Communications
block in obtaining the ketolactone with a free
hydroxy group at C5 had been our inability to
remove the benzyl ether with the ketolactone
moiety in place. As we were unable to deprotect
the alcohol from its benzyl ether, it would be
necessary to have the latent hydroxy group at
C5 protected in an acid-labile form (20).
Accordingly, in this connection, following pro-
tocols very similar to those described above and
previously,[1] compound 20, which contains the
acid-labile MOM group on the C5 alcohol, was
synthesized. Treatment of this compound with
acid did indeed result in liberation of the
primary alcohol by cleavage of its protecting
group. However, instead of prompting cycliza-
tion to lactonamycinone (2), the principal prod-
uct of treatment of 20 with acid in methanol was
21 (Scheme 5). Thus, the hemiacetal cyclization
step that we sought had occurred, but was
accompanied by opening of the lactone ring. We
Scheme 6. Proposed new strategy to form the E and F rings.
Our campaign commenced with the addition of
(1,3-dioxolan-2-ylmethyl)magnesium bromide[10]
to aldehyde 26[1] followed by oxidation to produce
quinone 27 in 55% yield over two steps
(Scheme 7). As described earlier, the strategic
alcohol function at the future C3’ in 27 directed the
Tamura reaction with the previously described
homophthalic anhydride 28,[1] thereby affording 29
in 42% yield. Subsequent oxidation followed by
Scheme 5. Failed attempt at lactonamycinone.
dihydroxylation afforded 30 (89%, two steps).
With diol 30 in hand, treatment with aqueous HCl
cannot speak rigorously as to the timing of these events, but
preliminary experiments suggest that lactone degradation
precedes hemiacetal formation.
(3n) led to deprotection of the octyloxymethyl ether with
concomitant intramolecular attack of the tertiary alcohol at
the proximal acetal to afford butenolide 24 in 82% yield.
Deprotection of the benzyloxy ether produced the pivotal
intermediate 31 (Scheme 8). Gratifyingly, when this com-
pound was heated at reflux with HCl/dioxane/methanol,
acetal 32 was obtained in 51% yield (over two steps, based on
recovered starting material). Hydrolysis of 32 gave rise to an
anomeric mixture of lactols in quantitative yield, with the C3’
methoxy group intact.
At this point, a new approach was needed to overcome the
limitations encountered in the previous routes. In compound
10, the hexacyclic ring system had been reached; however, all
efforts to introduce the angular methoxy were unsuccessful.
Methyl enol ether 19 contains the required methoxy func-
tional group and the correct oxidation state at C3’, but the
cyclization of the E ring did not occur, apparently due to the
stable nature of the vinylogous system. Ketolactone 20
contains the functionality to cyclize the E ring with the
addition of the angular methoxy group, but the lactone moiety
is vulnerable to the harshly acidic conditions of the cyclization
reaction.
The key departure in our new approach was the presen-
tation of the F ring as a vinylogous butenolide 24 (Scheme 6).
Given the instability of ketolactone 20, it was hoped that the
stable, structurally simpler vinylogous butenolide could
function as a more durable surrogate. Moreover, since
decarboxylation was a serious problem in reaching the
lactone, the vinylogous butenolide would prove to be more
amenable to a straightforward construction of the F ring. The
hope was that methanol-induced cyclization would lead to 25.
Conversion of 25 into 2 should be straightforward, even
though the oxidation state would require adjustment.
During the planning stages of the new strategy, the task of
oxidizing a lactol to a lactone seemed straightforward.
Unfortunately, what had initially been anticipated to be an
uneventful oxidation turned out to be anything but routine.
After a systematic survey of many oxidants, it was discovered
that only TEMPO[11]-mediated oxidation was able to com-
plete the transformation to lactonamycinone (2) (58% yield).
1
The H and 13C NMR spectra of the synthetic aglycone of
lactonamycin were similar to those of the natural product,
lactonamycin, which bears the carbohydrate attachment.[12]
Conclusive comparison data came from the identical NMR
spectra of synthetic peracetylated lactonamycinone 33 and
naturally derived peracetylated lactonamycinone, kindly
provided by Tekeuchi and co-workers.
In the total synthesis of lactonamycinone, it was found
that there was considerable kinetic resistance to fashioning a
5632
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2003, 42, 5629 –5634