Journal of the American Chemical Society
Article
a
amine A, is considerably more challenging in synthetic terms
for it contains an additional side chain terminating in a primary
amine as well as a quinoline moiety annulated to the
macrocyclic ring.
Scheme 3. Building Blocks (Ingenamine Series)
RESULTS AND DISCUSSION
■
Building Blocks and the Michael/Michael Cascade
Revisited. Our initial design had tried to match the reactivity
of the Michael acceptor and donor in the best possible way
(Scheme 2).1 To this end, the highly electrophilic alkylidene β-
ketoester 5 was deemed optimal, as it was expected to render
the first Michael addition particularly facile;50 at the same time,
the malonate-type anion 7 primarily formed should expedite
the subsequent intramolecular Michael addition that closes the
diaza-tricyclic core of 8.51 An allyl ester was chosen as the
exocyclic activating group in 5, as it lends itself to
decarboxylative allylation with formation of the required
quarternary C6-bridgehead position.30,31,52,53
Although successfully reduced to practice on gram scale, we
ultimately found this setting suboptimal. The fact that the
formation of the 2,7-diazadecaline core had to be carried out in
a stepwise manner was tentatively attributed to the fact that
enolate 7 derived from a 1,3-dicarbonyl derivative is actually
too stabilized. Although it likely engages in the second Michael
addition, it also seems to be too good a leaving group; as such,
it renders this step reversible, thus preventing an efficient
cyclization cascade from occurring.1 Moreover, one might want
to revisit the choice of the allyl ester: although the palladium-
catalyzed decarboxylative allylation worked perfectly well in
terms of yield and selectivity, this reaction is limiting in
conceptual terms, as it does not allow other substituents to be
attached to the bridgehead position. This handicap had already
surfaced in our original campaign: while the allyl handle was
ideal for the synthesis of nominal xestocyclamine A ((−)-3) via
hydroboration/cross coupling, a stepwise transposition of the
double bond by hydroboration/oxidation and subsequent
Wittig olefination with formation of 12 was necessary on the
way to actual xestocyclamine A ((−)-2).1
In an attempt to remedy these issues, to save steps in the
longest linear sequence, and gain higher flexibility at the same
time, it was decided to include the appropriate handle for
macrocyclization in the Michael acceptor from the very
beginning. For proof-of-concept, compound 17 carrying a
butenyl substituent was prepared by O-silylation of commercial
14 followed by regioselective C−H oxidation with RuO2 cat./
NaIO4 (Scheme 3).1,54 The elaboration of 15 thus formed into
16 was also high yielding. A particularly noteworthy improve-
ment concerns the use of a modified Saegusa-type decarbox-
ylative dehydrogenation catalyzed by Pd2(dba)3 to set the
internal double bond of the Michael acceptor 17;55 the
formation of the original building block 5, which is much more
electrophilic and hence more sensitive, had mandated
stoichiometric selenation/selenoxide elimination for this
purpose.1
a
Reagents and conditions: (a) TBSCl, Et3N, CH2Cl2, quant.; (b)
RuO2 (6 mol %), NaIO4, EtOAc/H2O, 55%; (c) LiHMDS, allyl
chloroformate, THF, −78 °C, 94%; (d) 4-bromo-1-butene, Cs2CO3,
DMF, 94%; (e) Pd2(dba)3·CHCl3 (5 mol %), MeCN, reflux, 83%; (f)
LiHMDS, allyl chloroformate, toluene, −78 °C → 0 °C, 50% (R =
COOMe); (g) NaH, diallyl carbonate, THF, 45% (R = Bn); (h) 1-
iodo-3-pentyne, K2CO3, acetone, reflux, 36%; (i) 1-iodo-3-pentyne,
Cs2CO3, DMF, 91%; (j) ClCOOMe, toluene, reflux, quant.; (k)
Pd2(dba)3·CHCl3 (5 mol %), MeCN, reflux, 96%
amine site from interfering with any downstream process, most
notably the palladium catalyzed Saegusa oxidation, 21 was
reacted with ClC(O)OMe in refluxing toluene: under these
conditions, the benzyl group is cleanly swapped for the
carbamate and the stage set for yet another palladium-catalyzed
decarboxylative dehydrogenation.55 This sequence to the
desired fragment 22 is considerably more efficient than the
original route. In addition, it is flexible with regard to the side
chain; this aspect is best illustrated by the total synthesis of
nominal njaoamine I outlined below, which would not have
been successful otherwise.
Gratifyingly, the redesigned building blocks could be coaxed
to participate in a true Michael/Michael cascade (Scheme
4).28,29 After some experimentation, it was found that the
reaction is best performed with LiOtBu as the base; although
the N-Boc group was also cleaved, the desired diaza-tricyclic
product 23 was the only discrete isomer detected in the crude
material (after reprotection); reduction with NaBH4 furnished
alcohol 24 as a single isomer, which is more polar than 23 and
hence easier to purify by flash chromatography. This key
compound was isolated in analytically pure form in 53% yield
over two steps (740 mg scale, single largest batch), which
marks yet another significant improvement over our original
foray.1 The base-induced elimination of the derived mesylate
25 required harsh conditions but proceeded cleanly; as the
−NBoc group was concomitantly cleaved, the stage was nicely
set for subsequent N-alkylation of 26 with formation of 27.
As expected, this substrate readily succumbed to RCAM
when treated with a catalyst generated by mixing complex 29
and trisilanol 30, which had also served our original study in
this field.56−58 Although the true nature of the active species
generated in situ is unknown, all evidence suggests that the
chosen silanolate ligand partly cross-links the active metal
fragments; the resulting mixture of (cyclo)oligomeric alkyli-
The preparation of the required Michael donor was also
much improved in practical terms. Direct alkylation of 19a (R
= −COOMe), as previously described, does work on scale but
is rather inefficient (36% yield);1 this poor outcome is
attributed to the good leaving group properties of the
carbamate adjacent to the enolate C formed upon deprotona-
tion. To circumvent this problem, the alkylation was carried
out with the N-Bn protected derivative 19b, which, indeed, led
to a much more favorable outcome. To prevent the basic
C
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX