Thienopyrimidines as β3-adrenoceptor agonists: Hit-to-lead optimization

Article abstract of DOI:10.1016/j.bmcl.2010.08.039

Resulting from a vHTS based on a pharmacophore alignment on known β3-adrenoceptor ligands, an aryloxypropanolamine scaffold comprising a thienopyrimidine moiety was further optimized as a human β3-AR agonist, yielding a lead compound with an excellent cellular activity of EC₅₀ = 20 pM, selectivity over hβ1- and hβ2-adrenoceptors and a promising safety profile.

Full text of DOI:10.1016/j.bmcl.2010.08.039

Bioorganic & Medicinal Chemistry Letters 20 (2010) 6108–6115
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Thienopyrimidines as b3-adrenoceptor agonists: Hit-to-lead optimization
Stefan Tasler a, , Roland Baumgartner a, Astrid Ammendola a, Josef Schachtner a, Tanja Wieber a,
⇑
Marcus Blisse a, Sandra Rath a, Mirko Zaja a, Philipp Klahn a, Udo Quotschalla b, Peter Ney b
a 4SC AG, Am Klopferspitz 19a, 82152 Planegg-Martinsried, Germany
b UCB Pharma S.A., Chemin du Foriest, 1420 Braine l’Alleud, Belgium
a r t i c l e i n f o
a b s t r a c t
Article history:
Resulting from a vHTS based on a pharmacophore alignment on known b3-adrenoceptor ligands, an aryl-
oxypropanolamine scaffold comprising a thienopyrimidine moiety was further optimized as a human b3-
AR agonist, yielding a lead compound with an excellent cellular activity of EC50 = 20 pM, selectivity over
hb1- and hb2-adrenoceptors and a promising safety profile.
Received 12 July 2010
Revised 6 August 2010
Accepted 7 August 2010
Available online 12 August 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
Adrenoceptor
b3-Adrenergic receptor
GPCR
Molecular modeling
vHTS
The human b3-adrenergic receptor (hb3-AR) is distributed in
different tissues like white and brown adipocytes, urinary bladder
detrusor, gastrointestinal tract, near-term myometrium, brain and
the heart. Consequently, several indications emerged for a poten-
tial treatment with hb3-AR ligands, in particular agonists: obesity,
type II diabetes, overactive bladder, irritable bowel syndrome (IBS),
preterm labor, anxiety and major depression disorder.1 Antago-
nists have recently been under evaluation for the treatment of
chronic heart failure.2 However, first clinical trials conducted with
hb3-AR agonists for the treatment of obesity and diabetes failed.
This was mainly due to differing expression levels of this receptor
especially on white adipocytes and a differently weighted role in
lipolysis as compared to rodents, in which respective animal mod-
els were performed. Given these results, the major focus shifted to-
wards the treatment of overactive bladder.1,3
Over the last decade, several phenylethanolamine hb3-AR ago-
nists (Fig. 1) advanced to clinical trials with variable success:
development of Solabegron was not continued upon completion
of phase II for the treatment of overactive bladder and IBS. Amibeg-
ron was not further pursued upon two phase III trials, one on anx-
iety, the other on major depressive disorder. Mirabegron and KUC-
7483 are still under evaluation in active phase III trials for the
treatment of overactive bladder, with a new drug application filing
being expected for the former to be submitted in 2011.4,5 Even
though hb3-AR agonists based on a second major general scaffold,
aryloxypropanolamines, emerged already in the mid 1980s,6 such
OH
O
H
N
Cl
Cl
N
H
OH
Solabegron
OH
OH
O
H
N
H
O
O
Amibegron
Mirabegron
H
N
S
O
NH
2
N
N
H
OH
H
N
R = Et: KUC-7483
R = H: Ritobegron
O
S
HO
O
R
O
Ki (hβ1-AR) = n.d.r.
Ki (hβ2-AR) = 700 nM
Ki (hβ3-AR) = 460 nM
OH
H
N
O
N
O
EC50 (hβ3-AR) = 120 nM.
N
N
1
(IA = 0.6)
⇑
Corresponding author. Tel.: +49 89 700763 0; fax: +49 89 700763 29.
E-mail address: stefan.tasler@4sc.com (S. Tasler).
Figure 1. hb3-AR agonists under clinical evaluation and hit compound 1 deduced
from a vHTS approach.7,9 IA = intrinsic activity; n.d.r. = no dose–response.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.08.039

S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
6109
case RL was chosen to be 0.25 (that is 25%) in order to allow for
an evaluation of partial agonists down to an IA of 0.25. Throughout
the text body, discussion of relative efficacy is based on the qResp
values calculated according to this equation.
With this concept established one is able to discriminate effi-
cacy-driven from affinity-driven agonists. Accordingly, tissue
selectivity can be addressed and target related side effects might
be avoided: for the activation of a certain cellular response, less
substrates have failed so far to reach any phase III clinical trials. To
our knowledge, LY-377604 is the only aryloxypropanolamine cur-
rently in phase II. It is under investigation for the treatment of
obesity in combination with Sibutramine, a serotonin and norepi-
nephrine reuptake inhibitor.4
As starting point for a program on hb3-adrenoceptors, a phar-
macophore alignment of a virtual library of 4.6 Mio commercially
available compounds on known hb3-AR agonists—mainly of the
phenylethanolamine array—was chosen. This led to the identifica-
tion of intriguing hit classes of the aryloxypropanolamine-type dis-
playing Ki values already in the submicromolar range.7 The further
structural evaluation of one of these hits, thienopyrimidine 1
(Fig. 1), is described within this contribution. It already displayed
a certain selectivity over hb1-AR and proved to be a partial agonist
at hb3-AR. As a commercially available compound, it has already
been described as an antagonist on neurokinin (=tachykinin)
receptors [Ki(NK1) = 53 nM; Ki(NK2) = 1700 nM].8 Notwithstand-
ing, it served as an excellent starting point for structural variations
leaving the realm of NK ligands towards selective hb3-AR agonists.
For an evaluation of the pharmacologically relevant molecular
texture of the compounds, a set of different assays has been estab-
lished. Binding constants (Ki) were obtained from radioligand bind-
ing assays for subtypes hb1-, hb2- and hb3-AR.10 A functional
cellular assay was used to assess hb3-AR agonism.11 Structure–
activity relationships (SAR) were based not only on affinity and
activity,12 but also on a combined metric which utilizes Ki and
EC50 values as well as the intrinsic activity (IA) of the respective li-
gand (relative to the maximum cellular response triggered by
efficacious ligands (those with higher qResp) would require a higher
receptor density on the respective cells, thus allowing for a selec-
tive stimulation of tissue with higher receptor expression levels.14
The primary focus, however, was defined as to identify highly po-
tent and efficacious ligands (that is, with low qResp) for triggering a
maximum cellular response, thus enabling an evaluation of one of
these ligands in a first tissue model.
A general synthetic route towards thienopyrimidine-substi-
tuted aryloxypropanolamines can be deduced from Scheme 1.
Additional details are elaborated within the Supplementary data.
Standard synthesis for aryloxymethyloxiranes 4 involved treat-
ment of phenols 2 with epibromohydrine (3) and K2CO3 in buta-
none at 80 °C—usually succeeding in good to excellent yields of
40–95%.15 N-Substituted 4-aminothienopyrimidines 7 were easily
attained by conversion of the respective 4-chlorothienopyrimidine
6
16 with a N-Boc-protected diamine like 5 in ethylene glycol17 and
subsequent in situ Boc-removal with HCl in dioxane. Yields of 77–
95% for this transformation represent the outcome for all combina-
tions of chloro-thienopyrimidines and secondary amines utilized
within the following SAR investigations. Only when utilizing pri-
mary amines like 1-Boc-4-aminopiperidine and 1-Boc-4-(amino-
methyl)piperidine (cf. derivatives in Fig. 2), yields deteriorated to
about 40%. For 4-chloro-2-methylthieno[2,3-d]pyrimidine (6) and
4-chlorothieno[3,2-d]pyrimidine (en route to compounds 49–51
and 52–54, respectively; Table 3), this sequence resulted in precip-
itation of the product as the respective hydrochloride upon Boc-re-
moval, which was simply filtered off. For all other amine building
blocks 7, alkaline extraction yielded the free amines. Finally,
amines 7 and oxiranes 4 were heated together up to 80 °C in iPrOH
or iPrOH/DMSO mixtures depending on solubility of the respective
reaction partners.18 Thus the desired aryloxypropanolamine sub-
strates for biological evaluation on adrenergic receptors were at-
tained. When amine 7 was used as the hydrochloride salt, DIEA
had to be added for achieving a smooth conversion. For multiple
step syntheses of more complexly substituted phenols 2 and addi-
tional functional group transformations upon construction of the
aryloxypropanol array, refer to the Supplementary data.
treatment with 100
flected by the term qResp
l
M isoproterenol). This relative efficacy13 is re-
:
RL ꢀ EC50
½
l
Mꢁ
qResp
¼
ꢀ 100
Ki ½
l
Mꢁ ꢀ ðIA ꢂ RLÞ þ ðRL ꢀ EC50
½lMꢁÞ
RL ¼ pre-defined response level; can be adjusted in between
0:01 and 1:0 to include partial agonists in analysis
IA ¼ intrinsic activity; with 1:0 representing the 100% level
achieved with 100lM isoproterenol
qResp represents the percentage of receptors required to be occupied
by a given ligand to result in a cellular response equal to a given
percentage of the isoproterenol signal (=response level RL). In our
O
OH
O
O
i
R
R
+
Br
Due to the commercial availability of numerous differently
substituted phenols 2, an extensive SAR analysis was initiated
around the aryloxy portion of the hit molecule 1, an excerpt of
which can be seen in Table 1.
40-95%
2
3
4
Based on binding data, it became obvious for basically all vari-
ations that selectivity for hb3-AR over hb1-AR might be achieved
more easily as over hb2-AR: affinities for hb2-AR were found to
be similar or even better as compared to the Ki values for hb3-AR
(except for only compound 24). When pure hydrogen bond accep-
tor groups were located in 3- or 4-position (OMe, OCF3, F; entries
12, 1, 13, 15 and 16), partial agonism on hb3-AR was detected in
a functional assay. Intrinsic activities ranged from 0.4 to 0.8, result-
ing in a poor relative efficacy reflected by qResp = 17–51% (i.e., 17–
51% of the b3-receptors in one cell have to be occupied by these
compounds in order to trigger a functional response equal to 25%
of the maximum response achieved with 100 lM isoproterenol).
3-CN substitution (compound 23) was in line with these findings,
even though now resulting in a full agonist and thus a somewhat
better relative efficacy. Interestingly, with hydrogen bond acceptor
groups installed in 2-position (OMe, F; entries 11 and 14), highest
affinities for hb3-AR (Ki values around 12 nM) and good one-digit
R'
H
N
R'
H2N
Boc
S
ii, iii
S
Cl
N
NH
+
N
N
77-95%
N
N
5
6
7
OH
H
N
R'
O
iv
13-80%
R
4
+
7
S
N
N
N
cf. Table entries
Scheme 1. Reagents and conditions: (i) K2CO3, butanone, 80 °C, 24–72 h; (ii)
ethylene glycol, 110 °C, 4 h; (iii) HCl in dioxane (4.0 M), rt, 3 h; (iv) depending on
solubility: iPrOH or iPrOH/DMSO (up to 1:1), DIEA (when amine present as
hydrochloride), 60–80 °C, 2–24 h; yields are not optimized.

6110
S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
Table 1
Mono-substitution at the aryloxy moiety9
OH
H
R
1
O
N
S
N
2
4
3
N
N
Compd
R
Binding assay, Ki [nM]
hb2-AR
Agonist assay hb3-AR
Antagonist assay hb3-AR, IC50 [nM]
a
hb1-AR
hb3-AR
EC50 [nM] (IA)
qResp [%]
8
9
10
11
12
1c
H
159
60
975
215
938
—
9.0
6.0
81
11
30
700
68
25
30
55
18
93
10
88
6.0
10
42
139
117
370
135
>10,000
17
18 (1.1)
0.62 (1.1)
1.6 (1.3)
2.5 (1.2)
21 (0.4)
120 (0.6)
113 (0.8)
8.3 (1.0)
41 (0.8)
92 (0.8)
26 (1.0)
—
23
2.1
0.26
5.5
44
17
49
16
51
32
15
—
—
—
—
—
—
—
—
—
—
—
—
162
—
8064
—
—
—
—
3-OHb
8.5
148
11
42
460
59
14
17
88
49
290
63
107
86
64
47
37
85
331
149
2646
4-OHb
2-OMe
3-OMe
4-OMe
3-OCF3
2-F
3-F
4-F
3-CH2OH
4-CH2OH
3-Et
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
996
433
293
268
34,400
—
289
462
150
100
700
2560
2018
1427
2902
—
264 (1.0)
—
58
—
4-Et
2-Allyl
2-iPr
79 (1.1)
100 (1.0)
35 (1.0)
150 (0.8)
—
22
35
19
64
—
—
9.4
36
3-CN
3-NHCONH2
3-NHAc
3-CONHMeb
3-CO-morpholinb
3-COOHb
107
402
—
—
55 (1.1)
2410 (0.7)
—
‘—’ refers to no dose–response in the respective assay.
a
qResp equation, RL set to 0.25.
For synthetic details, cf. Supplementary data.
Compound from a commercial source.
b
c
Table 2
Di-substitution at the aryloxy moietya,9
OH
OH
H
N
H
N
2
R
3
HO
O
O
3
S
N
S
N
N
4
6
HO
R
X1
X2
N
N
N
Compd
Scaffold
R
Binding assay, Ki [nM]
hb2-AR
Agonist assay hb3-AR
b
hb1-AR
hb3-AR
EC50 [nM] (IA)
qResp [%]
9
29
30
X1
X1
X1
H
60
—
—
6.0
259
517
8.5
620
919
0.62 (1.1)
97 (1.1)
n.d.
2.1
4.7
n.d.
4-CH2OH
6-COOH
10
31
32
33
34
35
36
37
X2
X2
X2
X2
X2
X2
X2
X2
H
3-OH
2-CH2OH
3-CH2OH
3-CH2OMe
2-Et
975
81
873
25
14
4.9
7.0
11
—
148
932
72
7.3
1.0
30
1.6 (1.3)
1.2 (1.3)
1.7 (1.1)
0.02 (1.1)
0.04 (1.1)
2.2 (1.1)
0.14 (1.1)
310 (1.0)
0.26
0.03
0.70
0.08
1.2
2.0
0.46
2.8
8560
1480
629
484
236
3-Et
3-COOH
287
6710
9.0
3500
‘—’ refers to no dose–response in the respective assay; n.d. = not determined.
a
For synthetic details, cf. Supplementary data.
qResp equation, RL set to 0.25.
b
nanomolar agonistic effects were achieved. These represented
slight improvements over the unsubstituted parent compound 8.
As to incorporation of hydroxy groups within the para-substituted
series, 4-hydroxymethyl substitution (compound 18) was not well
tolerated with regard to binding affinity for hb3-AR (Ki = 290 nM).
Furthermore, this derivative displayed antagonistic activity. In
contrast, direct 4-attachment of a hydroxy group (compound 10)
resulted in the most efficacious hb3-AR agonist of the whole SAR
series based on mono-substituted aryloxy derivatives reflected by
an qResp value of 0.26%. In comparison, 3-OH incorporation (com-

S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
6111
Table 3
Variations of the thienopyrimidine unita,9
OH
H
R
1
N
O
N
4
2
X
3
Compd
Scaffold X
R
Binding assay, Ki [nM]
hb2-AR
Agonist assay hb3-AR
b
hb1-AR
hb3-AR
EC50 [nM] (IA)
qResp [%]
8
10
H
4-OH
159
975
9
81
17
148
18 (1.1)
1.6 (1.3)
23
0.26
5'
33
3-CH2OH–4-OH
629
14
7.3
0.02 (1.1)
0.08
6'
S
N
N
2'
38
39
H
4-OH
114
600
19
175
23
75
10 (1.0)
3.1 (1.1)
13
1.2
F
S
N
N
S
40
41
H
4-OH
33
129
4
28
11
76
4.7 (1.0)
1.0 (1.3)
13
0.31
S
42
3-CH2OH–4-OH
1049
20
62
0.50 (1.2)
0.21
N
N
N
N
43
44
45
H
4-OH
3-CH2OH–4-OH
230
560
2810
100
270
94
90
499
300
96 (1.3)
7.8 (1.4)
3.1 (1.2)
20
0.35
0.27
Ph
S
46
47
48
H
4-OH
3-CH2OH–4-OH
450
3507
9330
80
2600
223
151
1390
989
7.1 (1.1)
20 (1.1)
2.9 (1.2)
1.4
0.41
0.08
S
S
N
N
N
N
49
50
51
H
4-OH
3-CH2OH–4-OH
72
4880
4980
189
589
440
113
1300
670
15 (1.2)
24 (1.1)
1.7 (1.1)
3.4
0.54
0.07
52
53
H
4-OH
30
5600
14
961
41
1385
2.8 (1.0)
33 (1.1)
2.2
0.74
S
54
3-CH2OH–4-OH
3970
165
600
2.0 (1.1)
0.10
N
N
a
For synthetic details, cf. Supplementary data.
qResp equation, RL set to 0.25.
b
pound 9) gave rise to a slightly superior EC50 value (0.62 vs 1.6 nM
for derivative 10), with a decreased relative efficacy, though, due to
a significantly enhanced hb3-AR affinity. As already found for the
4-hydroxymethyl derivative 18, hb1-AR affinity was virtually lost
as well for the respective 3-derivative 17. In the latter case, how-
ever, a decent agonistic effect on hb3-AR was maintained, compa-
rable to the unsubstituted variant 8. Other polar groups than
hydroxy functionalities in 3-position were not well tolerated (en-
tries 24–27): a 3-ureido group resulted in low relative efficacy
2-substitution (entries 21 and 22), affinities for hb1-AR improved
simultaneously, thus resulting in a rather general unselective
hb-AR binding. Agonistic activities on hb3-AR were diminished
for 2- and 3-alkyl derivatives as compared to the parent derivative
8 with EC50 values now found between 79 and 264 nM. In contrast,
respective 4-ethyl substitution resulted in comparable affinity
levels for hb1- and hb3-ARs with Ki values around 100 nM, but
led now to an antagonistic effect on hb3-AR in the functional assay.
Incorporation of carboxy groups (in 3- or 4-position) as well as of
(qResp only 64%; compound 24), 3-acetylamino and 3-(N-methyl-
endocyclic nitrogens (e.g., a 2-pyridyloxy unit) drastically
carboxamide) substitution again switched functional activity to
antagonism with IC50 values in the range of 100–400 nM (deriva-
tives 25 and 26). A 3-(morpholinocarbonyl) substitution proved
yet to be the best among those polar groups with an EC50 value
of 55 nM paralleled by a relative efficacy qResp around 10%. Placing
different alkyl groups in 2- or 3-position (entries 19, 21 and 22) re-
decreased affinities for hb3-AR. Within this subset of compounds,
best results were attained for derivative 28, still displaying a poor
Ki value of 2.6 lM.
With regard to selectivity for the hb3-AR over the other two b-
receptors, mono-substitution at the aryloxy-unit seemed appropri-
ate to serve a certain tuning of selectivity over hb1-AR based on Ki
values. A 3-hydroxymethyl group was identified as the most effec-
sulted in a certain selectivity for hb2-AR based on Ki values. As to

6112
S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
tive substituent in the latter context (compound 17) among hb3-
AR agonists. Best factor towards hb2-AR affinity was observed for
the urea derivative 24, displaying a factor of only 3.8—accompa-
nied by a weak hb3-AR partial agonism with an EC50 value of
merely 150 nM, though. Based on the SAR attained so far, a combi-
nation of two substituents within the aryloxy portion was envis-
aged next, with a 4-hydroxy-3-hydroxymethyl pattern appearing
to be most promising: the phenolic hydroxy group for attaining
highest relative efficacy and the hydroxymethyl unit for tuning
selectivity over hb1-AR.
3-hydroxymethyl substitution pattern in place (compounds 48,
51 and 54; Ki values between 600 and 990 nM). The diminution
in hb3-AR affinity observed upon incorporation of the 4-hydroxy
group into the parent molecules 46, 49 and 52 was accompanied
by a slight to significant decrease of functional activity (factor
1.5–12), which is in clear contrast to the results out of the 50-aryl
subsets. Agonistic effects, however, were still decent with EC50
values in the one-digit nanomolar range for the 4-hydroxy-
3-hydroxymethyl derivatives. This resulted in excellent relative
efficacies comparable to the respective compounds out of the
50-aryl series. Due to the generally elevated Ki level on hb3-AR
within the 50-non-aryl subsets, selectivity over hb1-AR based on
Ki values became less favorable (all below a factor of 10 as com-
pared to 86 for 33).
When comparing corresponding hydroxymethyl derivatives
within the 4-hydroxyaryloxy series X2 and the 3-hydroxyaryloxy
variations X1 (compound 29 vs 33; Table 2), the X2-derivative dis-
played significantly higher relative efficacy and activity on hb3-AR
as compared to the X1-compound, now even with an EC50 value of
20 pM (compound 33). The rather poor results attained for both
carboxylic acid derivatives 30 and 37 and the 2-ethyl compound
35 were perfectly in line with the trends already observed with
respective mono-substituted derivatives (Table 1): a poor affinity
and/or activity on hb3-AR for carboxylic acid derivatives and a loss
of selectivity for 2-alkyl variants were observed. For all other
examples, selectivity over hb1-AR (based on affinity data) was gen-
erally improved upon addition of another substituent as compared
to the respective parent compounds 9 and 10. Especially 4-hydroxy
variants X2 bearing a hydroxymethyl, methoxymethyl or ethyl
group in 3-position (entries 33, 34 and 36) should be highlighted
in this context—particularly compound 33, as originally
anticipated based on the data of Table 1: selectivity factors over
hb1-AR between 86 and 480 (ratio of Ki values) and sub-nanomolar
agonistic activity at hb3-AR were determined. Comparing the
relative efficacies of these three compounds, different additional
substitution of the 4-hydroxyphenyloxy unit might allow for an
efficacy fine-tuning in the end if required (1.2% and 0.46% vs
0.08% for compound 33)—on a highly efficacious level, though.
With regard to selectivity for hb3-Ar over hb2-AR based on Ki val-
ues, however, no significant improvement was observed as com-
pared to mono-substituted derivatives of Table 1. Affinities were
still found to be in the same range for both receptors. Best factor
of almost 5 was displayed by compound 34, which is still rather
similar to that identified for the 3-ureido substituted derivative
24 (factor 4).
The 50-methyl-60-phenyl derivatives displayed a similar devel-
opment of relative efficacy as observed for the 50-thienyl deriva-
tives 40–42, from the parent compound 43 to the 4-OH
derivative 44 and the 4-hydroxy-3-hydroxymethyl compound 45.
However, both the affinity and the activity on hb3-AR were dimin-
ished significantly for these derivatives, thus rendering this array
less favorable.
Upon altering both ‘ends’ of the molecule, the final set of varia-
tions was performed around the 2-propanolamine bridge and the
piperidine spacer unit. Several derivatizations are depicted in
Figure 2, all of which were quite detrimental to affinity and activity
on hb3-AR. Alkylations—with both small and large groups, lipo-
philic and polar—and acylation of the secondary amine of the pro-
panolamine unit, as well as cyclization including both the amine
and the hydroxy group resulted in derivatives with poor agonistic
activity on hb3-AR. This effect was paralleled by low relative effi-
cacy data (high qResp). An evaluation of different spacer units was
performed on the level of affinity for hb3-AR: inverting the 4-amin-
opiperidine unit resulted in a decrease of affinity from Ki = 23 nM
bridge
O
S
N
N
N
Effects of variations within the thienopyrimidine portion were
investigated next, the corresponding data is depicted in Table 3.
Two subsets of compounds might be formed for comparison: 50-
aryl derivatives showed very similar trends, as did all 50-non-aryl
variants, with the 50-methyl-60-phenyl set being somewhat in-be-
tween. Within the 50-Ar subset, 4-OH substitution always resulted
in superior EC50 values by a factor of 3–11 as compared to the
unsubstituted parent compounds, paralleled by a significant in-
crease in relative efficacy (compounds 8, 38 and 40 vs 10, 39 and
41, respectively). Both values were further optimized upon addi-
tional substitution with a 3-hydroxymethyl group, slightly for
derivative 42, significantly for compound 33. For the latter, the
EC50 value decreased by a factor of 80 (33 vs 10). Affinities for all
b-ARs deteriorated from the unsubstituted parent compounds 8,
38 and 40 to the 4-hydroxy derivatives 10, 39 and 41. By adding
the hydroxymethyl group, an increase of selectivity (based on Ki
values) for hb3-AR over hb1-AR was experienced for compounds
33 and 42 by a factor of at least 10, either by gaining hb3-AR affin-
ity or losing hb1-AR affinity. Still, no tuning of selectivity over hb2-
AR was detectable.
O
N
O
O
O
OH
R
N
N
N
R = Me, Bn, Ac, CH2CH2OH, (CH2)3-morpholine
EC50 (h
β
3-AR) = 150-1750 nM, ρResp = 15-46%
F
OH
spacer
S
O
N
N
H
N
N
H
N
N
N
N
N
N
Within the three sets of smaller thienopyrimidines, hb3-AR
Ki (hβ3-AR) =
770 nM
affinity was generally diminished as compared to the 50-aryl vari-
370 nM
8000 nM
3300 nM
ants, resulting in poor affinities of around 1.35 lM for all 4-hydro-
Figure 2. Bridge and spacer variations.9 For synthetic details, cf. Supplementary
data. qResp values were calculated using RL = 0.25.
xy derivatives 47, 50 and 53. Significantly better affinities could
not even be detected with the as yet most promising 4-hydroxy-

S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
6113
for compound 38 to 370 nM for the new derivative. Implementa-
tion of other diamines like (4-aminomethyl)piperidine, piperazine
or homopiperazine were tolerated even less. Keeping in mind the
central role of this spacer with regard to an overall orientation of
the thienopyrimidine moiety versus the aryloxy portion, such a
finding was not unexpected: these new spacer variants signifi-
cantly manipulate distance and angle between the two distal aro-
matic regions.
The next questions to be addressed were whether (1) the selec-
tivity over hb1-AR would prevail on a functional level (reflecting
relative efficacies on hb-AR subtypes), and (2) a certain selectivity
over hb2-AR might be achievable based on different relative effica-
cies on these two hb-AR subtypes.
And indeed, a certain selectivity for hb3-AR over hb1-AR could
be maintained on a cellular level, even though not predictably with
a loss or gain as compared to the binding events. The overall trend
was identified to be rather a (slight) deterioration of this selectivity
factor as compared to a selectivity ratio determined by Ki values:
for three out of twelve compounds, selectivity ratios based on Ki
values and on EC50 values were similar (compounds 12, 41 and
47; Table 4 vs Tables 1 and 3), for two data pairs, the EC50 ratio
was favorable (compounds 14 and 33; Table 4 vs Tables 1 and 2)
and for nine data pairs, the EC50 ratio was less favorable as com-
pared to the Ki ratios (compounds 9, 11, 17, 24, 38, 39 and 49;
Table 4 vs Tables 1 and 3), for compounds 17 and 24 even drasti-
cally. Such deviations between Ki ratios and EC50 ratios reflect dif-
ferences in relative efficacies of respective compounds on different
not only on binding data alone. Based on functional data, com-
pound 33 represented the most promising compound displaying
good to excellent selectivities towards hb1- and hb2-AR. Other thi-
enopyrimidine variants than 50-phenyl usually displayed no signif-
icant selectivity for hb3-AR over hb1-AR based on EC50 values
(compounds 38, 39, 41, 47 and 49; best selectivity factor of 3.3
for 47). Any slight selectivity based on Ki values was retained at
best, in most cases it was diminished. Selectivity for hb3-AR over
hb2-AR on a cellular level, on the other hand, seemed to be quite
decent for the latter compounds.
Concluding on these SAR investigations, compound 33 was se-
lected for further evaluations: a HitProfilingScreenÒ on 30 primary
molecular targets (GPCRs, ion channels and enzymes),19 a toxicity
profiling, and a determination of physicochemical properties were
initiated. With regard to toxicity uncritical ED50 values of 29 and
60 lM were measured in a PBMC viability and HepG2 assay,
respectively. Functional inhibition of the hERG channel was not
representing an issue either (determined electrophysiologically,
no inhibition at 2 lM). As to be expected from a log P of 2.90
(determined by capillary electrophoresis) and a TPSA of 110
(calcd),20 solubility in aq. media was found to be quite excellent
(>100 lM in a pH range from 4 to 7.4). Membrane permeability
as evaluated in a PAMPA was determined to be moderate to good
(about 10 nm/s at pH 5 and 7.4). The compound proved to be com-
pletely stable in artificial gastric juice, simulated intestinal fluids
and human plasma within a test period of 24 h. The HitProfiling-
ScreenÒ19 was further complemented by an evaluation on a set
of GPCRs comprising dopamine D4.2, 5-HT1 and the neurokinin
receptors NK1, NK2 and NK3. As the original hit molecule 1 result-
ing from the vHTS campaign was described in the literature to pos-
sess antagonistic effects on the latter three receptors,8 at least the
binding potential to these receptors had to be examined for com-
pound 33 as well. Compared to an on-target activity of 20 pM,
the whole set of binding data at a compound concentration of
hb-AR subtypes: relative efficacies qResp on hb1- and hb3-AR varied
by factors of 1.4–3.0 in favor of hb3-AR (=higher relative efficacy on
hb3-AR = lower qResp for hb3-AR in Table 4; valid for five out of
twelve compounds) and by factors of 1.5–100 in favor of hb1-AR
(for seven out of twelve compounds). The higher factors in favor
of hb1-AR reflect the more general loss of selectivity towards this
adrenoceptor on a functional level, now resulting in no selectivity
for hb3-AR over hb1-AR for compound 9 or even in a slight selec-
tivity for hb1-AR for derivatives 24, 38 and 49. As to selectivity
for hb3-AR over hb2-AR, antagonism at hb2-AR was identified for
all derivatives tested in a cellular assay. This result emphasizes
the importance of founding a hit-to-lead optimization process
10 lM proved to be quite promising: only at a1b-AR (rat), the hu-
man norepinephrine transporter and the dopamine receptor D4.2
radioligand displacement was observed with 88–97% at this con-
centration. For these off-targets, dose–response curves and func-
tional data have to be determined next. Affinities for neurokinin
Table 4
Functional data on hb1-, hb2- and hb3-ARs for selected compoundsa,9
OH
H
N
X
R
O
6'
5'
S
N
4
3
N
N
2'
Compd
X
R
Agonist assay hb1-AR
Antagonist assay hb2-AR, IC50 [nM]
Agonist assay hb3-AR
b
b
EC50 [nM] (IA)
qResp [%]
EC50 [nM] (IA)
qResp [%]
9
11
12
14
17
24
33
50-Ph
50-Ph
50-Ph
50-Ph
50-Ph
50-Ph
50-Ph
3-OH
2-OMe
3-OMe
2-F
3-CH2OH
3-NHCONH2
3-CH2OH–4-OH
0.57 (1.0)
9.6 (1.0)
450 (0.3)
600 (0.6)
70 (0.6)
0.31
1.5
71
50
0.15
2.3
100
0.62 (1.1)
2.5 (1.2)
21 (0.4)
8.3 (1.0)
26 (1.0)
2.1
5.5
44
16
15
205
250
220
245
1280
2250
110 (0.7)
3.5 (1.1)
150 (0.8)
0.02 (1.1)
64
0.08
0.15
38
39
41
47
49
50-(4-F-Ph)
50-(4-F-Ph)
20-(2-Thienyl)
50-Me
H
6.6 (1.0)
6.3 (1.0)
1.8 (1.1)
65 (1.0)
5.0 (1.0)
1.9
123
3000
327
5500
862
10 (1.0)
3.1 (1.1)
1.0 (1.3)
20 (1.1)
15 (1.2)
13
1.2
0.31
0.41
3.4
4-OH
4-OH
4-OH
H
0.34
0.43
0.61
2.2
20-Me
a
EC50 values given for agonistic effects, IC50 values represent antagonistic activity.
qResp equation, RL set to 0.25.
b

6114
S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
Scheme 2. Reagents and conditions: (i) epibromohydrine, K2CO3, acetone, 55 °C, 48 h; (ii) amine 7, iPrOH, 80 °C, 6 h; (iii) LiAlH4 (1 M in THF), CH2Cl2, 0 °C to rt, 6 h, 4–36%;
(iv) tBDMS-Cl, imidazole, CH2Cl2, rt, 18 h, 84%; (v) NaBH4, EtOH, 0 °C to rt, 1.5 h, 99% (crude); (vi) 2,2-dimethoxypropane, pTsOH (cat), CH2Cl2, rt, 48 h, 51%; (vii) KF, DMF, rt,
2 h, 75%; (viii) epibromohydrine, K2CO3, 2-butanone, 80 °C, 48 h, 81%; (ix) amine 7, LiClO4, DIEA, CH3CN, rt, 20 h; (x) HOAc/H2O (5:2), 70 °C, 0.5 h.
receptors were not significant, best binding event was identified
for NK2 with 52% displacement at 10 M.
Acknowledgments
l
Advancing towards a first tissue assay and in vivo investiga-
tions, bulk material of compound 33 was required. Following the
general synthetic route depicted in Scheme 1, which corresponds
to the sequence on the left in Scheme 2, a LiAlH4 reduction of the
ester group in compound 55 to give the hydroxymethyl unit of lead
compound 33 was scheduled as the final step (cf. also Supplemen-
tary data). This, however, was only achieved with low reproduc-
ibility and yield (4–36%; average yield 17%). The outcome was
compromised by an unfavorable workup procedure due to the
presence of aluminum salts remaining from the reducing agent
and the occasional necessity of removing byproducts. Conse-
quently, the total synthesis was re-evaluated, adapting the func-
tional group transformation strategy for the aryloxy portion from
the synthesis of Salmeterol:21 starting from aldehyde 2c, the carb-
aldehyde functionality was reduced to a hydroxymethyl group,22
which was incorporated into a cyclic acetal together with the adja-
cent phenolic hydroxy group (Scheme 2). However, this route
called for an extension of this sequence by two additional synthetic
steps as a direct generation and transformation of 2-(hydroxy-
methyl)benzene-1,4-diol was unsuccessful: a tBDMS protection
of the unhindered hydroxy group23 prior to the reduction of the
carbaldehyde functionality and its deprotection after acetal forma-
tion. Thus the preparation of the aryloxymethyloxirane 4c now be-
came a 5-step-synthesis with an overall yield of 26% as compared
to a one-step procedure (58%) for the respective 3-ethoxycarbonyl
derivative 4b incorporated within the original sequence. Despite
this drawback the total synthesis of 33 became advantageous fol-
lowing the alternative route due to its last step: final deprotection
of 56 to give 33 now succeeded with complete conversion without
any formation of byproducts. Nucleophilic oxirane opening of 4c
with amine 7 could be slightly optimized using LiClO4 as Lewis acid
catalyst.24
The authors thank Dr. Martin Lang, Dr. Wael Saeb and Oliver
Müller for synthetic support and Dr. Andrea Aschenbrenner for
helpful discussions.
Supplementary data
Supplementary data (additional synthetic Schemes, spectral
data for compound 33) associated with this article can be found,
in the online version, at doi:10.1016/j.bmcl.2010.08.039.
References and notes
1. Ursino, M. G.; Vasina, V.; Raschi, E.; Crema, F.; De Ponti, F. Pharmacol. Res. 2009,
59, 221.
2. (a) Wu, T.; Wang, L. Med. Hypotheses 2007, 69, 1114; (b) Lee, S.; Schwinger, R. H.
G.; Brixius, K. Pflugers Arch. Eur. J. Physiol. 2008, 455, 767.
3. Arch, J. R. S. Naunyn-Schmied. Arch. Pharmacol. 2008, 378, 225.
4. Information gathered from Thomson Reuters iddb3 database, MedTRACK and
www.clinicaltrials.gov.
5. Tyagi, P.; Tyagi, V.; Yoshimura, N.; Chancellor, M.; Yamaguchi, O. Drugs Future
2009, 34, 635.
6. Kordik, C. P.; Reitz, A. B. J. Med. Chem. 1999, 42, 181.
7. Tasler, S.; Baumgartner, R.; Aschenbrenner, A.; Ammendola, A.; Wolf, K.;
Wieber, T.; Schachtner, J.; Blisse, M.; Quotschalla, U.; Ney, P. Bioorg. Med. Chem.
Lett. 2010, 20, 3399.
8. Becker, O.; Chen, D.; Cheruku, S. R.; Dhanoa, D. S.; Heifetz, A.; Kalid, O.;
Kesavan, V.; Marantz, Y.; Mohanty, P.; Noiman, S.; Nudelman, R.; Shachem, S.;
Sharadendu A. WO 2004/014850; Chem. Abstr. 2004, 140, 199335; compounds
8 and 16 are likewise described as NK1 antagonists herein, Ki = 841 and 24 nM,
respectively. No functional data is presented.
9. All aryloxypropanolamine derivatives were tested as racemates. For a further
lead optimization process, the (S)-enantiomers proved to be more active than
the respective (R)-enantiomers.
10. Membrane preparations (CHO-K1 cell line) expressing human b1-, b2- or b3-
ARs (Bmax = 3.78, 1.68 and 47.2 pmol/mg protein, respectively), were purchased
from Euroscreen (now Euroscreen FAST or Perkin Elmer). Binding assays were
performed according to the manufacturer’s instructions. The radioligand for all
three receptor subtypes was
[
125I]-cyanopindolol (Amersham) (final
concentration of 0.05, 0.05 and 1.5 nM, respectively). Ki values were
calculated using the Cheng–Prusoff equation on IC50 determinations, which
were based on concentration curves using eight concentrations (half-
logarithmic) in duplicate.
Using a pharmacophore alignment on known hb3-AR agonists,
not including a single aryloxypropanolamine, a vHTS approach fol-
lowed by medicinal chemistry explorations around the resulting
hit molecule led to the identification of a highly potent and effica-
cious hb3-AR agonist (EC50 = 20 pM) of the aryloxypropanolamine
scaffold. This lead compound possesses promising selectivity over
hb1- and hb2-AR on a functional level and a favorable safety pro-
file. Based on these results, an evaluation of the relaxing potential
on pre-contracted human bladder detrusor strips was envisaged
and compound 33 advanced to the lead optimization stage, the re-
sults of which will be disclosed in due course.
11. Functional response of cells (agonistic or antagonistic) to the test compounds
was tested by measurement of cyclic AMP formation by HTRFÒ (Homogeneous
Time-Resolved Fluorescence) technology (Cisbio International) using a stable
cell line CHO-K1 expressing the human recombinant b3-AR (Euroscreen, now
Euroscreen FAST or Perkin Elmer; Bmax = 23 pmol/mg protein) according to the
manufacturer’s instructions. EC50 (agonists) and IC50 values (antagonists) were
determined by dose–response curves based on eight concentrations
(logarithmic) determined in quadruplicate in a 96 half-well plate in a final
volume of 100
ll. The antagonistic effect was determined by preincubation
with a test compound for 10 min followed by agonist stimulation (0.05 nM

S. Tasler et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6108–6115
6115
isoproterenol) for 30 min. All compounds were evaluated in both assay
formats.
16. Chlorothienopyrimidines 6 are commercially available, except for 4-chloro-5-
phenylthieno[2,3-d]pyrimidine, cf. Supplementary data.
Functional data on hb1-AR and hb2-AR were determined at Euroscreen (both,
in an agonist and an antagonist assay; Bmax = 11.1 and 2.3 pmol/mg protein for
hb1-AR and hb2-AR, respectively). EC50 and IC50 values, respectively, were
determined by dose–response curves based on seven concentrations
(logarithmic) determined in duplicate.
17. Kumagai, T.; Okubo, T.; Kataoka-Okubo, H.; Chaki, S.; Okuyama, S.; Nakazato, A.
Bioorg. Med. Chem. 2001, 9, 1349.
18. In analogy to, for example: (a) Yang, W.; Wang, Y.; Roberge, J. Y.; Ma, Z.; Liu, Y.;
Lawrence, R. M.; Rotella, D. P.; Seethala, R.; Feyen, J. H. M.; Dickson, J. K., Jr.
Bioorg. Med. Chem. Lett. 2005, 15, 1225; (b) Howe, R.; Rao, B. S.; Holloway, B. R.;
Stribling, D. J. Med. Chem. 1992, 35, 1751.
19. https://pharmacology.ricerca.com/Catalog/AssayCatalog/AssayCatalog.aspx, than
‘Molecular Packages’.
20. Program packages MOE and Maestro/QikProp.
21. (a) Procopiou, P. A.; Morton, G. E.; Todd, M.; Webb, G. Tetrahedron: Asymmetry
2001, 12, 2005; (b) Liu, J.; Zhou, D.; Jia, X.; Huang, L.; Li, X.; Chan, A. S. C.
Tetrahedron: Asymmetry 2008, 19, 1824.
22. Ono, M.; Kawashima, H.; Nonaka, A.; Kawai, T.; Haratake, M.; Mori, H.;
Kung, M.-P.; Kung, H. F.; Saji, H.; Nakayama, M. J. Med. Chem. 2006, 49,
2725.
23. Liu, A.; Dillon, K.; Campbell, R. M.; Cox, D. C.; Huryn, D. M. Tetrahedron Lett.
1996, 37, 3785.
24. Cribiu, R.; Allevi, P.; Anastasia, M. Tetrahedron: Asymmetry 2005, 16, 3059.
12. The term ‘affinity’ is strictly used for the description of binding data, whereas
‘activity’ refers to functional data from cellular assays throughout this Letter.
13. (a) Ehlert, F. J. Naunyn-Schmied. Arch. Pharmacol. 2008, 377, 549; (b) Lutz, M.;
Kenakin, T. Quantitative Molecular Pharmacology and Informatics in Drug
Discovery; Wiley & Sons: Chichester, 1999. Chapter 3.6, p 86ff; ISBN 0 471
98861 8.
14. Hanania, N. A.; Sharafkhaneh, A.; Barber, R.; Dickey, B. F. Am. J. Respir. Crit. Care
Med. 2002, 165, 1353.
15. In analogy to, for example: (a) Wagner, S.; Kopka, K.; Law, M. P.; Riemann, B.;
Pike, V. W.; Schober, O.; Schäfers, M. Bioorg. Med. Chem. 2004, 12, 4117; (b)
Elzein, E.; Shenk, K.; Ibrahim, P.; Marquart, T.; Kerwar, S.; Meyer, S.; Ahmed, H.;
Zeng, D.; Chu, N.; Soohoo, D.; Wong, S.; Leung, K.; Zablocki, J. Bioorg. Med. Chem.
Lett. 2004, 14, 973.