Evidence for the rapid conversion of stephacidin B into the electrophilic monomer avrainvillamide in cell culture

Article abstract of DOI:10.1021/ja0690971

We report that the dimeric alkaloid stephacidin B (1) and the monomeric alkaloid avrainvillamide (2) function equivalently within experimental error to inhibit the growth of four different cultured human cancer cell lines. We also show that the proportion

Full text of DOI:10.1021/ja0690971

Published on Web 03/31/2007
Evidence for the Rapid Conversion of Stephacidin B into the Electrophilic
Monomer Avrainvillamide in Cell Culture
Jeremy E. Wulff, Seth B. Herzon, Romain Siegrist, and Andrew G. Myers*
Department of Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138
Received December 19, 2006; E-mail: myers@chemistry.harvard.edu
Table 1. Measured GI50 Values for 1 and 2a
Here we report that the dimeric alkaloid stephacidin B (1) and
the monomeric alkaloid avrainvillamide (2) function equivalently
within experimental error to inhibit the growth of four different
cultured human cancer cell lines. We also show that the proportion
of the monomer greatly outweighs that of the dimer in the medium
of incubation, and that the half-life for the transformation of 1 to
2 is short relative to the half-life of cell division. Finally, using a
monomer-based affinity reagent, we provide evidence that the
monomer (2) interacts with intracellular thiol-containing proteins,
likely by covalent modification.
natural enantiomer
unnatural enantiomer
cell line
1
2
ent-1
ent-2
LNCaP
135 nM
241 nM
952 nM
1514 nM
(40-231) (159-323) (638-1266) (1243-1786)
âT-549
346 nM
621 nM
550 nM
786 nM
(321-372) (548-694) (488-611)
(717-855)
T-47D
91 nM 205 nM 942 nM
1485 nM
(30-152) (110-299) (583-1301) (1305-1665)
289 nM 406 nM 987 nM 1854 nM
(108-469) (206-607) (727-1247) (1750-1957)
MALME-3M
a Values in parentheses reflect the upper and lower limits for a 90%
confidence interval. The XLfit4 software package was used for curve-fitting
and estimation of confidence intervals.
Stephacidin B (1), isolated in 2001 from a culture broth of
Aspergillus ochraceus WC76466, was reported to inhibit the growth
of a testosterone-dependent prostate cancer cell line (LNCaP) at
nanomolar concentrations,1 while avrainvillamide (2), isolated in
2000 from a culture broth of Aspergillus sp. CNC358, was reported
to inhibit the growth of two breast cancer cell lines (âT-549 and
T-47D) as well as a metastatic malignant melanoma of the lung
(MALME-3M).2 It was recognized that stephacidin B (1) was
potentially the product of dimerization of avrainvillamide (2),1,3
and some support for this hypothesis was presented in 2005, when
two groups, having reported laboratory synthetic routes to 1 and 2,
described mild, albeit abiological conditions for their interconversion
(upon standing on silica gel or in the presence of triethylamine in
acetonitrile).4,5 To our knowledge, side-by-side comparisons of 1
and 2 in antiproliferative assays have not been reported, nor have
the rates of interconversion of 1 and 2 been established in a
physiologically relevant setting.
We first evaluated the growth-inhibitory properties of 1 and 2
in parallel, in the four cultured cancer cell lines for which potent
inhibition had been reported for either compound alone (Table 1).
In the same study, we measured the activities of the pure
enantiomers of 1 and 2 (ent-1 and ent-2, respectively). All four
compounds tested were synthesized by a stereodefined route.4 We
observed that 1 and 2 exhibited a range of growth-inhibitory
potencies (GI50 values) from 91-621 nM, depending on the cell
line (see Supporting Information). Notably, ent-1 and ent-2 were
hardly inactive, and in the âT-549 cell line in particular, the
unnatural enantiomers were only ∼2-fold less potent inhibitors of
growth. In all cell lines, and in both enantiomeric series, the dimeric
alkaloids ((+)- or (-)-1) were approximately equipotent to the
monomeric alkaloids ((-)- or (+)-2) in the corresponding series
when corrected for stoichiometry (i.e., solutions of 1 were twice
as potent as equimolar solutions of 2).
Figure 1. (A) Retrodimerization of (+)-stephacidin B (ent-1) to form (-)-
avrainvillamide (ent-2) in cell culture medium containing 6% DMSO at 23
°C; (B) structures and activities of simplified avrainvillamide analogues.
We followed the transformation of the dimeric alkaloid (+)-1
to the monomeric alkaloid (-)-2 in cell culture medium (10% fetal
bovine serum in RPMI 1640,6 containing 6% DMSO to solubilize
the alkaloids) by HPLC analysis (Figure 1A), and found that (+)-1
was converted to (-)-2 with a half-life of ∼50 min at 23 °C, and
∼10 min at 37 °C.7 Given these data, and the time necessary to
prepare solutions of the four compounds in cell culture medium,
serially dilute the solutions, and then add the diluted solutions to
the cells (∼3 h at 23 °C), it is unlikely that there was a substantial
fraction of the dimeric alkaloid present by the time drug effects
were first observed (morphological changes in the growing cells
were evident within hours at 37 °C; final measurements of cell
viability were performed after 48 h at 37 °C).
Further evidence that the growth-inhibitory activity of the dimeric
alkaloid stephacidin B (1) arises from the monomeric alkaloid
avrainvillamide (2) comes from our discovery that a number of
structurally simpler analogues, incapable of dimerization, exhibit
considerable potency in antiproliferative assays. Two representative
examples from several that we have identified are the 3-alkylidene-
3H-indole 1-oxides 3 and 4, both micromolar inhibitors of LNCaP
and T-47D cell lines (Figure 1B). Compound 3, first prepared in a
9
4898
J. AM. CHEM. SOC. 2007, 129, 4898-4899
10.1021/ja0690971 CCC: $37.00 © 2007 American Chemical Society

C O M M U N I C A T I O N S
LNCaP cell lysate with an IC50 value of ∼125 µM (see Supporting
Information). The inhibitory activity decreased with the addition
of glutathione disulfide, the natural substrate for GR, consistent
with active-site inhibition by 2. (+)-Avrainvillamide (2) was a poor
inhibitor of yeast GR.
Our results suggest that the antiproliferative activity of stepha-
cidin B (1) arises from its prior dissociation to form avrainvillamide
(2), which may then bind to one or more proteins with nucleophilic
cysteine residues, likely by covalent modification. The four specific
proteins identified here may or may not be important cellular targets
for 2 (the relatively weak inhibition of GR by 2 makes this protein
an unlikely candidate); further experiments are necessary. It is
interesting to note that while our early model studies showed that
both alcohols and thiols can add reversibly to the unsaturated nitrone
function of 2 and 3,4,8 thus far only cysteine-based nucleophilic
proteins have been identified as binding partners in our affinity
isolation studies.
Figure 2. Western-blot detection following affinity-isolations of four
cysteine-containing proteins with probe 5, in the absence and presence of
competitors (+)-avrainvillamide (2) and structural analogue 3.
Acknowledgment. We are indebted to Ross Tomaino for protein
sequencing and to the Bauer Center for Genomics Research for
the use of equipment. We gratefully acknowledge NSERC (J.E.W.),
NSF (S.B.H.), and SNSF (R.S.) for fellowship support. This work
was supported by a grant from the NIH.
synthetic model study,8 is comparable in activity in T-47D cells to
the unnatural enantiomer of avrainvillamide ((-)-2), and is only
one order of magnitude less potent than natural avrainvillamide
(+)-2.9 The data support the proposal that 2 is the biologically active
form of 1, but only insofar as it is true that 3 and 4 function by the
same mechanism as 1 and 2. Some evidence that 3 and 4 bind to,
and perhaps target, at least a subset of proteins that interact with 1
and 2 was gained from affinity-isolation experiments using the
reagent 5, synthesized (as a 1:1 mixture of diastereomers, and a
5.5:1 mixture of E/Z isomers; see Supporting Information) to
explore the hypothesis that the antiproliferative agents 1-4 might
covalently modify nucleophilic residues of one or more proteins
by addition to the electrophilic 3-alkylidene-3H-indole 1-oxide
function of these compounds.4
To identify potential binding proteins, an LNCaP cell lysate was
treated with the activity-based probe10,11 5 (Figure 2), in the presence
or absence of a 20-fold excess of compounds 2 or 3 as competitive
binders. Streptavidin-agarose was added, the resin was removed
by centrifugation, then was washed, and Laemmli buffer was added
to liberate any bound proteins. Following separation by SDS-PAGE,
protein bands were excised and sequenced by LC-MS/MS. Among
several candidate proteins identified in this manner (see Supporting
Information), four have thus far been confirmed by Western-blotting
(Figure 2).
The results of our study suggest that both (+)-avrainvillamide
(2) and the analogue 3 interact with a variety of cysteine-containing
proteins in the cell lysate, including heat-shock protein 60 (HSP60),
exportin 1 (XPO1), glutathione reductase (GR), and peroxiredoxin
1 (PRX1). GR12 and PRX113 are homodimeric oxidoreductase
enzymes, each with four active-site cysteine residues (distributed
between two active sites). HSP60 and XPO1 do not have cysteine
in their active sites, but both have accessible, nucleophilic cysteine
residues that are known to be alkylated by other electrophilic
inhibitors.14,15
Interestingly, while co-incubation with 2 or 3 led to a reduc-
tion in the amounts of HSP60 and XPO1 bound to probe 5 (com-
pare lane 3 of Figure 2 with lane 2, or compare lane 5 with lane 4)
the same conditions led to a marked increase in the binding of GR
and PRX1 to 5. This presumably relates to the fact that GR and
PRX1 have dual active sites, which may exhibit cooperative
binding. Protein binding to 5 was blocked in the presence of
iodoacetamide (10 mM), lending support to the hypothesis that 2-5
may function by alkylating cysteine residues.16 In enzymatic assays
(+)-avrainvillamide (2) reversibly inhibited human GR activity in
Supporting Information Available: Preparation of 3-6 and details
of affinity-isolations and enzymatic assays. This material is available
free of charge via the Internet at http://pubs.acs.org.
References
(1) (a) Qian-Cutrone, J.; Krampitz, K. D.; Shu, Y.-Z.; Chang, L.-P.; Lowe,
S. E. Stephacidin Antitumor Antibiotics. U.S. Patent 6,291,461, 2001.
(b) Qian-Cutrone, J.; Huang, S.; Shu, Y.-Z.; Vyas, D.; Fairchild, C.;
Menendez, A.; Krampitz, K.; Dalterio, R.; Klohr, S. E.; Gao, Q. J. Am.
Chem. Soc. 2002, 124, 14556-14557.
(2) (a) Fenical, W.; Jensen, P. R.; Cheng, X. C. Avrainvillamide, a Cytotoxic
Marine Natural Product, and Derivatives Thereof. U.S. Patent 6,066,635,
2000. (+)-Avrainvillamide (2) was independently isolated under the name
CJ-17,665: (b) Sugie, Y.; Hirai, H.; Inagaki, T.; Ishiguro, M.; Kim, Y.-
S.; Kojima, Y.; Sakakibara, T.; Sakemi, S.; Sugiura, A.; Suzuki, Y.;
Brennan, L.; Duignan, J.; Huang, L. H.; Sutcliffe, J.; Kojima, N. J. Antibiot.
2001, 54, 911-916.
(3) von Nussbaum, F. Angew. Chem., Int. Ed. 2003, 42, 3068-3071.
(4) Herzon, S. B.; Myers, A. G. J. Am. Chem. Soc. 2005, 127, 5342-5344.
(5) (a) Baran, P. S.; Guerrero, C. A.; Hafensteiner, B. D.; Ambhaikar, N. B.
Angew. Chem. Int. Ed. 2005, 44, 3892-3895. (b) Baran, P. S.; Hafen-
steiner, B. D.; Ambhaikar, N. B.; Guerrero, C. A.; Gallagher, J. D. J.
Am. Chem. Soc. 2006, 128, 8678-8693.
(6) Roswell Park Memorial Institute culture medium, series 1640, was
purchased from ATCC. For formulation, see: Moore, G. E.; Gerner, R.
E.; Franklin, H. A. J. Am. Med. Assoc. 1967, 199, 519-524.
(7) This experiment employed the unnatural enantiomers ((+)-1 and (-)-2),
which we had synthesized first.
(8) Myers, A. G.; Herzon, S. B. J. Am. Chem. Soc. 2003, 125, 12080-12081.
(9) Structural analogues were evaluated using a slightly different method than
that used for the data in Table 1 (see Supporting Information). In this
assay, the GI50 for (+)-2 was 330 nM vs. LNCaP and 420 nM vs. T-47D.
(10) Bogyo, M.; Baruch, A.; Jeffery, D. A.; Greenbaum, D.; Borodovsky, A.;
Ovaa, H.; Kessler, B. Curr. Protoc. Protein Sci. 2004, 21.17.1.
(11) (a) Liu, Y.; Patricelli, M. P.; Cravatt, B. F. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 14694-14699. (b) Kidd, D.; Liu, Y.; Cravatt, B. F. Biochemistry
2001, 40, 4005-4015.
(12) (a) Karplus, P. A.; Schulz, G. E. J. Mol. Biol. 1987, 195, 701-729. (b)
Becker, K.; Savvides, S. N.; Keese, M.; Schirmer, R. H.; Karplus, P. A.
Nat. Struct. Biol. 1998, 5, 267-271.
(13) Wood, Z. A.; Schro¨der, E.; Harris, J. R.; Poole, L. B. Trends Biochem.
Sci. 2003, 28, 32-40.
(14) (a) Nagumo, Y.; Kakeya, H.; Shoji, M.; Hayashi, Y.; Dohmae, N.; Osada,
H. Biochem. J. 2005, 387, 835-840. (b) Nagumo, Y.; Kakeya, H.;
Yamaguchi, J.; Uno, T.; Shoji, M.; Hayashi, Y.; Osada, H. Bioorg. Med.
Chem. Lett. 2004, 14, 4425-4429.
(15) Kudo, N.; Matsumori, N.; Taoka, H.; Fujiwara, D.; Schreiner, E. P.; Wolff,
B.; Yoshida, M.; Horinouchi, S. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
9112-9117.
(16) Control experiments (lanes 6-8 of Figure 2) with compounds lacking
either the biotin moiety (2 or 3) or the 3-alkylidene-3H-indole 1-oxide
function (6) did not afford detectable quantities of protein in subsequent
Western-blot analyses.
JA0690971
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4899