Franklinolidesa A - C from an Australian marine sponge complex: Phosphodiesters strongly enhance polyketide cytotoxicity

Article abstract of DOI:10.1002/anie.201005883

Rare discovery: Three novel polyketide phosphodiesters, franklinolidesa A-C, were identified as the cytotoxic components from an Australian marine sponge complex. Preliminary structure-activity relationship studies, using in vitro cytotoxicity and cell proliferation assays, demonstrated that the introduction of the phosphodiester moiety significantly improved the cytotoxicity by 30 to >300 fold. Copyright

Full text of DOI:10.1002/anie.201005883

Communications
DOI: 10.1002/anie.201005883
Cytotoxic Polyketides
Franklinolides A–C from an Australian Marine Sponge Complex:
Phosphodiesters Strongly Enhance Polyketide Cytotoxicity**
Hua Zhang, Melissa M. Conte, and Robert J. Capon*
During investigations into cytotoxic metabolites as potential
anticancer agents from Australian marine invertebrates and
algae, our attention was drawn to a sponge sample (CMB-
01989) collected during deepwater (À105 m) scientific trawl-
ing operations in the Great Australian Bight. The aqueous
EtOH extract of CMB-01989, a massive Geodia sp. thinly
encrusted with a Halichondria sp., returned a nBuOH
partition that displayed noteworthy cytotoxicity against
human colon (HT-29), prostate (DU145), ovary (JAM and
C180-13S), and lung (A549) cancer cell lines (GI50 < 5–
10 mgmLÀ1), with slightly lower potency against breast
(MDA-MB-231) and skin (SK-MEL-128) cancer cell lines
(GI50 ꢀ 30 mgmLÀ1). Fractionation of the aqueous EtOH
extract yielded franklinolides A–C (1–3) as unprecedented
polyketide phosphodiesters that are prone to esterification,
isomerization, and hydrolysis on exposure to protic solvents
and/or elevated temperatures in acidic media. Complete
absolute configurations were assigned to the franklinolides
based on detailed spectroscopic analyses, chemical derivati-
zation and degradation, and comparison to natural and
synthetic model compounds. What follows is an account of
the isolation, characterization, and structure elucidation of 1–
3, together with an assessment of chemical stability and
cytotoxic properties.
(ca. 20 mm) against rat normal fibroblast 3Y1 cells.[1] The
structure and absolute stereochemistry of bitungolide A (4;
and by inference the co-metabolites 5–7) was secured by
single-crystal X-ray analysis.[1]
The HRESI(À) mass spectrum for franklinolide A (1)
displayed a highest mass ion cluster (m/z 629/631) measuring
for an anion of composition C29H39ClO11P (Dmmu + 1.2), and
consistent with a C4H6O6P adduct to one of the isomeric
bitungolides 4–7. Supportive of a close structure relationship
between franklinolides and bitungolides, the NMR
A portion of the aqueous EtOH extract from CMB-01989
was subjected to solvent partitioning and sequential tritura-
tion, followed by gel chromatography and normal (silica) and
reverse phase (C8 and C18) SPE and HPLC, to yield
franklinolides A–C (1–3). Examination of the NMR
([D4]MeOH) data for 1–3 revealed a high degree of similarity
with data reported for the sponge metabolites bitungoli-
des A–D (4–7). First described in 2002 from an Indonesian
sponge Theonella cf. Swinhoei, the bitungolides 4–7 are
chlorophenol polyketides that weakly inhibit dual-specificity
phosphatase VHR, and exhibit modest cytotoxicity
([D4]MeOH) data for
1 (see Supporting Information
Table S1 and S2 and Figure S1 and S2, acid form) were very
similar to those for 4 (Table S4 and Figure S4),[1] with
diagnostic values for J12,13 and J14,15 (11.1 and 11.1 Hz)
confirming a common 12Z,14Z double-bond geometry. Dif-
ferences in the NMR data between 1 and 4 were limited to
1) a deshielded chemical shift and broadened multiplicity for
H-9 (1 dH = 4.55, br; 4 dH = 3.73, ddd (10.0, 4.7, 2.3)) and
2) the appearance of new resonances in 1 consistent with a
deshielded oxymethine (dH = 4.89, br, H-2’) coupled to a
methoxymethylene (dH = 3.83, br, H2-3’; dH = 3.40, s, 3’-
OCH3). Further interpretation of the NMR data for 1 were
complicated by broadening of key resonances.
Concerned that the initially purified 1 was either a
mixture of co-eluting or equilibrating isomers, and/or was
undergoing structure modification on storage, we undertook
to carefully reanalyze 1. An NMR sample of 1 (stored in
[D4]MeOH at RT for several days) was examined by HPLC-
[*] Dr. H. Zhang, M. M. Conte, Prof. R. J. Capon
Division of Chemistry and Structural Biology
Institute for Molecular Bioscience, The University of Queensland
St. Lucia, Queensland 4072 (Australia)
Fax: (+61)7-3346-2090
E-mail: r.capon@imb.uq.edu.au
[**] We thank CSIRO Marine Research and the crew of the RV Franklin
for assistance in sponge collection, L. Goudie (Museum Victoria)
for sponge taxonomy, C. Cuevas and colleagues (PharmaMar) for
preliminary in vitro anticancer screening, and A. M. Piggott (UQ) for
the acquisition of HRESI(Æ)MS data. This work was funded partially
by the Australian Research Council, with additional support from
PharmaMar (Madrid, Spain).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201005883.
9904
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 9904 –9906

Angewandte
Chemie
ESI(À)MS to reveal a chromatogram analyzing for ca. 45% 1
(m/z 629/631), but featuring two additional later-eluting
components, 1a (m/z 646/648, ca. 40%) and 4 (m/z 447/449,
ca. 15%)—neither of which were apparent in the originally
isolated sample. Significantly, over time the abundance of
these later-eluting components continued to increase, sug-
gesting that under these mild storage conditions 1 was
chemically unstable. Semipreparative HPLC fractionation
returned pure samples of 1, 1a, and 4. Spectroscopic analysis
of 4 proved that it was identical in all respects (UV/Vis, MS,
NMR, [a]D) to bitungolide A,[1] and its formation was
attributed to hydrolysis of 1. In contrast, while NMR analysis
for 1a (Table S1a and Figure S1a) showed almost identical
data with those for 1 (with reduced broadening), we were
perplexed by the MS data for 1a (17 amu heavier than 1). The
solution to this problem only became evident when attempts
the absolute stereochemistry for bitungolide A (4) had
previously been solved by X-ray spectroscopy,[1] the structure
and absolute configuration for franklinolide A (1) were
assigned as shown.
To further explore chemical stability, samples of 1 (100 mg)
were exposed for 5 days to aliquots of different solvents and
modifiers (100 mL; Figure 1) to reveal that esterification,
isomerization, and hydrolysis of 1 were promoted to varying
degrees by exposure to protic solvents and elevated temper-
atures in acidic media.
to re-isolate
1
from the crude extract using
a
H2O:MeOH:TFA gradient (TFA = trifluoroacetic acid),
rather than the H2O:MeCN:TFA gradient HPLC method
employed initially (the practical consequence of a worldwide
shortage of MeCN), yielded 1 and 1b (ESI(À)MS m/z 643/
645) instead of 1a. Taken together these observations
suggested that 1a and 1b were the deuteromethyl and
methyl derivatives of 1, respectively. Moreover, giving that
1 was stable to storage in the crude extract, in aqueous EtOH
at À308C over a period of 15 + years, the appearance of
hydrolysis (4) and methylation (1a and 1b) products during
fractionation were attributed to the use of TFA, MeOH, and
elevated temperatures during handling.
Alert to chemical instability at low pH values, but also
acknowledging that TFA was an essential modifier to achieve
HPLC fractionation, we established optimal conditions for
franklinolide purification. These conditions consisted of
gradient C8 MeCN:H2O (0.01% TFA) HPLC followed by
immediate postcolumn neutralization with aqueous NaHCO3
prior to in vacuo concentration and subsequent storage in
DMSO (dimethyl sulfoxide) at À308C. An unanticipated
bonus from these handling protocols, in particular post HPLC
neutralization, was a very significant decrease in the broad-
ening of the NMR signals. Applying this optimized protocol
facilitated purification and characterization of 1a (Table S1a
and Figure S1a) and 1b (Table S1b and Figure S1b), with the
latter exhibiting a HMBC correlation linking the ester methyl
protons (dH = 3.76 ppm) to the C-1’ carbonyl carbon atom
(dC = 172.4 ppm), necessitating that the C4H6O6P adduct
“subunit” in 1 incorporated a carboxylic acid moiety.
Figure 1. Chemical stability studies of franklinolide A (1). The asterisk
(*) denotes double bond isomers of 4 (bitungolides B–D). HPLC
conditions: Zorbax Eclipse C8 analytical column, 1 mLminÀ1, 10–100%
MeCN/H2O (0.01% TFA) over 15 min, with diode array detection
(displayed at 270 nm).
HRESI(À)MS measurements on the minor metabolites
franklinolides B (2) and C (3) confirmed them to be isomeric
to 1 (C29H39ClO11P, Dmmu À0.5 and À2.6, respectively), while
comparison of their respective NMR data ([D4]MeOH,
Tables S2 and S3) revealed very good correlations with
bitungolides B (5) and D (7), respectively. Furthermore, and
consistent with the major co-metabolite 1, both 2 and 3
displayed spectroscopic signatures for the (S)-3-O-methyl-2-
phosphoglyceric acid phosphodiester moiety, and were
assigned the structures as shown.
The franklinolides are polyketides that very likely incor-
porate a 3-hydroxybenzoyl-CoA starter unit, and are further
elaborated through a phosphodiester linkage to the primary
metabolite 2-phosphoglyceric acid (Figure 2). Given the
stability studies described above it is plausible that frankli-
nolide A (1) is the sole direct polyketide biosynthesis product,
with the franklinolides B (2) and C (3) and the bitungoli-
The 31P NMR ([D4]MeOH) spectrum of 1 displayed a
resonance (dP = 0.3, dd (JP-H = 8.9, 8.3 Hz)) comparable with
that measured for an authentic sample of commercially
available (S)-2-phosphoglyceric acid disodium salt (8; dP =
0.8, d (JP-H = 9.5 Hz)). Pursuing this reasoning, a JP-H coupling
to H-2’ (8.3 Hz) and a broadening of the H-9 resonance in 1
relative to 4 was taken as evidence of a C-9 to C-2’
phosphodiester linkage, consistent with 1 being the 3-O-
methyl-2-phosphoglyceric acid phosphodiester adduct of 4 as
shown. To confirm this assignment a sample of 1 was
subjected to acid-catalyzed hydrolysis at 408C to yield 3-O-
methyl-2-phosphoglyceric acid (9). Further acidic hydrolysis
of 9 at 1008C yielded (S)-3-O-methylglyceric acid (10).[2] As
Angew. Chem. Int. Ed. 2010, 49, 9904 –9906
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9905

Communications
isomer 2, 3) a 30- to 50-fold decrease on isomerization of 1 to
the 12E,14Z isomer 3, and 4) a limited 1.5- to 2-fold decrease
on esterification of 1 to 1b. A similar pattern of cytotoxicity
(IC50) was also observed against a human brain (SH-SY5Y)
cancer cell line (Table 1).
In conclusion, this work describes the discovery of three
novel compounds with a rare polyketide skeleton incorporat-
ing an unusual 3-O-methylglyceric acid phosphodiester
moiety. To the best of our knowledge the closest natural
phosphodiester analogues to the franklinolides are Strepto-
myces derived antibiotic phosphoglycolipids (i.e. moenomy-
cin A,[3] pholipomycin,[4] and noskomycins[5]). The franklino-
lides are the first examples of polyketide phosphodiesters
and, most importantly, our work highlights their relatively
fragile nature and very high cytotoxicity compared to the
bitungolides. These observations encourage speculation that
glyceric acid phosphodiesters may have broader application
in enhancing the potency and therefore therapeutic value of
other cytotoxic agents.
Figure 2. Plausible franklinolide biosynthesis.
des A–D (4–7) capable of being induced through non-
enzymatic isomerization and hydrolysis.
Preliminary structure–activity relationship (SAR) studies,
using in vitro cytotoxicity and cell proliferation assays against
stomach (AGS) and colon (HT-29) cancer cell lines, and a
noncancerous control cell line (HFF-1; Table 1), confirmed 1
Table 1: In vitro cytotoxicity[a] data (mm) for compounds 1–4.
No.
HFF-1
AGS
HT-29
SH-SY5Y
Experimental Section
1
GI50
TGI
IC50
GI50
TGI
IC50
GI50
TGI
IC50
GI50
TGI
IC50
GI50
TGI
IC50
0.2
1.0
0.3Æ0.2
1.5Æ0.8
1.1Æ0.1
0.5Æ0.4
3.0Æ1.4
1.6Æ0.4
0.7Æ0.2
6.0Æ5.3
8.0Æ3.5
9.3Æ4.9
>30
0.1Æ0.1
0.8Æ0.3
1.8Æ1.0
0.2Æ0.1
1.0Æ0.6
1.4Æ0.7
0.5Æ0.2
2.7Æ1.5
5.7Æ1.5
4.7Æ4.0
ꢀ20
Initial cytotoxicity screening on crude extract, general experimental
procedures, animal materials collection and taxonomy, extraction and
isolation of franklinolides A–C (1–3), chemical stability studies of 1,
and acid-catalyzed methanolysis and hydrolysis of 1, as well as
tabulated NMR data for compounds 1–4, 1a and 1b, full NMR
spectra for compound 1, and 1H NMR spectra for compounds 2–4, 1a,
1b, 9, and 10a are provided in the Supporting Information.
2.5Æ0.7
1.1Æ0.1
4.4Æ0.8
8.4Æ3.3
>30
1b
2
0.3
0.7
2.0Æ0.0
1.5
3.9
16Æ6
3
10
Received: September 20, 2010
Published online: November 16, 2010
23
>30
>30
>30
>30
>30
>30
4
9.7Æ1.5
>30
>30
>30
Keywords: cytotoxicity · franklinolides · natural products ·
.
25Æ3
phosphodiesters · polyketides
ꢀ16
>30
[a] HFF-1=human foreskin fibroblast; AGS=gastric adenocarcinoma
(stomach); HT29=colorectal adenocarcinoma (colon); SH-SY5Y neuro-
blastoma (brain). GI50 and TGI values were acquired from cell
proliferation assay; IC50 values were acquired from MTT cytotoxicity
assay. All compounds were tested at least twice in duplicate except in the
cell proliferation assay for HFF-1 cell line (once in duplicate).
[1] S. Sirirath, J. Tanaka, I. I. Ohtani, T. Ichiba, R. Rachmat, K. Ueda,
T. Usui, H. Osada, T. Higa, J. Nat. Prod. 2002, 65, 1820 – 1825.
[2] E. Chiellini, S. Faggioni, R. Solaro, J. Bioact. Compat. Polym.
1990, 5, 16 – 30.
[3] P. Welzel, F.-J. Witteler, D. Mꢀller, W. Riemer, Angew. Chem.
1981, 93, 130 – 131; Angew. Chem. Int. Ed. Engl. 1981, 20, 121 –
123.
[4] S. Takahashi, K. Serita, M. Arai, H. Seto, K. Furihata, N. Otake,
Tetrahedron Lett. 1983, 24, 499 – 502.
[5] R. Uchida, M. Iwatsuki, Y.-P. Kim, S. Omura, H. Tomoda, J.
Antibiot. 2010, 63, 157 – 163.
as the dominant cytotoxic agent (GI50 range from 0.1 to
0.3 mm). SAR analysis defined the relative importance of key
structural features: 1) a very significant 30- to > 300-fold
decrease in cytotoxicity following hydrolysis of 1 to 4, 2) a 2-
to 7-fold decrease on isomerization of 1 to the 12E,14E
9906 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 9904 –9906