acta 1991, 74,
241-254.
(33) Cha, M. Y.; Lee, I. Y.; Cha, J. H.; Choi, K. I.; Cho, Y. S.; Koh, H.
Y.; Pae, A. N. QSAR studies on piperazinylalkylisoxazole analogues
selectively acting on dopamine D3 receptor by HQSAR and CoMFA.
Bioorg. Med. Chem. 2003, 11, 1293-1298.
(34) Bostro¨m, J.; Gundertofte, K.; Liljeforsa, T. A pharmacophore model
for dopamine D4 receptor antagonists. J. Comput.-Aided Mol. Des.
2000, 14, 769-786.
(35) Bostro¨m, J.; Bo¨hm, M.; Gundertofte, K.; Klebe, G. A 3D QSAR
study on a set of dopamine D4 receptor antagonists. J. Chem. Inf.
Comput. Sci. 2003, 43, 1020-1027.
(36) Lo¨ber, S.; Hu¨bner, H.; Gmeiner, P. Azaindole derivatives with high
affinity for the dopamine D4 receptor: Synthesis, ligand binding
studies, and comparison of molecular electrostatic potential maps.
Bioorg. Med. Chem. Lett. 1999, 9, 97-102.
(37) Lo¨ber, S.; Hu¨bner, H.; Utz, W.; Gmeiner, P. Rationally based efficacy
tuning of selective dopamine D4 receptor ligands leading to the
complete antagonist 2-[4-(4-chlorophenyl)piperazin-1-ylmethyl]pyra-
zolo[1,5-a]pyridine (FAUC 213). J. Med. Chem. 2001, 44, 2691-
2694.
(38) Lo¨ber, S.; Aboul-Fadl, T.; Hu¨bner, H.; Gmeiner, P. Di- and
trisubstituted pyrazolo[1,5-a]pyridine derivatives: Synthesis, dopam-
ine receptor binding and ligand efficacy. Bioorg. Med. Chem. Lett.
2002, 12, 633-636.
(39) Bettinetti, L.; Schlotter, K.; Hu¨bner, H.; Gmeiner, P. Interactive SAR
studies: Rational discovery of super-potent and highly selective
dopamine D3 receptor antagonists and partial agonists. J. Med. Chem.
2002, 45, 4594-4597.
(40) Bettinetti, L.; Lo¨ber, S.; Hu¨bner, H.; Gmeiner, P. Parallel synthesis
and biological screening of dopamine receptor ligands taking
advantage of a click chemistry based BAL linker. J. Comb. Chem.
2005, 7, 309-316.
(21) Yang, G.-F.; Lu, H.-T.; Xiong, Y.; Zhan, C.-G. Understanding the
structure-activity and structure-selectivity correlation of cyclic
guanine derivatives as phosphodiesterase-5 inhibitors by molecular
docking, CoMFA and CoMSIA analyses. Bioorg. Med. Chem. 2006,
14, 1462-1473.
(22) Melville, J. L.; Lovelock, K. R. J.; Wilson, C.; Allbutt, B.; Burke,
E. K.; Lygo, B.; Hirst, J. D. Exploring phase-transfer catalysis with
molecular dynamics and 3D/4D quantitative structure selectivity
relationships. J. Chem. Inf. Model. 2005, 45, 971-981.
(23) Catto, M.; Nicolotti, O.; Leonetti, F.; Carotti, A.; Favia, A. D.; Soto-
Otero, R.; Mendez-Alvarez, E.; Carotti, A. Structural insights into
monoamine oxidase inhibitory potency and selectivity of 7-substituted
coumarins from ligand- and target-based approaches. J. Med. Chem.
2006, 49, 4912-4925.
(24) Mach, R. H.; Hammond, P. S.; Huang, Y.; Yang, B.; Xu, Y.; Cheney,
J. T.; Freeman, R.; Luedtke, R. R. Structure-activity relationship
studies of N-(9-benzyl)-9-azabicyclo[3.3.1]nonan-3-â-yl benzamide
analogues for dopamine D2 and D3 receptors. Med. Chem. Res. 1999,
9, 355-373.
(25) Wilcox, R. E.; Tseng, T.; Brusniak, M. Y.; Ginsburg, B.; Pearlman,
R. S.; Teeter, M.; DuRand, C.; Starr, S.; Neve, K. A. CoMFA-based
prediction of agonist affinities at recombinant D1 vs D2 dopamine
receptors. J. Med. Chem. 1998, 41, 4385-4399.
(26) Wilcox, R. E.; Huang, W. H.; Brusniak, M. Y.; Wilcox, D. M.;
Pearlman, R. S.; Teeter, M. M.; DuRand, C. J.; Wiens, B. L.; Neve,
K. A. CoMFA-based prediction of agonist affinities at recombinant
wild type versus serine to alanine point mutated D2 dopamine
receptors. J. Med. Chem. 2000, 43, 3005-3019.
(27) Ravina, E.; Negreira, J.; Cid, J.; Masaguer, C. F.; Rosa, E.; Rivas,
M. E.; Fontenla, J. A.; Loza, M. I.; Tristan, H.; Cadavid, M. I.; Sanz,
F.; Lozoya, E.; Carotti, A.; Carrieri, A. Conformationally constrained
butyrophenones with mixed dopaminergic (D(2)) and serotoninergic
(5-HT(2A), 5- HT(2C)) affinities: Synthesis, pharmacology, 3D-
QSAR, and molecular modeling of (aminoalkyl)benzo- and thieno-
cycloalkanones as putative atypical antipsychotics. J. Med. Chem.
1999, 42, 2774-2797.
(41) Loaiza, P. R.; Lo¨ber, S.; Hu¨bner, H.; Gmeiner, P. Click chemistry
on solid phase: Parallel synthesis of N-benzyltriazole carboxamides
as super-potent G-protein coupled receptor ligands. J. Comb. Chem.
2006, 8, 252-261.
(42) Lo¨ber, S.; Hu¨bner, H.; Gmeiner, P. Synthesis and biological
investigations of dopaminergic partial agonists preferentially recog-
nizing the D4 receptor subtype. Bioorg. Med. Chem. Lett. 2006, 16,
2955-2959.
(43) Koob, G. F.; Caine, S. B. Cocaine addiction therapysare we partially
there? Nat. Med. 1999, 5, 993-995.
(44) Chavez-Eng, C. M.; Constanzer, M. L.; Matuszewski, B. K. Picogram
determination of a novel dopamine D4 receptor antagonist in human
plasma and urine by liquid chromatography with atmospheric pressure
chemical ionization tandem mass spectrometry. J. Chromatogr., B.
1997, 691, 77-85.
(45) ASP V3.22, Oxford Molecular, Ltd.: The Medawar Centre, Oxford
Science Park, Sandford-on-Thames, Oxford OX4 4GA, U.K., 1998.
(46) TSAR V3.2, Oxford Molecular, Ltd.: The Medawar Centre, Oxford
Science Park, Sandford-on-Thames, Oxford OX44GA, U.K., 1998.
(47) Wang, R.; Gao, Y.; Liu, L.; Lai, L. All-orientation search and all-
placement search in comparative molecular field analysis. J. Mol.
Model. 1998, 4, 276-283.
(48) Thibaut, U.; Folkers, G.; Klebe, G.; Kubinyi, H.; Merz, A.; Rognan,
D. Recommendations to CoMFA studies and 3D QSAR publications.
In 3D QSAR in Drug Design. Theory, Methods and Applications;
Kubinyi, H., Ed.; ESCOM: Leiden, The Netherlands, 1993; pp 711-
716.
(49) Murray, P. J.; Harrison, L. A.; Johnson, M. R.; Robertson, G. M.;
Scopes, D. I.; Bull, D. R.; Graham, E.; Hayes, A. G.; Kilpatrick, G.
J. A novel series of arylpiperazines with high affinity and selectivity
for the dopamine D3 receptor. Bioorg. Med. Chem. Lett. 1995, 5,
219-222.
(28) Norinder, U.; Hogberg, T. A quantitative structure-activity relation-
ship for some dopamine D2 antagonists of benzamide type. Acta
Pharm. Nord. 1992, 4, 73-78.
(29) McGaughey, G. B.; Mewshaw, R. E. Application of comparative
molecular field analysis to dopamine D2 partial agonists. Bioorg.
Med. Chem. 1999, 7, 2453-2456.
(30) Homan, E. J.; Tulp, M. T.; Nilsson, J. E.; Wikstro¨m, H. V.; Grol, C.
J. C5-substituted derivatives of 5-OMe-BPAT: Synthesis and interac-
tions with dopamine D2 and serotonin 5-HT1A receptors. Bioorg.
Med. Chem. 1999, 7, 2541-2548.
(50) Yuan, Y.; Chen, C.; Brodbeck, R.; Primus, R.; Braun, J.; Wasley, J.
W. F.; Thurkauf, A. NGB 2904 and NGB 2849: Two highly selective
dopamine D3 receptor antagonists. Bioorg. Med. Chem. Lett. 1998,
8, 2718-2718.
(51) Dolci, L.; Dolle´, F.; Jubeau, S.; Vaufrey, F.; Crouzel, C. 2-[18F]-
fluoropyridines by no-carrier-added nucleophilic aromatic substitution
with [18F]FK-K222sa comparative study. J. Labelled Compd. Ra-
diopharm. 1999, 42, 975-985.

500 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 3
Salama et al.
(52) Hu¨bner, H.; Haubmann, C.; Utz, W.; Gmeiner, P. Conjugated enynes
as nonaromatic catechol bioisosteres: Synthesis, binding experiments,
and computational studies of novel dopamine receptor agonists
recognizing preferentially the D3 subtype. J. Med. Chem. 2000, 43,
756-762.
(53) Hayes, G.; Biden, T. J.; Selbie, L. A.; Shine, J. Structural subtypes
of the dopamine D2 receptor are functionally distinct: expression
of the cloned D2A and D2B subtypes in a heterologous cell line.
Mol. Endocrinol. 1992, 6, 920-926.
(54) Sokoloff, P.; Andrieux, M.; Besanc¸on, R.; Pilon, C.; Martres, M.-P.;
Giros, B.; Schwartz, J.-C. Pharmacology of human dopamine D3
receptor expressed in a mammalian cell line: Comparison with D2
receptor. Eur. J. Pharmacol. 1992, 225, 331-337.
(55) Asghari, V.; Sanyal, S.; Buchwaldt, S.; Paterson, A.; Jovanovic, V.;
Van Tol, H. H. M. Modulation of intracellular cyclic AMP levels
by different human dopamine D4 receptor variants. J. Neurochem.
1995, 65, 1157-1165.
(56) Schlotter, K.; Boeckler, F.; Hu¨bner, H.; Gmeiner, P. Fancy bioisos-
teres: Metallocene-derived G-protein-coupled receptor ligands with
subnanomolar binding affinity and novel selectivity profiles. J. Med.
Chem. 2005, 48, 3696-3699.
(57) Sybyl 6.9, Tripos, Inc.: 1699 South Hanley Road, St. Louis, Missouri,
63144, 2002.
(58) Clark, M.; Cramer, R.-D.; Van Opdenbosch, N. Validation of the
general purpose Tripos 5.2 force field. J. Comput. Chem. 1989, 10,
982-1012.
(59) Gasteiger, J.; Marsili, M. Iterative partial equalization of orbital
electronegativity: A rapid access to atomic charges. Tetrahedron
1980, 36, 3219-3222.
(64) Clark, T.; Alex, A.; Beck, B.; Chandrasekar, J.; Gedeck, P.; Horn,
A.; Huttern, M.; Martin, B.; Rauhut, G.; Sauer, W.; Schindler, T.;
Steinke, T. VAMP 7.0 Program Package; Oxford Molecular Plc.:
Oxford, 1998.
(65) Cho, S. J.; Tropsha, A. Cross-validated R2-guided region selection
for comparative molecular field analysis: A simple method to achieve
consistent results. J. Med. Chem. 1995, 38, 1060-1066.
(66) Viswanadhan, V. N.; Ghose, A. K.; Revankar, G. R.; Robins, R. K.
Atomic physichochemical parameters for three dimensional structure
directed quantitative structure-activity relationships. 4. Additional
parameters for hydrophobic and dispersive interactions and their
application for an automated superposition of certain naturally
occurring nucleoside antibiotics. J. Chem. Inf. Comput. Sci. 1989,
29, 163-172.
(67) Klebe, G. The use of composite crystal-field environments in
molecular recognition and the de novo design of protein ligands. J.
Mol. Biol. 1994, 237, 212-235.
(68) Wold, S. Validation of QSAR’s. Quant. Struct.-Act. Relat. 1991, 10,
191-193.
(69) Baroni, M.; Clementi, S.; Cruciani, G.; Costantino, G.; Riganelli,
D.; Oberrauch, E. Predictive ability of regression models. Part II.
Selection of the best predictive PLS model. J. Chemom. 1992, 6,
347-356.
(70) Gong, Y.; Pauls, H. W. A convenient synthesis of heteroaryl benzoic
acids via suzuki reaction. Synlett 2000, 6, 829-831.
(71) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. Protein
measurement with the folin phenol reagent. J. Biol. Chem. 1951, 193,
265-275.
(60) Hinze, J.; Jaffe, H. H. Electronegativity. I. Orbital electronegativity
of neutral atoms. J. Am. Chem. Soc. 1962, 84, 540-546.
(61) Stewart, J. J. MOPAC: A semiempirical molecular orbital program.
J. Comput.-Aided Mol. Design 1990, 4, 1-105.
(62) Stewart, J. J. P.; Frank, J. MOPAC; Seiler Bokstav Research
Laboratory: United States Air Force Academy, CO 80840, 1990.
(63) Beck, B.; Clark, T.; Glen, R. C. A detailed study of VESPA
electrostatic potential-derived atomic charges. J. Mol. Model. 1995,
1, 176-187.
(72) Cheng, Y. C.; Prusoff, W. H. Relationship between the inhibition
constant (Ki) and the concentration of inhibitor which causes 50%
inhibition (IC50) of an enzymatic reaction. Biochem. Pharmacol. 1973,
22, 3099-3108.
JM0611152