Solid-state nuclear magnetic resonance investigation of neurosteroid compounds and magnesium interactions

Article abstract of DOI:10.1002/jccs.201800458

The neurosteroid trans-dehydroandrosterone (DHEA) and its analogs with slightly different modifications in the side chain attached to C17, that is, (3S)-acetoxypregn-5-en-20-one (1) and (3S,20R)-acetoxypregn-5-en-20-ol (2), have been synthesized to investigate DHEA–cation interactions. In this study, we applied solid-state ¹H/¹³C cross-polarization/magic-angle spinning (CP/MAS) nuclear magnetic resonance (NMR) spectroscopy to a series of DHEA analog/Mg²⁺ mixtures at different Mg²⁺ concentrations. The high-resolution ¹³C NMR spectra of 1/Mg²⁺ mixtures exhibit two distinct ¹³C spectral patterns, one attributable to 1 free from Mg²⁺, and the other attributable to 1 with bound Mg²⁺. For 2, the ¹³C NMR spectra exhibit three distinct spectral patterns; besides that of the free form, the other two can be assigned to Mg²⁺-bound forms. Based on the analysis of the chemical shift deviations (CSDs), we conclude that both 1 and 2 might be subject to a cation–π interaction via the C5–C6 double bond, in contrast to that observed previously for DHEA. As demonstrated, DHEA possesses two Mg²⁺ binding sites, that is, C17–O and C5–C6 double bond, in which the binding affinity of the former is at least three times stronger than that of the latter. The solid-state ¹³C NMR investigation allows better understanding of the underlying cation binding effects of neurosteroid molecules in vitro.

Full text of DOI:10.1002/jccs.201800458

Received: 30 November 2018
DOI: 10.1002/jccs.201800458
Revised: 9 February 2019
Accepted: 4 March 2019
A R T I C L E
Solid-state nuclear magnetic resonance investigation of
neurosteroid compounds and magnesium interactions
Danny Wu1 | Kathleen Joyce D. Carillo2,3,4 | Shen-Long Tsai1 | Jiun-Jie Shie2,4
Der-Lii M. Tzou4,5
|
1Chemical Engineering Department, National
The neurosteroid trans-dehydroandrosterone (DHEA) and its analogs with slightly
different modifications in the side chain attached to C17, that is, (3S)-
acetoxypregn-5-en-20-one (1) and (3S,20R)-acetoxypregn-5-en-20-ol (2), have
been synthesized to investigate DHEA–cation interactions. In this study, we applied
solid-state 1H/13C cross-polarization/magic-angle spinning (CP/MAS) nuclear mag-
netic resonance (NMR) spectroscopy to a series of DHEA analog/Mg2+ mixtures at
different Mg2+ concentrations. The high-resolution 13C NMR spectra of 1/Mg2+
mixtures exhibit two distinct 13C spectral patterns, one attributable to 1 free from
Mg2+, and the other attributable to 1 with bound Mg2+. For 2, the 13C NMR spectra
exhibit three distinct spectral patterns; besides that of the free form, the other two
can be assigned to Mg2+-bound forms. Based on the analysis of the chemical shift
deviations (CSDs), we conclude that both 1 and 2 might be subject to a cation–π
interaction via the C5–C6 double bond, in contrast to that observed previously for
DHEA. As demonstrated, DHEA possesses two Mg2+ binding sites, that is, C17–O
and C5–C6 double bond, in which the binding affinity of the former is at least three
times stronger than that of the latter. The solid-state 13C NMR investigation allows
better understanding of the underlying cation binding effects of neurosteroid mole-
cules in vitro.
Taiwan University of Science and Technology,
Taipei, Taiwan
2Taiwan International Graduate Program of
Sustainable Chemical Science and Technology,
Taipei, Taiwan
3Department of Applied Chemistry, National
Chiao Tung University, Hsinchu, Taiwan
4Institute of Chemistry, Academia Sinica,
Nankang, Taipei, Taiwan
5Department of Applied Chemistry, National Chia-
Yi University, Chia-Yi, Taiwan
Correspondence
Der-Lii M. Tzou, Institute of Chemistry,
Academia Sinica, Taipei, Taiwan.
Email: tzougate@gate.sinica.edu.tw
Funding information
Ministry of Science and Technology, Grant/Award
Number: MOST 105-2113-M-011-022
KEYWORDS
neurosteroid, trans-dehydroandrosterone (DHEA), solid-state NMR,
conformational change, magnesium
pregnenolone (Preg), for the treatment of depressive disor-
ders and chronic Alzheimer's disease. For example, it has
1
| INTRODUCTION
Neurosteroids are important modulators of brain activity
and behavior and participate in the regulation of mood and
memory.[1–4] Neurosteroids derived from cholesterol or ste-
roidal precursors imported from peripheral sources are also
present in neurons.[5–9] They act as modulators of neuro-
transmitter receptors such as GABAA, NMDA, and sigma
1 receptors.[10–13] As reported in the literature, a deficiency
of active neurosteroids may be associated with aging-related
impairments.[6,14] Synthetic methodologies have been devel-
oped to produce neurosteroid molecules, such as
been shown that Preg molecules display antidepressant effi-
cacy in rats.[15] Upon inducing spinal cord injury, Preg mol-
ecules are able to reduce histopathological changes
in vivo.[16] This neuroprotective effect could result from the
direct action by Preg on spinal cord neurons, whereby it
modulates the neuronal cytoskeleton dynamics through
interaction with MAP2.[17] In association with microtubulin
in the brain, the actions of Preg indicate its potential roles in
brain development, plasticity, aging, and depression, as it
has been suggested that hippocampal MAP2 expression may
© 2019 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
J Chin Chem Soc. 2019;1–9.
http://www.jccs.wiley-vch.de
1

2
WU ET AL.
be involved in the pathogenesis and pharmacology of
depression.[18–21]
DHEA analogs (compounds 1 and 2) to further investigate the
underlying binding sites. In terms of molecular structure, 1 and
2 have a skeleton identical to that of DHEA, but differ in the
side chain attached to C17 as well as the substituent attached
to C3 (Scheme 1). For 1, high-resolution 13C NMR spectra of
1/Mg2+ mixtures exhibit two distinct 13C spectral patterns, one
attributable to 1 free from Mg2+, and the other attributable
1 with bound Mg2+, similar to that observed from DHEA. For
2, however, the 13C NMR spectra exhibit three distinct spectral
patterns; besides that of the free form, the other two can be
assigned to Mg2+-bound forms. As demonstrated, on the basis
It is well known that the first step in the cascade of the
biosynthesis process of steroidal hormones is to convert cho-
lesterol into Preg, which is mediated by mitochondrial
P450scc. Furthermore, Preg is transformed into progesterone,
dehydroepiandrosterone (DHEA), and testosterone.[22] The
synthesis normally takes place in steroidogenic tissues, such
as gonads, adrenal glands, and placenta.[23] Besides these tis-
sues, Preg can also be synthesized in the brain, which is well
equipped with the enzyme machinery of steroidogenesis. The
enzymes involved in the synthesis/metabolism pathways of
steroids mainly target the C3-position of progesterone and its
hydroxyl metabolites and the C17-position of androgens and
estrogens. Researchers have reported that the 3α-hydroxyl
configuration is required for binding and activity.[24,25]
Therefore, synthetic strategies have been developed to pre-
pare DHEA analogs with modification of C3–C17 positions,
to be used as antagonists.
Magnesium ions (Mg2+) have many diverse functions
within the central nervous system, including voltage-
dependent blocking of NMDA receptors.[26] In humans, Mg2+
deficiency is well documented to be correlated with a series of
neuropsychiatric disorders.[27] Mg2+ serves as a cofactor in
more than 300 enzymes, such as tyrosine and tryptophan
hydroxylases,[28] as well as enzymes ubiquitous in energy
metabolism.[26] Given that most of these enzymes are brain-
related, Mg2+ deficiency is likely to be involved in a variety
of neuroses. In this work, we aimed to exploit DHEA and the
interactions of its analogs with Mg2+ by solid-state nuclear
magnetic resonance (NMR) spectroscopy.
DHEA
Solid-state NMR has been developed to acquire structural
information on pharmaceutical and biomedical materials.[29]
For example, it can be used to determine the average orienta-
tion and dynamics of ergosterol[30–32] and cholesterol[33] in
model membranes and for examining steroid–lipid
interactions.[34–36] In 13C NMR solid-state spectroscopy, cross-
polarization/magic-angle spinning (CP/MAS) techniques pro-
vide high-resolution spectra comparable to those observed in
the solution phase, so that conformational information can be
directly obtained.[33] High-resolution 1H–13C CP/MAS spectra
of a variety of steroids, including testosterone, hydrocortisone,
DHEA, spironolactone, vitamin D, deflazacort, and Prd tert-
butyl acetate, show multiplet patterns with splittings of
0.2–2.1 ppm, which are indicative of different steroidal confor-
mations.[37,38] In our previous study of steroid–cation
interaction,[39] we demonstrated by solid-state NMR spectros-
copy that the DHEA molecule is able to interact with Mg2+
and Ca2+. The cation interaction induces different conforma-
tional changes in DHEA. And the most perturbed 13C chemical
shifts were resolved from the rings A and D, suggesting that
Ca2+ and Mg2+ might interact with DHEA through the oxygen
atom attached to C17 involving an interaction of the cation
and lone pair of electrons. To validate this, we synthesized two
1
2
SCHEME 1 Molecular structures of dehydroepiandrosterone (DHEA),
(3S)-acetoxypregn-5-en-20-one (1), and (3S,20R)-acetoxypregn-5-en-
20-ol (2)

WU ET AL.
3
samples following the same preparation process as described
for the DHEA analog/Mg2+ mixtures (see Section 2). High-
resolution 13C CP/MAS NMR spectra of compounds 1 and
2 are displayed in Figures 1a and 2a, respectively, which
represent fingerprints of the DHEA analogs. A typical spec-
tral feature is that most of the 13C resonances in the broad
downfield region (45–210 ppm) are well dispersed, whereas
the remaining 13C signals in the narrow upfield region
(20–45 ppm) are somewhat crowded and overlapped. This is
similar to the 13C spectra of other steroidal molecules, such
as DHEA,[30–32] SPI,[29] hydrocortisone,[18] estradiol,[40] and
prednisolone.[18]
To achieve better spectral resolution, we then per-
formed 1H-filtered 13C NMR experiments and normal
FIGURE 1 13C NMR signals of compound 1: (a) Normal 1H–13C
CP/MAS spectrum, (b) 1H-filtered 13C CP/MAS spectrum, and (c) short-
contact-time 1H–13C CP/MAS spectrum. The 1H-filtered 13C and short-
contact-time 1H–13C CP/MAS served as complementary experiments for
acquiring the signals of nonprotonated and protonated C atoms,
respectively. Superposition of the signals of the nonprotonated and
protonated C atoms yields the normal CP/MAS signals. As shown,
improved spectral resolution facilitated the assignment of 13C resonances.
The inset shows the same spectrum in the overcrowded region of
20–40 ppm. Signals for the methylated carbon atoms, C18 and C19, are not
completely suppressed in the “nonprotonated” spectra
of analysis of the chemical shift deviations (CSDs), the
double-bond π electron arising from C5–C6 at ring B contrib-
utes to the cation interaction as well. The solid-state NMR
study of the DHEA analog/Mg2+ interaction can further
advance our understanding of the cation binding effect in the
neurosteroid molecule DHEA.
FIGURE 2 13C NMR signals of compound 2: (a) Normal 1H–13C
CP/MAS spectrum, (b) 1H-filtered 13C CP/MAS spectrum, and (c) short-
contact-time CP/MAS spectrum. The 1H-filtered 13C and short-contact-time
1H–13C CP/MAS served as complementary experiments for acquiring the
signals of nonprotonated and protonated C atoms, respectively.
Superposition of the signals of the nonprotonated and protonated C atoms
yields the normal CP/MAS signals. As shown, improved spectral resolution
facilitated the assignment of 13C resonances. The inset shows the same
spectrum in the overcrowded region of 20–42 ppm. Signals for the
methylated carbon atoms C18 and C19 are not completely suppressed in the
“nonprotonated” spectra
2
| RESULTS AND DISCUSSION
2.1 | Chemical shift assignments
In this solid-state NMR approach, we aimed to unravel the
neurosteroid DHEA/Mg2+ binding sites by the two DHEA
analogs 1 and 2 in the solid phase. Prior to the analysis of
the DHEA analog/Mg2+ mixtures, we prepared Mg2+-free

4
WU ET AL.
TABLE 1 13C chemical shifts of the free forms and Mg2+-bound forms of DHEA, 1 (3β-acetoxypregna-5-en-20-one), and 2 (3β-acetyloxypregna-5-ene) and
their chemical shift deviations (CSDs)a
C atom
DHEAb
[1]
DHEA/Mgb
[2]
CSDb,c
[2] – [1]
1
[3]
1/Mg
[4]
CSDc
[4] – [3]
2
[5]
2/Mg
[6]
CSDc
[6] – [5]
C1
36.63
33.44
71.30
42.07
143.16
119.71
33.03
32.21
51.42
37.45
20.50
32.21
48.34
54.39
22.76
35.81
224.15
13.01
21.84
37.96
34.37
71.85
42.07
143.84
119.09
32.11
30.36
50.08
37.96
19.68
31.70
47.62
51.21
22.97
37.96
218.92
13.42
22.56
1.33 (25%)
0.93 (18%)
0.55 (11%)
0.00 (0%)
39.04
39.25
27.9
0.21 (5%)
−0.26 (6%)
−0.82 (20%)
1.44 (35%)
2.16 (53%)
−4.06 (100%)
0.41 (10%)
1.54 (38%)
2.36 (58%)
−0.62 (15%)
0.77 (19%)
0.72 (18%)
−0.88 (22%)
0.10 (2%)
37.72
38.63
0.91 (30%)
C2
28.16
74.06
39.3
28.44
75.65
39.53
140.56
123.51
31.93
34.01
50.75
37.37
22.61
40.99
43.23
57.04
25.49
25.54
58.4
26.82
−1.62 (53%)
C3
73.24
40.74
140.92
121.71
32.06
34.06
52.19
37.04
22.87
40.74
43.3
74.01/73.09
38.63
−1.64/−2.56 (54/20%)
−0.90 (30%)
C4
C5
0.68 (13%)
−0.62 (12%)
−0.92 (18%)
−1.85 (35%)
−1.34 (26%)
0.51 (10%)
−0.82 (16%)
−0.51 (10%)
−0.72 (14%)
−3.18 (61%)
0.21 (4%)
138.76
125.77
31.65
32.52
49.83
37.66
22.1
142.20
1.64 (54%)
C6
121.51/120.48
32.32
−2.00/−3.03 (66/100%)
0.39 (13%)
C7
C8
33.96/33.09
51.31/50.96
37.91
−0.05/−0.92 (2/30%)
0.56/0.21 (18/7%)
0.54 (18%)
C9
C10
C11
C12
C13
C14
C15
C16
C17
C18
C19
C20
C21
COd
22.56
−0.05 (2%)
40.02
44.18
57.89
26.46
22.35
65.02
13.98
21.22
206.44
31.29
169.52
23.12
—
39.04
−1.95 (64%)
43.56/43.0/42.59
56.81
0.33/−0.23/−0.64 (11/8/21%)
−0.23 (8%)
57.99
24.51
22.87
63.74
15.22
21.28
207.98
31.19
170.7
21.64
—
−1.95 (48%)
0.52 (13%)
−1.28 (32%)
1.24 (31%)
0.06 (1%)
25.85
0.36 (12%)
2.15 (41%)
−5.23 (100%)
0.41 (8%)
26.82
1.28 (42%)
59.53
1.13 (37%)
12.32
19.09
70.34
23.95
170.60
21.74
—
13.42/11.06
19.94/19.07
71.39
1.10/−1.26 (36/42%)
0.85/−0.02 (28/1%)
1.05 (35%)
0.72 (14%)
1.54 (38%)
−0.10 (2%)
1.18 (29%)
−1.48 (36%)
1.12
24.66/23.84
171.88/169.52
21.79/21.12
—
0.71/−0.11 (23/4%)
1.28/−1.08 (42/36%)
0.05/−0.62 (2/20%)
0.92
d
CH3
Ave
—
—
1.19
Note. Ave: average of absolute values of all the chemical shift differences.
aChemical shifts are in units of ppm referenced to the glycine carboxyl resonance at 176.4 ppm, within an uncertainty of 0.1 ppm.
bSolid-state 13C NMR data were previously determined for DHEA reported elsewhere.[39]
cCSD: 13C chemical shift differences determined by subtracting the chemical shift of the Mg2+-bound form of DHEA (1 or 2) from that of free form. CSD values greater
than the average are highlighted in bold type. The variations of CSDs in percentage are normalized according to the highest variation as 100% shown in bracket.
dCarbonyl or methyl carbon of the acetyl group attached to C3.
1H–13C CP/MAS experiment using a short contact time of
50 μs. These two experiments were complementary: The
1H-filtered 13C NMR spectroscopy targeted “non-proton-
ated” C atoms, and the short CP/MAS approach targeted
“protonated” C atoms. The 13C NMR spectra of 1 and
2 are shown in Figures 1b and 2b, respectively. By virtue
of the higher spectral resolution compared to normal
CP/MAS spectra, the nonprotonated C signals of 1, in
particular, those of C10, C19, C21, and CH3, are distinct
without any overlap (Figure 1a–c). Thus, selective obser-
vation of nonprotonated and protonated 13C signals in
1 and 2 was achieved. Because the 13C chemical shifts
observed in the solid phase are highly correlated with
those in the solution phase (see Supporting Information,
Figures S1 and S2), the solid-state 13C chemical shift
assignments can be inferred from the assignments deter-
mined from solution NMR experiments without ambigu-
ity. The full 13C chemical shift assignments of 1 and
2 are listed in Table 1.
2.2 | Chemical shift deviations
In our previous solid-state NMR study, we demonstrated that
Mg2+ is capable of inducing a conformational change in
DHEA.[39] It was suggested that Mg2+ may associate with the
oxygen atom attached to C17 through a lone pair of electrons.
To further exploit the underlying cation interaction, in the cur-
rent work we have investigated the effect of Mg2+ upon the
steroid ring conformation of DHEA analogs, that is, com-
pounds 1 and 2, which differ slightly in the substituent
attached at C17 (see Scheme 1). The 13C NMR spectra of
compounds 1 and 2 incubated with Mg2+ at different Mg2+
concentrations are displayed in Figures 3a–e and 4a–e, respec-
tively. For compound 1, the NMR spectra of the mixtures with
Mg2+ showed two sets of 13C spectral signals. For clarity,
these two sets will be hereafter referred to as pertaining to the
free form and the Mg2+-associated bound form. For the free
form, the spectral pattern is exactly identical to that seen for
1 free from Mg2+ (Figure 1a). For the bound form, however,

WU ET AL.
5
FIGURE 3 Solid-state 13C NMR analysis of mixtures of compound 1 and
Mg2+. 1H–13C CP/MAS spectra of 1 (0.124 M) mixed with Mg2+ at
concentrations of (a) 0, (b) 0.076, (c) 0.124, (d) 0.294, (e) 0.434 M. In
comparison with that of compound 1 alone (a), the spectra of 1/Mg2+
FIGURE 4 Solid-state 13C NMR analysis of mixtures of compound 2 and
Mg2+. 1H–13C CP/MAS spectra of 2 (0.174 M) mixed with Mg2+ at
concentrations of (a) 0, (b) 0.13, (c) 0.174, (d) 0.324, (e) 0.466 M. In
comparison with that of compound 2 alone (a), the spectra of 2/Mg2+
feature an additional set of 13C signals. The intensities of the additional 13
C
feature two additional sets of 13C signals. The signal intensities of these sets
are comparable and are dependent on the Mg2+ concentrations, suggesting
that 2/Mg2+ might adopt different steroidal ring conformations; for details,
see text. For clarity, the signals in the regions 10–50, 50–80, and
110–190 ppm are shown in different scales. The 13C chemical shift
assignments of the 2/Mg2+ mixture are indicated (see also Table 1)
signals are dependent on the Mg2+ concentration, suggesting that they arise
from the Mg2+-bound form; for details, see text. For clarity, the signals in
the regions 10–50, 50–80, and 110–220 ppm are shown in different scales.
The 13C chemical shift assignments of the 1/Mg2+ mixture are indicated
(see also Table 1)
the spectral pattern is highly distinct, with the 13C signals
being shifted either upfield or downfield by 0.1–4.0 ppm.
It is possible that due to co-crystallization or polymor-
phism, the molecular conformation might not be unique, in
which the signals of relevant carbon atoms are split. As
shown in Figures 3a and 4a, the 13C NMR spectra of 1 and
2 in the absence of Mg2+ (same sample preparation as that
in the presence of Mg2+) revealed singlet patterns, indicating
a single conformation. That basically ruled out such possibil-
ities. The 13C spectra of compounds 1 and 2 in the presence
of Mg2+ feature additional sets of 13C spectral signals, which
were therefore attributable to the Mg2+-associated bound
forms (Figures 3b–e and 4b–e). It is noteworthy that the
spectral patterns of the bound forms remained invariant in
the DHEA analog/Mg2+ mixtures. To confirm whether the
multiple splitting patterns detected from 2 are due to molecu-
lar packing effects or to different molecular conformations,
we further measured the resonance line widths (i.e., C3,
C17, and C18) at half-maximum, which were well resolved
in these spectra (Figure 3a–e) and free of any overlapping
signals. Similar to other signals, these signals gave rise to
line widths in the range of 20–35 Hz, independent of the cat-
ion concentration. The line broadening of the resonances
may be due to the homogeneous broadening arising from
heteronuclear dipolar interactions or to inhomogeneous
broadening arising from different molecular packing effects.
If the packing inhomogeneity occurred because of the pres-
ence of cations, the resulting 13C signals would be expected
to have higher line widths compared to the signals in the

6
WU ET AL.
absence of cations. However, our data indicated that the line
widths were unaffected by increasing cation concentration,
suggesting that line broadening is primarily due to homoge-
neous broadening. We therefore interpret the multiplet pat-
terns as an indication of different steroidal conformations.
In view of the fact that 13C chemical shifts have been
routinely used to obtain information about molecular struc-
ture in various biological systems,[41,42] in the present study
we used CSD values in an attempt to delineate Mg2+ binding
sites in the DHEA analogs. We deduced 13C CSDs by sub-
tracting the chemical shifts of the Mg2+-bound form from
those of the free form. The CSDs induced in these molecules
by Mg2+ are listed in Table 1. For compound 1, a number of
C atoms, namely C4, C5, C6, C8, C9, C15, C17, C18, C20,
CO, and CH3, showed CSDs greater than the average pertur-
bation value (1.12 ppm). Within these residues, the most
perturbed were C6 (4.06 ppm), C9 (2.36 ppm), and C5
(2.16 ppm) in ring B, with CSDs reaching at most three
times the average value. For compound 2, several C atoms,
namely C2, C3, C5, C6, C12, C16, C17, C18, C20, and CO,
showed CSDs greater than the average perturbation value
(0.92 ppm). Among them, the most perturbed were C6 (2.0
and 3.03 ppm), C3 (1.64 and 2.56 ppm), and C12
(1.95 ppm) in rings B, A, and C, with CSDs twice as large
as the average value. Thus, in both 1 and 2, C6 showed the
greatest perturbation in terms of the CSD value, which is
associated with the C5–C6 double bond. We therefore pre-
signal intensity of C6, representing the Mg2+ binding site, as
a function of Mg2+ concentration in Figure 5. It is seen that
the signals reveal a nonlinear dependence on the Mg2+ con-
centration. In the simulations, we utilized a nonlinear corre-
lation function for experimental curve-fitting, in which the
relative intensity arising from the Mg2+-bound form was
correlated with the Mg2+ concentration by an exponential
relationship [1 – exp(−κμ)], where κ represents the binding
affinity and μ the Mg2+ concentration. The binding affinity
κ was then deduced from experimental-curve fitting as a
measure of metal cation binding affinity: that is, the higher
the value, the stronger the binding affinity. As can be seen,
compounds 1 and 2 showed similar responses with respect
to Mg2+, although compound 1 gave a slightly steeper curve
than compound 2, suggesting that the former has a some-
what higher affinity. In these analyses, a binding affinity κ
of 8.3 M−1 was extracted from the best curve-fit of the rela-
tive intensities of C6 of compound 1, whereas a lower value
of 5.0 M−1 (0.75) was derived from the same C atom of
2. Based on the higher κ value, we surmise that compound
1 has a moderately stronger Mg2+ binding affinity than com-
pound 2. As illustrated in Figure 5, DHEA got converted
into the Mg2+-bound form at a higher rate, in which the
complete conversion was observed at a lower Mg2+ concen-
tration of 0.12 M for DHEA/Mg2+, as compared to 0.6 M
for DHEA analog/Mg2+, revealing a fivefold difference.
And a binding affinity κ of 25 M−1 (3.3) was deduced from
DHEA/Mg2+ (Figure 5), which is three times greater than
that for 1 and five times greater than that for 2. Thus, DHEA
indeed has a higher Mg2+ binding affinity as compared to
sume that this double bond is involved in binding to Mg2+
.
Unlike DHEA, the greatest perturbation was detected from
the C17–O double bond (see Scheme 1 and Table 1). On this
basis, we used the C6 signal variations in the Mg2+ titration
experiment to extract the binding affinity, as elaborated
below.
In this solid-state NMR study, we find that the 13C NMR
spectra of DHEA and its analogs show both positive and
negative CSDs on incubating with MgCl2. It is thus possible
that the negative CSDs were caused by the negatively
charged chloride anions. And the chloride anions might be
associated with the ion–dipole interaction regions where the
Mg2+ and steroidal molecules are interacting, such as the D
ring of DHEA. If indeed this is the case, the electrons sur-
rounding C17–O are to be pulled toward the Mg2+, making
the C17 partially electron-positive. And the chloride anions
are drawn to the C17 as well as the neighboring atoms of
C17–O, giving rise to negative CSDs. As indicated in
Table 1, negative CSD values were observed from the C17
of DHEA and 1 as well as those in the neighborhood of
C17, such as C13, C18, and C19 of DHEA, 1, and
2, suggesting this might be the case.
FIGURE 5 Investigation of DHEA analogs/Mg2+ interaction by solid-state
NMR spectroscopy. The 13C intensity ratios of the C17 signals of DHEA
[39]
▲
(
) and C6 signals of compounds 1 (♦) and 2 (×) are plotted as a
function of Mg2+ concentration. The signal intensities were calculated with
reference to that of the C14 signal. The Mg2+ binding effect was analyzed by
nonlinear exponential curve-fitting simulation for the two compounds. Note
that the two compounds gave rise to two slightly different nonexponential
growth curves for the 13C signals arising from their Mg2+-bound forms; for
details, see text. In the simulations, a nonexponential curve was used to
2.3 | Mg2+ binding affinity
In order to determine Mg2+ binding affinities, we further
monitored the signal intensities of C5, C6, C20, and CO as a
function of Mg2+ concentration. We have plotted the relative
optimize the best fit, and Mg2+ binding affinities (κ of 25, 8.3, and 5.0 M−1
were deduced for compounds DHEA, 1, and 2, respectively
)

WU ET AL.
7
the DHEA analogs. Based on these binding affinities, we
conclude that the Mg2+ binding affinity of DHEA is much
stronger than that of the DHEA analogs, following a
decreasing order of DHEA >> 1 > 2.
otherwise. All non-aqueous reactions were carried out in
oven-dried glassware under a slightly positive pressure of
argon unless otherwise noted. Reactions were magnetically
stirred and monitored by thin-layer chromatography on silica
gel. Column chromatography was performed on silica gel of
40–63 μm particle size. Yields are reported for spectroscopi-
cally pure compounds. Melting points were recorded on an
Electrothermal Mel-Temp 1101D melting point apparatus
and were not corrected.
Compared to the alkyl ketones (C–O) of 1 and the ester
group of 2, the cyclic ketone group (C17–O) of DHEA
is capable of inducing ring strain and increasing the
S-character in the C–C bonds of ring D, resulting in a higher
degree of acidity. Because of this, the C17–O binding to
basic Mg2+ is stronger than that of C–O. similarly, for 1 and
2, the alkene group (C5–C6) has a higher acidity than the
ketone (C–O) and ester group; this explains why the C5–C6
binding to basic Mg2+ is stronger than the C–O binding.
We here show that the Mg2+ binding sites of DHEA ana-
logs are different from those previously observed for
DHEA.[39] In the case of DHEA, the most perturbed CSD
value was detected at C17, indicating that C17–O is
involved in the Mg2+ interaction, whereas in the case of
1 and 2 the greatest perturbations in terms of CSD value
were detected at C5 and C6 (see Table 1), suggesting that
Mg2+ prefers to interact with C5–C6, albeit weakly, medi-
ated by a π-electron pair. Here we propose that DHEA pos-
sesses two Mg2+ binding sites, that is, the C17–O and C5–6
double bonds, and the two binding sites function in a com-
petitive manner. Since the binding affinity of the former is at
least threefold stronger than the latter, one is normally
expected to have the C17–O for Mg2+ binding while having
the C5–C6 silent due to its lower affinity. However, in the
absence of C17–O, for example, in the cases of 1 and 2, one
is then able to have the C5–C6 binding site observable for
the detection.
3.2 | NMR measurements
The 1H–13C CP/MAS NMR spectra were acquired on a
Bruker Avance 300 MHz NMR spectrometer (Bruker Spe-
ctrospin, Rheinstetten, Germany), equipped with a 4-mm
double-resonance probe operating at 1H and 13C Larmor
frequencies of 300.13 and 75.47 MHz, respectively. The
1H-to-13C polarization transfer was optimized to fulfill the
Hartman–Hahn matching condition.[43] For normal CP
experiments, the contact time was set to 1 ms, and an rf field
1
strength of 41.0 kHz was chosen for both the H and 13C
channels. In the 1H-filtered experiments, a 41.5 μs delay was
1
applied immediately before data acquisition, allowing H-
1
coupled 13C signals to be attenuated under strong H–13C
dipolar coupling and resulting in acquisition of only the 13C
signals of nonprotonated carbon atoms. A complementary
short CP experiment was performed, in which a contact time
of 50 μs was used to detect only the 13C signals of proton-
1
ated carbon atoms. During data acquisition, H decoupling
by two-pulse phase modulation[44] was applied, with an rf
field strength of 79.3 kHz. Unless otherwise specified, the
13C spectra were acquired with a sample spinning frequency
of 8 kHz, regulated by a spinning controller to within about
Even though both 1 and 2 share a high degree of similar-
ity in the steroidal ring skeleton, as mentioned in the intro-
duction, it is possible that both compounds bind Mg2+ but
1
1 Hz. All H–13C CP/MAS experiments were conducted at
ambient temperature. The 13C chemical shifts were
referenced to the glycine carboxyl carbon signal at
176.4 ppm.
mediated by different mechanisms. In the presence of Mg2+
,
1 showed a single conformation while 2 gave rise to two
conformations, suggesting that 2 might have a higher degree
of molecular plasticity. Aside from this, in the case of 1, the
most perturbed chemical shifts were found at ring B, such as
C5, C6, C8, and C9, whereas in the case of the two most
perturbed chemical shifts were found to a larger extent
including rings A, B and C, for instance, C2, C3, C5, C6,
and C12. Based on our solid-state NMR observations, we
therefore interpret that though both 1 and 2 bind Mg2+, and
yet their underlying binding mechanisms are somewhat
dissimilar.
3.3 | Sample preparations
To prepare powdered samples, compounds 1 and 2 were dis-
solved in anhydrous methanol and combined with MgCl2
solution to prepare DHEA analog/Mg2+ mixtures, in which
1 (0.124 M) was mixed with Mg2+ at six different concen-
trations (0, 0.076, 0.124, 0.294, 0.334, and 0.434 M), and
2 (0.174 M) was mixed with Mg2+ at five different concen-
trations (0, 0.13, 0.174, 0.324, and 0.466 M), respectively.
After lyophilization, the DHEA analog/Mg2+ mixtures in
powdered form were packed into a 4-mm zirconium MAS
rotor and analyzed by high-resolution 1H–13C CP/MAS
solid-state NMR spectroscopy.
3
| EXPERIMENTAL
3.1 | General
3.4 | (3S)-Acetoxypregn-5-en-20-one (1)
All reagents used were commercially available and used
without further purification unless indicated otherwise.
All solvents were anhydrous grade unless indicated
Acetic anhydride (13 mL, 134 mmol, 14.1 equiv) was added
to a solution of pregnenolone (3.0 g, 9.48 mmol, 1 equiv) in

8
WU ET AL.
ACKNOWLEDGMENTS
pyridine (26 mL, 315 mmol, 33.2 equiv) over a period of
10 min. After stirring at room temperature for 32 h, the reac-
tion mixture was concentrated to dryness under reduced
pressure. The residue was redissolved in CH2Cl2, and the
solution was washed with saturated aqueous NaHCO3 solu-
tion (2 × 50 mL) and 1 N HCl (2 × 50 mL). The organic
phase was dried over MgSO4, filtered, and concentrated to
afford the desired compound 1 (3.33 g, 98%) as a white
We thank Claire Yang (Institute of Chemistry, Academia
Sinica) for preparing the manuscript. This work was
supported by the grant MOST 105-2113-M-011-022 from
the Ministry of Science and Technology.
REFERENCES
[1] P. V. Piazza, M. Le Moal, Brain Res. Brain Res. Rev. 1997, 25, 359.
[2] E. R. de Kloet, M. Joels, F. Holsboer, Nat. Rev. Neurosci. 2005, 6, 463.
[3] B. S. McEwen, P. J. Gianaros, Annu. Rev. Med. 2011, 62, 431.
[4] C. Sandi, Nat. Rev. Neurosci. 2004, 5, 917.
1
solid. M.p. 145–148ꢀC. H and 13C NMR spectra and 2D
COSY and HSQC spectra were acquired, see Figures S1–S4.
1H and 13C chemical shifts (see Table S1) and the melting
points are consistent with those reported in the litera-
ture. [45,46]
[5] R. C. Agis-Balboa, G. Pinna, A. Zhubi, E. Maloku, M. Veldic, E. Costa,
A. Guidotti, Proc. Natl. Acad. Sci. USA 2006, 103, 14602.
[6] E. E. Baulieu, P. Robel, M. Schumacher, Int. Rev. Neurobiol. 2001, 46, 1.
[7] N. A. Compagnone, S. H. Mellon, Front. Neuroendocrinol. 2000, 21, 1.
[8] R. C. Melcangi, A. G. Mensah-Nyagan, Neurochem. Int. 2008, 52, 503.
[9] K. Shibuya, N. Takata, Y. Hojo, A. Furukawa, N. Yasumatsu, T. Kimoto,
T. Enami, K. Suzuki, N. Tanabe, H. Ishii, H. Mukai, T. Takahashi,
T. A. Hattori, S. Kawato, Biochim. Biophys. Acta 2003, 1619, 301.
[10] J. J. Lambert, D. Belelli, D. R. Peden, A. W. Vardy, J. A. Peters, Prog.
Neurobiol. 2003, 71, 67.
3.5 | (3S,20R)-Acetoxypregn-5-en-20-ol (2)
Sodium borohydride (0.11 g, 2.79 mmol, 1 equiv) was
slowly added to
a
cold (0ꢀC) solution of (3s)-
acetoxypregn-5-en-20-one (1) (1.0 g, 2.79 mmol, 1 equiv)
in methanol (30 mL) and CH2Cl2 (5 mL). After stirring
for 1.25 h in an ice bath, the reaction was quenched with
saturated aqueous NH4Cl solution and extracted with
CH2Cl2 (2 × 50 ml). The combined organic extracts were
washed with saturated aqueous NaHCO3 solution and
brine, dried over MgSO4, filtered, and concentrated under
reduced pressure. The residue was purified by column
chromatography on silica gel (hexane/EtOAc, 3:1) to
afford the desired compound 2 (0.82 g, 82%) as a white
solid. Melting point 152–155ꢀC. 1H and 13C NMR spectra
and the 2D COSY and HSQC spectra were determined,
see Figures S5–S8. 1H and 13C chemical shifts (see
Table S2) and the m.p. were consistent with that of the
20(R) form reported in the literature.[47]
[11] J. J. Lambert, M. A. Cooper, R. D. Simmons, C. J. Weir, D. Belelli,
Psychoneuroendocrinology 2009, 34(Suppl 1), S48.
[12] M. Sedlacek, M. Korinek, M. Petrovic, O. Cais, E. Adamusova,
H. Chodounska, L. Vyklicky Jr. , Physiol. Res. 2008, 57(Suppl 3), S49.
[13] A. I. Czlonkowska, P. Krzascik, H. Sienkiewicz-Jarosz, M. Siemiatkowski,
J. Szyndler, A. Bidzinski, A. Plaznik, Pharmacol. Biochem. Behav. 2000,
67, 345.
[14] M. Vallee, W. Mayo, G. F. Koob, M. Le Moal, Int. Rev. Neurobiol. 2001,
46, 273.
[15] M. Bianchi, E. E. Baulieu, Proc. Natl. Acad. Sci. USA 2012, 109, 1713.
[16] L. Guth, Z. Zhang, E. Roberts, Proc. Natl. Acad. Sci. USA 1994, 91,
12308.
[17] K. Murakami, A. Fellous, E.-E. Baulieu, P. Robel, Proc. Natl. Acad. Sci.
USA 2000, 97, 3579.
[18] V. Suitchmezian, I. Jess, J. Sehnert, L. Seyfarth, J. Senker, C. Näther, Cryst.
Growth Des. 2008, 8, 98.
[19] S. Swallow, Progr. Med. Chem. 2015, 54, 65.
[20] M. Bianchi, J. J. Hagan, C. A. Heidbreder, Curr. Drug Targets CNS Neu-
rol. Disord. 2005, 4, 597.
[21] M. Bianchi, A. J. Shah, K. C. Fone, A. R. Atkins, L. A. Dawson,
C. A. Heidbreder, M. E. Hows, J. J. Hagan, C. A. Marsden, Synapse, 2009,
63, 359.
[22] R. C. Melcangi, L. M. Garcia-Segura, A. G. Mensah-Nyagan, Cell. Mol.
Life Sci.: CMLS 2008, 65, 777.
4
| CONCLUSIONS
To unravel the Mg2+ binding sites of the neurosteroid, we
have applied high-resolution 13C CP/MAS NMR spectros-
copy to DHEA and its analogs 1 and 2. The side chain
attached to C17 and the substituent attached to C3 were
modified in 1 and 2. In DHEA, the most perturbed C atom
was found to be C17, suggesting the C17–O double bond to
be responsible for the interaction with Mg2+. In these DHEA
analogs, unlike DHEA, the most perturbed C atom was
found to be C6 in ring B, suggesting that the C5–C6 double
bond of distinct π-electron character is responsible for the
interaction. In this solid-state NMR study, we demonstrated
that DHEA possesses two Mg2+ binding sites, that is, C17–
O and C5–C6 double bonds, acting in a competitive manner.
The Mg2+ binding of C17–O is mediated by the cation/lone-
pair electron interaction, whereas the binding of C5–C6 is
through cation/π interaction, in which the former has a
higher binding affinity (at least three times stronger) than the
latter.
[23] A. H. Payne, D. B. Hales, Endocr. Rev. 2004, 25, 947.
[24] R. H. Purdy; A. L. Morrow; J. R. Blinn; S. Paul. J. Med. Chem. 1990, 33,
1572.
[25] M. Vallee, S. Vitiello, L. Bellocchio, E. Hebert-Chatelain, S. Monlezun,
E. Martin-Garcia, F. Kasanetz, G. L. Baillie, F. Panin, A. Cathala, et al.,
Science 2014, 343, 94.
[26] J. J. Haddad, Prog. Neurobiol. 2005, 77, 252.
[27] G. A. Eby, K. L. Eby, Med. Hypotheses 2006, 67, 362.
[28] K. M. Kantak, M. A. Edwards, T. P. O'Connor, Pharmacol. Biochem.
Behav. 1998, 59, 159.
[29] K. Paradowska, I. Wawer, J. Pharm. Biomed. Anal. 2014, 93, 27.
[30] M. P. Marsan, I. Muller, C. Ramos, F. Rodriguez, E. J. Dufourc,
J. Czaplicki, A. Milon, Biophys. J. 1999, 76, 351.
[31] M. P. Marsan, W. Warnock, I. Muller, Y. Nakatani, G. Ourisson, A. Milon,
J. Org. Chem. 1996, 61, 4252.
[32] I. Schuler, A. Milon, Y. Nakatani, G. Ourisson, A. M. Albrecht,
P. Benveniste, M. A. Hartman, Proc. Natl. Acad. Sci. USA 1991, 88, 6926.
[33] O. Soubias, F. Jolibois, S. Massou, A. Milon, V. Reat, Biophys. J. 2005,
89, 1120.
[34] D. L. Gater; V. Réat, G. Czaplicki, O. Saurel, F. Jolibois, V. Cherezov,
A. Milon, Langmuir 2013, 29, 8031.
[35] O. Soubias; F. Jolibois, V. Reat, A. Milon, Chemistry 2004, 10, 6005.
[36] O. Soubias, F. Jolibois, V. Réat, A. Milon, Chem. Eur. J. 2004, 10, 6005.

WU ET AL.
9
[37] J. H. Yang, Y. Ho, D. L. M. Tzou, Magn. Reson. Chem. 2008, 46, 718.
[38] P. C. Shih, G. C. Li, K. J. Yang, W. Chen, D. L. Tzou, Steroids 2011,
76, 558.
[39] D. Wang, M. Chen, R. J. Chein, W. M. Ching, C. H. Hung, D. L. Tzou,
Steroids 2015, 96, 73.
[40] P. Jeschke, Chembiochem: Eur. J. Chem. Biol. 2004, 5, 571.
[41] G. Cornilescu, F. Delaglio, A. Bax, J. Biomol. NMR 1999, 13, 289.
[42] F. Castellani, B. van Rossum, A. Diehl, M. Schubert, K. Rehbein,
H. Oschkinat, Nature 2002, 420, 98.
[43] S. R. Hartmann, E. L. Hahn, Phys. Rev. 1962, 128, 2042.
[44] A. E. Bennett, C. M. Rienstra, M. Auger, K. V. Lakshmi, R. G. Griffin,
J. Chem. Phys. 1995, 103, 6951.
[47] R. D. Dawe, J. L. C. Wright, Can. J. Chem. 1987, 65, 666.
SUPPORTING INFORMATION
Additional supporting information may be found online in
the Supporting Information section at the end of this article.
How to cite this article: Wu D, Carillo KJD,
Tsai S-L, Shie J-J, Tzou D-LM. Solid-state nuclear
magnetic resonance investigation of neurosteroid
compounds and magnesium interactions. J Chin
Chem Soc. 2019;1–9. https://doi.org/10.1002/jccs.
201800458
[45] N. V. Kovganko, Z. N. Kashkan, Y. G. Chernov, Chem. Nat. Compd.
1992, 28, 589.
[46] S. W. Maurya, K. Dev, K. B. Singh, R. Rai, I. R. Siddiqui, D. Singh,
R. Maurya, Bioorg. Med. Chem. Lett. 2017, 27, 1390.