Chemical synthesis of 7- and 8-dehydro derivatives of pregnane-3,17α,20-triols, potential steroid metabolites in Smith-Lemli-Opitz syndrome

Article abstract of DOI:10.1016/S0039-128X(02)00113-7

Pregnane-3,17α,20-triols bearing unsaturation at Δ⁷, Δ⁸, Δ^5,7, or Δ^5,8 have been tentatively identified as steroid metabolites in Smith-Lemli-Opitz syndrome (SLOS). Starting with 17α-hydroxypregnenolone diacetate, we have synthesized 13 unsaturated C₂₁ triols by four different routes in one to four steps. These multifunctional steroids were prepared by a series of regio- and stereoselective transformations chosen to minimize facile olefin isomerization and 17-deoxygenation. The results include a study of stereoselectivity in the reduction of 17α-hydroxy-20-ketosteroids, an alternative method for reducing diethyl azodicarboxylate adducts of Δ^5,7 steroids, and an efficient oxidation-isomerization of a Δ^5,7 steroid using cholesterol oxidase. The 13 triols and their synthetic precursors were fully characterized by ¹H and ¹³C nuclear magnetic resonance (NMR) spectroscopy. The NMR data, together with molecular modeling, indicated unanticipated conformational heterogeneity for two synthetic intermediates, 17α-hydroxypregna-4,7-diene-3,20-dione and 17α-hydroxy-5β-pregn-7-ene-3,20-dione. The unsaturated C₂₁ triols are useful as reference standards to study adrenal steroid production in SLOS and to develop methods for pre- and postnatal diagnosis of this congenital disorder.

Full text of DOI:10.1016/S0039-128X(02)00113-7

Steroids 68 (2003) 31–42
Chemical synthesis of 7- and 8-dehydro derivatives of
pregnane-3,17␣,20-triols, potential steroid metabolites
in Smith–Lemli–Opitz syndrome
a
a,∗
a
b
Li-Wei Guo , William K. Wilson , Jihai Pang , Cedric H.L. Shackleton
a
Department of Biochemistry and Cell Biology, Rice University, 6100 Main Street, Houston, TX 77005-1892, USA
Childrens Hospital Oakland Research Institute, 5700 Martin Luther King Jr. Way, Oakland, CA 94609, USA
b
Received 2 May 2002; accepted 27 June 2002
Abstract
Pregnane-3,17␣,20-triols bearing unsaturation at ꢀ , ꢀ , ꢀ5,7, or ꢀ5,8 have been tentatively identified as steroid metabolites in
Smith–Lemli–Opitz syndrome (SLOS). Starting with 17␣-hydroxypregnenolone diacetate, we have synthesized 13 unsaturated C21 tri-
ols by four different routes in one to four steps. These multifunctional steroids were prepared by a series of regio- and stereoselective
transformations chosen to minimize facile olefin isomerization and 17-deoxygenation. The results include a study of stereoselectivity
7
8
5
,7
in the reduction of 17␣-hydroxy-20-ketosteroids, an alternative method for reducing diethyl azodicarboxylate adducts of ꢀ steroids,
5,7
and an efficient oxidation–isomerization of a ꢀ steroid using cholesterol oxidase. The 13 triols and their synthetic precursors were
1
13
fully characterized by H and C nuclear magnetic resonance (NMR) spectroscopy. The NMR data, together with molecular model-
ing, indicated unanticipated conformational heterogeneity for two synthetic intermediates, 17␣-hydroxypregna-4,7-diene-3,20-dione and
1
7␣-hydroxy-5␤-pregn-7-ene-3,20-dione. The unsaturated C21 triols are useful as reference standards to study adrenal steroid production
in SLOS and to develop methods for pre- and postnatal diagnosis of this congenital disorder.
2002 Elsevier Science Inc. All rights reserved.
©
Keywords: Steroid metabolites; Smith–Lemli–Opitz syndrome; Conformational analysis; NMR; Mass spectrometry
1
. Introduction
intermediates accumulate, SLOS has the potential of pro-
ducing a new class of steroids if 7- or 8DHC can be utilized
as steroid precursor since the resulting metabolites would
Congenital adrenal hyperplasia and other classical en-
7
8
zyme disorders of adrenal steroid synthesis involve defects
in the pathway between cholesterol and either cortisol or
aldosterone [1,2]. An additional cause of attenuated steroid
likely have ꢀ or ꢀ unsaturation. This type of steroid was
unknown in humans but had long ago been found in the
equine [10]. In studies of steroids excreted by child and
adult SLOS patients, we found evidence for the presence of
ring B unsaturated compounds, particularly 17-hydroxylated
pregnene derivatives [11–13]. Additionally, in neonates and
during pregnancy we found evidence of 16␣-hydroxylated
dehydrosteroids, including a dehydroestriol [14]. A major
compound in SLOS urine was tentatively identified as a 7-
or 8-dehydro-3,17␣,20-triol, and this C21 steroid was poten-
tially a good analyte for diagnosing SLOS by urine analysis
[11–13]. The amount of isolated analytes was insufficient
for de novo structure determination, and no reference com-
pounds were available as aids for identification. To address
this deficiency, we have undertaken the preparation of au-
thentic standards. We describe here the chemical synthesis
of 7- and 8-dehydro derivatives of pregnane-3,17␣,20-
triols.
production was suggested by the identification of defective
7
7-dehydrocholesterol-ꢀ -reductase (DHCR7) as the cause
of Smith–Lemli–Opitz syndrome (SLOS) [3]. This genetic
disorder is characterized by mental retardation, physical
anomalies, and physiological aberrations [4] leading to
cholesterol deficiency and occasional adrenal insufficiency
[
7
(
4–6]. The hallmark of SLOS is an overproduction of
-dehydrocholesterol (7DHC) and 8-dehydrocholesterol
8DHC) [3,4,7], the two isomers being interconverted by
8
7
the ꢀ –ꢀ sterol isomerase [8,9].
In contrast to previously studied enzymatic defects of
adrenal steroid synthesis, in which only known and normal
∗
Corresponding author. Tel.: +1-713-348-4914; fax: +1-713-348-5154.
E-mail address: billw@rice.edu (W.K. Wilson).
0
039-128X/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved.
PII: S0039-128X(02)00113-7

3
2
L.-W. Guo et al. / Steroids 68 (2003) 31–42
2
. Experimental procedures and results
2.2.1. 3β,17α-Dihydroxypregna-5,7-dien-20-one
diacetate (2)
2
.1. Materials and methods
To a solution of 17␣-hydroxypregnenolone diacetate
(1; 2.03 g, 4.8 mmol) in benzene–hexane 1:1 (120 ml) was
added dibromantin (0.84 g, 2.92 mmol, 1.2 equivalents)
1
,3-Dibromo-5,5-dimethylhydantoin (dibromantin), di-
ꢀ
and 2,2 -azobisisobutyronitrile (32 mg). The mixture was
isobutylaluminum hydride (DIBALH), L-Selectride, diethyl
azodicarboxylate (DEAD), 4-methoxypyridine-N-oxide,
tetrabutylammonium bromide, tetrabutylammonium flu-
oride, palladium on carbon, and aluminum–nickel alloy
were purchased from Aldrich (Milwaukee, WI). TES buffer
N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid,
sodium salt), cholesterol oxidase from Streptomyces sp.,
catalase, and 17␣-hydroxypregnenolone diacetate (1) and
-acetate (17) were obtained from Sigma (St. Louis,
MO). Pyrophoric Raney nickel was prepared by slowly
adding sodium hydroxide pellets (27 g) to a suspension
of aluminum–nickel alloy 50/50 w/w (14 g) in water
200 ml), heating at 90 C for 1.5 h, and then wash-
◦
refluxed under nitrogen for 10 min in a preheated 100 C
oil bath and then placed in an ice bath to cool. Insoluble
material was removed by suction filtration, followed by
evaporation of the filtrate to a yellow solid. To a solution of
this yellow solid in anhydrous tetrahydrofuran (40 ml) was
added tetrabutylammonium bromide (0.4 g). The resulting
solution was stirred for 75 min under nitrogen at room
temperature. To this reaction mixture was added tetrabutyl-
ammonium fluoride (10 ml, 1 M solution in tetrahydrofu-
ran, 10 mmol, 2.1 equivalents). The resulting dark brown
solution was stirred for an additional 50 min, followed by
(
3
◦
◦
rotary evaporation at 40–45 C to a brown solid. A so-
(
lution of this solid in ethyl acetate (200 ml) was washed
with water (3 × 50 ml) and dried over anhydrous Na2SO4.
Evaporation of solvent gave crude 2 (2.03 g), which was
subjected to MPLC (370 mm × 25 mm i.d. column; elution
with hexane–ethyl acetate 95:5, 4000 ml and hexane–ethyl
acetate 92:8, 2000 ml). Fraction volumes were 20 ml. After
elution of 1 in fractions 161–198, the ꢀ5 product was
eluted in fractions 199–257, which were evaporated to give
ing with water (10 × 200 ml) and ethanol (4 × 200 ml).
4
-Methoxypyridine was prepared by hydrogenation of
4
-methoxypyridine-N-oxide over Raney nickel. Solvents
were Omnisolve grade from EM Science (Gibbstown,
NJ).
Low- and high-resolution mass spectra (HR-MS) were
acquired on a ZAB-HF or MAT-95 mass spectrome-
ter in the electron-impact mode. Mass spectral data are
presented as m/z (% relative abundance); fragment ions
representing loss of H2O, CH3, CH3COOH, CH3CHOH
,7
2
(0.97 g, 49% yield) of ∼96% purity (containing 0.5–1.3%
1 13
each of several minor olefins). H and C NMR (Tables 1
and 2); HR-MS, calculated for C23H30O3 (M–AcOH)
(side chain of 20-hydroxysteroids), CH3CO (side chain
∗
∗
354.2195, found 354.2190; MS m/z 354 (5), 294 (7), 279
of 20-ketosteroids), and combinations thereof are marked
by an asterisk (∗). Medium pressure liquid chromatog-
raphy (MPLC) was done on glass columns dry-packed
with silica gel (230–400 mesh). High performance liquid
chromatography (HPLC) was carried out on a preparative
∗
(25), 251 (100).
2.2.2. 3β,17α-Dihydroxypregna-5,7-dien-20-one (3)
To a solution of diacetate 2 (100 mg) in a 1:2 mixture of
tetrahydrofuran and methanol (24 ml) was added potassium
carbonate (140 mg). The resulting mixture was sparged with
nitrogen and then stirred at room temperature under nitrogen
for 46 h. After completion of the reaction as judged by TLC,
water (60 ml) was added. The resulting mixture was ex-
tracted with ethyl acetate (2×60 ml). The combined organic
phase was washed with water (2 × 30 ml) and dried over
anhydrous Na2SO4. Evaporation to a white solid (64 mg)
and recrystallization from methanol–water gave 3 as a white
solid (56 mg, 71% yield) of ∼90% purity (containing 1–3%
5
␮m Prodigy ODS(3) column (250 mm × 21.2 mm i.d.;
Phenomenex, Torrance, CA) with UV detection at 210 nm.
Nuclear magnetic resonance (NMR) spectra of steroids
◦
(2–15 mM in CDCl3) were acquired at 25 C as described
[15] on an AMX500, Avance 500, or Inova 500 spectrom-
1
eter and referenced to internal tetramethylsilane ( H) or
13
CDCl3 at 77.0 ppm ( C). NMR signal assignments were
made from 1D and 2D spectra as described previously [15],
and chemical shifts were corrected for effects of strong
coupling. The purity of steroid samples was estimated
1
13
1
each of several minor olefins). H and C NMR (Tables 1
and 2); HR-MS, calculated for C21H30O3 330.2195, found
by H NMR (500 MHz spectrum; methyl region and δ
2
.5–7.0 region). Steroid structures were modeled by molec-
+
∗
∗
∗
3
(
30.2200; MS m/z 330 (53, M ), 312 (60), 297 (46), 287
ular mechanics with PCMODEL 7.0 (Serena Software,
Bloomington, IN).
∗ ∗ ∗
45), 284 (45), 279 (63), 269 (100), 251 (99), 244 (76),
211 (99), 143 (73).
2
.2. Chemical syntheses
2.2.3. 17α-Hydroxypregna-4,7-diene-3,20-dione (4)
The chemical synthesis of compounds 2–15 are summa-
A solution of 3 (25 mg, 0.075 mmol) in butyl acetate
rized in Figs. 1 and 2. Because of the reported lability of
7-hydroxy-20-ketosteroids in acidic media [16], reactions
were conducted under neutral or basic conditions (except for
attempted diene isomerizations in HCl).
(7 ml) was added to a TES buffer solution (6 ml, 50 mM,
pH 7.5) containing cholesterol oxidase from Strepto-
myces sp. (50 units, 2.6 mg, 19 units/mg solid) and catalase
(40,000 units). The two-phase mixture was stirred in a vial
1

L.-W. Guo et al. / Steroids 68 (2003) 31–42
33
Fig. 1. Chemical synthesis of pregnane-3,17␣,20-triols with unsaturation in rings B and C.
at room temperature with a magnetic stirrer for 17 h. The
mixture was extracted with ethyl acetate (20 ml), and the
separated organic phase was washed with water and brine
and dried over anhydrous Na2SO4. Evaporation of solvent
gave crude 4 (24 mg) as a nearly white solid in ∼80%
and 2); HR-MS, calculated for C21H28O3 328.2038, found
+
∗
∗
328.2029; MS m/z 328 (38, M ), 310 (37), 285 (100),
∗
267 (50), 242 (53), 227 (58), 215 (43), 105 (38).
2.2.4. 17α-Hydroxy-5β-pregn-7-ene-3,20-dione (5)
To a solution of 4 (20 mg) in 4-methoxypyridine (1.5 ml)
was added palladium on carbon (20 mg; 10% Pd). This
4
,6
purity (containing 6% of the ꢀ isomer and 1–11% each
of several minor olefins). 1H and 13C NMR (Tables 1

3
4
L.-W. Guo et al. / Steroids 68 (2003) 31–42
mixture was stirred at room temperature under a hydrogen-
filled balloon for 16 h. The catalyst was removed by fil-
tration through Celite. Removal of solvent by bulb-to-bulb
distillation gave crude 5 (20 mg) as a nearly white solid
in ∼75% purity (containing 1–10% each of several minor
1
13
olefins). H and C NMR (Tables 1 and 2); HR-MS, calcu-
lated for C21H30O3 330.2195, found 330.2190; MS m/z 330
+ ∗ ∗ ∗ ∗
68, M ), 312 (12), 297 (13), 287 (85), 269 (66), 244
72), 229 (63), 211 (71), 105 (100). Several of the C NMR
(
(
1
3
signals were severely broadened, as shown in Fig. 3A.
2.2.5. 5β-Pregn-7-ene-3α,17α,20R-triol (6a) and
its 20S epimer 6b
To a solution of crude 5 (20 mg) in ether (5 ml) was added
LiAlH4 (50 mg). The resulting mixture was refluxed for 2 h,
followed by addition of cold 5% HCl (5 ml) to quench the re-
action. The organic phase was separated, washed with water,
and dried over anhydrous Na2SO4. Evaporation of solvent
gave a crude product (21 mg) comprising an 8:9 mixture of
Fig. 2. Lithium-ethylamine reductions of ꢀ5,8 DEAD adducts 10 and 15.
Table 1
1
H NMR chemical shifts of unsaturated pregnanetriols 6a–7b and their synthetic precursors 1–5a,b
5
5,7
5,7
4,7
5␤-ꢀ7
5␤-ꢀ7
5,7
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
1c
2c
3
4
5
3␣,20␤ (6a)
3␣,20␣ (6b)
3␤,20␤ (7a)
3␤,20␣ (7b)
†
†
†
†
†
†
H-1␣
H-1␤
H-2␣
1.166
1.878
1.879
1.378
1.908
1.933
1.317
1.892
1.893
1.982
1.915
2.371
2.350
2.053
1.458
2.466
2.284
1.760
1.753
1.315
1.317
†
†
†
†
†
†
1.058
1.451
1.063
1.437
1.888
1.892
†
†
†
1.899
1.892
†
†
H-2␤
H-3
1.600
4.611
1.582
4.715
1.498
3.651
1.776
3.612
1.777
3.613
1.499
3.650
1.500
3.651
†
†
†
H-4␣
H-4␤
2.344
2.317
2.514
2.367
2.481
2.285
5.805
2.694
2.349
2.11
1.364
1.361
2.475
2.286
2.478
2.286
†
†
1.515
1.518
†
†
H-5
H-6(␣)
1.817
1.655
1.361
1.622
1.361
1.633
5.386
5.577
5.582
5.573
5.580
†
H-6␤
H-7(␣)
3.159
5.265
2.415
2.415
5.128
2.415
5.166
1.638
1.053
5.451
2.060
5.455
2.008
5.211
–d
5.414
5.446
†
1.992
†
†
†
†
†
H-9␣
2.241
2.124
2.120
1.992
H-11␣
H-11␤
1.655
1.468
1.72
1.71
1.70
1.68
1.755
1.606
1.734
1.511
1.587
1.458
1.589
1.432
1.67
1.73
1.66
1.72
†
†
∗
H-12␣
H-12␤
H-14␣
H-15␣
H-15␤
H-16␣
1.981
1.558
1.700
1.704
1.277
1.753
2.047
1.650
2.616
1.853
1.579
1.840
1.776
1.473
2.608
1.961
1.669
1.705
1.856
1.479
2.562
1.823
1.688
1.684
1.864
1.464
2.609
1.713
1.755
1.598
2.560
1.693
1.527
1.769
1.70
1.74
†
†
∗
1.760
1.82
1.66
†
†
†
†
†
†
2.525
2.599
1.860
1.465
1.564
2.634
1.881
1.507
1.806
†
†
†
†
†
1.797
1.673
†
†
†
†
1.708
1.482
†
†
1.694
1.536
†
H-16␤
H-18
H-19
H-20
H-21
2.936
0.643
1.025
2.990
0.587
0.949
2.699
0.698
0.948
2.695
0.662
1.182
2.687
0.647
1.000
1.675
2.077
0.613
0.862
3.835
1.206
1.711
0.756
0.959
4.050
1.194
2.109
0.691
0.949
3.866
1.226
0.680
0.870
4.021
1.185
2.118
2.106
2.301
2.298
2.298
Chemical shifts given to two (three) decimal places are accurate to ± 0.01 (± 0.001) ppm except that values marked by dagger (†) are accurate to ca.
∗
±
0.003 ppm. Assignments of diastereotopic pairs marked with an asterisk ( ) may be interchanged.
a
◦
Data obtained at 500 MHz in 5–15 mM CDCl3 solution at 25 C and referenced to Si(CH3)4.
Coupling constants were generally similar to those of C27 sterols with the same double bond configuration (described in [15]) except for H-20 (q,
b
6
.3 Hz) and ring-A couplings of 5␤ steroids.
c
Acetate signals: 1, δ 2.039 (s), 2.043 (s); 2, δ 2.050 (s), 2.075 (s); other signals for 1: δ 1.553 (tdd, 12.0, 10.4, 5.1 Hz, H-8␤), 2.023 (m, H-7␤).
d
Not measured.

L.-W. Guo et al. / Steroids 68 (2003) 31–42
35
Table 2
1
3C NMR chemical shifts of unsaturated pregnanetriols 6a–7b and their synthetic precursors 1–5a,b
5
5,7
5,7
4,7
5␤-ꢀ7
5␤-ꢀ7
5,7
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
1
2
3
4
5c
3␣,20␤ (6a)
3␣,20␣ (6b)
3␤,20␤ (7a)
3␤,20␣ (7b)
†
C-1
C-2
C-3
C-4
36.97
27.72
73.79
38.04
37.88
28.03
72.59
36.59
38.28
31.94
70.32
40.73
33.16
34.08
199.21
122.88
35.5
34.71
31.49
71.31
37.65
34.67
31.45
71.26
37.61
38.32
31.97
70.39
40.75
38.32
31.96
70.37
40.75
37.70
212.46
43.47
∗
∗
†
C-5
C-6
C-7
C-8
C-9
C-10
139.63
122.20
31.79
31.92
49.39
36.56
139.13
120.02
117.76
139.52
45.65
140.11
119.45
117.37
140.17
45.93
168.28
32.80
116.72
138.71
45.51
41.9
40.96
28.66
115.65
137.53
36.69
40.90
28.66
139.73
119.51
116.58
141.48
46.11
139.82
119.55
116.93
140.94
46.01
28.34
115.99
137.25
∼36 ± 2
33.82
116.05
d
–
36.62
33.51
37.10
37.07
37.97
33.54
37.02
37.00
†
C-11
C-12
20.64
31.06
20.55
30.67
20.57
29.57
21.38
29.49
21.7
21.34
32.33
21.12
31.22
20.81
31.67
20.62
30.65
30.11
†
†
C-13
C-14
C-15
C-16
46.76
51.83
24.03
30.43
46.90
49.94
22.83
30.50
48.66
49.07
22.98
33.67
49.01
49.54
22.68
33.54
49.5
48.26
49.42
22.31
33.81
46.6
47.54
48.88
22.63
33.82
45.97
49.80
22.29
37.73
49.87
22.49
33.61
50.28
21.99
37.61
†
C-17
C-18
C-19
C-20
C-21
97.04
14.25
19.28
204.19
26.38
97.06
14.31
16.22
203.86
26.39
90.15
15.45
16.34
211.40
27.87
89.92
15.45
21.29
211.23
27.81
90.05
15.49
23.47
211.42
27.83
85.51
15.27
24.50
70.57
18.67
85.9
85.55
15.45
16.34
70.39
18.71
86.03
14.22
16.31
72.27
18.74
14.19
24.47
72.35
18.75
∗
Assignments marked by an asterisk ( ) may be interchanged. Values marked by dagger (†) were estimated to ca. ± 0.1 ppm from HSQC, HMBC, or 1D
spectra.
a
b
c
d
◦
Data obtained at 125 MHz at 25 C in 2–30 mM CDCl3 solution and referenced to the CDCl3 signal at 77.0 ppm.
Acetate signals: 1, δ 21.27, 21.42, 170.56, 170.79; 2, δ 21.21, 21.41, 170.60, 170.71.
Several signal of 5 were severely broadened at 25 C (see Fig. 3A).
Not measured.
◦
C-20 epimers 6a and 6b. A portion (10 mg) of this mixture
was subjected to preparative reverse-phase HPLC (elution
with methanol–water 8:2 at 9 ml/min) to give homogenous
samples of the 20R isomer 6a (tR 21.1 min, 3.4 mg, ∼97%
purity) and the 20S isomer 6b (tR 23.1 min, 3.1 mg, ∼98%
1
13
purity). H and C NMR for 6a and 6b (Tables 1 and 2);
HR-MS, calculated for C21H34O3 334.2508, found 334.2520
for 6a, found 334.2510 for 6b; MS of 6a and 6b (Table 3).
2
.2.6. Pregna-5,7-diene-3β,17α,20R-triol (7a) and
its 20S epimer 7b
To an ice-cold solution of ꢀ5,7 diacetate 2 (51 mg,
0.124 mmol) in anhydrous tetrahydrofuran (10 ml) was
added an excess of LiAlH4 (73 mg, 1.92 mmol). The re-
sulting mixture was stirred under nitrogen in an ice bath
for 1.5 h, followed by addition of cold 5% HCl (15 ml) and
extraction with ethyl acetate (3 × 15 ml). The combined
organic phase was dried over anhydrous Na2SO4 and evap-
orated to a white crystalline solid (41 mg) consisting of
a 3:4 mixture of 7a and 7b. This mixture was separated
by MPLC (elution with toluene–ethyl acetate 10:3; 6-ml
fraction volumes). Evaporation of fractions 42–57 gave 7a
Fig. 3. (A) 13C NMR spectrum (125 MHz; upfield portion) of
7
5
␤-ꢀ -3,20-diketone 5, showing the broadening of certain signals in
1
rings B and C. The position of C-9 is uncertain. (B) H NMR spectrum
500 MHz; upfield methyl region) of 9b, giving a representative illustra-
tion of the purity of the unsaturated pregnane-3,17␣,20-triols. Signals of
(
(
15 mg, 97% purity), and evaporation of fractions 73–100
1
13
gave 7b (16 mg, 98% purity). H and C NMR of 7a and
b (Tables 1 and 2); HR-MS, calculated for C21H32O3
13
the major contaminants (3% 13b and 1.5% 14a) are labeled, and
C
∗
7
satellites are marked by an asterisk ( ).

3
6
L.-W. Guo et al. / Steroids 68 (2003) 31–42
Table 3
Comparison of ion abundances in the electron-impact mass spectra of unsaturated pregnanetriolsa,b,c
Ion assignment
m/z
Relative abundance (%)
m/z
Relative abundance (%)
7
8
8(14)
5,7
5,8
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
6
a
6b
9a
9b
13b
0.4
3
12
1
1
4
100
2
3
14a
14b
7a
7b
11a
11b
+
M
334
316
301
298
289
283
271
255
253
246
231
228
213
5
3
1
2
7
3
1
7
100
4
23
4
2
0
4
0
36
13
18
10
33
12
31
7
28
17
16
9
29
11
73
11
11
100
28
18
35
332
314
299
296
287
281
269
253
251
244
229
226
211
4
7
3
3
1
8
100
5
21
1
0
9
1
22
1
46
78
47
33
3
8
12
36
2
10
16
6
2
8
14
4
M–H2O
M–H2O–CH3
M–2H2O
M–SC
M–2H2O–CH3
M–H2O–SC
M–2H2O–SC+2H
M–2H2O–SC
M–SC–C2H3O
M–SC–C3H6O
M–H2O–SC–C2H3O
M–H2O–SC–C3H6O
10
18
12
3
25
100
36
16
6
12
10
1
17
100
7
21
4
1
2
1
5
100
5
28
100
19
15
6
9
8
18
23
100
6
15
2
4
6
18
39
8
9
8
2
5
1
3
5
1
6
100
29
21
39
4
9
9
5
26
2
2
4
10
18
9
9
SC, side chain (C2H4O).
a
Spectra of underivatized triols were obtained by direct probe using electron-impact ionization (70 eV).
Molecular formulas of representative fragment ions were confirmed by HR-MS. Losses of C2H3O and C3H6O evidently arose from cleavage in ring D.
Additional ions: 6a, 105 (29); 6b, 105 (18); 7a, 143 (14); 7b, 263 (41), 143 (100); 9a, 105 (21); 9b, 105 (62); 11a, 143 (28); 11b, 143 (24); 13b,
b
c
1
05 (14); 14a, 147 (54); 14b, 147 (57).
3
32.2351, found 332.2355 for 7a, found 332.2351 for 7b;
(elution with toluene–ethyl acetate 10:3; 6-ml fraction vol-
MS of 7a and 7b (Table 3).
umes). Evaporation of fractions 49–70 gave 9a (13 mg,
95% purity, containing 3% of the 20R epimer of 13b),
2
.2.7. 3β,17α-Dihydroxy-5α-pregn-7-en-20-one
and evaporation of fractions 77–110 gave 9b (16 mg, 95%
purity, containing 3% 13b and 1.5% 14a, as shown in
diacetate (8)
1
13
Freshly prepared Raney nickel was added to a solution
Fig. 3B). H and C NMR of 9a and 9b (Tables 4 and 5);
HR-MS of 9a, calculated for C21H34O3 334.2508, found
334.2512; HR-MS of 9b, calculated for C21H32O2 (M–H2O)
316.2402, found 316.2408; MS of 9a and 9b (Table 3).
NMR data for the 20R epimer of 13b: δH 0.747 (s, H-18),
0.961 (s, H-19), 2.726 (m, H-14␣); δC 14.3 (C-18), 17.9
(C-19).
of ꢀ5 diacetate 2 (50 mg) in freshly distilled ethyl acetate
,7
(15 ml). The system was repeatedly evacuated and filled with
hydrogen. The mixture was then stirred at room tempera-
ture under a hydrogen-filled balloon for 1 h. After filtration
of the catalyst, the solvent was evaporated to give 8 as a
8
white solid (55 mg, ∼94% purity, containing 3% of the ꢀ
1
13
isomer). H and C NMR (Tables 4 and 5); HR-MS, cal-
ꢀ
culated for C H O 416.2563, found 416.2566; MS m/z
3
2.2.9. 3β,17α-Dihydroxy-7α-[N,N -bis(ethoxycarbonyl)
2
5
36
5
∗
8
∗
∗
∗
56 (27, M–AcOH), 341 (27), 313 (100), 281 (12), 253
hydrazino]pregna-5,8-dien-20-one 3,17-diacetate (10)
A solution of 2 (0.30 g, 0.72 mmol) and DEAD (0.45 g,
2.58 mmol, 3.6 equivalents) in benzene (5 ml) was re-
(
(
14); NMR data for the ꢀ isomer: δH 0.565 (s, H-18), 0.966
s, H-19); δC 13.7 (C-18), 17.8 (C-19). Similar hydrogena-
◦
tion of diol 3 resulted in concomitant hydrogenation of the
2
epimeric at C-20 (Table 6).
fluxed under nitrogen at 100 C for 4.5 h. After cooling,
7
8(14)
0-keto group, producing a mixture of ꢀ and ꢀ
triols
this solution was subjected directly to MPLC (elution
with hexane–ethyl acetate 7:3; 15-ml fraction volumes).
Fractions 11–13 contained the Diels–Alder adduct (41 mg;
NMR δH 0.748 (s), 0.954 (s), 2.031 (s), 2.059 (s), 2.178
(s), 6.092 (d, 8.3 Hz), 6.559 (d, 8.3 Hz)), and fractions
2
.2.8. 5α-Pregn-7-ene-3β,17α,20R-triol (9a)
and its 20S epimer 9b
5
,8(14)
To an ice-cold solution of diacetate 8 (50 mg, 0.120 mmol)
in anhydrous tetrahydrofuran (10 ml) was added an excess
of LiAlH4 (72 mg, 1.91 mmol). The resulting mixture was
stirred under nitrogen in an ice bath for 1.5 h, followed
by addition of cold 5% HCl (15 ml) and extraction with
ethyl acetate (3 × 15 ml). The combined organic phase
was dried over anhydrous Na2SO4 and evaporated to a
white crystalline solid (39 mg) consisting of a 2:3 mix-
ture of 9a and 9b. This mixture was separated by MPLC
17–18 contained the ꢀ
adduct (7 mg). Evaporation
of fractions 20–41 gave the ꢀ5 adduct 10 as the ma-
,8
1
jor product (0.32 g, ∼98% purity). H NMR, δ 0.613 (s,
H-18), 1.195 (s, H-19), 2.079 (s, H-21), 2.03 and 2.14
(br s, acetates), 3.04 (m), 4.0–6.2 (many broad signals);
HR-MS, calculated for C31H44N2O9 588.3047, found
∗
588.3036; MS m/z 412 (7, M–C H12N2O4), 397 (3),
6
∗
∗
∗
∗
353 (69), 293 (43), 277 (44), 250 (100), 235 (30),
157 (30).

L.-W. Guo et al. / Steroids 68 (2003) 31–42
37
Table 4
1
H NMR data for unsaturated pregnanetriols 9a–14b and 20-ketosteroids 8 and 15a
␣-ꢀ7
5␣-ꢀ7
ꢀ
5,8
ꢀ
5,7,9(11)
5␣-ꢀ8
5␣-ꢀ8(14)
ꢀ
5,8
5
3
9a
␤,20␤
3␤,20␣
9b
3␤,20␤
11a
3␤,20␣
11b
3␤,20␤
12a
3␤,20␣
12b
3␤,20␣
13b
3␤,20␤
14a
3␤,20␣
14b
8b
15b
†
H-1␣
H-1␤
H-2␣
1.160
1.858
1.832
1.093
1.836
1.800
1.095
1.829
1.802
1.370
1.368
1.529
1.526
1.228
1.761
1.851
1.114
1.715
1.826
1.115
1.703
1.827
1.378
1.886
1.900
†
†
†
†
†
†
1.889
1.875
1.712
1.701
†
†
1.892
1.892
1.934
1.933†
†
H-2␤
H-3␣
1.486
4.703
1.391
3.604
1.391
3.606
1.561
3.554
1.563
3.556
1.710
1.706†
3.613†
1.444
3.625
1.357
3.627
1.354
3.627
1.568
3.551
3.614
2.382
2.473
†
†
†
H-4␣
1.759
1.364
1.464
1.78
1.723
1.276
1.410
1.763
1.744
5.184
1.728
1.277
1.410
2.351
2.355
2.384
2.474
1.655
1.327
1.414
1.392
1.519
1.639
1.641
2.357
2.313
†
†
†
†
†
H-4␤
2.314
2.313
1.265
1.265
†
†
H-5␣
1.260
1.263
†
†
∗
H-6(␣)
H-6␤
1.77
5.434
5.440
5.682
5.432
5.689
5.453
1.339
1.344
5.433
∗
†
†
1.74
1.75
1.225
1.214
†
†
∗
†
†
H-7(␣)
H-7␤
5.232
5.221
2.544
2.555
1.99
1.834
2.400
1.837
2.410
2.530
2.580
†
†
∗
2.582
2.594
2.01
†
†
†
†
†
†
†
†
†
H-9␣
1.734
1.708
1.472
1.573
2.038
2.526
1.690
1.569
1.782
1.671
1.651
1.484
1.69
1.658
1.705
1.698
†
∗
†
†
H-11(␣)
H-11␤
H-12␣
H-12␤
H-14␣
H-15␣
H-15␤
H-16␣
1.652
2.18
2.18
5.579
5.542
2.202
1.676
1.508
1.593
1.693
1.675
2.22
∗
†
†
1.458
1.713
1.581
2.533
2.19
2.18
2.064
1.491
2.24
†
1.838
1.746
2.799
1.769
1.352
1.561
1.874
1.556
2.826
2.578
2.081
2.827
1.915
1.502
1.613
2.692
1.919
2.855
1.810
1.538
2.751
1.670
1.704
2.068
2.273
1.731
1.421
1.727
†
†
1.74
1.490
†
†
2.503
1.673
1.463
1.528
†
†
†
†
†
†
†
1.704
1.792
1.912
1.758
2.402
2.384
†
†
†
1.510
1.408
1.539
1.383
2.326
2.333
†
†
†
†
†
†
†
1.761
1.793
1.857
1.760
1.528
1.867
†
H-16␤
H-18
H-19
H-20
H-21
2.948
0.501
0.813
1.663
0.670
0.810
4.022
1.180
2.068
0.604
0.802
3.831
1.201
1.741
0.789
1.199
4.058
1.191
2.149
0.709
1.191
3.855
1.216
1.760
0.697
1.255
4.045
1.199
2.119
0.614
1.251
3.880
1.209
2.135
1.747
1.033
0.702
4.022
1.198
2.006
0.948
0.696
3.869
1.226
2.276
0.606
1.186
0.670
0.953
3.835
1.205
2.111
2.144
Chemical shifts given to two (three) decimal places are accurate to ± 0.01 (± 0.001) ppm except that values marked by dagger (†) are accurate to ca.
∗
±
0.003 ppm. Assignments of diastereotopic pairs marked with an asterisk ( ) may be interchanged.
a
◦
Data obtained at 500 MHz in 5–15 mM CDCl3 solution at 25 C and referenced to Si(CH3)4. Coupling constants were generally similar to those of
C27 sterols with the same double bond configuration (described in [15]) except for H-20 (q, 6.3 Hz) and ring-A couplings of 5␤ steroids.
b
Additional signals: acetate signals for 8: δ 2.033 (s), 2.053 (s); H-17␣ of 15, δ 2.601 (br t, 8.8 Hz).
2
.2.10. Pregna-5,8-diene-3β,17α,20R-triol (11a), its 20S
culated for C21H32O3 332.2351, found 332.2346; HR-MS
of 11b, calculated for C21H30O2 (M–H2O) 314.2246,
found 314.2229; MS of 11a and 11b (Table 3); HR-MS
of trienetriols, calculated for C21H30O3 330.2195, found
330.2189 for 12a, found 330.2191 for 12b; MS of 12a m/z
epimer 11b, pregna-5,7,9(11)-triene-3β,17α,20R-triol
(12a), and its 20S epimer 12b
To a solution of DEAD derivative 10 (170 mg, 0.29 mmol)
in anhydrous tetrahydrofuran (25 ml) was added dropwise
DIBALH (1.0 M solution in tetrahydrofuran, 7.2 ml, 25
equivalents). The resulting solution was stirred at room
temperature for 50 min, followed by addition of cold 5%
HCl (20 ml). After extraction with ethyl acetate (3×20 ml),
the combined organic phase was dried over anhydrous
Na2SO4. Evaporation of solvent gave a 3:3:3:4 mixture
of 11a, 11b, 12a, and 12b as a white solid (138 mg). Al-
though this mixture was essentially inseparable on silica
gel using toluene–ethyl acetate 10:3, a portion of the prod-
uct (60 mg) was separated on reverse-phase HPLC (elution
with methanol–water 65:35 at 8 ml/min) to give 12b (tR
+
∗
∗
∗
∗
330 (21, M ), 312 (10), 297 (22), 294 (18), 279 (90),
∗
276 (41), 267 (61), 261 (100), 251 (40), 209 (92); MS of
+
∗
∗
∗
12b m/z 330 (33, M ), 312 (5), 297 (19), 279 (73), 276
(47), 269 (100), 267 (63), 261 (70), 251 (25), 209 (54).
∗
2.2.11. 5α-Pregn-8-ene-3β,17α,20S-triol (13b)
To a solution of 11b (8 mg) in freshly distilled ethyl
acetate (2 ml) was added Raney nickel (∼5 mg). The re-
sulting mixture was stirred at room temperature under a
hydrogen-filled balloon. Analysis of reaction aliquots by
7
NMR after 10, 20, 30, and 40 min indicated a mixture of ꢀ ,
8
8(14)
2
8.5 min, 13 mg, 98% purity), 11b (tR 32.8 min, 12 mg,
97% purity), 12a (tR 33.9 min, 6 mg, 98% purity), and 11a
ꢀ , and ꢀ
triols (Table 7). The catalyst was removed
∼
by filtration through Celite, and the filtrate was evaporated
to a white solid (8 mg). This isomeric mixture was sub-
jected to reverse-phase HPLC (elution with methanol–water
1
13
(
tR 42.2 min, 10 mg, 97% purity). H and C NMR of 11a,
1
1b, 12a, and 12b (Tables 4 and 5); HR-MS of 11a, cal-

3
8
L.-W. Guo et al. / Steroids 68 (2003) 31–42
Table 5
1
3C NMR data for unsaturated pregnanetriols 9a–14b and 20-ketosteroids 8 and 15a
␣-ꢀ7
5␣-ꢀ7
ꢀ
5,8
ꢀ
5,7,9(11)
5␣-ꢀ8
5␣-ꢀ8(14)
ꢀ
5,8
5
3
9a
␤,20␤
3␤,20␣
9b
3␤,20␤
11a
3␤,20␣
11b
3␤,20␤
12a
3␤,20␣
12b
3␤,20␣
13b
3␤,20␤
14a
3␤,20␣
14b
8b
36.81
15
35.62
C-1
C-2
C-3
C-4
C-5
C-6
C-7
C-8
37.12
37.12
31.43
70.98
37.92
40.21
29.65
118.18
139.09
49.17
34.20
20.97
30.85
46.38
50.22
22.11
37.51
86.05
14.17
13.06
72.34
18.74
35.76
31.96
71.45
42.22
138.88
119.35
29.11
126.68
131.62
37.43
22.04
29.80
46.68
46.17
22.65
34.68
84.50
14.41
22.87
70.80
18.59
35.69
31.90
71.43
42.18
138.82
119.34
29.08
126.40
131.80
37.41
22.04
28.75
45.39
46.99
22.20
38.54
84.99
13.69
22.80
72.49
18.67
38.45
32.24
72.24
41.57
141.51
118.14
115.99
135.47
143.58
39.40
122.70
35.27
47.07
45.69
22.30
34.24
85.15
14.87
30.41
70.28
18.61
38.43
32.22
72.25
41.55
141.54
118.16
116.15
135.10
143.70
39.41
122.05
34.26
45.30
46.44
21.73
38.05
85.25
13.97
30.38
72.21
18.32
35.20
31.66
71.15
38.31
40.77
25.45
27.26
128.29
134.75
35.77
22.49
28.68
45.36
47.09
23.06
38.51
85.07
13.47
17.87
72.49
18.71
36.46
31.52
71.15
38.25
44.12
28.66
29.50
129.50
49.01
36.81
19.57
29.33
47.74
139.10
24.45
32.25
85.75
22.74
12.85
69.43
19.06
36.45
31.50
71.13
38.24
44.09
28.64
29.55
129.35
48.79
36.75
19.26
28.15
46.23
139.68
24.07
34.71
86.03
21.67
12.86
71.45
18.62
27.42
73.29
33.69
40.05
29.46
119.20
138.02
48.81
34.21
20.96
30.90
47.26
50.53
22.72
30.38
97.15
14.31
12.94
203.95
26.34
31.45
71.00
37.95
40.25
29.64
117.76
139.61
49.28
34.22
21.19
31.95
47.95
49.34
22.41
33.65
85.52
15.24
13.06
70.54
18.65
31.88
71.34
42.16
138.82
119.26
29.01
125.97
132.22
37.42
22.21
35.87
43.56
51.90
23.29
23.18
62.26
12.62
22.84
209.43
31.39
C-9
C-10
C-11
C-12
C-13
C-14
C-15
C-16
C-17
C-18
C-19
C-20
C-21
a
◦
Data obtained at 125 MHz at 25 C in 2–30 mM CDCl3 solution and referenced to the CDCl3 signal at 77.0 ppm.
b
Acetate signals for 8: δ 21.44, 170.72, 21.22, 170.72.
5
5:45) to give 13b (4 mg, containing 13% 9b). H and 13
1
C
Table 7
5,8
Products formed in the catalytic hydrogenation of ꢀ
Raney nickel after reaction times of 10–40 mina
triol 11b over
NMR (Tables 4 and 5); HR-MS, calculated for C21H32O2
M–H2O) 316.2402, found 316.2398; MS (Table 3).
(
Reaction time
min)
Percentage composition
(
2
.2.12. 5α-Pregn-8(14)-ene-3β,17α,20R-triol (14a)
5,8
7
8
8(14)
ꢀ
(11b)
ꢀ
(9b)
ꢀ
(13b)
ꢀ
(14b)
The ꢀ5
,7,9(11)
triol 12a (10 mg) was hydrogenated for
1
2
0
0
23
5
0
37
44
44
48
30
37
45
30
10
13
10
22
3
.5 h over PtO2 (7 mg) as described for the conversion of
1
1b to 14b. MPLC (elution with hexane–ethyl acetate 7:3)
30
40
of this crude product (10 mg) gave 14a (6.5 mg, 98% purity).
0
1
13
H and C NMR (Tables 4 and 5); HR-MS, calculated for
a
Reaction aliquots were taken after the indicated reaction time and
analyzed by 1H NMR.
C21H34O3 334.2508, found 334.2530; MS (Table 3).
Table 6
a,b
Ratio of 20R and 20S isomers produced by reduction of 3␤,17␣-dihydroxy-20-ketosteroid derivatives
Steroid
Reagent
20R (%)
20S (%)
5
38c
62c
1
1
1
1
3
(ꢀ -diacetate)
LiAlH4, 60 equivalents, 0.5 h
LiAlH4, 58 equivalents, 1.5 h
DIBALH, 80 equivalents, 0.5 h
DIBALH, 86 equivalents, 1.5 h
H2, Raney nickel
5
c
c
7 (ꢀ -3-acetate)
61
80
66
39
20
34
5
c
c
(ꢀ -diacetate)
5
c
c
7 (ꢀ -3-acetate)
5
,7
d
d
(ꢀ -diol)
21
79
a
All hydride reductions were conducted in tetrahydrofuran at room temperature, followed by addition of water, extraction with ethyl acetate, evaporation
of the organic extracts, and NMR analysis. In the case of catalytic hydrogenation in ethyl acetate, the reaction mixture was filtered to remove the catalyst
and then evaporated to dryness for NMR analysis.
b
A large excess of hydride reducing agent was used to insure complete reduction of acetate and keto groups. Use of 15–20 equivalents of hydride
resulted in mixtures containing little 20-hydroxysteroid.
c
These percentages exclude a small amount (∼10%) of an unidentified byproduct: NMR, δ 0.664 (s), 1.020 (s), 1.285 (d, 6.5 Hz), 4.426 (q, 6.5 Hz).
d
These percentages were derived from the total product composition of 55% 9b, 15% 9a, 13% 14b, 0.3% 14a, 2% 20-ketosteroid species, and 16%
unidentified.

L.-W. Guo et al. / Steroids 68 (2003) 31–42
39
2
.2.13. 5α-Pregn-8(14)-ene,3β,17α,20S-triol (14b)
two conformers: (I) the usual 1␣,2␤-half chair and (II)
a 1␤,2␣-half chair at a slightly higher relative energy
(0.5 kcal/mol for MMX; 0.3 kcal/mol for MM3). In con-
former I, C-19 is axial in both rings A and B, whereas C-19
is equatorial in ring A of conformer II. Predicted couplings
for MM3 geometries of conformer I (J1␣–2␣ 4.49 Hz; J1␣–2␤
13.24 Hz; J1␤–2␣ 2.69 Hz; J1␤–2␤ 3.99 Hz) and conformer
II (J1␣–2␣ 3.72 Hz; J1␣–2␤ 2.92 Hz; J1␤–2␣ 13.32 Hz; J1␤–2␤
4.30 Hz) gave the best fit to the observed data when the
ratio of I:II was 43:57. This model roughly corresponds to
the experimental data but does not exclude the possibility
of an additional low-energy conformer.
To a solution of ꢀ5 triol 11b (10 mg) in freshly dis-
,8
tilled ethyl acetate (2 ml) was added PtO2 (8 mg). The re-
sulting mixture was stirred at room temperature under a
hydrogen-filled balloon for 2 h. The catalyst was removed
by filtration through Celite, and the filtrate was evaporated to
1
13
give 14b as a white solid (8 mg, ∼95% purity). H and
C
NMR (Tables 4 and 5); HR-MS, calculated for C21H34O3
3
34.2508, found 334.2498; MS (Table 3).
2
.2.14. Lithium-ethylamine reductions of DEAD
adducts 10 and 16
Freshly cut lithium (88 mg, 12.7 mmol) was added to a
2
.4. Side-chain conformations of 17α-hydroxysteroids
solution of 10 (44 mg, 0.075 mmol) in ethylamine (2 ml) at
◦
−
78 C under nitrogen. After the blue color of the reac-
Molecular mechanics energies were calculated for the
tion solution appeared, the reaction temperature was ele-
◦
rotamers about the C17–C20 bond of 5␣-pregnane-3␤,17␣,
20R-triol, its 20S epimer, and 3␤,17␣-dihydroxy-5␣-
pregnan-20-one. The best conformers are shown in Fig. 4.
Other rotamers were at least 1 kcal/mol higher in energy and
represent minor conformers (≤15% of the total population).
vated to −20 C. The reaction solution was stirred for an
◦
additional 30 min, cooled to −78 C, and quenched with
t-butanol (2 ml). After aqueous workup, the crude product
was subjected to MPLC (elution with hexane–ethyl acetate
:3) to give 3␤-hydroxypregna-5,8-dien-20-one (15; 15 mg).
7
1
13
+
H and C NMR (Tables 4 and 5); MS m/z 314 (48, M ),
∗
∗
2
99 (8), 281 (84), 255 (32), 211 (27), 159 (36), 143 (58).
Similar reduction of 16 (14 mg, 0.028 mmol) with lithium
14 mg, 2 mmol) in ethylamine (1 ml) and anhydrous tetrahy-
drofuran (0.5 ml) gave a 2:1 mixture (6 mg) of 11b and 12b.
3. Discussion
(
We describe the chemical synthesis of a series of C21
steroids having hydroxyl at C-3, C-17, and C-20 and
unsaturation in rings B and/or C. These synthetic tar-
5
2
.3. Conformational analysis of 4
gets are candidate metabolites for SLOS. Apart from ꢀ
steroids, very few C21 steroids of any kind have been
reported with double bonds in rings B and C. Conse-
quently, we designed synthetic routes based on known
methods for preparing unsaturated sterols in the choles-
terol and ergosterol series [15,17–21]. As shown in
Fig. 1, all the unsaturated C21 triols were made from
a single precursor, 17␣-hydroxy-7-dehydropregnenolone
diacetate (2). This intermediate was prepared on a
multi-gram scale in ∼50% yield from the commer-
cially available 17␣-hydroxypregnenolone diacetate (1).
The conversion of 1 to 2 was based on the analogous
bromination–dehydrobromination of pregnenolone acetate
using tetrabutylammonium fluoride [22].
1
H NMR analysis of the ꢀ4,7-3-ketosteroid 4 indicated
1
the following H NMR coupling constants: J1␣–2␣ 5.8 Hz;
J1␣–2␤ 7.2 Hz; J1␤–2␣ 8.7 Hz; J1␤–2␤ 6.0 Hz. Molecular
modeling using the MMX or MM3 force fields suggested
Key intermediate 2 was converted to various unsaturated
pregnanetriols by four different methods (Fig. 1). Reduction
of 2 with excess LiAlH4 gave the ꢀ5 triols 7a and 7b,
which were isolated by MPLC in 37% and 39% yield, re-
,7
spectively. Hydrogenation of 2 over Raney nickel gave the
7
ꢀ
-20-keto-diacetate 8, which was also reduced with ex-
cess LiAlH4 to provide, after chromatographic separation,
7
the ꢀ triols 9a and 9b in 33% and 42% isolated yields.
Interestingly, when the hydrogenation was carried out on
diol 3 instead of diacetate 2, the carbonyl group at C-20
was also reduced, giving 9a and 9b directly as a 1:4 mix-
ture (Table 6). Despite this favorable ratio of C-20 epimers,
the product was contaminated by ꢀ8
(14)
steroids and other
Fig. 4. Newman projections showing the best conformers of
byproducts, a result that led us to avoid hydrogenation of
17␣-hydroxy-20-ketosteroids in favor of LiAlH reductions.
1
7␣-hydroxy-20-ketosteroids (A) and 17␣,20-dihydroxysteroids (B, C).
The view is from C20 to C17.
4

4
0
L.-W. Guo et al. / Steroids 68 (2003) 31–42
8
5,7,9(11)
The synthesis of 6a and 6b, the 3␣-hydroxy-5␤ stereoiso-
ꢀ triol 13b. Similar hydrogenation of the ꢀ
triol
triol 14a. Nevertheless,
hydrogenation of 11b over Raney nickel in redistilled ethyl
12a also produced mainly the ꢀ8
(14)
mers of 9a and 9b, began with solvolysis of diacetate 2 to
diol 3, followed by conversion to the ꢀ4 -3-ketosteroid 4
with cholesterol oxidase under conditions we have used pre-
viously for the selective oxidation of oxysterols [23]. The
enzymatic oxidation–isomerization was mild enough to min-
imize formation of the unwanted ꢀ4 isomer. Hydrogena-
tion of 4 over palladium on carbon in 4-methoxypyridine
,7
7
8(14)
acetate gave 13b, accompanied by the ꢀ and ꢀ
9b and 14b. As indicated in Table 7, a mixture of mainly
triols
7
8
ꢀ and ꢀ isomers formed initially, but extended reaction
,6
8(14)
times resulted in isomerization to the ꢀ
byproduct.
Our syntheses of unsaturated triols produced both C-20
epimers. We initially hoped to reduce the 20-ketosteroid
intermediates stereoselectively to the natural 20S iso-
mer without resorting to microbiological methods [33].
Compared with hydride reductions of 17-deoxysteroids,
in which the 20R isomer is strongly favored [34],
17␣-hydroxy-20-ketosteroids give increased amounts of 20S
epimer [16]. However, bulky hydrides lead predominantly
or exclusively [16] to the 20R epimer, as confirmed by our
results with large (DIBALH) and small (LiAlH4) reagents
and with a 17␣-hydroxy or 17␣-acetoxy group (Table 6).
Acetylation of the 17-hydroxyl appeared to favor formation
of the 20S epimer in LiAlH4 reductions (Table 6), and we
observed somewhat lower 20R:20S ratios in reductions of
17␣-acetoxysteroids (7a:7b, 3:4; 9a:9b, 2:3) than in that
of a 17␣-hydroxysteroid (6a:6b, 8:9). Although the model
DIBALH reductions described in Table 6 favored the 20R
epimer, DIBALH reduction of adduct 10 gave a 1:1 ratio of
[
24] produced predominantly the desired 5␤-steroid 5. Re-
duction of this 3,20-diketone with LiAlH4 gave a 9:10 mix-
ture of 20R and 20S epimers 6a and 6b, which were isolated
by preparative reverse-phase HPLC. The 3␣-hydroxy-5␤-H
stereochemistry was confirmed by comparison with NMR
7
data reported for sterols having similar 3-hydroxy-ꢀ func-
tionality [15,25]. We also attempted to prepare 6a and 6b by
5
,7
a remarkable transformation in which a 3␤-hydroxy-ꢀ
7
sterol is converted to the 3␣-hydroxy-5␤-ꢀ sterol by heat-
ing with a copper/alumina catalyst in cyclohexanol [26]. In
our hands, the reported transformation with ergosterol gave
a more complex mixture than originally described [26],
although some 5␤-ergosta-7,22E-dien-3␣-ol was observed.
However, treatment of either 3 or 7a under the same re-
action conditions gave a complex mixture, devoid of the
7
desired ꢀ stereoisomers.
We initially attempted to synthesize the ꢀ5,8 triols 11a
5,7
5,8
5,7,9(11)
and 11b by the classical method for converting ꢀ
to
ꢀ
triols and a 3:4 ratio of ꢀ
steroids epimeric at
5,8
ꢀ
sterols via an ene adduct with DEAD [27,28]. Al-
C-20 (20R:20S). In one example, catalytic hydrogenation
gave the largest percentage of 20S product, but this mate-
rial was accompanied by isomeric byproducts that were not
easily removed chromatographically.
though the DEAD adduct 10 was readily prepared in good
yield from the ꢀ5 diacetate 2, reduction of 10 with
,7
◦
lithium in ethylamine at −20 C led, as expected [29], to
the formation of the 17-deoxygenated product 15 (Fig. 2).
All pairs of C-20 epimers could be resolved chromato-
graphically and spectroscopically. Some epimeric pairs
Lithium-ethylamine reduction of the unprotected DEAD
5
,7
7
adduct 16 did give 11b, albeit accompanied by consid-
(ꢀ and 5␣-ꢀ triols) were resolved on MPLC (tR(20S) >
tR(20R)) but others required separation on reverse-phase
HPLC. The 20R and 20S isomers were easily distinguished
erable ꢀ5
,7,9(11)
triol 12b (Fig. 2). However, we found
that adduct 10 could be reduced directly to triols 11a and
1b with DIBALH. This single step achieved reduction
1
13
1
by H and C NMR, as described previously [16,35,36].
NMR signals for H-12␤ and H-18 were deshielded in the
20R epimer due to their proximity to the 20-hydroxyl in
the major conformer (Fig. 4), and H-16␤ was similarly
deshielded in the 20S epimer. These and other effects re-
sulted in notable chemical shift differences between the
C-20 epimers for H-16␣, H-16␤, H-18, H-20, C-18, and
C-20 (Tables 1, 2, 4, and 5). The best NMR reporter peaks
were the H-20 signals, which resonated at δ 4.0–4.1 in the
20R isomers and δ 3.8–3.9 in the 20S isomers. These H-20
chemical shift differences are much larger than in the corre-
sponding 17-deoxy analogs (δ 3.73 versus 3.72). The C-20
epimers (and most other pairs of isomers) were not readily
distinguished from their mass spectra (Table 3).
of the 20-keto group, the 3␤- and 17␣-acetates, and the
ꢀ
7
␣-[N,N -bis(ethoxycarbonyl)hydrazino] moiety without
5,8
any deoxygenation at C-17. Although the ꢀ
products
byprod-
were accompanied by similar amounts of ꢀ5
,7,9(11)
ucts, the mixture was separable on reverse-phase HPLC.
Interestingly, attempted reduction of 10 with sodium boro-
hydride, lithium aluminum hydride, or L-Selectride gave
none of the target ꢀ5 steroids.
,8
8
Preparation of the ꢀ triol 13b was more challenging
8
than the other syntheses. Although ꢀ sterols are accessible
from ꢀ5 sterols by acid isomerization [17–20], followed
by careful hydrogenation of the resulting ꢀ8 dienol [21],
our efforts to isomerize either 2 or 7a with aqueous or an-
hydrous HCl gave only mixtures containing little or none
of the ꢀ8 product. This isomerization also failed for C19
,7
,14
In the course of characterizing the steroids by 1D and
2D NMR, we encountered two examples of conforma-
,14
1
steroids [30]. Few other methods are available for preparing
tional heterogeneity. In the H NMR spectrum of the
8
ꢀ
4,7-3-ketosteroid 4, coupling constants for the C-1 and C-2
ꢀ
steroids. Our attempts to follow a questionable [19,31]
5,8
report [32] describing the catalytic hydrogenation of a ꢀ
protons showed major deviations from the values normally
observed in ring A of ꢀ -3-ketosteroids [37,38]. Molecular
modeling suggested that 4 is populated by a 1␤,2␣-half chair
8
4
sterol to a ꢀ sterol using platinum oxide gave almost ex-
clusively the ꢀ8
(14)
triol 14b from 11b, with no detectable

L.-W. Guo et al. / Steroids 68 (2003) 31–42
41
as well as the usual 1␣,2␤-half chair conformer adopted
[10] Bhavnani BR. The saga of the ring B unsaturated equine estrogens.
Endocr Rev 1988;9:396–416.
4
7
by most ꢀ -3-ketosteroids. This influence of the ꢀ bond
on the conformation of ring A may have been responsible
for the reduced stereoselectivity in the hydrogenation of 4
[
11] Shackleton CH, Roitman E, Kratz L, Kelley R. Dehydro-oestriol and
dehydropregnanetriol are candidate analytes for prenatal diagnosis
of Smith–Lemli–Opitz syndrome. Prenat Diagn 2001;21:207–12.
12] Shackleton CH, Roitman E, Kratz LE, Kelley RI. Midgestational
maternal urine steroid markers of fetal Smith–Lemli–Opitz (SLO)
syndrome (7-dehydrocholesterol 7-reductase deficiency). Steroids
4
compared to similar reductions of ꢀ -3-ketosteroids lack-
[
7
ing a ꢀ bond [24]. Effects of ring-B double bonds [38] and
other functionality ([37,39] and references therein) on the
ring-A conformation of 3-ketosteroids have been described
previously. The ring-A conformers of 4 interconvert rapidly
on the NMR time scale, and only one set of averaged chem-
1999;64:446–52.
[
13] Shackleton CH, Roitman E, Kelley R. Neonatal urinary steroids in
Smith–Lemli–Opitz syndrome associated with 7-dehydrocholesterol
reductase deficiency. Steroids 1999;64:481–90.
13
ical shifts was observed. By contrast, several C signals in
[14] Guo L-W, Shackleton CH, Wilson WK. Synthesis of ring B
unsaturated estriols. Confirming the structure of a diagnostic analyte
for Smith–Lemli–Opitz syndrome. Org Lett 2001;3:2547–50.
7
the NMR spectrum of 5␤-ꢀ -3,20-diketone 5 were severely
broadened or missing entirely (Fig. 3A). A ‘nonsteroidal’
[
15] Wilson WK, Sumpter RM, Warren JJ, Rogers PS, Ruan B, Schroepfer
Jr GJ. Analysis of unsaturated C27 sterols by nuclear magnetic
resonance spectroscopy. J Lipid Res 1996;37:1529–55.
conformation for rings A and B is possible for 5␤-sterols
7
[
25,40]. In the case of a 3␣-hydroxy-ꢀ sterol, only the
usual steroidal conformation was observed [25], whereas a
nonsteroidal conformation was found in an X-ray structure
of a derivative of 5␤-cholesta-8,14-dien-3␤-ol [40]. The
[16] Shull BK, Koreeda M. (E)-3␤-Hydroxy-5,17(20)-pregnadiene
-pivaloate and (Z)-3␤,16␣-dihydroxy-5,17(20)-pregnadiene 3-piva-
3
loate. J Org Chem 1990;55:99–105.
[17] Wilson WK, Schroepfer Jr GJ. Acid-catalyzed isomerization of
latter report also noted severe broadening of certain 13
NMR signals.
C
7-dehydrocholesteryl benzoate. A revised mechanism and improved
synthetic procedure. J Org Chem 1988;53:1713–9.
The unsaturated pregnanetriols were isolated by chro-
[18] Dolle RE, Schmidt SJ, Eggleston D, Kruse LI. Studies on the
acid-catalyzed homonuclear steroidal diene isomerization. J Org
Chem 1988;53:1563–6.
matographic methods in 95–98% purity (except for 13b) as
1
judged by 500 MHz H NMR spectra (see Fig. 3B). Consid-
[
[
[
19] Fieser LF, Fieser M. Steroids. New York, NY: Reinhold; 1959,
ering that these steroids are labile and differ little in chro-
matographic mobility, this level of purity is quite high and
should be ample for the intended use as reference standards.
p. 111–9 and 161.
20] Fieser LF, Ourisson G. Cholesterol and companions. Oxidation of ꢀ
7
sterols with selenium dioxide. J Am Chem Soc 1953;75:4404–14.
21] Lee WH, Kammereck R, Lutsky BN, McCloskey JA, Schroepfer
Jr GJ. Studies on the mechanism of the enzymatic conversion
8
7
of ꢀ -cholesten-3␤-ol to ꢀ -cholesten-3␤-ol.
969;244:2033–40.
J Biol Chem
Acknowledgments
1
[
22] Siddiqui AU, Wilson WK, Swaminathan S, Schroepfer Jr GJ.
Efficient preparation of steroidal 5,7-dienes of high purity. Chem
Phys Lipids 1992;63:115–29.
This work was supported by the National Institutes of
Health (Grants HL-49122 and R03HD-39707).
[
23] Li S, Pang J, Wilson WK, Schroepfer Jr GJ. Sterol synthesis.
Preparation and characterization of fluorinated and deuterated
analogs of oxygenated derivatives of cholesterol. Chem Phys Lipids
References
1
999;99:33–71.
24] Tsuji N, Suzuki J, Shiota M. Highly stereoselective hydrogenation of
-oxo-4-ene and -1,4-diene steroids to 5␤ compounds with palladium
[
[
[
1] White PC. Genetic diseases of steroid metabolism. Vitam Horm
994;49:131–95.
2] New MI. Genetic disorders of adrenal hormone synthesis. Horm Res
992;37(Suppl 3):22–33.
3] Tint GS, Irons M, Elias MR, Batta AK, Frieden R, Chen TS,
Salen G. Defective cholesterol biosynthesis associated with the
Smith–Lemli–Opitz syndrome. N Engl J Med 1994;330:107–13.
4] Kelley RI, Hennekam RC. The Smith–Lemli–Opitz syndrome. J Med
Genet 2000;37:321–35.
[
[
[
1
3
catalyst. J Org Chem 1980;45:2729–31.
1
25] Andrews AT de B, Boul AD, Meakins GD, Sledge MJ.
Hydroxysteroids. Part XIII. 5␤-Ergosta-7,22-dien-3-one and related
compounds. J Chem Soc (C) 1970;1052–5.
26] Ravasio N, Gargano M. Selective hydrogenations promoted by copper
catalysts. Hydrogen-transfer reactions leading to stereoselective
hydrogenation of ꢀ -3␤-sterols to 5␤ derivatives. J Org Chem
1
27] Anastasia M, Fiecchi A. Synthesis of lichesterol [ergosta-5,8(9),22-
trien-3␤-ol]. J Chem Soc, Perkin Trans 1 1981:2125–6.
28] Anastasia M, Fiecchi A, Galli G. Synthesis of cholesta-5,8-dien-
[
[
5
5] Andersson HC, Frentz J, Martinez JE, Tuck-Muller CM, Bellizaire J.
Adrenal insufficiency in Smith–Lemli–Opitz syndrome. Am J Med
Genet 1999;82:382–4.
993;58:1259–61.
[
[6] Nowaczyk MJ, Siu VM, Krakowiak PA, Porter FD. Adrenal
[
insufficiency and hypertension in
a
newborn infant with
3
␤-ol. J Org Chem 1981;46:3421–2.
Smith–Lemli–Opitz syndrome. Am J Med Genet 2001;103:223–5.
7] Ruan B, Wilson WK, Pang J, Gerst N, Pinkerton FD, Tsai J, Kelley
RI, Whitby FG, Milewicz DM, Garbern J, Schroepfer Jr GJ. Sterols
in blood of normal and Smith–Lemli–Opitz subjects. J Lipid Res
[
[29] Dryden Jr HL. Reductions of steroids by metal–ammonia solutions:
In: Fried J, Edwards JA, editors. Organic reactions in steroid
chemistry, vol. 1. New York, NY: Van Nostrand Reinhold; 1972,
p. 1–60.
2
001;42:799–812.
5,7
[
8] Batta AK, Tint GS, Shefer S, Abuelo D, Salen G. Identification
of 8-dehydrocholesterol (cholesta-5,8-dien-3␤-ol) in patients with
Smith–Lemli–Opitz syndrome. J Lipid Res 1995;36:705–13.
9] Ruan B, Wilson WK, Tsai J, Schroepfer Jr GJ. Aberrant pathways in
the late stages of cholesterol biosynthesis. Origin and metabolic fate
of unsaturated sterols relevant to the Smith–Lemli–Opitz syndrome.
J Lipid Res 2000;41:1772–82.
[30] Antonucci R, Bernstein S, Giancola D, Sax KJ. ꢀ
Steroids. X.
Transformation products of ꢀ5,7-androstadiene-3␤,17␤-diol. J Org
Chem 1951;16:1891–8.
[
[31] Barton DHR, Corrie JET, Widdowson DA, Bard M, Woods RA.
Biosynthesis of terpenes and steroids. Part IX. The sterols of some
mutant yeasts and their relationship to the biosynthesis of ergosterol.
J Chem Soc, Perkin Trans 1 1974:1326–33.

4
2
L.-W. Guo et al. / Steroids 68 (2003) 31–42
[
[
[
32] Tsuda K, Arima K, Hayatsu R. Cholesterol and related
5-pregnen-20-one from 15␤,17␣-dihydroxy-4-pregnen-3,20-dione.
Steroids 1996;61:18–21.
[37] Marat K, Templeton JF, Kumar VPS. An NMR study of ring-A
conformation in some 4-en-3-one steroids. Magn Reson Chem
1987;25:25–30.
compounds. Structure of a new non-conjugated cholestadienol from
7-bromocholesterol. J Am Chem Soc 1954;76:2933–8.
33] Madyastha KM, Joseph T. Transformation of 16-dehydroprogesterone
and 17␣-hydroxyprogesterone by Mucor piriformis. Appl Microbiol
Biotechnol 1994;41:170–7.
34] Wheeler DMS, Wheeler MM. Reductions of steroidal ketones. In:
Fried J, Edwards JA, editors. Organic reactions in steroid chemistry,
vol. 1. New York, NY: Van Nostrand Reinhold; 1972, p. 61–
[38] Ruan B, Wilson WK, Pang J, Schroepfer Jr GJ. Synthesis of
3
[3␣- H]cholesta-5,8-dien-3␤-ol and tritium-labeled forms of other
sterols of potential importance in the Smith–Lemli–Opitz syndrome.
Steroids 2000;65:29–39.
1
10.
[39] Marat K, Templeton JF, Gupta RK, Kumar VPS. Ring-A conforma-
tion in steroids. NMR study of C-2 monomethyl- and dimethyl-
substituted 5␣-androstan-3-ones. Magn Reson Chem 1987;25:730–3.
[40] Wilson DK, Wilson WK, Quiocho FA, Schroepfer Jr GJ. Concerning
the structure of 3␤-benzoyloxy-5␤-cholesta-8,14-diene, a major
byproduct in the chemical synthesis of 5␣-cholest-8(14)-en-3␤-ol-
15-one. Chem Phys Lipids 1988;47:273–82.
[
35] Joannou GE, Reeder AY. 15␤-Hydroxysteroids (part III).
Steroids of the human perinatal period: the synthesis of
3␤,15␤,17␣-trihydroxy-5-pregnen-20-one. Application of n-butyl
boronic acid protection of a 17,20-glycol. Steroids 1996;61:11–7.
36] Joannou GE, Reeder AY. 15␤-Hydroxysteroids (part V). Steroids of
the human perinatal period: the synthesis of 3␤,15␤,17␣-trihydroxy-
[