Apratoxin E, a cytotoxic peptolide from a guamanian collection of the marine cyanobacterium Lyngbya bouillonii

Article abstract of DOI:10.1021/np700717s

A collection of the marine cyanobacterium Lyngbya bouillonii from Guam afforded apratoxin E (1), a new peptide-polyketide hybrid of the apratoxin class of cytotoxins. The planar structure of 1 was elucidated by NMR spectroscopic analysis and mass spectrometry. Configurational assignments of stereocenters in the peptide portion were made by chiral HPLC analysis of the acid hydrolysate. The relative configuration in the polyketide moiety was assigned by comparison of NMR data including proton-proton coupling constants with those of the known analogues. Apratoxin E (1) displayed strong cytotoxicity against several cancer cell lines derived from colon, cervix, and bone, ranging from 21 to 72 nM, suggesting that the α,β-unsaturation of the modified cysteine residue is not essential for apratoxin activity. The 5- to 15-fold reduced activity compared with apratoxin A (2) is attributed to the dehydration in the long-chain polyketide unit, which could affect the conformation of the molecule.

Full text of DOI:10.1021/np700717s

J. Nat. Prod. 2008, 71, 1113–1116
1113
Apratoxin E, a Cytotoxic Peptolide from a Guamanian Collection of the Marine
Cyanobacterium Lyngbya bouillonii
Susan Matthew,† Peter J. Schupp,‡ and Hendrik Luesch*,†
Department of Medicinal Chemistry, UniVersity of Florida, 1600 SW Archer Road, GainesVille, Florida 32610, and UniVersity of Guam Marine
Laboratory, UOG Station, Mangilao, Guam 96923
ReceiVed December 13, 2007
A collection of the marine cyanobacterium Lyngbya bouillonii from Guam afforded apratoxin E (1), a new
peptide-polyketide hybrid of the apratoxin class of cytotoxins. The planar structure of 1 was elucidated by NMR
spectroscopic analysis and mass spectrometry. Configurational assignments of stereocenters in the peptide portion were
made by chiral HPLC analysis of the acid hydrolysate. The relative configuration in the polyketide moiety was assigned
by comparison of NMR data including proton-proton coupling constants with those of the known analogues. Apratoxin
E (1) displayed strong cytotoxicity against several cancer cell lines derived from colon, cervix, and bone, ranging from
21 to 72 nM, suggesting that the R,ꢀ-unsaturation of the modified cysteine residue is not essential for apratoxin activity.
The 5- to 15-fold reduced activity compared with apratoxin A (2) is attributed to the dehydration in the long-chain
polyketide unit, which could affect the conformation of the molecule.
Cyanobacteria produce an array of bioactive secondary metabo-
lites with characteristic biosynthetic signatures, which include a
high degree of modified amino acids in a peptide-polyketide hybrid
backbone.1 Representative examples are the apratoxins, which are
of mixed biosynthetic origin and exhibit potent cancer cell growth
inhibitory activity by inducing G1 phase specific cell cycle arrest
and apoptosis.2 Apratoxins A-C (2-4) were isolated from the
remarkably prolific Lyngbya bouillonii collected in Guam and
Palau3,4 and apratoxin D from a variety of the same species
collected in Papua New Guinea.5 The apratoxins have become
attractive targets for synthetic efforts, culminating in several total
syntheses.6–8 In our search for natural apratoxin analogues aimed
at expanding our knowledge about structure-activity relationships
within this intriguing compound class, we have encountered a new
natural apratoxin from a variety of L. bouillonii from Guam. Here,
we report the isolation, structure determination, and evaluation of
the antiproliferative activity of the new apratoxin designated as
apratoxin E (1), which was the predominant apratoxin in L.
bouillonii inhabiting deeper water and which was not present in
the apratoxin A (2) producer that lives in shallower water at the
same collection site (Figure 1).
To determine if the structural diversity of apratoxins may also
be a function of collection depth, we carried out collections of the
apratoxin-producing variety of L. bouillonii from shallow depths
of 1-3 m and deeper waters of 12-14 m at Finger’s Reef, Apra
Harbor, Guam. The freeze-dried cyanobacterial collections were
extracted with a 1:1 solvent mixture of MeOH and EtOAc and crude
extracts fractionated by normal-phase chromatography. The most
cytotoxic fraction was subjected to reversed-phase HPLC to afford
apratoxin E (1) as an amorphous, colorless solid, along with only
trace amounts of apratoxin A (2) in the deep collection, and only
apratoxin A (2) from the shallow collection. Compound 1 and
lyngbyastatin 29 accounted for the majority of the extract’s growth-
inhibitory activity (Figure 1).
Figure 1. Comparison of HPLC profiles derived from two varieties
of L. bouillonii from Finger’s Reef, Apra Harbor (Guam), encoun-
tered at different depths. The shallow L. bouillonii was collected
at 1-3 m depth and the deep L. bouillonii at 12-14 m depth.
Extracts were separated by silica gel chromatography and the
cytotoxic fractions subjected to reversed-phase HPLC (see Experi-
mental Section). Apratoxin E (1) is unique to the deep L. bouillonii
(red trace) and also contained higher amounts of lyngbyastatin 2
compared with the shallow L. bouillonii (blue trace).
niscent of those of apratoxin A (2) and the presence of conformers
in a 3:2 ratio. Detailed 1H and 13C NMR analysis of 1 in conjunction
with COSY, TOCSY, ROESY, HSQC, and HMBC data revealed
a hybrid structure comprised of peptide (O-Me-Tyr, N-Me-Ile,
N-Me-Ala, and Pro) and polyketide fragments (C27-C31 and
C33-C45) that were similar to apratoxin A (2)3 (Table 1), yet
exhibited a lower degree of C-methylation and a different position
of the double bond.10 The apparent disparity between 1 and 2 was
the replacement of the olefinic methine and allylic methyl proton
signals of the R,ꢀ-unsaturated modified cysteine (moCys) spin
system (C27-C32) (δΗ-29 6.35, δΗ-30 5.25, and δΗ-32 1.96) by a
pair of methylene signals H2-28 and H2-29 resonating at δH 2.76/
2.49 and 2.21/1.67 (major conformer), respectively.10 Thus, in
compound 1, this unit is neither R,ꢀ-unsaturated nor methylated at
C28, in contrast to apratoxin A (2). NMR analysis further suggested
that the substituted C9 carboxylic acid unit (C33-C45) more closely
resembled the semisynthetic (E)-dehydroapratoxin A (5);4 both 1
and 5 lack the hydroxyl group at C35 present in 2 and possess a
double bond between C34 and C35.10 The allylic methyl group
(C44) was missing, so that compound 1 consisted of a 7-hydroxy-
5,8,8-trimethyl-2-nonenoic acid residue (C33-C45). The geometry
The molecular formula of 1 was established as C43H65N5O7S,
on the basis of an [M + H]+ peak in the HRESI/APCIMS at m/z
1
796.4683 (calcd for C43H66N5O7S, 796.4683). H NMR data in
CDCl3 showed the presence of an amide NH and two N-Me groups
indicative of a peptide, with characteristic chemical shifts remi-
* To whom correspondence should be addressed. Tel: (352) 273-7738.
Fax: (352) 273-7741. E-mail: luesch@cop.ufl.edu.
† University of Florida.
‡ University of Guam.
10.1021/np700717s CCC: $40.75
2008 American Chemical Society and American Society of Pharmacognosy
Published on Web 05/08/2008

1114 Journal of Natural Products, 2008, Vol. 71, No. 6
Notes
of the acid hydrolysate. The configuration at C30 is speculated to
be S based on biosynthetic grounds, since the other natural
apratoxins from Apra Harbor also occur with 30S configuration.3,4
The potent cytotoxicity of 1 (see below) furthermore required a
37S configuration since recent structure-activity relationship studies
for apratoxin analogues have shown that S configuration at C37 is
essential for activity and inversion would lead to complete loss of
activity.7 Consequently, and consistent with biosynthetic consid-
erations as well, the proposed absolute configuration is 37S,39S,
as in the other apratoxins from the same collection site.3,4
Apratoxin E (1) was tested for antiproliferative activity against
several human cancer cell lines including HT29 colon adenocar-
cinoma, HeLa cervical carcinoma, and U2OS osteosarcoma cells,
alongside with apratoxin A (2) and E-dehydroapratoxin A (5) as
well as other natural products with nanomolar activity (actinomycin
D, paclitaxel) for comparison. Compound 1 displayed lower IC50
values than its closest analogue, compound 5 (Table 2), yet was 5-
to 15-fold less active than apratoxin A (2). The reduced activity
compared with apratoxin A (2) was expected since the effect of
dehydration in the long-chain polyketide unit (2 vs 5) was
previously established.4 Interestingly, the cytotoxicity appears to
correlate with the proportion of the trans amide isomer [in NMR
solvent CDCl3, apratoxins A (2) and C (4), >90%; apratoxin B
(3), ∼75%; apratoxin E (1), ∼60%; E-dehydroapratoxin A (5),
∼40%], which may or may not have biological relevance, as the
aqueous solution structure could be different.4 Furthermore, the
retention of significant activity despite modification of the moCys
unit unique to apratoxin E (1) indicates that the conjugated carbonyl
system is not required for activity, which excludes a mechanism
of action that involves a Michael addition on this residue. The
discovery of this new analogue provides an opportunity to expand
the structure-activity relationship in this series of compounds
derived both naturally and produced synthetically.4,7
As in the case of the apratoxins, cyanobacteria oftentimes produce
more than one member of a certain structural class, which is
partially due to relaxed specificity of biosynthetic enzymes that can
be exploited for precursor-directed biosynthesis.11 Particularly,
polyketide synthases of prokaryotes can show considerable substrate
tolerance,12 but also substrate binding pockets of nonribosomal
peptide synthetase adenylation domains may accept different amino
of the double bond could not be readily deduced from NMR data
in CDCl3 because of complete overlap of 1H NMR signals for H-34
and H-35 for the major conformer; yet the signal for H-34 of the
minor conformer (δH 6.54) appeared as a doublet with a 15.6 Hz
coupling, suggestive of E configuration. To confirm, NMR data
were recorded in benzene-d6 (Table S1, see Supporting Informa-
tion); the conformational ratio changed to 7:3, and signals for H-34
and H-35 were separated for both conformers. The coupling constant
between both protons could then be measured in signals of both
conformers (15.7 Hz), proving the hypothesized E configuration.
The sequence between the spin systems leading to the cyclic
structure for 1 was established by HMBC analysis in CDCl3 and
benzene-d6 (Table 1 and Table S1). Furthermore, strong ROESY
correlations between H-28a of the moCys unit (δ 2.76/2.52 in
CDCl3) and the NH of the O-Me-Tyr residue (δ 5.89/6.18 in CDCl3)
for both conformers confirmed the assignments.
acids within a group of polar or nonpolar amino acids,13
a
phenomenon that commonly leads to replacement of valine with
isoleucine in a given cyanobacterial structure, for example.14 Natural
structural diversification may also be due to slight genetic variation
or a slightly changed environment. For example, comparison of
apratoxin A (2) and E (1) structures suggests that two C-methylation
domains are inactivated in the biosynthetic gene cluster from the
apratoxin E (1) producer or that acetate is incorporated during chain
elongation instead of propionate. Conversely, dehydratase and enoyl
reductase activity are putatively responsible for the observed
unsaturation in the polyketide section of 1, and an additional enoyl
reductase, which then appears only functional in the apratoxin E
(1) producer, could account for the saturation of the moCys unit;15
however, this remains to be shown in biosynthetic studies. Many
filamentous cyanobacteria including L. bouillonii also grow in
patches, and each patch may be different genetically, or at least
chemically, as a result of differential gene transcription in response
to different environmental stimuli. Even within one homogeneous-
appearing collection made in a confined area of Apra Harbor, we
found heterogeneity based on 16S rDNA sequences.4 Some of these
circumstances may be the reason that we have discovered apratoxin
E (1) and encountered only traces of apratoxin A (2) from this
collection; we had found the latter in virtually every sample of this
organism obtained since 1991 at this site,3,16 however, at slightly
shallower depths.
1
Close similarity of H and 13C NMR shifts and JH,H values of 1
and with E-dehydroapratoxin A (5) were indicative of the same relative
configuration in the polyketide chain. Furthermore, both compounds
exist as conformers (3:2) in CDCl3 (Table 1) due to restricted rotation
around the O-Me-Tyr-N-Me-Ala amide bond.4 However, in contrast
to compound 5, the trans isomer was preferred in apratoxin E (1) since
a ROESY cross-peak between the R-protons H-14 and H-18 was
observed for the minor (cis) but not the major conformer. 1D TOCSY
experiments assisted in deducing the coupling constants of those
complex multiplets that otherwise were obscured due to extensive
signal overlap for both conformers (Table 1).
The absolute configurations of all the amino acid units in the
peptide portion of 1 were established as L by chiral HPLC analysis

Notes
Journal of Natural Products, 2008, Vol. 71, No. 6 1115

1116 Journal of Natural Products, 2008, Vol. 71, No. 6
Notes
not those of N-Me-D-Ala (8.8), D-Pro (21.0), N-Me-L-allo-Ile (24.0), N-Me-
D-Ile (34.5), N-Me-D-allo-Ile (34.6) (solvent 95:5, flow rate 0.75 mL/min).
The HPLC profile of the hydrolysate (solvent 90:10, flow rate 1.0 mL/
min) also showed additional peaks corresponding to O-Me-L-Tyr and L-Tyr
(arising from O-demethylation under the reaction conditions) at 105 and
26.2 min, respectively, while a peak for D-Tyr (27.8) was not detected in
the hydrolysate.
Cell Culture. Cell culture medium was purchased from Invitrogen
and fetal bovine serum (FBS) from Hyclone. Cells were propagated
and maintained in DMEM medium (high glucose) supplemented with
10% FBS at 37 °C humidified air and 5% CO2.
Table 2. Antiproliferative Activity Data for Various Apratoxins
and Other Antineoplastic Natural Products
IC50 in nM
(HT29)
IC50 in nM
(HeLa)
IC50 in nM
(U2OS)
apratoxin E (1)
apratoxin A (2)
E-dehydroapratoxin A (5)
actinomycin D
paclitaxel
21
1.4
41
2.9
1.9
72
10
121
0.4
1.5
59
10
160
3.2
24
Cell Viability Assays. Cells were plated in 96-well plates (HeLa, 3000
cells; U2OS, 5000 cells; HT29, 10 000 cells) and 24 h later treated with
various concentrations of compound 1, other apratoxins, actinomycin D,
paclitaxel, or solvent control (1% EtOH for apratoxins or 1% DMSO for
other drugs). After 48 h of incubation, cell viability was measured using
MTT according to the manufacturer’s instructions (Promega).
Experimental Section
General Experimental Procedures. Optical rotations were measured
on a Perkin-Elmer 341 polarimeter. IR data were recorded on a Bruker
Vector 22 IR spectrometer. UV spectra were recorded on a SpectraMax
M5 (Molecular Devices). 1H, 13C, and 2D NMR spectra were recorded
in CDCl3 and benzene-d6 on a Bruker Avance II 600 MHz spectrometer
equipped with a 1 mm triple-resonance high-temperature supercon-
ducting cryogenic probe using residual solvent signals (δH 7.26, δC
Acknowledgment. Funding was provided by the James & Esther King
Biomedical Research Program, Grant No. 06-NIR07 (to H.L.). The authors
gratefully acknowledge the NSF for funding through the External User
Program of the National High Magnetic Field Laboratory, which supported
our NMR studies at the Advanced Magnetic Resonance Imaging and
Spectroscopy (AMRIS) facility in the McKnight Brain Institute of the
University of Florida. We thank J. R. Rocca for assistance in acquiring
NMR data. The 600 MHz 1 mm triple-resonance HTS cryogenic probe
was developed through collaboration between the University of Florida,
NHMFL, and Bruker Biospin. Mass spectral analyses were performed at
the UCR Mass Spectrometry Facility, Department of Chemistry, University
of California at Riverside. P.J.S. acknowledges NIH MBRS SCORE grant
SO6-GM-44796-16a for support. This is University of Guam Marine
Laboratory contribution number 614.
77.0 for chloroform; δH 7.16, δC 128.0 for benzene) as internal
1
standards. HSQC and HMBC experiments were optimized for JCH
)
n
145 and JCH ) 7 Hz, respectively. Mixing times of 60 and 500 ms,
respectively, were used to acquire TOCSY and ROESY data. HRMS
data was obtained using an Agilent LC-TOF mass spectrometer
equipped with an APCI/ESI multimode ion source detector.
Biological Material. Lyngbya bouillonii strain PS372 was collected
by snorkeling from 12 to 14 m depth at Finger’s Reef, Apra Harbor,
Guam, during June 2006 and May 2007. Strain PS372 was found only
deeper (12-14 m depth) under overhangs or other partially shaded
reef structures, unlike the previously described L. bouillonii from
shallow reef parts, which was growing between the branches of
scleractinian corals such as Porites sp. and Acropora sp. Strain PS372
was mainly found overgrowing the green alga Halimeda sp. and
occasionally crustose coralline algae. Unlike L. bouillonii from shallow
parts of Finger’s Reef (1-3 m), no shrimps were found to be associated
with the deep strain PS372. Microscopic evaluation of the voucher
material confirmed that size and morphology of the filaments and cells
were consistent with L. bouillonii described by Hoffmann and Demoulin
in 1991.17 Cell size of L. bouillonii strain PS372 ranged from 6 to 13
µm length and 20 to 24 µm width with a cell length-to-width ratio
ranging from 0.27 to 0.53. A voucher specimen is retained at the
University of Guam Marine Laboratory.
Supporting Information Available: 1D and 2D NMR spectra in
CDCl3 and in benzene-d6; NMR data in benzene-d6 (Table S1). This
material is available free of charge via the Internet at http://pubs.acs.org.
References and Notes
(1) Tan, L. T. Phytochemistry 2007, 68, 954–979.
(2) Luesch, H.; Chanda, S. K.; Raya, R. M.; DeJesus, P. D.; Orth, A. P.;
Walker, J. R.; Izpisu´a Belmonte, J. C.; Schultz, P. G. Nat. Chem. Biol.
2006, 2, 158–167.
(3) Luesch, H.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J.; Corbett, T. H.
J. Am. Chem. Soc. 2001, 123, 5418–5423.
(4) Luesch, H.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J. Bioorg. Med.
Chem. 2002, 10, 1973–1978.
(5) Gerwick, W. H., presented at the US-Japan Symposium on Marine
Natural Products, Park City, UT, July 2007.
(6) Chen, J.; Forsyth, C. J. J. Am. Chem. Soc. 2003, 125, 8734–8735.
(7) Ma, D.; Zou, B.; Cai, G.; Hu, X.; Liu, J. O. Chem.-Eur. J. 2006, 12,
7615–7626.
(8) Doi, T.; Numajiri, Y.; Munakata, A.; Takahashi, T. Org. Lett. 2006,
8, 531–534.
Extraction and Isolation. The freeze-dried organism was extracted
twice with EtOAc-MeOH (1:1) to afford 6.92 and 1.89 g of extracts,
respectively. The combined extract was subjected to flash chromatog-
raphy over silica gel, eluting with CH2Cl2, followed by increasing
amounts of i-PrOH in CH2Cl2, and finally with MeOH. The mixtures
that eluted with 6% i-PrOH (122 mg) and 8% i-PrOH (88 mg) were
further subjected to semipreparative reversed-phase HPLC (Phenomenex
Ultracarb, ODS 250 × 10 mm, 5 µm, 3.0 mL/min; UV detection at
220, 240 nm) using an isocratic system of 80% aqueous MeCN for 30
min; 80-100% MeCN for 30-40 min; and 100% MeCN for 40-60
min to afford apratoxin E (1) (5.1 mg, tR 34.0 min).
(9) Luesch, H.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J. J. Nat. Prod.
1999, 62, 1702–1706.
Apratoxin E (1): colorless, amorphous solid; [R]20 -69 (c 0.12,
(10) The numbering system for compounds 2-5 was adapted to allow direct
D
comparison of NMR data.
MeOH); UV (MeOH) λmax (log ꢀ) 220 (4.28), 270 (3.34) nm; IR (film)
(11) Magarvey, N. A.; Beck, Z. Q.; Golakoti, T.; Ding, Y.; Huber, U.;
Hemscheidt, T. K.; Abelson, D.; Moore, R. E.; Sherman, D. H. ACS
Chem. Biol. 2006, 1, 766–779.
(12) For a recent example, see: Watts, R. E.; Tse, M. L.; Khosla, C.
Biochemistry 2007, 46, 3385–3393.
ν
max 2956, 2923, 2868, 1740, 1642, 1441, 1247, 1175 cm-1; 1H NMR,
13C NMR, COSY, and HMBC data, see Table 1 (in CDCl3) and Table
S1 (benzene-d6); HRESI/APCIMS m/z [M + H]+ 796.4699 (calcd for
C43H66N5O7S, 796.4683).
Acid Hydrolysis and Chiral HPLC Amino Acid Analysis. A
sample of compound 1 (0.3 mg) was treated with 6 N HCl (0.5 mL) at
110 °C for 18 h. The hydrolysate was concentrated to dryness, resuspended
in H2O (100 µL), filtered, and analyzed by chiral HPLC for amino acids
[column, Phenomenex Chirex phase 3126 (D) (4.6 × 250 mm); solvent,
2 mM CuSO4-MeCN (95:5 or 90:10); detection at 254 nm]. The retention
times (tR, min) of the amino acids in the hydrolysate of apratoxin E (1)
matched those of N-Me-L-Ala (8.4), L-Pro (10.5), N-Me-L-Ile (25.4), but
(13) Challis, G. L.; Ravel, J.; Townsend, C. A. Chem. Biol. 2000, 7, 211–224.
(14) Hoffmann, D.; Hevel, J. M.; Moore, R. E.; Moore, B. S. Gene 2003,
311, 171–180.
(15) For a review on polyketide biosynthesis, see, e.g.: Hopwood, D. A.
Chem. ReV. 1997, 97, 2465–2497, or other reports in the same issue.
(16) Luesch, H. Ph.D. Thesis, University of Hawaii, 2002.
(17) Hoffmann, L.; Demoulin, V. Belg. J. Bot. 1991, 124, 82–88.
NP700717S