7-endo radical cyclizations catalyzed by titanocene(III). Straightforward synthesis of terpenoids with seven-membered carbocycles

Article abstract of DOI:10.1021/ja054316o

We describe a novel procedure for the straightforward synthesis of seven-membered carbocycles via free-radical chemistry, based on titanocene(III)-catalyzed 7-endo-dig and 7-endo-trig cyclizations. This procedure has proved to be useful for the chemical preparation of terpenoids with different skeletons containing cycloheptane rings, including the first total syntheses of dauca-4(11),8-diene (2), barekoxide (3), authentic laukarlaol (81), and a valparane diterpenoid (72), as well as a substantially improved synthesis of karahanaenone (1). We also provide theoretical and experimental evidence in support of a plausible mechanism, which may rationalize the preference for the unusual 7-endo cyclization mode shown by radicals with substitution patterns characteristic of the linalyl, nerolidyl, and geranyl linalyl systems. In light of these chemical findings, we discuss the potential involvement of radical cyclizations in the biosynthesis of some terpenoids containing seven-membered carbocycles.

Full text of DOI:10.1021/ja054316o

Published on Web 09/30/2005
7-endo Radical Cyclizations Catalyzed by Titanocene(III).
Straightforward Synthesis of Terpenoids with
Seven-Membered Carbocycles
Jose´ Justicia,† Juan L. Oller-Lo´pez,† Araceli G. Campan˜a,† J. Enrique Oltra,*,†
Juan M. Cuerva,*,† Elena Bun˜uel,‡ and Diego J. Ca´rdenas*,‡
Contribution from the Department of Organic Chemistry, UniVersity of Granada, Faculty of
Sciences, Campus FuentenueVa s/n, E-18071 Granada, Spain, and Departamento de Qu´ımica
Orga´nica, C-I, UniVersidad Auto´noma de Madrid, Cantoblanco, E-28049 Madrid, Spain
Received June 30, 2005; E-mail: jmcuerva@ugr.es; joltra@ugr.es
Abstract: We describe a novel procedure for the straightforward synthesis of seven-membered carbocycles
via free-radical chemistry, based on titanocene(III)-catalyzed 7-endo-dig and 7-endo-trig cyclizations. This
procedure has proved to be useful for the chemical preparation of terpenoids with different skeletons
containing cycloheptane rings, including the first total syntheses of dauca-4(11),8-diene (2), barekoxide
(3), authentic laukarlaol (81), and a valparane diterpenoid (72), as well as a substantially improved synthesis
of karahanaenone (1). We also provide theoretical and experimental evidence in support of a plausible
mechanism, which may rationalize the preference for the unusual 7-endo cyclization mode shown by radicals
with substitution patterns characteristic of the linalyl, nerolidyl, and geranyl linalyl systems. In light of these
chemical findings, we discuss the potential involvement of radical cyclizations in the biosynthesis of some
terpenoids containing seven-membered carbocycles.
Chart 1. Chemical Structures of Terpenoids with Different
Introduction
Skeletons Containing a Seven-Membered Carbocycle
Seven-membered carbocycles (as well as five- and six-
membered ones) are often classified as “common rings” because
of their relatively low ring strain.1 It is not surprising therefore
that they are quite widespread in nature, where they can be found
in the carbon skeleton of different alkaloids2 and specially
numerous terpenoids, including monocyclic substances such as
the monoterpenoid karahanaenone (1),3 bicyclic sesquiterpenoids
such as daucadiene 2,4 tricyclic diterpenoids such as barekoxide
(3),5 tetracyclic sesterterpenoids such as aspergilloxide (4) (Chart
1),6 and many others.7
Nevertheless, as compared to cyclopentanes and cyclo-
hexanes, general methods for the synthesis of cycloheptane
systems are still notoriously scarce. To repair this deficit, several
researchers have made considerable efforts to develop methods
based mainly on metal-mediated cycloadditions, ring expansion
strategies, and metathesis reactions.8 Alternative methods based
on simple or cascade cyclizations have received much less
attention, however, possibly because of the general assumption
† University of Granada.
‡ Universidad Auto´noma de Madrid.
(1) Eliel, E. L.; Wilen, S. H.; Doyle, M. P. Basic Organic Stereochemistry;
Wiley-Interscience: New York, 2001.
(2) Mann, J.; Davidson, R. S.; Hobbs, J. B.; Banthorpe, D. V.; Harborne, J. B.
Natural products: their chemistry and biological significance; Longman
Scientific & Technical: Harlow, U.K., 1994.
(3) Naya, Y.; Kotake, M. Tetrahedron Lett. 1968, 9, 1645-1649.
(4) (a) Warmers, U.; Ko¨nig, W. A. Phytochemistry 1999, 52, 99-104. (b) Cool,
L. G. Phytochemistry 2001, 58, 969-972.
(5) (a) Kuniyoshi, M.; Marma, M. S.; Higa, T.; Bernardinelli, G.; Jefford, C.
W. J. Nat. Prod. 2001, 64, 696-700. (b) Kuniyoshi, M.; Marma, M. S.;
Higa, T.; Bernardinelli, G.; Jefford, C. W. Chem. Commun. 2000, 1155-
1156. (c) Rudi, A.; Kashman, Y. J. Nat. Prod. 1992, 55, 1408-1414.
(6) Cueto, M.; Jensen, P. R.; Fenical, W. Org. Lett. 2002, 4, 1583-1585.
(7) For an exhaustive overview of natural terpenoids classified by skeleton
type, including most of those containing seven-membered rings, see:
Connolly, J. D.; Hill, R. A. Dictionary of Terpenoids; Chapman & Hall:
London, 1991.
(8) For reviews, see: (a) Yet, L. Chem. ReV. 2000, 100, 2963-3007. (b)
Hoshomi, A.; Tominaga, Y. In ComprehensiVe Organic Synthesis; Trost,
B. M., Fleming, I., Paquette, L. A., Eds.; Pergamon Press: Oxford, U.K.,
1991; Vol. 5, pp 593-615. (c) Hoffmann, H. M. R. Angew. Chem., Int.
Ed. Engl. 1984, 23, 1-19. For selected recent reports: (d) Prie´, G.; Pre´vost,
N.; Twin, H.; Fernandes, S. A.; Hayes, J.; Shipman, M. Angew. Chem.,
Int. Ed. 2004, 43, 6517-6519. (e) Dzwiniel, T. L.; Stryker, J. M. J. Am.
Chem. Soc. 2004, 126, 9184-9185. (f) Barluenga, J.; Alonso, J.; Fan˜ama´s,
F. J.; Borge, J.; Garc´ıa-Granda, S. Angew. Chem., Int. Ed. 2004, 43, 5510-
5513. (g) Lo´pez, F.; Castedo, L.; Mascaren˜as, J. L. Chem.-Eur. J. 2002, 8,
884-899. (h) Barluenga, J.; Alonso, J.; Rodr´ıguez, F.; Fan˜ama´s, F. Angew.
Chem., Int. Ed. 2000, 39, 2460-2462. (i) Wender, P. A.; Glorius, F.;
Husfeld, C. O.; Langkopf, E.; Love, J. A. J. Am. Chem. Soc. 1999, 121,
5348-5349. For synthesis of diterpenoids with a seven-membered car-
bocycle by ring closing metathesis, see: (j) Pfeiffer, M. W. B.; Philips, A.
J. J. Am. Chem. Soc. 2005, 127, 5334-5335. (k) Nakashima, K.; Inoue,
K.; Sono, M.; Tori, M. J. Org. Chem. 2002, 67, 6034-6040.
9
10.1021/ja054316o CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 14911-14921
14911

A R T I C L E S
Justicia et al.
Scheme 1. Hypothetical 7-endo Cyclization of Radical 5 Derived
from Linalyl Pyrophosphate
that entropic factors do not favor cyclizations leading to seven-
membered rings.9 Nevertheless, it is generally believed that
nature uses these types of cyclization processes to build
terpenoids such as 1-4.2-7
The homolytic opening of epoxides mediated by bis(cyclo-
pentadienyl)titanium(III) chloride, introduced by Nugent and
RajanBabu,10 has become a formidable tool in organic synthe-
sis.11 The reaction proceeds under mild conditions compatible
with many functional groups, and the titanocene(III) complex
employed can be generated in situ by simply stirring commercial
Cp2TiCl2 with Mn, Zn, or Al in THF to form an equilibrium
mixture of the monomer Cp2TiCl and the dimer (Cp2TiCl)2.12
The process can also be carried out with substoichiometric
proportions of the titanocene catalyst by adding either a protic
titanocene-regenerating agent such as 2,4,6-collidine hydro-
chloride,13 or an aprotic one such as the combination Me3SiCl/
collidine.14 Relying on this latter catalytic version, we have
recently developed a novel strategy for the synthesis of
terpenoids that mimics oxidosqualene cyclase enzymes using
free-radical chemistry.15 This procedure, based on the radical
cascade cyclization of suitable epoxypolyprenes, has provided
satisfactory yields of relatively complex substances such as
malabaricane triterpenoids, which contain trans-fused five- and
six-membered rings in their carbon skeleton, among others. At
first glance, however, the possibility of adapting this method
to the synthesis of seven-membered carbocycles seemed unlikely
because of the general tendency of 6-heptenyl radicals to
undergo 6-exo instead of 7-endo cyclizations.16 Nevertheless,
a careful inspection of the literature revealed that 5,5-dioxy-
genated 6-heptenoyl radicals inverted their cyclization tendency
in favor of the 7-endo mode.17 In light of this observation, we
realized that a 5,5-disubstituted alkyl radical such as 5,
hypothetically derived from the known terpenoid precursor
linalyl pyrophosphate, should be prone to 7-endo cyclization
processes (Scheme 1).
As preliminary experiments carried out in our laboratory
supported this hypothesis,18 we decided to try to develop a
bioinspired strategy that was generally valid for the synthesis
of terpenoids with different skeletons containing seven-
membered carbocycles. Thus, our aim in this paper is to describe
the titanocene(III)-catalyzed cyclization of epoxides derived
from simpler analogues (acetates) of linalyl, nerolidyl, and
geranyl linalyl pyrophosphates (putative biogenetic precursors
of terpenoids)2 and prove the synthetic usefulness of this
procedure in the preparation of mono- (such as 1), bi- (such as
2), and tricyclic (such as 3) terpenoids containing cycloheptane
systems.
Results and Discussion
Titanocene(III)-Catalyzed 7-endo-trig Cyclizations. In
preliminary assays made in our laboratory, a considerable excess
(2.2 equiv) of the transition-metal complex was needed to
complete the cyclization of 6,7-epoxylinalyl (9) and 10,11-
epoxynerolidyl (10) acetates.18 Moreover, dilution levels to the
order of 10-3 M were required to reduce formation of byprod-
ucts derived from the premature trapping of intermediate radicals
by the excess of titanocene(III).19 To avoid these drawbacks,
we decided to use the catalytic version based on the aprotic
combination Me3SiCl/2,4,6-collidine, which is compatible with
oxiranes and capable of regenerating Cp2TiCl2 from both
Cp2Ti(Cl)H and oxygen-bound titanium derivatives including
Cp2Ti(Cl)OAc.15 Thus, we prepared a set of epoxyalkenes (7-
13) by regioselective epoxidation of commercially available raw
materials, and treated them with a substoichiometric quantity
of Cp2TiCl2 (0.2 equiv), Mn dust, and the Me3SiCl and 2,4,6-
collidine mixture in dry THF (10-1 M substrate concentration).
The results are summarized in Table 1.
Titanocene(III)-catalyzed cyclization of epoxyalkenes 7 and
8 gave cyclohexanols 14 and 15, respectively, but, as hoped,
cyclization of the allylically disubstituted substrates 9-13
mainly provided products (16, 18, and 20) containing the desired
seven-membered ring in acceptable yields.20 Physical and
spectroscopic data of both the synthetic aroma chemical
karahanaenol21 (16) and the widdrol-related22 sesquiterpenoid
1818 matched those previously reported. The HRMS of 20
indicated a C20H34O molecular formula corresponding to four
(9) For a review about the Perkin ring-closure reaction, see: (a) Byrne, L. A.;
Gilheany, D. G. Synlett 2004, 933-943. For a review on unusual radical
cyclizations, including 7-endo, see: (b) Srikrishna, A. In Radicals in
Organic Synthesis; Renaud, P., Sibi, M. P., Eds.; Wiley-VCH: Weinheim,
Germany, 2001; Vol. 2, pp 151-187. There are only a few reports dealing
with the application of simple or cascade cyclizations for the synthesis of
terpenoids containing seven-membered carbocycles. For a seminal work
on the synthesis of serratenediol via cationic 7-endo cyclization, see: (c)
Prestwich, G. D.; Labovitz, J. N. J. Am. Chem. Soc. 1974, 96, 7103-7105.
For an improved synthesis of serratenediol, see: (d) Zhang, J.; Corey, E.
J. Org. Lett. 2001, 3, 3215-3216. For the synthesis of two sesquiterpene
guaianolides via tandem 5-exo/7-endo radical cyclization, see: (e) Lee,
E.; Lim, J. W.; Yoon, C. H.; Sung, Y.; Kim, Y. K.; Yun, M.; Kim, S. J.
Am. Chem. Soc. 1997, 119, 8391-8392.
(10) RajanBabu, T. V.; Nugent, W. A. J. Am. Chem. Soc. 1994, 116, 986-997.
(11) For recent reviews covering this subject, see: (a) Gansa¨uer, A.; Lauterbach,
T.; Narayan, S. Angew. Chem., Int. Ed. 2003, 42, 5556-5573. (b) Gansa¨uer,
A.; Rinker, B. In Titanium and Zirconium in Organic Synthesis; Marek, I.,
Ed.; Wiley-VCH: Weinheim, Germany, 2002; pp 435-450. (c) Gansa¨uer,
A.; Narayan, S. AdV. Synth. Catal. 2002, 344, 465-475. (d) Gansa¨uer, A.;
Rinker, B. Tetrahedron 2002, 58, 7017-7026. (e) Gansa¨uer, A.; Pierobon,
M. In Radicals in Organic Synthesis; Renaud, P., Sibi, M. P., Eds.; Wiley-
VCH: Weinheim, Germany, 2001; Vol. 2, pp 207-220 (f) Gansa¨uer, A.;
Bluhm, H. Chem. ReV. 2000, 100, 2771-2788.
(17) Crich, D.; Fortt, S. M. Tetrahedron 1989, 45, 6581-6589.
(18) Barrero, A. F.; Cuerva, J. M.; Herrador, M. M.; Valdivia, M. V. J. Org.
Chem. 2001, 66, 4074-4078.
(19) There is theoretical and experimental evidence to suggest that free-radical
cyclizations take place in a stepwise manner via discrete carbon-centered
radicals; see the corresponding discussion in ref 15.
(20) Even the lowest yield of 39% obtained in the preparation of 20 from 12
can be regarded as satisfactory if we bear in mind that in just one step the
reaction selectively provided a product containing three fused (trans/anti/
trans) six-/six-/seven-membered rings, an endocyclic double bond, and five
stereogenic centers, among 96 potential regio- and stereoisomers.
(21) Wang, D.; Chan, T. H. Chem. Commun. 1984, 1273.
(22) Widdrol, a unique sesquiterpenoid found in plants of the genus Widdring-
tonia, has the same carbon skeleton as 18, see: Uyehara, T.; Yamada, J.;
Furuta, T.; Kato, T.; Yamamoto, Y. Tetrahedron 1987, 43, 5605-5620
and references therein.
(12) Enemærke, R. J.; Larsen, J.; Skrydstrup, T.; Daasbjerg, K. J. Am. Chem.
Soc. 2004, 126, 7853-7864.
(13) (a) Gansa¨uer, A.; Bluhm, H.; Pierobon, M. J. Am. Chem. Soc. 1998, 120,
12849-12859. (b) Gansa¨uer, A.; Pierobon, M.; Bluhm, H. Angew. Chem.,
Int. Ed. 1998, 37, 101-103. (c) Gansa¨uer, A.; Bluhm, H. Chem. Commun.
1998, 2143-2144.
(14) Barrero, A. F.; Rosales, A.; Cuerva, J. M.; Oltra, J. E. Org. Lett. 2003, 5,
1935-1938.
(15) Justicia, J.; Rosales, A.; Bun˜uel, E.; Oller-Lo´pez, J. L.; Valdivia, M.;
Ha¨ıdour, A.; Oltra, J. E.; Barrero, A. F.; Ca´rdenas, D. J.; Cuerva, J. M.
Chem.-Eur. J. 2004, 10, 1778-1788.
(16) (a) Bailey, W. F.; Longstaff, S. C. Org. Lett. 2001, 3, 2217-2219. (b)
Curran, D. P.; Porter, N. A.; Giese, B. Stereochemistry of Radical Reactions;
VCH: Weinheim, Germany, 1996; pp 77-82.
9
14912 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Synthesis of Seven-Membered Carbocycles
A R T I C L E S
Table 1. Titanocene(III)-Catalyzed Cyclization of Epoxyalkenes
7-13
Scheme 2. Hanessian’s Transition-State Model Adapted to the
7-endo-trig Cyclization of Radical 22
The results summarized in Table 1 merit some comments.
First, the titanocene(III)-catalyzed cyclizations of 9 and 10
provided yields similar to those obtained under stoichiometric
conditions18 but required lesser quantities of Cp2TiCl2 and
dilution levels lower by 1 and 2 orders of magnitude, respec-
tively. Second, the allylic disubstitution pattern characteristic
of linalyl, nerolidyl, and geranyl linalyl derivatives such as 9-13
clearly favored 7-endo cyclizations versus 6-exo ones, and, in
this sense, an allylic ester group seemed to be more effective
than a hydroxyl one. The following mechanistic discussion may
rationalize these observations.
In cyclization processes where either nonsubstituted23 or
monosubstituted16a 6-heptenyl radicals are involved, the relative
proportions of 6-exo and 7-endo cyclization products are a
consequence of the Kexo/Kendo ratios, which are always higher
than 1.16a Moreover, the stereochemistry of the six-membered
carbocycles obtained can be rationalized by the chairlike
transition-state model reported by Hanessian et al.24 In our case,
however, the results of Table 1 suggest that in a 6-heptenyl
radical such as 22, with a 1,1,5,5-tetrasubstitution pattern, the
value of the Kexo/Kendo ratio is lower than 1 and consequently
the 7-endo cyclization product predominates. Hanessian’s model
can be adapted to explain this phenomenon (see Scheme 2).
In a chairlike transition state such as 24, its particular
substitution pattern inevitably creates a 1,3-diaxial interaction,
which presumably will increase the activation energy and should
consequently reduce the reaction rate of the 6-exo cyclization
process. In contrast, a seven-membered ring can adopt confor-
mations such as 23 in which this diaxial interaction is released.25
Therefore, the 7-endo cyclization rate should not be altered (as
compared to that of nonsubstituted or monosubstituted radicals)
and the Kexo/Kendo ratio will be inverted toward values lesser
than 1, thus giving proportions of the secondary radical (25)
which are higher than those of the primary one (26). There is
strong evidence to support the fact that primary alkyl radicals
a Mixture of E/Z isomers (9:1 ratio). b Three-stereoisomer mixture (5.5:
4:1 ratio). c This yield refers to a mixture of 17 and two minor stereoisomers.
d Epimeric mixture (1:1 ratio) of products described in ref 15. e Only the
3S*,13R* diastereomer15 was detected.
degrees of unsaturation, whereas its 13C NMR spectrum showed
only two signals of sp2 carbons (δ 122.7 and 141.4) assignable
to a trisubstituted double bond, thus revealing the tricyclic nature
of the product. In the IR spectrum, a hydroxyl absorption band
appeared at 3415 cm-1, and in the 1H NMR spectrum the
chemical shift, multiplicity, and coupling constant values of H-3
(δ 3.22, dd, J ) 11.5, 4.8 Hz) indicated that the OH group was
attached at C-3 in an equatorial position. Apart from the
differences derived form this hydroxyl group and the double
bond at ∆13, the NMR data of 20 agreed well with those reported
for the tricyclic trans/anti/trans-fused six-/six-/seven-membered
terpenoid 3.5 We therefore tentatively assigned to it structure
20, depicted in Table 1. This structure was subsequently
confirmed by the chemical correlation established during the
synthesis of barekoxide (see below).
(23) Beckwith, A. L. J.; Moad, G. Chem. Commun. 1974, 472-473.
(24) Hanessian, S.; Dhanoa, D. S.; Beaulieu, P. L. Can. J. Chem. 1987, 65,
1859-1866.
(25) For conformational analysis of seven-membered rings, see: Eliel, E. L.;
Wilen, S. H.; Doyle, M. P. Basic Organic Stereochemistry; Wiley-
Interscience: New York, 2001; pp 485-487.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14913

A R T I C L E S
Justicia et al.
Scheme 3. Catalytic Cycle for the Radical Cyclization of 9 to 16 Mediated by Titanocene(III)
Scheme 4. Titanocene(III)-Mediated Radical Cascade Cyclization
of 10 and 11 into 18
formed after titanocene-mediated 5-exo-trig cyclizations are
trapped by a second titanocene(III) species to give an alkyl-
TiIV complex, the C-Ti bond of which finally undergoes
hydrolysis during the aqueous workup to give the corresponding
alkane.10 In a similar way, the secondary radical 25 could be
trapped by another Cp2TiCl species in a process facilitated by
titanium coordination with the acetate group, which would lead
to the bicyclic TiIV complex 27 (Scheme 3). This complex obeys
the 18-electron rule26 and might quickly evolve toward alkene
28 by Cp2Ti(Cl)OAc elimination, thus avoiding further hy-
drolysis of the C-Ti bond.27 The formation of a supposedly
strong Ti-O bond28 might act as the driving force for this
elimination process.
Finally, Cp2TiIVCl2 would be regenerated from 28 and
Cp2Ti(Cl)OAc, by reaction with 29 (presumably derived from
the Me3SiCl/collidine couple) and subsequently reduced to
Cp2TiIIICl by the Mn dust present in the medium. In this way,
the catalytic cycle depicted in Scheme 3 would be complete.
Thus, the two electrons needed for each turnover would be
provided by the oxidation of Mn0 to MnII with a concomitant
transference of two chlorine atoms from the reactive Me3SiCl
(26) Hegedus, L. S. Transition Metals in the Synthesis of Complex Organic
Molecules; University Science Books: Sausalito, CA, 1999; pp 2-4.
(27) There is strong evidence to support the idea that, in the presence of water,
the C-Ti bond of an alkyl-TiIV complex easily undergoes hydrolysis to
give the corresponding alkane, see: Barrero, A. F.; Oltra, J. E.; Cuerva, J.
M.; Rosales, A. J. Org. Chem. 2002, 67, 2566-2571 and ref 10.
Nevertheless, when we repeated the titanocene(III)-catalyzed cyclization
of 9 in the presence of water, apart from 16 and 17, we only detected a
trace of 5-hydroxy-1,4,4-trimethylcycloheptyl acetate, the expected hy-
drolysis product. This result supported the intermediacy of complex 27
and confirmed that the rearrangement of â-acetoxyalkyl-TiIV complexes
(such as 27) toward alkenes (such as 28) is a relatively fast process, as we
had previously suggested (cf., ref 14).
(28) With some exceptions, such as those derived from stable nitroxyl radicals,
Ti-O bonds are generally quite strong. For relevant discussions on this
subject, see: (a) Huang, K.-W.; Han, J. H.; Cole, A. P.; Musgrave, C. B.;
Waymouth, R. M. J. Am. Chem. Soc. 2005, 127, 3807-3816. (b) Gansa¨uer,
A.; Rinker, B.; Ndene-Schiffer, N.; Pierobon, M.; Grimme, S.; Gerenkamp,
M.; Mu¨ck-Lichtenfeld, C. Eur. J. Org. Chem. 2004, 2337-2351. (c)
Gansa¨uer, A.; Rinker, B.; Pierobon, M.; Grimme, S.; Gerenkamp, M.;
Mu¨ck-Lichtenfeld, C. Angew. Chem., Int. Ed. 2003, 42, 3687-3690. (d)
Huang, K.-W.; Waymouth, R. M. J. Am. Chem. Soc. 2002, 124, 8200-
8201. (e) Calhorda, M. J.; Carrondo, M. A. A. F. C. T.; Dias, A. R.;
Domingos, A. M. T. S.; Simoes, J. A. M.; Teixeira, C. Organometallics
1986, 5, 660-667.
to the relatively weak Lewis acid MnCl2. This process,
highlighted in Scheme 3, might be the driving force for the
cycle.
The above discussion suggests that the preference for a seven-
membered carbocycle observed in the titanocene-catalyzed
rearrangement of 9 derives not from an increased 7-endo
cyclization rate but rather from a restricted 6-exo cyclization.
This restraint should be even more acute with radicals 33, 34
(Scheme 4), 41, and 42 (Scheme 5) because of the enhanced
1,3-diaxial interaction caused by the higher rigidity introduced
into the potential transition states 36 and 44 by the preexisting
cyclohexane rings.29 In fact, in the titanocene(III)-catalyzed
cyclizations of 10-13, we detected no product that might have
been derived from these transition states (see Table 1).
9
14914 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Synthesis of Seven-Membered Carbocycles
A R T I C L E S
Scheme 5. Titanocene(III)-Mediated Radical Cascade Cyclization
of 12 and 13 into 20
Scheme 6. Titanocene(III)-Trapping of Radical 42 Facilitated by
Ti-O Coordination
obtained are summarized in Figures 1 and 2. Reaction and
activation energies are referred to the most stable conformer of
the corresponding substrate.31 Computational results are in
agreement with experimental observations and support our
hypothesis about the effect of the disubstitution pattern at C-5
and related positions on the cyclization mode.
Thus, 7-endo cyclization of I (the model for the 2S*,5S*
stereoisomer of radical 22) is more favorable than the 6-exo
process due to both kinetic and thermodynamic reasons. Both
reactions are exothermic and show relatively low activation
energies (11.3 versus 13.5 kcal mol-1). The energy difference
Moreover, from the titanocene(III)-catalyzed cyclizations of
alcohols 11 and 13, we isolated considerable proportions of the
partially cyclized products 19 and 21 (Table 1). Partially cyclized
byproducts closely related to these compounds have been
obtained before from the radical cascade cyclizations of other
epoxypolyenes and presumably derive from the premature
trapping of radical intermediates (such as 34 or 42) by a second
titanocene(III) species (cf., ref 15). It should be noted, however,
that in the latter case we started from a stereoisomeric mixture
of 13 (3S*,14S*/3R*,14S* in a ratio of roughly 1:1) but only
detected the 3S*,13R* diastereomer of the bicyclic product 21
(see Table 1). This result suggests that the stereogenic C-13
center (labdane numbering system) bearing the OH group is
possibly involved in the process through an alkyl-TiIV cyclic
intermediate (46), facilitated by Ti-O coordination, which only
admits the 13R* relative configuration for steric reasons (see
Scheme 6).
between transition states TSI-II and TSI-III is 2.2 kcal mol-1
,
and the seven-membered ring III is 8.0 kcal mol-1 more stable
than the six-membered ring II. These values suggest that the
2S*,5S* stereoisomer of 22 should preferentially evolve to give
16. In the cyclization of the 2S*,5R* diastereomer IV, however,
the energy difference between TSIV-V and TSIV-VI is favorable
for the formation of the six-membered ring V by only 0.5 kcal
mol-1. This low value suggests that the cyclization of the
2S*,5R* stereoisomer of radical 22 should give 17 accompanied
by a considerable proportion of 16. As in our experiments, we
employed an isomeric mixture of 9 in a 1:1 ratio;18 our
theoretical calculations predict the formation of a mixture
containing the 7-endo cyclization product, as the major com-
ponent, and a lesser amount of the 6-exo cyclization derivative
(coming from the 2S*,5R* stereoisomer), as it was experimen-
tally observed in the formation of 16 and 17 from 9 (see Table
1). The different behavior we have found for I and IV is related
to the different 1,3 repulsive interactions mentioned above
(shown for the 2S*,5S* stereoisomer 24 in Scheme 2). The 1,3-
diaxial arrangement of two methyl groups in TSI-II disfavors
the 6-exo cyclization from I, whereas lower repulsions are
exerted by the axial acetate in TSIV-V. Natural bonding orbital
analysis (NBO) of transition states did not show any significant
interaction between the acetate group and the incipient radical,
Theoretical Calculations. We performed calculations at DFT
level on model compounds to determine the activation and
reaction energies for 6-exo-trig and 7-endo-trig radical cycliza-
tions.30 Given that titanium lies far away from the reaction
centers involved in the cyclization process, it was removed from
the model structures to facilitate our calculations. Results
(29) The stereochemistry and nonconcerted character of 6-endo-trig cyclizations
leading to intermediate radicals closely related to 33, 34, 41, and 42 have
been discussed elsewhere; see ref 15.
(30) For a recent theoretical and experimental study of titanocene-catalyzed 3-exo
cyclizations, see: Friedrich, J.; Dolg, M.; Gansa¨uer, A.; Geich-Gimbel,
D.; Lauterbach, T. J. Am. Chem. Soc. 2005, 127, 7071-7077.
(31) Although IRC studies afforded different conformers, they all can be easily
interconverted. For the sake of simplicity, we have only depicted the energy
of the most stable conformer. Full details can be found in the Supporting
Information.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14915

A R T I C L E S
Justicia et al.
Figure 2. Activation and reaction energies for the formation of bicyclic
derivatives from VII and XI.
Figure 1. Activation and reaction energies calculated for the formation of
monocyclic derivatives from the diastereomers I and IV.
intramolecular allylsilylation of alkynes developed by Yama-
moto’s group, a process which takes place via carbocationic
intermediates.33 From a synthetic point of view, therefore, novel
alternative methods using free-radical chemistry seemed desir-
able. In this context, and bearing in mind the mechanisms
discussed above, we anticipated that the cyclization of 5,5-
disubstituted 6-heptynyl radicals might preferably take place
in a 7-endo-dig manner to give seven-membered cycloalkenes.
To check the feasibility and scope of this radical cyclization,
we prepared epoxyalkyne 55 and epoxyalkenynes 56 and 57,
via a simple sequence starting from the commercially available
ketones 49-51 (Scheme 7), and treated them with a sub-
stoichiometric quantity of Cp2TiCl2 (20 mol %), Mn dust, and
the Me3SiCl/2,4,6-collidine couple in dry THF.34 The results
are summarized in Table 2.
After 4 h of reaction, the titanocene(III)-catalyzed cyclization
of epoxyalkyne 55 provided the seven-membered cycloalkene
58 (38% isolated yield), together with a lower proportion of
59. Under the same conditions, epoxyalkenyne 56 gave a
mixture containing acetate 60 and conjugated diene 61, which
became the main reaction product after a longer reaction time
(16 h). A conjugated cycloheptadiene (64) was also the main
cyclization product after 16 h of treatment of 57. There is solid
which further supported that axial repulsion was the main reason
for the observed selectivity.
We have also computed the reaction pathways for the second
ring formation in the radical cyclization of diastereomers VII
(model for the 3′S* stereoisomer of 33) and XI (model for the
3′R* isomer of 33) (Figure 2).
More possibilities have to be considered in these cases
because of the formation of two new stereogenic centers in the
hypothetical 6-exo cyclizations. Nevertheless, 7-endo cyclization
is the preferred pathway regardless of the starting diastereo-
isomer and the relative configuration of the new centers.
Formation of the two possible six-member rings derivatives
shows higher activation energies in all cases. In addition, seven-
membered ring derivatives VIII and XII are the most stable
isomers in each case. Experimental results paralleled calculated
relative energies because in the titanocene-catalyzed cyclization
of 10, only the seven-membered ring 18 was observed (Table
1). The same stereochemical reasons as for the case of 9 lie
behind these data.
Results presented in this section support that radicals such
as 22 and 33, with a substitution pattern characteristic of linalyl,
and nerolidyl pyrophosphate systems,2 have an intrinsic tendency
to undergo 7-endo-trig cyclizations. Moreover, it is presumable
that radical 41 should have the same cyclization tendency. As
we will see later, these observations might have not only
synthetic but also biogenetic relevance.
(32) (a) Clayden, J.; Greeves, N.; Warren, S.; Wothers, P. Organic Chemistry;
Oxford University Press: Oxford, 2001; pp 1140-1144. (b) Baldwin, J.
E.; Lusch, M. J. Tetrahedron 1982, 38, 2939-2947. (c) Baldwin, J. E.
Chem. Commun. 1976, 734-736. (d) Baldwin, J. E.; Cutting, J.; Dupont,
W.; Kruse, L.; Silberman, L.; Thomas, R. C. Chem. Commun. 1976, 736-
738.
Titanocene(III)-Catalyzed 7-endo-dig Cyclizations. Despite
the fact that 7-endo-dig cyclizations are considered to be favored
by Baldwin’s rules,32 methods to achieve 7-endo-dig carbo-
cyclizations are few and far between and the only one we have
found to offer any synthetic usefulness is the HfCl4-catalyzed
(33) Imamura, K.; Yoshikawa, E.; Gevorgyan, V.; Yamamoto, Y. J. Am. Chem.
Soc. 1998, 120, 5339-5340.
(34) It should be noted that when epoxyalkynes 55 and 56 were treated with
stoichiometric quantities of titanocene(III), no seven-membered ring
products were detected. These results, probably due to an increased
premature trapping of intermediate radicals by the high concentration of
titanocene species, reinforce the synthetic value of the catalytic procedure.
9
14916 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Synthesis of Seven-Membered Carbocycles
A R T I C L E S
Table 2. Titanocene(III)-Catalyzed Cyclization of Epoxyalk(en)ynes 55-57
a Diastereomer mixture (2:1 ratio). b Mixture of 1R*,4S* and 1S*,4S* diastereomers (2:1 ratio). c Only one isolated diastereomer (8R* probable configuration).
Scheme 7. Synthesis of Epoxyalk(en)ynes 55-57
Scheme 8. Titanocene(III)-Mediated Cyclization of 56 into 61
radical cyclizations against stereochemical factors, as occurred
with the 6-exo-trig and 7-endo-trig processes.
To the best of our knowledge, the results presented in this
section represent the first examples of 7-endo-dig carbocycliza-
tions using free-radical chemistry reported to date.
Synthesis of Terpenoids Containing Seven-Membered
Carbocycles. With the valuable intermediates 16, 18, and 20
in our hands, the synthesis of natural products 1-3 and related
terpenoids containing seven-membered carbocycles seemed
relatively easy. In fact, simple Dess-Martin oxidation (DMO)
of 16 gave karahanaenone (1).35 Thus, we completed the total
synthesis of this aroma chemical from commercial linalyl acetate
in only three steps (m-CPBA epoxidation,36 titanocene-catalyzed
evidence to indicate that terminal vinyl radicals formed in
titanocene-mediated 5-exo-dig cyclizations are reduced to the
corresponding exocyclic alkenes by hydrogen abstraction from
the THF used as solvent.10 The results shown in Table 2 suggest
that internal vinyl radicals (such as 68) derived from 7-endo-
dig cyclizations might also be reduced by hydrogen abstraction
from THF or collidine present in the medium to give the
corresponding endocyclic alkenes (such as 69) (see Scheme 8).
Subsequently, the allylic tertiary acetate 69 would evolve toward
the conjugated diene 70 by AcOH elimination. It should be noted
that when we treated acetate 60 under similar conditions but in
the absence of Cp2TiCl2, we recovered unchanged starting
material and no diene 61 was detected, thus revealing the crucial
role played by titanocene species in the elimination process.
Moreover, the formation of a single stereoisomer of 60 and 62
showed the high sensitivity of the 6-exo-dig and 7-endo-dig
(35) Justicia, J.; Oltra, J. E.; Cuerva, J. M. J. Org. Chem., in press.
(36) Kametani, T.; Kurobe, H.; Nemoto, H. J. Chem. Soc., Perkin Trans. 1 1981,
756-760.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14917

A R T I C L E S
Justicia et al.
Chart 2. Structures Proposed for Natural 3-Bromo-barekoxide
Scheme 9. Synthesis of 2 from 18 via Carbocation 71
(74)5a,b and Laukarlaol (75)43
Scheme 10. Total Synthesis of 72 and Its Chemical Correlation41
with Natural Valparane 73
group,41 thus confirming the structure proposed for the natural
product. To the best of our knowledge, this is the first total
synthesis of a valparane terpenoid described to date. It was
achieved in only three steps from geranyl-linalyl acetate with a
21% overall yield (Scheme 10).
Marine natural products constitute one of the most exciting
examples of chemical diversity. Many of them are biosynthe-
sized by metabolic pathways exclusive to marine organisms and
have shown an astonishing array of biological properties.42 In
1992, while studying pharmacologically active metabolites from
sponges, Rudi and Kashman isolated barekoxide (3),5c a marine
diterpenoid with an unprecedented carbon skeleton, the structure
of which was established with the aid of X-ray analysis of its
3-bromo-derivative 74.5a,b Three years later, Su et al. isolated
the closely related terpenoid laukarlaol from the red alga
Laurencia karlae and proposed the structure 75 on the basis of
spectral data (Chart 2).43
Barekoxide and laukarlaol attracted our attention not only
because of the synthetic challenge they posed but also because
of their intriguing biogenesis.5a,b Therefore, we decided to
attempt their chemical synthesis from 20 trying to corroborate
structures 3 and 75, provide chemical evidence concerning their
biogenesis, and facilitate further biological analysis.44 Dess-
Martin oxidation of 20 gave ketone 76, which, after treatment
with p-toluenesulfonyl hydrazide, furnished tosylhydrazone 77
(Scheme 11). Catecholborane reduction45 of 77 gave hydrocar-
bon 78, and, finally, m-CPBA epoxidation of 78 took place
selectively at the R-face of the molecule, providing a 71% yield
of 3. The spectroscopic data of synthetic 3 were in accordance
with those reported for natural barekoxide,5 and thus we
achieved the first chemical synthesis of this marine product in
only six steps from geranyl linalyl acetate at an 8% overall yield.
Tosylhydrazone 77 also served as a branch point to start a
divergent route toward authentic laukarlaol (Scheme 12). Thus,
the treatment of 77 with sodium hydride gave diene 79, which,
as occurred with the closely related alkene 78, underwent
selective epoxidation at the R-face of the ∆13 double bond. NOE
observed between H-14 and H3-17 confirmed the R-disposition
of the oxirane ring of 80 (see Figure 3).46 Finally, the perchloric
acid opening of epoxide 80 completed the total synthesis of
alcohol 81 (seven steps with more than 8% overall yield), which
cyclization, and DMO) with a 38% overall yield, substantially
improving previously described procedures.37
Among the sesquiterpenoids with a daucane skeleton7 re-
ported, one can find some confusing structural assignments.38
Thus, the daucadiene structure 2 was originally assigned to a
metabolite from the liverwort Bazzania trilobata,4a but, two
years later, its mirror image was proposed for another metabolite
with different NMR data found in the hybrid cypress species
×Cupressocyparis leylandii.4b In this context, we deemed that
the chemical synthesis of 2 from 18 might help to resolve this
discrepancy and establish an unambiguous synthetic reference
for facilitating further elucidation of the structure of related
natural products. For this purpose, we chose the well-established
six- to five-membered ring contraction procedure employed by
Ourisson and others.39 Thus, we treated 18 with PCl5 obtaining
a 75% yield of 2, presumably via tertiary carbocation 71
(Scheme 9). As this rearrangement does not affect the inter-
annular junction,39 the trans stereochemistry of 2 was guaran-
teed. NMR data of this synthetic product matched those reported
for the metabolite found in the cypress species, thus confirming
the structure and relative stereochemistry proposed by Cool for
this natural product.4b To the best of our knowledge, this is the
first synthesis reported for dauca-4(11),8-diene (2) (three steps
and 19% overall yield from nerolidyl acetate).
In 1990, Urones et al.40 discovered a new type of diterpenoids
with a novel trans/anti/trans-fused five-/six-/seven-membered
tricyclic skeleton called valparane, which has eluded chemical
synthesis until now. The success obtained in the preparation of
2 prompted us to choose valaparane 72 as a synthetic target to
check the usefulness of our method to build this type of
terpenoid. Thus, we treated 20 with PCl5 obtaining an excellent
95% yield of the expected product 72 (Scheme 10). Spectro-
scopic data of this product matched those of the semisynthetic
derivative (72) prepared from natural valparane 73 by Urones’
(41) Urones, J. G.; Marcos, I. S.; Basabe, P.; Alonso, C.; Oliva, I. M.; Garrido,
N. M.; Mart´ın, D. D.; Lithgow, A. M. Tetrahedron 1993, 49, 4051-4062.
(42) (a) Salomon, C. E.; Magarvey, N. A.; Sherman, D. V. Nat. Prod. Rep.
2004, 21, 105-121. (b) Blunt, J. W.; Copp, B. R.; Munro, M. H. G.;
Northcote, P. T.; Prinsep, M. R. Nat. Prod. Rep. 2004, 21, 1-49 and
previous issues in this series.
(43) Su, J. Y.; Zhong, Y. L.; Zeng, L. M.; Wu, H. M.; Ma, K. Phytochemistry
1995, 40, 195-197.
(44) Barekoxide was isolated from the sponge Chelonaplysilla erecta, the extract
of which showed antitumoral activity (cf., ref 5c).
(45) Yee, N. K. N.; Coates, R. M. J. Org. Chem. 1992, 57, 4598-4608.
(46) The interatomic distance between H-14 and H-17 of 80 is 2.757 Å in the
minimized energy conformation shown in Figure 3 (CS ChemBats3D Pro,
MOPAC).
(37) (a) Barrero, A. F.; Alvarez-Manzaneda, E. J.; Palomino, P. L. Tetrahedron
1994, 50, 13239-13250. (b) Uneyama, K.; Date, T.; Torii, S. J. Org. Chem.
1985, 50, 3160-3163.
(38) See, for example, the discussion concerning the relative stereochemistry
of dauca-7,11-diene: (a) Cassidy, M. P.; Ghisalberti, E. L. J. Nat. Prod.
1993, 56, 1190-1193. (b) Bohlmann, F.; Zdero, C. Phytochemistry 1982,
21, 2263-2267.
(39) (a) Harayama, T.; Shinkai, Y.; Hashimoto, Y.; Fukushi, H.; Inubushi, Y.
Tetrahedron Lett. 1983, 24, 5241-5244. (b) Biellmann, J. F.; Ourisson,
G. Bull. Soc. Chim. Fr. 1962, 331-336.
(40) Urones, J. G.; Marcos, I. S.; Basabe, P.; Alonso, C. A.; Diez, D.; Garrido,
N. M.; Oliva, I. M.; Rodilla, J. S.; Slawin, A. M. Z.; Williams, D. J.
Tetrahedron Lett. 1990, 31, 4501-4504.
9
14918 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Synthesis of Seven-Membered Carbocycles
A R T I C L E S
Scheme 11. Synthesis of Barekoxide (3) from 20
Figure 3. Minimized energy conformation of 80. NOE between H-14 and
H-17. Red ) O, gray ) C, blue ) H.
and triterpenes,49 there is no conclusive evidence proving the
true nature of the cyclization mechanisms involved in the
biogenesis of many other terpenoids. In this context, it has been
proposed that the biosynthesis of daucane sesquiterpenoids (such
as 2) and the diterpenoid barekoxide (3) occurs via conventional
carbocationic cyclization of the corresponding nerolidyl and
geranyl linalyl precursors.5a,b,50 On subjecting linalyl, nerolidyl,
and geranyl linalyl derivatives to carbocationic-type cyclizations,
however, different researchers have obtained six-membered
cyclic ethers (presumably via intramolecular oxygen trapping
of the corresponding carbocation intermediates), but seven-
membered carbocycles have never been reported.51 These results
suggest that the behavior shown by 22, 33, 34, 41, and 42 only
applies to free-radical intermediates and not to their carbo-
cationic counterparts. Obviously, these observations do not
constitute a conclusive argument in favor of radical metabolite
intermediates but suggest that the possibility that radical
cyclizations are involved in the biosynthesis of some terpenoids
with seven-membered carbocycles cannot be completely ruled
out. In this sense, the biogenesis of marine barekane diterpenoids
is especially interesting.52 Thus, the co-occurrence of the bromo-
derivative 74 together with barekoxide (3) in the red alga
Laurencia luzonensis5a,b might well point to a biogenetic
cyclization initiated by bromine radical addition, a well-known
chemical process.53 In light of recent findings, the alternative
bromonium-ion-initiated cationic cyclization proposed by
Kuniyoshi et al.5a,b seems quite unlikely because Carter-Franklin
and Butler have shown how enzymes such as vanadium
bromoperoxidase, which catalyzes this type of cationic process,
produce tetrahydropiranyl ethers and snyderol derivatives but
not seven-membered carbocycles.54
Scheme 12. Synthesis of Authentic Laukarlaol (81) from 20
showed spectroscopic data matching those reported for natural
laukarlaol.47 Therefore, the relative 14S* stereochemistry (75)
originally proposed for this natural product43 should be revised
to the 14R* one depicted in 81.
The results shown in this section confirm the utility of
titanocene(III)-catalyzed radical cyclizations for the straight-
forward synthesis of mono (1), sesqui (2), and diterpenoids (3,
72, and 81) with different carbon skeletons containing a seven-
membered carbocyle.
Biogenetic Discussion. The new findings concerning the
intrinsic tendency shown by radicals 22, 33, 34, 41, and 42 (with
a substitution pattern characteristic of the linalyl, nerolidyl, and
geranyl linalyl systems) to undergo 7-endo cyclizations led us
to consider the possible implications of this phenomenon in the
biosynthesis of 1-3. Linalyl, nerolidyl, and geranyl linalyl
pyrophosphate esters and the corresponding alcohols are rec-
ognized among the acyclic biogenetic parents of mono-, sesqui-,
and diterpenoids, together with their allylic isomers, the geranyl,
farnesyl, and geranyl geranyl systems, respectively.2,48 Further-
more, it is generally assumed that the enzyme-catalyzed
cyclization of these precursors toward the different terpenoid
skeletons takes place via hypothetical carbocation intermedi-
ates.2,48 Nevertheless, apart from the currently well-studied
enzymatic cyclization of squalene and oxidosqualene to sterols
(49) For excellent reviews on this subject, see: (a) Hohino, T.; Sato, T. Chem.
Commun. 2002, 291-301. (b) Wendt, K. U.; Schulz, G. E.; Corey, E. J.;
Liu, D. R. Angew. Chem., Int. Ed. 2000, 39, 2812-2833. (c) Abe, I.;
Rohmer, M.; Prestwich, G. D. Chem. ReV. 1993, 93, 2189-2206. For more
recent reports, see: (d) Nishizawa, M.; Yadav, A.; Iwamoto, Y.; Imagawa,
H. Tetrahedron 2004, 60, 9223-9234. (e) Hess, B. A., Jr. Org. Lett. 2003,
5, 165-167. (f) Hess, B. A., Jr. J. Am. Chem. Soc. 2002, 124, 10286-
10287.
(50) Ghisalberti, E. L. Phytochemistry 1994, 37, 597-623.
(51) (a) Uyanic, M.; Ishibashi, H.; Ishihara, K.; Yamamoto, H. Org. Lett. 2005,
7, 1601-1604. (b) Khomenko, T. M.; Tatarova, L. E.; Korchagina, D. V.;
Barkhash, V. A. Russ. J. Org. Chem. 2002, 38, 498-506. (c) Polovinca,
M. P.; Korchagina, D. V.; Gatilov, Y. V.; Bagrianskaya, I. Y.; Barkhash,
V. A.; Shcherbukhin, V. V.; Zefirov, N. S.; Perutskii, V. B.; Ungur, N. D.;
Vlad, P. F. J. Org. Chem. 1994, 59, 1509-1517. (d) Kametani, T.;
Fukumoto, K.; Kurobe, H.; Nemoto, H. Tetrahedron Lett. 1981, 22, 3653-
3656.
(52) We propose the name “barekane” for the diterpenoid skeleton shared by 3,
74, and 81.
(47) A copy of the 1H NMR spectrum of synthetic 81 and a table with 13C
NMR data of 81 and natural laukarlaol43 are given for comparison as
Supporting Information.
(53) Carey, F. A.; Sundberg, R. J. AdVanced Organic Chemistry, 3rd ed.; Plenum
Press: New York, 1990; Part A, pp 695-701.
(48) For a review on the biosynthesis of C5-C25 terpenoids, see: Dewick, P.
(54) Carter-Franklin, J. N.; Butler, A. J. Am. Chem. Soc. 2004, 126, 15060-
15066.
M. Nat. Prod. Rep. 2002, 19, 181-222 and previous issues in this series.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14919

A R T I C L E S
Justicia et al.
with 2 N HCl and extracted with t-BuOMe. The organic layer was
washed with brine, dried with anhyd Na2SO4, and the solvent was
removed. The residue was dissolved in THF (20 mL) and stirred with
Bu4NF (10 mmol) for 2 h.64 The mixture was then diluted with
t-BuOMe, washed with brine, dried with anhyd Na2SO4, and the solvent
was removed. The products were isolated by flash chromatography of
the residue to give the corresponding carbocycles at the yields referred
to in Tables 1 and 2.
Synthesis of Daucadiene 2. A solution of 18 (70 mg, 0.31 mmol)
and PCl5 (72 mg, 0.34 mmol) in a benzene:toluene mixture (15 mL,
2:1) was stirred for 1 h at 0 °C. The reaction was quenched with 5%
aqueous K2CO3, extracted with t-BuOMe, washed with water, dried
with anhydrous Na2SO4, and the solvent was removed. The residue
was submitted to flash chromatography (hexane) to give 24b (47 mg,
75%) as a colorless oil.
Synthesis of Valparane 72. A solution of 20 (17 mg, 0.06 mmol)
and PCl5 (15 mg, 0.07 mmol) in a benzene:toluene mixture (15 mL,
2:1) was stirred for 1 h at 0 °C. The reaction was quenched with 5%
aqueous K2CO3, extracted with t-BuOMe, washed with water, dried
with anhydrous Na2SO4, and the solvent removed. The residue was
chromatographed (hexane) to give 7241 (15 mg, 95%) as a colorless
oil.
Conclusions
We have developed a new strategy for the straightforward
synthesis of seven-membered carbocycles, based on titano-
cene(III)-catalyzed 7-endo-dig and 7-endo-trig cyclizations via
free-radical chemistry. This procedure has proved to be useful
for the preparation of several natural products including mono-,
sesqui-, and diterpenoids with different skeletons containing
cycloheptane rings. We have also provided theoretical and
experimental evidence in support of a plausible mechanism to
rationalize the preference for the unusual 7-endo cyclization
mode shown by radicals with substitution patterns characteristic
of the linalyl, nerolidyl, and geranyl linalyl systems. In light of
these chemical findings, we discuss the potential involvement
of radical cyclizations in the biosynthesis of some terpenoids
with seven-membered carbocycles. We are currently engaged
in the total synthesis of aspergilloxide (4) in the hope of
extending the scope of our method to the chemical preparation
of bioactive sesterterpenoids found in marine organisms.
Computational Methods
Preparation of Ketone 76. A solution of 20 (290 mg, 1.06 mmol)
and Dess-Martin periodinane (1.50 g, 2.84 mmol) in wet CH2Cl2 (33
mL) was stirred for 45 h at room temperature. The reaction was diluted
with t-BuOMe, washed with a 1:1 mixture of 10% aqueous Na2S2O3
and saturated NaHCO3 solutions, washed with brine, dried with
anhydrous Na2SO4, and the solvent was removed. The residue was
submitted to flash chromatogrphy (4:1 hexane:t-BuOMe) to give 76
(253 mg, 88%).
Preparation of Tosylhydrazone 77. A solution of 76 (231 mg, 0.80
mmol) and TsNHNH2 (223 mg, 1.20 mmol) in absolute EtOH (10 mL)
was refluxed for 2 h. The solvent was removed, and the residue was
submitted to flash chrmotagraphy (4:1 hexane:t-BuOMe) to give 77
(334 mg, 92%).
Calculations were performed with the Gaussian 03 series of
programs.55 The geometries of all intermediates were optimized at the
DFT level using the B3LYP hybrid functional,56 using the standard
6-31G(d) basis set for C, H, and O. Transition states were graphically
located. Harmonic frequencies were calculated at the same level of
theory to characterize the stationary points and to determine the zero-
point energies (ZPE). Final energies include ZPE correction without
scaling. Intrinsic reaction coordinate (IRC) studies were performed to
ensure connection between reagents and products. The bonding
characteristics of the different stationary points were analyzed by means
of the Natural Bond Orbital (NBO) analysis of Weinhold et al.57
Experimental Section
Preparation of Tricyclic Alkene 78. Catecholborane (0.21 mL of
1 M in THF, 0.21 mmol) was added to a solution of 77 (32 mg, 0.07
mmol) in CH2Cl2 at 0 °C. The mixture was stirred for 30 min at 0 °C
and then for 1 h at room temperature. MeOH (1 mL) and NaOAc (44
mg, 0.42 mmol) were then added. This mixture was stirred at room
temperature for another 16 h, diluted with t-BuOMe, washed with water,
dried with anhydrous Na2SO4, and the solvent was removed. The residue
was chromatographed (hexane) to give 78 (12 mg, 62%).
Synthesis of Barekoxide (3). A mixture of 78 (8 mg, 0.03 mmol)
and m-CPBA (18 mg, 0.07 mmol) in CH2Cl2 (5 mL) was stirred for 1
h at room temperature. The reaction was diluted with CH2Cl2, washed
with water, dried with anhydrous Na2SO4, and the solvent removed.
The residue was submitted to flash chromatography (95:5 hexane:t-
BuOMe) to give barekoxide5 (3) (6 mg, 71%) as a white solid, mp
86-90 °C (lit.5a,b mp 140 °C).
Preparation of Tricyclic Diene 79. A mixture of 77 (290 mg, 0.64
mmol) and NaH (763 mg, 31.7 mmol) in toluene (10 mL) was boiled
under reflux for 1 h. The reaction was diluted with t-BuOMe, washed
with water, dried with anhydrous Na2SO4, and the solvent removed.
The residue was chromatographed (hexane) to give 79 (121 mg, 70%).
Preparation of Epoxide 80. A solution of 79 (70 mg, 0.26 mmol)
and m-CPBA (44 mg, 0.26 mmol) in CH2Cl2 (5 mL) was stirred for 30
min at 0 °C. The reaction was diluted with CH2Cl2, washed with water,
dried with anhydrous Na2SO4, and the solvent removed. The residue
was submitted to flash chromatography (95:5 hexane:t-BuOMe) to give
80 (69 mg, 93%).
General. For the reactions with titanocene, all solvents and additives
were thoroughly deoxygenated prior to use. Although all structures are
drawn as one enantiomers, the synthesized compounds are racemic.
Substances 7,58 8,59 9,18 and 1018 were prepared according to published
procedures. The following known compounds were isolated as pure
samples and their NMR spectra were identical to the reported
compounds: 1,3 2,4b 3,5 11,60 16,18 17,18 18,18 19,15 21,15 52,61 53,62
55,63 and 72.41 Preparation procedures for substrates 11-13 and 52-
57 as well as physical and spectroscopic data for 12-15, 20, 56-65,
76-80, and 5-hydroxy-1,4,4-trimethylcycloheptyl acetate are given as
Supporting Information.
Model Procedure for Titanocene(III)-Catalyzed Cyclizations.
Thoroughly deoxygenated THF (20 mL) was added to a mixture of
Cp2TiCl2 (0.5 mmol) and Mn dust (20 mmol) under an Ar atmosphere,
and the suspension was stirred at room temperature until it turned lime
green (after about 15 min). A solution of epoxide (2.5 mmol), 2,4,6-
collidine (20 mmol), and Me3SiCl (10 mmol) in THF (2 mL) was then
added, and the mixture was stirred for 16 h. The reaction was quenched
(55) Frisch, M. J.; et al. Gaussian 03, revision B.03; Gaussian, Inc.: Pittsburgh,
PA, 2003.
(56) (a) Becke, A. D. Phys. ReV. A 1988, 38, 3098-3100. (b) Lee, C.; Yang,
W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(57) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88, 899-926.
(58) Rodriguez, J.; Dulcore, J. P. J. Org. Chem. 1991, 56, 469-471.
(59) Cernigliaro, C. J.; Kocienski, P. J. J. Org. Chem. 1977, 42, 3622-3624
(60) Miyase, T.; Ueno, A.; Takizawa, N.; Kobayashi, H.; Karasawa, H. Chem.
Pharm. Bull. 1987, 35, 1109-11017.
Synthesis of Authentic Laukarlaol (81). Two drops of 60% HClO4
were added to a solution of 80 (17 mg, 0.06 mmol) in DMF (2 mL) at
(61) Anjum, S.; Marco-Contelle´s, J. Tetrahedron 2005, 61, 4793-4803.
(62) Fu¨rstner, A.; Hammen, P. Chem. Commun. 2004, 2546-2547. See also
ref 60.
(63) Torii, S.; Uneyama, K.; Tanaka, H.; Yamanaka, T.; Yasuda, T.; Ono, M.;
Kohmoto, Y. J. Org. Chem. 1981, 46, 3312-3315.
(64) In some experiments, acidic quenching was enough to promote the
desilylation reaction.
9
14920 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Synthesis of Seven-Membered Carbocycles
A R T I C L E S
room temperature. The reaction was stirred for 16 h and then diluted
with t-BuOMe, washed with water, dried with anhydrous Na2SO4, and
the solvent was removed. The residue was submitted to flash chroma-
tography (4:1 hexane:t-BuOMe) to give 81 (12 mg, 71%) as a white
solid, mp 105-106 °C (lit.43 mp 128-129) with NMR and mass spectra
matching those reported for natural laukarlaol.43 EIHRMS calcd for
C20H32O m/z 288.2453, found m/z 288.2453.
for Project CTQ2004-02040/BQU and the “Centro de Com-
putacio´n Cient´ıfica (UAM)” for computation time.
Supporting Information Available: Preparation procedures
for substrates 11-13 and 52-57 as well as physical and spec-
troscopic data for 12-15, 20, 56-65, 76-80, and 5-hydroxy-
1,4,4-trimethylcycloheptyl acetate. A comparison between 13C
NMR data reported for natural laukarlaol43 and those from
synthetic 81. Complete ref 55 with the full authors list. Copies
Acknowledgment. We thank the Spanish “Junta de Anda-
luc´ıa” for financial support (research group FQM339), Dr. A.
F. Barrero for his collaboration in the early stage of this work,
and our English colleague Dr. J. Trout for revising our text.
J.J. and J.L.O.-L. thank the Spanish Ministry of Education and
Science for grants enabling them to pursue these studies. D.J.C.
and E.B. thank the Spanish Ministry of Education and Science
1
of the H or 13C NMR spectra for all of the new compounds.
Atomic coordinates and energies for stationary points. This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA054316O
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14921