Synthesis of (-)-deoxypukalide, the enantiomer of a degradation product of the furanocembranolide pukalide

Article abstract of DOI:10.1021/jo016048s

A convergent stereoselective synthesis of (-)-deoxypukalide is described. This substance has not yet been found in Nature but is obtained through deoxygenation of pukalide, the first naturally occurring furanocembrane to be structurally elucidated. The route features a new intraannular furan synthesis that entails treatment of a 4-oxopropargylic β-keto ester with silica gel. The product of this novel reaction, a 3-carboxy 2,5-bridged furan, is formed in 96% yield. The synthetic strategy was strongly directed by molecular mechanics calculations, which provided valuable insight into stereodefining steps including double bond stereochemistry and butenolide configuration.

Full text of DOI:10.1021/jo016048s

J . Org. Chem. 2001, 66, 8037-8041
8037
Syn th esis of (-)-Deoxyp u k a lid e, th e En a n tiom er of a Degr a d a tion
P r od u ct of th e F u r a n ocem br a n olid e P u k a lid e
J ames A. Marshall* and Elva A. Van Devender
Department of Chemistry, University of Virginia, PO Box 400319, Charlottesville, Virginia 22904
jam5x@virginia.edu
Received August 22, 2001
A convergent stereoselective synthesis of (-)-deoxypukalide is described. This substance has not
yet been found in Nature but is obtained through deoxygenation of pukalide, the first naturally
occurring furanocembrane to be structurally elucidated. The route features a new intraannular
furan synthesis that entails treatment of a 4-oxopropargylic â-keto ester with silica gel. The product
of this novel reaction, a 3-carboxy 2,5-bridged furan, is formed in 96% yield. The synthetic strategy
was strongly directed by molecular mechanics calculations, which provided valuable insight into
stereodefining steps including double bond stereochemistry and butenolide configuration.
Furanocembranolides are marine natural products
isolated from various soft coral species in tropical and
temperate waters throughout the world.1 These com-
pounds have received relatively little attention, in part,
because of their limited availability. Studies to date
indicate that certain members of the family exhibit potent
neurotoxicity, antiinflammatory, and antifeedant activ-
ity.2
The furanocembranolides may be divided into two
classes: those in which the substituent at the C4 position
is a CH3 and those with a more highly oxidized CHO or
CO2Me C4 substituent. Rubifolide3 is a representative of
the former class, whereas the latter more populated class
includes pukalide,4 acerosolide,5 and lophotoxin.6
The few recorded syntheses of furanocembranolides
have employed two differing solutions to the challenging
problem of constructing the strained furanocyclic struc-
ture that typifies these substances. In the first a 2,3,5-
trisubstituted furan ring is constructed, the appropriate
appendages at C2 and C5 are elaborated, and the
macrocycle is closed. This strategy suffers from two
problems: (1) the relatively sensitive furan moiety is
introduced early in the sequence, which limits the range
of conditions that can be employed for subsequent steps,
and (2) the bond forming step in the cyclization must be
sufficiently energetic to overcome the unfavorable en-
thalpic and entropic factors attending formation of the
F igu r e 1. Representative furanocembranolides.
strained ring. Paquette and Astles utilized this approach
in their synthesis of acerosolide (eq 1).7
(1) Marshall, J . A. Recent Res. Dev. Org. Chem. 1997, 1, 1.
(2) (a) Wright, A. E.; Burres, N. S.; Schulte, G. K. Tetrahedron Lett.
1989, 3491. (b) Abramson, S. N.; Trischman, J . A.; Tapiolas, F. M.;
Harold, E. E.; Fenical, W.; Taylor, P. J . Med. Chem. 1991, 34, 1798.
(c) Groebe, D. R.; Dumm, J . M.; Abramson, S. N. J . Bio. Chem. 1994,
269, 8885. (d) Hyde, E. G.; Boyer, A.; Tang, P.; Xu, Y.; Abramson, S.
N. J . Med. Chem. 1995, 38, 2231. (e) Hyde, E. G.; Thornhill, S. M.;
Boyer, A.; Abramson, S. N. J . Med. Chem. 1995, 38, 4704. (f) Look, S.
A.; Burch, M. T.; Fenical, W.; Qi-tai, Z.; Clardy, J . J . Org. Chem. 1985,
50, 5741. (g) Fenical, W. J . Nat. Prod. 1987, 50, 1001.
(3) Williams, D.; Anderson, R. J .; Van Duyne, G. D.; Clardy, J . J .
Org. Chem. 1987, 52, 332.
(4) Missakian, M. G.; Burreson, B. J .; Scheuer, P. J . Tetrahedron
1975, 31, 2513.
(5) Chan, W. R.; Tinto, W. F.; Laydoo, R. S.; Manchand, P. S.;
Reynolds, W. F.; McLean, S. J . Org. Chem. 1991, 56, 1773.
(6) Fenical, W.; Okuda, R. K.; Bandurraga, M.; Culver, P.; J acobs,
R. S. Science 1981, 212, 1512.
An alternative solution to bridged furan formation
entails macrocyclization of an intermediate in which the
furan ring is absent but elements of its eventual forma-
tion are present. Following cyclization, these elements
are modified to enable intraannular furan formation to
be effected. This approach offers two advantages: (1) the
(7) Paquette, L. A.; Astles, P. C. J . Org. Chem. 1993, 58, 165.
10.1021/jo016048s CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/03/2001

8038 J . Org. Chem., Vol. 66, No. 24, 2001
Marshall and Van Devender
F igu r e 3. Possible strategies for the synthesis of 2,3,5-
trisubstituted furans related to furanocembranolides.
F igu r e 2. Calculated energy differences (Macromodel 5.5)
between (E)- and (Z)-deoxypukalide (the enantiomer is de-
picted). Note the significant twist angle a/b/c/d of the vinyl-
furan moiety in the (E) isomer.
enthalpic energy cost of macrocyclization is considerably
diminished, and (2) the resonance energy associated with
furan formation partially or fully offsets the resultant
increase in strain-energy. Our previous synthesis of
rubifolide embodied this “furan-last” strategy.8
F igu r e 4. Alternative strategy for the synthesis of 3-carboxy-
2,5-disubstituted furans.
for rubifolide (Figure 3). However, we harbored grave
misgivings about the probable success of the furan-
forming allenone cyclization step in which the C3 methyl
substituent in C derived from A would be replaced by a
CO2Me group in D derived from B. Subsequent elimina-
tion of the OR substituent in furans C or D affords the
vinylfuran products. Two problems were envisioned in
the ester application: (1) unlike A, the precursor allenone
ester B would be highly susceptible to nucleophilic 1,4-
addition, and (2) the developing positive charge at the
CdO position in B would be destabilized by the electron-
withdrawing CO2Me substituent. For these reasons we
sought alternative methodology more amenable to the
preparation of 3-carboxy-2,5-disubstituted furans.
A simple solution to this problem was arrived at by
“rearranging” the allenone keto ester structure to an
R-propargylic â-keto ester G or H (Figure 4). The pres-
ence of a leaving group at the 4-position of the propargylic
substituent, as in G, should favor the SN2′ heterocycliza-
tion as illustrated. In fact, Wipf and co-workers have
effected such a conversion.10 To further facilitate the
cyclization process, we introduced a carbonyl substituent
as in H for our “leaving group”, thereby profiting from
the favorable energetics of 1,4-additions and considerably
expanding the range of conditions that might be em-
ployed for the cyclization. Subsequent enolization toward
the furan ring should be favored by conjugation and the
(Z) double bond geometry would be favored in consider-
ation of the strain noted in Figure 2. The feasibility of
this approach was demonstrated with simple acyclic
â-keto ester ynones,11 but it remained to be tested in an
intramolecular context.
The target of the present studies is deoxypukalide. This
furanocembranolide, not yet found in Nature, can be
prepared from pukalide through treatment with Zn and
EtOH (eq 3).9 The double bond stereochemistry was
established by Tius using NOE analysis. That the product
of this deoxygenation possesses a (Z)-double bond is not
surprising given the probable stepwise nature of the
deoxygenation reaction and the high strain-energy of the
(E) isomer (Figure 2).
Our projected synthesis of deoxypukalide parallels our
previous route to rubifolide in which (S)-(-)-perillyl
alcohol 1 served as the starting material.8 Through this
choice we would arrive at the enantiomer of deoxy-
pukalide derived from natural material.4 However, there
was no compelling reason to favor the “natural” isomer,
In our projected synthesis of deoxypukalide we envi-
sioned a “furan-last” strategy similar to that employed
(8) Marshall, J . A.; Sehon, C. A. J . Org. Chem. 1997, 62, 4313.
(9) Tius, M. Department of Chemistry, University of Hawaii.
Unpublished results.
(10) Wipf, P.; Rahman, L. T.; Rector, S. R. J . Org. Chem. 1998, 63,
7132.
(11) Marshall, J . A.; Zou, D. Tetrahedron Lett. 2000, 41, 1347.

Synthesis of (-)-Deoxypukalide
J . Org. Chem., Vol. 66, No. 24, 2001 8039
Sch em e 1
Sch em e 3
Sch em e 2
centers are relatively modest. Acetal hydrolysis and
subsequent protection of the secondary alcohol 15 led to
aldehyde 16. The â-keto ester 18 was efficiently prepared
in one step through SnCl2-promoted homologation with
tert-butyl diazoacetate (17).15 We initially conducted this
step with ethyl diazoacetate, which yielded the ethyl
â-keto ester. However, this derivative was less robust
than the tert-butyl analogue and a number of the
subsequent steps proceeded in lower yield. We also had
misgivings about the ultimate conversion of the ethyl
ester to its methyl analogue, although this was not a
major consideration. Methyl diazoacetate, if it were
commercially available, would presumably suffer the
same drawbacks encountered with the ethyl ester. As will
be seen, the tert-butyl ester proved highly satisfactory.
Ensuing selective cleavage of the PMB ether 18 with
DDQ at pH 7 and subsequent iodide formation yielded
the macrocyclization precursor, â-keto ester 20. The
cyclization was effected (Scheme 3) in 83% yield upon
treatment of 20 with 0.9 equiv of KO-t-Bu in THF at -78
°C at moderately high dilution (0.005 M). The TBS ether
in the resulting cyclic ynone 21 was slowly, but selec-
tively, cleaved to alcohol 22 by treatment with PPTS in
EtOH at room temperature over a period of 10 days.
Increasing the temperature of this reaction led to con-
siderable byproduct formation at the expense of 22.
Exposure of the Dess-Martin oxidation16 product,
ynone 23, to silica gel in hexanes gave the carboxy furan
24 in 96% yield for the two steps! As already stated,
molecular mechanics calculations17 clearly showed that
a cis double bond at the 7,8-position, as in 26, would be
highly favored over the corresponding trans isomer (see
Figure 2). Treatment of ketone 24 with the Comins
pyridyl triflimide reagent 2518 yielded a 1:1 mixture of
diastereomeric enol triflates 26, which afforded a com-
so the choice was made on a purely economic basis.12 The
starting sequence, depicted in Scheme 1, is an improved
version of our previous route to the alkynyl acetal 6 in
which the Ohira methodology13 for alkynylation (4 f 6)
replaces the previous Corey-Fuchs methodology.14 Ad-
dition of the lithioalkyne 7 to aldehyde 8 followed by
protecting group adjustments afforded alcohol 11 in high
overall yield.
The derived aldehyde 12 was converted to alcohol 14
through addition of the lithio derivative of propargylic
ether 13 (Scheme 2). Although alcohol 14 and the derived
intermediates 15-20 are most likely mixtures of four
diastereoisomers, the 1H NMR spectra of these com-
pounds were remarkably free of extraneous peaks, sug-
gesting that interactions between the various stereo-
(15) Holmquist, C. R.; Roskamp, E. J . J . Org. Chem. 1989, 54, 3258.
(16) Dess, D. B.; Martin, J . C. J . Org. Chem. 1983, 48, 4155. We
prepared the Dess-Martin periodinane reagent used in this study from
a modified procedure that utilizes oxone: Frigerio, M.; Santagostino,
M.; Sputore, S. J . Org. Chem. 1999, 64, 4537.
(12) Current prices for (R)- and (S)-perillyl alcohol from the 2000-
2001 Aldrich catalogue are $33.60/g and $1.59/g.
(13) Ohira, S. Synth. Commun. 1989, 19, 561.
(14) Corey, E. J .; Fuchs, P. L. Tetrahedron Lett. 1972, 3769.

8040 J . Org. Chem., Vol. 66, No. 24, 2001
Marshall and Van Devender
Sch em e 4
F igu r e 5. Conformational basis for the stereoselective reduc-
tion of ynone 29. The methyl ester was employed for the
calculations.
parable mixture of methylated products 27 in high yield
through Pd-catalyzed coupling with Me2Zn. Cleavage of
the propargylic DPS ether 27 with TBAF and Dess-
Martin periodinane oxidation of the propargylic alcohol
28 afforded ketone 29 as a single stereoisomer. The
stereochemistry of the double bond was confirmed by
NOE studies.
F igu r e 6. Calculated energies (Macromodel 5.5) of diastere-
omeric allenic acids.
Installation of the embedded butenolide moiety was
effected along the lines previously employed in our
rubifolide synthesis.8 Thus reduction of ketone 29 with
K-Selectride gave rise to propargylic alcohol cis-28, as a
single diastereomer (Scheme 4). The basis for this
selectivity can be seen in the calculated structure, shown
in Figure 5, in which the C8 substituent adopts a nearly
perfect Felkin-Anh orientation19 with respect to the C10
carbonyl.
Conversion of alcohol 28 to the trifluoroacetate 30 and
in situ Pd-catalyzed carbohydroxylation yielded the al-
lenoic acid 31, which without purification was stirred
with 10% AgNO3 on silica gel in hexanes to afford the
butenolide 32 in 58% yield for the three steps. At this
point our choice of a tert-butyl ester proved remarkably
prescient. Pyrolysis in a sealed glass tube at 210 °C for
20 min followed by treatment of the acid pyrolysate 33
with TMSCHN2 afforded crystalline deoxypukalide 34 in
92% yield. The relative stereochemistry of this interme-
diate was confirmed through single-crystal X-ray struc-
ture analysis. Unfortunately, we have been unable to
secure spectral data for the authentic material as the
F igu r e 7. Equilibration of the intermediate allenylpalladium
stereoisomers derived from trifluoroacetates 36.
degradation studies were never published and the origi-
nal spectra have been either discarded or misplaced.9
A slight variation on the route to butenolide starting
from the 1:1 mixture of diastereomeric propargylic alco-
hols 35 was also pursued (Figure 7). These alcohols were
synthesized by a route parallel to that described for the
tert-butyl ester 28 starting from the â-keto ethyl ester
analogue of 18, mentioned previously.20 The lower yield
in several of the steps in the ethyl ester route precipitated
our switch to the more efficent tert-butyl route. The
impetus for the study came from molecular mechanics
(17) The program Macromodel 5.5 was employed for these calcula-
tions. Global minimum multiple conformer searching was achieved
with the Monte Carlo option in CSRCH (Conformer Search) employing
the MM2 force field in conjunction with the Truncated Newton
Conjugate Gradient (TNCG) minimization protocal through multiple
step interations (typically 1000 or more) until the minimum energy
conformer was found multiple times. For a description of the program,
see: (a) Mohamadi, F.; Richards, N. G. J .; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J .
Comput. Chem. 1990, 11, 440. (b) Chang, G.; Guida, W. C.; Still, W.
C. J . Am. Chem. Soc. 1989, 111, 4379.
(18) Comins, D. L.; Dehghani, A. Tetrahedron Lett. 1992, 6299.
(19) Cherest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1971,
379. Anh, N. T. Top. Curr. Chem. 1980, 88, 145.
(20) The synthesis of this is described in: Van Devender, E. A. Ph.D.
Dissertation, University of Virginia, 2002.

Synthesis of (-)-Deoxypukalide
J . Org. Chem., Vol. 66, No. 24, 2001 8041
calculations,17 which showed that allenic acid 40 was
favored by more than 1 kcal/mol over the diastereomer
(Figure 6). We assumed that the allenylpalladium pre-
cursor 38 to this acid would be similarly favored over 37.
Acid 40 is the precursor of butenolide 39. Our previous
studies indicated that conversion of enantioenriched
propargylic mesylates to allenic esters through Pd-
catalyzed carbomethoxylation proceeds with varying
amounts of racemization.21 The percent racemization
increased with the mol % of catalyst in keeping with the
known behavior of chiral allenylpalladium intermedi-
ates.22 We also noted that allenic esters could be racem-
ized by treatment with Ph3P, the ligands employed in
the carbonylation chemistry. Thus the allenic acid itself
is formed in an isomerizing environment. Accordingly,
we felt there was a good chance that allene equilibration
could be induced and the result would favor the requisite
butenolide precursor. In fact when the three-step acyla-
tion, carbonylation, and allenic acid cyclization sequence
was performed on the 1:1 mixture of alcohols 35 (trans-
35 and cis-35 in Figure 7) with a full equivalent of Pd
catalyst, butenolide 39 was obtained as the sole product.
Furthermore, when the pure alcohol cis-35 was subjected
to the three-step sequence with a full equivalent of
catalyst, butenolide 39 was again exclusively formed.
These results strongly suggest that equilibration of the
transient allenylpalladium intermediates 37 and 38 leads
to a predominance of 38, as surmised from our calcula-
tions. However, the yields of these reactions were in the
20-30% range. Thus it is possible that small amounts
of a minor isomer could have escaped detection. It is also
possible that a minor isomer, if present, could have
undergone decomposition during the course of the equili-
bration. Despite repeated efforts, we were unable to
separate the mixture of cis and trans alcohols 35 to
secure a sample of pure trans isomer for a more rigorous
test of the equilibration hypothesis. Given the low ef-
ficiency of the equilibration sequence, we did not pursue
the matter further.
The foregoing synthesis of (-)-deoxypukalide is notable
for its efficiency and high degree of stereocontrol, despite
the use of nonselective reactions that led to intermediates
consisting of up to eight diastereomers. The ultimate
success of the route can be attributed to favorable
conformational and steric factors that facilitated highly
controlled introduction of the key C10 stereocenter and
the C7 double bond.
Ack n ow led gm en t. We thank Dr. Michal Sabat for
his heroic efforts in solving the difficult X-ray structure
of our synthetic ent-deoxypukalide. This research was
supported by Grant R01-GM29475 from the National
Institute for General Medical Sciences and a grant from
the National Science Foundation (CHE-9901319). We
gratefully acknowledge the NIH for 20 continuous years
of support for our research on cembrane and cembra-
nolide synthesis. The synthesis reported in this article
marks the termination of this project. We also acknowl-
edge the kind cooperation of Marcus Tius for permission
to cite his unpublished findings on the structure of
deoxypukalide.
Su p p or tin g In for m a tion Ava ila ble: Complete experi-
mental details for the preparation of all intermediates in the
sequence leading to ent-deoxypukalide, selected 1H NMR
spectra, and an ORTEP diagram and structure parameters
for synthetic ent-deoxypukalide 34. This material is available
free of charge via the Internet at http://pubs.acs.org.
(21) Marshall, J . A.; Wolf, M. A.; Wallace, E. M. J . Org. Chem. 1997,
62, 367.
(22) Racemization of allenyl/propargyl palladium intermediates has
been studied previously. See ref 21 and (a) Elsevier, C. J .; Mooiweer,
H. H.; Kleijn, H.; Vermeer, P. Tetrahedron Lett. 1984, 5571. (b)
Elsevier, C. J .; Vermeer, P. J . Org. Chem. 1985, 50, 3042. (c) Elsevier,
C. J .; Kleijn, H.; Boersma, J .; Vermeer, P. Organometallics 1986, 5,
716.
J O016048S