A R T I C L E S
Nicolaou et al.
of the desired ∆7,8 isomer (38). No optimization was attempted
at this point, neither for the installation of the endocyclic double
bond nor for the deoxygenation, since we were only one step
away from the targeted molecule. Indeed, desilylation of 38
through the action of TBAF afforded the desired ABCD
fragment 26, which was further elaborated to the fully saturated
compound 4 by catalytic hydrogenation (H2, 10% Pd/C) in 95%
yield. NMR spectroscopic comparison of synthetic 26 with its
naturally derived counterpart suggested that the A regions of
the two molecules were closer, but still not identical with each
other. Furthermore, the hydrogenated product 4 differed in its
1H NMR spectrum from the naturally derived material, recon-
firming the nonidentity of the natural and synthetic ABCD
domains and underscoring the fact that something more than
the double bond location was wrong with the proposed structure.
Figure 3 summarizes the state of affairs at this stage of the
azaspiracid-1 project. The first and second proposed structures
of the ABCD domain, as represented by fragments 3 and 26,
respectively, were proven wrong by synthesis. However, those
studies led us to believe that the endocyclic double bond was
at the C7-C8 site, rather than the C8-C9 site as originally
proposed. Our reality check had also pointed to a thermody-
namically most stable structure, as both the natural azaspiracid-1
and its ABCD fragment 3 (derived by degradation of the natural
product by Satake et al.4) were stable under acidic conditions,
in contrast to our synthetic ABCD fragments, whose fleeting
nature under such conditions was in line with their lower
thermodynamic stability. In addition to this information, we also
had knowledge of an NOE exhibited by both the natural
azaspiracid-1 and its degradatively derived ABCD fragment 3
between H-6 and the C-14 methyl group. This intelligence
allowed us, through manual molecular modeling, to narrow
down the most likely structures for the ABCD domain from
the 128 (27) possible diastereomers to the two shown in Figure
3 (i.e., 39 and 40). Thus, both of these structures are thermo-
dynamically favored by two anomeric effects, and they should,
by virtue of space proximity, exhibit the obligatory NOE effect
between H-6 and the C-14 methyl group protons as indicated
on their structures, 39 and 40 (see Figure 3). The first to be
targeted for synthesis was isomer 39.
The construction of 39 began with ketophosphonate 41
(derived from L-malic acid as previously described9) and
aldehyde 42 and proceeded as summarized in Schemes 5 and
6. The sequence leading to aldehyde 59 (Scheme 5) is similar
to that described elsewhere9 for the preparation of a close relative
of this compound and, therefore, will not be described here in
more detail.
Scheme 6 describes the advancement of aldehyde 59 to
ABCD fragment 67, beginning with the addition of the lithium
anion derived from dithiane 60 and n-BuLi to afford a mixture
of diastereomeric alcohols (61). Oxidation of this mixture with
DMP led to the cyclization precursor, ketone 62 (93% yield),
whose exposure to TMSOTf in CH2Cl2 at -78 °C, followed
by warming to -30 °C, furnished the ABCD tetracycle 63 as
the major product (70% yield). NMR spectroscopic analysis of
63, however, showed that the stereocenter at C-13 had a
configuration opposite to that of the desired structure. This was
further confirmed by analyzing the spectral data of a descendant
Scheme 5. Synthesis of Aldehyde 59, a Precursor for the Third
Proposed Structure of the ABCD Domain of Azaspiracid-1a
a Reagents and conditions: (a) 41 (0.67 equiv), 42 (1.0 equiv), LiCl (1.3
equiv), i-Pr2NEt (1.0 equiv), MeCN, 25 °C, 12 h, 86% based on 41; (b)
LiAlH4 (10 equiv), LiI (8.0 equiv), Et2O, -100 °C, 30 min, 98%; (c) AcOH:
H2O (2:1), 25 °C, 5 h, 97%; (d) NIS (5.0 equiv), NaHCO3 (10 equiv), THF,
0 °C, 2.5 h, 70%; (e) TBDPSCl (1.4 equiv), Et3N (3.0 equiv), 4-DMAP
(0.1 equiv), CH2Cl2, -10 f 0 °C, 3 h, 90%; (f) TBSOTf (1.6 equiv), 2,6-
lutidine (4.0 equiv), CH2Cl2, -10 °C, 30 min, 100%; (g) H2, Raney-Ni (30
equiv), EtOH, 25 °C, 1 h, 99%; (h) H2, 20% Pd(OH)2/C (25% w/w), EtOH,
25 °C, 3 h, 88%; (i) DMP (2.0 equiv), CH2Cl2, 25 °C, 2 h, 99%; (j) 52 (6.0
equiv), THF, -78 f -10 °C, 3 h, 92%; (k) DMP (2.0 equiv), CH2Cl2, 25
°C, 2 h, 98%; (l) (HOCH2)2 (7.0 equiv), triethyl orthoformate (3.0 equiv),
p-TsOH (0.1 equiv), 55 °C, 98%; (m) TBAF (1.0 M in THF, 4.0 equiv),
THF, 25 °C, 48 h, 96%; (n) TBDPSCl (1.4 equiv), Et3N (3.0 equiv),
4-DMAP (0.1 equiv), CH2Cl2, 0 °C, 97%; (o) TESCl (1.5 equiv), imidazole
(3.0 equiv), 4-DMAP (0.1 equiv), CH2Cl2, 0 °C, 30 min, 98%; (p) OsO4
(0.03 equiv), NMO (2.0 equiv), t-BuOH:THF:H2O (10:2:1), 25 °C, 12 h,
then NaIO4 (5.0 equiv), pH 7 buffer, 25 °C, 5 h, 89%.
compound (67), which was prepared as follows. Reaction of
63 with PivCl, py, and 4-DMAP furnished pivaloate ester 64
(95% yield), from which the dithiane group was removed (NBS,
2,6-lutidine, 80% yield) to afford ketone 65. The reduction of
65 with NaBH4 in MeOH proceeded stereoselectively, leading
to hydroxy pivaloate 66 in 81% yield. Finally, exposure of 66
to TFA in CH2Cl2 at 25 °C for 4 h resulted in an equilibrium
mixture9 in which the C-13 epimer 67 predominated in a 55:45
(9) Nicolaou, K. C.; Qian, W.; Bernal, F.; Uesaka, N.; Pihko, P. M.; Hinrichs,
J. Angew. Chem., Int. Ed. 2001, 40, 4068.
9
2864 J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006