Total Synthesis of Bafilomycin A1
FULL PAPER
to obtain alkene 27 by base-induced elimination of the cor-
responding mesylate under a variety of different conditions
were not successful. Displacement of the mesylate with phe-
nylselenide and in situ oxidation to the selenoxide furnished
alkene 27 in low and varying yields. A three-step sequence
proved to be more promising: oxidation of alcohol 25 with
TPAP and NMO as stoichiometric oxidant gave the inter-
mediate ketone in 93% yield. Access to vinyl triflate 26 was
addition reaction. Instead of an alkene as latent aldehyde
functionality, we intended to form the aldehyde in the ulti-
mate step by oxidative cleavage of an appropriate diol.
Two routes with similar efficiency to the required chiral
allylic alcohol 29 were independently developed (Scheme 7).
Wittig olefination of known bis-acetal protected (R)-glycer-
aldehyde 30,[40] prepared in two steps from d-mannitol, pro-
vided the corresponding (Z)-alkene with 92:8 d.r. in 65%
yield over two steps. Deprotection using a mixture of acetic
acid and water liberated diol 31, the primary hydroxyl group
of which was selectively protected as trityl ether, giving al-
lylic alcohol 29 in 88% yield. Alternatively, the same allylic
alcohol 29 could be synthesized under the conditions of the
zinc triflate mediated addition of terminal alkynes to alde-
hydes,[16] starting from gaseous propyne and previously de-
scribed trityl-protected aldehyde 32.[41] In this transforma-
tion, the argon atmosphere above the reaction was simply
exchanged for propyne after standard preformation of the
catalyst. The generated zinc acetylide then added efficiently
to aldehyde 32, giving propargylic alcohol 33 in 88% yield
and 94% ee.[42] Carefully monitored Lindlar reduction of the
alkyne then efficiently installed the (Z)-alkene.
ꢀ
Scheme 5. Model studies on C C bond scission for the synthesis of the
C14–C20 aldehyde fragment. a) TPAP, NMO, 4 ꢂ MS, CH2Cl2, 08C !
RT, 2.5 h, 93%; b) KHMDS, THF, ꢀ788C, 30 min, then PhNTf2, ꢀ788C,
45 min, 80%; c) PdACHTUNGTRENNUNG(OAc)2 (10 mol%), HCO2H, NEt3, DMF, 608C,
15 min, 77%; d) O3, MeOH, ꢀ788C, 50 min, then NaBH4, ꢀ788C ! RT,
1 h; e) PivCl, pyridine, RT, 11 h, 65% (two steps).
then obtained by regioselective enolate formation using
LDA and trapping with PhNTf2, providing the desired prod-
uct 26 in 80% yield under optimized conditions.[39] Palladi-
um-catalyzed reduction with formic acid as reducing agent
then furnished the targeted alkene 27. However, careful
monitoring of the reaction progress in the reduction of vinyl
triflate 26 was crucial to prevent partial epimerization of the
resulting vinyl isoxazoline 27 on prolonged exposure to the
reaction conditions. Mechanistically, the epimerization to
the thermodynamically preferred 3,4-trans-substituted isoxa-
zoline trans-27 presumably occurs via intermediate forma-
tion of the allyl palladium species (Scheme 6). Oxidative
cleavage of the double bond in 27 was then accomplished
using ozone. After reductive work-up with NaBH4, the inter-
mediate primary alcohol was protected as the pivaloate to
give fully protected intermediate 28.
Scheme 7. Synthesis of chiral allylic alcohol 29. a) Ph3PEtBr, nBuLi,
THF, 08C, 30 min, then aldehyde 30, ꢀ788C ! RT, 6.5 h, d.r. 92:8, 65%
(two steps); b) AcOH/H2O 1:1, 608C, 6.5 h, 93%; c) TrCl, pyridine, cat.
DMAP, CH2Cl2, RT, 5 h, 88%; d) ZnACTHNUGRTNEUNG(OTf)2, (+)-NME, NEt3, toluene,
RT, 2 h, then propyne, RT, 15 min, then aldehyde 32, RT, 1 h, 94% ee,
88%; e) Lindlar catalyst, H2, EtOAc, RT, 25 min, 97%.
The oxime reaction partner 34 which served as precursor
for the nitrile oxide was assembled using an Evans aldol ap-
proach (Scheme 8).[43] Treatment of (R)-propionyl oxazolidi-
none 35 with Bu2BOTf and NEt3 and subsequent aldol reac-
tion with acetaldehyde provided multigram quantities of the
syn-aldol product 36 in 84% yield as a single diastereoiso-
mer. Subsequent silyl protection under standard conditions
with TBDPSCl and imidazole in DMF furnished silyl ether
38 in 95% yield.
The failure to access the desired alkene functionality di-
rectly from the secondary alcohol led to the modification of
the allylic alcohol reaction partner in the nitrile oxide cyclo-
Removal of the chiral auxiliary proved to be more diffi-
cult than expected. Attempted reductive cleavage with
LiBH4 in the presence of either EtOH[44] or H2O[45] afforded
the expected alcohol 39[46] in only 50 and 61% yield, respec-
tively. Under all conditions tested, formation of the corre-
sponding amide, resulting from reductive opening of the iso-
xazoline auxiliary, could be observed.[47] A two-step ap-
proach via the intermediate benzyl ester, originally reported
by Evans[48] slightly improved the overall yield of the trans-
Scheme 6. Epimerization of vinylisoxazoline 27.
Chem. Eur. J. 2012, 18, 3598 – 3610
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3601