Enantioselective synthesis of the central ring system of lomaiviticin A in the form of an unusually stable cyclic hydrate

Article abstract of DOI:10.1002/anie.200704830

(Chemical Equation Presented) A model system representing a protected form of the four central rings of the antitumor compound lomaiviticin A has been synthesized (outlined in red in the picture). The approach features an intramolecular furan Diels-Alder reaction, a stereoselective oxidative enolate coupling to dimerize the halves, and a base-initiated cascade reaction. The 1,4-diketone of the central ring system exists as a stable cyclic hydrate.

Full text of DOI:10.1002/anie.200704830

Communications
DOI: 10.1002/anie.200704830
Natural Product Synthesis
Enantioselective Synthesis of the Central Ring System of
Lomaiviticin A in the Form of an Unusually Stable Cyclic Hydrate**
Evan S. Krygowski, Kerry Murphy-Benenato, and Matthew D. Shair*
The lomaiviticin family of natural products are potent
cytotoxic molecules with remarkable C2-symmetric structures
(Scheme 1). They were isolated from a strain of actino-
mycetes, Micromonospora lomaivitiensis, which was itself
isolated from the inner core of a host ascidian.[1] The GI50
values of 1 against a panel of 24 cultured cancer cell-lines are
0.007–72 nm, and both 1 and 2 are potent antibiotics against
Gram-positive bacteria. He et al.[1] reported that 1 and 2
damage DNA, although detailed studies of their interactions
with nucleic acids or other biopolymers have not been
disclosed. The diazobenzofluorene ring system of the lomai-
viticins would appear to be responsible for their cytotoxicity.[2]
This rare ring system has only ever been found in the
kinamycin family of antibiotics (see kinamycin C in
Scheme 1),[3] which resemble the monomeric subunits of the
lomaiviticins.
Compounds 1 and 2 are daunting synthetic targets
because of their size, potential lability, and juxtaposition of
diverse functional groups. Of particular complexity are the
central CD/C’D’-ring systems of the lomaiviticins. These rings
are linked by a sterically congested and synthetically chal-
ꢀ
lenging C2 C2’ s bond, the center of which is the axis of
symmetry for 1 and 2. To date, Nicolaou et al. have reported
the only approach to the lomaiviticins with the syntheses of
model D/D’-ring systems of 1 and 2.[4]
Considering a global strategy for syntheses of 1 and 2, we
concluded that the most convergent approach to prepare
these C2-symmetric molecules would be to stereoselectively
ꢀ
form the C2 C2’ bond at the latest possible stage, thereby
reducing the amount of double processing (Scheme 2). Since
ꢀ
the C2 C2’ bond is part of a 1,4-diketone (C1-C2-C2’-C1’),
our desire was to link the two tetracyclic “halves” of 1 and 2 in
a late-stage stereoselective oxidative enolate coupling reac-
tion (see 3 in Scheme 2). However, this transformation has
two serious problems. Firstly, the ketone enolate 4 will be
prone to b elimination, thus leading to aromatization of the D
ring. Secondly, there is no obvious means of controlling the
configurations of the newly formed stereocenters at C2 and
C2’. Our hypothesis is that linking the C3 tertiary carbinol to
C6 and forming the 7-oxanorbornanone 5 will resolve both
issues. b Elimination of the C3 alkoxy group is prevented in
enolates of 5 by the nearly orthogonal orientation of the
enolate p system and the antibonding s* orbital of the
Scheme 1. Structures of lomaiviticins A (1) andB ( 2) as well as a
relatedkinamycin.
[*] E. S. Krygowski, Prof. M. D. Shair
Department of Chemistry andChemical Biology
HarvardUniversity
12 OxfordStreet, Cambridge, MA 02138 (USA)
Fax: (+1)617-496-4591
E-mail: shair@chemistry.harvard.edu
bridging C O bond.[5] Oxidative enolate coupling should also
ꢀ
be stereoselective in this system, with dimerization occurring
Dr. K. Murphy-Benenato
AstraZeneca Pharmaceuticals LP
35 Gatehouse Drive, Waltham, MA 02451 (USA)
from the convex, a faces (syn to the oxygen bridge), thus
ꢀ
delivering the desired a,a stereochemistry across the C2 C2’
bond. Herein we report the synthesis of the central ring
system of lomaiviticin A (1) using this strategy.
[**] Financial support for this project was provided by the NIH
(CA125240), Merck Research Laboratories, andNovartis. K.M.-B.
acknowledges AstraZeneca and the NIH for postdoctoral fellow-
ships. Douglas Ho andRichardStaples are acknowledgedfor X-ray
crystallographic analysis.
Initially, we sought to determine whether oxidative
enolate coupling of 7-oxanorbornanones could be accom-
plished stereoselectively and without b elimination. Conver-
sion of 6 (85% ee)[6] into 7 and exposure to Ag2O in DMSO[7]
at 1008C afforded the desired oxidative enol coupling adduct
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1680
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1680 –1684

Angewandte
Chemie
Scheme 2. Retrosynthesis of lomaiviticin A showing a strategy basedon oxidative enolate coupling of a 7-oxanorbornanone. PG =protecting
group.
as a 7:1 mixture of the C2-symmetric and meso products, 8 and
9,[8] respectively (Scheme 3). Gratifyingly, the reaction oc-
curred with complete facial selectivity from the convex, exo
faces of both bicycles. Attempted reductive cleavage of the
ether bridge, between the C6 and C6’ carbons, with SmI2 did
not afford 10, the central ring system of 2, but instead
delivered cyclobutanediol 11. Apparently, the close proximity
of the ketones renders pinacol coupling a faster process than
Scheme 4. Cleavage of the oxygen bridge using an exocyclic enolate.
Reagents andconditions: a) NiCl (5 mol%), CrCl2 (6 equiv), 13
2
(1.2 equiv), THF, 238C, 18 h, 35%; b) TPAP (2.5 mol%), NMO
(1.5 equiv), 4-Š M.S., CH2Cl2, 238C, 1 h, 77%; c) TFA, CH2Cl2, 08C,
58%; d) K2CO3 (3 equiv), MeOH, 08C, 10 min, 95%. M.S.=molecular
sieves, NMO=N-methylmorpholine-N-oxide, TFA=trifluoroacetic
acid, TIPS=triisopropylsilyl, TPAP=tetrapropylammonium perruthen-
ate.
ꢀ
C O bond cleavage.
We then tested an oxygen-bridge-opening strategy that
did not rely on generating reactive intermediates of the C1
ether bridge opened as desired and the C4 oxygen was
installed with the desired syn relationship to the C3 tertiary
carbinol. Following the discovery of this cascade reaction we
then sought to apply the same transformation to a C2-
symmetric dimer of 15, which represents the central ring
system of the lomaiviticins.
An enantioselective synthesis of the central ring system of
lomaiviticin A began with Michael addition of the lithium
enolate of furanone 17[10] to the oxazolidinone acrylate 16,[11]
to deliver 18 in 88% yield as a 1:1 mixture of diastereomers
(Scheme 5). To prevent the furanone of 18 from participating
in the ensuing aldol addition reaction, it was protected as the
acetoxyfuran 19. Using the Evans protocol,[12] a MgII-cata-
lyzed anti-aldol reaction was performed between 19 and
b-thiophenylacrolein (20).[13] Following TBS protection of the
aldol adduct 21, acetoxyfuran 22 was isolated as a separable
5.5:1 mixture of anti-aldol diastereomers in 60% yield over
two steps. Conversion of 22 to furanone 23 was accomplished
by exposure to catalytic potassium cyanide in iPrOH. The
same reaction in EtOH resulted in competitive opening of the
oxazolidinone. Treatment of 23 with mCPBA oxidized the
vinyl sulfide to afford vinyl sulfone 24. Upon gentle warming
24 underwent tautomerization, intramolecular furan Diels–
Alder cycloaddition, and enol tautomerization to afford 25 as
a 3:1 mixture of separable diastereomers in 53% overall yield
from 22. The furan Diels–Alder reaction afforded only endo
Scheme 3. Stereoselective dimerization of 7-oxanorbornanone.
Reagents andconditions: a) LDA (1.3 equiv), TMSCl (8 equiv), THF,
ꢀ788C, 5 min; b) Ag2O (1.2 equiv), DMSO, 1008C, 2 h, 85% over 2
steps; c) SmI2 (6 equiv), THF/MeOH (3:1), ꢀ788C, 20 min, 60%.
DMSO=dimethyl sulfoxide, LDA=lithium diisopropylamide,
TMS=trimethylsilyl.
ketone, thereby avoiding proximity-induced reactions like the
aforementioned pinacol coupling. A model substrate was
constructed (Scheme 4) by performing a tandem Kishi–
Nozaki–Hiyama coupling/intramolecular furan Diels–Alder
reaction between aldehyde 12 and vinyl iodide 13,[9] to afford
cycloadduct 14 after alcohol oxidation and enol silane
hydrolysis. In an attempt to b-eliminate the phenylsulfonyl
group, we were surprised yet happy to discover that treatment
of 14 with K2CO3 in MeOH at 08C led to 15 in 95% yield. The
Angew. Chem. Int. Ed. 2008, 47, 1680 –1684
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1681

Communications
Scheme 5. Enantioselective synthesis of the central ring system of lomaiviticin A. Reagents andconditions: a) LDA (1 equiv), THF, ꢀ788C,
30 min; then 16 (0.5 equiv), THF, ꢀ788C, 3 h, 90%; b) Ac2O (6 equiv), DMAP (10 mol%), pyr, 238C, 48 h, 92%; c) MgCl2 (10 mol%), NaSbF6
(30 mol%), Et3N (2 equiv), TMSCl (1.5 equiv), 20 (1.5 equiv), EtOAc, 238C, 5 days, d.r. 5.5:1, 60–77%; d) TBSCl (3 equiv), imidazole (6 equiv),
DMAP (5 mol%), DMF, 238C, 14 h, 78%; e) KCN (10 mol%), iPrOH, 508C, 3 days, 70–95%; f) mCPBA (3 equiv), CH2Cl2, 238C, 20 min; g) neat,
508C, 3 days, d.r. 3:1, 53% over 2 steps; h) 1. 1,4-dioxane/water/conc. HCl (4:4:1), 1008C, 2 days; 2. TBSOTf (3 equiv), iPr2NEt (6 equiv), CH2Cl2;
3. K2CO3 (5 equiv), THF/MeOH/water (2:1:1), 238C, 30 min, 71% over 3 steps; i) 1. oxalyl chloride (3 equiv), DMF (3 equiv), C6H6, 238C, 2 h; 2.
2-mercaptopyridine-1-oxide sodium salt (1.2 equiv), tBuSH (10 equiv), DMAP (20 mol%), C6H6, 808C, hn, 50 min, 64% over 2 steps; j) LiHMDS
(1.7 equiv), HMPA (5 equiv), THF, ꢀ788C, 1.5 h; then [Cp2Fe]PF6 (3 equiv), ꢀ208C, 20 h, 45–51%; k) 1. 48% aq. HF (20 equiv), MeCN, 238C,
3 d; 2. DMP (3 equiv), CH2Cl2, 238C, 3 h, 26% over 2 steps (four reactions); l) K2CO3 (6 equiv), MeOH, 08C, 30 min, 85%; m) K2CO3 (6 equiv),
MeOH, 238C, 2 h, 17%; n) K2CO3 (6 equiv), MeOH, 08C to 238C, 3 h, 14%. Bn=benzyl, [Cp2Fe]PF6 =ferrocenium hexafluorophosphate,
DMAP=4-dimethylaminopyridine, DMF=dimethylformamide, DMP=Dess–Martin periodinane, HMDS=hexamethyldisilazane, HMPA=hexame-
thylphosphoramide, mCPBA=3-chloroperbenzoic acid, pyr=pyridine, TBS=tert-butyldimethylsilyl.
cycloadducts; unlike most furan Diels–Alder reactions that
typically afford exo cycloadducts because they are reversible
and under thermodynamic control.[14] Exposure of the minor
diastereomer from the furan Diels–Alder reaction to the
above reaction conditions only led to starting material, thus
suggesting that this furan Diels–Alder reaction is not
reversible and therefore not susceptible to equilibration.
This may result from the low propensity of 25 to enolize under
the reaction conditions, a requirement for retrocycloaddition.
Unfortunately, attempts to generate carboxylic acid 26
directly from 25 by treatment with either lithium hydro-
peroxide or hydroxide led to undesired reactions involving
the oxygen bridge. However, the oxazolidinone auxiliary
could be removed with concomitant TBS removal using
aqueous acid. Reintroduction of the TBS group and decar-
boxylation using the Barton conditions[15] delivered 27 in 64%
yield.
consistently led to products in which the oxygen bridge had
been compromised. From these experiments we learned that
the lithium enolate of 27 was not stable above ꢀ208C, and we
reasoned that an oxidant capable of electron transfer below
ꢀ208C would be required to achieve dimerization of 27. With
this in mind, the powerful oxidant [Cp2Fe]PF6 was exposed to
the lithium enolate of 27 at ꢀ208C for 20 h to finally afford
the C2-symmetric molecule 28 as a single diastereomer.[16] The
oxidative enolate coupling was fully stereoselective and
1
afforded only the desired a,a adduct. The H and 13C NMR
spectra of 28 clearly indicates that a C2-symmetric compound
had been formed. A strong NOE (% enhancement) between
H2 and the phenylsulfonyl group, which is also possible with
H2 in an endo orientation, supported the stereochemical
assignment at C2 and C2’. Compound 28 is only moderately
stable to silica gel chromatography, which affected the yield of
isolated product.
Following our earlier success with 7-oxanorbornanone, we
initially attempted to achieve oxidative dimerization of the
enol silane of 27, but this failed and gave primarily starting
material. The lithium enolate of 27 could be generated using
LHMDS, but further exposure to common oxidants used for
such reactions, including copper(II) salts, iron(III) salts, and
iodine, which all required warming to 08C or higher,
Double processing of 28 commenced with desilylation
using aqueous HF in acetonitrile followed by oxidation with
Dess–Martin periodinane to afford 29. To our surprise, a
water molecule added to the C1 and C1’ ketones, forcing
compound 29 to exist as a cyclic hydrate. Apparently, the
cyclic hydrate has a stabilizing effect because, unlike 28,
compound 29 is now stable to silica gel chromatography.
1682
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1680 –1684

Angewandte
Chemie
X-ray crystallographic analysis provided confirmation of
structure 29, most importantly the stereochemical assignment
at C2 and C2’ and the presence of the cyclic hydrate
(Figure 1).[17]
are part of a vinylogous 1,2-diketone system. Also, 1,2-
diketones are known to have a higher propensity for
hydration.[18] In order to convert 31 (a lomaiviticin A-type
system) into 32 (a lomaiviticin B-type system) it may be
necessary to remove the C7 and C7’ ketones, thereby reducing
the propensity for hydration at C1 and C1’.
In conclusion, an enantioselective synthesis of the central
ring system of lomaiviticin A has been achieved using a
stereoselective oxidative enolate coupling of a 7-oxanorbor-
nanone to solve the problems of b elimination and facial
selectivity. One important discovery from this study is that
[Cp2Fe]PF6 promotes oxidative enolate coupling at low
temperature before oxygen-bridge-opening occurs; these
mild conditions will be useful in a total synthesis involving
more complex units. The resulting C2-symmetric molecule
was converted to the central ring system of lomaiviticin A
using a mild base-initiated cascade reaction. Our discovery
that the lomaiviticin A central ring system forms a stable
cyclic hydrate may also prove to be useful in a total synthesis
of 1 since it effectively prevents interconversion to the
lomaiviticin B structure, thus allowing the C3 and C3’ tertiary
carbinols to remain free for eventual glycosylation. We are
using the reactions and strategies reported herein to achieve
total syntheses of 1 and 2, which will enable more careful
scrutiny of their chemical and biological properties.
Figure 1. Representation of 29 derived from X-ray crystallographic
analysis (some hydrogen atoms are omitted). C gray, H blue, O red, S
yellow.
Three transformations are required to convert 29 to the
ꢀ
central ring structure found in 1: installation of the C5 C6
ꢀ
double bond, fragmentation of the C6 O bond, and con-
version of the C4 phenylsulfonyl group to an ether with
inversion of configuration. Based upon studies on the
monomeric model system (14), we found that all three
transformations can be achieved by exposure to K2CO3 in
methanol. Treatment of 29 with K2CO3/MeOH at 08C!238C
afforded 31 in 14% yield. Insight into the sequence of events
for conversion of 29 to 31 came from conducting the reaction
at 08C, which cleanly afforded the E2 reaction product 30 in
85% yield (Scheme 5). Re-exposure of 30 to K2CO3 in MeOH
Received: October 18, 2007
Published online: January 23, 2008
Keywords: antitumor agents · cycloaddition · natural products ·
.
radical reactions · total synthesis
[1] H. He, W.-D. Ding, V. S. Bernan, A. D. Richardson, C. M.
Ireland, M. Greenstein, G. A. Ellestad, G. T. Carter, J. Am.
Chem. Soc. 2001, 123, 5362 – 5363.
[2] a) R. S. Laufer, G. I. Dmitrienko, J. Am. Chem. Soc. 2002, 124,
1854 – 1855; b) K. S. Feldman, K. J. Eastman, J. Am. Chem. Soc.
2005, 127, 15344 – 15345; c) K. S. Feldman, K. J. Eastman, J. Am.
Chem. Soc. 2006, 128, 12562 – 12573.
[3] a) S. J. Gould, N. Tamayo, C. R. Melville, M. C. Cone, J. Am.
Chem. Soc. 1994, 116, 2207 – 2208; b) S. Mithani, G. Weeratunga,
N. J. Taylor, G. I. Dmitrienko, J. Am. Chem. Soc. 1994, 116,
2209 – 2210; c) J. Marco-Contelles, M. T. Molina, Curr. Org.
Chem. 2003, 7, 1433 – 1442.
Scheme 6. Attempted dehydration of the lomaiviticin A central ring
system. Reagents andconditions: a) Sc(OTf) (cat.), MeOH, 658C or
3
TsOH (cat.), 4-Š M.S., C6H6, 808C. TsOH=p-toluenesulfonic acid.
[4] K. C. Nicolaou, R. M. Denton, A. Lenzen, D. J. Edmonds, A. Li,
R. R. Milburn, S. T. Harrison, Angew. Chem. 2006, 118, 2130 –
2135; Angew. Chem. Int. Ed. 2006, 45, 2076 – 2081.
[5] P. Vogel, J. Cossy, J. Plumet, O. Arjona, Tetrahedron 1999, 55,
13521 – 13642.
[6] A. Warm, P. Vogel, Tetrahedron Lett. 1986, 27, 5615 – 5616.
[7] Y. Ito, T. Konoike, T. Saegusa, J. Am. Chem. Soc. 1975, 97, 649 –
651.
[8] CCDC 669765 contains the supplementary crystallographic data
for this structure. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via http://www.
ccdc.cam.ac.uk/data_request/cif.
[9] T. Zoller, D. Uguen, Eur. J. Org. Chem. 1999, 1545 – 1550.
[10] H. H. Wasserman, G. M. Lee, Tetrahedron Lett. 1994, 35, 9783 –
9786.
at 238C induced displacement of the allylic phenylsulfonyl
groups with inversion of configuration to afford 31 in 17%
yield (41% over two SN2 reactions). Similar to 29, the cyclic
hydrate of 31 is stable to silica gel chromatography. In fact,
attempts to dehydrate compound 31 with acid catalysis and
convert it to 32, the core system of lomaiviticin B, have only
led to recovery of starting material, or if more forcing
conditions were applied, they led to decomposition
(Scheme 6). We cannot exclude the possibility that equilibri-
um is established between 31 and 32 under acid catalysis,
favoring 31, and eventually decomposition pathways predom-
inate under harsh conditions. We suspect that the stability of
the cyclic hydrate of 31 results from a combination of effects:
the C1 and C1’ ketones are held in close proximity, and they
[11] D. A. Evans, K. T. Chapman, J. Bisaha, J. Am. Chem. Soc. 1988,
110, 1238 – 1256.
Angew. Chem. Int. Ed. 2008, 47, 1680 –1684
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1683

Communications
[12] D. A. Evans, J. S. Tedrow, J. T. Shaw, C. W. Downey, J. Am.
f) U. Jahn, P. Hartmann, I. Dix, P. G. Jones, Eur. J. Org. Chem.
2001, 3333 – 3335; g) U. Jahn, P. Hartmann, I. Dix, P. G. Jones,
Eur. J. Org. Chem. 2002, 718 – 735; h) U. Jahn, P. Hartmann, E.
Kaasalainen, Org. Lett. 2004, 6, 257 – 260; i) P. S. Baran, J. M.
Richter, D. W. Lin, Angew. Chem. 2005, 117, 615 – 618; Angew.
Chem. Int. Ed. 2005, 44, 609 – 612.
Chem. Soc. 2002, 124, 392 – 393.
[13] H. Danda, M. M. Hansen, C. H. Heathcock, J. Org. Chem. 1990,
55, 173 – 181.
[14] C. O. Kappe, S. S. Murphree, A. Padwa, Tetrahedron 1997, 53,
14179 – 14233.
[15] D. H. R. Barton, D. Crich, W. B. Motherwell, Tetrahedron 1985,
41, 3901 – 3924.
[17] CCDC 669499 contains the supplementary crystallographic data
for this structure. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via http://www.
ccdc.cam.ac.uk/data_request/cif.
[18] For an example of a similar cyclic hydrate of 1,2-diketones, see
M. A. Ernst-Russell, J. A. Elix, C. L. L. Chai, A. C. Willis, N.
Hamada, T. H. Nash III, Tetrahedron Lett. 1999, 40, 6321 – 6324.
[16] For examples of enolate oxidation using ferrocenium cation, see
a) U. Jahn, P. Hartmann, Chem. Commun. 1998, 209 – 210; b) U.
Jahn, J. Org. Chem. 1998, 63, 7130 – 7131; c) U. Jahn, M. Muller,
S. Aussieker, J. Am. Chem. Soc. 2000, 122, 5212 – 5213; d) P. Q.
Nguyen, H. J. Schafer, Org. Lett. 2001, 3, 2993 – 2995; e) U. Jahn,
P. Hartmann, J. Chem. Soc. Perkin Trans. 1 2001, 2277 – 2282;
1684
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1680 –1684