Stereochemical analysis of the 3α- and 3β-hydroxy metabolites of tibolone through NMR and quantum-chemical investigations. An experimental test of GIAO calculations

Article abstract of DOI:10.1002/mrc.1064

The configuration at C-3 of the 3α- and 3β-hydroxy metabolites of tibolone was studied by extensive application of one- and two-dimensional ¹H and ¹³C NMR spectroscopy combined with molecular modeling performed at the B3LYP/6-31G(d)

Full text of DOI:10.1002/mrc.1064

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2002; 40: 581–588
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1064
Stereochemical analysis of the 3a- and 3b-hydroxy
metabolites of tibolone through NMR and
quantum-chemical investigations. An experimental test
of GIAO calculations
Diego Colombo,1 Patrizia Ferraboschi,1∗ Fiamma Ronchetti1 and Lucio Toma2
1
Dipartimento di Chimica e Biochimica Medica, Universita` di Milano, Via Saldini 50, 20133 Milan, Italy
Dipartimento di Chimica Organica, Universita` di Pavia, Via Taramelli 10, 27100 Pavia, Italy
2
Received 16 January 2002; Revised 15 May 2002; Accepted 24 May 2002
The configuration at C-3 of the 3a- and 3b-hydroxy metabolites of tibolone was studied by extensive
application of one- and two-dimensional 1H and 13C NMR spectroscopy combined with molecular
modeling performed at the B3LYP/6–31G(d) level. Using HF and DFT GIAO methods, shielding tensors
of the two molecules were computed; comparison of the calculated NMR chemical shifts with the
experimental values revealed that the density functional methods produced the best results for assigning
proton and carbon resonances. Although steroids are relatively large molecules, the present approach
appears accurate enough to allow the determination of relative configurations by using calculated 13C
resonances; the chemical shift of pairs of geminal a/b hydrogen atoms can also be established by using
calculated 1H resonances. Copyright  2002 John Wiley & Sons, Ltd.
KEYWORDS: NMR; 1H NMR; 13C NMR; tibolone metabolites; stereochemistry; steroids; molecular modeling; HF
calculations; DFT calculations
(5), of which the tibolone is the 7˛-methyl analogue and
which on metal hydride reduction affords as preferred
product the 3˛-hydroxy derivative owing to the quasi-chair
conformation assumed by the A ring.9,10 Although tibolone
and norethinodrel share the same A ring, the presence of a
methyl group at position 7 could, in principle, modify the A
ring quasi-chair conformation and hence the stereochemical
outcome of the 3-ketone reduction.
INTRODUCTION
The synthetic steroid tibolone (Org OD 14) (1) is widely
used in hormone replacement therapy (HRT) of menopausal
complaints1 and it is metabolized mainly affording the 4-ene
isomer 2 and the 3˛- and the 3ˇ-alcohols 3 and 4 obtained
by reduction of the 3-keto group. The hormonal activities
of these three steroids have been extensively evaluated2 and
more recently the role of tibolone and its metabolites in the
protection of breast tissue in postmenopausal women with
HRT has been studied.3–7
Considering the pharmacological significance of tibolone
metabolites and the few available chemico-physical data, we
decided to study the 3-hydroxy derivatives, verifying the
configuration at C-3 of both epimers. They can be easily
prepared from 1, the first by reduction with lithium tri-
tert-butoxyaluminum hydride that affords a predominant
product purified by crystallization. Its 3-epimer can be
obtained by inversion of the configuration at C-3 performed
through a Mitsunobu reaction.8 The 3˛ configuration,
represented by structure 3, might be assigned to the
main product of reduction on the basis of the structural
analogy of 1 with the antifertility steroid norethinodrel
We report here a detailed NMR study of diol 3 and its
epimer 4 combined with a modeling investigation through
quantum-chemical calculations that allowed us to confirm
the assignment of the relative configuration at C-3 and to
1
explore the usefulness of theoretical calculations of H and
13C chemical shifts in relation to stereochemical studies of
steroidal compounds.
RESULTS AND DISCUSSION
Reduction of tibolone (1) with lithium tri-tert-butoxyalumi-
num hydride yielded two epimeric diols (3 and 4) in a ratio
of ca 96 : 4. The main product 3 was easily obtained pure
by crystallization from hexane–acetone whereas its epimer
4 was prepared by treatment of 3 with benzoic acid, diiso-
propyl diazadicarboxylate and triphenylphosphine followed
by hydrolysis of the recovered benzoate.
Complete 1H and 13C NMR signal assignments (Tables 1
and 2) of the spectra of 3 and 4 were achieved using a combi-
nation of 1D and 2D (COSY, HSQC and NOESY) experiments
ŁCorrespondence to: Patrizia Ferraboschi, Dipartimento di Chimica
e Biochimica Medica, Universita` di Milano, Via Saldini 50, 20133
Milan, Italy. E-mail: patrizia.ferraboschi@unimi.it
Contract/grant sponsor: Universita` degli Studi di Milano.
Contract/grant sponsor: Universita` degli Studi di Pavia.
Copyright  2002 John Wiley & Sons, Ltd.

582
D. Colombo et al.
HO
HO
O
O
1
2
172
HO
HO
18
172
18
12
12
HO
171
16
171
16
11
11
17
17
13
13
14
1
4
1
4
10
10
14
15
15
9
9
2
3
2
8
8
7
7
71
3
71
HO
HO
5
5
6
6
O
3
4
5
Table 1. GIAO-calculated 1H NMR chemical shifts (υ, in ppm relative to TMS) for 3 and 4 based on geometries optimized at the
B3LYP/6–31G(d) level in comparison with the experimental values from the spectra recorded in chloroform–pyridine (1 : 1)
HF/
6–31G(d)
HF/
6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
Compound
Ha
Exp.
3
1˛
1ˇ
2˛
2ˇ
3
4˛
4ˇ
6˛
6ˇ
7
1.97
2.17
1.57
2.08
3.94
2.05
2.33
1.62
2.23
1.81
1.47
1.69
1.96
1.20
1.90
1.71
1.90
1.66
1.34
2.40
2.13
1.02
0.81
2.84
1.64
1.96
1.14
1.87
3.59
1.33
2.10
1.33
1.98
1.49
1.26
1.42
1.62
1.05
1.71
1.24
1.51
1.48
1.25
2.24
2.05
0.92
0.85
2.75
1.55
1.88
1.04
1.77
3.44
1.24
2.00
1.21
1.90
1.34
1.07
1.28
1.53
0.95
1.64
1.14
1.35
1.41
1.17
2.18
1.95
0.86
0.79
2.69
1.92
2.26
1.21
1.87
3.79
1.50
2.26
1.56
2.22
1.78
1.71
1.97
1.82
1.33
1.90
1.45
2.13
1.66
1.46
2.30
2.12
0.95
0.91
1.94
1.90
2.23
1.12
1.83
3.72
1.43
2.21
1.47
2.19
1.71
1.57
1.85
1.79
1.27
1.87
1.41
2.03
1.64
1.42
2.29
2.09
0.93
0.87
2.05
1.91
2.25
1.22
1.87
3.81
1.51
2.26
1.57
2.21
1.79
1.73
1.97
1.82
1.32
1.89
1.45
2.13
1.65
1.44
2.31
2.11
0.93
0.88
2.03
1.89
2.22
1.13
1.82
3.74
1.44
2.22
1.48
2.18
1.72
1.59
1.84
1.79
1.26
1.86
1.40
2.02
1.63
1.40
2.30
2.08
0.91
0.85
2.13
8
9
11˛
11ˇ
12˛
12ˇ
14
15˛
15ˇ
16˛
16ˇ
18
71
172
4
1˛
1ˇ
2˛
2ˇ
3
4˛
4ˇ
6˛
6ˇ
7
1.84
2.43
1.74
1.87
4.16
2.21
2.13
1.56
2.31
1.81
1.50
1.59
1.99
1.31
1.79
3.69
1.81
1.75
1.33
1.95
1.50
1.24
1.52
1.92
1.20
1.68
3.57
1.72
1.66
1.21
1.88
1.35
1.07
1.79
2.32
1.55
1.73
3.82
2.16
1.89
1.57
2.20
1.79
1.70
1.76
2.31
1.48
1.67
3.81
2.11
1.83
1.48
2.17
1.73
1.58
1.79
2.28
1.53
1.73
3.83
2.14
1.91
1.58
2.19
1.81
1.73
1.76
2.28
1.46
1.68
3.83
2.09
1.84
1.49
2.16
1.74
1.59
8
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

Stereochemical analysis of tibolone metabolites 583
Table 1. (Continued)
HF/
6–31G(d)
HF/
6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
Compound
Ha
Exp.
9
1.73
1.99
1.21
1.91
1.70
1.90
1.67
1.34
2.40
2.13
0.98
0.83
2.83
1.48
1.69
0.99
1.74
1.25
1.52
1.49
1.25
2.25
2.05
0.91
0.87
2.76
1.34
1.61
0.90
1.66
1.16
1.36
1.41
1.17
2.19
1.96
0.86
0.80
2.70
2.02
1.89
1.28
1.93
1.46
2.15
1.67
1.46
2.31
2.13
0.95
0.92
1.95
1.90
1.88
1.24
1.90
1.42
2.05
1.65
1.43
2.30
2.09
0.93
0.89
2.06
2.02
1.89
1.27
1.92
1.46
2.15
1.65
1.45
2.32
2.12
0.93
0.90
2.04
1.90
1.88
1.22
1.89
1.42
2.04
1.64
1.41
2.32
2.08
0.91
0.87
2.14
11˛
11ˇ
12˛
12ˇ
14
15˛
15ˇ
16˛
16ˇ
18
71
172
a Numbered according to IUPAC–IUB Joint Commission on Biochemical Nomenclature (www.chem.qmul.ac.uk/iupac/steroid/).
Table 2. GIAO-calculated 13C NMR chemical shifts (υ, in ppm relative to TMS) for 3 and 4 based on geometries optimized at the
B3LYP/6–31G(d) level in comparison with the experimental values from the spectra recorded in chloroform–pyridine (1 : 1)
HF/
6–31G(d)
HF/
6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
Compound
C
Exp.
3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
27.5
33.1
67.3
41.2
124.6a
38.9
27.5
42.0
40.2
128.7a
25.7
33.6
47.8
46.3
22.3
39.5
79.0
13.3
13.0
89.5
73.2
25.0
29.9
59.7
37.4
125.3
34.4
24.3
34.8
34.7
129.0
23.7
28.7
40.8
38.6
21.0
34.5
69.2
14.6
13.8
78.1
75.3
25.4
30.5
60.6
38.2
126.4
35.0
25.2
35.9
35.7
130.1
24.2
29.4
42.0
39.7
21.3
35.0
70.6
14.6
13.9
79.9
76.1
29.7
34.3
66.6
42.8
123.4
39.6
30.8
42.9
42.0
127.0
28.1
33.8
50.8
47.5
24.4
39.3
79.4
15.4
14.4
78.1
67.7
30.2
34.8
67.7
43.5
124.8
40.3
31.7
44.1
43.2
128.4
28.6
34.3
52.4
48.6
24.8
39.7
81.2
15.3
14.3
80.4
69.0
29.6
33.5
66.3
42.3
123.9
39.3
30.2
42.3
41.5
127.3
27.6
33.4
50.6
47.0
24.4
38.9
79.7
15.3
14.4
79.7
70.4
30.1
34.0
67.3
43.0
125.2
39.9
31.0
43.4
42.6
128.6
28.0
33.9
52.1
48.0
24.7
39.2
81.3
15.3
14.3
81.9
71.6
71
171
172
4
1
2
3
4
5
6
7
8
9
23.2
30.8
65.1
40.1
123.4a
39.5
27.5
41.9
39.9
128.6a
25.6
21.3
27.2
58.1
35.2
124.6
34.5
24.3
34.9
34.7
128.8
23.7
28.8
21.8
27.7
58.9
35.8
125.7
35.2
25.2
35.9
35.8
129.9
24.3
29.4
25.1
31.0
64.9
40.2
122.4
39.8
30.8
43.0
42.1
126.8
28.1
33.8
25.6
31.4
65.9
41.0
123.8
40.5
31.6
44.2
43.3
128.3
28.6
34.3
25.0
30.5
64.7
39.9
122.9
39.5
30.2
42.4
41.6
127.2
27.6
33.4
25.4
30.9
65.6
40.6
124.1
40.1
31.0
43.5
42.8
128.5
28.0
33.9
10
11
12
33.6
(continued overleaf)
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

584
D. Colombo et al.
Table 2. (Continued)
HF/
6–31G(d)
HF/
6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
Compound
C
Exp.
13
14
15
16
17
18
71
171
172
47.7
46.4
22.3
39.4
79.0
13.2
13.2
89.6
73.2
40.8
38.7
21.0
34.6
69.2
14.7
13.8
78.1
75.4
42.0
39.7
21.4
35.0
70.5
14.7
13.9
79.9
76.2
50.8
47.5
24.5
39.4
79.4
15.4
14.4
78.1
67.7
52.4
48.7
24.9
39.8
81.2
15.4
14.3
80.4
69.0
50.6
47.0
24.4
39.0
79.7
15.4
14.4
79.7
70.5
52.1
48.1
24.7
39.3
81.3
15.3
14.3
81.9
71.7
a Assigned through comparison with the calculated values.
Table 3. Experimental 1H NMR coupling constants (Hz) for 3
and 4 in comparison with the values calculated with the
electronegativity-modified Karplus relationship
recorded in a chloroform–pyridine (1 : 1) mixture, as this sol-
vent gave the best spread of proton resonances of the two
steroids. Starting from the characteristic resonances of the
7˛-methyl group and of the H-3 proton, it was possible to
assign the resonances of H-1, H-2, H-4, H-6, H-7 and H-8 of
both 3 and 4 on the basis of their COSY and HSQC spec-
tra. Also, even if many protons in the 1H NMR spectrum
resonated as complex multiplets in the range 1.2–2.5 ppm,
some of these (Table 1) resulted in well resolved signals the
coupling of which could be measured (Table 3). In particular,
the assignments of some pairs of geminal protons (H-6, H-11,
H-15 and H-16 of both 3 and 4 and H-2 and H-4 of 3) were
made by comparison (Table 3) of the experimental values
of the vicinal coupling constants with the values calculated
through the electronegativity-modified Karplus relationship
(see below). NOE contacts from NOESY spectra were useful
for the assignment of other geminal protons, i.e. H-12 of both
3 and 4 (NOE between H-12ˇ and H3-18), H-1 of 3 (NOE
between H-1ˇ and H-3) and H-4 of 4 (NOE between H-4˛
and H-6˛), while the pairs of geminal H-1 and H-2 protons
J
3 (exp.)
3 (calcd)
4 (exp.)
4 (calcd)
2˛,2ˇ
1˛,2˛
1˛,2ˇ
1ˇ,2˛
1ˇ,2ˇ
2˛,3
11.5
5.5
5.5
2.5
11.3
6.1
10.7
3.3
5.8
2.5
11.0
6.4
2.0
5.3
11.5
11.5
3.5
2.0
6.0
2ˇ,3
4˛,4ˇ
3,4˛
3,4ˇ
16.8
9.0
5.5
17.0
3.5
4.5
9.3
5.5
4.0
3.5
6˛,6ˇ
6˛,7
6ˇ,7
7,71
16.5
<1.0
7.0
17.0
<1.0
6.0
1.4
6.1
1.4
6.1
7.0
7.0
7,8
8,9
8,14
11˛,11ˇ
9,11˛
9,11ˇ
11˛,12˛
11˛,12ˇ
11ˇ,12˛
11ˇ,12ˇ
15˛,15ˇ
14,15˛
14,15ˇ
16˛,16ˇ
15˛,16˛
15˛,16ˇ
15ˇ,16˛
15ˇ,16ˇ
3.0
3.0
12.0
12.0
3.0
2.4
12.0
12.0
1
of 4 were tentatively assigned from the H NMR chemical
11.0
11.0
12.0
11.0
11.0
13.0
3.5
13.0
3.5
3.5
13.0
3.5
shifts (Table 1) calculated through the GIAO approach (see
below). Finally, a cross peak between H-11˛ and one of the
H-1 protons in the NOESY spectrum of the isomer 4 was sig-
nificant for the assignment of the protons H-11 vs H-15 and,
consequently, of H-9 vs H-14 and H-12 vs H-16, of the C and
D rings. As this part of the molecule is identical for the two
isomers, the protons of the C and D rings were assigned for 3
on the basis of the resonances already established for 4, even
though the NOESY cross peak between ˛H-11 and one of the
H-1 protons, which is assumed from the computed distances
(data not shown), was not evidenced because of resonance
overlapping in the corresponding proton spectrum (Table 1).
The H-3 signal is of special interest as the four vicinal
coupling constants of H-3 (Table 3) can be diagnostic for
the configuration at C-3. This configurational assignment
relies heavily, however, on the knowledge of the exact
conformational preferences of 3 and 4 and, in particular, of
the A ring. In fact, owing to the presence of the 5(10) double
bond, two half-chair conformations can be envisaged (A and
B type, Figure 1), the relative stability of which derives from a
fine balance between steric and electronic factors. The vicinal
coupling constants indicate a pseudo-axial orientation of the
3.2
12.3
4.0
2.7
13.1
4.0
3.2
12.3
4.0
2.7
13.1
4.0
12.0
12.0
3.0
12.0
12.5
6.0
11.4
6.0
11.4
12.0
15.0
9.0
12.5
13.0
10.0
4.0
5.5
12.5
12.1
3.0
4.7
12.1
3.0
4.8
5.5
12.0
12.2
12.2
H-3 atom for 3 (and hence a hydroxy group equatorially
oriented) and vice versa for 4. However, these data cannot
be of help until the conformational preferences of 3 and 4
have been established.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

Stereochemical analysis of tibolone metabolites 585
(GIAO)14 method is the more widely used; Cheeseman
et al.15 recommended the following procedure to give a reli-
able estimate of shielding constants: after an optimization
at the B3LYP/6–31G(d) level, the optimized geometries
should be used to compute the NMR properties at the
HF/6–31G(d) level, predicting the isotropic chemical shifts
for carbon and hydrogen atoms with respect to tetramethyl-
silane (TMS). However, other workers suggested the use of
different models and/or basis sets also in relation to the
nuclei which are to be predicted.16 Hence, in this work two
DFT functionals,11,17 B3LYP and B3PW91, together with the
traditional Hartree–Fock method were used for GIAO cal-
culations combined with the 6–31G(d) and 6–31G(d,p) basis
sets.
Table 4. Relative energy (kcal
molꢀ1) and population percentages
at 298K of the B3LYP/6–31G(d)
optimized conformations of 3 and 4
Conformation
Erel
%
3A
3B
4A
4B
0.00
1.18
0.00a
1.22
88.0
12.0
88.7
11.3
a Erel D 0.03 kcal molꢀ1 with respect
to 3A (1 kcal = 4.184 kJ).
We computed both the 1H and 13C chemical shifts for each
pair of conformations of 3 and 4 and weighted averaged them
on the basis of the population percentages; the results are
reported in Tables 1 and 2. The shifts for the carbon atoms
computed with the HF approach appear prevalently at higher
fields than those measured experimentally with an error that
can become higher than 10 ppm. This disagreement does not
depend on the fact that GIAO calculations do not explicitly
consider the solvent, in our case a 1 : 1 mixture of pyridine
and chloroform, as 13C chemical shifts are not sensitive to the
solvent, as can be seen from the data in Table 5, where the
13C resonances in chloroform and pyridine are reported; the
solvent effect is in general limited to less than 1 ppm with
the only exception of the quaternary acetylenic atom. The
change of the basis set from 6–31G(d) to 6–31G(d,p) slightly
improved the results that, however, remain unsatisfactory.
An improvement could be observed by turning to the density
functional methods which presented fairly good agreement
of the calculated and experimental values; however, the
prediction of the acetylenic carbon atoms still remains
unsatisfactory.
The relative stability of conformers A and B was
determined within the DFT framework using a hybrid
exchange-correlation functional, B3LYP,11 at the 6–31G(d)
level as implemented in Gaussian 98.12 The relative energies
of these conformers are reported in Table 4 together with the
population percentages, calculated through the Boltzmann
equation, and their 3D representations are reported in
Fig. 1. It can be seen that both compounds prefer by
about 90% a conformation of type A that makes the OH
group equatorial in 3 and axial in 4. For each conformer
the 1H vicinal coupling constants were calculated with
the electronegativity-modified Karplus relationship13 and
were weighted averaged on the basis of the population
percentages. The values obtained are reported in Table 3 in
comparison with the experimental constants for 3 and 4. The
close agreement of the experimental and calculated values
confirms that the configuration at C-3 of diol 3 is ˛ and
similarly the configuration of diol 4 is ˇ. A number of NOE
contacts (e.g. between H-4ˇ and H-3 in 3 and between H-4˛
and H-3 in 4) further confirm the assigned structures. These
˚
contacts correspond to distances of <3 A as measured on the
To allow an easier comparison of methods and basis sets,
we determined the root mean square (r.m.s.) errors between
calculated and experimental 13C resonances (Table 6). The
values were calculated by inclusion or exclusion of the data
for the two acetylenic carbon atoms. It appears that the use
of a density functional method is largely to be preferred
computed 3A and 4A conformations of 3 and 4, respectively.
Ab initio computation of NMR chemical shifts is becom-
ing a convenient alternative tool for facilitating spectral
assignments and rationalizing experimental chemical shifts,
but has been infrequently applied to steroidal compounds.
For these calculations, the gauge-including atomic orbital
3A
3B
4A
4B
Figure 1. 3D plots of the minimum energy conformations of 3 and 4.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

586
D. Colombo et al.
Table 5. 13C NMR chemical shifts (υ, in ppm relative to TMS)
for 3 and 4 from the spectra recorded in chloroform and in
pyridine
and calculated υ values can be observed at all levels of
calculation.
As far as the 1H NMR resonances are concerned, also in
this case the density functional methods with the 6–31G(d)
basis set work better than the Hartree–Fock method as
r.m.s. errors less than 0.2 ppm are observed (Table 6) if the
acetylenic proton is excluded from the computation. In the
case of the proton resonances, comparison of experimental
and calculated values cannot be a safe tool for configurational
assignments; however, it should be pointed out that in each
pair of geminal ˛–ˇ hydrogen atoms, the relative order in
the chemical shifts is correctly predicted (Table 1). Hence
this can become a method for the assignment of ˛- and ˇ-
hydrogens in cases in which other methods, such as vicinal
coupling constant analysis or NOE contacts, fail.
Compound 3
Compound 4
C
Chloroform
Pyridine
Chloroform
Pyridine
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
71
171
172
27.2
32.6
68.1
40.6
124.1
38.7
27.0
41.7
39.8
128.6
25.3
33.2
47.5
46.2
22.1
39.0
79.8
12.9
12.9
87.9
73.7
27.8
33.7
67.4
41.7
124.9
39.1
27.7
42.2
40.5
129.0
25.9
34.0
48.1
46.7
22.6
40.0
79.2
13.6
13.1
90.3
74.1
22.3
29.9
65.9
39.7
122.6
39.1
27.3
41.7
39.5
128.5
25.3
33.1
47.4
46.3
22.1
39.0
79.9
12.9
13.0
87.8
73.8
23.6
31.4
65.3
40.7
123.7
40.0
27.8
42.2
40.2
128.9
25.9
34.0
48.0
46.8
22.6
39.7
79.2
13.3
13.5
90.3
74.1
CONCLUSIONS
The C-3 configuration of the diols 3 and 4, obtained by
reduction of tibolone, was assigned through a detailed
modeling study combined with the analysis of the vicinal
coupling constants of the ring A protons compared with the
theoretical J values. It has been shown that B3LYP/6–31G(d)
optimization followed by GIAO NMR calculations with the
same method or with the other DFT approach (B3PW91) is no
doubt a better way to carry out the theoretical determination
of 1H and 13C resonances. These methods are accurate
enough to permit the stereochemical assignment of the
configuration of diastereoisomeric steroidal compounds by
using 13C resonance differences; the predicted 1H resonances
appear less precise but allow the assignment of the chemical
shift within pairs of geminal ˛–ˇ hydrogen atoms.
over the HF method as r.m.s. errors of 1.5 ppm are observed
with the B3PW91 method if only sp3 and sp2 carbon atoms
are considered. The use of the more extended basis set
6–31G(d,p) seems unnecessary as it gives a worsening of the
errors.
EXPERIMENTAL
We now address the question of whether the theoretical
calculation of the 13C resonances can be used for the assign-
ment of the relative configuration of the diastereoisomeric
pair 3–4. We think that they can, in particular through mini-
mization of the systematic errors by expressing the chemical
shifts of the carbon atoms of one isomer relative to the values
of the other isomer. Table 7 reports calculated and experi-
mental υ(˛ ꢀ ˇ): obviously, these υ values are significant
only for the carbon atoms of ring A, as the other rings
are identical. Very close agreement between experimental
All solvents and reagents were purchased from Sigma.
Tibolone (1) was obtained according to Ref. 18. All reac-
tions were monitored by TLC on silica gel 60 F254 plates
(Merck) with detection by spraying with 10% phospho-
°
molybdic acid in ethanol solution and heating at 110 C.
Column chromatography was performed on silica gel 60
(0.063–0.200 mm) (Merck). Differential scanning calorimetry
(DSC) was performed on a Perkin-Elmer DSC-7 instrument.
GC analysis was performed on a Hewlett-Packard HP5890
°
instrument at 260 C oven temperature, with an HP5 capillary
Table 6. Comparison of the different methods for prediction of 1H and 13C chemical shifts by r.m.s. errors (in ppm)
HF/
6–31G(d)
HF/
6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
Compound
3
All 13
C
5.3
4.9
0.29
0.30
4.6
4.2
0.39
0.40
3.2
1.7
0.26
0.19
3.1
2.3
0.26
0.20
2.6
1.5
0.25
0.19
2.5
2.0
0.25
0.20
sp2 and sp3 13
All 1H
C
C
sp3 1H
All 13
4
C
5.3
4.9
0.29
0.30
4.6
4.3
0.38
0.39
3.2
1.7
0.23
0.15
3.1
2.3
0.22
0.16
2.6
1.5
0.22
0.16
2.5
2.0
0.21
0.15
sp2 and sp3 13
All 1H
sp3 1H
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

Stereochemical analysis of tibolone metabolites 587
Table 7. Experimental and calculated 13C NMR chemical shifts differences, υ(˛ ꢀ ˇ) (in ppm), between the resonances of 3 and 4
HF/
6–31G(d)
HF/
m6–31G(d,p)
B3LYP/
6–31G(d)
B3LYP/
6–31G(d,p)
B3PW91/
6–31G(d)
B3PW91/
6–31G(d,p)
C
Exp.
1
2
3
4.3
2.3
2.2
3.7
2.7
1.6
3.6
2.8
1.7
4.6
3.3
1.7
4.6
3.4
1.8
4.6
3.0
1.6
4.7
3.1
1.7
4
1.1
2.2
2.4
2.6
2.5
2.4
2.4
5
1.2
0.7
0.7
1.0
1.0
1.0
1.1
6
7
ꢀ0.6
0.0
ꢀ0.1
0.0
ꢀ0.2
0.0
ꢀ0.2
0.0
ꢀ0.2
0.1
ꢀ0.2
0.0
ꢀ0.2
0.0
8
9
0.1
0.3
0.1
ꢀ0.1
0.0
0.2
0.0
ꢀ0.1
0.2
ꢀ0.1
ꢀ0.1
0.2
ꢀ0.1
ꢀ0.1
0.1
ꢀ0.1
ꢀ0.1
0.1
ꢀ0.1
ꢀ0.2
0.1
10
11
12
13
14
15
16
17
18
71
171
172
0.1
0.0
0.1
ꢀ0.1
0.0
0.1
0.0
0.1
0.0
ꢀ0.1
0.0
ꢀ0.1
0.0
ꢀ0.1
0.0
ꢀ0.1
0.0
0.0
ꢀ0.1
ꢀ0.1
0.0
0.0
0.0
ꢀ0.1
0.0
0.1
ꢀ0.1
0.0
0.0
ꢀ0.1
0.0
0.0
0.0
0.0
ꢀ0.1
ꢀ0.1
0.0
0.0
0.0
0.0
0.0
0.0
ꢀ0.1
ꢀ0.1
ꢀ0.1
0.0
ꢀ0.1
0.0
0.0
0.0
0.0
0.0
0.0
ꢀ0.1
0.0
ꢀ0.1
0.0
0.0
0.0
0.0
0.0
ꢀ0.1
0.0
ꢀ0.1
0.0
0.0
0.0
0.0
ꢀ0.2
ꢀ0.1
0.0
0.0
0.0
0.0
0.0
ꢀ0.1
ꢀ0.1
chromatography [silica gel, 1 : 10; hexane–ethyl acetate (7 : 3) as elu-
column (25 m ð 0.32 mm i.d., 0.52 µm film thickness). Opti-
cal rotations were determined on a Perkin-Elmer model 241
polarimeter in ethyl acetate solutions (c D 1.0) in a 1 dm cell
ant] and crystallization (2-propanol–water) afforded pure 3ˇ-diol 4
25
°
°
(0.15 g, 53%). Endothermic peak fusion (DSC) at 142 C; [˛]D C15.7 ;
25
C
°
[˛]546 C18.5 ; EI-MS: m/z 314 (M , 65%), 296 (67%), 288 (100%); GC,
tR D 12.56 min; TLC [CHCl3 –AcOEt (7 : 3)], Rf D 0.40.
°
at 25 C. Electron ionization mass spectrometry (EI-MS) was
carried out at 70 eV by LC particle beam introduction with
a Hewlett-Packard HP 5988A quadrupolar mass spectrom-
eter equipped with a PB 59980A interface and an HP 1050
low-pressure liquid chromatograph.
NMR spectroscopy
All NMR spectra were recorded at 297 K with a Bruker AM-
500 spectrometer operating at 500.13 and 125.76 MHz for
1H and 13C, respectively, using a 5 mm broadband reverse
probe. Chemical shifts are reported on the υ (ppm) scale and
are relative to TMS as an internal reference. Compounds
3 and 4 (ca 15 mg) were dissolved in CDCl3 –pyridine-d5
(1 : 1) (0.5 ml) under N2, and their assignments were given
by a combination of 1D and 2D COSY, HSQC and NOESY
experiments, using standard Bruker pulse programs. The
Compounds
17˛-Ethynyl-7˛-methyl-5(10)-estren-3˛,17ˇ-diol (3)
A
solution of lithium tri-tert-butoxyaluminum hydride (1.1 M,
10.7 ml) in tetrahydrofuran was added dropwise to a solution of
tibolone (1) (1 g, 3.2 mmol) in anhydrous tetrahydrofuran (10 ml)
°
kept under N2, at ꢀ70 C. After 2 h the reaction mixture was poured
into 10% aqueous acetic acid (30 ml) and disodium ethylenedi-
aminetetraacetate (0.2 g) was added. The mixture was extracted
with chloroform (5 ð 25 ml). The collected organic phases were dried
over sodium sulfate; evaporation of the solvents afforded a crude
product: by trituration with three portions of methylene chloride
(5 ml) and crystallization from acetone–hexane pure 3˛-diol 3 (0.7 g,
pulse widths were 7.5 µs (90 ) and 9.6 µs (90 ) for 1H and
13C, respectively. Typically 16K and 32K data points were
collected for one-dimensional proton and carbon spectra,
respectively. Spectral widths were 5747 Hz for 1H NMR
°
°
25
°
70%) was recovered. Endothermic peak fusion (DSC) at 187 C; [˛]D
(digital resolution: 0.70 Hz per point) and 38 461 Hz for 13
C
25
C
°
°
C67.1 ; [˛]546 C79.9 ; EI-MS: m/z 314 (M , 43%), 296 (100%), 288
(100%); GC, retention time ꢀtRꢁ D 12.75 min; TLC [CHCl3 –AcOEt
(7 : 3)], Rf D 0.49.
NMR (digital resolution: 2.34 Hz per point). 2D experiments
parameters were as follows. For 1H–1H correlations (COSY
and NOESY): relaxation delay 1.2 s, data matrix 1K ð 2K (512
experiments to 1K, zero filling in F1, 2K in F2), 16 transients
in each experiment, spectral width 5.9 ppm (2958.6 Hz). The
NOESY spectra were generated with a mixing time of 1.0 s
and acquired in the TPPI mode. There were no significant
differences in the results obtained at different mixing times
(0.5–1.5 s). For 13C–1H correlations (HSQC): relaxation delay
1.5 s, data matrix 0.5K ð 2K (256 experiments to 0.5K, zero
filling in F1, 2K in F2), 32 transients in each experiment,
spectral width 5.9 ppm (2958.6 Hz) in the proton domain
and 147.2 ppm (18 518.5 Hz) in the carbon domain. All 2D
17˛-Ethynyl-7˛-methyl -5(10)-estren-3ˇ,17ˇ-diol (4)
A solution of 3˛-diol 3 (0.44 g, 1.4 mmol) and triphenylphos-
phine (0.474 g, 1.81 mmol) in anhydrous diethyl ether (8 ml) was
added dropwise to
a solution of diisopropyl azodicarboxy-
late (0.37 g, 1.81 mmol) and benzoic acid (0.222 g, 1.81 mmol) in
diethyl ether (0.8 ml). The reaction mixture was kept at room tem-
perature with stirring overnight. The solvent was removed under
reduced pressure and the crude product was purified by column
chromatography (silica gel, 1 : 10); elution with hexane–ethyl acetate
(9 : 1) afforded the benzoate (0.38 g, 65%). The ester was treated with
°
sodium carbonate (0.42 g) in methanol–water (9 : 1) (13 ml) at 50 C
for 4 h. The reaction mixture was poured into cool water (20 ml) and
the precipitated crude product was recovered by suction. Column
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588

588
D. Colombo et al.
4. Pasqualini JR, Paris J, Sitruk-Ware R, Chetrite G, Botella J. J.
Steroid Biochem. Mol. Biol. 1998; 65: 225.
5. Chetrite GS, Kloosterboer HJ, Philippe JC, Pasqualini JR.
Anticancer Res. 1999; 19(1A): 261.
spectra were weighted with sine-bell squared and shifted
(ꢂ/2 in both dimensions) window functions, and processed
with the Bruker software package.
6. Chetrite GS, Kloosterboer HJ, Philippe JC, Pasqualini JR.
Anticancer Res. 1999; 19(1A): 269.
Calculations
7. Gompel A, Siromachkova M, Lombert A, Kloosterboer HJ,
Rostene W. Eur. J. Cancer 2000; 36(Suppl 4): S76.
8. Mitsunobu O. Synthesis 1981; 1.
9. Palmer KH, Ross FT, Rhodes LS, Baggett B, Wall ME. J. Pharmacol.
Exp. Ther. 1969; 167: 207.
All calculations were carried out with the Gaussian 98
program.12 Geometry optimization of the conformations
of 3 and 4 was performed without constraints at the
B3LYP/6–31G(d) level. The population percentages were
calculated from the gas-phase electronic energies of the
conformers through the Boltzmann equation at 298K; the
entropic terms were neglected. Attempts to evaluate the
influence of the solvent on the relative energies of the
conformers were made using a continuum solvent model
(C-PCM)19 at different dielectric constant values, but the
runs stopped without completion owing to the molecular
size of 3 and 4. However, solvent calculations on smaller
models of 3 and 4 lacking the D ring could be performed and
confirmed the preference for conformers such as 3A and 4A.
NMR chemical shifts were calculated at the Hartree–Fock
and density functional levels with the 6–31G(d) or the
10. Palmer KH, Cook CE, Ross FT, Dolar J, Twine ME, Wall ME.
Steroids 1969; 14: 55.
11. (a) Lee C, Yang W, Parr RG. Phys. Rev.
(b) Becke AD. J. Chem. Phys. 1993; 98: 5648.
B 1988; 37: 785;
12. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA,
Cheeseman JR, Zakrzewski VG, Montgomery JA Jr, Stratmann
RE, Burant JC, Dapprich S, Millam JM, Daniels AD, Kudin KN,
Strain MC, Farkas O, Tomasi J, Barone V, Cossi M, Cammi
R, Mennucci B, Pomelli C, Adamo C, Clifford S, Ochterski J,
Petersson GA, Ayala PY, Cui Q, Morokuma K, Malick DK,
Rabuck AD, Raghavachari K, Foresman JB, Cioslowski J, Ortiz
JV, Baboul AG, Stefanov BB, Liu G, Liashenko A, Piskorz P,
Komaromi I, Gomperts R, Martin RL, Fox DJ, Keith T, Al-Laham
MA, Peng CY, Nanayakkara A, Challacombe M, Gill PMW,
Johnson B, Chen W, Wong MW, Andres JL, Gonzalez C, Head-
Gordon M, Replogle ES, Pople JA. Gaussian 98, Revision A.9.
Gaussian Inc.: Pittsburgh, PA, 1998.
13. Haasnoot CAG, de Leeuw FAAM, Altona C. Tetrahedron 1980;
36: 2783.
14. (a) Wolinski K, Hilton JF, Pulay P. J. Am. Chem. Soc. 1990; 112:
8251; (b) Ditchfield R. Mol. Phys. 1974; 27: 789.
15. Cheeseman JR, Trucks GW, Keith TA, Frisch MJ. J. Chem. Phys.
1996; 104: 5497.
1
6–31G(d,p) basis sets using the GIAO method. All the H
and 13C chemical shifts are referenced to those of TMS.
The absolute 1H and 13C shielding of TMS, based on the
B3LYP/6–31G(d) optimized geometry, were calculated at
the same level/basis set used in the calculation to which they
refer.
Acknowledgments
This work was financially supported by the Universita` degli Studi
di Milano and Universita` degli Studi di Pavia.
16. (a) Eloranta J, Hu J, Suontamo R, Kolehmainen E, Knuutinen J.
Magn. Reson. Chem. 2000; 38: 987; (b) Casanovas J, Namba AM,
Leo´n S, Aquino GLB, da Silva GVJ, Alema´n C. J. Org. Chem. 2001;
66: 3775; (c) Kupka T, Pastema G, Jaworska M, Karali A, Dais P.
Magn. Reson. Chem. 2000; 38: 149; (d) Smith WB. Magn. Reson.
Chem. 1999; 37: 103.
REFERENCES
1. Tax L, Goorissen EM, Kicovic PM. Maturitas 1987; 1(Suppl): 3.
2. (a) Markievicz L, Gurpide E. J. Steroid Biochem. 1990; 35: 535;
(b) Kloosterboer HJ. J. Steroid Biochem. Mol. Biol. 2001; 76: 231.
3. Chetrite G, Kloosterboer HJ, Pasqualini JR. Anticancer Res. 1997;
17(1A): 135.
17. Perdew JP, Wang Y. Phys. Rev. B 1992; 45: 13 244.
18. De Jongh HP, van Vliet NP. US Patent 3340279, Organon, 1967.
19. Barone V, Cossi M. J. Phys. Chem. A 1998; 102: 1995.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 581–588