Evolution of a synthetic strategy: Total synthesis of (±)- welwitindolinone A isonitrile

Article abstract of DOI:10.1021/ja076663z

An efficient and highly stereoselective total synthesis of the natural product (±)-welwitindolinone A isonitrile (1) is described. The bicyclo[4.2.0]octane core of 1 was established by a regio- and diastereoselective [2+2] ketene cycloaddition. The C12 quaternary center and vicinal stereogenic chlorine were installed in a single operation with excellent stereocontrol via a chloronium ion mediated semipinacol rearrangement. Described strategies for construction of the spiro-oxinole include a Sml ₂-LiCl mediated reductive cyclization and a novel anionic cyclization that simultaneously constructs the spiro-oxindole and vinyl isonitrile moieties.

Full text of DOI:10.1021/ja076663z

Published on Web 01/17/2008
Evolution of a Synthetic Strategy: Total Synthesis of
(()-Welwitindolinone A Isonitrile
Sarah E. Reisman, Joseph M. Ready, Matthew M. Weiss, Atsushi Hasuoka,
Makoto Hirata, Kazuhiko Tamaki, Timo V. Ovaska, Catherine J. Smith, and
John L. Wood*,†
Sterling Chemistry Laboratory, Department of Chemistry, Yale UniVersity,
New HaVen, Connecticut 06520-8107
Received September 19, 2007; E-mail: john.l.wood@colostate.edu
Abstract: An efficient and highly stereoselective total synthesis of the natural product (()-welwitindolinone
A isonitrile (1) is described. The bicyclo[4.2.0]octane core of 1 was established by a regio- and
diastereoselective [2+2] ketene cycloaddition. The C12 quaternary center and vicinal stereogenic chlorine
were installed in a single operation with excellent stereocontrol via a chloronium ion mediated semipinacol
rearrangement. Described strategies for construction of the spiro-oxinole include a SmI2-LiCl mediated
reductive cyclization and a novel anionic cyclization that simultaneously constructs the spiro-oxindole and
vinyl isonitrile moieties.
Introduction
In 1994, as a result of a screen for novel compounds
displaying multiple drug resistance (MDR) reversing activity,
Moore and co-workers reported the structural elucidation of the
welwitindolinones, a family of unusual oxindole alkaloids
isolated from the cyanobacteria Hapalosiphon welwischii and
Westiella intricate (Figure 1, 1-7).1 Five years later, the
structures of three additional highly oxidized welwitindolinones
were reported (8-10).2 Of the ten welwitindolinones isolated
to date, nine comprise a bridged 3,4-oxindole carbon skeleton
typified by N-methyl welwitindolinone C isothiocyanate (7), the
compound responsible for the MDR reversing activity of the
algal extracts. Interestingly, a single welwitindolinone, termed
welwitindolinone A isonitrile (1), consisted of a unique carbon
skeleton in which the oxindole is spiro-fused to a densely
functionalized bicyclo[4.2.0]octane core. Despite their obvious
differences, the homology of the relative stereochemistry at C12,
C13, and C15 led Moore to speculate that 1 is the biogenetic
precursor to 2-10. While a number of synthetic strategies
toward 7 have been reported in recent years,3-9 Baran and
Richter’s biomimetic synthesis of 1 in 2005 was the first
Figure 1. The welwitindolinone alkaloids.
completed total synthesis of any member of this family of
molecules.10,11
Several years ago we initiated a general program for the
synthesis of the welwitindolinones consisting of independent
approaches to both the bridged 3,4-oxindole carbon-skeleton
of 2-109 and the spiro-cyclobutane oxindole core of 1.12 The
latter approach resulted in a highly stereoselective synthesis of
1 that utilized an unprecedented anionic cyclization of an
isocyano-isocyanate precursor to provide both the vinyl isoni-
† Current address: Department of Chemistry, Colorado State University,
Fort Collins, CO 80523.
(1) Stratmann, K.; Moore, R. E.; Bonjouklian, R.; Deeter, J. B.; Patterson, G.
M. L.; Shaffer, S.; Smith, C. D.; Smitka, T. A. J. Am. Chem. Soc. 1994,
116, 9935.
(2) Jimenez, J. I.; Huber, U.; Moore, R. E.; Patterson, G. M. L. J. Nat. Prod.
1999, 62, 569.
(3) Baudoux, J.; Blake, A. J.; Simpkins, N. S. Org. Lett. 2005, 7, 4087.
(4) Deng, H. B.; Konopelski, J. P. Org. Lett. 2001, 3, 3001.
(5) Holubec, A. A. Progress Toward the Total Synthesis of the Welwitindolinone
Alkaloids: Efficient Construction of the Carbocyclic Skeleton; Yale
University: New Haven, CT, 2000.
(6) Konopelski, J. P.; Deng, H. B.; Schiemann, K.; Keane, J. M.; Olmstead,
M. M. Synlett 1998, 10, 1105.
(7) MacKay, J. A.; Bishop, R. L.; Rawal, V. H. Org. Lett. 2005, 7, 3421.
(8) Weiss, M. M. Progress Toward a Total Synthesis of the Welwitindolinone
Alkaloids; Yale University: New Haven, CT, 2001.
(9) Wood, J. L.; Holubec, A. A.; Stoltz, B. M.; Weiss, M. M.; Dixon, J. A.;
Doan, B. D.; Shamji, M. F.; Chen, J. M.; Heffron, T. P. J. Am. Chem. Soc.
1999, 121, 6326.
(10) Baran, P. S.; Richter, J. M. J. Am. Chem. Soc. 2005, 127, 15394.
(11) Baran, P. S.; Maimone, T. J.; Richter, J. M. Nature 2007, 446, 404.
(12) Ready, J. M.; Reisman, S. E.; Hirata, M.; Weiss, M. M.; Tamaki,
K.; Ovaska, T. V.; Wood, J. L. Angew. Chem., Int. Ed. 2004, 43, 1270.
9
10.1021/ja076663z CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 2087-2100
2087

A R T I C L E S
Scheme 1
Reisman et al.
Scheme 2
necessarily reside in a cyclobutene ring (see 12, Scheme 1).
Concern about the propensity of the cyclobutene substrate to
undergo electrocyclic ring opening at the temperatures required
for cyclization led us to consider alternative methods.18
trile and spiro-oxindole moieties in a single operation.13 This
paper describes the evolution of our synthetic strategy for 1
and highlights the advances in both methods and strategy that
typically result from total syntheses of complex natural products.
At the outset of this project, Pd-catalyzed enolate arylation
was just emerging as a powerful method for the construction
of R-aryl carbonyl compounds. In 1998, Hartwig et al. reported
the intramolecular R-arylation of amides to give oxindoles.19
Importantly, this method is suitable for the preparation of several
3,3-disubstituted oxindoles and would not require a cyclobutene
substrate. To test the feasibility of employing this method to
access a hindered, spiro-cyclobutane oxindole such as 11, model
pentamethylated cyclobutyl-amide 16 was prepared.20 Indeed,
treatment of amide 16 with Pd2(dba)3 and BINAP in the presence
of NaOt-Bu delivered a 10:1 mixture of diastereomers from
which spiro-oxindole 17 was isolated in 38% yield (Scheme
2).
Discussion
I. Retrosynthetic Considerations. From a retrosynthetic
perspective, we initially expected that the potentially labile vinyl
isonitrile of 1 would derive from ketone 11 in the final stage of
the synthesis; thus 11 was considered a key synthetic intermedi-
ate (Scheme 1). For construction of 11, we anticipated that the
required bicyclo[4.2.0]octane carbon core would be accessible
from a [2+2] ketene cycloaddition of known diene 14 and
dimethylketene (15). With the core structure in place we would
be in position to experimentally assess various methods for
converting the derived ketone to 11. Initial efforts focused on
advancing tricycle 13 to a spiro-oxindole-containing intermediate
(e.g., 11). Strategically, this required formation of either the
C3-C9 (aryl) bond or C3-C2 (acyl) bond in a key cyclization
event. Given the numerous reports of methods for preparing
3,3-disubstituted oxindoles by formation of an sp3-aryl carbon-
carbon bond, we first considered strategies that would rely on
cyclization of arylhalide intermediates such as 12.
II. Construction of the Spiro-Oxindole by Formation of
the C3-Aryl Bond. Of particular relevance to the cyclization
of amide 12 is the work of Overman and co-workers, which
has thoroughly demonstrated the utility of intramolecular Heck
cyclizations for enantio- and diastereoselective preparation of
a variety of 3,3-disubstitued oxindoles.14-17 However, Heck
cyclizations, particularly of hindered substrates, often require
prolonged heating at elevated temperatures. Furthermore, in the
context of our synthetic target (1), the requisite olefin would
Having established the Pd-catalyzed enolate arylation as a
suitable strategy for constructing sterically hindered spiro-
oxindoles, we began to focus attention on the preparation of a
substrate suited for advancement to 1. At the outset of these
studies it was recognized that the concave/convex nature of the
requisite bicyclo[4.2.0]octane substrate (e.g., 18) would likely
direct aryl bond formation to the less hindered, convex face
(via 19) and lead to oxindole 20, which possesses the undesired
stereochemistry at C3. Hoping to overcome this intrinsic
stereochemical bias by adjusting the ligands on palladium we
proceeded with the preparation of a more advanced model
cyclization substrate. To this end, treatment of acetonide 14 with
excess isobutyryl chloride and triethylamine in refluxing THF
provided tricyclic ketone 13 in 85% yield as a single regio-
and diastereo-isomer (Scheme 3). Hydrogenation of the olefin
furnished cyclobutanone 21, which was subjected to Takai
(13) Reisman, S. E.; Ready, J. M.; Hasuoka, A.; Smith, C. J.; Wood, J. L. J.
Am. Chem. Soc. 2006, 128, 1448.
(14) Dounay, A. B.; Hatanaka, K.; Kodanko, J. J.; Oestreich, M.; Overman, L.
E.; Pfeifer, L. A.; Weiss, M. M. J. Am. Chem. Soc. 2003, 125, 6261.
(15) Ashimori, A.; Bachand, B.; Overman, L. E.; Poon, D. J. J. Am. Chem.
Soc. 2000, 122, 192.
(18) Early experiments in our laboratory demonstrated that model cyclobutene
Heck substrates were unstable to the reaction conditions.
(19) (a) Shaughnessy, K. H.; Hamann, B. C.; Hartwig, J. F. J. Org. Chem. 1998,
63, 6546. (b) For leading references on Pd-catalyzed arylation of carbonyl
compounds, see: Culkin, D. A.; Hartwig, J. F. Acc. Chem. Res. 2003, 36,
234-245 and references cited therein.
(16) Ashimori, A.; Overman, L. E. J. Org. Chem. 1992, 57, 4571.
(17) Dounay, A. B.; Overman, L. E. Chem. ReV. 2003, 103, 2945.
(20) See Supporting Information for details.
9
2088 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 3
Table 1. Catalyst Screeninga
catalyst
(mol %)
ligand
conversion
(%)
27
(%)b
28
(%)b
29
(%)b
(mol %)
1
2
3
4
5
6
7
8
9
Pd2(dba)3 (30)
Pd2(dba)3 (40)
Pd2(dba)3 (70)
Pd(OAc)2 (30)
t-BnClPd(PPh3)2 (30) PCy3 (30)
t-BnClPd(PPh3)2 (30) BINAP (40)
t-BnClPd(PPh3)2 (30) Dppe (40)
t-BnClPd(PPh3)2 (30) Dppe (40)
t-BnClPd(PPh3)2 (30) (o-tol)3P
BINAP (45)
BINAP (30)
BINAP (100)
PCy3 (30)
88
100
100
32
4
16
2
0
0
6
3
2
4
25
18
26
0
9
0
10
6
22
5
5
7
5
8
8
60
0
100
100
100
100
100
100
59
59
54
79
77
65
10 t-BnClPd(PPh3)2 (30)
11 Pd(PPh3)4 (30)
8
10
a Conditions: Reactions were performed with 1.5 equiv of t-BuONa in
refluxing dioxane [0.01 M]. b Isolated yield.
Scheme 5
Scheme 4
III. A Second Generation Strategy: Development of a
SmI2/LiCl-Mediated Preparation of Spiro-Oxindoles. Our
revised retrosynthetic strategy now called for the construction
of spiro-oxindole 11 via the intermediacy of enone 30 or 31,
wherein an isonitrile or isocyanate would serve as N-acyl
surrogates (Scheme 5). A survey of the literature provided
limited precedent for this type of bond construction but did
illustrate the viability of aryl isonitriles as acyl radical
precursors.22-25 Unfortunately, all examples required elevated
temperatures (>80 °C) that were considered incompatible with
the cyclobutene-containing substrate. However, we became
intrigued with a report by Kim and co-workers detailing the
SmI2-mediated intermolecular reductive coupling of isocyanates
and R,â-unsaturated esters to give amides.26 While the Kim
report described an intermolecular reaction, application to an
intramolecular substrate (i.e., 31) was expected to deliver spiro-
oxindole 11 under very mild conditions.
olefination conditions21 to give exo-olefin 22. Hydroboration/
oxidation of olefin 22 smoothly provided primary alcohol 23,
which after a two-step adjustment of oxidation state, delivered
carboxylic acid 24.
Carboxylic acid 24 was subsequently coupled with PMB-
protected o-bromoaniline 25 to give cyclization substrate 26
(Scheme 4). As illustrated in Table 1, exposure of amide 26 to
the conditions originally reported by Hartwig (entry 1) provided
a modest 29% yield of spiro-oxindoles 27 and 28 as a 1:6
mixture of diastereomers accompanied with a significant
quantity of hydrodebromination product 29; as anticipated the
major isomer (28) results from carbon-carbon bond formation
on the convex face of the bicyclic molecule. Further investiga-
tion of these conditions initially showed some promise with
regard to overriding the intrinsic facial bias (Table 1, entry 2);
however, subsequent screening of several palladium sources and
ligands failed to produce favorable yields of the desired isomer
(27) and led only to our improving the production of 28 (see
Table 1 entry 9 and conditions illustrated in Scheme 4).20
Given our inability to identify a ligand system that would
overcome the diastereofacial bias imparted by the bicyclic
system, we turned to an alternative strategy wherein construction
of the spiro-oxindole follows from cyclization via the C3-acyl
bond (see Scheme 5).
To investigate this reaction, model substrate 33 (a nonisolated
intermediate) was readily prepared by Pd-catalyzed Heck
coupling of cyclohex-2-en-1-one with o-iodoaniline followed
by treatment with phosgene and triethylamine (Scheme 6).
(22) Fukuyama, T.; Chen, X. Q.; Peng, G. J. Am. Chem. Soc. 1994, 116, 3127.
(23) For an alternative acyl radical precursor, see Herzon, S. B.; Myers, A. G.
J. Am. Chem. Soc. 2005, 127, 5342.
(24) Tokuyama, H.; Fukuyama, T. Chem. Record 2002, 2, 37.
(25) Tokuyama, H.; Yamashita, T.; Reding, M. T.; Kaburagi, Y.; Fukuyama,
T. J. Am. Chem. Soc. 1999, 121, 3791.
(26) Kim, Y. H.; Park, H. S.; Kwon, D. W. Syn. Comm. 1998, 28, 4517.
(21) Takai, K.; Kakiuchi, T.; Kataoka, Y.; Utimoto, K. J. Org. Chem. 1994,
59, 2668.
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2089

A R T I C L E S
Reisman et al.
Table 2. Additive Screeninga
Scheme 7
entry
additive
yield 34 (%)b
1
2
3
4
5
none
5
15
32
47
88
c
NiI2
HMPAd
LiBre
LiCle
a Reactions carried out with 1 equiv 32, 1 equiv t-BuOH, and 2.1 equiv
SmI2. b Isolated yield. c Run using 0.05 equiv. d Run using 6.2 equiv. e Run
using 8.4 equiv.
Scheme 6
Scheme 8
Unfortunately, exposure of the crude isocyanate to SmI2
provided only trace amounts of the desired reductive cyclization
product 34 (Table 2, entry 1). Subjection of isocyanate 33 to
the conditions reported by Kim et al. (Table 2, entry 3) provided
a modest increase to 32% yield.
Despite the low yields, these initial results appeared promis-
ing. In an effort to further improve the chemical yield of 34,
other additives known to alter the reactivity of SmI2 were
screened (Table 2). Gratifyingly, treatment of the crude isocy-
anate 33 to a preformed mixture of SmI2 and LiCl provided
spiro-oxindole 34 in 88% yield. This increased reactivity is
consistent with previous findings that the combination of SmI2
and LiCl generates SmCl2 in situ, a stronger reductant than SmI2
alone (the E1/2 for SmI2/LiCl and SmI2 are -1.78 V and -0.98
V, respectively).27
In considering the mechanism for cyclization, it was recog-
nized that substrate 33 possesses two potential electron accep-
tors: the R,â-unsaturated ketone and the isocyanate. In an effort
to distinguish which functional group is reduced by the SmI2/
LiCl mixture, two experiments were performed. In the first,
aniline 32 (lacking the isocyanate functionality) was exposed
to the optimized reaction conditions and found to furnish diol
35 (70% yield),28 the product of pinacol dimerization, and
aminal 36 (19% yield), the result of net-1,4-reduction of the
enone (Scheme 7).29 Alternatively, subjection of phenyl isocy-
anate (37, lacking a reactive enone) to the same reactions
conditions resulted in no reaction. These results suggest that
the conversion of isocyanate 33 to spiro-oxindole 34 proceeds
first by reduction of the enone, followed by cyclization into
the isocyanate.30
Although the mild conditions required for SmI2/LiCl-mediated
reductive cyclization suggested it would be tolerated by a highly
functionalized substrate (e.g., 31, Scheme 5), we sought
confirmation and prepared a more relevant, cyclobutane-
containing substrate (43, Scheme 8). To this end, cyclobutanone
13 was treated with triazene-protected aryl Grignard reagent
38 to give tertiary alcohol 39 in excellent yield as a single
diastereomer. Reductive deprotection of the triazene, followed
by urethane formation furnished 40, which upon acidic hy-
drolysis of the acetonide and selective oxidation of the allylic
alcohol provided hydroxy enone 41. Using this efficient four-
step sequence, multigram quantities of enone 41 could routinely
be prepared from 39 with only a single chromatographic
purification.
(27) (a) Miller, R. S.; Sealy, J. M.; Shabangi, M.; Kuhlman, M. L.; Fuchs, J.
R.; Flowers, R. A. J. Am. Chem. Soc. 2000, 122, 7718. (b) On the basis of
the emperical rules of House the reduction potentials of 33 and 43 are
predicted to be 1.7 V and 1.8 V, respectively, see: House, H. O.; Huber,
L. E.; Umen, M. J. J. Am. Chem. Soc. 1972, 94, 8471.
(28) Diol 35 was isolated as a mixture of diastereomers.
(29) On the basis of a theoretical yield of 50% for dimerization.
(30) It remains unclear whether the reductive cyclization of 33 to 34 occurs by
a 1 or 2 e- process.
9
2090 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 9
Scheme 10
Dithiolane protection of enone 41 followed by oxidation under
Swern conditions provided ketone 42, a masked substrate for
our newly developed SmI2/LiCl-mediated reductive cyclization
protocol. In the event, treatment of ketone 42 with DBU effected
smooth elimination of CO2 to provide the corresponding aniline,
which was converted in situ (Et3N, COCl2) to isocyanate 43.
Removal of the triethylamine salts and exposure of the crude
isocyanate 43 to a preformed mixture of SmI2/LiCl delivered
spiro-oxindole 44 in 71% yield as a 7:1 mixture of diastereo-
mers. The major diastereomer possessed C13/C15 relative
stereochemistry consistent with bond formation on the less-
hindered, convex face of 43 and was thus poised for advance-
ment to 1. This stereochemical assignment was confirmed by
single-crystal X-ray diffraction.31
At this juncture, a more detailed retrosynthesis was devised
(Scheme 9). While spiro-oxindole 44 was considered the most
advanced synthetic intermediate en route to 1, R-hydroxy enone
41, which was easily accessible in multigram quantities (five
steps and 65% overall yield) from cyclobutanone 13, was
envisioned to be an ideal substrate to explore the functional-
ization of the northern portion of the molecule. The illlustrated
strategy relies on accessing oxindole 11 by employing the SmI2/
LiCl-mediated reductive cyclization on a completely function-
alized aryl isocyanate (31). Aryl isocyanate 31 was envisioned
to arise from urethane 45, which we hoped to prepare from
enone 41. Though it was considered somewhat risky to delay
construction of the spiro-oxindole to the end of the synthesis,
the mild reaction conditions used for this key cyclization were
expected to be compatible even with such an advanced
intermediate.
was subsequently quenched with NCS to furnish R-chloroketone
48 in 71% yield as a 7:1 mixture of diastereomers. Unfortu-
nately, analysis of the coupling constant and NOE data indicated
that the major diastereomer contained the incorrect stereochem-
istry at C13.32 This diastereoselectivity, while undesired, was
not entirely unexpected as it resulted from trapping of the lithium
enolate on the less hindered, convex face of the bicyclic ring
system (see Scheme 10).
While R-acetoxy enone 46 could theoretically be advanced
to the desired C13 epimer through a two-step sequence involving
R-hydroxylation followed by chlorination with inversion of
stereochemistry, it was postulated that protection of the C11
hydroxyl of enone 41 with a suitably large protecting group
could over-ride the inherent facial bias of the bicyclic skeleton
and promote chlorination of the concave face. Thus, R-hydroxy
enone 41 was converted to the triisopropyl silyl ether 49
(Scheme 11). Subjection of silyl ether 49 to the conditions used
previously (Scheme 10, LHMDS, L-selectride, then NCS) failed
to provide significant quantities of chlorination product. How-
ever, after considerable experimentation it was found that
conversion of enone 49 to tert-butyldimethylsilyl enol ether 50
and treatment with NCS at room-temperature provided R-chlo-
roketone 51 in 86% yield as a single diastereomer possessing
the desired stereochemistry at C13. While speculative, the
selectivity in this reaction likely derives from the TIPS ether
of 50 residing in the pseudoaxial position, a conformation which
minimizes A1,2 strain between the two large silyl groups. In
this conformation, reaction by chloronium ion formation on the
convex face of 50, opposite of the large TIPS group, provides
the observed diastereomer.
IV. A Chloronium-Ion-Mediated Semipinacol Rearrange-
ment. The focus of initial studies for advancing R-hydroxy
enone 41 to isocyanate 31 was the stereoselective introduction
of the C13 chlorine. To investigate this chemistry, the C11
secondary alcohol of 41 was first protected as the acetate (46,
Scheme 10). Sequential treatment of 46 with LHMDS (to
transiently protect the urethane as the lithium amide) and
L-selectride regioselectively provided lithum enolate 47, which
Pleased with our ability to control the diastereoselectivity of
the chlorination event, we considered methods to install the all
carbon quaternary center at C12. After contemplating sequences
to convert chloroketone 51 to urethane 45, we decided to pursue
a more streamlined strategy in which the quaternary center and
vicinal stereogenic chlorine would be incorporated simulta-
neously. Specifically, we recognized that one approach to
(31) See Supporting Information for crystallographic details.
(32) See Supporting Information for details.
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2091

A R T I C L E S
Scheme 11
Reisman et al.
Scheme 13
Scheme 12
reoselective chlorination of structurally similar silyl enol ether
50 to give chloroketone 51 (Scheme 9) led us to hypothesize
that reaction via chloronium ion formation on the concave face
of 54 (producing intermediate 55), would result in chloroketone
56 wherein the relative stereochemistry between C12, C13, and
C15 is correct for advancement to 1 (Scheme 12).
To implement this strategy, tertiary allylic alcohol 54 was
prepared by a sequence which commenced by treating R-siloxy
ketone 49 with LHMDS (to deprotonate the urethane) followed
by L-selectride then N-phenyltriflimide to give enol triflate 57
in 89% yield (Scheme 13). Subsequent exposure of triflate 57
to Pd2(dba)3 and 1,1′-bis(diphenylphosphino)ferrocene (dppf)
in the presence of methanol, DIPEA and carbon monoxide
furnished enoate 58 (69% yield). Subjection of enoate 58 to
excess methyl magnesium bromide and anhydrous cerium
trichloride provided tertiary allylic alcohol 54, the required
semipinacol rearrangement substrate.
Initial screening of tertiary alcohol 54 with the conditions
described by Wang (chloramine-T, ZnCl2, MeCN) provided
minor quantities (<15%) of the rearrangement product 56;
however, separation from chloramine-T byproducts was difficult.
Switching to NaOCl/AcOH as the chlorine source improved the
overall conversion, unfortunately formation of multiple byprod-
ucts was observed. Use of CeCl3 in biphasic CH2Cl2/H2O and
cooling the reaction to 0 °C significantly improved the yield of
56 to 50%. Upon further optimization, using MeCN as the
solvent and maintaining the reaction temperature between -10
and 0 °C provided chloroketone 56 in 78% isolated yield as a
single diastereomer.35
construct R-halo quaternary centers is the semipinacol rear-
rangement of tertiary allylic alcohols (Scheme 12, 52 f 53).33
In analogy to the Lewis-acid-mediated semipinacol rearrange-
ment of epoxyl alcohols,34 the diastereoselectivity is a result of
stereospecific alkyl migration anti to the chloronium ion. It was
therefore anticipated that treatment of tertiary alcohol 54 with
a source of electrophilic chlorine would provide chloroketone
56 with the desired relative stereochemical relationship between
carbons C12 and C13 (Scheme 12). Furthermore, the diaste-
(33) Wang, B. M.; Song, Z. L.; Fan, C. A.; Tu, Y. Q.; Chen, W. M. Synlett
2003, 10, 1497.
(34) Marson, C. M.; Walker, A. J.; Pickering, J.; Hobson, A. D.; Wrigglesworth,
R.; Edge, S. J. J. Org. Chem. 1993, 58, 5944.
(35) The stereochemistry was assigned by extensive nOe experimentation, for
example, irradiation of the psuedo-axial H13 showed significant enhance-
ments of the bridgehead H15, H14eq, and H19 (methyl protons) (see
Supporting Information for details).
9
2092 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 14
Scheme 15
In this remarkable reaction, the relative stereochemistry of
the C12 quaternary center and vicinal chlorine are simulta-
1
neously set with excellent stereocontrol. Analysis of the H
NMR spectrum of the crude product mixture indicated that the
mass balance was a mixture of several minor byproducts (<2%
yield per byproduct); the major and only cleanly isolable
byproduct was olefin 59 (<5% yield), resulting from competitive
deprotonation R to the chloronium ion.36
With chloroketone 56 in hand, attention turned to accessing
fully functionalized reductive cyclization substrate 45 (see
Scheme 9) by converting the C20 ketone to the requisite vinyl
functionality. Disappointingly, all attempts to reduce ketone 56
resulted in recovery of starting material. Reasoning that the TIPS
protecting group was blocking access to the C20 ketone, 56
was desilylated by treatment with fluorosilicic acid in warm
acetonitrile to give hydroxy ketone 60 (Scheme 14).37 Interest-
ingly, prolonged heating (>12 h) resulted in formation of
cyclobutane quinoline 62.38 This unanticipated byproduct likely
arises by acid-mediated retro-aldol/chloride elimination to give
putative intermediate 61, followed by CO2 extrusion and
intramolecular condensation.
64 using Dess-Martin periodinane40 furnished ketone 45 and
set the stage for the previously developed SmI2/LiCl-mediated
reductive cyclization.
In the event, we were gratified to find that treatment of ketone
45 with DBU smoothly effected the elimination of CO2 and
provided the corresponding aniline, which was converted in situ
to isocyanate 31 (Et3N/COCl2). Upon removal of the triethy-
lamine salts, cooling to -78 °C and subjection to a preformed
mixture of SmI2/LiCl, spiro-oxindole 11 was isolated in 75%
yield as a single diastereomer. The stereochemical assignment
of oxindole 11 was confirmed by X-ray crystallographic analysis
and is consistent with bond formation at C3 occurring on the
less hindered, convex face of the bicyclic molecule. Importantly,
there was no detection of products wherein the C13 chlorine
had been reduced under the reaction conditions, an observation
which is consistent with literature reports that SmI2/LiCl
mixtures reduce enones significantly faster than alkyl chlorides.27
V. Attempts to Convert Ketone 11 to Welwitindolinone
A Isonitrile (1). Having developed an efficient and stereose-
lective preparation of highly functionalized spiro-oxindole 11
(17 steps and 10.3% overall yield from cyclohexadiene 14),
completion of (()-welwitindolinone A (1) isonitrile required
only conversion of the C11 ketone to the corresponding vinyl
To avoid complications that could arise by the retro-aldol
pathway illustrated in Scheme 14, hydroxy ketone 60 was
reduced with Me4NHB(OAc)3 to give diol 63 (Scheme 15).
Fortuitously, diol 63 crystallized from CDCl3, and single-crystal
X-ray diffraction provided definitive proof of the assigned
stereochemistry at C12 and C13 set by the semipinacol
rearrangement. Exposure of diol 63 to Martin sulfurane39
resulted in selective dehydration of the C20 alcohol, furnishing
the required vinyl moiety of 1. Subsequent oxidation of alcohol
(36) Attempts to assign the C12 stereochemistry of byproduct 59 using nOe
techniques were inconclusive, and 59 was unstable to recrystallization
conditions.
(37) Pilcher, A. S.; Deshong, P. J. Org. Chem. 1993, 58, 5130.
(38) Compound 62 was isolated as a single olefin isomer, however the bond
geometry was not confirmed.
(39) Martin, J. C.; Arhart, R. J. J. Am. Chem. Soc. 1971, 93, 4327.
(40) Dess, D. B.; Martin, J. C. J. Org. Chem. 1983, 48, 4155.
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2093

A R T I C L E S
Scheme 16
Reisman et al.
Scheme 19
extreme conditions (i.e., heating >150 °C), decomposition. For
example, heating ketone 11 in neat pyridine with 100 equiv of
hydroxylamine hydrochloride in a microwave reactor at 100 °C
for 12 h provided only recovered starting material! Attempts to
condense other amine sources documented to be successful with
sterically hindered substrates also failed to provide any detect-
able quantities of product.46 Interestingly, under all conditions
evaluated, there was no evidence of Cl-elimination, suggesting
that the C13-Cl is locked in a pseudo-equatorial position by
the conformationally rigid bicyclic skeleton.
Scheme 17
To determine whether iminium ion formation was occurring,
but disfavored under equilibrium conditions, 11 was subjected
to reductive amination conditions that had been successfully
employed with similar ketones in the synthesis of related
fischerindoles.47 Ketone 11 again failed to produce any amine
product 70 (R ) H). These results suggested that iminium ion
formation was strongly disfavored for ketone 11. Alternatively,
it is conceivable that iminium ion formation occurs, but the
intermediate is too hindered for hydride reduction; however,
the observation that ketone 11 undergoes facile NaBH4 reduction
to a single diastereomer of alcohol 71 suggests this is unlikely
(Scheme 19).
Scheme 18
isonitrile (see, 11 f 1, Scheme 16). Noting that the carbonyl
carbon of 11 and isonitrile-bearing carbon of 1 are in the same
oxidation state, formally the conversion of ketone 11 to vinyl
isonitrile 1 should require condensation with 1 equiv of
formamide (65) and the net loss of 2 equiv of water (Scheme
16). However, in a practical sense this straightforward approach
is limited by the poor nucleophilicity of formamide.
Given that dehydration of vinyl formamides is facile and can
be accomplished under mild conditions,41-43 a variety of routes
to convert ketones to vinyl formamides have been developed
for the purpose of synthesizing vinyl isonitriles. One such
method was developed by Barton et al., wherein a ketone (66)
is first converted to the corresponding oxime (70, Scheme 17),
a transformation which is generally high yielding and applicable
to a variety of substrates.42 Subsequent treatment of the oxime
67 with Ti(OAc)3 in the presence of acetic formic anhydride
(AFA) delivers the enamide (68). Recent advances have shown
that commercially available and inexpensive iron(0) can be used
as the reductant instead of Ti(OAc)3.44,45
Further probing of the general reactivity of ketone 11 revealed
that exposure to a variety of carbon-based nucleophiles (e.g.,
Grignard reagents, organolithiums) also results in the recovery
of starting material. Suspecting that competitive enolization was
responsible for the lack of reactivity, deuterium-trapping experi-
ments were conducted. Indeed, treatment of 11 with vinylmag-
nesium bromide followed by quenching with CD3OD provided
11-C10-D with near quantitative deuterium incorporation.
Recognizing that welwitindolinone A isonitrile (1) contains
10-11
∆
unsaturation, we sought to capitalize on the facile
enolization of ketone 11 by trapping with triflating reagents. In
the event, N-silylation of ketone 11 followed by treatment with
LHMDS cleanly generated the bridgehead lithium enolate (73),
which could be quenched with Comins’ reagent to give vinyl
triflate 74 (Scheme 20).48 With vinyl triflate 74 in hand,
investigations of metal-catalyzed amidation reactions were
undertaken.
In recent years, there have been numerous reports detailing
Pd-catalyzed coupling of vinyl triflates49 and vinyl tosylates50-52
with amides or carbamates to give enamide products. While
the Pd-catalyzed amidations were found to be quite general with
respect to the amide partner, until recently, they had been limited
Initial efforts to convert ketone 11 to 1 focused on application
of the aforementioned Barton sequence. Unfortunately, ketone
11 proved to be remarkably unreactive; all efforts to access
oxime 69 (Scheme 18, R ) OH) using a variety of conditions
simply resulted in the recovery of starting material or, under
(46) For a table of failed conditions for the conversion of 11 to 69 or 70, see
the Supporting Information.
(47) Baran, P. S.; Richter, J. M. J. Am. Chem. Soc. 2004, 126, 7450.
(48) Comins, D. L.; Dehghani, A. Tetrahedron Lett. 1992, 33, 6299.
(49) Wallace, D. J.; Klauber, D. J.; Chen, C. Y.; Volante, R. P. Org. Lett. 2003,
5, 4749.
(41) Baldwin, J. E.; Oneil, I. A. Synlett 1990, 10, 603.
(42) Barton, D. H. R.; Bowles, T.; Husinec, S.; Forbes, J. E.; Llobera, A.; Porter,
A. E. A.; Zard, S. Z. Tetrahedron Lett. 1988, 29, 3343.
(43) Creedon, S. M.; Crowley, H. K.; McCarthy, D. G. J. Chem. Soc., Perkin
Trans. 1 1998, 6, 1015.
(44) Burk, M. J.; Casy, G.; Johnson, N. B. J. Org. Chem. 1998, 63, 6084.
(45) Yoshida, M.; Watanabe, T.; Ishikawa, T. Heterocycles 2001, 54, 433.
(50) Klapars, A.; Campos, K. R.; Chen, C. Y.; Volante, R. P. Org. Lett. 2005,
7, 1185.
(51) Movassaghi, M.; Ondrus, A. E. J. Org. Chem. 2005, 70, 8638.
(52) Willis, M. C.; Brace, G. N.; Holmes, I. P. Synth. Stuttg. 2005, 19, 3229.
9
2094 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 20
Scheme 22
Scheme 23
Scheme 21
VI. An Intramolecular Condensation Approach. Unable
to advance ketone 11 or various derivatives to 1 via intermo-
lecular reactions (presumably due to steric and entropic factors)
we began to consider strategies involving the intramolecular
condensation of tethered amine nucleophiles (see Scheme 22).
It was envisioned that our current route to ketone 11 could easily
be modified to incorporate the required tethers at either C20
(e.g., 78) or C21 (e.g., 79). Of particular interest was tethered
amine 78, which was envisioned to arise in a straightforward
manner by selective functionalization of the less-hindered C20
alcohol of diol 63 (see Scheme 15).
Functionalization of diol 63 commenced by selective mono-
protection of the C20 hydroxyl as the TBS ether, and oxidation
of the remaining C11 alcohol with Dess-Martin periodinane
to provide ketone 80 (Scheme 23).40 Exposure of ketone 80 to
our previously optimized conditions for SmI2/LiCl-mediated
oxindole formation provided oxindole 81 in 91% yield, once
again with complete stereocontrol of C3 spiro-center. The silyl
protecting group was subsequently removed using fluorosilicic
acid to give hydroxy ketone 82, a substrate well-suited for
nitrogen incorporation.
to activated vinyl triflates or tosylates.52 Unfortunately, attempts
to couple vinyl triflate 74 and formamide (65) using a number
of catalysts, ligands, and solvents failed to produce any
observable quantities of vinyl formamide 75 (Scheme 21).53 At
low temperatures (<80 °C), the vinyl triflate 74 could be
recovered; however, at higher temperatures decomposition of
the starting material was observed. Even the use of stoichio-
metric palladium and ligand failed to provide any trace of
desired product or any byproducts indicative of oxidative
addition. On the basis of the success earlier in the synthesis for
promoting Pd-catalyzed CO-insertion of sterically hindered vinyl
triflate 57 (see Scheme 13), vinyl triflate 74 was also subjected
to a variety of Pd-catalyzed CO-insertion conditions. The
complete failure of these latter reactions suggested that vinyl
triflate 74 was too hindered to undergo oxidative addition to
palladium, and once again other routes were evaluated.
Because hydroxylamines are known for their exceptional
nucleophilicity, and N-O bond cleavage can be carried out
under a variety of conditions, initial efforts focused on prepara-
(53) At the time of this work, there were no examples of Pd-catalyzed amidation
of unactivated vinyl triflates.
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2095

A R T I C L E S
Scheme 24
Reisman et al.
Scheme 26
Scheme 27
Scheme 25
further corroborated by IR spectroscopy, which showed strong
carbonyl stretching frequencies at 1703 and 1780 cm-1, the latter
of which is consistent with literature values for cyclobutanones.58
The proposed structure was eventually confirmed by single-
crystal X-ray analysis. A proposed mechanism for formation
of this structurally intriguing compound is shown in Scheme
26. As excess Martin sulfurane was used, activation of the
secondary alcohol was accompanied by Lewis acid promoted
enolization of the ketone to give enolate 86. Subsequent
intramolecular nucleophilic displacement would then deliver
cyclobutanone 85.59
The difficulties encountered in efforts to access and cyclize
tethered amine substrates from hydroxy ketone 82 were
postulated to derive from the sterically hindered environment
of the C20 neopentyl alcohol. As a result, we turned our
attention to a primary hydroxylamine tether (e.g., 88, Scheme
27).
Unfortunately, attempts to access alcohol 87 directly by
hydroboration of the olefin in oxindole 11 were unproductive;
the starting material was recovered unchanged using mild
conditions or went to multiple uncharacterized products under
more forcing conditions (Scheme 27). However, by turning to
an earlier olefin substrate, alcohol 64, we were gratified to find
that treatment with BH3‚DMS in THF and warming from 0 °C
tion of hydroxylamine derivative 83 (Scheme 24).54-57 To
incorporate the hydroxylamine, hydroxy ketone 82 was treated
with N-hydroxyphthalimide under a variety of Mitsunobu
conditions; disappointingly, starting material was recovered
unchanged. Subsequent attempts to effect alternative cyclizations
with carbamate-derivatives or silyl tethered amines were also
unsuccessful.
In addition to evaluating tether attachment, we briefly
explored the feasibility of eventually dehydrating derivatives
of 82 to the requisite olefin (i.e., 11) (Scheme 25). In the event,
it was found that exposure of a benzene solution of ketone 82
to Martin sulfurane provided a single, less polar product in 73%
yield. While low-resolution mass spectrometry (ES+) indicated
the correct mass (m/z calcd for C20H22ClNNaO2 [M+Na]+
366.1, found 366.1), analysis of the 1H NMR spectrum clearly
indicated that the product of this reaction was not known ketone
11. Notably, there were no vinyl protons; instead, the C10 R-keto
bridgehead proton signal had disappeared, and the H20 carbinol
signal had shifted upfield by approximately 1.2 ppm.
On the basis of extensive spectroscopic studies (NMR: 1H,
13C, 1H-1H COSY, DEPT, HMQC), the structure was ultimately
assigned as the pentacyclic ketone 85. This assignment was
(54) Aschwanden, P.; Kvaerno, L.; Geisser, R. W.; Kleinbeck, F.; Carreira, E.
M. Org. Lett. 2005, 7, 5741.
(55) Curran, D. P.; Fenk, C. J. Tetrahedron Lett. 1986, 27, 4865.
(56) Curran, D. P. J. Am. Chem. Soc. 1983, 105, 5826.
(57) Curran, D. P.; Singleton, D. H. Tetrahedron Lett. 1983, 24, 2079.
(58) Pretsch, E.; Buhlmann, P.; Affolter, C. Structure Determination of Organic
Compounds, 3rd ed.; Springer-Verlag: Berlin, 2000.
(59) While the observed stereochemistry is consistent with an SN2 mechanism
(inversion at C20), it is also possible an SN1 mechanism could be at work.
9
2096 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 28
Scheme 30
Scheme 29
yield over two steps (Scheme 30). Subsequent oxidation of
alcohol 97 proceeded smoothly with Dess-Martin periodinane40
to give ketone 98. As a proof-of-concept for the intramolecular
condensation, phthalimide 98 was deprotected by treatment with
hydrazine hydrate in deuterated chloroform at room temperature.
to room-temperature provided the desired primary alcohol 90,
albeit in low yield (ca. 25% by 1H NMR) owing to competitive
formation of a diastereomeric mixture of the secondary alcohol
regioisomer (91, Scheme 28).60
1
After 30 min, H NMR showed complete conversion to a new
product proposed to be hemiacetal 99. Furthermore, upon
removal of the hydrazine by rotary evaporation and standing in
fresh CDCl3 overnight, hemiacetal 99 underwent complete
conversion to oxazine 100.
Pleased by the mild conditions under which oxazine formation
occurred, methods for accessing oxindole 89 were evaluated.
Treatment of ketone 98 with DBU followed by phosgene and
triethylamine provided the required aryl isocyanate (101,
Scheme 31). Unfortunately, subjection of the crude isocyanate
to our optimized reaction conditions (SmI2/LiCl, t-BuOH,
-78 °C) provided products resulting from N-O bond cleavage
and none of the desired spiro-oxindole 102.
While this result was initially frustrating, the electron-
withdrawing nature of the phthalimide protecting group was
deemed responsible for the facile reduction of the N-O bond.
Previous experiments had shown that oxime ethers were stable
to the SmI2/LiCl at low temperatures; therefore protection of
the hydroxylamine functionality as an oxime ether (i.e., 103,
Scheme 31) was expected to render the N-O bond inert to the
SmI2/LiCl conditions. In the event, oxime 103 was prepared in
excellent yield by treating alcohol 97 with hydrazine hydrate
in chloroform for 30 min, then, after removal of excess
hydrazine by rotary evaporation, dissolution of the residue in
acetone.
This poor regioselectivity was surprising considering the high
levels of regiocontrol generally found for primary olefin
substrates in hydroboration reactions.61,62 Screening a number
of dialkylboranes known to increase the regioselectivity of
hydroboration failed to result in significant improvement.63
Suspecting that the free alcohol could be acting as a directing
group to favor Markovnikov addition,64 alcohol 64 was con-
verted to bis-Boc derivative 92 and subjected to the same
hydroboration conditions (Scheme 28). Thus, exposure of 92
to BH3‚DMS provided a much improved 63% isolated yield of
primary alcohol 93 along with <10% of minor regioisomer 94.
The balance of the mass was starting material and small
quantities of Boc-deprotected products.
VII. Synthesis of a Tethered Hydroxylamine Substrate.
With primary alcohol 93 in hand, efforts turned to the synthesis
of oxindole 95 (Scheme 29) by installation of the required
hydroxylamine and construction of the spiro-oxindole using our
SmI2/LiCl-mediated reductive cyclization.
Primary alcohol 93 proved to be an excellent substrate for
Mitsunobu reaction with N-hydroxyphthalimide, providing ph-
thalimide-protected hydroxylamine derivative 96, which upon
treatment with trifluoroacetic acid gave free alcohol 97 in 83%
(60) The major diastereomer is identical by 1H NMR to compound 63.
(61) Brown, H. C., W. A. Hydroboration, Benjamin: New York, 1962.
(62) Brown, H. C. Science 1980, 210, 485.
(63) Screening of dialkylboranes included: Et2BH, (Cy)2BH, 9-BBN, catechol
borane. With Et2BH, regioselection was comparable to BH3, while the other
three reagents provided no reaction at ambient or elevated temperatures.
(64) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. ReV. 1993, 93 (4), 1307.
Oxidation of the derived alcohol (103) with Dess-Martin
periodinane40 provided ketone 104 which, when exposed to
DBU followed by phosgene was converted to isocyanate 105.
Gratifyingly, subjection of the crude isocyanate to a preformed
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2097

A R T I C L E S
Scheme 31
Reisman et al.
Scheme 32
Scheme 33
mixture of SmI2/LiCl furnished spiro-oxindole 106 in 78% yield.
Having devised an appropriate sequence of transformations to
access hydroxylamine derivative 106, our attention turned to
formation of the critical C11-nitrogen bond. To this end,
treatment of a methanolic solution of oxindole 106 with
hydroxylamine hydrochloride (to scavenge liberated acetone)
cleanly provided oxazine 89 in 87% yield (Scheme 31).
With access to oxindole 89, the first intermediate in the
synthesis to contain both the oxindole and the requisite C11
nitrogen functionality, end-game strategies were reconsidered.
Formally, transformation of oxindole 89 to 1 would require
reductive formylation of the oxazine to give formamide 107,
followed by simultaneous loss of two molecules of water to
provide (()-welwitindolinone A isonitrile (1, Scheme 32).
Unfortunately, in analogy to ketone 11, oxazine 89 proved
to be extraordinarily unreactive. Implementation of a number
of reductive conditions reported in the literature to cleave similar
N-O bonds failed to provide any significant reduction
products.55-57 In addition, attempts to reduce the C-N double
bond with hydride sources were unproductive.65 At this time,
due in part to the fact that preparation of oxazine 89 now
required 23 steps and there was no clear route from 89 to 1, we
again considered alternative end-game scenarios.
VIII. Completion of the Synthesis of 1. Although nonpro-
ductive, our efforts to convert ketone 11 to (()-welwitindolinone
A isonitrile (1) did illustrate the need to incorporate the C11
nitrogen prior to oxindole formation. Thus, we considered the
possibility of utilizing nitrogen-containing substrates to access
analogues of 11 via the SmI2/LiCl-mediated reductive cycliza-
tion protocol (cf., 31 f 11 and 109 f 111 in Scheme 33).66-68
However, the difficulties encountered while attempting to
functionalize ketone 11 and oxazine 89 caused great concern
about the prospect of converting imine 111 to 1. Instead,
consideration of the mechanistic details illustrated in Scheme
33 led us to recognize that an intermediate similar to 110 could
(66) Dahlen, A.; Hilmersson, G. Chem.sEur. J. 2003, 9, 1123.
(67) Knettle, B. W.; Flowers, R. A. Org. Lett. 2001, 3, 2321.
(68) Shiraishi, H.; Kawasaki, Y.; Sakaguchi, S.; Nishiyama, Y.; Ishii, Y.
Tetrahedron Lett. 1996, 37, 7291.
(65) Conditions included: NaBH4/TFA, NaBH3CN/H+, NaBH(OAC)3, LAH,
DIBAL, LiEt3BH, Raney Ni/B(OH)3.
9
2098 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008

Total Synthesis of (
±
)-Welwitindolinone A Isonitrile
A R T I C L E S
Scheme 34
Scheme 36
Scheme 35
Scheme 37
115, wherein the aniline nitrogen is positioned to undergo ring
closure to the corresponding hemiaminal (116). Subsequent
Grob-type fragmentation of 116 to diene 117 followed by
aromatization then delivers hydroxyquinoline 119.71 Notably,
this reaction only requires a catalytic nucleophile and is
consistent with the observed formation of the same product (119)
under different conditions.72
In light of the proposed mechanism for the rearrangement of
114 to hydroxyquinoline 119, protection of the aniline nitrogen
was expected to disfavor formation of hemiacetal 116, thus
preventing the deleterious Grob-type fragmentation. Conversion
of carbamate 45 to N-Boc-imide 120 followed by in situ
treatment with DBU at room-temperature delivered N-Boc-
aniline 121 in 92% yield (Scheme 36). As predicted, 121 was
a stable intermediate and readily purified by silica gel chroma-
tography.
be accessed by taking advantage of the known propensity for
alkyl isocyanides to undergo R-deprotonation when exposed to
strong base.69 In this scenario, access to 1 would entail the
preparation of isocyano-isocyanate 112, which was envisioned
to undergo cyclization upon deprotonation via the intermediacy
of anion 113 (Scheme 34).
Of the many intermediates we had prepared that could
potentially give rise to isocyano-isocyanate 112, we initially
focused on aniline 114, the CO2-elimination product derived
from ketone 45 (Scheme 35). Unfortunately, aniline 114 was
unstable to a variety of conditions, including silica gel chro-
matography. While initially puzzling, our eventual isolation of
byproduct 119 upon exposure of aniline 114 to DMAP was
enlightening.70 As illustrated in Scheme 35, conversion of aniline
114 to hydroxyquinoline 119 is proposed to proceed first by
nucleophilic attack at the enone of 114 to provide intermediate
Having accessed enone 121, efforts next focused on installing
the C11 nitrogen. Unfortunately, attempts to affect the direct
reductive amination of 121 employing NH4OAc/NaBH3CN
(69) Hoppe, D. Angew. Chem., Int. Ed. 1974, 13, 789.
(70) Exposure of a dichloromethane solution of aniline 114 to DMAP followed
by Et3N and methyl chloroformate for 10 min provided hydroxyquinoline
119 in 65% yield.
(71) Grob, C. A. Angew. Chem., Int. Ed. 1969, 8, 535.
(72) Subjection of aniline 114 to TMSCN also provides hydroxyquinoline 119.
9
J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008 2099

A R T I C L E S
Reisman et al.
failed to provide amine 123.73 Resorting to a two step method,
enone 121 was readily converted to oxime 122 and represented
the first successful installation of the required C11-nitrogen bond
in an intermolecular reaction manifold. However, we were
disappointed to find that exposure of oxime 122 to a variety of
hydride sources failed to provide the desired amine (123). In
contrast, advancement of 121 to the corresponding methyl oxime
(124) followed by exposure to NaBH3CN/AcOH resulted in the
smooth formation of methoxylamine 125 (Scheme 37). Subse-
quent formylation of 125 with acetic formic anhydride (AFA)
provided a near quantitative yield of methoxyamide 126 which,
Scheme 38
74
upon sequential exposure to SmI2 followed by formic acid,
underwent N-O bond cleavage and Boc-deprotection to furnish
aniline 127.
In the final-ring closing event, treatment of formamide 127
with excess phosgene and triethylamine promoted simultaneous
dehydration of the formamide to the isocyanide and conversion
of the aniline to the isocyanate to deliver isocyano-isocyanate
112 (Scheme 38). Upon removal of the triethylamine salts by
filtration and drying in vacuo, we were delighted to find that
treatment of 112 with excess LHMDS at -78 °C provided 1 in
47% yield.13 However, attempts to further optimize the reaction
by changing the base or rigorously drying the intermediate
isocyano/isocyanate 112 did not significantly improve the yield
of 1.
install the C12 quaternary center and C13 stereogenic chlorine
with excellent stereocontrol. Furthermore, the spiro-oxindole and
vinyl isonitrile moieties were accessed in a single operation by
means of an unprecedented anionic cyclization of isocyano-
isocyanate 112, providing (()-welwitindolinone A isonitrile (1)
as a single diastereomer. Though not utilized in the successful
total synthesis, this work also resulted in the discovery of a
mild SmI2/LiCl-mediated synthesis of spiro-oxindoles.
Acknowledgment. Financial support was provided by Bristol-
Myers Squibb, Yamanouchi, Merck, Amgen, Pfizer, and the
NIH (Grant No. 1 RO1 CA/GM 93591-01A). S.E.R. thanks
Bristol-Myers Squibb for a graduate student fellowship. J.M.R.
was the recipient of a NIH postdoctoral fellowship. In addition,
we acknowledge and thank C.D. Incarvito for X-ray crystal-
lographic analysis.
Conclusions
A synthesis of (()-welwitindolinone A (1) from readily
accessible cyclohexadiene 14 has been developed. Notably, this
synthesis furnishes a 2.5% overall yield of 1 with an average
yield of 81% and features a chloronium ion mediated semipi-
nacol rearrangement of tertiary alcohol 54 to simultaneously
Supporting Information Available: Experimental and char-
acterization details. This material is available free of charge via
the Internet at http://pubs.acs.org.
(73) The 1H NMR spectrum for the crude reaction mixture indicated that
reduction of ketone 121 to the corresponding alcohol had occured.
(74) Keck, G. E.; Wager, T. T.; McHardy, S. F. Tetrahedron 1999, 55, 11755.
JA076663Z
9
2100 J. AM. CHEM. SOC. VOL. 130, NO. 6, 2008