Total synthesis of (+)-aquaticol by biomimetic phenol dearomatization: Double diastereofacial differentiation in the Diels-Alder dimerization of orthoquinols with a C2-symmetric transition state

Article abstract of DOI:10.1002/anie.200604610

(Chemical Equation Presented) Double or nothing: The bissesquiterpene (+)-aquaticol was synthesized by biomimetic oxidative dearomatization of 1 via an orthoquinol intermediate that undergoes spontaneous and selective dimerization; only like diastereomers

Full text of DOI:10.1002/anie.200604610

Angewandte
Chemie
DOI: 10.1002/anie.200604610
Diels–Alder Cycloaddition
Total Synthesis of (+)-Aquaticol by Biomimetic Phenol
Dearomatization: Double Diastereofacial Differentiation in the Diels–
Alder Dimerization of Orthoquinols with a C2-Symmetric Transition
State**
Julien Gagnepain, FrØdØric Castet, and StØphane Quideau*
(+)-Aquaticol (1a) is a bissesquiterpene that was isolated
from Veronica anagallis-aquatica, a plant used in traditional
Chinese medicine.[1] Retrosynthetic analysis of its structure
quickly revealed that it should be accessible through a Diels–
Alder dimerization of the orthoquinol (6R,7S)-2,[2] which
could itself be derived from natural (ꢀ)-hydroxycuparene
((ꢀ)-3) by hydroxylation accompanied by dearomatization
(Scheme 1).[3] Although biosynthetic evidence is still lacking,
there is little doubt that this disconnection pathway relates
directly to the route by which 1a is produced naturally.
(that is, 6-alkyl 6-hydroxycyclohexa-2,4-dienones) by a [4+2]
cyclodimerization. These intermediates would be generated
in vivo by a stereoselective dearomatizing ortho hydroxyl-
ation of phenolic monomers. Examples of such a hydroxyl-
ation/Diels–Alder sequence can be found in the chemistry of
the natural phenols curcuphenol,[4a] ferruginol,[4b] and sorbi-
cillin.[4c] All of these cycloadditions occur with endo selectiv-
ity, and the orthoquinol that acts as the dienophile always
reacts through its D-4,5 bond in a site-selective manner. Most
remarkably, the two orthoquinols approach one another with
their hydroxy groups oriented towards each other. No sound
rationale for this double diastereofacial selectivity has yet
been proposed, despite the cornucopia of experimental data
that has been gathered during the last fifty years.[2,5] The
synthesis of 1a gave us the opportunity to investigate further
these intriguing aspects of orthoquinol dimerization. We
report herein the first orthoquinol-based biomimetic syn-
thesis of (+)-aquaticol (1a), and discuss the structural
features that underlie the extraordinary level of diastereo-
facial selectivity observed in this Diels–Alder dimerization
process.
Scheme 1. Biomimetic retrosynthesis of (+)-aquaticol (1a)
In fact, 1a is just one example of several natural
Hydroxycuparene (3) was prepared as the racemate from
(ꢁ )-cuparene (4) (Scheme 2).[6] The nitration of 4[7] was
followed by conversion of the resulting nitrocuparene into an
acetate through the elegant two-step one-pot procedure
described by Glatzhofer et al.[8] Alkaline hydrolysis of this
acetate then furnished 3 in 57% yield from 4 (see the
Supporting Information). This racemate was then resolved by
HPLC on a chiral phase, and each enantiomer was submitted
at room temperature to the dearomatizing ortho-selective
products[4] that can be derived from orthoquinol derivatives
[*] J. Gagnepain, Dr. S. Quideau
Institut EuropØen de Chimie et Biologie
2 rue Robert Escarpit, 33607 Pessac Cedex (France)
Fax: (+33)5-4000-2215
E-mail: s.quideau@iecb.u-bordeaux.fr
and
Laboratoire de Chimie Organique et OrganomØtallique
(UMR CNRS5802)
UniversitØ Bordeaux 1
351 cours de la LibØration, 33405 Talence Cedex (France)
Dr. F. Castet
Laboratoire de Physico-Chimie MolØculaire (UMR CNRS5803)
UniversitØ Bordeaux 1
351 cours de la LibØration, 33405 Talence Cedex (France)
[**] Support for this research was provided by the French Ministry of
Research and Simafex. We thank Prof. Jean-Michel LØger for X-ray
analyses, Anne-Marie Lamidey for her contribution to the synthesis
of 4, and Dr. Raphal MØreau for discussions on calculations
performed at the M3PEC computational center of the UniversitØ
Bordeaux 1.
Supporting information for this article (including experimental
procedures, characterization data for all new compounds, ORTEP
diagrams of X-ray structures, and optimized transition-state geo-
metries) is available on the WWW under http://www.angewand-
te.org or from the author.
Scheme 2. Synthesis of (+)-aquaticol (1a) from (ꢀ)-hydroxycuparene
((ꢀ)-3): a) HNO3/AcOH, Ac2O, 57%; b) 1) H2 (4.5 bar), 10% Pd/C
(80% w/w), Ac2O/AcOH (2:1); 2) Pd/C filtration; 3) NaNO2 (4 equiv),
100%; c) 1) KOH, MeOH/H2O, 100%; 2) separation by HPLC on a
chiral phase. TFA=trifluoroacetic acid.
Angew. Chem. Int. Ed. 2007, 46, 1533 –1535
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1533

Communications
hydroxylation conditions that we developed previously with
facial selectivity as those derived from 3 (see the Supporting
Information). It is unlikely that any hydrogen bonding
between the hydroxy groups at the two C6 centers of the
analogues of 2 has anything to do with this facial selectivity,
for the replacement of the hydroxy group with a fluorine atom
in analogous cases of dimerization led to the same facial
selectivity.[5c] The difference in steric bulkbetween the methyl
and hydroxy groups at C6 has also been used as an argument
for the fact that the bulkier methyl groups end up oriented
away from each other.[5c] However, this simple rationale is
again not fully satisfying, as the dimers we prepared initially
by using SIBX still had even bulkier IBX-derived l3-iodanyl
iodosylbenzoic acid units bonded to the C6 oxygen atom.[10]
The reaction mixture is treated with TFA to induce the
cleavage of these units in situ. Furthermore, when the
hydroxy groups were replaced with bulkier acetate groups
in analogous systems, the methyl groups were still oriented
anti to each other in the cyclodimers formed kinetically.[5b]
We then decided to use computational chemistry to gather
further insight into this facial selectivity. Thus, we carried out
a search at the B3LYP/6-31G(d) level of all transition states
(TSs) that lead experimentally to cyclodimers from the
orthoquinols 2. A natural bond orbital (NBO) analysis[11]
was also carried out to obtain additional details on the
electronic structure of these TSs. These calculations indicated
that the TSs have similar energies and are asynchronous with
the l5-iodane SIBX (stabilized IBX), a stabilized (that is,
nonexplosive) version of o-iodoxybenzoic acid (IBX).[9]
Following treatment with TFA and workup with aqueous
solutions of NaOH and Na2S2O4, reversed-phase HPLC of the
reaction mixture generated from (ꢀ)-3 furnished pure 1a[1] in
6% yield from 4 along with the all-S dimer 1b. 1,2-
Dihydroxycuparene ((ꢀ)-5) was also formed as a result of
unavoidable hydroxylation at the unsubstituted position ortho
to the hydroxy group in (ꢀ)-3.
This one-pot dearomatizing hydroxylation/Diels–Alder
transformation of (ꢀ)-3 led to only two of eight possible endo
cyclodimers. Of course, no stereocontrol was imposed during
the SIBX-mediated hydroxylation at the C6 center of (ꢀ)-3,
but the two diastereomeric orthoquinol intermediates
(6R,7S)-2 and (6S,7S)-2 thus produced recognized each
other and reacted to furnish the cyclodimers (+ )-1a and
(ꢀ)-1b, which were expected on the basis of the regioselec-
tivity, site selectivity, and diastereofacial selectivity observed
previously with many related orthoquinol models.[2,5] When
(ꢁ )-3 was submitted to the same reaction conditions, the four
possible stereoisomers of 2 were formed, and again only those
with the same configuration at their stereogenic C6 center
combined with each other to furnish the expected racemic
mixture of the four endo cyclodimers 1a–d (Scheme 3 and
Supporting Information).
ꢀ
an initial formation of the C5 C5’ bond, the length of which
ranges from 1.92 to 1.95 Š (see the Supporting Information).
Most remarkably, the TS-a structure, which leads to the
natural dimer 1a, has a twofold axis of symmetry (Scheme 4),
Scheme 4. Structure of TS-a, which leads to (+)-aquaticol (1a), with
dihedral angles and interatomic distances (Š), and schematic view of
“Cieplak–Fallis” interactions (in kcalmolꢀ1).
so that it is not possible to differentiate the diene from the
dienophile in this TS.[12] A structure analogous to TS-a, but
with opposite configurations at the two C6 centers, was also
calculated to examine any perturbation brought about by this
stereochemical change. This TS-a’ structure was found to be
9.9 kcalmolꢀ1 higher in energy than TS-a (see the Supporting
Information). This large energy difference can be explained,
at least in part, in terms of hyperconjugative effects, which can
also serve as a basis for rationalizing the facial selectivity.[13] In
all TSs that lead to observed dimers, the D-4,5 bond of one
orthoquinol follows an approach anti to the sC,C bond of the
allylic methyl substituent at C6 of the other orthoquinol unit.
This approach could benefit from Cieplakhyperconjugation,
first discussed by Macaulay and Fallis,[14] since the aforemen-
tioned sC6,Me orbital is suitably aligned with the s#* orbital of
Scheme 3. SIBX-mediated oxidation of (ꢁ)-3. One enantiomer of each
of the racemic homo- and heterodimers is shown; the configurations
of all of them are given at the bottom.
It can be deduced from the above results that the steric
bulkand the chirality of the cyclopentyl unit have no apparent
influence on the observed diastereofacial selectivity. To
convince ourselves further of this observation, we carried
out the same SIBX-mediated reaction on a series of phenols
with different substituents, such as a hydrogen atom, or a
methyl, isopropyl, or tert-butyl group, replacing the cyclo-
pentyl group. All cyclodimeric products, which were isolated
in yields ranging from 30 to 96%, were formed with the same
ꢀ
the incipient C5 C5’ bond, as in TS-a (Scheme 4). The
1534
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 1533 –1535

Angewandte
Chemie
[3] B. J. Hopkins, G. W. Perold, J. Chem. Soc. Perkin Trans. 1 1974,
relevant dihedral angles aCMe-C6-C5-C5’ closely reflect the
ideal antiperiplanar atom positioning required for optimal
hyperconjugation (see the Supporting Information).
32 – 36.
[4] For some examples of orthoquinol-derived natural products, see:
a) C. Zdero, F. Bohlmann, H. M. Niemeyer, Phytochemistry
1991, 30, 1597 – 1601; b) C. P. Falshaw, A. Franklinos, J. Chem.
Soc. Perkin Trans. 1 1984, 95 – 100; c) K. C. Nicolaou, G.
Vassilikogiannakis, K. B. Simonsen, P. S. Baran, Y.-L. Zhong,
V. P. Vidali, E. N. Pitsinos, E. A. Couladouros, J. Am. Chem. Soc.
2000, 122, 3071 – 3079.
[5] a) D. Deffieux, I. Fabre, A. Titz, J.-M. Lꢀger, S. Quideau, J. Org.
Chem. 2004, 69, 8731 – 8738; b) K. Holmberg, Acta Chem. Scand.
Ser. B 1974, 28, 857 – 865, and other articles in the same series;
c) A. S. Kende, P. MacGregor, J. Am. Chem. Soc. 1961, 83, 4197 –
4204.
Our NBO analysis indicated that Cieplak sC6,Me!s#*
interactions amount to approximately 3.0 kcalmolꢀ1, and
apply mutually to both reaction partners (see the Supporting
Information). Felkin–Anh interactions between the incipient
s# bond and the electron-withdrawing s*CO orbitals of the C6-
linked OH groups also occur, but to a much smaller extent
(ca. 1.1 kcalmolꢀ1).[15] However, Felkin–Anh-type interac-
tions that involve instead the more electron donating but
better aligned s*C6-Me bond amount to approximately 2.6 kcal
molꢀ1. We then wondered whether or not these last inter-
actions stabilize these TSs. In this regard, the NBO analysis of
TS-a’ was highly informative and revealed that a change in the
configuration at C6 of one reaction partner does, as expected
from geometrical considerations, reinforce its Felkin–Anh
s#!s*CO interaction (4.4 kcalmolꢀ1). However, the higher-
energy TS-a’ structure, which exhibits a significantly longer
[6] A. Krief, P. Barbeaux, Synlett 1990, 511 – 514.
[7] M. Tashiro, T. Yamato, J. Org. Chem. 1979, 44, 3037 – 3041.
[8] D. T. Glatzhofer, R. R. Roy, K. N. Cossey, Org. Lett. 2002, 4,
2349 – 2352.
[9] a) A. Ozanne, L. Pouysꢀgu, D. Depernet, B. Franꢁois, S.
Quideau, Org. Lett. 2003, 5, 2903 – 2906; b) S. Quideau, L.
Pouysꢀgu, D. Deffieux, A. Ozanne, J. Gagnepain, I. Fabre, M.
Oxoby, ARKIVOC 2003, 6, 106 – 119.
[10] For previous experimental evidence of such a result, see: D.
Magdziak, A. A. Rodriguez, R. W. Van De Water, T. R. R.
Pettus, Org. Lett. 2002, 4, 285 – 288.
ꢀ
C5 C5’ bond (1.99 Š), does not lead to any cyclodimer under
the kinetic conditions used.
Other factors, such as electrostatic and steric effects
induced by the allylic substituents and shown to control single
facial selectivity in some [4+2] cycloaddition systems,[16]
might also contribute to the double diastereofacial selectivity
observed in the [4+2] cyclodimerization described herein.
However, our analysis shows that a double “Cieplak–Fallis”
hyperconjugation appears to be the determining factor in this
stereoselectivity, which was also observed in all cases reported
to date of the kinetically controlled [4+2] dimerization of
chiral orthoquinols.[2,4,5] Finally, we emphasize that we have
described the first example of the construction of a natural
product on the basis of theoretical bispericyclic cycloaddition
models, represented in our case by the C2-symmetric TS-a.[12]
[11] J. P. Foster, F. Weinhold, J. Am. Chem. Soc. 1980, 102, 7211 –
7218.
[12] This C2-symmetric TS is bispericyclic, as it has the ideal
geometry for both the [4+2] and [2+4] pathways on the basis
of both the Woodward–Hoffmann (C3–C3’) and Salem–Houk
(C2–C4’ and/or C2’–C4) secondary orbital interactions, which
also control its preference for endo cycloaddition; see: a) P.
Caramella, P. Quadrelli, L. Toma, J. Am. Chem. Soc. 2002, 124,
1130 – 1131; b) D. M. Birney, K. N. Houk, J. Am. Chem. Soc.
1990, 112, 4127 – 4133.
[13] For a review on models used to describe facial selectivity in
addition reactions, see: M. Kaselj, W.-S. Chung, W. J. le Noble,
Chem. Rev. 1999, 99, 1387 – 1413.
[14] a) J. B. Macaulay, A. G. Fallis, J. Am. Chem. Soc. 1990, 112,
1136 – 1144; b) K. Ohkata, Y. Tamura, B. B. Shetuni, R. Takagi,
W. Miyanaga, S. Kojima, L. A. Paquette, J. Am. Chem. Soc. 2004,
126, 16783 – 16792.
Received: November 13, 2006
Published online: January 16, 2007
[15] For a recent NBO analysis on the predominance of Cieplak
versus Felkin–Anh effects in late TSs, see: V. K. Yadav, A.
Gupta, R. Balamurugan, V. Sriramurthy, N. V. Kumar, J. Org.
Chem. 2006, 71, 4178 – 4182.
[16] a) S. C. Data, R. W. Frank, R. Tripathy, G. J. Quigley, L. Huang,
S. Chen, A. Sihaed, J. Am. Chem. Soc. 1990, 112, 8472 – 8478;
b) L. A. Paquette, B. M. Branan, R. D. Rogers, A. H. Bond, H.
Lange, R. Gleiter, J. Am. Chem. Soc. 1995, 117, 5992 – 6001.
Keywords: cycloaddition · dearomatization ·
diastereoselectivity · hyperconjugation · hypervalent compounds
.
[1] B.-N. Su, Q.-X. Zhu, Z.-J. Jia, Tetrahedron Lett. 1999, 40, 357 –
358.
[2] For a recent review on the chemistry of orthoquinols and related
species, see: S. Quideau in Modern Arene Chemistry (Ed.: D.
Astruc), Wiley-VCH, Weinheim, 2002, pp. 539 – 573.
Angew. Chem. Int. Ed. 2007, 46, 1533 –1535
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1535