On the influence of unsaturation on the macrolactonization of hydroxy fatty acids

Article abstract of DOI:10.1002/poc.1548

Regioselectivity and yields of Yamaguchi cyclization of (15S, 16S)-15, 16-dihydroxy-otadecyl-(6Z, 9Z, 12Z)-trienoic acid and its saturated homologue have been compared at different temperatures and concentrations. For the unsaturated species the regiosele

Full text of DOI:10.1002/poc.1548

Research Article
Received: 31 October 2008,
Revised: 21 January 2009,
Accepted: 19 February 2009,
Published online in Wiley InterScience: 27 April 2009
(www.interscience.wiley.com) DOI 10.1002/poc.1548
On the influence of unsaturation on the
macrolactonization of hydroxy fatty acids
Tonino Carusoa, Carmen Donnamariaa, Antonietta Artilloa,
a
a
a
*
*
Andrea Peluso , Aldo Spinella and Guglielmo Monaco
Regioselectivity and yields of Yamaguchi cyclization of (15S,16S)-15,16-dihydroxy-otadecyl-(6Z,9Z,12Z)-trienoic acid
and its saturated homologue have been compared at different temperatures and concentrations. For the unsaturated
species the regioselectivity of cyclization depends on temperature to such an extent that the preferred regioisomer
changes as the temperature changes, whereas for the saturated homologue the even-membered ring turns out to be
always the preferred one. Molecular mechanics is used to interpret these findings. For both dihydroxyacids, there is
experimental evidence that the yields of the two monocyclic regioisomers are thermodynamically controlled, unlike
those of the analogue monohydroxy lactones. The higher cyclization yield of the unsaturated monohydroxyacid is
interpreted in terms of conformational pre-organization. Copyright ß 2009 John Wiley & Sons, Ltd.
Keywords: Yamaguchi macrocyclization; conformational analysis; kinetics; thermodynamic control; regioselectivity;
conformational pre-organization; vic-dihydroxyacid; coarse graining
INTRODUCTION
RESULTS AND DISCUSSION
Regioselectivity
Macrocyclization is a fundamental process in organic synthesis.
Since the pioneering investigation of Jacobson and Stock-
mayer,[1] which opened the way towards understanding the
distribution of open chain oligomers with respect to cyclic
products, macrocyclization has been thoroughly investigated
both theoretically and experimentally.[2,3,4,5] Much of the work
has been concerned with open precursors bearing only two
functional groups for ring closure. The case of precursors with
three or more functional groups presents the further challenge of
understanding the regiochemistry of the ring closure reactions.
As a contribution along this line, in this paper we report the
results of experimental and theoretical studies on the regio-
chemistry and yields of macrolactonizations of vicinal dihydroxy
saturated and unsaturated fatty acids.
Yamaguchi macrocyclization of 4 has been carried out at 298 and
384 K (the solvent boiling point), and at different concentrations
of the open chain precursors, 1 and 10 mM. Reaction yields and
concentration ratios of 5 and 6 regioisomers have been
1
determined by H NMR.
As in the case of macrolactonization of 1,[6] we checked that by
putting in the same reaction conditions only one of the two
saturated lactones 5 or 6, the same concentration ratio [6]/[5] is
obtained as by Yamaguchi cyclization of 4. Thus, both the [3]/[2]
and the [6]/[5] ratios are the equilibrium constants of the ring
expansion reactions: 2 fi 3 and 5 fi 6, here denoted by K16!17
.
The equilibrium constants for the two ring expansions are
reported in Table 1. For the saturated precursor, regioselectivity is
always in favor of the smaller ring. The equilibrium constant
slightly increases as the temperature increases, but the
concentration ratio remains lower than one, in contrast with
the unsaturated precursor, which exhibits a stronger temperature
dependence of K16!17, leading to change the predominant
product as the temperature increases.[6] From the temperature
dependence of K16!17’s, the enthalpy and entropy changes for
the two ring expansion processes have been determined; the
results are reported in Table 1.
In a previous letter we have reported that the product
distribution of the Yamaguchi macrocyclization of
(15S,16S)-15,16-dihydroxy-octadecyl-(6Z,9Z,12Z)-trienoic
acid
(1), leading to macrolides 2 and 3, Scheme 1, significantly
depends on temperature.[6] At room temperature macrolide 2 is
the favored product, but at higher temperature macrolide 3
becomes the predominant one. Here, we extend those studies to
the macrolactonization of the saturated analogue of 1,
compound 4, the precursor of macrolides 5 and 6.
It has already been reported that the change from a saturated
to an unsaturated chain can considerably speed up the
macrolactonization of monohydroxyacids,[7] but nothing is
known about the possible effect of unsaturations on the
regioselectivity and on the yields of dihydroxyacid macrocycliza-
tion.
*
Correspondence to: A. Spinella or G. Monaco, Dipartimento di Chimica,
`
Universita di Salerno, via ponte Don Melillo Fisciano, 84084 (SA), Italy.
E-mails: spinella@unisa.it; gmonaco@unisa.it
Macrolactonization of two monohydroxyacid analogues has
also been performed for a better understanding of the effects
due to the presence of the vicinal hydroxyl groups.
a
T. Caruso, C. Donnamaria, A. Artillo, A. Peluso, A. Spinella, G. Monaco
`
Dipartimento di Chimica, Universita di Salerno, via ponte Don Melillo
Fisciano, 84084 (SA), Italy
J. Phys. Org. Chem. 2009, 22 978–985
Copyright ß 2009 John Wiley & Sons, Ltd.

INFLUENCE OF UNSATURATION ON MACROLACTONIZATION
Scheme 1.
Inspection of Table 1 shows that enthalpy and entropy changes
of the two ring expansion reactions are significantly different
from each other; although enthalpy and entropy changes are
positive for both systems, for the saturated ring they are both
significantly lower than for the unsaturated one.
Understanding that behavior at microscopic level is quite a
difficult task because of the very large number of conformers
which are expected to significantly contribute to the thermo-
dynamics of ring expansions. We have therefore considered two
different approaches. In the first one, the separated fragments
approach, the conformational partition function of each of the
macrocycles of Scheme 1 is obtained from calculations on two
parent fragments. In the second approach, the thermodynamic
properties are determined from the partition functions of the
cyclic regioisomers.
The separated fragments approach consists in (i) dividing
the macrocycle in two fragments, one of which, the
(S,S)-4-acetoxy-3-hexanol (7), is common to all species and
the other one is either 8 or 9 (Scheme 2); (ii) finding all the
conformations of the fragments and grouping them in N different
intervals of either r0,5 or r0,6 distances; (iii) computing the
conformational partition function of a cycle by a piecewise
factorization:
fragment 7 and are not expected to be significantly affected
by the presence of the other fragment. A further simplification of
Equation 1 considers the long fragments to be arbitrarily flexible
in the range of interest for cycle formation (zD,i independent of
the i-th interval). In this case the partition functions of the small
and large cycles are different only because the conformational
partition function of fragment 7 is differently coarse grained.
We have then computed the values of the z7,i’s and their
@ ln z
corresponding energies E7;i ¼ ꢀ
7;i ; with b ¼ 1/kT, considering
@b
either r0,5 or r0,6 as the coarse graining variable. The width of the
˚
intervals has been fixed to 0.5 A. The results are given in the bar
plots of Figure 1.
Figure 1 shows that the r0,5 values not only have smaller
average than the r0,6 values, but they also occur on a narrower
range. This suggests a higher entropy for the large ring
and indeed, using the values plotted in Figure 1, we get
DS816!17 ¼ 1.0 cal molꢀ1 Kꢀ1 and DE816!17 ¼ 0.93 kcal molꢀ1, in
remarkable semi-quantitative agreement with the experimental
values.
Obviously, the different experimental values of DH816!17 and
DS816!17 for saturated and unsaturated chains cannot be
interpreted in terms of this simple model which only considers
the role of fragment 7. It is formally possible to consider Equation
1 without the assumption of perfectly flexible chains for 8 and 9,
although in this case the matching criterion should be
investigated in detail. We did not consider this aspect here,
since the separated fragments method was only introduced to
obtain a qualitative insight into the origin of enthalpy and
entropy increases in the ring expansion processes of both the
saturated and unsaturated macrocycles.
N
X
zcycle
¼
z7;izD;i
;
D ¼ 8; 9; cycle ¼ 2; 3; 5; 6; (1)
i¼1
where z7,i (zD,i) is the partition function of those conformers of
species 7 (D) which fall in the i-th interval of r0,5 or r0,6. Equation 1
holds only if, for each conformation of the cycle, similar
conformations of the parent chains can be identified—which
is the usual assumption for strainless rings—and if the
interaction energy between the two fragments is the same for
all conformers and can therefore be neglected. The latter
condition is indeed our case, since the most important
electrostatic and H-bond contributions are all localized in
As a second approach toward a better understanding of the
thermodynamics of ring expansions, we have performed an
extensive Monte Carlo search of the conformational spaces of the
four macrocycles of Scheme 1. The number of conformers of
macrocycles 2, 3 and 5, 6 within a threshold of 5 kcal molꢀ1
is in qualitative agreement with the positive entropy of ring
Table 1. Equilibrium constants, enthalpies, and entropies of ring expansion reactions for the unsaturated[6] and saturated cycles of
Scheme 1
K16!17 (298 K)
K16!17 (384 K)
DH816!17 (kcal molꢀ1 Kꢀ1
)
DS816!17 (cal molꢀ1 Kꢀ1
)
Unsaturated
Saturated
0.67 ꢁ 0.04
0.59 ꢁ 0.05
1.48 ꢁ 0.05
0.72 ꢁ 0.03
2.2
0.53
6.7
0.72
J. Phys. Org. Chem. 2009, 22 978–985
Copyright ß 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/poc

T. CARUSO ET AL.
Scheme 2.
expansion: we have 635 and 1367 conformers for 2 and 3, 1389
and 2018 conformers for 5 and 6. The three conformers of lower
energy are shown in Figure 2 for the four rings under study. The
conformers have been classified into three classes. Those
conformers in which the hydroxyl group forms hydrogen
bonding with either the carbonyl or the esteric oxygen are
included in class C1 and C2, respectively, whereas those
conformers without any intramolecular hydrogen bond are
included in class C3. Conformers are then denoted as I-CJ-K,
where I is the number of the species, CJ is the class and K is an
index denoting the relative stability of the conformer within a
given species. Conformer energies (kcal molꢀ1), relative to the
lowest energy conformers of the smaller rings, are also reported
in Figure 2 in parentheses.
The energy differences between conformers of isomeric rings
with the same hydrogen pattern (2-C1-1, 3-C1-1 and 5-C1-1,
6-C1-1) stem from cycle conformations. In the case of the
saturated rings, the even/odd conformational effect, pointed
out by Dale for saturated cycles,[8] holds in the present case.
According to Dale, the lowest energy conformations of medium
and large-sized even cycles have a quadrangular aspect (with 8
gauche conformations) while conformations of odd cycles
either have a pentagonal shape, with two more gauche
conformations, or a triangular one, with highly unfavorable
steric interactions near the angles. The minimum energy
conformations of 5 and 6 do have these geometric features. As
concerns the unsaturated cycles, the presence of conformers of
different classes among the low energy conformers of 2 and 3
allows to ascribe part of the positive enthalpy of reaction to the
absence of the stronger intramolecular hydrogen bonding
between the hydroxyl and the carbonyl groups in some low
energy conformers of 3.[6]
All the lowest energy conformers belong to the C1 class, but for
3, for which there are two quasi degenerate conformers of class
C3. Moreover 3 has a conformer of class C2, whose energy is
predicted to be only 0.59 kcal molꢀ1 higher than that of 3-C1-1.
P
Figure 1. Bar plots of probabilities ðpi ¼ z7;i
=
end-to-end distance. Energies have been computed by MMFF. The temperature has been assumed to be 298 K
z7;iÞ and energies of conformers of 7 coarse grained by (a) the r0,5 end-to-end distance, (b) the r0,6
i
www.interscience.wiley.com/journal/poc
Copyright ß 2009 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2009, 22 978–985

INFLUENCE OF UNSATURATION ON MACROLACTONIZATION
Figure 2. Lowest energy conformers of the rings found by MMFF Monte Carlo searches. Conformers are designed as I-CJ-K, where I is the number of the
species, CJ is the class (the kind of hydrogen bonding as explained in the text) and K is the number of the conformer taken from a list sorted by increasing
energy. The numbers in parentheses are the MMFF energies in kcal molꢀ1 relative to the lowest energy conformer of the small rings
The energy difference between the minimum energy
conformers (DE8T¼0 K) is close to the experimental enthalpy
difference. Encouraged by this close agreement, we have
computed enthalpies and entropies of ring expansions, including
all the conformers, by either neglecting vibrational effects or
modelling them in terms of harmonic oscillators. While for the
unsaturated species the inclusion of all conformations and
that in the reaction mixture there are acid and basic molecules,
whose effects have not been taken into account. Indeed, as we
will consider shortly, the trichlorobenzoic acid (TCB) is expected
to be H-bonded in some amount to the cycles.
Yields
vibrations produces
a slight better agreement with the
experimental quantities, this is not the case for the saturated
chain. Indeed it has been reported that the harmonic model is
unsuitable in saturated rings.[9] In any case, the computed
enthalpies of reaction differ from the experimental ones by less
than 1 kcal molꢀ1, while reaction entropies are underestimated
by a factor of 0.3–0.5 (Table 2). The overall agreement with
experimental results is satisfactory; especially if one considers
At room temperature, the cyclization yields of 1 and 4 are very
similar to each other, 56 and 59, respectively, whereas in toluene
at reflux (384 K) our results are in line with what Corey found on
different substrates and with a different reaction procedure[7]: the
yield of the unsaturated species is significantly higher than the
yield of the saturated homologues, 80 vs 62 for 1 and 4,
respectively.
Table 2. Computed enthalpies and entropies of ring expansion of 2 and 5. Computations have been performed either considering
the lowest energy conformers only (column DE8T¼0 K) or all the Boltzmann averaged conformers (columns DH8conf and DS8conf) or all
the Boltzmann averaged conformers together with the vibrational contribution (columns DH8 and DS8)
DE8T¼0 K
DH8conf
DH8
DS8conf
DS8
Reaction
(kcal molꢀ1 Kꢀ1
)
(kcal molꢀ1 Kꢀ1
)
(kcal molꢀ1 Kꢀ1
)
(cal molꢀ1 Kꢀ1
)
(cal molꢀ1 Kꢀ1
)
2 ! 3
5 ! 6
2.81
0.35
2.88
0.08
2.44
ꢀ0.23
2.35
0.37
2.76
0.33
J. Phys. Org. Chem. 2009, 22 978–985
Copyright ß 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/poc

T. CARUSO ET AL.
equilibrium constants of cyclization Kintra,i are to be introduced.
This is the case of 4, especially in the run at high temperature and
concentration, but certainly it is not the case of 1, whose reaction
mixture had in the apolar fraction the monocycle only. We have
thus considered for species 1 a simplification of Scheme 3, where
all Kintra,1 are set to zero, but
Table 3. Cyclization yields obtained at different temperature
and concentration of the precursors
Linear
Species
therm.
model
kinetic
model
Conc. mM)
T (K)
Yield
½C1ꢂ
1
1
1
1
1
1
10
12
1
1
1
298
343
363
384
298
384
384
384
56
63
72
80
49
71
71
53
54
68
72
76
49
71
—
—
68
84
88
90
23
54
—
—
Kintra;1
¼
¼ EM Kinter
;
½M1ꢂ
where the effective molarity EM, a compact indicator of the ease
of cyclization compared to oligomerization, can be further
analyzed in terms of an enthalpic and an entropic contri-
bution:[11]
1
10
10
1
1
EM
M0
¼ EMHEMS
ꢀ
ꢁ
ꢀ
ꢁ
DHintra ꢀ DHinter
DSintra ꢀ DSinter
¼ exp ꢀ
exp
:
RT
R
The effect of a 10-fold increase of the precursor concentration
has also been considered. A significant concentration depen-
dence of the cyclization yields is observed only for the saturated
species at higher temperature (30 at T ¼ 384 vs 62 at T ¼ 298),
while for the unsaturated species cyclization yields are still
remarkably high at higher concentration (Table 3). For
unsaturated species, the retention of high yields at high
temperature and at 10 mM precursor concentration is of
considerable practical relevance, because it indicates that the
high dilution condition generally adopted for macrocyclization
can be weakened.
The fact that the regiochemistry is determined by thermo-
dynamics suggests that the yields of cyclization of 1 and 4 are
thermodynamically controlled too. Ercolani et al. have thoroughly
investigated the equilibrium for a system of open chain precursor
(M1), open chain oligomers (M2-Mn) and cyclic species of different
ring sizes (C1-Cn), see Scheme 3, considering that the equilibrium
constant for lengthening the linear species Mi,
The equilibrium concentration of cyclic species can then be
determined for different values of the reaction thermodynamics
parameters.[10]
We
obtained
,
the
following
results:
EMS ¼
DHinter ¼ 2.7 kcal molꢀ1 Kꢀ1
DSinter ¼ 15 cal molꢀ1 Kꢀ1
,
0.049. The inclusion of an EMH different from one did not allow
any relevant improvement of the fit. This indicates an almost
strainless cyclization, in accord with the assumption made in the
separated fragments method. The values of cyclization yield
computed with these parameters compare well with the
experimental ones (Table 3, therm. model column, estimated
standard deviation s ¼ 2.7). According to the optimized thermo-
dynamic parameters the yield of linear dimers exceeds unity only
for the two runs at higher concentration, while the yield of trimers
is always smaller than 1. Indeed a truncation of Scheme 3 which
considers species up to dimers gives substantially unchanged
results (still s ¼ 2.7). On the other hand, the simple kinetic model
of Stoll,[2,12] which considers that monocycles and dimers are
formed irreversibly, gives an unacceptably worse fit (s ¼ 15) with
unlikely values of the optimized kinetic parameters (in the
notation of Reference [11] we find EMS ¼ kintra/kinter ¼ 2.7 which is
at least an order of magnitude higher than what is expected[11]).
½Miꢂ
expðꢀDHinter=RTÞ expðDSinter=RÞ
Kinter
¼
¼
;
M0
½Miꢀ1ꢂ½M1ꢂ
where M0 ¼ 1 mol Lꢀ1, is equal for all linear species.[10]
The above scheme should be extended with care to the case of
dihydroxyacids, where in correspondence to a single species
Mnþ1 or Cn formed from monohydroxyacids one has up to 2n
different species because each esteric bond can be realized in
two different ways. Moreover the reaction of a cycle and an open
species should also be considered. The collected data are too
scarce to investigate any extension of Scheme 3 and even the
application the above scheme is strongly limited if many
Reversibility vs. irreversibility
The regioselectivity of macrolactonization of 1 and 4 is
thermodynamically controlled and that is somewhat in contrast
with the fact that Yamaguchi lactonization is considered an
irreversible, kinetically controlled reaction.[13,14] This anomalous
behavior of 1 and 4 can be very probably attributed to the
presence of a vicinal dihydroxyl group.
Scheme 3.
www.interscience.wiley.com/journal/poc
Copyright ß 2009 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2009, 22 978–985

INFLUENCE OF UNSATURATION ON MACROLACTONIZATION
Scheme 4.
To better address this point we have considered Yamaguchi
cyclization of (15S)-15-hydroxy-hexadecyl-(4Z,7Z,10Z,13Z)-
cycles of different length with no need to consider the
conformations of the open chain.[2] Considering only monocycles
and dimers, one would simply have[2]
tetraenoic acid 10 and its saturated homologue 12
(Scheme 4). 11 and 13 do not react for 2 days in the same
conditions adopted for cyclization and thus the macrolactoniza-
tion of the monohydroxyacids 10 and 12 is an irreversible
process, as usually reported in the literature.
d½C1ꢂ
EM
¼
;
d½M2ꢂ ½M1ꢂ
where EM ¼ kintra/kinter, kintra and kinter are the kinetic constants for
cyclization and dimerization, respectively, and linear and cyclic
species are indicated as above.
The ring opening reaction leading to the thermodynamic
control of the macrolactonization regioselectivity of 1 and 4 is
therefore catalyzed by the hydroxyl group. Two mechanisms can
be envisaged: a direct translactonization or a full reversible
Yamaguchi macrolactonization. The latter one should be
catalyzed by the free OH group, because in toluene TCB is
strongly associated to N,N-dimethyl-4-aminopyridine DMAP (the
association constant of DMAP with perfluoroterbutanol in
toluene is 2.7ꢃ105),[15] making the concentration of unassociated
TCB very low for the reaction. The vicinal OH group can break the
strong TCBꢀDMAP complex because, together with the esteric
group, it can form a strong bidentate hydrogen bond with TCB.
Semiempirical computations, carried out using the MNDO
method with the AM1 parameterization,[16] support the above
hypothesis: for the two most stable conformers of the TCBꢀ7
H-bonded complex, the reaction DMAPꢀTCB þ 7 ! DMAP
þ TCBꢀ7 is predicted to be exothermic (DE ¼ ꢀ2.9 and
ꢀ1.8 kcal molꢀ1).
Thus, a higher cyclization yield can be interpreted in terms of a
higher effective molarity. Clearly, if conformational pre-
organization is relevant, its role must be somehow embedded
in the effective molarities. In order to unravel the relationship
between conformational pre-organization and effective molarity,
we adopt the following simple kinetic Scheme 5, where MO1 and
MC1 indicate open and coiled conformations of the monomer,
respectively.
Under the Curtin–Hammett conditions,[21] i.e., when the
interconversion of conformers is very fast compared to bond
rearrangement reactions (k2, k3 >> k4, k1), one has
d½C1ꢂ k4 KC
k4 KC
¼
ffi
;
(2)
½MO1 ꢂ
d½M2ꢂ k1
k1 ½M1ꢂ
where ½M1ꢂ ¼ ½M1Oꢂ þ ½M1Cꢂ and KC ¼ ½M1Cꢂ=½MO1 ꢂ is the equilibrium
constant for the coiling of the open conformers of the
monomer. The final approximation in Equation 2 (½MO1 ꢂ ꢄ ½M1ꢂ)
holds in the usual case that KC<<1. Thus, according to this simple
model
Although translactonization appears a simpler and more
appealing mechanism, the excellent fit of the reaction yields
obtained by employing the full reversible model suggests that
the latter mechanism can play an important role too.
k4
EM ¼ KC:
k1
Conformational pre-organization
The cyclization of 10 and 12 requires breaking and forming the
same kind of bonds and therefore the ratio k4/k1 is expected to be
Corey suggested that the higher rate of cyclization of unsaturated
secoacids should be attributed to a higher percentage of suitably
folded conformers, a suggestion which is also in line with the fact
that arachidonic peracids undergo selective internal epoxida-
tion.[17] In effect the conformation has been recognized as a main
factor determining regioselectivity.[4,5,18,19,20]
On the other hand, the established quantitative treatment of
kinetics of macrocyclization considers effective molarities for
Scheme 5.
J. Phys. Org. Chem. 2009, 22 978–985
Copyright ß 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/poc

T. CARUSO ET AL.
Figure 3. Boltzmann probabilities to have conformers of 8 and 9 within different intervals of the end-to-end distance r0,11. Energies have been
computed by MMFF. The temperature has been assumed to be 298 K
roughly equal for the two precursors. A higher effective molarity
should then be ascribed to a higher value of KC, which means a
higher percentage of suitably folded conformers.
This unprecedented finding has been interpreted in terms of the
ability of a vic-dialcohol to break the strong complex DMAPꢀTCB,
thus producing a significant amount of the active species needed
for lactonization.
The computation of this percentage is complicated by the fact
that the conformer distribution will be strongly influenced by the
TCB and DMAP present in solution. Nevertheless, in order to
appreciate the relevance of conformational pre-organization, we
have performed a conformational analysis for fragments 8 and 9.
The distribution functions of the end-to-end distances of the
saturated and unsaturated chains 8 and 9 are considerably
different (Figure 3); while the histogram of the former shows a
For irreversible macrocyclizations, it has been often reported
that conformational pre-organization is a relevant characteristic
of an effective reaction.[7,18,22] However, that molecular aspect is
not included in established kinetic modeling. We have proposed
a novel kinetic model, which explicitly considers conformational
pre-organization and is compatible with current models. In terms
of this model, the higher yield of the unsaturated cycle can be
interpreted by the conformational freedom of single bonds next
to double ones: although the unsaturated chain has less degrees
of torsional freedom, its torsional potentials have degenerate
minima and this leads on average to more coiled chains.
Finally, the regioselectivity of ring closure of dihydroxyacids
has markedly different outcomes for the two chains, due to the
different magnitude of enthalpy and entropy of ring expansion,
which are, however, positive in both cases. The sign of these
variations has been interpreted by a coarse grained approxi-
mation of the conformational partition function of fragment 7,
which is common to all cycles 2, 3, 5, and 6. The higher values of
the thermodynamic reaction parameters for the unsaturated ring
expansion have been recovered by MMFF computations on the
cycles.
˚
broad peak in the region 10–12 A, the latter one is very flat,
showing peaks of alternating height in the whole region between
˚
5 and 11 A. That is because the saturated chain can only coil
paying the energy penalty of adopting some gauche confor-
mations, while the unsaturated chain can coil without energy
penalties, because the minimum energy conformations for bonds
adjacent to double bonds are the anticlinal ꢁ120. Noticeably, for
fragment 8, the probability to have conformations with small
end-to-end distances is more than 10 times that of fragment 9. A
similar trend could be expected also for 10 and 12, when proper
account is taken for the acid and basic molecules in the reaction
medium. Therefore we conclude that the conformational
pre-organization is indeed well capable of explaining the
increased cyclization rate.
CONCLUSIONS
EXPERIMENTAL
Yamaguchi macrolactonizations of saturated and unsaturated di-
and monhydroxysecoacids 1, 4, 10, and 12 have been performed.
At 384 K the yields of unsaturated precursors are higher than the
corresponding ones for the saturated homologues. Notably, the
yield of macrocyclization of 1 remained high at 10 mM, in
contrast to 4.
NMR spectroscopy
NMR spectra were recorded at room temperature on a Brucker
DRX 400 MHz. Chemical shifts are reported in ppm: the CHCl3
signal (7.26 ppm per 1H) and the CDCl3 signal (77.0 ppm per 13C)
have been used as internal references. Ratios of regioisomers of
the cycles have been determined by integration of the H NMR
signals of the protons bounded to the carbinol.
1
In contrast with the monohydroxyacids, Yamaguchi lactoniza-
tion of dihydroxyacid 1 is well described as a reversible reaction.
www.interscience.wiley.com/journal/poc
Copyright ß 2009 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2009, 22 978–985

INFLUENCE OF UNSATURATION ON MACROLACTONIZATION
Syntheses
mations, respectively. In the case of 9 the multiplicity of the
conformations to be used in the computation of the probabilities
had to be recovered by molecular geometry. Conformational
searches on cycles 2, 3, 5, and 6 have been performed with 50
000-step Monte Carlo searches. In all cases only the conformers
within 5 kcal molꢀ1 above the minimum have been kept.
Semiempirical calculations have been performed at the AM1
level.
All reactions were performed under dry nitrogen. Products have
been purified by chromatography on silica gel (Merck
70–230mesh and Merck 230–400mesh).
The synthesis of 1, 2, and 3 has been performed according to
[23]
Reference
The synthesis of 10 has been performed according to
.
[24]
Reference
.
(15S,16S)-15,16-dihydroxy-octadecanoic acid (4). A mixture
of the methyl ester of 1 (130 mg, 0.4 mmol), BaSO4 (catalytic
amount) in methanol (2 ml) was stirred under a H2 atmosphere
for 2 h at 258C. The mixture was filtered through silica gel to
eliminate the catalyst, using methanol as eluent and concen-
trated to give the saturated ester of 1 (126 mg, 0.38 mmol, 95%).
The product was then dissolved in 1,2-dimethoxyethane (25 ml)
with lithium hydroxide (3 M, 67 ml) and stirred at 258C for 30 min.
The mixture was acidified with hydrochloric acid to pH 1 and the
water layer was extracted with CH2Cl2 (3 Â 25 ml). The combined
organic layers were dried (Na2SO4), filtered, and concentrated to
give dihydroxy acid 4 (119 mg, 0.38 mmol, 97%) which was used
without further purification.
Acknowledgements
Financial support from the MIUR is gratefully acknowledged.
REFERENCES
[1] H. J. Jacobson, W. H. Stockmayer, J. Chem. Phys. 1950, 18, 1600–1606.
[2] L. Mandolini, Adv. Phys. Org. Chem. 1986, 22, 1–111.
[3] K.-S. Yeung, I. Paterson, Angew. Chem. 2002, 114, 4826–4847.
[4] A. Parenty, X. Moreau, J.-M. Campagne, Chem. Rev. 2006, 106,
911–939.
[5] S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders, J. F. Stoddart,
Angew. Chem. Int. Ed. 2002, 41, 898–952.
[6] T. Caruso, G. Monaco, A. Peluso, A. Spinella, Org. Biomol. Chem. 2004,
2, 3425.
[7] E. J. Corey, S. Iguchi, J. O. Albright, B. De, Tetrahedron Lett. 1983, 24,
37–40.
[8] J. Dale, Acta Chem. Scand. 1973, 27, 1115.
[9] H. F. Dos Santos, W. R. Rocha, W. B. De Almeida, Chem. Phys. 2002, 280,
31–42.
[10] G. Ercolani, L. Mandolini, P. Mencarelli, S. Roelens, J. Am. Chem. Soc.
1993, 115, 3901–3908.
[11] C. Galli, L. Mandolini, Eur. J. Org. Chem. 2000, 3117–3125.
1H NMR d (CDCl3): 3.45-3.41 (1H, m); 3.39-3.32 (1H, m); 2.34 (2H,
t, J ¼ 7.6 Hz); 1.68-1.54 (6H, m); 1.51-1.42 (6H, m); 1.34-1.29 (12H,
m); 1.02 (3H, t, J ¼ 7.6 Hz).
13C NMR d (CDCl3): 177.2, 75.8, 74.00, 36.1, 34.00, 33.3, 30.4
(2CH2), 30.2, 29.3, 29.2, 29.1, 26.3, 25.6, 25.7, 19.8, 13.0, 10.7 (CH3).
15-hydroxy-esadecanoic acid (12). A mixture of the methyl
ester of 10 (130 mg, 0.4 mmol), BaSO4 (catalytic amount) in
methanol (2 ml) was stirred under a H2 atmosphere for 2 h at
258C. The mixture was filtered through silica gel to eliminate the
catalyst, using methanol as eluent and concentrated to give the
saturated ester of 10 (126 mg, 0.38 mmol, 95%). The product was
then dissolved in 1,2-dimethoxyethane (25 ml) with lithium
hydroxide (3 M, 67 ml) and stirred at 258C for 30 min. The mixture
was acidified with hydrochloric acid to pH 1 and the water layer
was extracted with CH2Cl2 (3 Â 25 ml). The combined organic
layers were dried (Na2SO4), filtered, and concentrated to give
hydroxy acid 12 (119 mg, 0.38 mmol, 97%) which was used
without further purification.
´
[12] M. Stoll, A. Rouve, G. Stoll-Comte, Helv. Chim. Acta 1934, 17, 1289.
[13] D. Seebach, U. Bra¨ndli, H.-M. Mu¨ller, M. Dobler, M. Egli, M. Przybylski,
K. Schneider, Helv. Chim. Acta 1989, 72, 1704–1717.
[14] S. J. Rowan, J. K. M. Sanders, J. Org. Chem. 1998, 63, 1536–1546.
´
[15] A. Demeter, V. Mile, T. Berces, J. Phys. Chem. A 2007, 111, 8942–8949.
[16] M. J. S. Dewwar, E. G. Zoebisch, E. F. Healy, J. J. P. Stewart, J. Am. Chem.
Soc. 1985, 107, 3908.
[17] E. J. Corey, H. Niwa, J. R. Falck, J. Am. Chem. Soc. 1979, 101, 1586.
[18] J. Mulzer, H. M. Kirstein, J. Buschmann, C. Lehmann, P. Luger, J. Am.
Chem. Soc. 1991, 113, 910.
[19] J. D. White, P. R. Blakemore, C. C. Browder, J. Hong, C. M. Lincoln, P. A.
Nagornyy, L. A. Robarge, D. J. Wardrop, J. Am. Chem. Soc. 2001, 123,
8593.
[20] J. Blankenstein, J. Zhu, Eur. J. Org. Chem. 2005, 1949–1964.
[21] J. I. Seeman, Chem. Rev. 1983, 83, 83–134.
[22] T. Matsushima, K. Horita, N. Nakajima, O. Yonemitsu, Tetrahedron Lett.
1996, 37, 383.
1H NMR d (CDCl3): 3.82-3.76 (1H, m), 2.34 (2H, t, J ¼ 7.6 Hz),
1.67-1.44 (2H, m); 1.44-1.24 (22H, m), 1.18 (3H, d, J ¼ 6.0 Hz).
13C NMR d (CDCl3): 178, 68.07 (CH), 39.04, 33.89, 29.54 (5CH2),
29.44, 29.26, 28.93, 25.58, 24.62, 23.07 (CH3).
Computational methods
All calculations have been performed with SPARTAN 04.[25]
Molecular mechanics calculations have been performed by the
Merck molecular force field (MMFF).[26] Systematic conformation-
al searches have been performed for compounds 7, 8, and 9,
corresponding to the investigation of 35, 37, and 39 confor-
[23] T. Caruso, A. Spinella, Tetrahedron: Asymmetry, 2002, 13, 2071–2073.
[24] A. Spinella, T. Caruso, M. Martino, C. Sessa, Synlett. 2001, 12,
1971–1973.
[25] SPARTAN ’04, Wavefunction Inc. 18401 Von Karman Avenue, Suite
370; Irvine, CA 92612.
[26] T. A. Halgren, J. Comp. Chem. 1996, 17, 490.
J. Phys. Org. Chem. 2009, 22 978–985
Copyright ß 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/poc