€
and Hunig’s base in THF. After stirring for 1 h at room
data of the diastereomeric precursors 25b and 28bꢀ32b
all very similar to those of 25a and 28aꢀ32a. The Stille
macrocyclization was performed under the same condi-
tions, affording macrocycle 1b as a colorless oil, albeit
contaminated by two minor diastereoisomers, which most
likely arise from epimerization of the N,O-acetal stereo-
genic centers C-10 and C-32, and a small amount of AsPh3.
In this instance the product was soluble in CD3OD en-
abling a direct comparison with natural ‘upenamide to be
made, but unfortunately, the NMR data were significantly
different to those of the natural compound for each of
the three compounds in this mixture (see Supporting
Information).
Thus, the NMR data of both 1a and 1b does not match
that reported for ‘upenamide, suggesting that neither is the
natural product.24 It is regrettable that neither synthetic
structure could be unambiguously verified by X-ray crys-
tallography, but as both compounds are highly unstable
and degrade quickly in solution,25 this was not possible,
despite considerable effort. We therefore cannot state
categorically that ‘upenamide has been misassigned, but
at the very least, this work means that the proposed struc-
tures of ‘upenamide should be treated with caution and
that a structural re-evaluation may be required.
temperature, the reaction mixture was analyzed by high
resolution ESIþ mass spectrometry, and we were delighted
to observe a major signal consistent with‘upenamide20 and
only trace amounts of the starting material. Following stan-
dard workup and column chromatography, the desired
macrocycle 1a was isolated in 74% yield as a white amor-
phous solid.
Unfortunately a direct comparison between the spectro-
scopic data of 1a and natural ‘upenamide could not be
made, given that 1a is insoluble in CD3OD, the NMR
solvent used to characterize ‘upenamide by the isolation
chemists. Its NMR data was instead measured in CDCl3
and showed considerable differences to that of the natural
product, which allied to its solubility properties, makes it
likely that 1a is not ‘upenamide.21
Aside from these comparisons, all of the spectroscopic
data accrued for 1a supports its structural assignment as
being that shown. The methods described above that
were used to confirm the stereochemsitry of each of the 8
stereogenic centers in its precursors all continue to apply
throughout the synthesis, and crucially in 1a, suggesting
that any unexpected epimerization is unlikely. Evidence
for the geometry of the C-16/C-17 alkene can be found in
the 1H NMR spectra, as H-16 appears as a discrete signal
with a large coupling constant typical of an E-alkene.22
Unfortunately the geometry of remaining two alkenes of
the triene could not be directly observed because of the over-
lap of signals, although all of the corresponding protons in
its precursor 32a were clearly visible, supporting the all-
trans-assignment.23
In summary, we have completed the synthesis of the two
proposed structures of ‘upenamide 1a and 1b. Noteworthy
steps include the DIA reaction used to install the A ring, a
20-membered ring Stille macrocyclization and each of the
two SnCl2 2H2O-mediated stereocontrolled N,O-acetal
3
formations. Future work will focus on establishing unam-
biguously the structures of macrocycles 1a and 1b
before, if necessary, seeking to re-evaluate the structure of
‘upenamide through total synthesis.
The synthesis of the other proposed structure 1b
(Figure 1) was therefore completed using the same route
described above, but using S-Alpine Borane to generate
19b, the epimer of alcohol 19a (Schemes 3 and 4).
The synthesis proceeded similarly well, with the spectroscopic
Acknowledgment. We thank EPSRC for studentship
(K.A.G.) and postdoctoral funding (W.P.U., EP/G068313/1),
and AstraZeneca for CASE support (K.A.G.). We are also
grateful to Dr. Adrian C. Whitwood (University of York)
for X-ray crystallography studies, Dr. Ian Ashworth, Jan
Cherrywell and Andy Hard (all AstraZeneca) for assistance
with the enzymatic desymmetrization and chiral HPLC
(20) Found MHþ 523.3521, C32H47N2O4þ requires 523.3530.
(21) We wish to thank Mr. Wesley Yoshida, the only author of the
‘upenamide isolation paper (ref 1) remaining at the University of
Hawaii, for his attempts to find a sample of natural ‘upenamide or
any NMR data recorded in CDCl3. Unfortunately neither could be
found.
ꢀ
analysis and Dr. Cecilia Menard-Moyon and Mark Reid
(both University of York) for preliminary studies.
ꢀ
(22) δH 6.31 (1H, dd, J = 14.4, 10.3, H-16)
(23) J(16ꢀ17) = 14.3 Hz, J(18ꢀ19) = 18.8 Hz, J(20ꢀ21) = 15.0 Hz.
(24) It was considered that amine salt formation may account for the
differences in NMR data; however, while the addition of TFA to each of
1a and 1b led to a change in their NMR spectra, neither accorded with
natural ‘upenamide, and product decomposition was clearly evident.
(25) Within a few hours of standing in CDCl3, a clean sample of 1a
contained additional signals in its 1H NMR spectrum, thought to arise
from C-10 and/or C-32 epimerization. After 1 week in DCM/hexane
(attempted crystallization), complete degradation was observed.
Supporting Information Available. Synthetic proce-
dures and spectral data. This material is available free
The authors declare no competing financial interest.
Org. Lett., Vol. 15, No. 2, 2013
265