Comparative molecular field analysis of fenoterol derivatives: A platform towards highly selective and effective β2-adrenergic receptor agonists

Article abstract of DOI:10.1016/j.bmc.2009.11.062

Purpose: To use a previously developed CoMFA model to design a series of new structures of high selectivity and efficacy towards the β₂-adrenergic receptor. Results: Out of 21 computationally designed structures 6 compounds were synthesized and characterized for β₂-AR binding affinities, subtype selectivities and functional activities. Conclusion: the best compound is (R,R)-4-methoxy-1-naphthylfelnoterol with K_iβ₂-AR = 0.28 μm, K_iβ₁-AR/K_iβ₂-AR = 573, EC_50cAMP = 3.9 nm, EC_50cardio = 16 nm. The CoMFA model appears to be an effective predictor of the cardiomocyte contractility of the studied compounds which are targeted for use in congestive heart failure.

Full text of DOI:10.1016/j.bmc.2009.11.062

Bioorganic & Medicinal Chemistry 18 (2010) 728–736
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Comparative molecular field analysis of fenoterol derivatives: A platform
towards highly selective and effective b2-adrenergic receptor agonists
b
c
c
c
Krzysztof Jozwiak a, , Anthony Yiu-Ho Woo , Mary J. Tanga , Lawrence Toll , Lucita Jimenez ,
*
Joseph A. Kozocas c, Anita Plazinska a, Rui-Ping Xiao b, Irving W. Wainer d
a Department of Chemistry, Medical University of Lublin, Lublin, Poland
b Laboratory of Cardiovascular Science, National Institute on Aging Intramural Research Program, Baltimore, MD, United States
c SRI International, Menlo Park, CA, United States
d Laboratory of Clinical Investigation, National Institute on Aging Intramural Research Program, Baltimore, MD, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 29 July 2009
Accepted 27 November 2009
Available online 6 December 2009
Purpose: To use a previously developed CoMFA model to design a series of new structures of high selec-
tivity and efficacy towards the b2-adrenergic receptor. Results: Out of 21 computationally designed struc-
tures 6 compounds were synthesized and characterized for b2-AR binding affinities, subtype selectivities
and functional activities. Conclusion: the best compound is (R,R)-4-methoxy-1-naphthylfelnoterol with
Kib2-AR = 0.28 lm, Kib1-AR/Kib2-AR = 573, EC50cAMP = 3.9 nm, EC50cardio = 16 nm. The CoMFA model
appears to be an effective predictor of the cardiomocyte contractility of the studied compounds which
are targeted for use in congestive heart failure.
Keywords:
Comparative molecular fields analysis
b2 selective agonist
Drug design
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
transmembrane (TM) components of the b2-AR. These interactions
included hydrogen bond formation between the p-oxygen moiety
In the initial studies in this series, (R,R)-fenoterol, Fig. 1, a po-
tent and selective b2-adrenoceptor (b2-AR) agonist was identified
as a potential drug for the treatment of congestive heart failure.1,2
(R,R)-Fenoterol is currently entering initial clinical trials for use in
this disease. Based on the potential therapeutic use of (R,R)-feno-
terol, the stereoisomers of fenoterol and a series of fenoterol ana-
logues were synthesized and tested for b2-AR activity and
selectivity.3 Three of the 26 compounds investigated in the latter
study had submicromolar affinity for the b2-AR, a >40-fold selectiv-
ity for the b2-AR relative to the b1-AR and were full agonists of the
b2-AR. One of these compounds, (R,R)-methoxyfenoterol was also
identified as a potential candidate for drug development and pre-
liminary pharmacokinetic and toxicity studies have been initiated.
The binding affinities, Ki values, of the 26 fenoterol analogues
were used to develop a comparative molecular field analysis model,
CoMFA(n = 26), to describe and predict the binding of the stereoiso-
mers of fenoterol and fenoterol analogues to the b2-AR.3 The
CoMFA(n = 26) model explained the role of stereoconfiguration
and structural modifications of studied molecules on observed
b2-AR affinities. The model was also consistent with earlier findings
from other laboratories,4,5 which indicated that the b2-AR selectiv-
ity of the fenoterol analogues was due to interactions of the amino-
alkyl portion of the molecules with different sites within the
on the phenyl ring in fenoterol and methoxyfenoterol and tyrosine
308 (Y308) in TM7 and/or histidine 296 (H296) in TM6 and and/
or -hydrogen bond interactions between the aromatic system of
p–p
p
the ligand and aromatic residues located in this area of the binding
site. In addition, the data from the previous study3 and an earlier
study6 indicated that the binding process of the fenoterol analogues
with the b2-AR included the interaction of the chiral center of the
aminoalkyl portion with a sterically-restricted site on the receptor.
In this study, the CoMFA(n = 26) model was used to design 21
additional fenoterol analogues as potential selective and potent
b2-AR agonists. Based on the CoMFA(n = 26) model and the indi-
cated binding interactions, three sites on the aminoalkyl portion
of the fenoterol molecule were chosen for alteration, Figure 1.
The modifications were utilized to explore the effect on b2-AR
binding affinity, selectivity and agonist activity of the steric bulk
OH
NH
HO
*
*
CH3
OH
OH
* Corresponding author. Tel.: +48 81 7423686; fax: +48 81 5357350.
E-mail address: krzysztof.jozwiak@umlub.pl (K. Jozwiak).
Figure 1. The structure of (R,R)-fenoterol in which the sites on the molecule probed
in this study are circled in red.
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.11.062

K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
729
at the second chiral center, of a 1-napthyl versus a 2-napthyl
substituent, and of the effect of a substituent at the 4-position of
a 1-naphthyl substituent. A subset of 6 compounds, composed of
the (R,R)- and (R,S)- isomers of 3 of the designed compounds,
was synthesized using the previously described approach.3 The
b2-AR and b1-AR binding affinities and the EC50 values associated
with induced cAMP accumulation (EC50cAMP) were determined.
The (R,R)-isomers of the subset were tested for their activity in a
previously described cardiomyocyte contractility model3 and their
EC50cardio values determined.
reductions in the affinity to the b1-AR were reflected in the relative
binding affinities as the Ki(b1-AR)/Ki(b2-AR) for the (R,R)-isomers
increased from 43 (1) to 334 (52) and from 5 (1) to 20 (52) for
the (R,S)-isomers. Thus, a modest increase in the steric bulk at
the chiral center on the aminoalkyl portion of the molecule in-
creased the relative energies of the agonist-b2-AR and agonist-b1-
AR complexes, irrespective of the configuration at that site. These
results are consistent with the existence of a sterically-restricted
site on the b2-AR and suggest that the site is also present in the
b1-AR and that the steric interaction has a stronger effect on the
binding to the b1-AR.
The substitution of an ethyl moiety for the methyl group pro-
duced a slight but significant increase in the EC50cAMP values of
(R,R)- and (R,S)-52 relative to (R,R)- and (R,S)-1, Table 3. However,
the increase in steric bulk produced a profound decrease in the
activity of (R,R)-52 in the cardiomyocyte contractility model as
the EC50cardio of (R,R)-52 was 8551 nM, a 128-fold increase relative
to (R,R)-1, EC50cardio = 83 nM, Table 3. Thus it appears that steric
interactions involving the chiral center on the aminoalkyl portion
of the fenoterol molecule had a greater effect in b2-AR associated
cardiomyocyte contractility system than in the displacement bind-
ing or cAMP stimulation studies. It is interesting to note that the
activity of (R,R)-52 in the cardiomocyte contractility model was
antagonized by the b2-AR selective antagonist ICI 118,551, but
(R,R)-52 did not antagonize the activity of zinterol in this model
(data not shown).
When the effect of the change in the orientation of a naphthyl
moiety on the amionoalkyl portion of the molecule from the
1- to the 2-position on the ring, 5 and 53, was investigated, the
data indicate that this change had only a slight effect on the bind-
ing affinities to the b2-AR and b1-AR, Table 2. In these studies, the
differences between the Ki(b2-AR) for the (R,R)- and (R,S) isomers of
53 and the corresponding isomers of 5 were less than twofold as
was the difference in the Ki(b1-AR) for (R,S)-53 and (R,S)-5. The
greatest effect was seen in the Ki(b1-AR) of (R,R)-53 which was
ꢁ4-fold higher relative to that of (R,R)-5, and reflected in an
increase in the relative binding affinities, Ki(b1-AR)/Ki(b2-AR) from
14 (5) to 40 (53). The results suggest that the change from a
1-naphthyl to a 2-naphthyl substituent on the fenoterol molecule
2. Results
In the previous study, the CoMFA(n = 26) model was derived
using the experimentally determined b2-AR binding affinities of a
training set of 26 fenoterol derivatives and stereoisomers.3 The
validity parameters of the model were high (R2 = 0.920,
Q2 = 0.847, F = 60.380) and the standard error of prediction was
low (SEP = 0.309). In the current study, the CoMFA(n = 26) model
was used to estimate the pKi values for the (R,R)-isomers of a set
of 12 new molecular structures, and the (R,S)-, (S,R)- and (S,S)-iso-
mers of three of these compounds for a set of 21 unique com-
pounds, 52–63, Table 1.
The (R,R)- and (R,S)-isomers of three of the designed fenoterol
analogues, 52, 53, 54, were synthesized using
a previously
described synthetic scheme in which the epoxide formed from
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin was coupled with
the (R)- or (S)-enantiomer of the appropriate N-benzylamino-al-
kanes.3 The chirality of the first asymmetric center was set as (R)
based on the data from the previous study that fenoterol analogues
with this configuration had higher affinities and activities at the
b2-AR relative to the corresponding analogues with a (S)-configura-
tion at this carbon.3
The binding affinities of the synthesized compounds were
determined using membranes derived from HEK cells stably trans-
fected with human b2-AR and b1-AR as previously described3 and
the corresponding Ki values were calculated, Table 2. The maxi-
mum difference,
D value, between computationally predicted
and experimentally determined affinities, expressed as pKi values
(ꢀlog Ki), was ꢀ0.6 and in some cases did not exceed the standard
error of experimental determination, Table 1. The external valida-
tion of the CoMFA(n = 26) model with the newly synthesized test
had a relatively small effect on the
p–p and/or p-hydrogen bond
interactions between the naphthyl substituent and aromatic resi-
dues located in this area of the binding site of the b2-AR. The data
also suggest that these interactions occur between the agonists and
the b1-AR.
While the change from a 1-naphthyl to a 2-naphthyl substituent
has less than a twofold affect on the Ki(b2-AR) of the (R,R)-isomers, a
greater effect was seen in the EC50cAMP values as the EC50cAMP of
(R,R)-53 was 30-fold lower than that of (R,R)-5, Table 3. However,
the inverse was observed with cardiomyocyte contractility as the
EC50cardio of (R,R)-53 was threefold higher than that of (R,R)-5, Table
3, but it is not clear if this difference has any pharmacological signif-
icance. There was less than a threefold difference in the EC50cAMP
values (R,S)-53 and (R,S)-5, Table 3.
The data from the initial studies of the aminoalkyl analogues of
fenoterol, indicated that optimum binding affinities and b2-AR
selectivities were observed with an electronegative substituent in
the 4-position of the benzyl moiety or with a 1-naphthyl moiety.3
In this study, a series of 4-substituted-1-naphthyl analogues of fen-
oterol, 54–63, were docked in the CoMFA(n = 26) model and their
pKi values predicted, Table 1. The results indicated that the pres-
ence of an electronegative substituent increased the binding affin-
ity of the compound and that the pKi values of the (R,R)-isomers of
54–59 would be greater than the corresponding values of 1, 2 and
5. It is of interest to note that 4-amino-1-naphthylfenoterol (60)
had the lowest predicted pKi which is consistent with the previous
set was calculated and the results, R2PRED ¼ 0:583, SPRESS = 0.337,
P
where PRESS is predictive residual sum of squares,
(D
2)7, indi-
cated that the CoMFA(n = 26) model had a high predictive power
for the computational design of congeneric b2-AR agonists based
upon fenoterol.
The b2-AR and b1-AR binding affinities, Ki(b2-AR) and Ki(b1-AR),
and relative binding affinities, Ki(b1-AR)/Ki(b2-AR), of the (R,R)- and
(R,S)- isomers of 52, 53 and 54, were determined, Table 2. In addi-
tion, the b2-AR agonist activities of the (R,R)-isomers of the previ-
ously reported compounds 1, 2, 5 and 52, 53, 54 were examined
using a previously described cardiomyocyte contractility model,
Table 3.3 In these studies, the agonist-stimulated b2-AR contractil-
ity response was distinguished from that of b1-AR activity using ICI
118,551, a b2-AR specific antagonist.
The data demonstrate that the substitution of the methyl group
by an ethyl moiety at the chiral center on the aminoalkyl portion of
the fenoterol backbone, 1 and 52, respectively, reduced the
calculated binding affinities to the b2-AR. The Ki(b2-AR) value for
(R,R)-52 was ꢁ4-fold greater than that of (R,R)-1 and the value
for (R,S)-52 was 1.5-fold greater than that of (R,S)-1, Table 2. The
binding affinities to the b1-AR were also affected, as the Ki(b1-AR)
of (R,R)-52 was 28-fold higher then that of (R,R)-1 and the Ki(b1-
AR) for (R,S)-52 was sixfold greater than that of (R,S)-1. The greater

730
K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
Table 1
Computationally predicted pKi values {pKi(CoMFA)} for the compounds designed using the previously reported CoMFA(n = 26) model3
No.
Structure
Stereochemistry/substitution
pKi CoMFA
pKi exper.
D
Old compounds
(R,R)
(R,S)
(S,R)
(S,S)
5.84
5.48
5.02
4.66
6.46
5.43
4.99
4.56
ꢀ0.62
OH
OH
0.05
NH
NH
HO
0.03
0.10
1
2
3
CH3
OH
OH
OH
(R,R)
(R,S)
(S,R)
(S,S)
6.17
5.80
5.34
4.99
6.32
5.71
5.28
4.80
ꢀ0.15
0.09
HO
HO
0.06
0.19
CH3
O
CH3
(R,R)
(R,S)
(S,R)
(S,S)
5.57
5.21
4.75
4.39
5.53
5.10
4.64
4.54
0.04
0.11
0.11
ꢀ0.15
OH
OH
NH
NH
CH3
OH
OH
(R,R)
(R,S)
(S,R)
(S,S)
6.72
6.36
5.90
5.54
6.62
6.47
5.75
5.60
0.10
ꢀ0.11
0.15
HO
ꢀ0.06
5
CH3
New compounds
(R,R)
(R,S)
(S,R)
(S,S)
5.91
5.47
5.08
4.65
5.895
5.240
0.015
0.23
OH
OH
NH
HO
52
H3C
OH
OH
OH
(R,R)
(R,S)
(S,R)
(S,S)
6.04
5.7
5.2
5.2
6.393
6.293
ꢀ0.353
ꢀ0.593
NH
NH
HO
53
CH3
(R,R)
(R,S)
(S,R)
(S,S)
6.93
6.6
6.24
5.91
6.556
6.498
0.374
0.102
OH
HO
54
CH3
O
CH3
OH
55
56
R = OH
R = F
6.79
6.76
OH
57
58
R = Cl
R = Br
6.92
6.94
6.95
5.79
6.41
6.38
6.40
NH
HO
59
60
61
62
63
R = CF3
R = NH2
R = CH3
R = SH
R = SCH3
CH3
R
OH
(R,R) isomers only
study in which the Ki(b2-AR) of (R,R)-4-aminobenzylfenoterol (4)
was greater than the corresponding values for 1, 2 and 5.3
In this study, (R,R)-54 and (R,S)-54 were synthesized and tested.
The data obtained with (R,R)-54 indicate that the b2-AR binding

K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
731
Table 2
considered, then the data suggest that fenoterol analogues contain-
ing a p-oxy-phenyl moiety have a greater effect on cAMP stimula-
tion than fenoterol analogues containing a 1-naphthyl moiety, 5,
and that the addition of a 4-methoxy substituent in the naphthyl
ring has a positive effect on the EC50cAMP value, 54. It is of interest
to note that there is no significant difference in the EC50cAMP values
of (R,R)-2 and (R,R)-53, which contains a 2-naphthyl ring. This sug-
gests that the difference in the EC50cAMP values between 2 and 5 is a
function of the relative position of the naphthyl group within the
b2-AR binding area rather than the lack of a p-oxy-substituent.
When the effect on cardiomyocyte contractility was investi-
gated, the relative EC50cardio values were 54 < 5 < 2, although the
difference between 54 and 2 was only ꢁ10-fold. These results indi-
cate that the presence of the 1-naphthyl ring and the 4-methoxy
substituent had a synergistic effect in this model system.
No significant differences were observed in the cAMP stimula-
tion produced by the (R,S)-isomers of 2, 5, and 54, Table 3. For
these compounds, the relative differences between the (R,R)- and
(R,S)-isomers of each compound mirrored the relative differences
in Ki(b2-AR) values, that is, (R,R)-2 < (R,S)-2; (R,R)-5 > (R,S)-5;
(R,R)-54 ꢂ (R,S)-54, Tables 2 and 3. These results suggest that the
displacement binding studies and the induced stimulation of cAMP
accumulation may reflect the same binding process.
Experimentally determined binding affinities to the b2-adrenergic receptor (Ki b2-AR)
and b1-adrenergic receptor (Ki b1-AR) of the compounds synthesized in this study and
previously reported compounds, the relative b1-AR and b2-AR binding affinities (Ki b1/
Kib2)
Ki b1-AR (
l
M)
Ki b2-AR (
l
M)
Ki b1/Ki b2
(R,R)-1
(R,S)-1
(R,R)-2
(R,S)-2
(R,R)-5
(R,S)-5
(R,R)-52
(R,S)-52
(R,R)-53
(R,S)-53
(R,R)-54
(R,S)-54
14.8 2.5a
18.9 2.4a
21.9 3.1a
30.8 2.5a
0.35 0.03a
3.69 0.25a
0.47 0.04a
1.93 0.14a
0.24 0.03a
0.34 0.03a
1.27 0.81
5.76 0.83
0.41 0.09
0.51 0.06
0.28 0.01
0.32 0.01
43a
5a
47a
16a
14a
46a
334
20
3.35 0.13a
15.8 2.5a
424.3 70.7
116.7 44.4
16.6 1.6
40
53
573
46
27.1 2.6
160.5 54.1
14.8 1.6
a
Kib2-AR and Kib1-AR values obtained from Ref. 3.
Table 3
The activity (presented as EC50 values) of the (R,R)- and (R,S)-isomers of fenoterol and
selected fenoterol analogues in the stimulation of cAMP accumulation in HEK cells
containing human b2-adrenergic receptors (EC50cAMP
contractility model system (EC50cardio
) and in a cardiomyocyte
)
The pKi data from this study was used to refine the
CoMFA(n = 26) model. A total of 32 structures (26 original and 6
newly synthesized) were used as the training set for a revised
3D-QSAR model, CoMFA(n = 32). Partial Least Square methodology
extracted four statistically significant components of the model,
and all of the validation parameters were high and indicated that
the model had a stronger predictive power than the CoMFA(n = 26)
model, Table 4.
Induced stimulation of
cAMP accumulation
Cardiomyocyte contractility
EC50cAMP (nM)
% Stimulation
123
EC50cardio (nM)
(R,R)-1
(R,S)-1
(R,R)-2
(R,S)-2
(R,R)-5
(R,S)-5
(R,R)-52
(R,S)-52
(R,R)-53
(R,S)-53
(R,R)-54
(R,S)-54
0.30 0.09
4.70 0.50
0.30 0.23
2.00 0.45
12.50 3.50
2.70 0.84
2.80 0.90
16.60 4.90
0.40 0.12
7.60 2.76
3.90 1.80
4.00 1.29
9
73 18
131 30
136 11
120 19
129 24
112 42
575 122
186 50
506 107
59 16
103 34
8551 4992
125
98
6
4
Table 4
Quantitative comparison of the refined CoMFA(n = 32) model with the original model,
CoMFA(n = 26)3
90 10
89 12
106 11
118 30
133 25
CoMFA(n = 26)
CoMFA(n = 32)
16
4
n
26
32
PLS components
4
4
R2
0.920
60.380 (4, 21)
0.847
0.223
0.309
48.1%
51.9%
0.925
64.984 (4, 27)
0.851
0.216
0.305
38.3%
61.7%
F (n1, n2)
Q2 (loo)
affinity of this compound was equivalent to (R,R)-5 and slightly
stronger than (R,R)-2, Table 2. The same trend was observed with
the (R,S)-isomers as the Ki(b2-AR) values of (R,S)-54 and (R,S)-5
were equivalent and both were about sixfold stronger than (R,S)-
2. The results suggest that when both a p-methoxy moiety and a
1-naphthyl moieties are present in the molecule, interactions with
the naphthyl group appear to play a more significant role in the
stabilization of the agonist-b2-AR complex.
SEE
SEP
Electrostatic contribution
Steric contribution
The Ki(b1-AR) value of (R,R)-54 was significantly weaker than
(R,R)-2 (19-fold) and (R,R)-5 (141-fold), while there was no signif-
icant difference between (R,S)-54 and the (R,S)-isomers of 2 and 5,
Table 2. The weaker b1-AR binding affinity for (R,R)-54 was re-
flected in the relative binding affinity ratios, Ki(b1-AR)/Ki(b2-AR)
which increased to 573, as compared to 47 and 14, for (R,R)-2
and (R,R)-5, respectively. The results indicate that both the p-meth-
oxy and 1-naphthyl moieties in (R,R)-54 had a destabilizing effect
in the binding interactions with the b1-AR relative to the (R,R)-iso-
mers of 2 and 5. This effect was not observed in the relative binding
of the (R,S)-isomers of 54, 2 and 5 to the b1-AR suggesting that the
S-configuration in the aminoalkyl portion of the molecule altered
the role of the 4-oxy and/or 1-naphthyl substituents in the binding
of (R,S)-54 to the b1-AR.
Figure 2. The CoMFA(n = 32) model with (R,R)-54 included in the figure. The fields
are color coded in the following manner: green represents a region of favorable
steric (bulk) interactions, unfavorable steric interactions are denoted by yellow,
electrostatic interactions with a positive charge (or H-bond donors) are blue, and
electrostatic interactions with a negative charge (or H-bond acceptors) are red.
While the Ki(b2-AR) values of the (R,R)-isomers of 2, 5 and 54
were essentially equivalent, the EC50cAMP values of 5 and 54 were
significantly higher, 40-fold and 10-fold, respectively, than the
EC50cAMP value of 2, Table 3. If the EC50cAMP value of (R,R)-1 is also

732
K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
The CoMFA(n = 32) model is presented in Fig. 2. For visual clar-
binding studies with the wild-type b1-AR and b2-AR and salmeterol
and a salmeterol derivative that lacked an ether linkage found in
the parent compound.5 The Ki(b2-AR) of salmeterol and the deriv-
ative were 1.5 nM and 220 nM, respectively, while the Ki(b1-AR)
values were both ꢁ2300 nM. The differences in the Ki(b2-AR) val-
ues were reflected in the Ki(b1-AR)/Ki(b2-AR) ratios which were
>1000 for salmeterol and ꢁ10 for the derivative. The authors of this
study concluded that the large difference in the relative binding ra-
tios was a result of the inability of the derivative to interact with
Y308 in the b2-AR due to the absence of the oxygen atom of the
ether linkage.
ity only (R,R)-54 is included in the figure. The interaction fields are
color coded in the following manner: green represents a region of
favorable steric (bulk) interactions, unfavorable steric interactions
are denoted by yellow, electrostatic interactions with a positive
charge (or H-bond donors) are blue, and electrostatic interactions
with a negative charge (or H-bond acceptors) are red.
As previously demonstrated with the CoMFA(n=26) model, the
CoMFA(n = 32) model explains the source of stereoselectivity ob-
served with this group of compounds. At the first center of chiral-
ity, an R-configuration directs the hydroxyl moiety towards the
negative charge field (red), a favorable interaction, while an S-con-
figuration at this center directs the hydroxyl moiety towards the
positive charge field, a destabilizing interaction. Steric fields are
associated with both chiral centers and an S-configuration at both
of the sites directs substituents toward an unfavorable steric inter-
action site, denoted by the yellow fields.
The >500-fold b2-AR selectivity observed with (R,R)-54 and the
>40-fold b2-AR selectivity observed with compounds that do not con-
tain p-oxygen moieties, that is, (R,S)-5, (R,R)-53, (R,S)-53, Table 2,
indicate that
a role in the relative stabilities of the two b-ARs. The contribution
of and/or -hydrogen bond interactions in the binding process
p–p and/or p-hydrogen bond interactions also play
p–p
p
The interactions of the aminoaryl portion of the fenoterol ana-
logues with the b2-AR and the relative stabilities of the agonist-
receptor complexes were also refined in the CoMFA(n = 32) model.
These interactions are illustrated by the proposed stabilizing inter-
actions associated with the 4-methoxy-1-naphthyl substituent of
(R,R)-54, Fig. 2. In this model, the substituent is placed in the vicin-
ity of electrostatic interactions with a positive charge or H-bond
donors (blue regions), suggesting hydrogen bond interaction with
the oxygen atom. Additional stabilizing interactions are suggested
by the presence of two large electronegative fields (red regions)
surrounding both faces of the naphthyl ring of the (R,R)-54 indicat-
is indicated by the two electronegative fields in the CoMFA(n = 32)
model, which are located on both faces of the docked (R,R)-54,
Fig. 2. The electronegative fields in the CoMFA(n = 32) model
reflect a network of
p–p interactions which include Y308 in the
b2-AR and other aromatic residues such as H296, Y316, F289,
F193, F290 and W109 which are in close proximity to Y308.9 The
higher Ki(b1-AR) values for the same compounds suggest that
p–p interaction networks based upon F325 and other aromatic res-
idues also exist in the b1-AR, but that the interactions are weaker
than the corresponding ones in the b2-AR. In addition, the presence
of extended electronegative fields in both b-ARs also explains the
lack of a significant difference in the affinities of the 1-naphthyl
and 2-naphthyl analogues, except for the interaction of (R,R)-54
with the b1-AR. In this case, the 4-methoxy substituent may play
a more significant role in the increased Ki(b1-AR) value than the
1-naphthyl ring.
In the initial study involving the stereoisomers of fenoterol and
fenoterol analogues3 and with the compounds synthesized in this
study, the b2-AR binding affinities of the (R,R)-isomers were greater
than the corresponding (R,S)-isomers. The results support the
hypothesis that the stereochemistry of the chiral carbon on the
aminoalkyl moiety positions this portion of the molecule relative
to the interactive moieties in TM3, TM6 and TM7 and that this
effect reflects interactions with a sterically-restricted site on the
b2-AR.3 These data indicate that this effect was more pronounced
with compounds containing a phenyl ring with a p-hydroxy or
p-methoxy substituent, that is, 1, 2, 52, Table 2. This may reflect
the fact that the hydrogen bonding interactions between the
p-substituents on 1, 2 and 52 and the b2-AR occur at more defined
sites, that is, Y308 and H296, than the sites binding to non-oxygen
ing the possibility of p–p interactions with aromatic residues with-
in the binding site. Steric favoring regions (green regions) are also
present at both edges of the naphthyl ring, which most probably
reflects the preference of 1-naphthyl versus 2-naphthyl in the
b2-AR binding. These regions may also affect the conformational
mobility of a molecule during the optimization of electrostatic
and
p–p interactions. The three interaction sites are independent
and appear to act cooperatively in the binding of (R,R)-54.
3. Discussion
The results of this study demonstrate that the CoMFA(n = 26)
model can be used to design selective b2-AR agonists based upon
fenoterol or compounds that contain interactive sites equivalent
to those found on the aminoalkyl portion of the fenoterol molecule.
This design strategy is reflected by the synthesis and properties of
the b2-AR agonist (R,R)-54, which is highly selective (Kib1/Kib2 =
573) and effective (EC50 = 16 nM, cardiomyocyte contractility).
The revised molecular model, CoMFA(n = 32) is consistent with
the earlier model3 and with the previously reported models
describing the molecular basis of b2-AR selectivity,4–6 although a
recent paper has questioned the role that Y308 plays in this pro-
cess.8 The data from this study supports the hypothesis that hydro-
gen bond formation between the p-oxygen moiety on the aromatic
substituents of the (R,R)-isomers of 1, 2, 52, and 54 and Y308 in
TM7 and/or H296 in TM6 contribute to the >40-fold b2-AR selectiv-
ity. The role of Y308 in b2-AR selective binding was also supported
by the data from a study of the binding of salmeterol to chimeric
b1/b2-ARs and alanine-substituted b2-AR mutants.5 In these stud-
ies, the Ki(b2-AR) changed from 1.5 nM in the wild-type receptor
to 184 nM in the Y308A mutant, the largest effect observed in
the series of alanine-substituted b2-AR mutants used in the study.
The data also indicate that the >300-fold b1-AR/b2-AR selectivity
of (R,R)-52 and (R,R)-54 was primarily due to a large decrease in
b1-AR binding affinities, Table 2. It is reasonable to assume that
the weaker affinities reflect the loss of hydrogen bond interactions
due to the replacement of Y308 and H296 in the b2-AR by F359 and
K347, respectively, in the b1-AR.6 This hypothesis is supported by
containing fenoterol analogues, that is, a diffuse
p–p interaction
area. Therefore, subtle changes in the binding interactions pro-
duced by alterations in the relative positions of the (R,R)- and
(R,S)-isomers should have a greater effect in 1, 2, and 52 than 5
and 53. The minimal differences between the Ki(b2-AR) values of
(R,R)-54 and (R,S)-54 suggests that the p–p interactions play a rel-
atively larger role in the stabilization of the 54-b2-AR complex than
the hydrogen bonding interactions.
In this study, the Ki(b1-AR) values of the (R,R)-isomers of 1, 2, 5,
and 53 were lower than the corresponding (R,S)-isomers although
the calculated differences were <5-fold, Table 2. The opposite effect
was observed with 52 and 54 where the Ki(b1-AR) of the (R,R)-iso-
mers were 4-fold and 10-fold larger, respectively, Table 2. The
Ki(b1-AR)/Ki(b2-AR) ratios of 52 and 54 were also significantly
greater than those observed with the other fenoterol analogues,
by 334- and 573-fold, respectively. The results obtained with 52
may reflect the effect of the increase in steric bulk at the chiral
center and the resulting destabilization of the 52-b1-AR complexes,
irrespective of the chirality, as both isomers have Ki(b1-AR) values

K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
733
>100
l
M. This suggests that the postulated sterically-restricted
The binding of an agonist to the b2-AR has been previously de-
scribed as a multi-step sequential process that produces a series of
conformational changes in the receptor.14,15 This descriptionreflects
the extensive study of the binding of agonists to the ‘first’ binding
area and it is important to note that the majority of the compounds
used in these studies did not have large amino alkyl side chains con-
taining oxygen or aromatic moieties, cf. Ref. 13. Thus it is reasonable
to assume that with compounds such as the fenoterol analogues, the
sequential binding mechanism will also include interactions with
the ‘second’ binding area and that these interactions will also result
in distinct conformational changes in the b2-AR molecule. However,
the data from the current study are unable to differentiate between
these biding interactions. This is in part due to the fact that the bind-
ing studies were conducted using [3H]CGP-12177, a highly selective
b2-AR antagonist, which, unfortunately, does not contain a feno-
terol-like alkylaminomoiety.16 This issue is being addressed in bind-
ing studies using a [3H]-labeled fenoterol analogue as the marker
ligand and in molecular modeling and thermodynamic studies,
which have also been initiated.
site on the b2-AR also exist in the b1-AR.
When the results obtained with 54 are compared to 2 and 5, it is
evident that for (R,S)-54, the presence of both a 4-methoxy moiety
and a 1-naphthyl moiety had little effect on the Ki(b1-AR) relative
to the presence of only a 4-methoxy moiety or a 1-naphthyl moi-
ety, (R,S)-2 and (R,S)-5, respectively, Table 2. The more significant
effect is seen with (R,R)-54 as the calculated Ki(b1-AR) increased
from 21.9 lM (2) and 3.35 lM (5) to 160.5 lM. The results indicate
that the presence of the 4-methoxy and 1-napthyl moieties act in a
synergistic manner to reduce the stability of the (R,R)-54-b1-AR
complex. The mechanism associated with this phenomenon is cur-
rently unknown.
The 12 compounds used in this study stimulated cAMP accumu-
lation with EC50(cAMP) values that ranged from 0.30 nM to
16.60 nM, Table 3, and all of these compounds can be classified
as full b2-AR agonists. There was no direct relationship between
the Ki(b2-AR) and EC50(cAMP) values probably due to the fact that
the binding studies used a radiolabeled antagonist. Planned studies
with a radiolabeled fenoterol analog may produce a better correla-
tion between binding affinity and stimulation of cAMP accumula-
tion. The fact that fenoterol analogs stimulate cAMP affinity at
considerably lower concentrations than they stimulate cardiomyo-
cyte contractility suggests either a far greater degree of amplifica-
tion in the cAMP activity pathway, or potentially a different signal
transduction pathway altogether.
In the cardiomyocyte contractility model system, 8 of the 9
compounds studied had stimulatory effects with EC50cardio values
ranging from 16 ((R,R)-54) to 575 nM ((R,S)-1), Table 3. The clear
outlier was (R,R)-52 with an EC50cardio value of 8551 nM. When
the Ki(b2-AR) and EC50cardio values for the 8 compounds were com-
pared, there was a statistically significant relationship between the
values, r2 = 0.8804, p = 0.006. The data indicate that in a series of
4. Experimental section
4.1. b2-AR binding assays
Compounds synthesized in this study were tested up to three
times each to determine their binding affinities at the b1- and
b2-ARs following
a
previously described approach.3 In brief,
b2-AR binding was conducted on membranes derived from HEK
cells containing human b2-AR (provided by Dr. Brian Kobilka, Stan-
ford Medical Center, Palo Alto, CA). Cells were grown in Dulbecco’s
Modified Eagle Medium (DMEM) containing 10% fetal bovine ser-
um (FBS) and 0.05% penicillin–streptomycin with 400 lg/mL
G418. b1-AR binding was conducted using rat cortical membranes.
The binding assays contained 0.3 nM or 1.4 nM [3H]CGP-12177 in a
volume of 1.0 mL for binding to b2- and b1-ARs, respectively. Non-
fenoterol analogues a Ki(b2-AR) of ꢃ0.500
lM should correspond
to a EC50cardio of <0.200 M and that the CoMFA models described
l
specific binding was determined using 10 lM propranolol. To
block the b2-AR sites present in the cortical membrane preparation,
30 nM ICI 118-551 was added to the assay buffer.
Competition curves with standard and unknown compounds in-
cluded at least six concentrations, in triplicate. IC50 values and Hill
coefficients were calculated using Prism software. Ki values were
calculated using the Chang-Prusoff transformation.17
in these studies are capable of designing compounds which are ac-
tive agonists in the b2-AR cardiomyocyte contractility model. The
EC50cardio value obtained with (R,R)-52 was similar to previous data
obtained with (S,R)-1 and (S,R)-2, EC50(cardio) = 2340 nM and
3160 nM, respectively.10 It is of interest to note that all three com-
pounds can be classified as full b2-AR agonists with respect to
cAMP stimulation, EC50(cAMP) = 2.80 ((R,R)-52), 8.50 nM ((S,R)-1),
7.20 nM ((S,R)-2). The data indicate that the chirality at both of
the chiral centers on the fenoterol backbone and the steric bulk
at the second chiral center play limited roles in cAMP stimulation,
but a major role in cardiomyocyte contractility. The role of the chi-
ral centers in cardiomyocyte contractility has been previously
demonstrated in which (R,R)-1 and (R,R)-2 were shown to prefer-
entially activate GS signaling while the corresponding (S,R)-isomers
activated both GS and Gi proteins.10 Our initial data indicate that
the cardiomyocyte contractility of (R,R)-52 is consistent with the
selectivity of the (S,R)-isomers as this activity is sensitive to pertus-
sis toxin, which indicates that this compound activates both GS and
Gi proteins (data not shown).
The data from this study are consistent with the hypothesis
that there are two binding areas within the b2-AR molecule.
One area is the site that interacts with the ‘catechol’ portion of
an agonist and is created by residues in TM3, TM5 and TM6,
(the ‘first’ binding area).7,11–15 The ‘‘second” binding area is cre-
ated by residues in TM3, TM6 and TM7 and interacts with the
aminoalkyl side chain of the fenoterol analogues.3–6 The results
also suggest that the binding process between the fenoterol ana-
logues and the b2-AR involves a number of multi-step interac-
tions. In addition, these interactions and their functional
consequences vary based upon the composition and stereochem-
istry of the fenoterol analogue.
4.2. b2-AR mediated stimulation of cAMP accumulation
The HEK cells expressing b2-ARs were grown on 96-well plates
in DMEM containing 10% FBS, 0.05% penicillin–streptomycin, and
400 lg/ml G418. When the cells reached confluence, the medium
was removed and each well was rinsed with 0.1 mL of Krebs-
HEPES buffer (130 mM NaCl, 4.8 mM KCl, 1.2 mM KH2PO4,
1.3 mM CaCl2, 1.2 mM MgSO4, 25 mM HEPES, and 10 mM glucose,
pH 7.3). The test compound was diluted in Krebs-HEPES buffer
containing 0.1% ascorbic acid, 10 lM pargyline, and 50 lM 3-iso-
butyl-1-methylxantine (IBMX), and 0.1 mL of the solution was
added to each well. In the agonist assay, the plates were preincu-
bated for 10 min at room temperature (rt) with buffer alone; then
test compound diluted in buffer was added to the wells. In the
antagonist assay, the plates were preincubated (for 20 min at
37 °C) with test compound diluted in buffer; then buffer plus
30 nM formoterol was added. The plates were incubated for an
additional 10 min with the test compound. After incubation, the
medium was removed and 0.1 mL of 0.5 M formic acid added. After
a minimum of 1 h, the supernatant was removed and lyophilized.
cAMP was quantified using the protein kinase binding assay of
Gilman.18 The amount of protein per well was determined and
used to calculate the amount of cAMP/mg/well.

734
K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
4.3. Cardiomyocyte contractility
4.6. Synthesis of (R,R)- and (R,S)-fenoterol analogues
These experiments were carried out as previously described.3 In
brief, single ventricular myocytes were isolated from 2–4 month
old rat hearts by a standard enzymatic technique. The isolated cells
were resuspended in HEPES buffer solution [20 mM, pH 7.4] con-
taining, NaCl (137 mM), KCl (5.4 mM), MgCl2 (1.2 mM), NaH2PO4
(1.0 mM), CaCl2 (1.0 mM), and glucose (20 mM). All experiments
were performed within 8 h of cell isolation.
The cells were placed on the stage of an inverted microscope
(Zeiss model IM-35, Zeiss, Thornwood, NY), perfused with the
HEPES- buffered solution at a flow rate of 1.8 mL/min, and electri-
cally stimulated at 0.5 Hz at 23 °C. Cell length was monitored by an
optical edge-tracking method using a photodiode array (Model
1024 SAQ, Reticon, Boston, MA) with a 3 ms time resolution. Cell
contraction was measured by the percent shortening of cell length
following electrical stimulation.
The fenoterol analogues used in this study were prepared
following
a previously reported synthetic scheme in which
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin or (S)-(+)-30,50-dib-
enzyloxyphenylbromohydrin was condensed with the appropriate
free N-benzylaminoalkane.3 The present work only utilized
(R)-(ꢀ)-30,50-dibenzyloxy-phenylbromohydrin, which was synthe-
sized as previously reported.3 The syntheses of the new N-ben-
zylaminoalkanes and fenoterol analogues are reported below.
4.7. General procedure for the preparation of aryl ketones
The appropriate phenylacetic acid and alkyl anhydride were
combined in pyridine and heated to reflux under argon for 6 h. Sol-
vents were removed and the residue was partitioned between
dichloromethane and 1 N NaOH. The organic fraction was washed
with water, dried (Na2SO4), filtered, and evaporated. Silica gel
chromatography (1:30) eluting with 1/1 dichloromethane/hexanes
affords the pure aryl ketone.
4.4. Comparative molecular field analysis (CoMFA)
SYBYL 8.0. (TRIPOS Inc., St. Louis, MO) was used for 3D QSAR
modeling. The previously described CoMFA model3 was used for
pKi prediction for the set of newly designed derivatives. Briefly,
molecular models of structures were prepared in HyperChem v.
6.03 (HyperCube Inc., Gainesville, FL) using ModelBuild procedure
to ensure the same conformation of the common scaffold. The
models were extracted to SYBYL and the Gasteiger—Huckel atomic
charges were calculated. The models were aligned with molecules
of original training set using the two asymmetric carbon atoms in
the core of the fenoterol molecule (–C –CH2–NH–C –CH2–) as a
common substructure. A PREDICT procedure implemented in CoM-
FA package was used to compute the estimated pKi values.
The revised CoMFA(n = 32) model was developed using a train-
ing set that included the new structures and pKi values examined
in this study, that is, the (R,R)- and (R,S) isomers of 52, 53, 54.
Two types of molecular fields (steric and electrostatic) were sam-
pled on the grid (2 Å spacing) lattice surrounding each structure.
Distance-dependent dielectric constant was used in electrostatic
calculations and energetic cutoffs of 30 kcal/mol for both the steric
and the electrostatic energies were set.
4.7.1. 1-(4-(Benzyloxy)phenyl)butan-2-one
Prepared from 7.5 g (31 mmol) of 4-benzyloxyphenylacetic acid
(Aldrich) and 17.2 mL of propionic anhydride in 17 mL of pyridine.
Yield 1.92 g (24%). 1H NMR (CDCl3) d 1.00 (t, 3H, J = 7.5 Hz), 2.44 (q,
2H, J = 7.5, Hz), 3.59 (s, 2H), 5.02 (s, 2H), 6.90 (d, 2H, J = 8.7 Hz),
7.09 (d, 2H, J = 8.7 Hz), 7.25–7.40 (m, 5H) ppm.
4.7.2. 1-(Naphthalen-2-yl)propan-2-one
*
*
Prepared from 25.0 g (134 mmol) of 2-naphthylacetic acid
(Aldrich) and 100 mL of acetic anhydride in 100 mL of pyridine.
Yield 3.68 g (15%). 1H NMR (CDCl3) d 2.19 (s, 3H), 3.86 (s, 2H),
7.35 (dd, 1H, J = 2.1, 8.7 Hz), 7.45–7.50 (m, 2H), 7.68 (s, 1H),
7.79–7.85 (m, 3H), ppm.
4.7.3. 1-(4-Methoxynaphthalene-1-yl)propan-2-one
Prepared from 5.04 g (23.3 mmol) of 4-methoxy-1-naphthylace-
tic acid (ACB Blocks Ltd.) and 30 mL of acetic anhydride in 30 mL of
pyridine. Yield 1.54 g (31%). 1H NMR (CDCl3) d 2.09 (s, 1H), 4.01 (s,
3H), 4.02 (s, 2H), 6.78 (d, 1H, J = 7.8 Hz), 7.29 (d, 1H, J = 7.8), 7.47–
7.56 (m, 2H), 7.78–7.82 (m, 1H), 8.30–8.33 (m, 1H) ppm.
The application of the partial Least Square correlation procedure
to the resultant database extracted four significant components and
validation parameters for this solution (CoMFA(n = 32)), which are
presented in Table 4. For the sake of comparison the validity param-
eters obtained for the original solution (CoMFA(n = 26))3 are also
included in Table 4.
4.8. General procedure for the preparation of N-benzylamino-
alkanes
To the appropriate aryl ketone in dichloromethane (c = 0.5 M),
cooled to 0 °C was added glacial acetic acid (1 equiv), followed by
benzylamine (1 equiv) and Na(AcO)3BH (1.4 equiv). The reaction
mixture was warmed to RT and stirred under argon for 20 h.
Cooled reaction mixture (ice bath) and added dropwise 10% NaOH
(5 equiv), then extracted into dichloromethane, washed with brine,
dried (Na2SO4), filtered, and evaporated. The resulting product was
used without further purification.
4.5. Chemistry
All reactions were carried out using commercial grade reagents
and solvents. Ultraviolet spectra were recorded on a Cary 50 Concen-
tration spectrophotometer. Optical rotations were done at 25 °C on a
Rudolph Research Autopol IV. NMR Spectra were recorded on a Var-
ian Mercury VMX 300-MHz spectrophotometer. In reporting the
NMR multiplicities, we used the following abbreviations: s, singlet;
d, doublet; t, triplet; q, quartet; p, pentet; m, multiplet; apt., appar-
ent; and br, broad. Low resolution mass spectra (MS) were obtained
on a Finnigan LCQDuo LC MS/MS system equipped with an electro-
spray ionization (ESI) probe. High resolution mass spectra (HRMS)
were obtained by the University of Minnesota Mass Spectrometry
Service, also with an ESI probe. Analytical HPLC data were obtained
using a Waters 2690 Separations Module with PDA detection. Meth-
od (a): Varian Sunfire 100 ꢄ 4.6 mm C18 column. Method (b): Ther-
moHypersil BDS 100 ꢄ 4.6 mm C18 column. Method (c): Chiralpak
IA 250 ꢄ 10 mm. Merck silica gel (230–400 mesh) was used for open
column chromatography.
4.8.1. rac-N-Benzyl-1-(4-(benzyloxy)phenyl)butan-2-amine
Prepared from 1.88 g (7.4 mmol) of 1-(4-(benzyloxy)phenyl)-
butan-2-one, 0.81 mL (7.4 mmol) of benzylamine, 2.40 g (11.3
mmol) of Na(OAc)3BH and 0.5 mL of acetic acid. Yield 2.55 g
(100%). 1H NMR (CDCl3) d 0.93 (t, 3H, J = 7.2 Hz), 1.45–1.55 (m,
2H), 2.60–2.75 (m, 2H, 1H), 3.76 (dd, 2H, J = 6.6, 9.9 Hz), 5.03 (s,
2H), 6.88 (d, 2H, J = 8.7 Hz), 7.06 (d, 2H, J = 8.7 Hz), 7.18–7.45 (m,
10H) ppm. MS (ESI+) m/z (rel): 346 (100, M+H).
4.8.2. rac-N-Benzyl-1-(naphthylen-2-yl)propan-2-amine
Prepared from 3.68 g (20 mmol) of 1-(naphthalen-2-yl)propan-
2-one, 2.18 mL (20 mmol) of benzylamine, 6.35 g (30 mmol) of

K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
735
Na(OAc)3BH and 1.3 mL of acetic acid. Yield 5.04 g (92%). 1H NMR
(CDCl3) d 1.07 (d, 3H, J = 6.0 Hz), 2.70–2.91 (m, 2H), 2.97 (sextet,
1H, J = 6.6 Hz), 3.39 (s, 1H, NH), 3.75 (dd, 2H, J = 13.2, 34.2 Hz),
7.10–7.24 (m, 5H, 1H), 7.34–7.39 (m, 2H), 5.30 (s, 1H), 7.67-7.77
(m, 3H) ppm. MS (ESI+) m/z (rel): 276 (100, M+H).
coupling of the epoxide formed from (R)-(ꢀ)-30,50-dibenzyloxy-
phenylbromohydrin with the (R)- or (S)-enantiomer of the
appropriate N-benzylaminoalkanes. The epoxide was formed by
combining (R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin (1 equiv)
with K2CO3 (1.4 equiv) in 1:1 THF/MeOH (c 0.3 M) and stirring
for 2 h under argon at rt. The solvent was removed and the residue
partitioned between toluene and H2O, the toluene fraction dried
(Na2SO4), filtered, and evaporated. The residue was dissolved with
0.95 equiv of the appropriate free N-benzylaminoalkane in toluene
and evaporated again under high vacuum to remove trace H2O. The
resulting colorless residue was heated to 120 °C under argon for
24–48 h, monitoring by 1H NMR and mass spectrometry to afford
the coupled product. The residue was dissolved in EtOH (c 0.07
M) with heat and transferred to a Parr flask, where it was hydroge-
nated at 50 psi of hydrogen over 10% (wt) Pd/C (10 mg cat/65 mg
bromohydrin) for 24 h to afford the debenzylated product, as con-
firmed by 1H NMR and mass spectrometry. The mixture was
filtered through CeliteÒ, the filter cake rinsed with iPrOH, and then
concentrated. The free base was purified on silica gel (1:20) eluting
with 15% methanol/chloroform. The resulting material was con-
verted to the fumarate salt by heating with 0.5 eq of fumaric acid
in 1:1 iPrOH/EtOH (c 0.2 M) to afford the 0.5 fumarate salt of the
fenoterol analog.
4.8.3. rac-N-Benzyl-1-(4-methoxynaphthalen-1-yl)propane-2-
amine
Prepared from 1.48 g (6.9 mmol) of 1-(4-methoxynaphthalene-
1-yl)propan-2-one, 0.75 mL (6.9 mmol) of benzylamine, 2.20 g
(10.4 mmol) of Na(OAc)3BH and 0.45 mL of HOAc. Yield 1.96 g
(93%). 1H NMR (CDCl3) d 1.15 (d, 3H, J = 6.0 Hz), 2.98–3.15 (m,
2H), 3.20–3.24 (m, 1H), 3.82 (dd, 3H, J = 13.8, 47.7 Hz), 3.99 (s,
3H), 6.73 (d, 1H, J = 7.8 Hz), 7.17–7.26 (m, 5H, 1H), 7.45–7.48 (m,
2H), 7.88–7.91 (m, 1H), 8.28–8.32 (m, 1H) ppm. MS (ESI+) m/z
(rel): 306 (100, M+H).
4.9. General procedure for the enantiomeric resolution of N-
benzylaminoalkanes
The appropriate racemic N-benzylaminoalkane was combined
with 1 equiv of either optically active mandelic acid or naproxen
in 1:2 iPrOH/MeOH (c 0.5 M), refluxed until full dissolution, then
cooled to rt. The crystals were filtered, collected, and recrystallized
twice from MeOH (c 0.3 M) to afford the optically active salts. The
salts were converted to the free amine by partitioning between
10% K2CO3 and CHCl3, drying organic extracts (Na2SO4), filtering
and evaporating.
4.10.1. (R,R)-(ꢀ)-5-(1-Hydroxy-2-(1-hydroxyphenyl)butan-2-
ylamino)ethyl) benzene-1,3-diol [(R,R)-52]
Prepared from 282 mg (0.82 mmol) of (R)-(ꢀ)-N-benzyl-1-(4-
(benzyloxy)phenyl)butan-2-amine and 260 mg (0.78 mmol) of
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 175 mg (60%).
1H NMR (300 MHz, CD3OD): d 0.950 (t, 3H, J = 7.5 Hz), 1.67 (m,
2H), 2.83–3.18 (m, 4H), 3.33–3.40 (m, 1H), 3.37 (s, 4H), 4.82 (m,
1H), 6.24 (d, 1H, J = 2.1 Hz), 6.37 (d, 2H, J = 1.8 Hz), 6.73 (s, 2H,
fum), 6.76 (d, 2H, J = 8.4 Hz), 7.05 (d, 2H, J = 8.7 Hz) ppm. 13C
NMR (75 MHz, CD3OD) d 9.43, 23.28, 36.56, 52.29, 62.16, 70.02,
103.4, 105.3, 116.7, 127.8, 131.3, 136.5, 144.6, 157.6, 159.9,
4.9.1. (R)-(ꢀ)-N-Benzyl-1-(4-(benzyloxy)phenyl)butan-2-amine
Prepared from 2.56 g (7.4 mmol) of rac-N-benzyl-1-(4-(benzyl-
oxy)phenyl)butan-2-amine and 1.12 g (7.4 mmol) of (R)-(ꢀ)-man-
delic acid (Aldrich). Yield 433 mg (33%), [
a]
D = ꢀ18.0 (0.7% MeOH).
4.9.2. (S)-(+)-N-Benzyl-1-(4-(benzyloxy)phenyl)butan-2-amine
Prepared from 2.11 g (6.1 mmol) of rac-N-benzyl-1-(4-(benzyl-
oxy)phenyl)butan-2-amine and 928 mg (6.1 mmol) of (S)-(+)-man-
172.3 ppm. UV (MeOH) kmax 206 nm (e 22,500), 223 (12,300),
278 (2460); [
a
]
D = ꢀ15.6 (free amine, 0.5% MeOH); HPLC: (a)
delic acid (Aldrich). Yield 407 mg (39%), [a]D = +21.4 (0.8% MeOH).
H2O/ACN/TFA 70/30/0.01, 1.0 mL/min, 278 nm, tR = 3.57 min,
97.1% pure. (c) ACN/MeOH/TFA 90/10/0.05, 2.0 mL/min, 278 nm,
tR = 5.26 min, 97.5% enantiomeric purity. HRMS (m/z) calcd for
C18H24NO4 [M+H]+ 318.1700; found 318.1699.
4.9.3. (R)-(ꢀ)-N-Benzyl-1-(naphthylen-2-yl)propan-2-amine
Prepared from 1.05 g (3.8 mmol) of rac-N-benzyl-1-(naphthy-
len-2-yl)propan-2-amine and 875 mg (3.8 mmol) of (S)-(+)-na-
proxen (Aldrich). Yield 310 mg (61%), [
a
]D = ꢀ11.2 (0.7% MeOH).
4.10.2. (R,S)-(+)-5-(1-Hydroxy-2-(1-hydroxyphenyl)butan-2-
ylamino)ethyl) benzene-1,3-diol [(R,S)-52]
4.9.4. (S)-(+)-N-Benzyl-1-(naphthylen-2-yl)propan-2-amine
Prepared from 2.10 g (7.6 mmol) of rac-N-benzyl-1-(naphthy-
len-2-yl)propan-2-amine and 1.76 g (7.6 mmol) of (R)-(ꢀ)-naprox-
Prepared from 352 mg (1.02 mmol) of (S)-(+)-N-benzyl-1-(4-
(benzyloxy)phenyl)butan-2-amine and 321 mg (0.97 mmol) of
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 142 mg (38%).
1H NMR (300 MHz, CD3OD): d 0.972 (t, 3H, J = 7.5 Hz), 1.70 (p,
2H, J = 6.9 Hz)), 2.86–3.22 (m, 4H), 3.32–3.37 (m, 1H), 3.34 (s,
4H), 4.82 (m, 1H), 6.25 (t, 1H, J = 2.1 Hz), 6.36 (d, 2H, J = 1.8 Hz),
6.74 (s, 2H, fum), 6.77 (d, 2H, J = 8.4 Hz), 7.08 (d, 2H, J = 8.7 Hz)
ppm. 13C NMR (75 MHz, CD3OD) d 9.820, 24.16, 36.48, 52.30,
62.32, 69.92, 103.3, 105.3, 116.8, 127.7, 131.3, 136.1, 144.4,
en (Aldrich). Yield 630 mg (60%), [a]D = +12.0 (0.7% MeOH).
4.9.5. (R)-(ꢀ)-N-Benzyl-1-(4-methoxynaphthalen-1-yl)propa-
ne-2-amine
Prepared from 1.04 g (3.4 mmol) of rac-N-benzyl-1-(4-meth-
oxynaphthalen-1-yl)propane-2-amine and 520 mg (3.4 mmol) of
(S)-(+)-mandelic acid. Yield 389 mg (77%),
MeOH).
[a
]D = ꢀ12.1 (1.0%
157.6, 159.8, 171.3 ppm. UV (MeOH) kmax 204 nm (e 26,900), 224
(11,500), 278 (2320). [
a
]
D = ꢀ7.2 (free amine, 0.5% MeOH). HPLC:
4.9.6. (S)-(+)-N-Benzyl-1-(4-methoxynaphthalen-1-yl)propane-
2-amine
Prepared from 1.95 g (6.4 mmol) of rac-N-benzyl-1-(4-meth-
oxynaphthalen-1-yl)propane-2-amine and 971 mg (6.4 mmol) of
(a) H2O/ACN/TFA 70/30/0.01, 1.0 mL/min, 278 nm, tR = 3.56 min,
99.5% pure. (c) ACN/MeOH/TFA 90/10/0.05, 2.0 mL/min, 278 nm,
tR = 5.88 min, 99.0% enantiomeric purity. HRMS (m/z): calcd for
C18H24NO4 [M+H]+ 318.1700; found 318.1706.
(R)-(ꢀ)-mandelic acid. Yield 988 mg (67%), [
a]D = +11.7 (1.0%
MeOH).
4.10.3. (R,R)-5-(1-Hydroxy-2-(1-(naphthalen-2-yl)propan-2-
ylamino)ethyl) benzne-1,3-diol [(R,R)-53]
4.10. Preparation of fenoterol analogs
Prepared from 310 mg (1.12 mmol) of (R)-(ꢀ)-N-benzyl-1-
(naphthylen-2-yl)propan-2-amine and 355 mg (1.06 mmol) of
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 186 mg (44%).
1H NMR (300 MHz, CD3OD): d 1.06 (d, 3H, J = 6.3 Hz), 2.73–2.90
In this study, the fenoterol analogues were prepared using a
previously described approach.3 The key step involved the

736
K. Jozwiak et al. / Bioorg. Med. Chem. 18 (2010) 728–736
(m, 4H), 3.07 (m, 1H), 3.30 (m, 3H), 4.05 (dd, 1H, J = 5.1, 8.4 Hz),
6.11 (t, 1H, J = 2.4 Hz), 6.28 (d, 2H, J = 2.1 Hz), 7.26 (dd, 1H,
J = 1.5, 8.4 Hz), 7.38–7.42 (m, 2H), 7.57 (s, 1H), 7.72–7.77 (m, 3H)
ppm. 13C NMR (75 MHz, CD3OD) d 15.84, 40.67, 52.70, 56.66,
70.46, 103.27, 105.26, 126.97, 127.39, 128.22, 128.64, 128.70,
129.20, 129.64, 134.04, 135.05, 135.11, 136.81, 144.95, 160.13,
(53%). 1H NMR (300 MHz, CD3OD): d 1.20 (t, 3H, J = 6.6 Hz), 3.07–
3.21 (m, 3H), 3.52–3.75 (m, 2H), 3.97 (s, 3H), 4.69–4.83 (m, 1H),
6.24 (t, 1H, J = 2.1 Hz), 6.39 (dd, 2H, J = 2.4, 5.4 Hz), 6.74 (s, 1H),
6.84 (d, 1H, J = 7.8 Hz), 7.31 (d, 1H, J = 8.1 Hz), 7.48 (t, 1H,
J = 6.9 Hz), 7.56 (t, 1H, J = 6.9 Hz), 8.01 (dd, 1H, J = 8.4, 13.5 Hz),
8.27 (d, 1H, J = 7.8 Hz) ppm. 13C NMR (75 MHz, CD3OD) d 15.77,
36.64, 52.37, 55.94, 70.46, 103.4, 104.5, 105.3, 123.8, 124.3,
124.9, 126.2, 127.4, 128.1, 129.4, 133.8, 135.5, 144.7, 156.6,
173.20 ppm. UV (MeOH), kmax: 276 nm (e 6170), 224 (90,680),
206 (50,290; [
a
]
D = ꢀ27.4 (0.5% MeOH). HPLC: (b) H2O/ACN/TFA
70/30/0.1, 1.0 mL/min, 275 nm, tR = 3.96 min, 95.3% purity. (c)
ACN/iPrOH/DEA 95/5/0.05, 3.0 mL/min, 275 nm, tR = 8.58 min,
95.6% enantiomeric purity. HRMS (m/z): calcd for C21H24NO3
[M+H]+ 338.1751; found 338.1745.
160.0, 169.0 ppm. UV (MeOH), kmax
:
298 nm
(e 5430), 286
(5710), 233 (25,100), 210 (43,200). [
a
]
D = ꢀ15.8 (free amine; 0.5%
MeOH). HPLC: (a) H2O/ACN/TFA 50/50/0.01, 1.0 mL/min, 286 nm,
tR = 3.087 min, 96.0% pure (c) gradient 0–20% iPrOH/hexane/
0.05% TFA, 3.0 mL/min, 275 nm, tR = 9.69 min 99.0% enantiomeric
purity. HRMS (m/z): calcd for C22H26NO4 [M+H]+ 368.1856; found
368.1844.
4.10.4. (R,S)- 5-(1-Hydroxy-2-(1-(naphthalen-2-yl)propan-2-
ylamino)ethyl) benzne-1,3-diol [(R,S)-53]
Prepared from 626 mg (2.27 mmol) of (S)-(+)-N-benzyl-1-
(naphthylen-2-yl)propan-2-amine and 719 mg (2.16 mmol) of
(R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 578 mg (68%).
1H NMR (300 MHz, CD3OD): d 1.09 (d, 3H, J = 6.3 Hz), 2.60–2.95
(m, 4H), 3.03 (m, 1H), 3.30 (m, 3H), 4.07 (dd, 1H, J = 5.1, 8.4 Hz),
6.14 (t, 1H, J = 2.4 Hz), 6.28 (d, 2H, J = 2.1 Hz), 7.33 (dd, 1H,
J = 1.8, 8.4 Hz), 7.40–7.44 (m, 2H), 7.64 (s, 1H), 7.77–7.81 (m, 3H)
ppm. 13C MNR (75 MHz, CD3OD) d 16.52, 40.28, 52.82, 56.88,
70.75, 103.27, 105.29, 126.97, 127.39, 128.22, 128.64, 128.69,
129.20, 129.65, 134.03, 135.05, 135.18, 136.89, 144.95, 160.00,
Acknowledgments
This work was supported in part by funds from the National
Institute on Aging, Intramural Research Program, NIA Contract
N01-AG-3-1009 and the Foundation for Polish Science (FOCUS
4/2006 programme)
Supplementary data
173.46 ppm. UV (MeOH), kmax: 276 (
e 6310), 224 (93,530), 204
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.11.062.
(50,580). [
a
]
D = ꢀ18.8 (0.5% MeOH). HPLC: (b) H2O/ACN/TFA 70/
30/0.1, 1.0 mL/min, 275 nm, tR = 3.94 min, 98.6% purity. (c) ACN/
iPrOH/DEA 95/5/0.05, 3.0 mL/min, 275 nm, tR = 8.04 min, 96.4%
enantiomeric purity. HRMS (m/z): calcd for C21H24NO3 [M+H]+
338.1751, found 338.1746.
References and notes
1. Xiao, R.-P.; Zhu, W.; Zheng, M.; Chakir, K.; Bond, R.; Lakatta, E. G.; Cheng, H.
Trends Pharmacol. Sci. 2004, 25, 358–365.
2. Beigi, F.; Bertucci, C.; Zhu, W.; Chakier, K.; Wainer, I. W.; Xiao, R.-P.; Abernethy,
D. A. Chirality 2006, 18, 822–827.
4.10.5. (R,R)-(ꢀ)-5-(1-Hydoxy-2-(1-(4-methoxynaphthalen-1-
yl)propan-2-ylamino)ethyl)benzene-1,3-diol [(R,R)-54]
3. Jozwiak, K.; Khalid, C.; Tanga, M. J.; Berzetei-Gurske, I.; Jimenez, L.; Kozocas, J.
A.; Woo, A.; Zhu, W.; Xiao, R.-P.; Abernethy, D. R.; Wainer, I. W. J. Med. Chem.
2007, 50, 2903–2915.
4. Kikkawa, H.; Isogaya, M.; Nagao, T.; Kurose, H. Mol. Pharmacol. 1998, 53, 128–
134.
5. Isogaya, M.; Yamagiwa, Y.; Fujuita, S.; Sugimoto, Y.; Nago, T.; Kurose, H. Mol.
Pharmacol. 1998, 54, 616–622.
6. Furse, K. E.; Lybrand, T. P. J. Med. Chem. 2003, 46, 4450–4462.
7. Kubinyi, H. Comparative Molecular Field Analysis (COMFA). In Encyclopedia of
Prepared from 390 mg (1.28 mmol) of (R)-(ꢀ)-N-benzyl-1-(4-
methoxynaphthalen-1-yl)propane-2-amine and 404 (1.22 mmol)
of (R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 309 mg
(60%). 1H NMR (300 MHz, CD3OD): d 1.22 (t, 3H, J = 6.6 Hz), 3.09–
3.21 (m, 3H), 3.59–3.69 (m, 2H), 3.99 (s, 3H), 4.74–4.83 (m, 1H),
6.23 (t, 1H, J = 2.4 Hz), 6.37 (dd, 2H, J = 2.4, 5.7 Hz), 6.74 (s, 1H),
6.86 (d, 1H, J = 7.8 Hz), 7.32 (d, 1H, J = 7.8 Hz), 7.48 (t, 1H, J =
6.9 Hz), 7.56 (t, 1H, J = 6.9 Hz), 8.02 (dd, 1H, J = 8.4, 12.0 Hz), 8.27
(d, 1H, J = 8.7 Hz) ppm. 13C NMR (75 MHz, CD3OD): d 15.78, 36.66,
52.39, 55.96, 70.20, 103.4, 104.5, 105.3, 123.8, 124.3, 124.9, 126.2,
127.4, 128.1, 129.5, 133.8, 135.2, 144.6, 156.6, 160.0, 168.3 ppm.
Computational Chemistry, 1998, John Wiley
www.wiley.com//legacy/wileychi/eec/.
&
Sons, Ltd. http://
8. Seifert, R.; Dove, S. Mol. Pharmacol. 2009, 75, 13–18.
9. Cherezov, V.; Rosenbaum, D. M.; Hanson, M. A.; Rasmussen, S. G.; Thian, F. S.;
Kobilka, T. S.; Choi, H. J.; Kuhn, P.; Weis, W. I.; Kobilka, B. K.; Stevens, R. C.
Science 2007, 318, 1258–1265.
10. Woo, A. Y.-H.; Wang, T.-B.; Zeng, X.; Weizgog, Z.; Abernethy, D. R.; Wainer, I.
W.; Xiao, R.-P. Mol. Pharmacol. 2009, 75, 158–165.
11. Wieland, K.; Zuurmond, H. M.; Krasel, C.; Ijzerman, A. D. P.; Lohse, M. J. Proc.
Natl. Acad. Sci. U.S.A. 1996, 93, 9276–9281.
12. Zuurmond, H. M.; Hessling, J.; Blüml, K.; Lohse, M.; Ijzerman, A. P. Mol.
Pharmacol. 1999, 56, 909–916.
UV (MeOH), kmax: 298 nm (e 4,970), 286 (9920), 234 (22,600), 210
(42,500). [
a
]
D = ꢀ28.8 (free amine; 0.5% MeOH). HPLC: (a) H2O/
ACN/TFA 50/50/0.01, 1.0 mL/min, 286 nm, tR = 3.087 min, 99.0%
pure. (c) Gradient 0–20% iPrOH/hexane/0.05% TFA, 3.0 mL/min,
275 nm, tR = 8.62 min, 95.0% enantiomeric purity. HRMS (m/z): calcd
for C22H26NO4 [M+H]+ 368.1856; found 368.1845.
13. Kontoyianni, M.; DeWeese, C.; Penzotti, J. E.; Lybrand, T. P. J. Med. Chem. 1996,
39, 4406–4420.
14. Swaminath, G.; Xiang, Y.; Lee, T. W.; Steenhuis, J.; Parnot, C.; Kobilka, B. K. J.
Biol. Chem. 2004, 279, 686–691.
15. Kobilka, B. Mol. Pharmacol. 2004, 65, 1060–1062.
16. Staehelin, M.; Simons, P.; Jaeggi, K.; Wigger, N. J. Biol. Chem. 1983, 258, 3496–
3502.
4.10.6. (R,S)-(+)-5-(1-Hydoxy-2-(1-(4-methoxynaphthalen-1-yl)-
propan-2-ylamino)ethyl)benzene-1,3-diol [(R,S)-54]
Prepared from 422 mg (1.38 mmol) of (S)-(+)-N-benzyl-1-(4-
methoxynaphthalen-1-yl)propane-2-amine and 438 (1.31 mmol)
of (R)-(ꢀ)-30,50-dibenzyloxyphenylbromohydrin. Yield 302 mg
17. Cheng, Y.-C.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22, 3099–3108.
18. Gilman, A. G. Proc. Natl. Acad. Sci. U.S.A. 1970, 67, 305–312.