Liphagal, a selective inhibitor of PI3 kinase α isolated from the sponge Aka coralliphaga: Structure elucidation and biomimetic synthesis

Article abstract of DOI:10.1021/ol052744t

Liphagal (1), a selective inhibitor of PI3K α, has been isolated from the marine sponge Aka coralliphaga collected in Dominica. The "liphagane" meroterpenoid carbon skeleton of liphagal (1) is new. A biomimetic total synthesis has been used to confirm the

Full text of DOI:10.1021/ol052744t

ORGANIC
LETTERS
2006
Vol. 8, No. 2
321-324
Liphagal, a Selective Inhibitor of PI3
Kinase Isolated from the Sponge Aka
r
coralliphaga: Structure Elucidation and
Biomimetic Synthesis
Frederic Marion,† David E. Williams,† Brian O. Patrick,† Irwin Hollander,‡
Robert Mallon,‡ Steven C. Kim,‡ Deborah M. Roll,‡ Larry Feldberg,‡
Rob Van Soest,§ and Raymond J. Andersen*,†
Departments of Chemistry & EOS, UniVersity of British Columbia, 2036 Main Mall,
VancouVer, British Columbia, Canada V6T 1Z1, Institute for Systematics and Ecology,
UniVersity of Amsterdam, 1090 GT Amsterdam, The Netherlands, and Wyeth Research,
401 North Middletown Road, Pearl RiVer, New York 10965
randersn@interchange.ubc.ca
Received November 12, 2005
ABSTRACT
Liphagal (1), a selective inhibitor of PI3K
r, has been isolated from the marine sponge Aka coralliphaga collected in Dominica. The “liphagane”
meroterpenoid carbon skeleton of liphagal (1) is new. A biomimetic total synthesis has been used to confirm the constitution of liphagal (1)
and support a proposed biogenesis.
The phosphatidylinositol-3-kinase (PI3K) signaling pathway
plays a central role in regulating cell proliferation, cell
survival, adhesion, movement, differentiation, membrane
trafficking, glucose transport, neurite outgrowth, and super-
oxide production.1-3 A key second messenger in this pathway
is phosphatidylinostitol-3,4,5-trisphosphate (PI-3,4,5-P3), which
is present at low concentrations in unstimulated cells but is
rapidly synthesized from PI-4,5-P2 by PI3K in response to a
wide array of extracellular stimuli. To tightly regulate PI3K
signaling, cells degrade PI-3,4,5-P3 to PI-4,5-P2 with the
tumor suppressor PTEN and to PI-3,4-P2 with SHIP, sSHIP,
or SHIP2.4
† University of British Columbia.
‡ Wyeth Research.
§ University of Amsterdam.
(1) Ward, S. G.; Sotsoios, Y.; Dowden, J.; Bruce, I.; Finan, P. Chem.
Biol. 2003, 10, 207-213.
(2) Ward, S. G.; Finan, P. Curr. Opin. Pharmacol. 2003, 3, 426-
434.
(3) Wymann, M. P.; Zvelebil, M.; Laffargue, M. Trends Pharmacol. Sci.
2003, 24, 366-376.
There are several closely related PI3K isoforms that are
thought to have different biological activities. Isoform
10.1021/ol052744t CCC: $33.50
© 2006 American Chemical Society
Published on Web 12/18/2005

selective drugs capable of attenuating PI3K signaling should
have significant therapeutic potential for the treatment of
inflammatory and autoimmune disorders as well as cancer
and cardiovascular diseases.1-4 Wortmannin (2) and LY294002
(3), a synthetic analogue of the flavanoid quercetin (4), are
two first-generation PI3K inhibitors that have been widely
employed as chemical genetics probes to elucidate the
biological roles of PI3K signaling. Second-generation iso-
form-selective PI3K inhibitors, whose structures appear to
be based on quercetin and LY294002, have been described
in the patent literature.5-7 It has also been reported that AS-
605240 and related thiazolidine-2,4-diones are selective
submicromolar inhibitors of PI3K γ.8
We have screened a library of marine invertebrate extracts
in a fluorescent polarization assay using human PI3K R
expressed in SF9 insect cells as part of a program designed
to find new isoform-selective PI3K inhibitors. The MeOH
extract of the sponge Aka coralliphaga collected in Dominica
showed promising activity in the assay. Bioassay-guided
fractionation of the extract identified liphagal (1), a mero-
terpenoid with an unprecedented carbon skeleton, as the only
PI3K inhibitory component. The details of the structure
elucidation, proposed biogenesis, biomimetic synthesis, and
biological activity of liphagal (1) are presented below.
were not correlated to carbon resonances in the HMQC
spectrum, were assigned to the OH protons.
Preliminary analysis of the 1H and 13C NMR spectra (Table
1, Supporting Information) revealed that liphagal (1) had both
substituted benzene and terpenoid fragments. One aliphatic
methyl doublet (δ 1.36, J ) 7.1 Hz, Me-21) and three
aliphatic methyl singlets (δ 0.91, Me-19; 0.94, Me-20; 1.28,
Me-22) were present in the 1H NMR spectrum. Resonances
that could be assigned to a pentasubstituted benzene ring (δ
107.9, C-14; 114.9, C-17; 119.2, C-12; 140.8, C-16; 145.8,
C-13; 147.2, C-15), a tetrasubstituted olefin (δ 124.4, C-10;
155.2, C-9), and a conjugated aldehyde (δ 189.7, C-18) were
observed in the 13C NMR spectrum. The benzene, olefin,
and aldehyde functionalities accounted for six of the nine
required sites of unsaturation, and the remaining three sites
had to be rings. Structural features of 1 identified from the
NMR data were similar to the structural components of
siphonodictyal B (5),9 previously isolated from specimens
of A. coralliphaga collected in Belize, suggesting that the
compounds were related.
A proton resonance at δ 7.43 (H-17) showed strong
HMBC correlations to carbon resonances at δ 140.8 (C-16),
147.2 (C-15), and 145.8 (C-13) and a weak correlation (4
bond) to 107.9 (C-14), and the aldehyde proton resonance
at δ 10.40 (H-18) showed correlations to carbon resonances
at δ 107.9 (C-14), 147.2 (C-15), and 145.8 (C-13). This set
of HMBC correlations supported the presence of a penta-
subsituted benzene ring in 1 having the same substitution
pattern found in the aromatic ring of siphonodictyal B (5).
COSY correlations identified the spin system extending from
the methylene protons at C-1 (δ 1.38 and 2.47) through to
the methylene protons at C-3 (δ 1.21 and 1.43), and starting
from H-5 (δ 1.50) through to H-8 (δ 3.16) and Me-21 (δ
1.36). HMBC correlations observed between proton reso-
nances at δ 0.91 (Me-19), 0.94 (Me-20), and 1.28 (Me-22)
and a carbon resonance at δ 53.4 (C-5); between the proton
resonances at δ 0.91 (Me-19) and 0.94 (Me-20) and the
carbon resonance at δ 41.3 (C-3); between proton resonances
at δ 1.50 (H-5) and 1.78 (H-6R) and a carbon resonance at
38.9 (C-11); between a proton resonance at δ 1.50 (H-5)
and carbon resonances at δ 39.7 (C-1) and 19.9 (C-22); and
between the proton resonance at δ 1.52 (H-6â) and the carbon
resonance at δ 34.4 (C-4) confirmed the presence of a six-
membered A ring with dimethyl substitution at C-4 and
monomethyl substitution at C-11.
Specimens of A. coralliphaga were harvested by hand
using scuba on reefs in Prince Rupert Bay, Portsmouth,
Dominica. Freshly collected sponge was frozen on site and
transported to Vancouver over dry ice. Thawed sponge was
extracted exhaustively with MeOH, and the combined MeOH
extracts were concentrated in vacuo to give an aqueous
suspension that was partitioned between H2O and EtOAc.
The H2O-soluble material, which showed strong PI3K
inhibition, was partitioned between BuOH and H2O. Frac-
tionation of the enzyme-inhibitory BuOH-soluble portion by
sequential application of Sephadex LH20 chromatography
(eluent: MeOH) and reversed-phase HPLC (eluent: 17:3
MeOH/H2O) gave a pure sample of liphagal (1).
Liphagal (1) was obtained as an optically active ([R]25
D
+12.0 (c 3.7, MeOH)) amorphous yellow solid that gave a
[M]+ ion in the HREIMS at m/z 356.1994 appropriate for a
molecular formula of C22H28O4 (calcd 356.1988), requiring
nine sites of unsaturation. The 13C NMR spectrum (Sup-
porting Information) obtained for 1 contained resonances that
could be assigned to 22 carbon atoms consistent with the
HRMS data. HMQC data showed that only 26 of the
hydrogen atoms were attached to carbon (4 × CH3, 5 × CH2,
4 × CH, 9 × C), requiring the presence of 2 OHs. Broad
The H-8 (δ 3.16) and Me-22 (1.28) resonances showed
HMBC correlations to the C-10 (δ 124.4) resonance, and
the H-7R (δ 2.12), H-8 (3.16), and Me-21 (1.36) resonances
showed correlations to the C-9 (δ 155.2) resonance, dem-
onstrating the presence of a seven-membered ring B with a
tetrasubstituted ∆9,10 olefin. An HMBC correlation observed
between the H-17 (δ 7.43) and C-10 (δ 124.4) resonances
indicated that the benzene ring carbon C-12 was bonded to
C-10. Connecting the oxygenated benzene carbon C-13 (δ
1
singlets at δ 9.29 and 10.28 in the H NMR spectrum, that
(4) Yang, L.; Williams, D. E.; Mui, A.; Ong, C.; Krystal, G.; van Soest,
R.; Andersen, R. J. Org. Lett. 2005, 7, 1073-1076.
(5) Hayakaa, M.; Kaizawa, H.; Kawaguchi, K. I.; Matsuda, K.; Ishikawa,
N.; Koizumi, T.; Yamanao, M.; Ohta, M. U.S. Patent US 6403588, 2002.
(6) Robertson, A. J.; Jackson, S.; Kenche, V.; Yaip, C. Parbaharan, H.;
Thompson, P. Int. Patent Appl. WO 0181346 A2, 2001.
(7) Sadhu, C.; Masinovsky, B.; Dick, K.; Sowell, G. C.; Staunton, D. E.
J. Immunol. 2003, 170, 2647-2654.
(8) Camps, M.; Ruckel, T.; Ji, H.; Ardissone, V.; Rintelen, F.; Shaw, J.;
Ferrandi, C.; Chabert, C.; Gillieron, C.; Francon, B.; Martin, T.; Gretener,
D.; Perrin, D.; Leroy, D.; Vitte, P.-A.; Hirsch, E.; Wymann, M. P.; Cirillo,
R.; Schwarz, M. K.; Rommel, C. Nature Med. 2005, 11, 936-943.
(9) a) Sullivan, B. W.; Faulkner, J. J. Org. Chem. 1986, 51, 4568-4573,
and b) Sullivan, B.; Djura, P.; McIntyre, D.; Faulkner, D. J. Tetrahedron
1981, 37, 979-982.
(10) Inoue, M.; Carson, M. W.; Frontier, A. J.; Danishefsky, S. J. J.
Am. Chem. Soc. 2001, 123, 1878-1889.
322
Org. Lett., Vol. 8, No. 2, 2006

145.8) and the olefinic carbon C-9 (δ 155.2) through an ether
linkage to give a furan completed the final ring required by
the unsaturation number of liphagal (1) and generated a
constitution that was consistent with all of the spectroscopic
data.
1D NOESY data provided evidence for the relative
stereochemistry of liphagal (1). Correlations observed be-
tween Me-22 (δ 1.28) and H-1eq (2.47), H-2ax (1.68), Me-
19 (0.91), and H-6â (1.52) when Me-22 was irradiated;
between Me-21 (δ 1.36) and H-5 (δ 1.50) when Me-21 was
irradiated; and between H-3ax (δ 1.21) and both H-1ax (δ
1.38) and H-5 (δ 1.50) when H-3ax was irradiated were only
possible if ring A adopted a chair conformation, the 6,7 ring
junction was trans, and Me-21 was R as shown.
Liphagal (1) inhibited PI3K R with an IC50 of 100 nM in
the primary fluorescent polarization enzyme assay. In the
same assay, wortmannin (2) had an IC50 of 12 nM and
LY294002 (3) had an IC50 of 550 nM. Liphagal (1) was
approximately 10-fold more potent against PI3K R than
against PI3K γ. Secondary in vitro cell assays showed that
liphagal (1) was cytotoxic to LoVo (human colon: IC50 0.58
µM), CaCo (human colon: IC50 0.67 µM), and MDA-468
(human breast: IC50 1.58 µM) tumor cell lines.
biogenesis of siphonodictyal B (5). Siphonodictyal B (5)
could be converted to liphagal (1) via the epoxide III
followed by ring expansion to ketone IV, epimerization at
C-8 and hemiketal formation to give V, and dehydration. In
pathway A, it is the 2′ carbon of the prenyl residue that acts
as a nucleophilic center in the polyene cyclization step.
Pathway B outlines a more direct route to liphagal (1) that
starts with the conversion of I to the ketone VII, perhaps
via the epoxide VI. The ketone VII should spontaneously
form the hemiketal VIII, which upon dehydration yields the
benzofuran IX. Proton-initiated cyclization of IX leads
directly to liphagal (1). Pathway B suggests that the
preorganization of a benzofuran fragment represents a way
to create nucleophilicity at the 1′ carbon of the prenyl side
chain (see IX, Scheme 1), thereby committing the polyene
cyclization to direct formation of the 6,7 ring system.
There has been significant interest in the synthesis of
frondosin B (6).10-12 The methodologies used to construct
the fused 6,7 ring system in frondosin B (6) were not readily
applicable to the synthesis of liphagal (1). We were intrigued
by the possibility that cation-initiated cyclization reactions
involving trapping by a benzofuran as shown in biogenetic
pathway B (Scheme 1) might represent an efficient synthetic
approach to liphagal (1).
Liphagal (1) represents the first example of the “liphagane”
meroterpenoid carbon skeleton. Scheme 1 shows two possible
The biomimetic synthesis of racemic liphagal (1) started
with preparation of the desired cyclization precursor 16
(Scheme 2). Commercially available 2,4,5-trimethoxybenz-
aldehyde 7 was selectively demethylated at the 2 position
Scheme 1. Proposed Biogenetic Pathways to Liphagal (1)
Scheme 2. Synthesis of the Biomimetic Cyclization Precursor
16
biogenetic routes to liphagal. Pathway A involves a proton
initiated polyene cyclization of farnesylated trihydroxy-
benzaldehyde I with trapping of the intermediate carbocation
by the ∆2′3′ olefin to give II, a putative intermediate in the
Org. Lett., Vol. 8, No. 2, 2006
323

with BBr3, and the resulting phenol was brominated at C-3
to give 8. Direct reduction of the aldehyde 8 led to an
unstable diol,13 so it was necessary to use a protection
sequence to get the desired phosphonium salt 10. Protection
of the phenol in 8 with TBS, followed by reduction of the
aldehyde with sodium borohydride, produced the benzyl
alcohol 9. Treatment of 9 with triphenylphosphine/HBr
followed by removal of the TBS protecting group with HF/
pyridine complex in THF gave the phosphonium salt 10.
Scheme 3. Synthesis of Liphagal (1) from Intermediate 16
Synthesis of the isoprenoid fragment started with a Wittig
reaction on geranylacetone 11 to give the enol ether 12
(Scheme 2). Direct hydrolysis of the enol ether did not
proceed cleanly, so a two-step hydrolysis, first with PPTS
and MeOH to give the dimethoxy acetal and then with PPTS
in a mixture of acetone and H2O, was used to generate the
aldehyde 13. Oxidation with NaClO2 converted the aldehyde
13 into the corresponding acid 14. Coupling the phenol 10
with the acid 14 using DCC gave the ester 15,13 which was
not isolated. Deprotonation of the phosphonium salt 15 with
Et3N in refluxing THF brought about an intramolecular
Wittig reaction resulting in the formation of the desired
benzofuran 16.14
The key cyclization step was first effected by refluxing
16 in a biphasic system of formic acid and cyclohexane
(Scheme 3)15 for 14 days to give a racemic mixture 18 of
C-8 epimers as a minor product accompanying a mixture of
inseparable partially cyclized alkenes 17, suggesting that the
conversion of 16 to 18 proceeded in two steps. In support
of this suggestion, it was found that upon treatment with
formic acid in refluxing cyclohexane, compound 16 gave
within 2 h only a complex mixture of cyclohexenes 17. The
alkenes 17 reacted slowly under the same conditions to give
after one month the tetracyclic system 18 in 40% isolated
yield as a 1/1 mixture of C-8 epimers 18a (Me-21R) and
18b (Me-21â). To shorten the time required for the polyene
cyclization, benzofuran 16 was treated with chlorosulfonic
acid at -78 °C in nitropropane. Under these conditions,
cyclization was complete within 30 min yielding 18 in 43%
yield as a 2/5 mixture of the C-8 epimers 18a and 18b.
trans 6,7 ring junction and the Me-21 R configuration.
Compound 19b obtained from the HPLC purification crys-
tallized and X-ray diffraction analysis confirmed that it had
the Me-21 â configuration. Demethylation of 19a with BI3
followed by reductive workup with aqueous sodium thio-
sulfate cleanly generated racemic liphagal (1) that was
identical to the natural product by HPLC, 1H NMR, and 13
NMR analyses.
C
The biomimetic synthesis of liphagal (1) efficiently
assembled the tetracyclic skeleton of the natural product with
stereoselective formation of the trans fusion between the six-
and seven-membered rings. It provides strong chemical
support for the proposal that the biogenesis of liphagal might
involve the use of a benzofuran residue to guide a biological
cation-initiated polyene cyclization reaction toward the
formation of the fused 6,7 ring substructure. Liphagal (1) is
a more potent inhibitor of PI3K R than LY294002 (3), more
selective than wortmannin (2), and shows significant in vitro
cytotoxicity against a small panel of human tumor cell lines,
making it a promising lead structure for the development of
a new class of PI3K inhibitors.
The aldehyde functionality was introduced by treatment
of the isomeric mixture of bromobenzenes 18 with n-
butyllithium to give the corresponding phenyllithium mixture
that was condensed with DMF followed by hydrolysis. The
resulting aldehydes 19a and 19b were separated by normal-
phase HPLC to give the desired product 19a, which had the
Acknowledgment. Financial support was provided by a
NIH-NCDDG grant (CA 67786). F.M. was supported by a
Fellowship from Pierre Fabre.
(11) Hughes, C. C.; Trauner, D. Angew. Chem., Int. Ed. 2002, 41, 1569-
Supporting Information Available: Experimental de-
tails, NMR assignments for liphagal, NMR spectra of 1 and
synthetic internediates, and an ORTEP drawing for 19b. This
material is available free of charge via the Internet at
http://pubs.acs.org.
1572.
(12) Kerr, D. J.; Willis, A. C.; Flynn, B. L. Org. Lett. 2004, 6, 457-
460.
(13) Kraus, G. A.; Nguyen, T.; Bae, J.; Hostetter, J.; Steadham, E.
Tetrahedron 2004, 60, 4223-4225.
(14) Yuan, Y.; Men, H.; Lee, C. J. Am. Chem. Soc. 2004, 126, 14720-
14721.
(15) Hercouet, A.; Le Corre, M. Tetrahedron Lett. 1979, 20, 2145-2148.
OL052744T
324
Org. Lett., Vol. 8, No. 2, 2006