A. Fꢁrstner et al.
methyl ketone 25 was reacted with aldehyde 20 to provide
product 31 (Scheme 8). This transformation takes advantage
of the acetyl group in the substrate, which migrates from the
phenolic site in 20 to the more basic alkoxide generated in
the initial step. The released phenoxide then promotes elim-
ination of the resulting acetate ester to give the required
enone 31. Subsequent treatment with HCl/MeOH provided
compound 32 as a single isomer in 92% yield through an
obviously well-orchestrated triple-deprotection/1,4-addition/
spirocyclization cascade; no sign of trans-esterification was
observed.
Scheme 9. a) OsO4 (5 mol%), NMO, acetone; b) PbACHTNUGRTNE(UGN OAc)4, CH2Cl2; c)
NaClO2, NaH2PO4, tBuOH/H2O 2:1, (2E)-butene; d) TMSCHN2,
MeOH, 08C, 96% (over four steps); e) Bu4NF, THF; f) H2 (1 atm),
Pd(OH)2 cat., EtOAc, 85% (over both steps).
At this juncture we considered a minor diversion from
our initial synthesis, the latter stages of which had suffered
from the fact that the free phenol in the assembled core was
highly susceptible to oxidation.[33] As such, protection of the
phenol as its TBS ether 33 resulted in a vast improvement,
with the three-stage conversion to 34 proceeding in high
yield. Despite the advantages offered by the new protecting
group, its lability required some optimization of the reaction
conditions. Specifically, during the oxidative cleavage of the
intermediate diol formed by OsO4-catalyzed dihydroxylation
of 33, the TBS group could be completely retained only if
the stoichiometry of the oxidant was rigorously controlled.
was easily attained by subjecting the crude aldehyde formed
by the oxidative cleavage of the olefinic terminus in 33 to
Pinnick oxidation[36] and esterification with trimethylsilyl di-
azomethane. Subsequent cleavage of the silyl group in 37
and hydrogenolysis of the remaining benzyl ester gave the
truncated congener 38 for biological testing. With this mate-
rial and our synthetic samples of analytically pure (ꢁ)-1,
(+)-1 and the 22S-diastereomer thereof in hand (cf. Experi-
mental Section), we hope to resolve a puzzle resulting from
contradictory reports from the groups of Stierle and Snider.
Whereas the isolation team claimed highly selective cytotox-
icity for berkelic acid against the OVCAR-3 tumor cells, as
outlined in the Introduction,[4] the latter group found that
their synthetic material did not show significant activity at
10ꢁ5 m concentration when profiled in the NCI 60 cancer
cell-line assay.[12b] The results of our investigation will be re-
ported in a separate publication in due time.
To this end, one equivalent of a stock solution of PbACTHNUTRGNEUNG(OAc)4
in CH2Cl2 was added dropwise and the reaction monitored
closely by TLC. Reduction of the resulting crude aldehyde
with NaBH4 in MeOH also led to partial deprotection; how-
ever, employing a milder mixed-phase THF/H2O system left
the TBS ether untouched. Conversion of alcohol 34 thus
formed into iodide 35 again occurred with partial TBS de-
protection, which was attributed to residual hydrogen iodide
in the commercial iodine and/or to impurities in the imida-
zole. Gratifyingly, the use of freshly sublimed and recrystal-
lized reagents allowed for full retention of the protective
group.
Conclusion
A concise and productive total synthesis of berkelic acid (1)
has been accomplished that takes the structure revision pre-
viously proposed by our group into account and remedies
the shortcomings of the original route.[10] Specifically, the
tetracyclic core 32 of the target was assembled as a single
diastereomer by a triple-deprotection/1,4-addition/spiroace-
talization cascade, which proceeded with remarkable effi-
ciency. The required cyclization precursor 31 was obtained
from segments 20 and 25 by an aldol condensation, in which
the phenolic acetate moiety adjacent to the aldehyde in 20
not only serves as a passive protecting group but also takes
an active and productive role. The preparation of the
methyl ketone 25 benefits from the Collum–Godenschwager
modification of the ester enolate Claisen rearrangement,
which proved superior to the classical Ireland–Claisen pro-
cess in terms of diastereoselectivity. Although our previous
work had shown that the elaboration of 32 into the target
can be accomplished without any extra protecting group ma-
nipulation, the short detour of blocking the free phenolic
site as a TBS-ether was found to be highly beneficial. Whilst
this tactic added two extra steps to the longest linear se-
quence, it tempered the sensitivity of the pentasubstituted
arene ring toward oxidation and hence greatly improved the
For the final fragment coupling, the pre-nucleophile 35
was subjected to halogen–metal exchange at low tempera-
ture followed by the addition of aldehyde 30. Oxidation of
the resulting secondary alcohol with Dess–Martin periodin-
ACHTUNGTRENNUNG
ane[34] afforded fully protected berkelic acid 36. Fluoride-in-
duced desilylation followed by hydrogenolysis of the benzyl
ester over Pearlmanꢀs catalyst[35] proceeded uneventfully to
provide berkelic acid (ꢁ)-1. The recorded spectra for the
synthetic samples were fully consistent with the published
data, and the optical rotation of 1 was also in good agree-
ment with those for natural and synthetic berkelic acid re-
ported in the literature.[4,12,13] It should be noted, however,
that we encountered problems in the purification, handling
and storage of berkelic acid such that oxidative degradation
was a common frustration. Only the use of carefully de-
gassed solvents and storage in a matrix of frozen and de-
gassed benzene resolved these issues. Suffice it to mention
that the synthesis of non-natural berkelic acid (+)-1 has
also been accomplished by the same route (see the Experi-
mental Section).
Finally, we prepared an analogue deprived of the side
chain carrying the quaternary center (Scheme 9). This goal
12138
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 12133 – 12140