32
Z. Szendi et al. / Steroids 67 (2002) 31–38
The present report describes the reaction of (20R)-3,
in the transition state had one and only one negative eigen-
value.
20,26-trihydroxy-27-norcholest-5-en-22-one [22] (1) un-
der Wolff-Kishner reduction conditions to produce quan-
titatively the unexpected (20R,22RS)-27-norcholest-5-
ene-3,20,22-triol (2). These results suggested that a
base-catalyzed 1,5-hydrogen shift mechanism promoted
this reaction. Quantum chemical calculations by AM1
[23] and PM3 [24] methods were used in the present
study to obtain the theoretically optimum molecular ge-
ometries of the starting material and the most probable
transition states for the reactive intermediates. The mo-
lecular structures corresponding to the calculated energy
minima were validated by comparing them with those
obtained earlier by X-ray crystallography [25]. This, in
turn, permitted us to calculate the activation energies
from which we could derive the most probable reaction
paths for the C-22 diastereoisomers of 2.
Measurement of the amounts of the two diastereoisomers
of 2 in the crude product by 1H NMR spectroscopy permit-
ted us to corroborate the quantum chemical predictions.
Quantum chemical data also predicted the most likely path
and charge for a hydrogen atom during an intramolecular
migration. The results of both chemical and computational
experiments were compared to appraise the validity of our
hypothesis for the mechanism of this unusual Wolff-Kish-
ner reduction.
2.1. (20R,22RS)-27-Norcholest-5-ene-3-20,22-triol 3,22-
diacetate (3)
Hydrazinehydrate (85%, 2.0 ml) and KOH (1 g; 19.46
mmol) were added to 0.70 g (1.67 mmol) of (20R)-
3,20,26-trihydroxy-27-norcholest-5-en-22-one (1) [22] in
20 ml tri(ethylene glycol). The reaction mixture was heated
for 0.5 h at 165°C. Then, the condenser was removed, and
the reaction bath temperature was kept at 195°C for 2 h,
during which time all of the starting material was con-
sumed.1 After cooling the mixture to room temperature, it
was poured into a mixture of ice-cold 3 ml of conc. HCl in
97 ml of water whereupon a solid precipitated. The precip-
itate was washed with water until neutral and then dried to
yield 0.59 g (1.46 mmol, 87% yield) (C26H44O3, MW:
404.638). Mp: 108–111°C. Compound 2 was obtained as a
light yellow powder that migrated as one spot on TLC.
Product 2 was acetylated at room temperature with 1.5
ml of pyridine and 1 ml of acetic anhydride, and after 24 h,
it was poured onto a mixture of 2 ml of conc HCl in 48 ml
of water. The precipitate was filtered and washed with water
until neutral. 1H NMR analysis of the crude products
showed ratio (20R, 22S)-3: (20R, 22R)-3-triol 3,22-diac-
etates to be 54:46. The dry product (Mp: 143–150°C) was
dissolved in benzene, evaporated onto the surface of silica
(200–400 mesh), and purified by column chromatography
by elution with 4% acetone-petroleum ether (v/v). The
(20R,22R)-3 and (20R,22S)-3 products had the same Rf
value. The weight of crystalline isomeric product mixture
was 0.446 g (54.6% based on 1). The ratio of (20R,22S)-3:
(20R,22R)-3 triol 3,22-diacetates after column chromatog-
raphy was 61:39. Mp: 147–151°C. Rf: 0.72 (30% acetone/
benzene, v/v). IR (KBr) cmϪ1: 3450 (OH), 1730 and 1740
(C ϭ O), 1250 (–C–O–C–), 1030 (–C–O(H)). (20R,22S)-3
was obtained from (3) by fractional crystallization from
methanol. Mp: 160–163°C. The same isomer could simi-
larly be crystallized from acetone/petroleum ether. 1H NMR
(CDCl3): (20R,22S)-3, see Table 1. The NMR data and
assignments for (20R,22R)-32 were obtained using signals
due to (20R,22RS)-3 as a reference. (20R,22R)-3: ␦: 0.87 (s,
3H, H-18), 0.88 (t, 3H, H-26, J ϭ 6.92 Hz), 1.01 (s, 3H,
H-19), 1.25 (s, 3H, H-21), 4.60 (m, 1H, 3-␣H), 4.84 (dd,
1H, H-22, J ϭ 12.0 Hz), 5.37 (m, 1H, H-6), 2.03 and 2.10
(s, 3H, 3- and 22-OAc, respectively) ppm. MHϩ calculated
2. Experimental
Reagents and solvents were used as received from
commercial sources unless otherwise indicated. Analyti-
cal TLC was carried out on 0.20 mm (Merck) silica gel
precoated plates (60-F254) and visualized under UV light
or by spraying the plates with concentrated sulfuric acid
and then heating them in an efficient fume hood. Melting
points (Mp) were determined on a Kofler apparatus and
are uncorrected. IR spectra were recorded in KBr pellets
on a UNICAM SP 200 instrument. The starting material
(1) was prepared by methods described previously [22].
NMR spectra were recorded with a BRUKER DRX 500
instrument. Chemical shifts are expressed in ppm relative
to TMS. Mass spectra were recorded with electron impact
at 70 eV, and the fast atom bombardment (FAB) mass
spectra were recorded using VG ZAB 2SEQ hybrid tan-
dem mass spectrometers (Cs ion gun 30 keV).
Quantum chemical calculations were performed at the
level of AM1 [23] and PM3 [24] semi-empirical quantum
chemical methods with the PcMOL and MOPAC 6.0 pro-
grams. The keyword TS in MOPAC was used to find the
transition states. The gradient norm was less than 0.1 in the
stationary points of the potential energy surface (PES; min-
ima and transition state). The calculated force matrices of
the stable isomers were positive definite. The force matrices
1 The reaction went to completion at 160°C.-
2 The 1H NMR spectrum of (20R,22R)-3 was obtained with substrac-
tion of the 1H NMR spectrum of the pure (20R,22S) compound from the 1H
spectrum of the mixture.