the a unit contained a 6,8-dioxabicyclo[3.2.1]octane skeleton
(Figure1), which is contained in known serinolipids, didemni-
serinolipids AꢀC11 and cyclodidemniserinol trisulfate.12
Further, the COSY spectrum and HMBC cross peaks from δ
1.52/1.74 (H2-13) to δ 38.3 (C-15) suggested that another
linear carbon chain was connected at C-14, but the COSY
correlation did not define the length of the carbon chain. From
these data, the a unit possesses the 6,8-dioxabicyclo[3.2.1]-
octane core with two long carbon chains, one of which has an
ester linkage. In a second unit b, the COSY spectrum implied
the presence of 1,2,3-trisubstituted propyl [δ 4.22 (dd, J =
12.0, 6.5 Hz, H-10), 4.38 (dd, J = 12.0, 3.7 Hz, H-10), 5.27 (m,
H-20),and4.03(2H,t,J= 5.8 Hz, H2-30)] and 1,2-disubstitued
ethyl [δ4.27 (2H, m, H2-100) and 3.65 (2H, m, H2-200)] moieties,
and the 13C chemical shifts clearly suggested that the five car-
bons were bonded to hetero atoms. A singlet signal at δ 3.23
matched nine hydrogens, which defined the presence of three
equivalent methyl groups. The signal (δ 3.23) had an HSQC
cross peak with a carbon at δ 54.6, which indicated that the
singlet signal was a methyl signal accommodated in a tri-
methylamino group. The HMBC spectrum showed a correla-
(δ 70.9) in 1 clearly showed that the linear aliphatic chain
attached to C-14 in the a unit was connected to the c unit
through an ether linkage. In addition, ESI MS2 and MS3
spectra of 1 were measured to confirm the connection among
the a, b, and c units and their observed peaks established a
planar structure, which was analyzed by NMR experiments
(Figures 2 and S10, Supporting Information). The lack of
coupling between H-9 and H-10 and between H-10 and H-11
revealed that the dihedral angles, H-9/H-10 and H-10/H-11,
were approximately 90ꢀ. Therefore, the relative configurations
were determined to be 9R*,10S*,11R*,14S*, which was sup-
ported by the NOE correlation between H-9 and H-11. In
order to determine the absolute configuration, a modified
Mosher’s method13 was applied to acidic methanolysis pro-
duct 13 prepared from 1 (Scheme 1). The differences in
chemical shifts (Δδ = δS ꢀ δR) between (R)- and (S)-MTPA
esters (14a and 14b, respectively) clearly indicated 9R,10S,
11R,14S,30S for 1. In order to determine the absolute config-
uration of C-20, the alkaline hydrolysis product 16 prepared
from 1 was subjected to esterification to give dibenzoate 17
(Scheme 2). The CD spectrum of 17 was identical to that
of a dibenzoate 19 derived from a commercially available
3-sn-phosphatidylcholine (Figure 3). Thus, the structure of 1,
14
tion from NMe3 to C-200 and a 1Hꢀ N HSQC (long-range)
spectrum showed correlations from H2-100 and NMe3 to NMe3
(Figures 1 and S8, Supporting Information), which showed
that the trimethylamino group was attached to C-200. The
presence of an acetoxy group at C-10 was revealed by HMBC
cross peaks from δ 4.22/4.38 (H2-10) and 2.03 (10-OCOMe) to
δ 172.3 (10-OCOMe). It was noted that the carbon signals
at δ 60.5 (C-100), 64.8 (C-30), 67.3 (C-200), and 71.7 (C-20)
appeared as doublet signals, which indicated that these car-
bons were coupled with a phosphorus atom. This was con-
firmed by detection of a 31P signal at δ ꢀ0.19, which showed
Scheme 1. Distribution of δΔSꢀR Values for the MTPA Deri-
vatives 14a and14b
31
1Hꢀ P HMBC cross peaks with δ 3.65 (H-200), 4.03 (H-30),
and 4.27 (H-100) (Figures 1 and S9, Supporting Information).
These data established the structure of a 20-O-substituted
glycerophosphocholine moiety. HMBC correlations from δ
5.27 (H-20) to δ 167.3 (C-1) clearly showed that the a and b
units were connected through an ester linkage at C-20. A third
unit cwas suggested to be a substituted propyl group (δH 3.51/
3.58 (H2-29), 3.43 (H-30), and 3.63/3.68 (H2-31); δC 70.9
(C-29), 56.0 (C-30), and 63.0 (C-31)), in which heteroatoms
were attached to each carbon judged by their chemical shifts.
Based on the molecular formula of 1, the c unit was implied to
consist of C3H8NO5S and the linear aliphatic chain linked to
C-14 in the a unit was composed of the remaining residue
C13H26, which thereby indicated that the c unit was a sulfo-
nated serinol derivative (Figure 1). Siladenoserinol G (7) has
the same molecular formula as 1, and the NMR spectra clearly
indicated that a structural difference between 1 and 7 was only
the sulfonated position (δH 3.62/3.66 (H2-29), 3.64 (H-30),
and 4.12/4.22 (H2-31); δC 68.6 (C-29), 52.3 (C-30), and 66.1
(C-31) in 7). In order to determine the sulfonated positions
in 1 and 7, the 13C NMR spectra in CD3OD and CD3OH
were measured. The differences [ΔδC = δC(CD3OD) ꢀ δC-
(CD3OH)] in the 13C chemical shift of C-31 were ꢀ0.24 and
0.00 in 1 and 7, respectively, which definitively showed that 1
had a hydroxy group at C-31 and a sulfamate group at C-30.
In contrast, 7 had amino and sulfate groups at C-30 and C-31,
respectively. The HMBC correlation from H2-28 to C-29
including the absolute configuration, was determined. The
structure determination of other siladenoserinols 2ꢀ6 and
8ꢀ12 is described in the Supporting Information.
The inhibitory effects of siladenoserinols AꢀL (1ꢀ12) on
p53ꢀHdm2 interaction were tested by ELISA10 (Table 1). In
spite of their structural similarities, they differed in their IC50
values from 2.0 to 55 μM. Compound 1 and 2 exhibited the
most significant inhibition, 8 was the next most potent
inhibitor, and 7 and 12 were the weakest inhibitors. Compar-
ing the sulfamate derivatives (R2 = SO3H, R3 = H) and the
corresponding sulfate derivatives (R2 = H, R3 = SO3H), the
sulfamate derivatives were more potent than the sulfate de-
rivatives (IC50 values: 2.0/53 μM for 1/7; 2.5/9.3 μM for 8/9;
(13) Ohtani, I.; Kusumi, T.; Kashman, Y.; Kakisawa, H. J. Am.
Chem. Soc. 1991, 113, 4092–4096.
Org. Lett., Vol. XX, No. XX, XXXX
C