Total synthesis and absolute configuration of the novel microtubule-stabilizing agent peloruside A

Article abstract of DOI:10.1002/anie.200351145

A unique configuration-dependent mechanistic switch, observed when epimeric seco-acids are subjected to typical Mitsunobu lactonization conditions, is just one of the many unexpected, yet insightful surprises of a synthetic adventure culminating in the fi

Full text of DOI:10.1002/anie.200351145

Communications
OH
PGO
8
OH
OH
OPG OPG
11
MeO
MeO
H
H
O
O
Me Me
OH
Me Me
OPG
CHO
13
OMe
Me
14
O
Me
MeO
MeO
CO2Me
1
O
O
I
HO
PGO
Me
Peloruside A (1)
II
20
Me
H-bonding
dipole–dipole
interaction
OPG
9
OPG
11
OH
H
OH
11
Me
Me
H
8
8
MeO
MeO
9
O
H
O
H
Me
OPG
Me
OH
A
B
syn pentane
interaction
Scheme 1. Peloruside A (1): Structure and Strategy. PG=protecting
group.
Peloruside A (1) is a potent cytotoxin and induces apoptosis
in cultured mammalian cells.[2] Of potential clinical relevance
is the recent observation that 1 affects microtubule dynamics
in a manner similar to paclitaxel (Taxol).[2b] The relative
stereochemistry of 1 was assigned based on extensive data
gathered from a variety of high-field NMR experiments,
although the absolute configuration remained to be solved. In
addition to the presence of an array of ten stereogenic centers
and a Z-configured trisubstituted olefin, we anticipated that
one of the main challenges of a peloruside synthetic venture
would be to find an efficient solution to the sterically crowded
C8–C11 unit. In the solution-phase structure of 1, this region
adopts a conformation determined by optimal minimization
of syn and gauche pentane interactions with potential intra-
molecular hydrogen bonding between the C9 and C11
hydroxy groups (A, Scheme 1). On a synthetic pathway, the
introduction of the C8, C9, and C11 hydroxy functionality will
by necessity introduce additional syn pentane and dipole–
dipole interactions depending on the protection scheme used
(B, Scheme 1). Herein, we communicate a solution to these
synthetic challenges that culminates in the first total synthesis
of (ꢀ)-peloruside A (1), and in doing so, document the
absolute configuration of the natural isolate (opposite to the
one shown in Scheme 1).[3]
Strategically, we opted for a late-stage aldol coupling
between a fully functionalized aldehyde I with methyl ketone
II. This approach is driven by the perception that access to
diastereomeric seco-acids, through reduction of a common
C15 ketone, will increase the odds for a successful macro-
cyclization by accessing two mechanistically distinct path-
ways, namely acylative or invertive macrolactonization. A
concise, enantioselective synthesis of fragment II is outlined
in Scheme 2. The known homoallylic alcohol 2, prepared in
enantiopure form according to an elegant procedure descri-
bed by Hoveyda et al.,[4] was acylated with methacryloyl
chloride followed by ring-closing olefin metathesis with
Grubb's second generation catalyst 4.[5] The resulting lactone
5 then provides a valuable entry to the Z-trisubstituted enone
Macrolide Total Synthesis
Total Synthesis and Absolute Configuration of the
Novel Microtubule-Stabilizing Agent
Peloruside A**
Xibin Liao, Yusheng Wu, and Jef K. De Brabander*
Recently, we became interested in peloruside A (1,
Scheme 1), a marine metabolite isolated from sponge speci-
mens of the species Mycale by Northcote and co-workers.[1]
[*] Prof. Dr. J. K. De Brabander, Dr. X. Liao, Dr. Y. Wu
Department of Biochemistry
University of Texas Southwestern Medical Center at Dallas
5323 Harry Hines Blvd, Dallas, TX 75390-9038 (USA)
Fax: (+1)214-648-6455
E-mail: jdebra@biochem.swmed.edu
[**] Financial support provided by the Advanced Research Program of
the Texas Higher Education Coordinating Board, the Robert A.
Welch Foundation, and the Howard Hughes Medical Institute
(junior faculty support). J.K.D.B. is a fellow of the Alfred P. Sloan
Foundation.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1648
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200351145
Angew. Chem. Int. Ed. 2003, 42, 1648 –1652

Angewandte
Chemie
6 by treatment with methyllithium and silylation of the
primary alcohol.
Me
HO
O
a) CH2CMeC(O)Cl
75%
1 step
ref. [4]
Our approach towards the densely functionalized tetra-
hydropyranyl fragment is shown in Scheme 3. Addition of
aldehyde 9 (derived from 7[6] as shown) to (Z)-alkoxyallyl-
borane 10 at low temperature produced the desired homo-
allylic alcohol diastereomer 11 (> 10:1).[7] The relative 1,3-syn
stereochemical relationship was demonstrated by 13C NMR
analysis of the acetonide derivative 13 (resonances for the
isopropylidene carbon atoms at d = 20.0, 30.3, and
98.8 ppm).[8] Compound 11 was advanced to aldehyde 15 by
methylation (!12), oxidative debenzylation (DDQ), and
oxidation of the resulting primary alcohol. Addition of
aldehyde 15 to a cold solution of the lithium enolate derived
from 14 and Dess–Martin oxidation[9] of the resulting aldol
product fashioned b-diketone 16. Transformation to dihydro-
pyranone 17 ensued by stirring 16 in an acidic toluene
O
O
Me
Me
2
3
99%ee
PCy3
Ph
RuCl2
b)
Me
O
O
Me
Mes
Mes
N
N
c) MeLi
d) TBSCl
50-70%
4
O
52-63%
TBSO
Me
Me
6
5
Scheme 2. a) CH2CMeC(O)Cl, iPr2NEt, DMAP, CH2Cl2 (75%); b) 10
mol% 4, CH2Cl2 (0.0025m), reflux, 17 h (50–70% + 20% dimer
derived from 3); c) MeLi, THF, ꢀ788C; or Me3SiCH2Li, pentane,
ꢀ788C; d) TBSCl, imidazole, DMAP, DMF (52–63% from 5).
DMAP=4-dimethylaminopyridine, TBS=tert-butyldimethylsilyl.
PMBO
OR
PMBO
OTES
PMBO
OTES
PMBO
b, c
80%
d, e
f
O
OMOM
OMOM
CHO
81%
44%
RO
O
Ipc2B
9
7: R = H
8: R = TES
11: R = H
12: R = Me
a
13
MOMO
10
97%
Me
Me
Me
Me
Me
Me
g, h
77%
OBn
O
OBn
OBn
TESO
l
k
92%
O
O
O
TESO
OBn
Me
Me
i, j
OHC
HO
O
OMOM
OMOM
OMOM
OMOM
76%
MeO
MeO
MeO
MeO
O
18
17
16
15
14
OBn
OBn
OBn
Me
Me
HO
Me
Me
TESO
Me
Me
Me
Me
TESO
CHO
OMe
O
m
72% from 17
OMe
O
OMe
O
OMe
O
TESO
n, o
75%
p-s
t, u
HO
MeO
MeO
MeO
80%
90%
OMOM
OMOM
OMOM
CO2Me
OMOM
MeO
MeO
MeO
MeO
CO2Me
19
20
21
22
OH
Naph
O
Me
v
Me
94%
TESO
H
O
OMe
O
TESO
w
x, y
90%
MeO
O
O
O
H8
Me
Me Me
OMOM
CHO
81%
5H
H
MeO
MeO
OMOM
CO2Me
O
H11
Me
OTES
MeO
MeO
CO2Me
25
23
NOEs are indicated
by the dashed arrows
24
Scheme 3. a) TESOTf, 2,6-lutidine, CH2Cl2 (97%); b) cat. OsO4, NMO, acetone/H2O; c) Pb(OAc)4, pyridine, CH2Cl2 (80% from 8); d) 10, sBuLi,
THF, ꢀ788C, 15 min, then (þ)-Ipc2BOMe, ꢀ788C, 1 h, ꢀ78!08C, 1.5 h, then 9, ꢀ958C, 3 h, slowly warm to RT; 30% H2O2, NaOH, 16 h (91%);
e) NaH, MeI, DMF, ꢀ58C (89%); f) PTSA, MeOH, 20 min, remove solvent, then Me2C(OMe)2 (44%); g) DDQ, CH2Cl2/H2O, 08C (88%);
h) py·SO3, Et3N, DMSO, CH2Cl2, 08C (87%); i) 14, LDA, THF, ꢀ788C, then 15, ꢀ788C (94%); j) Dess–Martin periodinane, CH2Cl2, ꢀ108C (81%
based on recovered starting material); k) PTSA, PhMe, RT (92%); l) NaBH4, CeCl3·7H2O, MeOH, ꢀ308C; m) mCPBA, NaHCO3, CH2Cl2/MeOH,
08C (72%, two steps); n) tBuOK, MeI, THF, 08C; o) TESOTf, 2,6-lutidine, CH2Cl2 (75% from 19); p) cat. OsO4, NMO, acetone/H2O; q) Pb(OAc)4,
pyridine, CH2Cl2; r) NaClO2, NaH2PO4, 2-Me-2-butene, tBuOH/H2O; s) CH2N2, Et2O, 08C (80% from 20); t) H2, Pd/C (10%), MeOH (quant.);
u) py·SO3, Et3N, DMSO, CH2Cl2, 08C (90%); v) allylBEt2, Et2O, ꢀ108C (94%); w) TESOTf, 2-naphthaldehyde, ꢀ788C, then 2,6-lutidine, TESOTf,
08C (81%); x) cat. OsO4, NMO, acetone/H2O; y) NaIO4 on silicagel, CH2Cl2 (90% from 24). PMB=p-methoxybenzyl, TES=triethylsilyl, Tf=tri-
fluoromethane sulfonate, NMO=4-methylmorpholine-N-oxide, MOM=methoxymethyl, Ipc=isopinocampheyl, DDQ=2,3-dichloro-5,6-dicyano-
1,4-benzoquinone, LDA=lithium diisopropylamide, PTSA=p-toluenesulfonic acid, mCPBA=m-chloroperbenzoic acid.
Angew. Chem. Int. Ed. 2003, 42, 1648 –1652
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1649

Communications
solution. It proved critical to perform the subsequent
assumed to be the expected 1,3-anti aldol product based on
extensive literature precedent.[15] Evidence provided below,
however, indicates that aldehyde 25 entered this reaction with
an unprecedented bias for the formation of the 1,3-syn b-
hydroxy ketone 26 instead (80% yield). Initially unaware of
this stereochemical outcome, we continued with methyl ether
formation (!27), CBS reduction,[16] and ester hydrolysis to
reach seco-acid 28. Slow addition of this compound to a
premixed solution of PPh3 and diisopropylazodicarboxylate
instigated formation of macrolactone 29 in 40–50% yield.[17]
We were able to deduce the stereochemistry of 29 based on a
series of NOE correlations that locate H11, H13, and H15 on
the same upper side of the macrolactone ring in agreement
with the assigned C13(S) configuration (Figure 1). At this
Luche[10] reduction (!18) at ꢀ308C to prevent decomposi-
tion. Most satisfactorily, hydroxy-directed epoxidation of
crude 18 followed by in situ methanolysis of the incipient
glycal–epoxide produced one single glycoside, 19, in high
yield (72% from 17).[11] Sequential methylation (equatorial
OH)/silylation (axial OH) of the a-diol (!20), oxidative
transformation of the double bond to a C1 carboxylic acid,
and diazomethane treatment furnished methyl ester 21.
Access to the aldehyde 22 by hydrogenolytic removal of the
benzyl ether and oxidation set the stage for progression
through C11–C12 bond formation. Concerns related to the
feasibility of this critical event[12] seemed justified when
various attempts to add carbon nucleophiles to the sterically
demanding aldehyde 22 failed.[13] Ultimately, a highly effi-
cient allyl transfer was achieved with allyldiethylborane.
Remarkably, a single alcohol diastereomer, 23, was produced
with C11 configuration ascertained by NOE analysis of
spirocyclic ketal 24 and advancement of 23 to peloruside A
(see below).[14]
With fragments 6 and 25 (derived from 24 as shown) at
hand, their union and completion of the peloruside macro-
cycle seemed an attainable goal, yet unexpected surprises lay
ahead (Scheme 4). Mukaiyama-type aldol reaction of alde-
hyde 25 with the enol silane derived from methyl ketone 6
afforded almost exclusively (14:1) the compound that was
Figure 1. Chem3D representation of 29 (C in dark gray, H in light gray,
and O in black). Important NOEs are illustrated with dashed arrows.
point, we were left with the obvious challenge of correcting
the stereochemistry at C13. Various attempts to remove the
(2-naphthyl)methylidene acetal failed.[18] Undeterred, we
embraced the opportunity to explore a more attractive
avenue that would eliminate this protecting group problem
altogether, that is, we decided to advance materials with a free
C11 alcohol through the remainder of the synthesis.
As illustrated in Scheme 5, b-hydroxy aldehyde 31
(derived from 23 as shown) was a viable partner for coupling
with the enolborinate derived from 6. Most gratifyingly, a
separable 2:1 mixture of C13 epimers (32a/32b) was gen-
erated in 87% yield. This rather unselective reaction was
welcomed when contrasted with the situation described above
(cf. 25 + 6 ! 26). At this stage, we left the stereochemical
assignment unresolved, and advanced both isomers 32a/32b
individually. Surprisingly, methylation of these compounds
occurred with concomitant hydrolysis to the C9 hemiketals
33a/33b. We did not reprotect these materials but instead
positioned ourselves to explore, in line with our original
conception, two mechanistically distinct macrocyclization
pathways. Intramolecular acylation through an active ester
intermediate required access to the C15(R)-configured seco-
acids 35a/35b, while Mitsunobu-type lactonization would
initiate from the C15-epimeric relatives 34a/34b. These
requirements were best fulfilled by asymmetric reduction of
enones 33a/33b by using (S)- or (R)-B-Me-CBS-oxazabor-
Scheme 4. a) Enolsilane derived from 6 (TMSOTf, Et3N, CH2Cl2,
ꢀ108C, 25 min, aqueous extraction), 25, CH2Cl2, ꢀ788C, add BF3·Et2O,
2 h (80%, 14:1 ratio); b) Me3OBF4, 1,8-bis(dimethylamino)naphtha-
lene, CH2Cl2, RT (92%); c) (S)-B-Me-CBS (20 equiv), BH3·SMe2
(7 equiv), CH2Cl2, ꢀ308C, 1 h, !RT (4 h), add MeOH (83%, 13:1
ratio); d) 0.3n aq. LiOH, THF, RT (quant.); e) PPh3, DIAD, THF
(0.05m), RT, 15 min, add 28 (0.003m in THF) through syringe pump
over 2 h, 1 h at RT (40–50%); f) 48% aq. HF, MeCN/H2O, RT, 20 h
(88%). TMS=trimethylsilyl, CBS=catalyst named after Corey, Bakshi,
and Shibata, DIAD=diisopropylazodicarboxylate.
1650
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 1648 –1652

Angewandte
Chemie
TESO
H
TESO
H
TESO
H
OMe OH
OMe OH
OH
OH
MeO
MeO
MeO
O
O
O
Me Me
Me Me
OMOM
OH
Me
OMe
Me
Me Me
OMOM
CHO
a, b
95%
d
c
OMOM
CO2Me
23
87%
85%
O
O
MeO
CO2Me
MeO
CO2Me
MeO
TBSO
TBSO
31
32a
32b
33a
33b
2:1 ratio
Me
Me
TESO
OH
OH
TESO
H
TESO
OH
OH
O
OH
OH
MeO
H
MeO
MeO
O
80-94%
H
e, f
Me Me
OMOM
O
OMe
Me
Me Me
OMe
Me
Me Me
OMOM
OMe
Me
g
35a
OMOM
CO2H
69%
O
MeO
HO
HO
MeO
MeO
CO2H
O
TBSO
TBSO
TBSO
35a
35b
34a
34b
Me
Me
Me
38
35b
34b
1RO
OH
OH
MeO
11
h
65%
H
g
g
O
Me Me
OR3
13
OMe
Me
47%
47%
15
(–)-Peloruside A (1)
O
MeO
g
35b:34b
(1:1)
O
2RO
52%
Me
36: R1 = TES, R2 = TBS, R3 = MOM
37: R1 = R2 = R3 = H
65%
h
Scheme 5. a) Cat. OsO4, NMO, acetone/H2O; b) Pb(OAc)4, pyridine, CH2Cl2 (95% from 23); c) 6, iPr2NEt, Et2BOTf, CH2Cl2, ꢀ788C, 15 min,
ꢀ308C, 45 min, ꢀ788C, add 31, ꢀ788C, 2 h (87%, 2:1 ratio); d) Me3OBF4 (20 equiv), 2,6-di-tert-butyl-4-methylpyridine, CH2Cl2, RT (85%); e) 33a
or 33b, (R)- or (S)-B-Me-CBS (20 equiv), BH3·SMe2 (7 equiv), CH2Cl2, ꢀ308C, 1 h, !RT (4 h), add MeOH (34a: 80%, 34b: 80%, 35a: 94%, 35b:
80%); f) 0.3n aq. LiOH, THF, RT (quant.); g) PPh3, DIAD, THF (0.05m), RT, 15 min, add seco-acid (0.003m in THF) through syringe pump over
2 h, then 1 h at 08C (36: 47% from 34b or 35b, 52% from ꢁ1:1 mixture of 34b:35b, 38: 69% from 35a); h) 4n HCl, THF, RT, 3 h (37: 65%, 1:
65%).
olidine and H3B·SMe2 (80–94% yield, ratios > 12:1).[16] Note
that no concomitant reduction of the C9 hemiketal was
observed under these conditions! The corresponding carbox-
ylic acids 34a,b/35a,b were liberated by saponification.[19]
Since the C15 configuration could not be unambiguously
inferred from the Corey model,[16] we explored both epimers
34b and 35b as substrates for a Mitsunobu-type macro-
lactonization, Remarkably, the same lactone 36 was produced
from either 34b, 35b, or an equimolar mixture of both.[20] This
lactone was later shown to have a C13 epimeric relationship
to peloruside A (see below).[21] To the best of our knowledge,
this represents the first documented case of a configuration-
dependent mechanistic switch for a Mitsunobu lactoniza-
tion.[22] We speculate that geometrical/conformational con-
straints preclude the formation of C15 epimeric lactones[23]
and enforce the cyclization of substrate 35b via an acyloxy-
phosphonium intermediate (retention) and 34b via the
familiar alkoxyphosphonium intermediate (inversion).[24,25]
The structure of 36 was assigned retrospectively,[21] that is,
after the completion of peloruside A (1) from precursor 35a.
Initial evidence for a C13 epimeric configuration surfaced,
however, when the 1H NMR data of the deprotected lactone
37 (aq. HCl/THF) did not match those reported for the
natural product. By inference, we realized peloruside A
would fortuitously derive from the major aldol diastereomer
32a. In the event, Mitsunobu lactonization of 35a (69%)
followed by simultaneous cleavage of the MOM and silyl
protecting groups in lactone 38 (65%), did produce a pure
compound (1) with spectroscopic properties (1H and
13C NMR, IR, HRMS) identical to those of the natural
isolate.[1] However, the optical rotation of synthetic peloru-
side A (1) was of opposite sign ([a]2D3 = ꢀ16; c = 0.12 in
CH2Cl2) to the one reported for the natural product ([a]2D0 =+
16; c = 0.30 in CH2Cl2).[1] Also, synthetic 1 (up to 10 mm) had
no effect on the growth of cultured human tumor cell lines
(SK-MEL-5, HeLa).[26] Based on these results, we assign the
absolute configuration of natural (þ)-peloruside A as 2S, 3R,
5R, 7R, 8R, 9R, 11S, 13S, 15S, 18R.
In summary, we accomplished the first total synthesis of
(ꢀ)-peloruside A and documented the absolute configuration
of the dextrorotatory natural product. The advancement of
minimally protected intermediates merits recapitulation, and
was key to a successful endgame. A particularly noteworthy
configuration-dependent mechanistic switch was observed
Angew. Chem. Int. Ed. 2003, 42, 1648 –1652
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1651

Communications
BF3·Et2O) or acetate aldol reactions (lithium or boron enolates)
resulted in recovery of starting material.
[14] This fortunate stereochemical outcome was not predicted a
priori, nor are we able to offer a plausible rationale for it. For a
related, nonselective allylation, see ref. [11a].
[15] D. A. Evans, M. J. Dart, J. L. Duffy, M. G. Yang, J. Am. Chem.
Soc. 1996, 118, 4322 – 4343, and references therein.
[16] For an excellent review, see: E. J. Corey, C. J. Helal, Angew.
Chem. 1998, 110, 2092 – 2118; Angew. Chem. Int. Ed. 1998, 37,
1986 – 2012.
when epimeric seco-acids 34b/35b were subjected to typical
Mitsunobu lactonization conditions. Current efforts focus on
gaining additional insights into the mechanistic aspects of this
reaction, and the synthesis of naturally configured peloruside
and tailored derivatives for structure–function studies.
Received: February 10, 2003 [Z51145]
Keywords: antitumor agents · configuration determination ·
.
cyclization · natural products · total synthesis
[17] O. Mitsunobu, Synthesis 1981, 1 – 28.
[18] We have explored a variety of solvolytic (protic or Lewis acid
mediated), oxidative, and reductive pathways. Other protecting
group schemes (cyclic and acyclic) were briefly investigated with
no indication for success. Details will be included in a later
disclosure.
[19] Reesterification with CH2N2 demonstrated that no epimeriza-
tion had occurred. The Z configuration of the double bond was
maintained throughout the synthesis and verified by NOE
interactions.
[1] L. M. West, P. T. Northcote, C. N. Battershill, J. Org. Chem. 2000,
65, 445 – 449.
[2] a) K. A. Hood, B. T. Bäckström, L. M. West, P. T. Northcote,
M. V. Berridge, J. H. Miller, Anti-Cancer Drug Des. 2001, 16,
155 – 166; b) K. A. Hood, L. M. West, B. RouwØ, P. T. Northcote,
M. V. Berridge, S. J. Wakefield, J. H. Miller, Cancer Res. 2002, 62,
3356 – 3360.
[3] For the synthesis of peloruside fragments, see: I. Paterson, M. E.
Di Francesco, T. Kühn, Org. Lett. 2003, 5, 599 – 602.
[4] Z. Xu, C. W. Johannes, A. F. Houri, D. S. La, D. A. Cogan, G. E.
Hofilena, A. H. Hoveyda, J. Am. Chem. Soc. 1997, 119, 10302 –
10316.
[20] No diastereomeric or transposed (allylic) macrolactones were
detected by NMR analysis of the nonpurified mixture.
[21] Clues about the stereochemistry of 36 were provided at this
point by virtue of mutual NOE enhancements between the H11–
H13 and H13–H15 proton pairs (cf. 29).
[5] a) A. K. Chatterjee, J. P. Morgan, M. Scholl, R. H. Grubbs, J.
Am. Chem. Soc. 2000, 122, 3783 – 3784; b) A. Bhattacharjee, O.
Soltani, J. K. De Brabander, Org. Lett. 2002, 4, 481 – 484.
[6] a) A. B. Smith III, K. P. Minbiole, P. R. Verhoest, M. Schelhaas,
J. Am. Chem. Soc. 2001, 123, 10942 – 10953; b) Y. Wu, X. Liao,
R. Wang, X.-S. Xie, J. K. De Brabander, J. Am. Chem. Soc. 2002,
124, 3245 – 3253.
[7] a) H. C. Brown, P. K. Jadhav, K. S. Bhat, J. Am. Chem. Soc. 1988,
110, 1535 – 1538; b) A. L. Smith, E. N. Pitsinos, C.-K. Hwang, Y.
Mizuno, H. Saimoto, G. R. Scarlato, T. Suzuki, K. C. Nicolaou, J.
Am. Chem. Soc. 1993, 115, 7612 – 7624.
[8] S. D. Rychnovsky, B. Rogers, G. Yang, J. Org. Chem. 1993, 58,
3511 – 3515.
[9] D. B. Dess, J. C. Martin, J. Am. Chem. Soc. 1991, 113, 7277 –
7287.
[10] J. L. Luche, L. Rodriguez-Hahn, P. CrabbØ, J. Chem. Soc. Chem.
Commun. 1978, 601 – 602.
[11] For related examples, see: a) P. A. Wender, J. De Brabander,
P. G. Harran, J.-M. Jimenez, M. F. T. Koehler, B. Lippa, C.-M.
Park, M. Shiozaki J. Am. Chem. Soc. 1998, 120, 4534 – 4535;
b) D. A. Evans, P. H. Carter, E. M. Carreira, A. B. Charette, J. A.
Prunet, M. Lautens, J. Am. Chem. Soc. 1999, 121, 7540 – 7552.
[12] Problematic carbon–carbon bond formation adjacent to related
a gem-dimethyl substituted C-glycoside units has been docu-
mented. For pertinent examples, see ref. [11] and M. Kageyama,
T. Tamura, M. H. Nantz, J. C. Roberts, P. Somfai, D. C.
Whritenour, S. Masamune, J. Am. Chem. Soc. 1990, 112, 7407 –
7408.
[22] For selected examples of Mitsunobu esterifications leading to
retention of configuration when hindered alcohols are involved,
see: a) C. Ahn, P. DeShong, J. Org. Chem. 2002, 67, 1754 – 1759;
b) A. B. Smith III, I. G. Safonov, R. M. Corbett, J. Am. Chem.
Soc. 2002, 124, 11102 – 11113.
[23] At a superficial level, molecular models of C15 epimeric lactones
appear to be highly strained. More comprehensive computa-
tional studies are currently pursued.
[24] Acyloxyphosphonium salts have been implicated as intermedi-
ates in Mitsunobu esterifications. For examples and mechanistic
studies, see: ref. [22] and a) D. Camp, I. D. Jenkins, J. Org. Chem.
1989, 54, 3049 – 3054; b) C. Ahn, R. Correia, P. DeShong, J. Org.
Chem. 2002, 67, 1751 – 1753; c) J. McNulty, A. Capretta, V.
Laritchev, J. Dyck, A. J. Robertson, J. Org. Chem. 2003, 68,
1597 – 1600.
[25] A referee rightfully pointed out that our observations could also
be explained by invoking a diastereoselective SN1 reaction
mechanism for the Mitsunobu lactonization with allylic alcohols.
We had tentatively ruled out this possibility because we have not
been able to detect[20] any products expected to derive to some
extent from allylic carbocation intermediates (allylic transposi-
tion, double bond isomerization, elimination). For a reference
pertinent to this issue, see: B. K. Shull, T. Sakai, J. B. Nichols, M.
Koreeda, J. Org. Chem. 1997, 62, 8294 – 8303, and references
therein.
[26] We are grateful to Dr. Michael G. Roth and Maria G. Kosfiszer
(UT Southwestern Medical Center) for assistance with the cell-
based assays.
[13] Allylmagnesium bromide reacted competitively with the methyl
ester. Lewis-acid catalyzed allyl transfer (allylSiMe3, TiCl4 or
1652
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 1648 –1652