Total synthesis and stereochemical reassignment of (±)-indoxamycin B

Article abstract of DOI:10.1002/anie.201109175

Revised version: The first total synthesis of indoxamycin B leads to a stereochemical reassignment of the natural product (see picture). The synthetic route features an efficient carboannulation sequence to rapidly access the dihydroindenone system. Moreover, a series of Au^I-catalyzed transformations served in the construction of the sterically congested core framework. Copyright

Full text of DOI:10.1002/anie.201109175

.
Angewandte
Communications
DOI: 10.1002/anie.201109175
Natural Product Synthesis
Total Synthesis and Stereochemical Reassignment of
(ꢀ)-Indoxamycin B**
Oliver F. Jeker and Erick M. Carreira*
The order Actinomycetales includes Gram-positive micro-
organisms that have been a rich source of biologically active
compounds, yielding more than 50% of all microbial anti-
biotics discovered to date.[1] Over the last five decades most
small-molecule discovery efforts have focused on bacteria of
terrestrial origin. However, microbes from aquatic environ-
ments have recently gained in importance as a source of
structurally diverse natural products.[2] In 2009 a research
group in Japan isolated a novel class of polyketides, sub-
sequently named indoxamycins, from saline cultures of
marine-derived actinomycetes (Scheme 1). Within this
The unique tricyclic carbon skeleton common to the
indoxamycins bears six contiguous asymmetric centers, an
a,b-unsaturated carboxylic acid side chain, and a trisubsti-
tuted alkene appendage. Three of the six stereogenic centers
are quaternary, of which two are vicinal. As outlined in
Scheme 2, the synthetic strategy envisioned relies on the
Scheme 2. Retrosynthetic analysis of indoxamycin B.
implementation of a number of key transformations, includ-
ing a Saucy–Marbet (propargyl Claisen) rearrangement,
which is followed by an allene hydroalkoxylation to install
the embedded tetrahydrofuran. Furthermore, an oxidative
carboannulation sequence would be employed to construct
the key dihydroindenone core in 3 from a C2v-symmetric
cyclohexa-2,5-dienone precursor.
The synthesis commenced with commercially available
methyl 3,5-dimethylbenzoate (4), which was converted to
cyclohexa-2,5-dienone 5 in four steps and 59% overall yield
(Scheme 3). In preliminary investigations, 5 proved reluctant
to undergo conjugate addition, which led to the examination
of alternatives, such as ketone crotylation followed by anionic
oxy-Cope rearrangement. The conditions identified for
ketone addition involved treatment of 5 with a 1,3-dimethyl-
allyltitanocene reagent generated in situ from (E)-4-methoxy-
pent-2-ene (6) and [Cp2TiCl2]/nBuLi.[4] Tertiary alcohol 7 was
thus isolated in 62% yield as a 3.8:1 mixture of olefin
diastereomers, as determined by 1H NMR spectroscopy.
Exposure of 7 to tBuOK/[18]crown-6 led to its participation
in a [3,3]-sigmatropic rearrangement to furnish the corre-
sponding ketone enolate, which was trapped with Et3SiCl to
afford 8 (70%). Enolsilane 8 readily underwent oxidative
cyclization to give dihydroindenone 3 in 74% yield upon
exposure to Pd(OAc)2 (10 mol%) in DMSO with O2 as
a terminal oxidant.[5]
Scheme 1. Originally assigned structures of indoxamycins A, B, and F.
Revised structure of indoxamycin B.
family, indoxamycins A and F have been shown to display
growth inhibition against HT-29 tumor cell lines (IC50
=
0.59 mm and 0.31 mm, respectively).[3] Their biological activity
in conjunction with the highly congested and stereochemical-
ly dense core render the indoxamycins notable as targets for
synthetic studies. Herein, we report a total synthesis of (ꢀ)-
indoxamycin B (1), which is not only the first synthesis of
a member of this unprecedented structural class, but has also
led to the stereochemical reassignment of the natural product.
[*] O. F. Jeker, Prof. Dr. E. M. Carreira
Laboratorium fꢀr Organische Chemie, ETH Zꢀrich, HCI H335
Wolfgang-Pauli Strasse 10, 8093 Zꢀrich (Switzerland)
E-mail: carreira@org.chem.ethz.ch
With a reliable route to dihydroindenone 3 established,
the full elaboration of the highly congested core could be
addressed (Scheme 4). Following hydrolysis of acetonide 3
(aq HCl, THF), the 1,3-diol produced was subjected to
vanadium-catalyzed epoxidation conditions. In the experi-
ment, a tandem reaction was observed wherein the inter-
mediate epoxide underwent a group-selective intramolecular
Homepage: http://www.carreira.ethz.ch
[**] We are grateful to Dr. M.-O. Ebert, R. Frankenstein, P. Zumbrunnen,
and R. Arnold for NMR spectroscopic measurements.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201109175.
3474
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 3474 –3477

Angewandte
Chemie
Scheme 3. Reagents and conditions: a) Li (2.2 equiv), tBuOH
(1.05 equiv), NH3/THF, ꢁ788C; then ICH2OC(O)tBu (1.0 equiv),
ꢁ788C; b) LiAlH4 (1.5 equiv), THF, 08C; c) TsOH (10 mol%), Me2C-
(OMe)2, RT, 96% over 3 steps; d) Pd/C (10 wt%, 2.5 mol%), tBuOOH
(2.5 equiv), K2CO3 (0.25 equiv), CH2Cl2, 08C, 61%; e) [Cp2TiCl2]
(3.0 equiv), nBuLi (6.0 equiv), (E)-4-methoxypent-2-ene (6) (1.5 equiv),
THF, ꢁ788C to RT; then 5, ꢁ408C to 108C, 3.8:1 ratio of olefin
diasteromers, 62%; f) tBuOK (3.0 equiv), [18]crown-6 (3.0 equiv), THF,
ꢁ788C to ꢁ408C, then Et3SiCl (3.0 equiv), ꢁ788C, d.r.=3.6:1, 70%;
g) Pd(OAc)2 (10 mol%), O2 atmosphere, Me2SO, 458C, 74%. Cp=
cyclopentadienyl, TsOH=p-toluenesulfonic acid.
Scheme 4. Reagents and conditions: a) HCl (aq, 1.0m, 2.9 equiv),
THF, RT, quant; b) [VO(acac)2] (5 mol%), tBuOOH (3.0 equiv),
4,4’-thiobis(2-tert-butyl-5-methylphenol) (2.5 mol%), 4 ꢁ M.S., CH2Cl2,
408C, 75%; c) tBuMe2SiCl (1.2 equiv), NEt3 (2.0 equiv), DMAP
(0.2 equiv), CH2Cl2, 08C to RT, 88%; d) DMP (1.5 equiv), CH2Cl2, 08C
to RT, 95%; e) KH (1.1 equiv), THF, RT; then [18]crown-6 (1.5 equiv),
propargyl bromide (1.2 equiv), 08C, 87%; f) [(Ph3PAu)3O]BF4
(1 mol%), 1,2-dichloroethane, 758C, 84%; g) LiBHEt3 (1.1 equiv),
THF, ꢁ788C, 80%; h) chloro[2-(di-tert-butylphosphino)biphenyl]gold(I)
(10 mol%), AgOTs (10 mol%), PhMe, 608C, d.r. =3.2:1, 72%. acac=
acetylacetonato, DMAP=4-dimethylaminopyridine, DMP=Dess–
Martin periodinane, M.S.=molecular sieves.
ring opening, affording 9 in 75% yield.[6–8] The primary
hydroxy group in 9 was then selectively protected as a silyl
ether (tBuMe2SiCl, NEt3, DMAP, 88%), and the remaining
secondary alcohol in 10 was oxidized to afford ketone 11
(DMP, 95%).[9] O-alkylation of the potassium enolate of 11
with propargyl bromide in the presence of [18]crown-6
furnished propargyl vinyl ether 12 in 87% yield. Attempts
at conducting the Saucy–Marbet rearrangement at elevated
temperatures (1608C, o-xylene) failed to give the desired
product. Gratifyingly, the rearrangement was observed to
proceed through the use of the trinuclear AuI-oxo complex
[(Ph3PAu)3O]BF4 as catalyst (1 mol%), yielding allene 13 as
structure of indoxamycin B ((1’’Z)-2-epi-1). Accordingly, 16
was taken forward in a two-step sequence involving regiose-
lective hydration ([Mn(dpm)3] (10 mol%), PhSiH3, O2, d.r. =
1:1, 49%),[14,15] followed by oxidation of the intermediate
secondary alcohols (DMP, 85%). The resulting methyl ketone
17 was exposed to equilibration conditions (DBU, toluene,
1008C) to obtain a 6:1 mixture of C(2) epimers, favoring 18.
Separation by chromatography on silica gel furnished 18 in
58% yield. When 18 was subjected to Wittig olefination,
a 1.6:1 mixture of olefin isomers was obtained (70%), which
could not be separated. Sequential deprotection of this
mixture was effected by ester saponification (aq LiOH)
followed by addition of aq HCl to attain removal of the silyl
protective group. Surprisingly, neither olefin isomer obtained
displayed spectral properties (1H and 13C NMR) that matched
those reported for the natural product. Careful reexamination
of the published NMR data (including NOESY spectra) of
several members of the indoxamycin family of natural
products led to the conclusion that the relative configuration
at C(2) had been misassigned.[3a] Furthermore, there was
considerable ambiguity regarding the geometry of the trisub-
stituted olefin side chain. We subsequently targeted the
revised structures (1’’E/Z)-1 for synthesis.
1
the only detectable diastereomer (84%) by H NMR spec-
troscopy.[10] Chemo- and diastereoselective reduction of the
cyclohexenone carbonyl in 13 was accomplished with
LiBEt3H, giving alcohol 14 in 80% yield. The formation of
the tetrahydrofuran ring of the indoxamycin framework was
then achieved through an intramolecular AuI-catalyzed
hydroalkoxylation of allene 14, to form tetracyclic intermedi-
ate 2 as a 3.2:1 mixture of inseparable diastereomers at C(2)
(72%).[11]
Reductive cleavage of the a-keto ether (SmI2, THF/
MeOH) in cyclopentanone 2 released a primary alcohol,
which was oxidized to the corresponding aldehyde (DMP,
97%; Scheme 5). The latter was subjected to Horner–Wads-
worth–Emmons olefination delivering the a,b-unsaturated
ester 15 in 92% yield. Chemoselective reduction of the
ketone in 15 (BH3·NH2tBu, CH2Cl2, reflux) furnished the
corresponding secondary alcohol (88%),[12] which was
exposed to Burgessꢀ reagent, yielding cyclopentene 16 in
44% yield.[13] It was envisioned that the conversion of the
terminal olefin in 16 to the corresponding methyl ketone 17
would not only set the stage for the introduction of the
trisubstituted (Z)-olefin side chain but also provide the
opportunity for C(2) epimerization to arrive at the reported
In contrast to the route previously discussed for the
conversion of 2 to (1’’E/Z)-2-epi-1 (Scheme 5), it seemed
prudent to install the side chain at C(2) prior to manipulation
of the cyclopentane ring in order to compare the spectro-
scopic properties of olefins 20a and 20b with those of the
natural product. Accordingly, tetracycle 2 was subjected to
Angew. Chem. Int. Ed. 2012, 51, 3474 –3477
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3475

.
Angewandte
Communications
olefin configuration unambiguously secured by NMR spec-
troscopic experiments after separation by chromatography on
silica gel (NOESY analysis: see the Supporting Information).
1
The similarity of the H NMR resonances of diastereomer
20a to those reported for natural indoxamycin B led to the
hypothesis that the natural product has the structure shown
for 1 in Scheme 1.
Olefin 20a was advanced by employing the sequence
described for the conversion of 2 to 15 (steps a–c; Scheme 5)
to furnish the corresponding enoate in 98% overall yield. The
introduction of the unsaturation in the five-membered ring
proceeded in 55% yield over two steps, following the
analogous reduction–dehydration protocol (steps d and e;
Scheme 5). Deprotection of 21 afforded carboxylic acid 1 in
96% yield, which, after conversion to the corresponding
potassium salt, exhibited 1H and 13C NMR spectra identical in
all respects to those reported for natural indoxamycin B.[19]
In conclusion we have described the first total synthesis
(ꢀ)-indoxamycin B (1), which has culminated in the struc-
tural reassignment of the natural product. The synthetic
strategy is based on the use of a highly symmetric cyclohexa-
2,5-dienone precursor and relies on a series of modern metal-
catalyzed reactions to construct the tricyclic core framework.
The salient features of the route include an efficient
carboannulation sequence involving a Ti-mediated ketone
crotylation and anionic oxy-Cope rearrangement, as well as
a Pd-catalyzed oxidative cycloalkenylation reaction to rapidly
access the key dihydroindenone intermediate. Moreover,
a highly diastereoselective vanadium-catalyzed tandem reac-
tion and a series of AuI-catalyzed transformations (Saucy–
Marbet rearrangement and allene hydroalkoxylation) allow
for elaboration of the sterically congested scaffold. We
propose that the structural revision for indoxamycin B
((1’’Z)-2-epi-1 to 1) described herein is also valid for the
other members of the natural product family, but validation of
this hypothesis awaits additional experimentation. Efforts to
expand the described strategy to the synthesis of other
indoxamycins are subject of current research in our labora-
tories and will be reported in due course.
Scheme 5. Reagents and conditions: a) SmI2 (1.0 equiv), THF/MeOH,
RT, d.r.=3.2:1, quant; b) DMP (1.25 equiv), CH2Cl2, 08C to RT,
d.r.=3.8:1, 97%; c) methyl diethylphosphonoacetate (5.0 equiv), NaH
(5.0 equiv), THF, RT, d.r.=3.6:1, 92%; d) BH3·tBuNH2 (2.0 equiv),
CH2Cl2, 408C, d.r.=9:1, 88%; e) Burgess’ reagent (2.0 equiv), PhMe,
1108C, 44%; f) Mn(dpm)3 (10 mol%), PhSiH3 (2.5 equiv), O2 atmos-
phere, EtOH, RT, d.r. =1:1, 49%; g) DMP (2.0 equiv), CH2Cl2; 08C to
RT, 85%. h) DBU (5.0 equiv), PhMe, 1008C, 58%; i) Ph3PEtBr
(6.0 equiv), KN(SiMe3)2 (5.0 equiv), THF/DMPU, RT, ꢁ788C to 08C,
1:1.6 ratio of olefin isomers, 70%; j) LiOH (18 equiv); then HCl (aq,
1.0m, 53 equiv), THF/MeOH/H2O, RT, 73%. dpm=2,2,6,6-tetra-
methyl-3,5-heptanedionato, DBU=1,8-diazabicyclo[5.4.0]undec-7-ene,
DMPU=1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone.
the hydration–oxidation protocol disclosed earlier to provide
methyl ketone 19 as a single diastereomer at C(2) (53% over
2 steps; Scheme 6).[16] Wittig olefination of 19 under the
conditions disclosed by Still proved capricious, giving 20a and
20b in a wide range of selectivity (2.5:1 to 15:1),[17,18] with the
Received: December 27, 2011
Published online: February 17, 2012
Keywords: antitumor agents · gold ·
.
sigmatropic rearrangement · structure elucidation ·
total synthesis
[1] J. Bꢁrdy, J. Antibiot. 2005, 58, 1 – 26.
[2] W. Fenical, P. R. Jensen, Nat. Chem. Biol. 2006, 2, 666 – 673.
[3] a) S. Sato, F. Iwata, T. Mukai, S. Yamada, J. Takeo, A. Abe, H.
Kawahara, J. Org. Chem. 2009, 74, 5502 – 5509; b) S. Sato, F.
Iwata, S. Yamada, J. Takeo, A. Abe, H. Kawahara (Nippon
Suisan Kaisha, Ltd.), WO 113725, 2010.
[4] a) A. Kasatkin, T. Nakagawa, S. Okamoto, F. Sato, J. Am. Chem.
Soc. 1995, 117, 3881 – 3882; b) Y. Yatsumonji, T. Nishimura, A.
Tsubouchi, K. Noguchi, T. Takeda, Chem. Eur. J. 2009, 15, 2680 –
2686.
Scheme 6. Reagents and conditions: a) [Mn(dpm)3] (10 mol%),
PhSiH3 (2.5 equiv), O2 atmosphere, EtOH, RT, d.r. =1:1, 73%; b) DMP
(1.5 equiv), CH2Cl2; 08C to RT, 72%; c) Ph3PEtBr (6.0 equiv), KN-
(SiMe3)2 (5.0 equiv), THF/HMPA (10:1), RT, ꢁ788C to ꢁ308C, 2.5:1–
15:1 ratio of olefin isomers, 80%; d) SmI2 (1.0 equiv), THF/MeOH
(7:3); RT, 99%; e) DMP (1.25 equiv), CH2Cl2; 08C to RT, quant;
f) methyl diethylphosphonoacetate (5.0 equiv), NaH (5.0 equiv), THF,
RT, 99%; g) BH3·tBuNH2 (2.0 equiv), CH2Cl2, 408C, 79%; h) Burgess’
reagent (2.0 equiv), PhMe, 1108C, 69%; i) LiOH (10 equiv), then HCl
(aq, 1.0m, 29 equiv), THF/MeOH/H2O, RT, 96%. HMPA=
(Me2N)3PO.
[5] a) Y. Ito, H. Aoyama, T. Hirao, A. Mochizuki, T. Saegusa, J. Am.
Chem. Soc. 1979, 101, 494 – 496; b) A. S. Kende, B. Roth, P. J.
Sanfilippo, J. Am. Chem. Soc. 1982, 104, 1784 – 1785; c) A. S.
3476
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 3474 –3477

Angewandte
Chemie
Kende, B. Roth, P. J. Sanfilippo, T. J. Blacklock, J. Am. Chem.
Soc. 1982, 104, 5808 – 5810; d) M. Toyota, T. Wada, K. Fuku-
moto, M. Ihara, J. Am. Chem. Soc. 1998, 120, 4916 – 4925; e) For
review see: M. Toyota, M. Ihara, Synlett 2002, 1211 – 1222.
[6] K. B. Sharpless, R. C. Michaelson, J. Am. Chem. Soc. 1973, 95,
6136 – 6137.
[7] The intermediate epoxide could be observed when the reaction
was stopped prior to completion.
[8] Y. Kishi, M. Aratani, H. Tanino, T. Fukuyama, T. Goto, S. Inoue,
S. Sugiura, H. Kakoi, J. Chem. Soc. Chem. Commun. 1972, 64 –
65.
Chem. Soc. Jpn. 1990, 63, 179 – 186; e) S. Inoki, K. Kato, S.
Isayama, T. Mukaiyama, Chem. Lett. 1990, 1869 – 1872.
[15] C. S. Schindler, C. R. J. Stephenson, E. M. Carreira, Angew.
Chem. 2008, 120, 8984 – 8987; Angew. Chem. Int. Ed. 2008, 47,
8852 – 8855.
[16] After hydration of 2, the pair of alcohols C(1’’)-R*/C(2)-R* and
C(1’’)-S*/C(2)-R* was separated from diastereomers C(1’’)-R*/
C(2)-S* and C(1’’)-S*/C(2)-S* by chromatography on silica gel
to furnish an inconsequential 1:1 mixture of C(1’’) epimers that
were subsequently oxidized to the corresponding ketone (see the
Supporting Information).
[9] D. B. Dess, J. C. Martin, J. Org. Chem. 1983, 48, 4155 – 4156.
[10] a) B. D. Sherry, F. D. Toste, J. Am. Chem. Soc. 2004, 126, 15978 –
15979.
[11] Z. Zhang, C. Liu, R. E. Kinder, X. Han, H. Qian, R. A.
Widenhoefer, J. Am. Chem. Soc. 2006, 128, 9066 – 9073.
[12] Ketone reduction in 15 furnished the corresponding secondary
alcohol in diastereoenriched form at C(2) (3.6:1 to 9:1) due to
a kinetic resolution of the initial isomer mixture.
[13] a) E. M. Burgess, H. R. Penton, Jr., E. A. Taylor, J. Am. Chem.
Soc. 1970, 92, 5224 – 5226; b) E. M. Burgess, H. R. Penton, Jr.,
E. A. Taylor, J. Org. Chem. 1973, 38, 26 – 31.
[14] a) T. Mukaiyama, S. Isayama, S. Inoki, K. Kato, T. Yamada, T.
Takai, Chem. Lett. 1989, 449 – 452; b) S. Inoki, K. Kato, T. Takai,
S. Isayama, T. Yamada, T. Mukaiyama, Chem. Lett. 1989, 515 –
518; c) S. Isayama, T. Mukaiyama, Chem. Lett. 1989, 1071 – 1074;
d) K. Kato, T. Yamada, T. Takai, S. Inoki, S. Isayama, Bull.
[17] C. Sreekumar, K. P. Darst, W. C. Still, J. Org. Chem. 1980, 45,
4260 – 4262.
[18] Preferential formation of diastereomer 20a stands in contrast to
the expected product on the basis of the substrates examined by
Still. However, the presence of a bishomoallylic ketone in the
substrates in this study may be responsible for the observed
outcome.
[19] We note that the 1H and 13C NMR spectra for synthetic
1 obtained from purification by preparative TLC (as previously
described for the natural product) did not match those reported,
albeit the observed differences in the resonances were minor
(1H: j Dd jꢂ 0.33 ppm; 13C: j Dd jꢂ 6.5 ppm). However, incre-
mental addition of 1.0 equiv of CD3OK to a solution of 1 in
CD3OD led to a new set of signals in the 1H and 13C NMR
spectra that were in full accordance with the reported data for
the natural product (see the Supporting Information).
Angew. Chem. Int. Ed. 2012, 51, 3474 –3477
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3477