Communication
Scheme 3. Mechanism of cycloaddition of 14.
tional diastereoselectivity issues, as well as the nucleophilic re-
activity of a proximate cyclopentene functional group. Gratify-
ingly, the key intramolecular [4+3] cycloaddition catalyzed by
TESOTf proved to be versatile enough and compatible with all
of the pre-existing functionalities, and afforded the desired cy-
cloadduct 15 in 87% yield as a single diastereomer, a reaction
which was carried out on as large as a 10 g scale. In this
manner, we successfully completed the assembly of rings B, C,
and D of the target molecule.
Scheme 5. Formal total synthesis of 1 and 2 intercepting 24 in the Nico-
laou–Chen total synthesis: a) NaBH4, MeOH, À788C, 100%; b) TBSCl, imida-
zole, DMAP, DMF, 258C, 86%; c) Et3N·3HF, THF, 258C, 96%; d) Dess–Martin
periodinane, NaHCO3, CH2Cl2, 258C, 100%; e) TFA, CH2Cl2, reflux, 85%.
Catalytic hydrogenation of cycloadduct 15 led to the forma-
tion of a fully saturated product, but with the undesired cis-
fused junction at the CD rings. To establish the trans-fused ste-
reochemistry at C13ÀC14, the siloxy group at C17 must be de-
protected. Therefore, dehydration of cycloadduct 15, followed
by desilylation with camphorsulphonic acid, provided 16 in
72% yield over two steps (Scheme 1). Catalytic hydrogenation
of triene 16 was not chemoselective, and competitive reduc-
tion at the electron-deficient double bond was also observed.
On the other hand, in the presence of Crabtree’s catalyst, only
the reduction of ring D and the dihydrofuran occurred to
afford diol 17 in 96% yield. Bis-oxidation of 17 with Dess–
Martin periodinane afforded the expected ketoaldehyde, which
spontaneously underwent intramolecular aldol cyclization
during chromatographic purification on silica gel, to afford 18
containing the full tetracyclic framework of the cortistatins in
84% overall yield. The stereoselectivity of the aldol cyclization
could be rationalized as shown in Scheme 4.
With an efficient and scalable route to tetracyclic 20 avail-
able, we proceeded to the synthesis of the Nicolaou–Chen in-
termediate 24 (Scheme 5). Reduction of ketone 20 with
sodium borohydride afforded b-alcohol 21a in quantitative
yield and as a single diastereomer. Quantitative protection
with TBSCl, followed by selective deprotection of the less hin-
dered of the triethylsiloxy groups, yielded alcohol 22 in 96%
yield. Oxidation of 22 with Dess–Martin periodinane under buf-
fered conditions afforded ketone 23, which was subjected to
treatment with trifluoroacetic acid to effect b-elimination to
afford 24 in 16.0% overall yield and in 19 steps by the longest
linear route. This synthetic sample of 24 exhibited identical
spectroscopic data (1H NMR, 13C NMR, IR, MS and optical activi-
ty) compared with those reported. According to the Nicolaou–
Chen synthesis, intermediate 24 can be converted to cortista-
tins A (1) and J (2) by a 12- and 14-step sequence, respective-
ly.[4c,d]
Although the cortistatins could be obtained via 24, evidently
this was a rather early intermediate, as more than 10 steps are
still required to complete the total synthesis of the natural
products. Therefore, we pursued a more efficient approach to
an advanced, isoquinoline-substituted synthetic intermediate,
the Yamashita–Hirama intermediate 31, which would consti-
Scheme 4. Aldol cyclization yielding 18. DMP=Dess–Martin periodinane.
tute
a more rapid access to both cortistatins A and J
(Scheme 6).
Ketone 20 was converted to vinyl triflate 25 in 95% yield by
using NaHMDS and PhNTf2. Suzuki–Miyaura coupling between
30 and pinacolboronate 26 afforded isoquinoline 27 in 85%
yield.
Whereas under the Luche conditions,[14] both the C17 and
C19 carbonyl groups of 18 were reduced, a chemoselective re-
duction was achieved by a hydroxyl-directed triacetoxyborohy-
dride reduction, affording 1,3-diol 19 in excellent yield, and as
a single diastereomer, distinguishing the C17 carbonyl group
for subsequent elaborations.[15] The stereochemistry of 19 was
deduced through computer modelling of both the cis- and
trans-diol diastereomers, which confirmed that the NOE be-
tween H1 and H19 as observed would only be found in cis-19
(see the Supporting Information). Bis-protection of cis-19 with
chlorotriethylsilane provided the key common intermediate 20
in 97% yield.
The conversion of 27 to 28 required a chemoselective reduc-
tion of the cyclopentene over the cyclohexene moiety. Notably,
the corresponding transformation in several previous literature
syntheses was the penultimate step of the total synthesis, that
is, reduction of 3!1 (Figure 1), which suffered either from low
conversion/yield and/or chemoselectivity.[4a,c–e] However, the
desired chemo- and stereoselective reduction was accom-
plished by catalytic hydrogenation of 27 to give 28 as a single
diastereomer in 84% yield, with no over-reduction. The stereo-
Chem. Eur. J. 2015, 21, 14287 – 14291
14289
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim