Synthesis of 2-aryloxypropanoic acids analogues of clofibric acid and assignment of the absolute configuration by 1H NMR spectroscopy and DFT calculations

Article abstract of DOI:10.1016/j.tetasy.2008.04.010

A new set of optically active 2-aryloxypropanoic acids has been synthesized through a simple strategy, in good yields and excellent enantiomeric excesses. Their absolute configuration was assigned by means of a NMR-based approach consisting of the derivatization of the carboxylic acids with ethyl mandelate and the comparison of the chemical shifts of the obtained diastereomers. The effectiveness of such an approach has been tested against a larger set of chiral α-substituted-carboxylic acids and by performing high level density functional theory (DFT) calculations on a set of low energy conformations for each diastereomeric derivative. The employed computational procedure has enabled us to find a semiquantitative relationship between the experimental NMR data and the theoretically calculated energy gaps which confirms the theoretical foundations of the NMR strategy and allows to understand when and why it is most likely to fail.

Full text of DOI:10.1016/j.tetasy.2008.04.010

Available online at www.sciencedirect.com
Tetrahedron: Asymmetry 19 (2008) 989–997
Synthesis of 2-aryloxypropanoic acids analogues of clofibric acid and
assignment of the absolute configuration by H NMR spectroscopy
and DFT calculations
1
Alessandra Ammazzalorso, Giancarlo Bettoni, Barbara De Filippis, Marialuigia Fantacuzzi,
Letizia Giampietro, Antonella Giancristofaro, Cristina Maccallini, Nazzareno Re,
Rosa Amoroso* and Cecilia Coletti
Department of Drug Sciences, University ‘G. D’Annunzio’, Via dei Vestini, 66100 Chieti, Italy
Received 26 February 2008; accepted 8 April 2008
Abstract—A new set of optically active 2-aryloxypropanoic acids has been synthesized through a simple strategy, in good yields and
excellent enantiomeric excesses. Their absolute configuration was assigned by means of a NMR-based approach consisting of the deriv-
atization of the carboxylic acids with ethyl mandelate and the comparison of the chemical shifts of the obtained diastereomers. The effec-
tiveness of such an approach has been tested against a larger set of chiral a-substituted-carboxylic acids and by performing high level
density functional theory (DFT) calculations on a set of low energy conformations for each diastereomeric derivative. The employed
computational procedure has enabled us to find a semiquantitative relationship between the experimental NMR data and the theoret-
ically calculated energy gaps which confirms the theoretical foundations of the NMR strategy and allows to understand when and why it
is most likely to fail.
Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction
cular mechanics (MM) or semiempirical methods5 and,
very recently, on molecular dynamics simulations6 or den-
sity functional theory (DFT) calculations4 have supported
the conformational hypothesis that the NMR strategy is
based upon. Compared to other substrates, relatively few
studies describing the determination of the absolute config-
uration of carboxylic acids by NMR spectroscopy have
been reported, where the acidic function has been deriva-
tized in ester5,7,8 or amide9–14 derivatives, by using chiral
auxiliaries leading to an effective aromatic shielding. Rigu-
era et al.5 described a powerful application of this method
to a series of a-chiral carboxylic acids by using a very effi-
cient chiral reagent, ethyl 2-hydroxy-2-(9-anthryl)acetate;
the use of methyl mandelate was also reported,15 showing
the ability to correlate the chemical shift with the stereo-
chemistry of the analyzed compounds, even if with very
small values of difference in chemical shift (DdRS), inter-
preted with a model where only steric considerations were
taken into account.
The NMR-based methodology for the stereochemical
assignment of chiral pharmaceutical compounds has
received great interest in the last years, because of its
simplicity and high reliability.1,2 Several reports have been
described, in which different substrates have been investi-
gated by NMR spectroscopy;3 these studies are generally
based on substrate derivatization with two enantiomers
of a suitable chiral reagent and the comparison of the
NMR spectra of the resulting pair of diastereomers.
Indeed, the preference of the latter for a specific conforma-
tion leads to a variation of the chemical shift of the substit-
uents of the asymmetric carbon of the substrate, which can
be directly related to its absolute configuration. This proce-
dure has been successfully employed for a number of chiral
substrates: primary and secondary alcohols, primary
amines, carboxylic acids, diols, triols, amino alcohols and
thiols.4 In some cases, theoretical studies based on mole-
Fibrates exert their hypolipidemic activity by the activation
of PPARa, a nuclear receptor involved in lipid metabolism
by regulating the transport and catabolism of fatty
*
Corresponding author. Tel.: +39 0871 3554686; fax: +39 0871
3554911; e-mail: ramoroso@unich.it
0957-4166/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2008.04.010

990
A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
O
4
O
O
O
O
O
3
O
HN
Br
O
4
HN
OH
6
OH
OH
OH
OH
OH
O
5
O
3
(R)-1a
(R)-1b
(S)-1d
(S)-1c
5
(R)-1e
(R)-1f
O
O
+
L1
L1
*
*
O
COOEt
O
COOEt
L1
L2
(X,R)-2a-f
(X,S)-2a-f
Figure 1. Training set of a-chiral carboxylic acids and general formula of their mandelic esters.
acids;16,17 more potent and selective PPARa ligands have
already been described,18–21 with a superior clinical profile
for a therapeutic intervention in dyslipidemia and other
metabolic disorders. With the aim to discover new PPARa
agonists, we have synthesized new optically active 2-aryl-
oxypropanoic acids analogues of clofibric acid, the active
metabolite of clofibrate. Herein, we focus on the applica-
tion of an NMR-based method to our 2-aryloxypropanoic
acids, with the aim of realizing a convenient and very sim-
ple determination of their absolute configuration. We
decided to utilize an inexpensive alcoholic chiral reagent,
easily accessible in both enantiomeric forms, the ethyl-2-
hydroxy-2-phenylacetate (ethyl mandelate): despite its
structural simplicity, it possesses the minimal requirements
to produce the anisotropic effect. In order to test the effi-
cacy and reliability of ethyl mandelate as a chiral reagent,
we first applied this methodology to a series of commer-
cially available a-chiral carboxylic acids with a different
steric hindrance (Fig. 1). We have synthesized a series of
chiral derivatives of clofibric acid (Fig. 2) with a simple
reaction scheme, in good yields and excellent enantiomeric
excess and applied the NMR-based approach to assign
their absolute configuration.
grees of freedom; (ii) selected the most representative con-
formations by applying a cluster analysis; (iii) performed
on these conformations a geometry optimization through
quantum mechanical density functional methods which
are well known to produce reliable geometries and accurate
energies for organic molecules. This is particularly impor-
tant because the energy differences between the conforma-
tions, responsible for the variations in the chemical shifts,
that is, the DdRS parameter, are reported to be within
1 kcal/mol, a value which is well below the accuracy of
MM and semiempirical methods. Furthermore, besides
performing DFT calculations in gas phase, we carried out
geometry optimizations in solution, using CHCl3 as a sol-
vent in agreement with the experimental NMR conditions,
by a Poisson-Boltzmann polarizable continuum model.
The results can thus be compared to the experimental data
in a much more consistent and reliable fashion.
Indeed, it has been possible to find a semiquantitative rela-
tionship between the experimental DdRS and the calculated
energy gaps, strongly supporting the hypothesis of a spe-
cific conformational preference underneath the present
NMR strategy. Furthermore, the analysis of the stereo-
chemistry of the diastereomers preferred conformations
and of their energy differences (i.e., their different popula-
tions) has also enabled us to deepen our understanding
of the applied NMR strategy for the assignment of the
absolute configuration of the chiral a-substituted-carbox-
ylic acids so as to foresee when it is expected to work best.
O
O
N
O
O
OH
OH
Cl
Cl
(R)-1h
(R)-1g
O
2. Results and discussion
O
O
N
S
O
OH
OH
In the literature,5 the observed chemical shift for L1/L2
groups (directly bonded to the stereogenic carbon of the
investigated carboxylic acids) upon derivatization with
the (R)- and (S)-enantiomer of a chiral aryl alcohol was ex-
plained according to the preference of the acids for the
antiperiplanar (ap) rather than the synperiplanar (sp) con-
formation of the Cꢀa hydrogen respect to the C@O oxygen
(Fig. 3a). This preference was shown to be maintained in
the esters, and, since it is known that the alcohol part of
an ester adopts a preferred conformation with the C–H
synperiplanar with respect to the C@O (Fig. 3b), it leads
to an antiperiplanar conformation, ap(H,H), for the alco-
holic C–H and the acidic Ca–H, and thus to a selective
shielding by the aryl group on only one of the L1/L2 groups
(R)-1i
(R)-1j
Figure 2. Selected chiral analogues of clofibric acid.
We also performed theoretical calculations on all the dia-
stereomeric couples of (R)- and (S)-ethyl mandelate with
both the first and second series of acids, so as to confirm
that the sign of the DdRS parameter is correlated to a con-
formational preference of the esters. Although the aim of
such calculations is very much alike Ref. 5, the procedure
we have used is much more general and accurate, since
we (i) performed a conformational analysis on all the de-

A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
991
O
H
O
H
H
R
O
O
H
O
C
O
L1
L2
L2
L1
≡
≡
H
≡
(a)
L2
(b)
HO
R2
HO
H
R
O
R1
R2
L1
L2
L1
H
R1
sp
ap
H
O
L1
L2
CO2Et
(R)-ethyl mandelate
L2
L1
EtO2C
Ph
≡
O
Ph
O
H
H
L1
L2
ΔδRS<0
ΔδRS>0
L1
L2
ap(H,H)
(c)
H
H
H
O
L1
L2
L2
L1
ap
Ph
≡
O
EtO2C
(S)-ethyl mandelate
EtO2C
Ph
ap(H,H)
H
H
Figure 3. (a) Antiperiplanar (ap) and synperiplanar (sp) conformations of a-chiral carboxylic acids. (b) Synperiplanar conformation (alcoholic C–H and
acidic C@O) preferentially adopted by the esters of a-chiral carboxylic acids with a chiral secondary alcohol. (c) Upon esterification of the ap
conformation of the a-chiral carboxylic acids with the (R)- or (S)-enantiomers of ethyl mandelate, the alcohol C–H and the acidic Cꢀa–H bonds adopt an
ap(H,H) conformation, so that in the two diastereomers the aromatic shielding is exerted on different L1/L2 groups, causing a different sign for the DdRS of
the two groups.
in turn (Fig. 3c, with ethyl mandelate as an esterifying
agent). As a consequence, hydrogens of the L1 group will
have a negative value of the DdRS parameter (DdRS < 0),
while those belonging to the L2 group will have DdRS > 0,
allowing a ready assignment of the absolute configuration
of the original carboxylic acid by the comparison of the
NMR spectra of both diastereomers. This conclusion was
supported by molecular mechanics and semiempirical
(AM1 and PM3) calculations on a number of carboxylic
acids and their methyl esters, all showing the ap as the most
stable conformation by 1 kcal/mol or less. The magnitude
of the DdRS parameter depends on the different populations
of the ap and sp conformers, and thus on the energy gap
between the corresponding conformations. However, the
ability of the chosen chiral agent to selectively shield the
signal also plays a significant role in determining the value
of DdRS, larger aromatic cone exerted by the aryl portion of
the chiral alcohol leading to more effective shielding.
the possible discrepancies in their observed behaviour as
well as to investigate the possibility of establishing a quan-
titative relationship between the energy gaps and experi-
mental DdRS values.
We first prepared the diastereomeric esters 2a–f by the
reaction of the commercially available a-homochiral car-
boxylic acids 1a–f with both (R)- and (S)-ethyl mandelate,
using N,N0-dicyclohexylcarbodiimide (DCC) and 4-dimeth-
ylaminopyridine (DMAP), in dichloromethane at room
temperature (Fig. 1). The reaction mixtures were purified
by silica gel, and protonic spectra of the corresponding dia-
stereomeric esters 2 have been recorded (300 MHz,
CDCl3), with the aim of applying the NMR-based ap-
proach to assign their absolute configuration. The values
of the chemical shifts and related DdRS for the aliphatic
protons of the L1/L2 substituents are reported in Table 1.
Due to their overlap with the signals of the aryl protons
of ethyl mandelate, the signals of the aromatic protons of
the L1/L2 substituents could not be unambiguously as-
signed for all the compounds. Figure 4a shows the DdRS
values for those acids in the training set which could be
unequivocally determined. In all the cases, the DdRS sign
correctly predicts the absolute configuration, although for
compound 2a its value is very close to zero. In general,
Herein, we investigate the possibility to use the ethyl
mandelate, which is commercially available in both
enantiomeric forms and less expensive than the ethyl
2-hydroxy-2-(9-anthryl)acetate5; although providing
a
smaller aromatic shielding cone and less steric hindrance
than 2-hydroxy-2-(9-anthryl)acetate, and thus leading to
smaller values of DdRS, it is reported that a very similar
chiral agent, methyl mandelate, has been successfully em-
ployed.15 However, in that study, no real understanding
of the factors determining the DdRS values was attained,
the results being interpreted with a model where only steric
considerations were taken into account, with no reference
to the acid/ester conformational preference. In order to
shed light on this point and because the energy gaps be-
tween the ap and sp conformers calculated previously5 at
a low level of theory, although showing the same trend,
lie below the accuracy of the employed method, we have
undertaken a series of quantum mechanical DFT calcula-
tions on the considered esters. Such an approach also al-
lowed us to find a more sound correlation and explain
Table 1. Chemical shifts and DdRS values for esters 2a–f
Compd
Signal
d (ppm) (X,R)
d (ppm) (X,S)
DdRS
2a
2b
2c
2c
2c
2d
2e
2f
CH3
CH3
CH3 (5)
CH2 (3)
CH3 (4)
CH3
1.893
1.575
1.240
1.515
0.922
1.574
1.572
1.851
1.626
0.997
1.890
1.597
1.192
1.566
0.990
1.563
1.458
1.704
1.594
0.946
+0.003
ꢁ0.022
+0.048
ꢁ0.051
ꢁ0.068
+0.011
+0.114
+0.147
+0.032
+0.051
CH3
CH2 (3)
CH (4)
CH3 (5,6)
2f
2f

992
A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
Table 2. Chemical shifts and DdRS values for esters 2g–j
- 0.025
- 0.042
- 0.020
- 0.025
DdRS
O
Compd
Signal
d (ppm) (R,R)
d (ppm) (R,S)
a
OR*
2g
2h
2i
CH3
CH3
CH3
CH3
1.745
1.797
1.815
1.898
1.656
1.703
1.790
1.814
+0.089
+0.094
+0.025
+0.084
- 0.051
+ 0.011
2d
2j
O
O
- 0.060
O
S
- 0.020
OR*
O
- 0.024
OR*
N
- 0.030
- 0.040
O
H
b
+ 0.025
Cl
+ 0.089
γ ω
α
Ph
- 0.040
L1
L2
O
CO2Et
2g
2i
H
Figure 4. DdRS values of aromatic and aliphatic protons determined for
the diastereomeric esters 2d of the training set (panel a), and esters 2g and
2i of the newly synthesized acids (panel b).
Figure 5. The three torsional angles determining the mutual position of
the alcoholic C–H and the acidic Cꢀa–H bonds in the esters. Distances
calculated with respect to these angles were used to form clusters whose
representative conformations were then minimized.
the absolute value of DdRS is rather small, due, as expected,
to the moderate anisotropic effect exerted by ethyl mandel-
ate, but it is, however, larger than that obtained when using
methyl mandelate.
preference of the esters, we first performed a conforma-
tional search where all the degrees of freedom were taken
into account; the results were clustered according to the
three dihedral angles a, c and x (Fig. 5) determining the
mutual positions of the phenyl ring and the L1/L2 substit-
uents to the Cꢀa of the carboxylic acid. Ten clusters were
considered for each diastereomer and the leading members
(i.e., the representative conformation of each cluster, cho-
sen as the lowest energy one) were qualitatively analyzed
by visual inspection. In particular, we examined the follow-
ing features: (i) the relative position of the alcoholic C–H
and the C@O, determined by both c and x, to verify if they
are indeed synperiplanar; (ii) the relative position of the
acidic Cꢀa–H hydrogen and the C@O oxygen around the a
torsional angle, to verify if they are periplanar (either ap
or sp); (iii) the relative position of the acidic Cꢀa–H hydro-
gen and the alcoholic C–H, determined by the joined values
of all the three angles a, c and x, which is crucial for ver-
ifying the correctness of the foundations of the NMR strat-
egy under study: the considered hydrogens have to be
periplanar to lead to the aromatic shielding of the acidic
L1/L2 groups by the phenyl ring. Throughout this analysis
and in the following we will use the term periplanar in a
slightly more relaxed sense than in the original Klyne and
Prelog23 acceptation : here we will use synperiplanar and
antiperiplanar to define a flexible dihedral angle of
Next, we performed the synthesis of the chiral derivatives
of clofibric acid (R)-1g–j (Fig. 2): Mitsunobu coupling22
of the appropriate commercial alcohols and (S)-ethyl lac-
tate, in the presence of triphenylphosphine and DIAD, in
THF dry at room temperature, furnished the (R)-3g–j
esters. After hydrolysis under acidic conditions, the desired
(R)-1g–j acids were obtained (Scheme 1); the enantiomeric
purity of (R)-1g–j was evaluated by a capillary electropho-
retic run, by using b-cyclodextrin as a chiral selector. Anal-
yses showed enantiomeric excesses up to 98% for all the
tested compounds. The (R,R)- and (R,S)-2g–j diastereo-
meric esters were afforded via Mitsunobu reaction with
(R)- and (S)-ethyl mandelate, and then analyzed by
NMR. The values of the chemical shifts and the related
DdRS for the aliphatic protons of the L1/L2 substituents
are reported in Table 2. Figure 4b again shows the aro-
matic DdRS values for those acids, the signals of which
could be unambiguously determined. Also in this case a
complete agreement between the DdRS sign and the stereo-
chemistry of substrates has been obtained. Besides, when
compared with the results of the training set (Table 1),
these compounds show larger values of DdRS
.
0
45° (syn) or 180 45° (anti). This, however, does not
O
O
HCl, 60˚
diminish the generality and the conclusions of the present
discussion since in spite of these less strict definitions, the
aromatic cone of the phenyl ring is directed towards a well
defined region, so as to selectively shield only one of the L1/
(S)-ethyl lactate
Ar
O
Ar
O
Ar OH
OEt
OH
Ph3P, DIAD,
THF dry
(R)-3g-j
(R)-1g-j
L2 substituents of the acidic Cꢀ.
a
(R)- and (S)-ethyl
The analysis of the clusters representative structures de-
scribed above highlighted some important qualitative fea-
tures shared by all the distereoisomers: (i) all the
representative structures within 25 kJ/mol from the mini-
mandelate
O
O
+
Ph3P, DIAD,
THF dry
Ar
O
Ar
O
O
COOEt
O
COOEt
(R,S)-2g-j
(R,R)-2g-j
Scheme 1.
It is noteworthy mentioning that, contrarily to its common usage which
identifies it with ‘coplanar’, the term periplanar was introduced to define
a torsional angle of 0 30° (syn) or 180 30° (anti), see, for example,
Ref. 24.
In order to confirm the hypothesis of a relationship be-
tween the sign of the DdRS values and a conformational

A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
993
Table 3. Energy differences (in kcal/mol) in solution between the ap(H,H)
and sp(H,H) conformations of the (R)- and (S)-ethyl mandelate deriva-
tives of acids in Figure 1
mum energy structure show a synperiplanar disposition of
the alcoholic C–H and the C@O group, which is in full
agreement with the previous studies5,7; (ii) the acidic
Cꢀa–H hydrogen and the C@O oxygen are, within the re-
laxed notation underlined above, periplanar. The largest
deviations from coplanarity are found for the diastereo-
mers of acids 1e and 1f, which are the compounds showing
the largest conformational flexibility. For most diastereo-
mers (but not all), the minimum energy structure has an
ap conformation in agreement with the results of the liter-
ature;5 (iii) as a consequence of the relative positions of the
groups described in the two preceding points, the alcoholic
C–H and the acidic Cꢀa–H hydrogens lie approximately in
the same plane. Here, deviations from coplanarity, being
the result of the joined deviations of three angles, are, in
some cases, slightly larger than those described before,
however, the relative position of the two hydrogens can al-
ways be assigned as syn or anti according to the definition
given above. Since it is the mutual position of these two
hydrogens to determine the direction of the aromatic
shielding cone, in the following we will analyze and classify
our results according to their anti periplanarity, ap(H,H),
or syn periplanarity, sp(H,H). It is interesting to note that
the phenyl ring responsible for the shielding, due to steric
reasons, is always oriented so as to direct its aromatic cone
towards the acidic moiety of the ester, the variations in the
various diastereomers amount to 15° at most and can
only account for the small differences in the numerical val-
ues of the DdRS (see below) passing from one system to
another.
Compd
DER(sol)
DES(sol)
2a
2b
2c
2d
2e
2f
1.79
ꢁ0.15
0.11
2.12
2.10
ꢁ1.31
0.96
0.90
ꢁ0.11
1.08
2.00
1.09
A positive value means that the sp(H,H) conformation is higher in energy.
Table 4. Energy differences (in kcal/mol) in solution between the ap(H,H)
and sp(H,H) conformations of the (R)- and (S)-ethyl mandelate deriva-
tives of acids in Figure 2
Compd
DER(sol)
DES(sol)
2g
2h
2i
2.72
4.33
a
0.71
ꢁ0.23
ꢁ1.23
0.53
2j
2.73
A positive value means that the sp(H,H) conformation is higher in energy.
a All the calculations converged to an ap(H,H) conformation.
The results for (R)-2-bromopropanoic acid are in agree-
ment with the very small experimental DdRS value, 0.003,
and can be easily explained by examining the geometries
of the most stable conformations for the (R,R)- and
(R,S)-esters (Fig. 6). For both diastereomers, the most sta-
ble conformations have the large electronegative bromine
atom located away from the C@O oxygen atom. In the case
of the (R,S)-ester, the ap(H,H) conformation has instead
the bromine atom close to the C@O oxygen and is thus
higher in energy (more than 1 kcal/mol) than the corre-
sponding sp(H,H) one. This results in a shielding effect of
the aryl ring on the same group for both the (R,S)- and
(R,R)-esters and in a DdRS value which is practically zero.
We have optimized the structures of the clusters’ represen-
tative members by DFT methods both in the gas phase and
in the solution. Although showing approximately the same
trend, only the latter, consistent with the experimental con-
ditions, will be discussed. Again, qualitatively, the results
were similar for all the diastereomers: upon minimization,
some of the structures converged to the same minimum;
variations, sometimes large, from the original geometry of-
ten occurred, but always preserving a well defined ap(H,H)
or sp(H,H) orientation. In many cases, especially when the
conformational flexibility was higher, the energy sequence
of the structures obtained from the MM conformational
search changed after QM minimization. This is remark-
able, as it shows that a rigorous approach to a theoretical
understanding of the present NMR strategy cannot be
dealt with, without the aid of accurate QM techniques
and appropriate treatment of the solvent effects.
We next calculated the energy gaps, DE, between the lowest
energy ap(H,H) conformation and the lowest energy
sp(H,H) conformation within the minimized set for each
diastereomer. Results are reported only for solution phase
in Table 3 for the esters derived from the training set of a-
chiral carboxylic acids and in Table 4 for those obtained
from the analogues of clofibric acid synthesized in the pres-
ent work.
Figure 6. Most stable conformations for the (S)-ethyl mandelate ester (a)
and (R)-ethyl mandelate ester (b) of the (R)-2-bromopropanoic acid. In
both cases the preferred conformations are those with the bromine atom
away from the C@O oxygen, corresponding, respectively, to a sp(H,H)
and an ap(H,H) conformations.
Data in Table 3 show in all the cases, except for acid 1a,
that is, (R)-2-bromopropanoic acid, a significant preference
for the ap(H,H) conformer, the small negative values for
ester 2d being within the accuracy of the method.
Data in Table 4, relative to the analogues of clofibric acid,
show a similar behaviour, however, a more homogeneous

994
A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
pattern can be found: for all the (R,R)-esters the ap(H,H)
conformation is much more stable—of at least 2.5 kcal/
mol—than the corresponding sp(H,H) one, meaning that
such diastereomers at room temperature can all be found
quantitatively in an ap(H,H) conformation. For the
(R,S)-esters the DdRS value, on the contrary, is less than
1 kcal/mol, and in two cases, compounds 2h and 2i, there
is a preference for the sp(H,H) conformation. For com-
pound 2h, this preference amounts to only 0.23 kcal/mol
in solution (the gas phase calculation shows a slight prefer-
ence for the ap(H,H) conformation) and does not seem to
have particular consequences on the experimental DdRS
which is rather large, while, for compound 2i, DERS values
are between ꢁ1 and ꢁ2 kcal/mol, leading to a significant
preference for the sp(H,H) conformation in the (R,S)-ester,
and thus to the smallest DdRS value for the set. In any case,
for all the analogues of clofibric acid, the ap(H,H) prefer-
ence in the (R,R)-esters is so overwhelming that the applied
NMR strategy can always be used to correctly predict the
absolute configuration of the acids.
the ap(H,H) and sp(H,H) conformers in both diastereo-
mers or to an equal excess of the ap(H,H) and sp(H,H) con-
former, respectively, in the two diastereomers (in both
cases this should lead to a null DdRS value); a value of
RS
X
¼ ꢁ1 corresponds to the presence of the sp(H,H)
ap–sp
conformer only for both diastereomers.
RS
ap–sp
The obtained X
values have been linearly correlated
with the experimental DdRS taken with a plus sign when-
ever predicting the right assignment of the L1 and L2
groups. Since a correct correlation can be only established
by comparing the signal of the hydrogens located at similar
distances from the phenyl, we have considered DdRS values
belonging to groups in a-position with respect to the stereo-
genic carbon atom, that is, we have excluded lines 5, 9 and
10 in Table 1 from the correlation. The results are reported
in Figure 7 for the training set, while, in Figure 8, the newly
synthesized analogues of clofibric acid have been included
in the set. Figure 7 shows a high correlation between the
RS
ap–sp
calculated X
and the experimental DdRS, with a R2 va-
lue of 0.846, showing that, although the correlation is sig-
nificant also in vacuum (R2 = 0.533), solvent effects do play
an important role and have to be included for a proper
treatment. When the set of compounds is increased by
the inclusion of the four analogues of clofibric acid the cor-
relation decreases slightly to 0.748, still remaining quite
large (Fig. 8). However, a close inspection of data for the
latter correlation shows that the value of DERS for com-
We attempted to find an approximate semiquantitative cor-
relation between the calculated energy differences between
the ap(H,H) and sp(H,H) conformations of Tables 3 and
4 and the experimental DdRS values of Tables 1 and 2,
respectively.
This has been done by evaluating the difference in the pop-
ulation between the ap(H,H) and sp(H,H) conformers for
all the compounds by using a Boltzmann distribution; the
fraction of the population of the conformation i is thus cal-
culated by
1
0.8
0.6
i
ni
N
eꢁE =RT
ap=RT
RS
y = 4.892x + 0.175
R2 = 0.846
xi ¼
¼
ð1Þ
p
s
-
eꢁE
þ eꢁE
sp=RT
ap
0.4
0.2
0
X
where i = ap(H,H), sp(H,H) and T is the temperature cor-
responding to the experimental conditions and R is the
Boltzmann constant. As we have considered the lowest en-
ergy ap(H,H) and sp(H,H) conformers only, this is a rather
crude estimate of xi, however, one has to bear in mind that
we do not expect this correlation to be fully quantitative
since, as underlined above, the difference between the pop-
ulation of the two conformers is only one of a number of
0
0.05
0.1
0.15
0.2
ΔδRS
Figure 7. Linear correlation between the experimental DdRS data (Table 1)
RS
ap–sp
and the calculated X
values for the ethyl mandelate esters of the
training set of a-carboxylic acids in Figure 1.
factors contributing to the numerical value of DdRS
.
The excess of population for the ap(H,H) conformer is gi-
ven by
1
0.8
0.6
Xap–sp ¼ xap ꢁ xsp
ð2Þ
(an excess in the population of the sp(H,H) conformer will
thus lead to a negative value of Xap–sp). Since DdRS depends
on the excess of population of both the (R)- and (S)-ethyl
mandelate derivatives, the corresponding Xap–sp values
have been algebraically summed
RS
y = 5.135x + 0.149
R2 = 0.748
p
s
-
ap
0.4
0.2
0
X
XRap–sp þ XSap–sp
0
0.05
0.1
0.15
0.2
RS
ap–sp
X
¼
ð3Þ
ΔδRS
2
Figure 8. Linear correlation between the experimental DdRS data (Tables 1
and the sum has been divided by 2 for normalizing pur-
RS
RS
poses: a value of X
¼ 1 corresponds to the presence
and 2) and the calculated X
values for the ethyl mandelate esters of the
ap–sp
ap–sp
of the ap(H,H) conformer only for both diastereomers; a
training set of a-carboxylic acids in Figure 1 with the addition of the
analogues of clofibric acid synthesized in this work (Fig. 2).
RS
value of X
¼ 0 means to have equal populations of
ap–sp

A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
995
RS
Table 5. Chemical shifts for the substituents to the Ca of the (S,R) and
(S,S) diastereomeric esters of acid 1c, recorded at different temperatures
We can again use the values of X
obtained with Eqs.
ap–sp
1–3 and correlate them with the experimental DdRS of
Temperature Signal
d (ppm) (S,R) d (ppm) (S,S) DdRS
Table 5. In this case, we should expect the correlation
RS
(°C)
between the calculated X
values and the DdRS of the
ap–sp
same groups at different temperatures to be complete
within the experimental and computational accuracy: the
increase of DdRS in this case is only due to the different
populations of the considered conformations and no other
effects come into play.
20
0
ꢁ10
ꢁ40
20
CH3 (5) 1.240
1.192
1.184
1.179
1.162
1.566
1.561
1.560
1.553
0.990
0.987
0.987
0.982
+0.048
CH3 (5) 1.237
CH3 (5) 1.235
CH3 (5) 1.231
CH2 (3) 1.515
CH2 (3) 1.506
CH2 (3) 1.501
CH2 (3) 1.487
CH3 (4) 0.922
CH3 (4) 0.913
CH3 (4) 0.908
CH3 (4) 0.895
+0.053
+0.056
+0.069
ꢁ0.051
ꢁ0.055
ꢁ0.059
ꢁ0.066
ꢁ0.068
ꢁ0.074
ꢁ0.079
ꢁ0.087
0
ꢁ10
ꢁ40
20
Figure 9 illustrates such a correlation for the three CHn
groups of the esters of compound 1c, which is very good
in each case (0.98 < R2 < 1). Although the number of
points to be correlated is limited, R2 values so close to 1
indicate that the temperature dependence of the experimen-
tal DdRS only depends on the different conformational pop-
ulations of the diastereomers, so that such correlations
could be used to quantitatively predict DdRS values upon
the temperature changes.
0
ꢁ10
ꢁ40
The absolute value of DdRS becomes larger as the temperature increases.
pound 2h is responsible for this discrepancy: the elimina-
tion of the corresponding point leads to a value of
R2 = 0.827, very similar to that obtained with the training
set only.
The existence of a qualitative relationship between DdRS
and DERS values and that, after some manipulations, a
quantitative relationship could as well be established,
underlines that the NMR strategy for the assignment of
the absolute configurations of a-chiral carboxylic acids is
founded on arguments based on a well defined conforma-
tional preference, as proposed.5 Furthermore, our calcula-
tions suggest that in order for this strategy to be applied to
a-chiral carboxylic acids, the related conformational con-
straints are not so strict: it is sufficient that the two hydro-
gens (the alcoholic C–H and the acidic Cꢀa–H) in the esters
lie approximately in an antiperiplanar position. Only in the
presence of bulky and/or very electronegative groups di-
rectly linked to the Ca can the most stable conformation
vary so as to determine a different aromatic shielding.
To further investigate the relationship between the differ-
ences in the calculated populations of the ap(H,H) and
sp(H,H) conformers in the chiral esters and the experimen-
tal values of DdRS we also performed low temperature
NMR experiments. Indeed, as highlighted in the litera-
ture,5 one should expect larger absolute values for DdRS
as the temperature decreases, due to the increased popula-
tion of the lowest energy preferred conformation. The
experimental results on the diastereomeric esters (S,R)-
and (S,S)-2c reported in Table 5 are fully coherent with this
picture.
0.44
0.44
a
b
0.43
0.42
0.41
0.4
0.43
0.42
p
0.41
s
-
p
-
y = 2.917x + 0.234
R2 = 0.981
y = 3.103x + 0.159
R2 = 0.993
ap
ap
0.4
0.39
0.38
0.37
0.36
RS
RS
0.39
0.38
0.37
0.36
X
X
0.045
0.055
0.065
0.075
0.065
0.075
0.085
0.095
ΔδRS
ΔδRS
0.44
c
0.43
0.42
0.41
0.4
p
s
-
y = 3.900x + 0.172
R2 = 0.993
ap
RS
0.39
0.38
0.37
0.36
X
0.05
0.055
0.06
0.065
0.07
ΔδRS
Figure 9. Linear correlation between the experimental DdRS data (Table 5) for the substituents (CH3 (5) panel (a), CH2 (3) panel (b), CH3 (4) panel (c)) to
the Ca of the (S,R)- and (S,S)-esters of acid 1c obtained from different temperatures NMR spectra and the corresponding theoretical XRapS–sp values.

996
A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
3. Conclusions
4.1.2. General procedure for the preparation of esters (R)-
3g–j. A solution of diisopropylazodicarboxylate (DIAD,
607 mg, 3.0 mmol) in anhydrous tetrahydrofuran (15 mL)
was added dropwise to a solution of (S)-ethyl lactate
(260 mg, 2.2 mmol), the appropriate alcohol (2.0 mmol)
and triphenylphosphine (786 mg, 3.0 mmol) in anhydrous
tetrahydrofuran (50 mL). The reaction mixture was stirred
at room temperature overnight, under a nitrogen atmo-
sphere. THF was evaporated in vacuo, and the residue
was purified by chromatography on silica gel (cyclohex-
ane/ethyl acetate 95:5) to afford the desired (R)-3g–j esters
in 36–67% yields.
Herein, we have performed the synthesis of some chiral
analogues of clofibric acid by a simple reaction scheme,
in good yields and with excellent enantiomeric excess,
and assigned their absolute configurations by a well-known
NMR-based methodology, consisting of the derivatization
of a-chiral carboxylic acids to couples of diastereomeric
esters through a suitable chiral alcohol. The comparison
between the diastereomers chemical shifts of the substi-
tuents to the Cꢀa of the carboxylic acid should lead to the
determination of the acid absolute configuration.
For that purpose we chose ethyl mandelate as an auxiliary
reagent since, besides having the minimal requirements to
provide a good aromatic shielding, it has the appealing
benefits of being little expensive and easy to obtain in both
enantiomeric forms. Although a similar reagent, methyl
mandelate, had already been employed, leading to fairly
good, although not excellent, results, this represents the
first attempt to use this chiral agent. For this reason, we
decided to thoroughly investigate the use of ethyl mandel-
ate by testing the procedure on a set of commercially avail-
able a-chiral carboxylic acids and by performing DFT
calculations on the preferred conformations of the diaste-
reomeric esters. The success of the NMR strategy is indeed
based on a well-defined conformational preference of the
diastereomers leading to a selective shielding of the substit-
uents to Cꢀa in turn.
4.1.3. General procedure for the preparation of acids (R)-1g–
j. 6 M HCl (20 mmol, 3.33 mL) was added to (R)-3g–j
(1 mmol), and the mixture was heated to 60 °C. When
the reaction was complete, the mixture was diluted and
extracted with diethyl ether (20 mL). The collected organic
layers were dried on sodium sulfate. After solvent evapora-
tion, crude materials were purified by crystallization to
obtain (R)-1g–j acids in good yields and excellent enantio-
meric excess.
4.1.3.1. (R)-2-(4-Chlorophenoxy)propanoic acid, (R)-
1g. White needles (crystallized from petroleum ether),
52% yield; mp 115–117 °C; IR (KBr) 1712 cmꢁ1
;
23
1
½aꢂD ¼ þ41:9 (c 1.2, CH3OH); H NMR (CD3OD) d 1.57
(d, 3H, J = 6.9 Hz, CH3), 4.78 (q, 1H, J = 6.9 Hz, CHCH3),
6.86 (d, 2H, J = 9.0 Hz, CH arom), 7.24 (d, 2H, J = 9.0 Hz,
CH arom); 13C NMR (CD3OD) d 17.6 (CH3), 72.4
(CHCH3), 116.4 (CH arom), 126.0 (C arom), 129.1 (CH
arom), 156.7 (C arom), 174.3 (C@O). Anal. Calcd for
C9H9ClO3: C, 53.88; H, 4.52. Found: C, 53.71; H, 4.53.
The experimental DdRS sign for both the training set and
the new chiral 2-aryloxypropanoic acids is in full agree-
ment with the stereochemistry of the substrates, although
with smaller absolute values than with ethyl 2-hydroxy-2-
(9-anthryl)acetate, thus showing that ethyl mandelate can
be successfully used, except in the case of bulky and/or very
electronegative substituents to Ca.
4.1.3.2. (R)-2-[(6-Chloroquinolin-2-yl)oxy]propanoic acid,
(R)-1h. White needles (crystallized from cyclohexane),
72% yield; mp 140–143 °C; IR (KBr) 1716 cmꢁ1;
23
1
½aꢂD ¼ þ110:7 (c 2.0, CHCl3); H NMR (CDCl3) d 1.69
(d, 3H, CH3CH), 5.51 (q, 1H, CH), 7.01 (d, 1H, CH arom),
7.50 (dd, 1H, CH arom), 7.68 (dd, 2H, CH arom), 7.95 (d,
1H, CH arom); 13C NMR (CDCl3) d 17.6 (CH3), 70.3
(CH), 113.8 (CH arom), 126.1 (C arom), 126.4 (CH arom),
128.9 (CH arom), 130.2 (C arom), 130.6 (CH arom), 138.7
(CH arom), 144.3 (C arom), 161.0 (C arom), 177.6 (C@O).
Anal. Calcd for C12H10ClNO3: C, 57.27; H, 4.01; N, 5.57.
Found: C, 57.15; H, 4.02; N, 5.58.
4. Experimental
4.1. Chemistry
Melting points were determined on a Buchi B-540 appara-
¨
tus and are uncorrected. Optical rotations were measured
on a Perkin Elmer model 241 polarimeter. Infrared spectra
were recorded on a FT-IR 1600 Perkin Elmer spectro-
meter. Microanalyses were carried out with an Eurovector
Euro EA 3000 model analyzer; the analytical results are
within 0.4% of the theoretical values. Commercial reagents
were used as received from Aldrich or Fluka. THF was dis-
tilled from sodium/benzophenone.
4.1.3.3. (R)-2-(1,3-Benzothiazol-2-yloxy)propanoic acid,
(R)-1i. White needles (crystallized from cyclohexane),
70% yield; mp 167–170 °C; IR (KBr) 1716 cmꢁ1
;
19
1
½aꢂD ¼ þ35:9 (c 2.0, CH3OH); H NMR (CD3OD) d 1.65
(d, 3H, CH3), 5.34 (q, 1H, CH), 7.18 (m, 2H, CH arom),
7.33 (dd, 1H, CH arom), 7.54 (dd, 1H, CH arom); 13C
NMR (CD3OD) d 13.3 (CH3), 51.4 (CH), 111.3 (CH
arom), 122.3 (C arom), 122.8 (CH arom), 123.3 (CH
arom), 126.4 (CH arom), 136.4 (C arom), 170.2 (C arom),
171.4 (C@O). Anal. Calcd for C10H9NO3S: C, 53.80; H,
4.06; N, 6.27. Found: C, 53.67; H, 4.05; N, 6.26.
4.1.1. General procedure for the preparation of diastereo-
meric esters (X,R)- and (X,S)-2a–f. A mixture of (R)- or
(S)-1a–f (1.0 mmol), (R)- or (S)-ethyl mandelate (180 mg,
1.0 mmol), DCC (206 mg, 1.0 mmol) and DMAP (1.2 mg,
0.1 mmol) was dissolved in dry dichloromethane and stir-
red at room temperature overnight. The reaction mixture
was filtered off, and after evaporation of solvent, the crude
material was purified by silica gel (cyclohexane/ethyl ace-
tate 9:1).
4.1.3.4. (R)-2-(1-Naphthyloxy)propanoic acid, (R)-
1j. White needles (crystallized from cyclohexane), 36%
22
yield; mp 126–127 °C; IR (KBr) 1706 cmꢁ1; ½aꢂD ¼ ꢁ61:3

A. Ammazzalorso et al. / Tetrahedron: Asymmetry 19 (2008) 989–997
997
(c 1.8, CHCl3); 1H NMR (CDCl3) d 1.80 (d, 3H,
Acknowledgements
J = 6.9 Hz, CH3), 4.98 (q, 1H, J = 6.9 Hz, CH), 6.73 (d,
1H, CH arom), 7.31–7.35 (m, 1H, CH arom), 7.47–7.51
(m, 3H, CH arom), 7.79–7.82 (m, 1H, CH arom), 8.30–
8.34 (m, 1H, CH arom); 13C NMR (CDCl3) d 18.7
(CH3), 72.5 (CH), 105.8, 121.7, 122.2, 125.7, 125.8 (CH
arom), 125.8 (C arom), 126.9, 127.7 (CH arom), 134.8,
153.2 (C arom), 177.8 (C@O). Anal. Calcd for C13H12O3:
C, 72.21; H, 5.59. Found: C, 72.34; H, 5.58.
The Italian Ministry for the University and the Research is
acknowledged for the financial support (Contracts
2006038520 and 2005033023). The authors gratefully
acknowledge Mrs. Maria Luisa Tricca for her precious
technical help.
References
´
1. Seco, J. M.; Ortiz, A.; Quinoa, E.; Riguera, R. Chem. Rev.
˜
4.1.4. General procedure for the preparation of diastereo-
meric esters (R,R)- and (R,S)-2g–i. Diastereomeric esters
(R,R)- and (R,S)-2g–i were readily obtained by Mitsunobu
coupling of (R)-1g–j with (R)- or (S)-ethyl mandelate, as
previously described for compounds 3g–j.
2004, 104, 17–118.
2. Wenzel, T. J.; Wilcox, J. D. Chirality 2003, 15, 256–270.
3. Seco, J. M.; Quinoa, E.; Riguera, R. Tetrahedron: Asymmetry
´
˜
2001, 12, 2915–2925.
´
4. Porto, S.; Seco, J. M.; Ortiz, A.; Quinoa, E.; Riguera, R. Org.
˜
Lett. 2007, 9, 5015–5018.
´
5. Ferreiro, M. J.; Latypov, Sh. K.; Quinoa, E.; Riguera, R. J.
˜
4.2. NMR spectroscopy
Org. Chem. 2000, 65, 2658–2666.
6. Iwamoto, H.; Kobayashi, Y.; Kawatani, T.; Suzuki, M.;
Fukazawa, Y. Tetrahedron Lett. 2006, 47, 1519–1523.
1H and 13C NMR spectra were recorded on a Varian
300 MHz instrument, using tetramethylsilane (TMS) as
an internal standard. All the compounds were dissolved
in CDCl3, (5 mg in 0.7 mL). For experiments at low tem-
perature, samples were allowed to equilibrate for 15 min
at each temperature before recording the spectra.
´
7. Latypov, Sh. K.; Seco, J. M.; Quinoa, E.; Riguera, R. J. Org.
˜
Chem. 1995, 60, 504–515.
´
8. Seco, J. M.; Latypov, Sh. K.; Quinoa, E.; Riguera, R.
˜
Tetrahedron 1997, 53, 8541–8564.
´
9. Seco, J. M.; Quinoa, E.; Riguera, R. J. Org. Chem. 1997, 6,
˜
7569–7574.
10. Trost, B. M.; Bunt, R. C.; Pulley, S. R. J. Org. Chem. 1994,
59, 4202–4205.
4.3. Computational methods
´
11. Latypov, Sh. K.; Seco, J. M.; Quinoa, E.; Riguera, R. J. Org.
˜
Chem. 1995, 60, 1538–1545.
´
All the calculations were performed with the Schro¨dinger
suite of programs. The conformational searches were car-
ried out on all the diastereomeric derivatives through MAC-
ROMODEL 9.0,25 using MMFF94s as force field and
Conjugate Gradient with Polak-Ribiere first derivatives
method as minimization algorithm. 3000 conformations
for each diastereomer were minimized, only those within
30 kJ/mol from the global minimum were recorded. Once
a conformational search was completed, the recorded con-
formations were subjected to a cluster analysis using
XCLUSTER.26 The conformations were clustered according
to their root mean square distances with respect to the
three dihedral angles determining the mutual positions of
12. Lopez, B.; Quinoa, E.; Riguera, R. J. Am. Chem. Soc. 1999,
˜
121, 9724–9725.
13. Yabuuchi, T.; Kusumi, T. J. Org. Chem. 2000, 65, 397–404.
14. Ichikawa, A.; Ono, H.; Hiradate, S.; Watanabe, M.; Harada,
N. Tetrahedron: Asymmetry 2002, 13, 1167–1172.
15. Tyrrell, E.; Tsang, M. W. H.; Skinner, G. A.; Fawcett, J.
Tetrahedron 1996, 52, 9841–9852.
16. Staels, B.; Dallongeville, J.; Auwerx, J.; Schoonjans, K.;
Leitersdorf, E.; Fruchart, J.-C. Circulation 1998, 98, 2088–
2093.
17. Chapman, M. J. Atherosclerosis 2003, 171, 1–13.
18. Brown, P. J.; Winegar, D. A.; Plunket, K. D.; Moore, L. B.;
Lewis, M. C.; Wilson, J. G.; Sundseth, S. S.; Koble, C. S.;
Wu, Z.; Chapman, J. M.; Lehmann, J. M.; Kliewer, S. A.;
Willson, T. M. J. Med. Chem. 1999, 42, 3785–3788.
19. Shi, G. Q.; Dropinski, J. F.; Zhang, Y.; Santini, C.; Sahoo, S.
P.Berger, J. P.;MacNaul, K. L.;Zhou, G.;Agrawal, A.;Alvaro,
R.; Cai, T.; Hernandez, M.; Wright, S. D.; Moller, D. E.; Heck,
J. V.; Meinke, P. T. J. Med. Chem. 2005, 48, 5589–5599.
20. Sauerberg, P.; Mogensen, J. P.; Jeppesen, L.; Bury, P. S.;
Fleckner, J.; Olsen, G. S.; Jeppesen, C. B.; Wulff, E. M.;
Pihera, P.; Havranek, M.; Polivka, Z.; Pettersson, I. Bioorg.
Med. Chem. Lett. 2007, 17, 3198–3202.
the phenyl ring and the L1/L2 substituents to the Cꢀ of
a
the carboxylic acid (Fig. 5). For each diastereomer, 10 clus-
ters were obtained, and their leading members, that is, the
minimum energy structure within each cluster, fully opti-
mized by means of density functional theory calculations
both in gas phase and in solution of CHCl3 using JAGUAR
6.0.27 Such calculations were performed using the B3LYP
exchange correlation functional, which is known to per-
form particularly well for organic molecules, and the
6-31G* basis set, except for compounds derived from acid
1a, for which an LACVP* basis set was employed, consist-
ing of the Los Alamos effective core potential and a double-
zeta valence basis set including the outermost core orbitals
for Br and the all electron 6-31G* basis set for all the
remaining atoms. Calculations in solutions were carried
out through the Poisson-Boltzmann continuum model
employing the following values for the dielectric constant
and probe radius to model CHCl3, e = 4.806 and
21. Yamazaki, Y.; Abe, K.; Toma, T.; Nishikawa, M.; Ozawa,
H.; Okuda, A.; Araki, T.; Oda, S.; Inoue, K.; Shibuya, K.;
Staels, B.; Fruchart, J.-C. Bioorg. Med. Chem. Lett. 2007, 17,
4689–4693.
22. Mitsunobu, O. Synthesis 1981, 1, 1–28.
23. Klyne, W.; Prelog, V. Experientia 1960, 16, 521–523.
24. Kane, S.; Hersh, W. H. J. Chem. Educ. 2000, 77, 1366.
25. MACROMODEL, Version 9.0; Schro¨dinger, LLC: New York,
NY, 2005.
26. Macromodel XCLUSTER, Version 9.0; Schro¨dinger, LLC: New
York, NY, 2005.
˚
27. JAGUAR, Version 6.0; Schro¨dinger, LLC: New York, NY, 2005.
r = 2.514 A.