Synthesis of phakellistatin 13 and oxidation to phakellistatin 3 and isophakellistatin 3

Article abstract of DOI:10.1021/ol049614p

Matrix presented. The natural product phakellistatin 13 cyclo-(TrpProPheGlyProThrLeu) was synthesized. Photosensitized oxidation of phakellistatin 13 gave the natural products phakellistatin 3 and isophakellistatin 3, demonstrating for the first time that

Full text of DOI:10.1021/ol049614p

ORGANIC
LETTERS
2004
Vol. 6, No. 11
1713-1716
Synthesis of Phakellistatin 13 and
Oxidation to Phakellistatin 3 and
Isophakellistatin 3
Kevin L. Greenman, Dana M. Hach, and David L. Van Vranken*
Department of Chemistry, UniVersity of California, IrVine, California 92697-2025
dlVanVra@uci.edu
Received March 1, 2004
ABSTRACT
The natural product phakellistatin 13 cyclo-(TrpProPheGlyProThrLeu) was synthesized. Photosensitized oxidation of phakellistatin 13 gave the
natural products phakellistatin 3 and isophakellistatin 3, demonstrating for the first time that a tryptophan residue can be directly converted
to the corresponding 3a-hydroxypyrrolo[2,3-b]indoline in a full length peptide. Competitive oxidation of the indoline product was identified as
the cause of low mass balance and is probably the source of low mass balance in the oxidative cyclization of all tryptamine derivatives.
Four groups of cyclic peptide natural products containing
the 3a-hydroxypyrrolidino[2,3-b]indoline (Hpi) moiety have
been isolated to date. These groups include himastatin,1,2
phakellistatin 3 and isophakellistatin 3,3 kapakahines C and
D,4 omphalotins B, C, and D,5 and chloptosin.6 Phakellistatin
3 was shown to inhibit P388 cell growth (ED50 0.4 µM) while
the diastereomer, isophakellistatin 3, was inactive. The Hpi
residue presumably arises from oxidation of a Trp-containing
peptide as opposed to enzymatic incorporation of the Hpi
subunit. The recent isolation7 of phakellistatin 13 (1) supports
this hypothesis, wherein the Trp residue of phakellistatin 13
has been oxidized to Hpi in phakellistatin 3 (2) and
isophakellistatin 3 (3, Figure 1).4,5 However, there are no
known synthetic transformations of Trp to Hpi in full length
peptides.
Phakellistatin 13 exhibits potent, selective cytotoxicity. It
is active against the hepatoma cell line BEL-7404 (ED50
<
12 nM) but inactive against the leukemia cell line HL-60.
Since phakellistatin 13 is comprised of ribosomally compat-
ible amino acids it may be mimicking an endogenous protein.
The hydrophobicity of phakellistatin 13 (calculated C log P
) 4.49) and phakellistatin 3 (calculated C log P ) 3.77)
suggest that the target could be intracellular.
(1) Leet, J. E.; Schroeder, D. R.; Krishnan, B. S.; Matson, J. A. J. Antibiot.
1990, 43, 961-966.
The peptide sequence of phakellistatin 13 (cyclo-WPF-
GPTL) shares homology with several known human proteins.
The selective cytotoxicity of phakellistatin 13 may be due
to mimicry of one or more of these proteins, particularly
proteins involved in protein recognition. We were further
intrigued by the potential biogenetic relationship of the Hpi-
containing product phakellistatin 3. Thus, we set out to study
the oxidative cyclization of phakellistatin 13 to phakellistatin
3 and isophakellistatin 3 and elucidate the conformation of
these molecules in solution.
(2) Leet, J. E.; Schroeder, D. R.; Golik, J.; Matson, J. A.; Doyle, T. W.;
Lam, K. S.; Hill, S. E.; Lee, M. S.; Whitney, J. L.; Krishnan, B. S. J.
Antibiot. 1996, 49, 299-311
(3) Pettit, G. R.; Tan, R.; Herald, D. L.; Cerny, R. L.; Williams, M. D.
J. Org. Chem. 1994, 59, 1593-1595.
(4) Nakao, Y.; Yeung, B. K. S.; Yoshida, W. Y.; Scheuer, P. J.;
Kellyborges, M. J. Am. Chem. Soc. 1995, 117, 8271-8272.
(5) Buchel, E.; Martini, U.; Mayer, A.; Anke, H.; Sterner, O. Tetrahedron
1998, 54, 5345-5352.
(6) Umezawa, K.; Ikeda, Y.; Uchihata, Y.; Naganawa, H.; Kondo, S. J.
Org. Chem. 2000, 65, 459-463.
(7) Li, W. L.; Yi, Y. H.; Wu, H. M.; Xu, Q. Z.; Tang, H. F.; Zhou, D.
Z.; Lin, H. W.; Wang, Z. H. J. Nat. Prod. 2003, 66, 146-148.
10.1021/ol049614p CCC: $27.50 © 2004 American Chemical Society
Published on Web 04/29/2004

There are two principle synthetic methods for oxidative
cyclization of tryptophan to Hpi: (1) reagent-based mono-
oxygenation8,9 and (2) photooxidative cyclization followed
by peroxide reduction (Scheme 1).10-12 Both methods have
The highest yielding methods for Hpi formation involve
a two-step sequence of photosensitized oxidative cyclization
followed by peroxide reduction. For example, L-tryptophan
has been oxidatively cyclized in water at low temperature
using rose bengal to afford the Hpi derivative in 86% yield.11
Photooxidations are capricious, and widely varying condi-
tions have been reported for photooxidative cyclization of
tryptophan derivatives. The highest isolated yield reported
for the photooxidative cyclization of a tryptophan derivative,
NR-Boc-tryptophan, is 90% (before drying).12 Dipeptides
(Xxx-Trp)18 and diketopiperazines19-22 fare much worse,
giving Hpi derivatives in yields ranging from 11 to 39%. In
our hands, photooxidative cyclization of NR-Boc-tryptophan
on a small scale (100 mg) gave 50-60% yields of the Hpi
derivative.12 However, the photooxidative cyclization of NR-
acetyltryptophan methyl ester gave none of the desired Hpi
product on a 100 mg scale. However, by decreasing the
temperature from 0 to -40 °C, the desired Hpi derivative
was obtained in a 15% yield. We were unable to obtain better
results on less than 1 mmol scale with any published
conditions11 or through independent variation of reaction
parameters.
Scheme 1
been widely applied to the amino acid tryptophan and simple
derivatives of tryptophan but not to tryptophan residues in
peptides. Multistep methods exist for the stereoselective
synthesis of Hpi from suitably protected Trp derivatives, but
incorporation into peptides is laborious.9,13
To investigate the oxidation of Trp to Hpi in the context
of a full-length peptide, the synthesis of phakellistatin 13
was undertaken. Mindful of the fact that epimerization of
activated amino esters can be problematic during slow
couplings,23 we chose to cyclize a linear peptide with glycine
at the carboxy terminus. ProThr(O-t-Bu)LeuTrpProPheGly
(4) was synthesized on chlorotrityl resin from Fmoc-protected
amino acids using HBTU24 and cleaved from resin using 1:1:
98 TFA/EDT/CH2Cl2 (69%, after HPLC purification). The
peptide was then subjected to macrocyclic ring closure at 1
mM in CH3CN with HBTU and HOBt over 4 days to give
cyclo-ProThr(Ot-Bu)LeuTrpProPheGly (5) in 60% yield
(Scheme 2). The substrate GlyProThr(O-t-Bu)LeuTrpProPhe
(6), with an unhindered glycine at the N-terminus, cyclized
in 16% yield without epimerization. Final deprotection of
the Thr hydroxyl with 0.01:1:1 EDT/TFA/CHCl3 proceeded
in 94% yield to give synthetic phakellistatin 13 (1). The 1H
and 13C NMR spectra of 1 were in agreement with those
reported for the natural product except for a minor conformer
Conceptually, the most direct route to Hpi is oxidative
cyclization of a Trp residue with an electrophilic oxygen
atom source. Dimethyldioxirane (DMDO) is the most ef-
ficient oxygen atom donor for cyclization of suitably
protected tryptophan derivatives, but even under ideal
conditions it is unsatisfactory for oxidative cyclization of
Trp in full length peptides.9 The putative epoxide intermedi-
ates can rearrange to oxindolylalanine14 and other products
that are further susceptible to oxidation and oligomerization.
Additionally, the desired product, an indoline, is likely to
react with electrophilic oxidants faster than the indole starting
material.15,16 Not surprisingly, we found that oxidation of
NR-acetyltryptophan methyl ester with one equiv. DMDO
at -78 °C generates a complex mixture of products and
starting material; reversed-phase HPLC reveals at least 10
new products. In contrast, oxidation of tryptamines can be
efficient when the indole has a quaternary substituent at the
2-position that slows cyclization to the indoline.17
1
present (4%) as indicated by the H NMR, although no
mention of multiple conformers was made in the original
isolation paper. At 100 °C, the 1H NMR revealed one set of
signals. This work confirms that Trp1 has the natural L
configuration, which could not be previously assigned
(8) Savige, W. E. Aust. J. Chem. 1975, 28, 2275-2287.
(9) Kamenecka, T. M.; Danishefsky, S. J. Chem.-Eur. J. 2001, 7, 41-
63.
(10) Saito, I.; Matsuura, T.; Nakagawa, M.; Hino, T. Acc. Chem. Res.
1977, 10, 346-352.
(18) Anthoni, U.; Christophersen, C.; Nielsen, P. H.; Christophersen, M.
W.; Sørensen, D. Acta Chim. Scand. 1998, 52, 958-960.
(19) Plate, R.; Akkerman, M. A. J.; Ottenheijm, H. C. J.; Smits, J. M.
M. J. Chem. Soc., Perkin Trans. 1 1987, 2481-2490.
(20) Maes, C. M.; Potgieter, M.; Steyn, P. S. J. Chem. Soc., Perkin Trans.
1 1986, 861-866.
(21) Bock, M. G.; DiPardo, R. M.; Pitzenberger, S. M.; Homnick, C.
F.; Springer, J. P.; Freidinger, R. M. J. Org. Chem. 1987, 52, 1644-1646.
(22) Baran, P. S.; Guerrero, C. A.; Corey, E. J. J. Am. Chem. Soc. 2003,
125, 5628-5629.
(23) Samy, R.; Kim, H. Y.; Brady, M.; Toogood, P. L. J. Org. Chem.
1999, 64, 2711-2728.
(24) Abbreviations used: 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyl-
uronium hexafluorophosphate (HBTU), N-hydroxybenzotriazole (HOBt),
and 1,2-ethanedithiol (EDT).
(11) Nakagawa, M.; Kato, S.; Kataoka, S.; Kodato, S.; Watanabe, H.;
Okajima, H.; Hino, T.; Witkop, B. Chem. Pharm. Bull. 1981, 29, 1013-
1026.
(12) Droste, H.; Wieland, T. Liebigs Ann. Chem. 1987, 901-910.
(13) Ley, S. V.; Cleator, E.; Hewitt, P. R. Org. Biomol. Chem. 2003, 1,
3492-3494.
(14) Witkop, B. Justus Liebigs Ann. Chem. 1947, 558, 98-109.
(15) Ritchie, R.; Saxton, J. E. Tetrahedron 1981, 37, 4295-4303.
(16) Somei, M.; Yamada, F.; Kurauchi, T.; Nagahama, Y.; Hasegawa,
M.; Yamada, K.; Teranishi, S.; Saito, H.; Kaneko, C. Chem. Pharm. Bull.
2001, 49, 87-96.
(17) Schkeryantz, J. M.; Woo, J. C. G.; Siliphaivanh, P.; Depew, K. M.;
Danishefsky, S. J. J. Am. Chem. Soc. 1999, 121, 11964-11975.
1714
Org. Lett., Vol. 6, No. 11, 2004

Scheme 2
Figure 1. Superposition of (a) phakellistatin 3 (tube) with
isophakellistatin 3 (wire) and (b) phakellistatin 3 (tube) with
phakellistatin 13 (wire). Leu and Phe side chains were omitted for
clarity.
because the acidic hydrolysis used for amino acid analysis
destroyed the tryptophan residue.
indoline with tert-butyl hydroperoxide in the absence of light.
No reaction was observed over 1 h. Next, we subjected
indoline to our photooxidative conditions in methanol-d4.
Within 15 min, half of the indoline was consumed, giving
primarily indole. This result clearly indicates that competitive
photooxidation of the indoline products is a major reason
for low mass balance in the photooxidative cyclization of
tryptamine derivatives.19-22 The problem with photooxidation
is that it can generate two types of peroxide intermediates:
a dioxetane that is poised to cleave to an N-formylkynurenine
through a retro-[2 + 2] reaction or a 3-hydroperoxypyrro-
lidino[2,3-b]indoline11 (Scheme 1) that is highly susceptible
to further oxidation. To avoid this problem, one must use
substrates that generate hindered/unreactive products or run
the reaction to very low conversion.
Having prepared phakellistatin 13, we next turned our
attention toward the oxidation of the Trp residue. Using the
optimized photooxidation conditions, phakellistatin 13 was
oxidized to a 1:1 mixture of phakellistatin 3 and isophakel-
listatin 3 in a combined yield of 20% (Scheme 3). Under
Scheme 3
The solution conformation of phakellistatin 13 was
determined using a combination of ROESY and variable-
temperature 1H NMR. NOEs were observed from the Trp1-
HR to both the Phe3-NH and the Gly4-NH. The Gly4-NH
appeared as a triplet (J ) 3.7 Hz) suggesting a φ angle near
180°. Both of the Gly4 R-protons showed NOEs to the δ
methylene protons of the adjacent Pro5, consistent with a Z
amide conformation. In addition, the backbone NHs of Phe3,
Gly4 and Leu7 exhibited low-temperature dependence (∆δ
< 2 ppb/K) suggestive of intramolecular hydrogen bonding
or solvent inaccessibility, whereas the backbone NHs of Trp1
and Thr6 were relatively large (∆δ > 4 ppb/K) suggesting
exposure to solvent.
the best HPLC conditions, phakellistatin 3 and isophakel-
listatin 3 elute as overlapping sharp and broad peaks,
respectively. After multiple rounds of HPLC purification,
1
phakellistatin 3 could be obtained in >95% purity. The H
and 13C NMR of synthetic phakellistatin 3 in methanol-d4
matched the published spectra for the natural product. In
contrast, isophakellistatin 3 proved to be much more difficult
to separate from phakellistatin 3. In fact, the NMR spectrum
published for isophakellistatin 3 is actually contaminated with
20% of phakellistatin 3. After five rounds of HPLC purifica-
tion, an 85:15 mixture of isophakellistatin 3 and phakellistatin
An attempt was made to obtain a diastereomerically pure
sample of isophakellistatin 3 using the stereoselective cy-
clization reported for protected tryptophan derivatives using
N-phenylselenophthalimide.13,25,26 Unfortunately, after Boc-
protection of the tryptophan side chain of peptide 6 an
attempt to effect oxidative cyclization of the tryptophan
residue with N-phenylselenophthalimide led to a mixture of
over 10 products (Scheme 4).
1
3 was obtained, giving a H NMR spectrum that closely
matched that previously published. When the photooxidation
was applied to the Boc-protected indole 7 over 6 h, no
reaction was observed.
In an attempt to explain the poor mass balance, two NMR
experiments were conducted to test the sensitivity of the
indoline intermediate to further oxidation under the condi-
tions of the reaction (methanol-d4, -40 °C). First, we treated
1
Phakellistatin 3 exhibited one set of H NMR signals in
methanol-d4 and two sets of signals in DMSO-d6 (>95:<5).
(25) Marsden, S. P.; Depew, K. M.; Danishefsky, S. J. J. Am. Chem.
Soc. 1994, 116, 11143-11144.
(26) Crich, D.; Huang, X. J. Org. Chem. 1999, 64, 7218-7223.
Org. Lett., Vol. 6, No. 11, 2004
1715

active against p388 and BEL-7404 cells, respectively. Since
Pro2, Phe3, and Gly4 have nearly identical conformations
in all three compounds it is tempting to attribute the
differences in biological activity to the Hpi and Thr residues.
The backbone conformations of these molecules lack com-
mon elements of protein secondary structure such as turns,
sheets, or helices. Of the human proteins with sequence
homology to phakellistatin 3 and 13, only RhoGAP proteins
have published crystallographic data. Since the sequence
FGPNL in P50 RhoGAP adopts a helical conformation,
phakellistatins 3 and 13 do not appear to mimic the Rho-
bound conformation of RhoGAP proteins.
Scheme 4
In conclusion, phakellistatin 13 was synthesized confirm-
ing the absolute stereochemistry of the Trp residue. Phakel-
listatin 13 was oxidatively cyclized to phakellistatin 3 and
isophakellistatin 3, demonstrating that full-length Trp-
containing peptides can be oxidized to the corresponding
Hpi-containing peptide. Over-oxidation of the Hpi residue
accounts for the low mass balance in the oxidative cyclization
of tryptophan residues in peptides. It is unclear whether the
biological formation of phakellistatin 3 suffers from similar
side reactions or whether Nature has found a way to cleanly
generate Hpi residues. Spectroscopic data in conjunction with
molecular modeling revealed a similar backbone conforma-
tion for all three cyclic peptides despite their markedly
different biological activity against tumor cell lines. Further
studies are required to elucidate the mode of action of
phakellistatin 3 and 13.
Due to signal overlap, fewer resonances could be assigned.
Two key NOEs could be discerned: between the Hpi â
proton and one of the Pro2 δ protons; between the Thr6 â
proton and one of the Pro5 â protons. The Phe3 NH exhibited
low temperature dependence (∆δ ) 1.7 ppb/K) whereas the
Thr6 NH did not (∆δ ) 4.1 ppb/K). The Gly4 NH and Leu7
NH were not sufficiently resolved for VT analysis.
Amber*/H2O does a good job of reproducing the confor-
mation of isophakellistatin 3 for which there is a crystal
structure. An unconstrained Monte Carlo search was used
to generate plausible conformations for phakellistatin 3 and
phakellistatin 13. The global minimum for phakellistatin 13
did not match all of the spectroscopic data; however, the
family of conformers 0.7 kcal/mol above the global minimum
(Figure 1) is consistent with all of the spectroscopic data.
The global minimum identified for phakellistatin 3 was
consistent with all of the spectroscopic data (Figure 1).
Phakellistatin 3 and isophakellistatin 3 adopt similar back-
bone conformations except around Thr6 and Pro5; the
threonine residue in isophekellistatin 3 is puckered toward
the Hpi residue (Figure 1). The conformations of phakel-
listatin 3 and phakellistatin 13 are much more similar, with
only subtle variations around Thr6 and Pro5.
Acknowledgment. This work was supported by NIH
GM-54523.
Supporting Information Available: Experimental pro-
cedures and spectroscopic data for compounds 1-7 and
details of computational methods. This material is available
free of charge via the Internet at http://pubs.acs.org.
Isophakellistatin is reported to be inactive against p388
cell lines whereas phakellistatin 3 and phakellistatin 13 are
OL049614P
1716
Org. Lett., Vol. 6, No. 11, 2004