D. M. Garrido et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1840–1845
1843
Table 2. Diastereo- and enantioselectivity for the asymmetric cyclopropanation of 4-substituted styrenes
Entry X =
Chirality of (bis)oxazoline R = Ratio of III/IVa % Yield of IV Chirality of major enantiomer of Vd % ee of Ve
1
2
3
4
5
NO2
NO2
NO2
(S,S)
(R,R)
(R,R)
t-Bu 1:4
t-Bu 1:4
24b
30b
31b
61c
na
(ꢀ)-(R,R)
(+)-(S,S)
(+)-(S,S)
(+)-(S,S)
(+)-(S,S)
95
95
83
94
57
Ph 1:5.7
t-Bu 1:2.3
NHC(O)CF3 (R,R)
NHC(O)CF3 (R,R)
Bn
1:2.3
na, not determined.
a Ratio estimated from 1H NMR of reaction mixture after workup.
b Mixture of IVA and diethylfumarate. Yield of IVA estimated by 1H NMR.
c Refers to the isolated yield of IVB in >95% purity.
d The absolute stereochemistry was assigned based on the VCD analysis Ref. 13.
e Determined by chiral SFC analysis Ref. 12.
oxazoline ligand (Table 2, entry 2). Unfortunately this
ligand is not commercially available and needed to be
prepared in three steps from (D)-tert-leucine.14 Since
unnaturally occurring (D)-tert-leucine is relatively
expensive, the use of the commercially available (R,R)-
phenyl-bisoxazoline ligand was also investigated (Table
2, entry 3). The ratio of products IIIA/IVA was slightly
improved with the phenyl- versus the tert-butyl-ligand
but the enantiomeric excess of product V was signifi-
cantly worse (83% vs 95% ee, respectively).
The compounds presented up to this point all contain a
carboxylic acid headgroup and thus resemble free fatty
acids themselves. The question remained whether this
functional group was critical for activity, so a set of
amide analogs was prepared. Replacement of the acid
would remove the potential for acyl glucuronide forma-
tion and perhaps lower binding to plasma proteins, such
as albumin, which is typically quite high (>99%) for this
class of compounds. The primary amide in series A
(16A) was equipotent to the corresponding acid 2A,
while in series B the primary amide (16B) was more po-
tent than the acid 2B (Table 3). Simple mono-alkylation
with a methyl group (derivative 17A) or a bulkier iso-
propyl substituent (compounds 18A and 18B) was well
tolerated displaying potencies similar to those of the
corresponding acids and efficacies similar to those of
the primary amides. Though not statistically significant,
there does appear to be a drop in maximal efficacy sug-
gesting that the primary and secondary amides may only
be partial agonists. This trend was much more pro-
nounced with the tertiary amides (the dimethyl-amide
19A and the pyrrolidine-amide 20A) whose maximal effi-
cacy deteriorated to almost half of that elicited by the
free carboxylic acid 2A (e.g., 50% max for 19A and
46% max for 20A vs 86% max for 2A). A similar trend
also appeared within the cyclopropyl-linked analogs in
series B (21B–23B). In addition, increasing the bulk of
the amide substituent detrimentally affects activity. For
instance, changing from the N-isopropyl (24B) to N-cyc-
lobutyl (25B) to N-(R)-phenethylamide (26B) resulted in
a progressive decrease in efficacy and potency. The more
polar analog 27B was worse still. On the contrary, the
smaller hydroxamic acids 28A and 28B resemble the free
carboxylic acids in both potency and efficacy at GPR40.
One drawback to the initial cyclopropanation route was
the poor isolated yield of compound V. It was hypothe-
sized that the nitrostyrene starting material IIA might be
polymerizing under the reaction conditions and there-
fore the 4-trifluoroacetamide-styrene IIB was subjected
to the cyclopropanation procedure described above
(Table 2, entry 4). Although the diastereoselectivity of
the reaction was a little worse than that observed for
substrate IIA, the desired trans-cyclopropyl analog
IVB was obtained in a 61% isolated yield, which was a
significant improvement. The trifluoroacetamide pro-
tecting group was cleaved with NaBH4 in EtOH in an
unoptimized 61% yield to afford compound V in 94%
ee. Although the overall yield is still modest, this method
has been used to produce over 30 g of the enantiomeri-
cally pure (S,S)-aniline V.15 The use of the commercially
available (R,R)-benzyl-bisoxazoline ligand was also
investigated with substrate IIB but the reaction proceed-
ed with poor enantioselectivity (Table 2, entry 5).
Both the (R,R)- and (S,S)-anilines of V as well as the
(ꢀ)-cis-isomer isolated as the minor product from the
cyclopropanation of IIB (Table 2, entry 4) were subse-
quently converted to the final products 13A, 14A, 14B,
and 15B in >95% ee. The (+)-(S,S)-enantiomer 14A be-
haved as a potent, full agonist at the GPR40 receptor
with a pEC50 = 7.91 (Table 1) and is >45-fold more ac-
tive than the corresponding (ꢀ)-(R,R)-isomer 13A
(pEC50 = 6.25). The (ꢀ)-(R,R)-analog 13A also appears
to be a partial agonist producing a maximal response
only 64% of that produced by compound 1. The (+)-
(S,S)-derivative in series B (compound 14B) was also ac-
tive with a pEC50 = 8.31, although in neither series was
the potency of the enantiomerically pure (S,S)-isomer
statistically different from the potency of the racemic
mixture. The (ꢀ)-cis-isomer 15B was less active than
both the corresponding (+)-(S,S)-trans-isomer 14B and
the simple ethyl-linked analog 2B.
In conclusion, a variety of novel compounds that acti-
vate GPR40 at low nanomolar concentrations have been
identified, the majority of which behave as full agonists
as compared to the endogenous long-chain fatty acid li-
gands. The introduction of the (S,S)-cyclopropyl acid
headgroup led to a significant improvement in potency
over the ethyl-linked analogs, while the (R,R)-cyclopro-
pyl enantiomer was only weakly active. The asymmetric
syntheses of the cyclopropyl intermediates were accom-
plished in high enantiomeric excess and good yield from
commercially available 4-aminostyrene. Structure–activ-
ity relationships revealed that the acid itself is not criti-
cal for activity but typically elicits a higher agonistic
response than that observed with the carboxamide