Total synthesis of (+/-)-frondosin B and (+/-)-5-epi-liphagal by using a concise (4+3) cycloaddition approach

Article abstract of DOI:10.1002/chem.201303273

A recently developed (4+3) cycloaddition between dienes and furfuryl alcohols, as precursors of oxyallyl-type cations, has been used as a key step in the racemic syntheses of two natural products: frondosin B and liphagal. This work demonstrates the synthetic potential of this cycloaddition reaction, and offers a short synthetic route to an interesting family of natural products. A full account of these synthetic studies is presented, further illustrating the mechanism, scope, and limitations of this straightforward synthetic method for seven-membered rings. Cycloheptanes: Furfuryl alcohols serve as effective three-carbon dienophiles for a wide range of dienes in stereoselective cycloaddition reactions. This efficient reaction has been used to prepare two cycloheptanoid natural products, offering a conceptually straightforward and short synthetic entry into a biologically interesting class of polycyclic meroterpenoids (see scheme; TFA = trifluoroacetic acid). Copyright

Full text of DOI:10.1002/chem.201303273

DOI: 10.1002/chem.201303273
Full Paper
&
Natural Products
Total Synthesis of (+/À)-Frondosin B and (+/À)-5-epi-Liphagal by
Using a Concise (4+3) Cycloaddition Approach
Duchan R. Laplace,[a] Bart Verbraeken,[a] Kristof Van Hecke,[b] and Johan M. Winne*[a]
Dedicated to Prof. Dr. Pierre De Clercq on the occasion of his retirement and his 65th birthday
Abstract: A recently developed (4+3) cycloaddition between
dienes and furfuryl alcohols, as precursors of oxyallyl-type
cations, has been used as a key step in the racemic synthe-
ses of two natural products: frondosin B and liphagal. This
work demonstrates the synthetic potential of this cycloaddi-
tion reaction, and offers a short synthetic route to an inter-
esting family of natural products. A full account of these syn-
thetic studies is presented, further illustrating the mecha-
nism, scope, and limitations of this straightforward synthetic
method for seven-membered rings.
Introduction
Cycloadditions are privileged synthetic methods that allow the
rapid elaboration of stereochemically complex carbocyclic
structures. This is exemplified by the Diels–Alder reaction,
which has had an enormous impact on the science of chemical
synthesis.[1] Indeed, an important step in planning the synthe-
sis of any target molecule that incorporates a cyclohexane sub-
structure, is giving due consideration to all possible (4+2) cy-
cloaddition routes. For the synthesis of seven-membered rings,
the situation is somewhat less straightforward.[2] Although
many useful synthetic methods have been developed that
allow the assembly of cycloheptenes through a (4+3) cycload-
dition of a conjugated diene and a suitable three-carbon dien-
ophile (e.g. oxyallyl cations),[3] these methods suffer from
a rather narrow reactant/substrate scope and require reaction
partners (or precursors thereof) that are synthetically quite
challenging. Their use in the total synthesis of natural products
has thus been limited,[4] especially in the later stages of multi-
step syntheses; that is, the convergent assembly of two rela-
tively advanced intermediates.
Scheme 1. The dehydrative (4+3) cycloaddition between furfuryl alcohols
and conjugated dienes and two cyclohepta[b]furan-containing natural prod-
ucts.
meswaralide.[6,7] Interestingly, Ivanova et al. independently
found a similar transformation by starting from 2-furyl-cyclo-
propane-1,1-dicarboxylates as furfuryl cation precursors.[8] Re-
cently, Wu and co-workers developed a very elegant three-
component coupling process which expands the scope of this
reaction to indole-3-carbinols, giving access to valuable syn-
thetic intermediates.[9] Similarly, benzofuran-3-carbinols have
also been shown to be viable substrates in this (4+3) cycload-
dition.[10]
Some time ago, we reported a novel (4+3) cycloaddition
that uses plain furfuryl alcohols as precursors for furfuryl cat-
ions, which were shown to be excellent three-carbon dieno-
philes for a relatively wide range of dienes (Scheme 1).[5] Pat-
tenden and Winne had originally observed this transformation
during synthetic studies towards the polycyclic diterpene ra-
As this new cycloaddition reaction affords products incorpo-
rating a furan-fused cycloheptene, our attention was drawn to
the natural products frondosin B (1)[11] and liphagal (2),[12]
which both contain this cyclohepta[b]furan substructure
(Scheme 1). Moreover, these tetracyclic meroterpenoids have
been popular targets for total syntheses.[13,14] Especially for li-
phagal, this synthetic interest has been spurred by reports of
a remarkable biological activity, as it selectively inhibits one
isoform of the phosphatidylinositide 3-kinase (PI3K) enzyme
[a] D. R. Laplace, B. Verbraeken, Prof. Dr. J. M. Winne
Department of Organic Chemistry
Ghent University, Krijgslaan 281 S4, 9000 Gent (Belgium)
E-mail: johan.winne@ugent.be
[b] Prof. Dr. K. Van Hecke
Department of Inorganic and Physical Chemistry
Ghent University, Krijgslaan 281 S3
9000 Gent (Belgium)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201303273.
Chem. Eur. J. 2014, 20, 253 – 262
253
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
family.[12] PI3K signaling has been implicated in many impor-
tant cellular functions, which makes liphagal an attractive lead
structure to develop novel therapeutics. In fact, after our initial
report of the (4+3) cycloaddition of furfuryl cations, a prelimi-
nary communication by Xue, Li and co-workers[13(l)] reported
the total synthesis of frondosin B, together with the synthesis
of an advanced intermediate for the synthesis of 5-epi-liphagal,
using this new reaction as a key step. Herein, we present a full
account of our own independent synthetic studies for these in-
teresting natural products, which have offered further insight
into the scope and limitations of the (4+3) cycloaddition reac-
tion of furfuryl cations, illustrating its usefulness in total syn-
thesis.
thylcyclohexanone.[16] The previously described vinylmagnesi-
um bromide addition and subsequent dehydration reaction
proceeded smoothly (Scheme 3), but we found that the ob-
Results and Discussion
The (4+3) cycloaddition of furfuryl cations was found to be
a remarkably efficient reaction.[5–7] These cations can be gener-
ated from straightforward synthetic intermediates and, in con-
trast to most oxyallyl-type cations, furfuryl cations show a rela-
tively wide scope of possible diene reaction partners, giving
synthetically useful yields and selectivities. For the synthesis of
the skeletal types of the title natural products (see retrosynthe-
sis in Scheme 2), we were encouraged by the fact that we had
Scheme 3. Alternative syntheses explored for vinylcyclohexene 3. DIAD= dii-
sopropyl azodicarboxylate.
tained vinylcyclohexene 3 contained considerable amounts of
inseparable impurities (>10%). These were traced back to re-
gioisomeric products in the starting material (which is only
available in ~90% purity from commercial sources). These re-
gioisomeric ketones could not be separated and our own at-
tempts at preparing the dimethylcyclohexanone as a pure re-
gioisomer, by controlled methylation of 2-methylcyclohexa-
none, did not give better results. We soon found that the pres-
ence of these impurities was quite problematic for our study
of the planned cycloadditions, as they gave rise to additional
reaction products, which were hard to separate from the de-
sired compounds. As furfuryl cation cycloadditions sometimes
require an excess of the diene reactant, we also observed
cases in which these regioisomers were actually enriched in
the isolated mixtures of cycloadducts, further complicating re-
action analyses and purification. Therefore, we explored an al-
ternative synthesis for the diene 3, which we expected to give
an isomerically pure final product without problems
(Scheme 3, bottom route). Thus, starting from the known cyclic
b-ketoester 6, easily obtained from very cheap 6-methyl-5-
hepten-2-one,[17] a sodium borohydride reduction and Mitsuno-
bu-type dehydration gave the cyclic enoate ester 7. Next,
a two-step redox adjustment gave the corresponding cyclo-
hexene–carbaldehyde 8 and, finally, a Wittig methylenation
gave the desired vinylcyclohexene 3 in high overall yield and
in very high purity. We later found that careful chromatogra-
phy over argentated silica gel of the previously obtained diene
3, contaminated with a mixture of regioisomers, also gave the
compound 3 in quite pure form (>95%), albeit in a relatively
low recovered yield (~35%). For large-scale (multigram) prepa-
ration, we still prefer the longer route, as this requires very
little and straightforward purification steps, and uses inexpen-
sive starting materials and reagents.
Scheme 2. Retrosynthesis for frondosin B and liphagal by using the (4+3) cy-
cloaddition of furfuryl alcohols gives straightforward fragments.
already shown the known vinyl cyclohexene 4 to react with
a simple furfuryl alcohol under standard conditions, giving
only the desired regioisomer.[5] The isolated yield was rather
low in this case (45%), but we have found that for most sub-
strate combinations the efficiency of the (4+3) cycloaddition
reaction can be optimized to satisfactory levels by changing
the reaction conditions and the type of (Lewis) acid promotor
used. We thus set out to prepare the diene and furfuryl alcohol
precursors required for the assembly of frondosin B (1) and li-
phagal (2).
The vinylcyclohexene 4, required for the liphagal synthesis
(Scheme 2), is a known compound and can be obtained in one
step from commercial b-cyclocitral by using a Wittig methyle-
nation.[15] Similarly, the related vinylcyclohexene 3, containing
one methyl substituent less, is a known compound that can be
prepared in two steps from commercially available 2,2-dime-
In our initial exploration of the key cycloaddition required
for the total synthesis of frondosin B, we had rapidly achieved
a promising model reaction. The reaction between the simple
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
254
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
benzofuran 9[18] and 1-vinyl-cyclohexene under our standard
conditions using stoichiometric amounts of titanium chloride
gave the expected polycyclic core in reasonable yield (11,
Scheme 4). The tetracycle 11 was obtained as an inseparable
dicated most of the undesired reaction products seemed to be
furfuryl oligomerization products (rather than the more usual
and easily removed diene oligomers). Switching to the milder
Lewis acid iron(III)chloride hexahydrate (Scheme 5), a reagent
Scheme 5. Optimization of the key cycloaddition step for frondosin B by
using milder Lewis acids and identification and mechanistic rationalization
of side products.
Scheme 4. Investigation of the key cycloaddition step for the assembly of
the frondosin B polycyclic skeleton by using a model benzofuran substrate 9
and with the fully functionalized substrates 10 and 3.
which we have found to be useful and sometimes superior in
these (4+3) cycloadditions, a much cleaner reaction mixture
was obtained. After chromatography, the O-methyl frondosin B
isomer 14 was isolated as a 1:1 mixture of C-11 epimers. Inter-
estingly, another isolated fraction was found to contain the bis-
(adduct) structure 15 (obtained as a mixture of isomers). Thus,
our suspicion of a Friedel–Crafts-type benzofuran–carbinol oli-
gomerization as the major competing reaction pathway in this
system was confirmed, implicating the methoxy-substituent as
the promotor of this electrophilic aromatic susbstitution reac-
tion. Finally, we found that a modest but reproducible yield of
the desired cycloadducts 14 could be obtained by using tri-
fluoroacetic acid as the cation-generating reagent (35% isolat-
ed yield).[20] This reaction has been optimized by Xue, Li, and
co-workers up to a yield of 50% in their independent studies
by using camphorsulfonic acid as a reagent.[13l,21]
mixture of regioisomeric olefins, and in rather low purity (even
after removal of oligomeric material by filtration over a small
plug of Florisil). This low purity was attributed to the harsh re-
action conditions. Although the titanium(IV)chloride reagent
has been found to be the most reliable and generally applica-
ble for a wide range of substrates,[5] this strong Lewis acid can
indeed be expected to give rise to various acid-catalyzed side
reactions. Nevertheless, this result clearly demonstrated the
overall feasibility of the required transformation.
In contrast to the results obtained with benzofuran 9, and
somewhat unexpectedly, only terribly complex reaction mix-
tures were obtained when the fully functionalized benzofuran
alcohol 10[19] was used. Reaction analyses indicated only minor
amounts of the expected cycloadducts were formed. However,
careful chromatography of the highly complex reaction mix-
ture obtained from the reaction with diene 3 (when used in its
highly pure form), gave a small but relatively clean isolated
fraction. This was found to contain a mixture of only three iso-
meric products, comprising (+/À)-O-methyl frondosin B 12 as
the major product next to two minor positional alkene isomers
(13). Interestingly, this exact mixture has been obtained in
three previous total syntheses of frondosin B, reported by the
groups of Danishefsky,[13a,b] Trauner,[13c,e] and MacMillan.[13j] This
mixture of 12 and 13 has previously been converted to frondo-
sin B by demethylation with boron tribromide.[13(a,b)] Thus,
a formal total synthesis of frondosin B was achieved in our first
attempt at the key cycloaddition.
Our total synthesis of (+/À)-frondosin B was then complet-
ed by following the literature procedure and subjecting the cy-
cloadducts 12/13 or 14 to the action of boron tribromide in di-
chloromethane, giving deprotection of the methyl ether,
and—in the case of 14—concomittant isomerization of the
C5,C6-double bond to C5,C11-position, along with a small
amount of the positional isomer (cf. 13). Thus, a convergent
four-step racemic total synthesis of frondosin B was achieved,
which should be amenable to skeletal diversifications and ana-
logue synthesis around this unique natural product scaffold.
The key cycloaddition for the assembly of the liphagal-skele-
ton from diene 4 (cf. Scheme 2) was initially found to be a lot
less straightforward. Although we had previously used vinyl cy-
clohexene 4 in reactions with various furfuryl alcohols, giving
the expected cycloadducts in modest yields,[5] we could only
observe the formation of traces of these cycloadducts when
the bis(methoxy)-substituted benzofuran–carbinol 16 was used
Unfortunately, the titanium(IV)chloride-promoted reaction
between diene 3 and benzofuran–carbinol 10 was not only
quite low yielding, but proved to be hard to control, as a large
variation in the isolated yield was observed. The reaction com-
plexity (as judged by TLC analysis and proton NMR spectra), in-
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
255
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
the bis(methoxy)–benzofurfuryl cation, relative to the mono-
methoxy benzofurfuryl cation. This explanation was further
supported by the successful synthesis of tetracycle 19. This
‘hybrid’ frondosin B–liphagal structure 19 was formed in good
yield, without apparent benzofuran oligomerization, albeit
with lower stereoselectivity. As expected, reactions of diene 4
with the benzofuran–carbinol 10 only gave complex mixtures
of mostly oligomeric material.
The cycloadduct 18 could be separated to some extent from
the noncyclized adducts (17) by careful chromatography over
silica gel. On a multigram scale; however, it was found to be
more practical to perform a simple radical hydrothiolation with
mercaptoethanol on the crude cycloaddition reaction mixture.
This reaction was serendipitously found to selectively trans-
form the conjugated dienes (17) into more polar mercaptoe-
thanol addition products, which facilitated their removal by fil-
tration over a plug of silica.
Upon establishing a stereoselective route for the tetracyclic
alkene 18 (of which more than 10 grams were prepared in
a single batch), we next explored the hydrogenation of the tri-
substituted olefin to get the saturated fused cyclohexane–cy-
cloheptane system present in liphagal 2 (Scheme 7). This
Scheme 6. Key cycloaddition step for the assembly of the liphagal polycyclic
skeleton (the relative stereochemistry for the major isomer is shown), and
synthesis of a frondosin B–liphagal hybrid analogue by using diene 3.
(Scheme 6). In fact, the major reaction products under our
standard conditions (using titanium(IV)chloride) were shown to
be noncyclized adducts of the diene and the benzofuran, ob-
tained as a mixture of stereoisomeric dienes (17). We rational-
ized this disappointing result in light of the two-step mecha-
nism we proposed for this reaction, which was also confirmed
by DFT calculations.[5] Clearly, these linear adducts arise from
an irreversible proton elimination from the initially formed al-
lylic cation intermediate (shown in Scheme 6). This can be ex-
plained by a slower second step (intramolecular electrophilic
aromatic addition), affected by the increased steric demand
around the receiving carbocationic center, and the altered nu-
cleophilicity of the aromatic system. We then reasoned that
using a protic acid rather than a Lewis acid as a reaction pro-
motor, would allow a reversible proton elimination of the inter-
mediate allylic cation, giving a better chance to the full cyclo-
addition pathway. In fact, using trifluoroacetic acid as the
cation-generating reagent, gave a remarkably improved yield
of the desired cycloadduct 18. To our delight, this reaction
proved to be highly stereoselective,[22] giving a major isomer
showing the relative configuration of the target natural prod-
uct (+/À)-2, even on a decagram scale.
Scheme 7. Attempted hydrogenation and functionalization of polycyclic
alkene 18. DMP=Dess–Martin periodinane.
simple transformation proved to be highly problematic. The
palladium or platinum(IV) oxide catalyzed hydrogenations
were unusually slow, and mostly gave alkene rearrangement to
the disubstituted C6,C7-alkene 21. This rearranged alkene 21
proved highly resistant to hydrogenation and more forcing
conditions resulted mainly in benzofuran hydrogenation. This
alkene rearrangement could be minimized in alcohols as the
solvent, but upon forcing a complete hydrogenation of the
slowly reacting trisubstituted alkene, we were again unable to
avoid significant concommittant hydrogenation of the benzo-
furan moiety. Furthermore, comparison of the proton NMR
spectra of the complex hydrogenation mixtures (and isolated
HPLC fractions), with those reported in the literature for the
trans-fused liphagal-type tetracycles, revealed that in all cases
we obtained major formation of the unnatural cis-fused ring
system (20), and not even a trace of the desired liphagal-type
relative stereochemistry could be observed.[23]
The high efficiency of the cycloaddition giving the liphagal-
type adduct 18 is somewhat surprising, considering our results
for the synthesis of (+/À)-frondosin B (vide supra). In fact, no
trace of Friedel–Crafts-type benzofuran oligomers could be de-
tected in any of the reactions between substrates 4 and 16.
This can be rationalized by the decreased electrophilicity of
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
256
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
A possible explanation for the problems in the hydrogena-
tion reactions described above was found in the bis(methoxy)-
substituted benzofuran. This very electron-rich p-system could
act as a catalyst poison and its coordination to the catalyst sur-
face might even influence the highly selective addition of hy-
drogen from the convex side of the polycyclic scaffold. We ex-
plored several alternatives to achieve the desired transforma-
tion, such as diazene reduction, cationic hydrogenation (TFA-
Et3SiH) or a radical hydrothiolation–hydrodesulfurization, but
the hindered alkene survived all of these reactions. In fact, one
of the few reactions that could be performed efficiently on this
unreactive alkene was a hydroboration. This gave the corre-
sponding secondary alcohol 22, obtained as a single diastereo-
mer in high yield. However, again an exclusive formation of
the unnatural ring fusion stereochemistry resulted. Neverthe-
less, the X-ray diffraction analysis of this crystalline intermedi-
ate did allow the confirmation of the initially made NMR spec-
tra based assignments of the relative stereochemistry in these
systems (Figure 1).[24] Oxidation of the secondary hydroxyl to
Scheme 8. Synthesis and hydrogenation of O,O-dimethyl-5,6-dehydrolipha-
gal 24 (shown yields and d.r.’s for the hydrogenation products are based on
NMR spectroscopic integration, except when mentioned otherwise). TME-
DA=N,N,N’,N’-tetramethylethylenediamine.
synthesis of (+)-liphagal.[14b] As this final substituent alters the
electron density of the aromatic system, we expected this to
affect the hydrogenation reaction and reduce the aforemen-
tioned assumed catalyst poisoning. Indeed, we obtained quite
different results in this case. With platinum(IV)oxide as a hydro-
genation catalyst, the aldehyde 24 was quickly reduced to the
corresponding alcohol 25, which did not undergo any further
reaction. The reactions using palladium on carbon as a catalyst
proved more interesting. Working in ethanol as a solvent,
a complex but interesting reaction mixture was obtained. Ex-
amination of the olefinic region of the crude NMR spectrum in-
dicated a fast and quite clean hydrogenation of the C5,C6-
alkene bond, apart from the formation of a small amount of re-
arranged C6,C7-alkene (cf. compound 21, Scheme 7). The ben-
zofuran system remained intact, but we found that the reac-
tion invariably gave major amounts of the products 27a and
27b, implicating the solvent-assisted formation of hemiacetal
intermediates and their fast hydrogenolysis. This aldehyde side
reaction was hard to control, but the resulting less polar prod-
ucts were easily separated from the reaction mixture, leaving
a quite clean mixture of products consisting mainly of three
diastereomers of compound 26 in a ~10:1:1 ratio (major ste-
reochemistry shown).[23] In contrast with the reactions of
alkene 18, we could now clearly observe the formation of the
natural trans-ring fused stereochemistry in one of the minor
diastereomers. Furthermore, by switching to the less nucleo-
philic solvent 2,2,2-trifluoroethanol, the solvent-assisted hydro-
genolysis side reaction could be completely avoided, resulting
Figure 1. Asymmetric unit of the crystal structure of alcohol 22, which
shows thermal displacement ellipsoids at the 50% probability level and
atom labeling scheme of the non-hydrogen atoms. Both methoxy groups
are found to be almost planar with the benzene ring (torsion angles C6-C1-
O1-C7 and C3-C2-O2-C8 of À2.9(3)8 and 6.1(3)8, respectively). The six-mem-
bered ring adopts a typical chair conformation with the C-22 methyl and
the furan ring as syndiaxial substituents. The configuration of the chiral
carbon atoms of the asymmetric unit was established as C11(R), C13(R),
C14(R) and C15(S), which confirms the relative orientation of the two methyl
groups on the seven-membered ring (corresponding to that found in lipha-
gal 2). The presence of both enantiomers of 22 in the crystal structure is ob-
vious from the centrosymmetric space group (C2/c).
the corresponding ketone 23 then gave an opportunity to epi-
merize the carbonyl a-position by enolization. However, no
trace of the epimerized trans-fused isomer could be detected
after prolonged heating in the presence of a catalytic base.
As the selective hydrogenation the C5,C6-alkene in cycload-
duct 18 proved to be highly problematic (vide supra); there-
fore, we decided to first install the C18 carbaldehyde group to
establish the full carbon framework of the target natural prod-
uct (Scheme 8). The desired tetracyclic aldehyde 24 was ob-
tained in a single step through ortho-lithiation and in situ for-
mylation, applying a procedure used by George et al. in their
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
257
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
in a clean hydrogenation of the C5,C6-double bond, without
affecting the benzofuran–carbaldehyde system. Despite some
effort, we were unable to change the stereoselectivity more in
favor of the natural product trans-fused stereochemistry (at
best, a ~10% conversion of 24 to 5-epi-26 could be obtained).
Furthermore, separation and unambiguous identification of
the minor isomer 5-epi-26 was complicated by the presence of
another minor isomer of aldehyde 26 (presumably 8-epi-26)
which was formed in comparable amounts to the liphagal-type
5-epi-26 diastereomer. However, semi-preparative reversed-
phase HPLC gave a sufficiently pure sample of 5-epi-26 to es-
tablish its identity with the known final intermediate in two
previous liphagal total syntheses.[12,14(d)] Thus, a formal total
synthesis of (+/À)-liphagal was completed.
Conclusion
It has been argued that the true test of any novel synthetic
methodology is its successful application in the total synthesis
of a complex natural product. Here, we have shown that the
(4+3) cycloaddition of furfuryl cations can be used to prepare
two particularly challenging natural products in a straightfor-
ward way. Although our route for liphagal suffers from a late-
stage unfavorable stereoselectivity, which may be difficult to
circumvent efficiently, we believe the overall conciseness and
convergent nature of the approach to these tetracyclic scaf-
folds should be quite useful for the rapid and stereoselective
generation of collections of molecules inspired by the biologi-
cally interesting liphagal/frondosin B framework. Furthermore,
due to the fact that furans are versatile synthetic intermedi-
ates,[25] which can be transformed into a wide range of differ-
ent skeletal structures (e.g. through cycloadditions, oxidative
ring openings and acid hydrolysis or rearrangements), this
methodology should be applicable for the synthesis of a wide
range of interesting scaffolds that contain a seven-membered
ring.
The 5-epi-liphagal dimethyl ether (26) could be obtained
from the hydrogenation mixture as a single diastereomer after
flash chromatography (61% isolated yield). Removal of the cat-
echol methyl ethers proceeded smoothly with boron tribro-
mide (Scheme 9), completing a seven-step stereoselective syn-
Experimental Section
General: For general details, and for the experimental details for
the synthetic procedures leading to compounds 3, 4, 10, 16, 25,
and 29, see the Supporting Information. This file also contains gen-
eral details and a further discussion of the crystallographic analysis
of intermediate 22. Also, images of the original NMR spectra of
final compounds and intermediates are provided.
Synthesis of (+/À)-O-methyl-frondosin B (12): A solution of ben-
zofuran–carbinol 32 (19.8 mg, 0.103 mmol) and 6,6-dimethyl-1-vi-
nylcyclohex-1-ene 3 (28.1 mg, 0.206 mmol; 2 equiv) in dichlorome-
thane (0.50 mL) was cooled to À788C. The reaction mixture was
stirred vigorously at this temperature and then a solution of tita-
nium(IV)chloride (13 mL, 1.1 equiv) in dichlorormethane (0.30 mL)
was added dropwise over 1 min. The resulting mixture was allowed
to warm slowly to À108C over a 90 min period, and then a saturat-
ed aqueous sodium bicarbonate solution (0.5 mL) was added all at
once. The resulting mixture was warmed to room temperature and
methyl tert-butyl ether (5 mL) and water (1 mL) was added. The
layers were separated and the aqueous phase was extracted with
methyl tert-butyl ether (3 ꢁ 1 mL). The organic phase was dried
over anhydrous sodium sulfate and concentrated under reduced
pressure. The residue is purified by flash chromatography over
silica gel, eluting with a gradient of 0 through to 5% of methyl
tert-butyl ether in petroleum ether (b.p. 40–608C). After discarding
a highly apolar fraction, eluting in pure petroleum ether, a slightly
more polar fraction eluted upon increasing the solvent polarity,
giving O-methyl-frondosin B (12) (8.6 mg, 26%, isolated as a 2.5:1
mixture with the isomers 13). This mixture showed a proton NMR
spectrum that corresponded to that reported in the literature for
the same mixture of compounds.[13a,b,c,e,j] Careful analysis of the res-
onances of the minor isomer further revealed a 2:1 ratio of diaste-
reomers. All later eluting (more polar) fractions showed similar
proton NMR resonances, but as broad and/or heavily splitted
bands, with extra peaks in the aromatic regions.
Scheme 9. Synthesis of 5-epi-liphagal, 5-epi-(6R)-hydroxy-liphagal, and at-
tempted synthesis of 5,6-dehydroliphagal.
thesis of (+/À)-5-epi-liphagal (28) in 11% overall yield (26%
from the benzofuran-carbinol 16). Similarly, intermediate 22
was converted to (+/À)-5-epi-(6R)-hydroxy-liphagal (30) by
ortho-lithiation, formylation and methyl ether deprotection.
Unfortunately, this final step also gave rise to significant de-
composition, which decreased the isolated yield. Surprisingly,
all attempts at the deprotection of 5,6-dehydroliphagal di-
methyl ether (24) resulted only in complete decomposition of
this compound.
1
Compounds 12 and 13: H NMR (300 MHz, CDCl3): the main isomer
12 was identified by its characteristic resonances, corresponding to
those reported in the literature:[13a,b] d=7.25 (1H, d, J=8.8 Hz;
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
258
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ArH), 7.12 (1H, d, J=2.5 Hz; ArH), 6.79 (1H, dd, J=8.8, 2.5 Hz; ÀOÀ
C(=CHÀ)CH=CH), 3.81 (3H, s; OÀCH3), 3.15 (1H, q, J=8.5 Hz; CH),
2.54 (2H, t, J=5,9 Hz; CH2-C=C-CH2); the minor isomer 13 showed
two closely related diastereomers; the major diastereomer (inte-
grating for 66%) showed d=7.24 (1H, d, J=8,8 Hz; ArH), 7.05 (1H,
d, 2,5 Hz; ArH), 6.772 (1H, dd, J=8.8, 2.5 Hz; ArH), 5.99 (1H, t, J=
3.5 Hz; C=CHÀCH2), 3.815 (3H, s; CH3OÀ), 3.10–2.95 (1H, m; Fur-
CH); and the minor diastereomer (integrating for 33%) showed:
d=7.23 (1H, d, J=8.8 Hz; ArH), 7.03 (1H, d, 2.5 Hz; ArH), 6.764
(1H, dd, J=8.8, 2.5 Hz; ArH), 5.98 (1H, t, J=3.5 Hz; C=CH-CH2),
3.813 (3H, s; CH3OÀ), 3.10–2.95 ppm (1H, m; Fur-CH); GCMS (EI):
(0.033 mL, 1m in dichloromethane, 0.033 mmol) was added drop-
wise at À558C. The reaction was slowly warmed to 08C over
45 min. After stirring for 30 min at 08C, a mixture of water and ace-
tonitrile (0.5/3.0 mL) was added all at once. The volatiles were re-
moved by using reduced pressure and the resulting residue was fil-
tered over a plug of silica, using a 1:10 ethyl acetate/hexane mix-
ture. Flash chromatography over silica gel, eluting with 6% ethyl
acetate in hexane gave (+/À)-frondosin B (1) (2.3 mg, ~75%, d.r.
3:1 with the positional isomer cf. 13) as a yellow oil. This showed
a proton NMR spectrum corresponding to that reported in the lit-
erature for the same mixture of diastereomers.[12a,b]
+
+
C
C
for 12: m/z: 310 (70) [M ], 295 (100) [M ÀCH3]; for 13: m/z: 310
+
+
C
C
(73) [M ]; 295 (100) [M ÀCH3].
Synthesis of the liphagal-precursor cycloadduct (18): A solution
of benzofuran–carbinol 16 (10.1 g, 45.5 mmol) and vinyl cyclohex-
ene 4 (13.65 g, 91.0 mmol, 2 equiv) in dichloromethane (300 mL)
was cooled to À788C. At this temperature, trifluoroacetic acid
(10.5 mL, 137 mmol) was added dropwise and the resulting mix-
ture was allowed to warm slowly to À258C over 2 h. While stirring
the reaction mixture at À258C, a saturated aqueous sodium bicar-
bonate solution (150 mL) was added all at once. The resulting
layers were separated and the aqueous phase was extracted with
dichloromethane (3ꢁ500 mL). The combined organic layers were
washed with brine, dried over magnesium sulfate, and concentrat-
ed under reduced pressure. The residue was purified over a plug
of silica gel, eluting with a gradient of 0.5 to 10% of ethyl acetate
in petroleum ether (b.p. 40–608C). This gave the cycloadduct 18
(12.94 g, 80.3%), contaminated with a small amount of noncyclized
adducts 17 (~5% by 1H NMR spectroscopic integration). These
noncyclized adducts were removed by first boiling a solution of
this mixture in ethanol (200 mL) and 2-mercaptoethanol (10 mL,
143 mmol) in the presence of azobisisobutyronitrile (250 mg,
1.5 mmol) for 2 h under reflux conditions. Removal of the volatiles
under reduced pressure, followed by filtration over a plug of silica
gel, washing with 8% ethyl acetate in petroleum ether (b.p. 40–
608C) gave cycloadduct 18 (12.08 g, 75%, d.r. 6:1) as a viscous col-
orless oil.
Synthesis of (+/À)-O-methyl-frondosin B alkene isomer (14): A
solution of benzofuran–carbinol 32 (25 mg, 0.13 mmol) and vinyl
cyclohexene 3 (36 mg, 0.26 mmol) in dichloromethane (0.30 mL)
was cooled to À788C. At this temperature, a solution of trifluoro-
acetic acid (0.030 mL, 0.39 mmol) in dichloromethane (0.30 mL)
was added and the resulting mixture was allowed to warm slowly
to À258C over 1 h. While stirring the reaction mixture at À258C,
a saturated aqueous sodium bicarbonate solution (5 mL) was
added all at once, followed by dichloromethane (5 mL). The result-
ing mixture was warmed to room temperature, the layers were
separated and the aqueous phase was extracted with dichlorome-
thane (3ꢁ15 mL). The combined organic phase was washed with
brine, dried on magnesium sulfate, and concentrated under re-
duced pressure. The residue was purified by flash chromatography
over silica gel, eluting with a gradient of 0.5% through to 10% of
ethyl acetate in petroleum ether (b.p. 40–608C). This gave the cy-
cloadduct 14 (14.1 mg, 35%), obtained as an almost 1:1 mixture of
epimers, and the derived Friedel–Crafts adducts 15 (6 mg, 10%),
obtained as a complex mixture of diastereomers.
Cycloadduct 14: The interpretation and assignment of split epimer
resonances (with half integrations) in the proton NMR spectra was
guided by comparison with literature data for these known inter-
mediates.[13j] 1H NMR (300 MHz, CDCl3): 7.27 (d, 0.5H, J=8.8 Hz;
ArH), 7.26 (d, 0.5H, J=8.8 Hz; ArH), 6.87 (d, 1H, J=2.5 Hz; ArH),
6.79 (dd, 1H, J=8.9, 2.5 Hz; ArH), 5.62 (app.brt, 0.5H, J=6.0 Hz;
C=CH-CH2), 5.58 (dd, 0.5H, J=7.7, 4.2 Hz; C=CHÀCH2), 3.86 (s, 3H;
ÀOCH3), 3.70 (brd, 1H, J=12.2 Hz; Fur-CH-C=CH), 3.29–3.09 (m,
1H; ÀCHCH3), 2.62–2.37 (m, 1H; CHH), 2.36–2.13 (band, 2H), 1.97–
1.79 (m, 1H), 1.73 (band, 2H), 1.60–1.55 (band, 2H), 1.35 (d, 3H, J=
6.8 Hz; ÀCHCH3), 1.20 (s, 3H; ÀCCH3CH3), 1.15 (s, 1.5H; CCH3CH3),
1.13 ppm (s, 1.5H; CCH3CH3); HRMS (ESI): m/z: calcd for C22H26O2:
311.2001 [M+H]+; found: 311.2006.
Cycloadduct 18: IR n˜ =2958, 2866, 1623, 1485 cmÀ1 1H NMR
;
(300 MHz, CDCl3): for the major isomer: d=7.14 (s, 1H; ArH), 6.93
(s, 1H; ArH), 5.95 (dd, 1H, J=9.7, 5.8 Hz; C=CHCH2), 3.91 (s, 3H;
OCH3), 3.88 (s, 3H; OCH3), 3.13 (dqd, 1H, J=10.6, 6.8, 3.8 Hz; À
CHMe), 2.70–2.64 (m, 1H; ÀCMe2CH2CH2CHH), 2.65 (ddd, 1H, J=
13.6, 5.8, 3.2 Hz; =CHCHH), 2.19 (ddd, 1H, J=13.6, 9.7, 4.0 Hz; =
CHCHH), 1.89–1.79 (band, 2H), 1.67–1.64 (m, 1H; CHH), 1.62 (s, 3H;
OÀC=C-CCH3), 1.57–1.53 (m, 1H; CHH), 1.46–1.40 (m, 1H; CHH),
1.30 (d, 3H, J=7.0 Hz; CHCH3), 1.24 (s, 3H; CCH3CH3), 1.21 (s, 3H;
CCH3CH3); the minor isomer showed the following major resonan-
ces: 7.15 (s, 1H; ArH), 6.95 (s, 1H; ArH), 6.05 (dd, 1H, J=10.0,
6.0 Hz; C=CHCH2), 3.92 (s, 3H; ÀOCH3), 3.90 (s, 3H; ÀOCH3), 2.97
(dqd, 1H, J=11.6, 7.4, 2.6 Hz; FurCHMe), 2.32 (ddd, 1H, J=13.8,
7.7, 6.0 Hz; C=CHCHH), 1.59 (s, 3H; OÀC=C-CCH3), 1.32 (d, 3H, J=
7.2 Hz; FurCHCH3), 1.22 (s, 3H; CCH3CH3), 1.15 ppm (s, 3H;
CCH3CH3); 13C NMR (75 MHz, CDCl3): for the major isomer: d=156.7
(C), 155.9 (C), 154.0 (C), 153.3 (C), 148.0 (C), 146.8 (C), 144.7 (C),
121.1 (CH), 106.0 (CH), 94.9 (CH), 57.0 (CH3), 56.0 (CH3), 40.3 (C),
38.2 (CH2), 37.5 (CH2), 34.8 (C), 33.7 (CH3), 33.5 (CH3), 33.2 (CH), 31.4
(CH2), 26.9 (CH3), 18.4 (CH3), 18.1 ppm (CH2); an analytical sample
of the minor epimer could not be obtained, hampering the direct
structural assignment through NOE-correlations. However, the as-
signment was obvious through analogy and remarkable similarity
to known synthetic intermediates. Conclusive proof was obtained
in the synthesis of intermediate 22 and of the final compound li-
phagal (2). HRMS (ESI): m/z: calcd for C23H31O3: 355.2268 [M+H]+;
found: 355.2268.
Friedel–Crafts adduct 15: Most diastereomers show closely overlap-
ping resonances, allowing interpretation of the spectrum, guided
by the similarity with those obtained for compound 14. The ob-
served chemical shifts and partial multiplicities of the aromatic
proton resonances are consistent with alkylation on the frondosin
1
C17-position, as shown in Scheme 5. H NMR (300 MHz, CDCl3): d=
7.29–7.23 (m, 1H; ArH), 7.17–7.12 (m, 1H; ArH), 6.99–6.95 (m, 1H;
ArH), 6.88–6.86 (m, 1H; ArH), 6.82–6.76 (m, 1H; ArH), 6.45–6.36 (m,
1H; FurH), 5.64–5.55 (m, 1H; C=CH-CH2-), 4.79 (q, 1H, J=7.0 Hz; À
ArÀCHMeÀFur), 3.93 (s, 3H; ÀOCH3), 3.83 (s, 3H; ÀOCH3), 3.77–3.68
(m, 1H; FurÀCHÀC=CH), 3.26–3.05 (m, 1H; FurCHMe), 2.61–2.12
(band, 3H), 2.00–1.81 (m, 1H), 1.76–1.69 (m, 1H); 1.67–1.13 ppm
(band, 15H; including 4ꢁCH3); HRMS (ESI): m/z: calcd for C32H37O4:
485.2686 [M+H]+; found: 485.2668.
Synthesis of (+/À)-frondosin B (1): A solution of (+/À)-O-methyl
frondosin B (12) (3.2 mg, 0.010 mmol) in dichloromethane (1.0 mL)
was cooled to À558C. Then, a solution of boron tribromide
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
259
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Synthesis of the liphagal–frondosin hybrid cycloadduct (19): A
solution of benzofuran–carbinol 16 (25 mg, 0.113 mmol) and vinyl
cyclohexene 3 (30 mg, 0.226 mmol) in dichloromethane (0.3 mL)
was cooled to À788C. At this temperature, a solution of trifluoro-
acetic acid (25 mL, 0.34 mmol) in dichloromethane (0.2 mL) was
added dropwise and the resulting mixture was allowed to warm
slowly to À258C over 1 h. Then, a saturated aqueous sodium bicar-
bonate solution (2 mL) was added all at once, followed by di-
chloromethane (10 mL). The resulting layers were separated and
the aqueous phase was extracted with dichloromethane (3ꢁ
10 mL). The combined organic layers were washed with brine,
dried over magnesium sulfate and concentrated under reduced
pressure The residue was purified by flash chromatography over
silica gel, eluting with a gradient of 0.5 to 10% of ethyl acetate in
petroleum ether (b.p. 40–608C). This gave the cycloadduct 19
(19.5 mg, 51%, d.r. 2:1) as a viscous colorless oil.
Cycloadduct 19: 1H NMR (300 MHz, CDCl3): for the major isomer:
d=6.99 (s, 1H; ArH), 6.84 (s, 1H; ArH), 5.62 (app.t, 1H, J=6.0 Hz;
C=CHCH2), 3.94 (s, 3H; OCH3), 3.89 (s, 3H; OCH3), 3.69 (brd, 1H, J=
12.1 Hz; Fur-CH-C=CH), 3.19 (dqd, 1H, J=11.7, 7.0, 3.8 Hz; CHMe),
2.41 (ddd, 1H, J=15.2, 7.0, 3.0 Hz; CMe2CH2CH2CHH), 2.34–2.12
(band, 2H; C=CHCH2), 1.98–1.79 (m, 1H), 1.75–1.36 (band, 4H;
CH2CH2), 1.33 (d, 3H, J=7.0 Hz; CHCH3), 1.21 (s, 3H; CCH3CH3),
1.13 ppm (s, 3H; CCH3CH3); the minor isomer showed: d=6.97 (s,
1H; ArH), 6.84 (s, 1H; ArH), 5.58 (dd, 1H, J=7.5, 4.5 Hz; C=CHCH2),
3.16 (dqd, 1H, J=11.7, 7.0, 3.0 Hz; CHMe), 2.60–2.50 ppm (m, 1H);
HRMS (ESI): m/z: calcd for C22H29O3: 341.2111 [M+H]+; found:
341.2114.
Synthesis of (+/À)-O,O-dimethyl-5,6-dehydroliphagal (24): A so-
lution of n-butyllithium in hexanes (2.5m, 3.39 mL, 8.47 mmol) was
added dropwise to a stirring solution of cycloadduct 19 (1.0 g,
2.82 mmol) and tetramethylethylenediamine (1.27 mL, 8.47 mmol)
in THF (25 mL), and the resulting mixture was cooled to 08C and
stirred for 30 min at this temperature. Then, dimethylformamide
(2.17 mL, 28.2 mmol) was added dropwise to the reaction at 08C.
The resulting mixture was warmed to room temperature and
stirred another 20 min at this temperature. Then, a saturated aque-
ous ammonium chloride solution (20 mL) was added and the re-
sulting mixture was stirred vigorously for 5 min. The aqueous
phase was extracted with methyl-tert-butylether (100 mL and 2ꢁ
50 mL) and the combined organic layers were dried over magnesi-
um sulfate and concentrated under reduced pressure. The residue
was purified by flash chromatography over silica gel, eluting with
5% ethyl acetate in petroleum ether (b.p. 40–608C). This gave (+
/À)-O,O-dimethyl-5,6-dehydroliphagal (24) as a slightly yellow oil
(761 mg, 71%, d.r 10:1).
(+/À)-O,O-Dimethyl-5,6-dehydroliphagal (24): 1H NMR (300 MHz,
CDCl3): d=10.55 (s, 1H; CHO), 7.45 (s, 1H; ArH), 5.96 (dd, 1H, J=
9.8, 5.8 Hz; C=CH-CH2), 3.96 (s, 3H; OCH3), 3.93 (s, 3H; OCH3), 3.26
(dqd, 1H, J=10.6, 6.8, 3.8 Hz; ÀCHMe), 2.60 (ddd, 1H, J=13.9, 5.8,
3.0 Hz; C=CHCHH), 2.60–2.53 (m, 1H; ÀCMe2CH2CH2CHH), 2.21
(ddd, 1H, J=13.9, 9.8, 4.1 Hz; C=CHCHH), 1.83–1.76 (m, 1H; CHH),
1.83–1.77 (m, 1H; CHH), 1.68–163 (m, 1H; CHH), 1.62 (s, 3H; OÀC=
C-CCH3), 1.59–1.53 (m, 1H; CHH), 1.44–1.37 (m, 1H;
CMe2CH2CH2CHH), 1.29 (d, 3H, J=7.0 Hz; ÀCHCH3), 1.24 (s, 3H;
CCH3CH3), 1.21 ppm (s, 3H; CCH3CH3); the minor isomer (not isolat-
ed), integrating for about 10%, showed: d=10.56 (s, 1H; ÀCHO),
7.47 (s, 1H; ArH), 6.06 (dd, 1H, J=9.8, 6.2 Hz; C=CH-CH2), 3.97 (s,
3H; OCH3), 3.93 (s, 3H; OCH3), 3.09–2.95 (dqd, 1H, J=11.5, 7.0,
3.2 Hz; CHMe), 1.66 (s, 3H; OÀC=C-CCH3), 1.33 (d, 3H, J=6.9 Hz; À
CHCH3), 1.22 (s, 3H; CCH3CH3), 1.15 ppm (s, 3H; CCH3CH3; 13C NMR
(75 MHz, CDCl3): d=188.4 (CH), 159.7 (C), 153.6 (C), 149.5 (C), 147.8
(C), 146.0 (C), 125.3 (C), 121.5 (CH), 121.1 (C), 114.8 (C), 113.5 (CH),
62.8 (CH3), 57.4 (CH3), 40.3 (C), 38.3 (CH2), 37.7 (CH2), 37.4 (C), 33.9
(CH3), 33.6 (CH3), 33.3 (CH), 31.4 (CH2), 27.0 (CH3), 18.4 (CH3),
18.2 ppm (CH2); HRMS (ESI): m/z: calcd for C24H31O4: 383.2217
[M+H]+; found: 383.2219.
Synthesis of the tetracyclic alcohol 22: A solution of borane in
THF (1.0m, 6.75 mL, 6.75 mmol) was slowly added to a stirring so-
lution of cycloadduct 18 (0.80 g, 2.15 mmol, 6:1 d.r.) in THF
(30 mL), kept at 08C. The reaction was heated and boiled under
reflux for 1 h. After cooling to 08C, an aqueous solution of sodium
hydroxide (15%, 24 mL) was added to the reaction mixture and
stirring was continued for 5 min at 08C. Then an aqueous solution
of hydrogen peroxide (10%, 32 mL) was added and stirring was
continued for 1 h at 08C. Finally, a saturated sodium thiosulfate so-
lution (10 mL) was added and stirring was continued for 5 min
before removing the bulk of the THF under reduced pressure. The
aqueous residue was extracted with ethyl acetate (3ꢁ150 mL) and
the combined organic layers were dried over magnesium sulfate
and concentrated under reduced pressure. The residue was puri-
fied by flash chromatography over silica gel, eluting with 30%
ethyl acetate in hexane. This gave the tetracyclic alcohol 22
(630 mg, 77%) in diastereomerically pure form as a white crystal-
line solid. The expected other diastereomer (cf. d.r. starting materi-
al) was not isolated or detected. Recrystallization from hexane/
ethyl acetate gave a single crystal suitable for X-ray analysis.
Synthesis of (+/À)-O,O-dimethyl-5-epi-liphagal (26), (+/À)-O,O-
dimethyl-liphagal (5-epi-26), and (+/À)-O,O-dimethyl-8-epi-li-
phagal (5-epi-8-epi-26): A solution of O,O-dimethyl-5,6-dehydroli-
phagal 24 (115 mg, 0.30 mmol, 10:1 ratio with the C8-epimer) in
trifluoroethanol (10 mL) in a Schlenk flask (2 neck) was degassed
and backfilled with argon three times. Palladium on carbon (10
w%, 63.6 mg, 0.06 mmol) was added while keeping a gentle posi-
tive flow of argon. The Schlenk flask was cooled down to À788C
upon which the content solidified. The entire set-up was again de-
gassed and backfilled with argon three times. The stop cork was
swiftly replaced by a three way adapter equipped with a hydrogen
filled balloon under a gentle flow of argon. The vessel was de-
gassed and backfilled with hydrogen three times and was then al-
lowed to warm to room temperature. The mixture was stirred vig-
ourously for 2.5 h at room temperature. The reaction mixture was
then filtered over a plug of Celite and concentrated under reduced
pressure. The residue was purified by flash chromatography over
silica gel, eluting with 4% ethyl acetate in hexane. This gave (+
/À)-O,O-dimethyl-5-epi-liphagal 26 (70 mg, 61%), collected as
a single diastereomer, followed by impure fractions also containing
the two minor diastereomers of 26 as well as the alkene rear-
ranged compounds (as judged by proton NMR spectroscopy, these
rearranged alkenes make up 15–20% of the initial crude reaction
mixture). Reversed-phase semi-preparative HPLC of the fractions
Tetracyclic alcohol 22: Rf =0.41 (hexane/ethyl acetate 7:3); m.p.
1
1638C (from hexane/ethyl acetate); H NMR (500 MHz, CDCl3): d=
7.15 (s, 1H; ArH), 6.97 (s, 1H; ArH), 4.60 (dt, 1H, J=5.3, 2.9 Hz; À
CHOH), 3.88 (brs, 6H; 2ꢁÀOCH3), 3.55 (dqd, 1H, J=10.6, 7.2,
4.0 Hz; ÀCHMe), 2.67 (brd, 1H, J=14.2 Hz; CMe2CH2CH2CHH), 2.08–
1.92 (band, 3H), 1.67–1.60 (m, 1H; CHH), 1.57–1.53 (band, 2H, ring-
fusion CH and CHH), 1.50 (s, 3H; OÀC=C-CCH3), 1.51–1.43 (m, 1H;
CMe2CH2CH2CHH), 1.44 (d, 3H, J=7.0 Hz; ÀCHCH3), 1.39 (dd, 1H,
J=12.8, 3.5 Hz; CHH), 1.37–1.32 (m, 1H; CHH), 1.01 (s, 3H;
CCH3CH3), 0.68 ppm (s, 3H; CCH3CH3); 13C NMR (125 MHz, CDCl3):
d=156.3 (C), 148.1 (C), 146.5 (C), 145.3 (C), 119.8 (C), 119.3 (C),
104.7 (CH), 95.0 (CH), 70.9 (CH), 61.7 (CH), 56.6 (CH3), 56.1 (CH3),
45.4 (CH2), 42.7 (CH2), 39.5 (CH2), 38.1 (C), 34.7 (C), 33.3 (CH3), 33.0
(CH3), 27.7 (CH), 24.0 (CH3), 21.4 (CH2), 18.4 ppm (CH3); HRMS (ESI):
m/z: calcd for C23H33O4: 373.2373 [M+H]+; found: 373.2373.
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
260
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
enriched in the minor diastereomers of 26 (Luna, 4.6 mm, gradient
75 to 100% acetonitrile in water with 0.1% trifluoroacetic acid,
15 min) gave small amounts of sufficiently pure samples of these
isomers, which allowed the identification of O,O-dimethylliphagal
(5-epi-26) by comparison with the literature NMR spectroscopic
data.[12] The other diastereomer of 26 was tentatively assigned as
O,O-dimethyl-5-epi-8-epi-liphagal (8-epi-26), further supported by
the fact that this isomer showed a different proton NMR spectrum
than the known O,O-dimethyl-8-epi-liphagal.[12] The ratio of the
major and the two minor diastereomers of 26 can be judged as
~10:1:1 based on the proton NMR spectra of the initial reaction
mixture (see the Supporting Information).
CCH3), 0.97 (s, 3H; CCH3CH3), 0.64 ppm (s, 3H; CCH3CH3); 13C NMR
(75 MHz, CDCl3): d=192.5 (CH), 156.9 (C), 147.5 (C), 145.2 (C), 140.0
(C), 120.2 (C), 120.0 (C), 115.6 (CH), 106.3 (C), 53.9 (CH), 45.2 (C),
40.9 (C), 38.4 (CH2), 35.5 (CH2), 34.9 (CH3), 32.4 (CH3), 32.1 (CH2),
30.4 (CH3), 26.2 (CH2), 25.4 (CH), 21.2 (CH2), 18.4 ppm (CH3).
Acknowledgements
B. Denoo, I. Ascoop, and R. De Wolf are acknowledged for per-
forming intial proof-of-concept studies during their undergrad-
uate Bachelor projects. K.V.H. thanks the Hercules Foundation
(project AUGE/11/029 “3D-SPACE: 3D Structural Platform
Aiming for Chemical Excellence”) for funding. UGent is ac-
knowledged for funding and D.R.L. thanks UGent for a scholar-
ship.
(+/À)-O,O-Dimethyl-5-epi-liphagal (26): Rf =0.27 (hexane/ethyl ace-
1
tate 92:8); H NMR (300 MHz, CDCl3): d=10.58 (s, 1H; ÀCHO), 7.50
(s, 1H; ArH), 3.97 (s, 3H; OCH3), 3.91 (s, 3H; OCH3), 3.17 (dqd, 1H,
J=11.7, 6.9, 4.0 Hz; ÀCHMe), 2.68 (brd, 1H, J=14.1 Hz;
CMe2CH2CH2CHH), 2.26–2.17 (m, 2H), 2.00–1.71 (band, 2H), 1.70–
1.60 (band, 2H), 1.58–1.54 (m, 1H; ring fusion CH), 1.49 (d, 3H, J=
6.9 Hz; ÀCHCH3), 1.37–1.20 (band, 3H), 1.30 (s, 3H; OÀC=C-CCH3),
0.97 (s, 3H; CCH3CH3), 0.64 ppm (s, 3H; CCH3CH3); 13C NMR
(75 MHz, CDCl3): d=188.4 (CH), 159.2 (C), 149.0 (C), 148.3 (C), 145.7
(C), 124.9 (C), 119.1 (C), 114.8 (C), 122.2 (CH), 62.7 (CH3), 56.8 (CH3),
53.9 (CH3), 45.0 (C), 41.2 (C), 38.5 (CH2), 35.5 (CH2), 35.0 (CH), 32.3
(CH3), 32.0 (CH2), 30.4 (CH3), 26.2 (CH2), 25.4 (CH), 21.2 (CH2),
18.4 ppm (CH3); HRMS (ESI): m/z: calcd for C24H33O4: 385.2373
[M+H]+; found: 385.2394.
Keywords: cycloaddition reactions · cycloheptanes · furans ·
natural products · total synthesis
[1] K. C. Nicolaou, S. A. Snyder, T. Montagnon, G. Vassilikogiannakis, Angew.
Chem. 2002, 114, 1742–1773; Angew. Chem. Int. Ed. 2002, 41, 1668–
1698.
[2] V. N. Thanh, J. M. Hartmann, D. Enders, Synthesis 2013, 45, 845–873.
[3] a) M. Harmata, Chem. Commun. 2010, 46, 8886–8903; b) M. Harmata,
Chem. Commun. 2010, 46, 8904–8922.
[4] For an overview of a few remarkable exceptions, see references [3a]
and [3b]; for more recent examples, see: a) J. Wang, S.-G. Chen, B.-F.
Sun, G.-Q. Lin, Y.-J. Shang, Chem. Eur. J. 2013, 19, 2539–2547; b) M. G.
Nilson, R. L. Funk, J. Am. Chem. Soc. 2011, 133, 12451–12453.
[5] J. M. Winne, S. Catak, M. Waroquier, V. Van Speybroeck, Angew. Chem.
2011, 123, 12196–12199; Angew. Chem. Int. Ed. 2011, 50, 11990–11993.
[6] G. Pattenden, J. M. Winne, Tetrahedron Lett. 2009, 50, 7310–7313.
[7] For a related recent study, see: M. J. Palframan, G. Pattenden, Tetrahe-
dron Lett. 2013, 54, 324–328.
(+/À)-O,O-Dimethyl-11-epi-liphagal (8-epi-26): Rf =0.27 (hexane/
ethyl acetate 92:8);1H NMR (300 MHz, CDCl3): d=10.552 (s, 1H;
CHO), 7.53 (s, 1H; ArH), 3.98 (s, 3H; OCH3), 3.91 (s, 3H; OCH3), 3.40
(app.sext, 1H, J=7.2 Hz; ÀCHMe), 2.72 (brd, 1H; J=14.4 Hz;
CMe2CH2CH2CHH), 2.32–2.17 (band, 3H), 2.00–1.93 (band, 2H),
1.77–1.71 (m, 1H), 1.66–1.62 (m, 1H), 1.50–1.40 (band, 3H), 1.39 (s,
3H; OÀC=C-CCH3), 1.37 (d, 3H, J=7.2 Hz; CHMe), 0.97 (s, 3H;
CCH3CH3), 0.67 ppm (s, 3H; CCH3CH3).
(+/À)-O,O-Dimethylliphagal (5-epi-26): Rf =0.27 (hexane/ethyl ace-
tate 92:8); this compound showed all resonances reported for the
known literature compound:[12] 1H NMR (300 MHz, CDCl3): d=
10.547 (s, 1H; CHO), 7.46 (s, 1H; ArH), 3.96 (s, 3H; OCH3), 3.92 (s,
3H; OCH3), 3.30 (appsext, 1H, J=7.2 Hz; CHMe), 2.54 (brd, 1H, J=
12.7 Hz; CMe2CH2CH2CHH), 2.21–2.16 (m, 1H), 1.85–1.81 (m, 1H),
1.75–1.70 (m, 1H), 1.63–1.47 (band, 6H), 1.45 (d, 3H, J=7.2 Hz; À
CHMe), 1.36 (s, 3H; OÀC=C-CCH3), 1.25 (m, 1H), 0.98 (s, 3H;
CCH3CH3), 0.95 ppm (s, 3H, CCH3CH3).
[8] O. A. Ivanova, E. M. Budynina, A. O. Chagarovskiy, A. E. Kaplun, I. V.
Trushkov, M. Y. Melnikov, Adv. Synth. Catal. 2011, 353, 1125–1134.
[9] X. Han, H. Li, R. P. Hughes, J. Wu, Angew. Chem. 2012, 124, 10536–
10539; Angew. Chem. Int. Ed. 2012, 51, 10390–10393.
[10] W. Gong, Y. Liu, J. Zhang, Y. Jiao, J. Xue, Y. Li, Chem. Asian J. 2013, 8,
546–551.
[11] A. D. Patil, A. J. Freyer, L. Killmer, P. Offen, B. Carte, A. J. Jurewicz, R. K.
Johnson, Tetrahedron 1997, 53, 5047–5060.
[12] F. Marion, D. E. Williams, B. O. Patrick, I. Hollander, R. Mallon, S. C. Kim,
D. M. Roll, L. Feldberg, R. Van Soest, R. J. Andersen, Org. Lett. 2006, 8,
321–324.
[13] For total syntheses of frondosin B, see: a) M. Inoue, A. J. Frontier, S. J.
Danishefsky, Angew. Chem. 2000, 112, 777–780; Angew. Chem. Int. Ed.
2000, 39, 761–764; b) M. Inoue, M. W. Carson, A. J. Frontier, S. J. Dani-
shefsky, J. Am. Chem. Soc. 2001, 123, 1878–1889; c) C. C. Hughes, D.
Trauner, Angew. Chem. 2002, 114, 1639–1642; Angew. Chem. Int. Ed.
2002, 41, 1569–1572; d) D. J. Kerr, A. C. Willis, B. L. Flynn, Org. Lett.
2004, 6, 457–460; e) C. C. Hughes, D. Trauner, Tetrahedron 2004, 60,
9675–9686; f) X. Li, T. V. Ovaska, Org. Lett. 2007, 9, 3837–3840; g) J. P.
Olson, H. M. Davies, Org. Lett. 2008, 10, 573–576; h) G. Mehta, N. S. Li-
khite, Tetrahedron Lett. 2008, 49, 7113–7116; i) T. V. Ovaska, J. A. Sullivan,
S. I. Ovaska, J. B. Winegrad, J. D. Fair, Org. Lett. 2009, 11, 2715–2718;
j) M. Reiter, S. Torssell, S. Lee, D. W. C. MacMillan, Chem. Sci. 2010, 1, 37–
42; k) K. S. Masters, B. L. Flynn, Org. Biomol. Chem. 2010, 8, 1290–1292;
l) J. Zhang, L. Q. Li, Y. X. Wang, W. J. Wang, J. J. Xue, Y. Li, Org. Lett. 2012,
14, 4528–4530.
Synthesis of (+/À)-5-epi-liphagal (28): A solution of boron tribro-
mide (0.040 mL, 0.416 mmol) in dichloromethane (0.8 mL) was
added dropwise to a solution of O,O-dimethyl-5-epi-liphagal (26)
(40 mg, 0.104 mmol) in dichloromethane (15 mL), which was stirred
and cooled to À558C. The reaction was slowly warmed to 08C
over 30 min and then stirred for another 30 min at 08C. Then,
a mixture of water (0.75 mL) and acetonitrile (4.50 mL) was added
all at once to the reaction mixture and the volatiles were removed
under reduced pressure. The resulting oily residue was filtered
through a small plug of silica gel, washing with 10% ethyl acetate
in hexane. Concentration of the filtrate under reduced pressure
and purification of the residue by flash chromatography over silica
gel, eluting with 8% ethyl acetate in hexane, gave (+/À)-5-epi-li-
phagal (28) (30 mg, 81.0%) as a yellow film.
5-epi-Liphagal (26): Rf =0.41 (hexane/ethyl acetate 7:3);1H NMR
(300 MHz, CDCl3): d=11.23 (s, 1H; HCO-C=C-OH), 10.46 (s, 1H;
CHO), 7.54 (s, 1H; ArH), 5.40 (s, 1H; OH), 3.16 (dqd, 1H, J=11.9,
7.0, 3.8 Hz; CHMe), 2.63 (brd, 1H, J=14.1 Hz; ÀCMe2CH2CH2CHH),
2.27–2.12 (band, 2H), 1.91–1.73 (m, 2H), 1.69–1.58 (m, 2H), 1.43 (d,
3H, J=7.0 Hz; ÀCHCH3), 1.35–1.29 (m, 2H), 1.29 (s, 3H; OÀC=C-
[14] For total syntheses of liphagal, see reference [12], and also: a) G. Mehta,
N. S. Likhite, C. S. A. Kumar, Tetrahedron Lett. 2009, 50, 5260–5262;
b) J. H. George, J. E. Baldwin, R. M. Adlington, Org. Lett. 2010, 12, 2394–
2397; c) E. Alvarez-Manzaneda, R. Chahboun, E. Alvarez, M. J. Cano, A.
Haidour, R. lvarez-Manzaneda, Org. Lett. 2010, 12, 4450–4453; d) J. J.
Day, R. M. McFadden, S. C. Virgil, H. Kolding, J. L. Alleva, B. M. Stoltz,
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
261
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Angew. Chem. 2011, 123, 6946–6950; Angew. Chem. Int. Ed. 2011, 50,
6814–6818.
[22] Using camphorsulfonic acid at elevated temperatures, as in reference
[13l], results in an unselective formation of a 1:1 mixture of epimers, as
reported in that reference.
[23] The natural C5-epimer of compound 20 is a known synthetic intermedi-
ate for which the NMR spectroscopic data are available, see references
[14b,d]. The unnatural C5-epimer of 20 was also obtained in reference
[13l].
[15] N. Maugel, F. M. Mann, M. L. Hillwig, R. J. Peters, B. B. Snider, Org. Lett.
2010, 12, 2626–2629.
[16] S. Knapp, S. Sharma, J. Org. Chem. 1985, 50, 4996–4998.
[17] T. Hosoya, K. Sumi, H. Doi, M. Wakao, M. Suzuki, Org. Biomol. Chem.
2006, 4, 410–415.
[24] Crystal data for compound 22: C23H32O4, M=372.49, monoclinic, space
group C2/c (No. 15), a=27.995(2), b=11.6489(6), c=14.1728(10) ꢂ, b=
[18] The benzofuran 9 has been successfully used in our previous studies
(see reference [5]), giving an almost quantitative (4+3) cycloaddition re-
action with cyclohexadiene.
120.952(10)8, V=3963.7(6) ꢂ3, Z=8, T=100(2) K, 1calcd =1.248 gcmÀ3
,
m(CuKa)=0.667 mmÀ1, F(000)=1616, 16742 reflections measured, 3881
unique (Rint =0.0545), which were used in all calculations. The final R1
was 0.0552 (I>2s (I)) and wR2 was 0.1535 (all data). CCDC-948989 con-
tains the supplementary crystallographic data for this paper. These data
can be obtained free of charge from The Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[19] D. S. Noyce, R. W. Nichols, J. Org. Chem. 1972, 37, 4311.
[20] Trifluoroacetic acid was the original reagent found to promote these
cycloadditions (see ref. [6]). We have found that the efficiency of this re-
action is often superior to that with titanium chloride, but its success is
more strongly depending on the nature of the furfuryl cation. Con-
versely, titanium chloride will effect a cycloaddition for almost all furfur-
yl alcohols we have used so far (albeit sometimes in modest to poor
yield).
[25] D. L. Wright, in Progress in Heterocyclic Chemistry, Vol. 17 (Eds: G. W.
Gribble, J. A. Joule), Elsevier, 2005, pp. 1–32.
[21] Using camphorsulfonic acid at elevated temperatures, as in reference
[13l], also results in the formation of a 1:1 mixture of epimers, as re-
ported in that reference.
Received: August 20, 2013
Published online on November 28, 2013
Chem. Eur. J. 2014, 20, 253 – 262
www.chemeurj.org
262
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim