Substrate interaction with 5alpha-reductase enzyme: influence of the 17beta-chain chirality in the mechanism of action of 4-azasteroid inhibitors.

Article abstract of DOI:10.1016/S0039-128X(01)00114-3

A series of steroidal compounds were synthesized in order to evaluate the possible influence of the configuration of a stereocenter in the 17beta-side chain on the inhibitory activity on the enzyme 5alpha-reductase (5AR). For this purpose diastereomerically pure 4-azasteroids epimers at C-22 were prepared (compounds 1-11) and tested as inhibitors of 5AR in "in vitro" tests. The obtained data showed that in most cases the couples of epimers possess a significant difference in their biological activity. We also considered, for the tested molecules, a series of chemico-physical parameters in order to find a possible correlation with their biological activity. The findings allowed us to propose a model of the binding site of 5AR which comprises also, for 4-azasteroid inhibitors, the configurational aspect of the 17beta-side chain.

Full text of DOI:10.1016/S0039-128X(01)00114-3

Steroids 66 (2001) 803–810
Substrate interaction with 5␣-reductase enzyme: influence of the 17␤-
chain chirality in the mechanism of action of 4-azasteroid inhibitors
Paride Grisentia,*, Ambrogio Magnia, Vincenzo Olgiatia, Ada Manzocchib,
Patrizia Ferraboschib, Vincenzo Villanic, Rachele Pucciarielloc, Fabio Celottid
aPoli Industria Chimica SpA, Via Volturno 45/48, Quinto de Stampi, Rozzano, Milano, Italy
bDipartimento di Chimica e Biochimica Medica, Universita` degli Studi di Milano, Via Saldini 50, Milano, Italy
cDipartimento di Chimica, Universita` della Basilicata, Via N. Sauro 85, Potenza, Italy
dIstituto di Endocrinologia, Universita` degli Studi di Milano, Via Balzaretti 9, Milano, Italy
Received 17 May 2000; received in revised form 11 January 2001; accepted 25 January 2001
Abstract
A series of steroidal compounds were synthesized in order to evaluate the possible influence of the configuration of a stereocenter in the
17␤-side chain on the inhibitory activity on the enzyme 5␣-reductase (5AR). For this purpose diastereomerically pure 4-azasteroids epimers
at C-22 were prepared (compounds 1–11) and tested as inhibitors of 5AR in ‘in vitro’ tests. The obtained data showed that in most cases
the couples of epimers possess a significant difference in their biological activity. We also considered, for the tested molecules, a series of
chemico-physical parameters in order to find a possible correlation with their biological activity. The findings allowed us to propose a model
of the binding site of 5AR which comprises also, for 4-azasteroid inhibitors, the configurational aspect of the 17␤-side chain. © 2001
Elsevier Science Inc. All rights reserved.
Keywords: 4-Azasteroids; Testosterone 5␣ reductase; Inhibitors; Molecular modeling; Chirality; Biological activity
1. Introduction
and cloning of their cDNA’s, but only indirect information
is available on the three-dimensional structure of their bind-
ing sites. Data obtained from structure-activity relationship
studies with several synthetic derivatives showing variable
inhibitory activity on 5AR [13–14] are the only source of
information so far. Further structure-activity data would
provide considerable help to the drug designer. Several
studies [2–7] are available on the relationship between the
structure and the inhibitory activity of a class of steroidal
compounds, the 4-azasteroids. From these studies it appears
that, while the chemical nature and size of the functional
groups in rings A and B were extensively analyzed (even if
not completely defined), the effect of the size and the con-
figuration of the 17␤ side chain on the activity of 4-aza-
steroids has not yet been clarified. In fact, the binding site of
5AR seems able to accept both ‘small’ and ‘large’ substitu-
ents to position 17␤. In particular, groups of high steric
hindrance, like adamantane [15] or the long aliphatic chain
of cholesterol [16], can be inserted without loss of the
inhibitory activity. We therefore decided to study the inhib-
itory effect of 4-azasteroids differently substituted at posi-
tion 17␤. We have focused our attention on amides of
The study of selective inhibitors of 5␣-reductase (5AR),
the enzyme which catalyses the NADPH-dependent reduc-
tion of testosterone to dihydrotestosterone (DHT) [1], is at
present the focus of growing interest [2–4]. These mole-
cules show therapeutic potential in the treatment of several
diseases, whose pathogenic mechanism seems to be an ab-
normal formation of dihydrotestosterone (as probably oc-
curs in benign prostatic hyperplasia, acne, hyrsutism and
male pattern baldness) [5–7]. The recently discovered exis-
tence of two different genes encoding for two 5AR isoen-
zymes, types 1 and 2 [8], with different biochemical char-
acteristics and tissue distribution [9], gave a further impulse
to this research [10–12], since the possibility to discrimi-
nate between the two enzymes could allow the use of highly
selective drugs with minimal side effects. The primary
structure of these enzymes was determined after isolation
* Corresponding author. Tel.: ϩ39-02-8227-2530; fax: ϩ39-02-8246-
183.
E-mail address: pgrisenti@poli.it (P. Grisenti).
0039-128X/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved.
PII: S0039-128X(98)00114-3

804
P. Grisenti et al. / Steroids 66 (2001) 803–810
Series FTIR instrument. MS spectra were recorded on a
Hewlett-Packard 5988A instrument. Analytical TLC were
performed on silica gel Merck 60 F254 plates and column
chromatographies were carried out on silica gel Merck 60
(230–400 mesh). The MOLGEN-3 software [18] has been
used for both the optimization of the molecular structures
and the evaluation of physicochemical properties.
2.2. Preparation of compounds 1–16
Compounds 1–6, 9–11 were prepared by reaction of the
thiopyridyl ester [3] 12 with the required amino derivatives,
as reported in Scheme 1. Compounds 7 and 8 were prepared
according to Scheme 2 starting from the carboxy-derivative
13: intermediate 14 was prepared by reaction of acyl chlo-
ride derivative (obtained from 13 by reaction with oxalyl
chloride) and the ethyl ester of the required R- or S-amino-
acid. The epimers 14 were transformed in the corresponding
secosteroids 15 by the use of KMnO4/NaIO4 [2]. The next
step, the reaction with methylamine, leads to the lactamiza-
tion of the A ring and a complete conversion of the ester
moiety to methylamide. The unsaturated azasteroids 16
were then hydrogenated to obtain the desired compounds 7
and 8.
2.3. N-[(S)-phenylalanyl methyl amide]-4-methyl-3-oxo-4-
aza-5␣-androstane-17␤-carboxamide (1)
Fig. 1.
A suspension of S-phenylalanine hydrochloride (1.46 g,
7.282 mmol) in a mixture of dry THF (26 ml) and pyridine
(0.6 ml) was vigorously stirred at room temperature. After
10 min (2-pyridyl)-3-oxo-4-aza-5␣-androstane-17␤-thio-
carboxylate (12) (499 mg, 1.214 mmol) was added. The
reaction mixture was maintained under stirring at room
temperature for 18 h. After this period CH2Cl2 (5 ml) was
added and the reaction was maintained for an additional
18 h under stirring at room temperature. The reaction was
then heated for 4 h and the solvents evaporated under
vacuum. The residue was dissolved in CH2Cl2, washed with
an aqueous saturated NaHCO3 solution, dried with Na2SO4,
filtered and evaporated under vacuum to give a crude which
was purified by column chromatography (basic aluminium
oxide, CH2Cl2/CH3OH ϭ 98:2 v/v as eluant) affording pure
1 (0.55 g, 94%). For analytical purposes the product was
crystallized from ethyl acetate: mp 225°C; 1H NMR ␦ 0.60
(3 H, s, CH3-18), 0.87 (3 H, s, CH3-19), 2.44 (2 H, m, H-2),
2.69 (3 H, d, J ϭ 4.91 Hz, CH3-NH), 2.92 (3 H, s, CH3-4),
2.97–3.05 (3 H, overlapped, H-5 and CH2Ph), 4.57 (1 H, m,
CH-NH), 5.62 (1 H, broad q, NH-CH3), 5.98 (1 H, d, J ϭ
7.7 Hz, NH-CH), 7.18–7.32 (5 H, m, arom); IR 1657, 1622,
1496 cmϪ1; MS m/z 493 (Mϩ), 435, 332, 316, 288. Ele-
mental analysis calculated for C30H43O3N3: theoretical C ϭ
72.99%, H ϭ 8.78%, N ϭ 8.51%; found C ϭ 72.97%, H ϭ
8.76%, N ϭ 8.50%.
17␤-carboxylic acid with different aminoacids (1–9 in Fig.
1) and R- and S-phenyl ethyl amine (compounds 10–11 in
Fig. 1). The above compounds show the common feature of
possessing a stereogenic center in the alpha position (con-
ventionally named C-22) to the 17␤-carboxamide group.
The presence of a stereogenic center in the substituents
provides the opportunity of studying the effect of the con-
figuration at this position on the activity of the drug. Liter-
ature data report no examples of such chiral discrimination:
in fact even if a molecule with a chiral center at C-22 [17]
has been described, only the mixture of two epimers was
biologically tested.
2. Experimental
2.1. Materials and methods
Solvents and reagents were purchased from Fluka (Swit-
zerland). Melting points were determined with a Bu¨chi
apparatus (model 535) and are uncorrected. 1H NMR spec-
tra were obtained in CDCl3 on a Bruker AM-500 spectrom-
eter at 303 K or, when indicated, on a Varian 360 L spec-
trometer with tetramethylsilane as internal standard. 1H
NMR spectra (500 MHz) were also used to check the
diastereomeric purity of obtained products. IR spectra were
recorded in chloroform solution on a Perkin Elmer 1600
The following amino acid derivatives were prepared with
the same procedure: N-[(R)-phenylalanyl methyl amide]-

P. Grisenti et al. / Steroids 66 (2001) 803–810
805
4-methyl-3-oxo-4-aza-5␣-androstane-17-carboxamide (2):
mp 139–141°C; 1H NMR ␦ 0.47 (3 H, s, H-18), 0.86 (3 H,
s, H-19), 2.44 (2 H, m, H-2), 2.71 (3 H, d, J ϭ 4.9 Hz,
CH3-NH), 2.91 (3 H, s, CH3-4), 3.01 (1 H, dd, J ϭ 3.5, 12.6
Hz, H-5), 3.05 (2 H, d, J ϭ 7.4 Hz, CH2Ph), 4.68 (1 H, m,
CH-NH), 5.93 (1 H, d, J ϭ 7.7 Hz, NH-CH), 6.03 (1 H,
broad q, NH-CH3), 7.19–7.30 (5 H, arom); IR 1657, 1622,
1497 cmϪ1; MS m/z 493 (Mϩ), 435, 332, 316, 288. Ele-
mental analysis calculated for C30H43O3N3: theoretical C ϭ
72.99%, H ϭ 8.78%, N ϭ 8.51%; found C ϭ 73.01%, H ϭ
8.75%, N ϭ 8.47%.
(3 H, s, H-18), 0.90 (3 H, s, H-19), 2.40 (2 H, m, H-2), 3.05
(1 H, dd, J ϭ 3.5, 12.6 Hz, H-5), 3.97 (2 H, m, COCH2NH),
5.5 (1 H, s, 4-NH), 6.09 (1 H, broad t, NH-CH2); IR
1684(sh), 1653.5 cmϪ1; MS m/z 376 (Mϩ), 375, 361, 358,
318, 302, 274. Elemental analysis calculated for
C21H32O4N2: theoretical C ϭ 66.99%, H ϭ 8.57%, N ϭ
7.44%; found C ϭ 66.95%, H ϭ 8.54%, N ϭ 7.40%.
N-[(R)-␣-methylbenzyl]-4-methyl-3-oxo-4-aza-5␣-
1
androstane-17␤-carboxamide (10) H NMR ␦ 0.68 (3 H, s,
H-18), 0.84 (3 H, s, H-19), 1.45 (3 H, d, J ϭ 7.0 Hz,
CH3-CH), 2.38 (2 H, m, H-2), 2.87 (3 H, s, CH3N), 2.98 (1
H, dd, J ϭ 3.5, 12.6 Hz, H-5), 5.12 (1 H, m, CHPh), 5.57 (1
H, d, J ϭ 8.1 Hz, NH), 7.19–7.33 (5 H, m, arom); IR 1660,
1621, 1496 cmϪ1; MS m/z 436 (Mϩ), 421, 331, 317, 288.
Elemental analysis calculated for C28H40O2N2: theoretical
C ϭ 77.02%, H ϭ 9.23%, N ϭ 6.42%; found C ϭ 77.05%,
H ϭ 9.27%, N ϭ 6.39%.
N-[(R)-leucyl ethyl ester]-4-methyl-3-oxo-4-aza-5␣-an-
1
drostane-17␤-carboxamide (3) H NMR ␦ 0.70 (3 H, s,
H-18), 0.89 (3 H, s, H-19), 0.96 (3 H, d, J ϭ 7 Hz, CH3CH),
0.95 (3 H, d, J ϭ 6 Hz, CH3CH), 1.27 (3 H, t, J ϭ 7 Hz,
CH3CH2), 2.40 (2 H, m, H-2), 2.89 (3 H, s, 4-CH3), 3.0 (1
H, dd, J ϭ 3.5, 12.6 Hz, H-5), 4.20 (2 H, q, J ϭ 7 Hz,
CH3CH2O), 4.65 (1 H, m, CH-NH), 5.67 (1 H, d, J ϭ 8 Hz,
NH-CH); IR 1732, 1672, 1622, 1503.5 cmϪ1; MS m/z 474,
459, 429, 418, 401, 316, 288. Elemental analysis calculated
for C28H46O4N2: theoretical C ϭ 70.05%, H ϭ 9.77%, N ϭ
5.90%; found C ϭ 70.01%, H ϭ 9.75%, N ϭ 5.84%.
N-[(S)-leucyl ethyl ester] 4-methyl-3-oxo-4-aza-5␣-andro-
stane-17␤-carboxamide (4) mp 161–162°C; 1H NMR ␦
0.68 (3 H, s, H-18), 0.88 (3 H, s, H-19), 0.94 (3 H, d, J ϭ
6.0 Hz, CH3CH), 0.95 (3 H, d, J ϭ 6.0 Hz, CH3CH), 1.27
(3 H, t, J ϭ 7.0 Hz, CH2CH3), 2.40 (2 H, m, H-2), 2.97 (3
H, s, 4-H), 3.02 (1 H, dd, J ϭ 3.5, 12.6 Hz, H-5), 4.18 (2 H,
m, CH3CH2O), 4.64 (1 H, m, CH-NH), 5.59 (1 H, d, J ϭ 8.4
Hz, NH); IR 1732, 1671.5, 1621, 1503.5 cmϪ1; MS m/z 474
(Mϩ), 459, 429, 418, 401, 316, 288. Elemental analysis
calculated for C28H46O4N2: theoretical C ϭ 70.05%, H ϭ
9.77%, N ϭ 5.90%; found C ϭ 70.08%, H ϭ 9.78%, N ϭ
5.85%. N-[(S)-phenylalanyl methyl ester]-3-oxo-4-aza-5␣-
N-[(S)-␣-methylbenzyl]-4-methyl-3-oxo-4-aza-5␣-
1
androstane-17␤-carboxamide (11) mp 155°C; H NMR ␦
0.60 (3 H, s, H-18), 0.86 (3 H, s, H-19), 1.48 (3 H, d, J ϭ
7.0 Hz, CH3CH), 2.43 (2 H, m, H-2), 2.91 (3 H, s, CH3N),
3.01 (1 H, dd, J ϭ 3.5, 12.6 Hz, H-5), 5.15 (1 H, m, CHPh),
5.49 (1 H, d, J ϭ 7.7 Hz, NH), 7.2–7.3 (5 H, m, arom); IR
1659, 1621, 1495 cmϪ1; MS m/z 436 (Mϩ), 421, 331, 317,
288. Elemental analysis calculated for C28H40O2N2: theo-
retical C ϭ 77.02%, H ϭ 9.23%, N ϭ 6.42%; found C ϭ
77.07%, H ϭ 9.29%, N ϭ 6.40%.
N-[(S)-leucyl ethyl ester]-androst-4-en-3-oxo-17␤-car-
boxamide (14). A solution of androst-4-en-3-oxo-17␤-car-
boxylic acid (13) (6.92 g, 21.9 mmol) in dry toluene (104
ml) and pyridine (2.43 ml) was cooled at 10°C and a
solution of oxalyl chloride (2.822 ml) in toluene (11.66 ml)
was added under vigorous stirring. The reaction was stirred
at 10°C (1 h), then a suspension of S-leucine ethyl ester
hydrochloride (20.85 g, 106.8 mmol) in toluene (104 ml)
and pyridine (8.807 ml) was added. After heating at 40°C
for 4 h, under vigorous stirring, the reaction mixture was
cooled at room temperature and acidified to pH 2, filtered by
suction and the precipitate washed with toluene. The col-
lected filtrates were washed with aqueous saturated
NaHCO3 to neutrality, dried on Na2SO4, filtered and the
solvents were evaporated under vacuum to give the crude
product (14.6 g). After silica gel column chromatography
(CH2Cl2/CH3OH ϭ 9:1 v/v), pure 14 was obtained (6.79 g,
66% yield). 1H NMR (60 MHz) ␦ 0.75 (3 H, s, H-18), 1.15
(3 H, s, H-19), 0.9–2.8 (complex), 5.1 (1 H, s, exch, NH),
1
androstane-17␤-carboxamide (5) mp 175–176°C; H NMR
␦ 0.65 (3 H, s, H-18), 0.88 (3 H, s, H-19), 2.39 (2 H, m,
H-2), 3.04 (1 H, dd, J ϭ 4.2, 11.9 Hz, H-5), 3.07–3.17 (2 H,
m, CH2Ph), 3.72 (3 H, s, COOCH3), 4.89 (1 H, m, CH (L)),
5.51 (1 H, s, 4-NH), 5.67 (1 H, d, J ϭ 7.7 Hz, NH), 7.09 (2
H, m, arom), 7.27 (3 H, m, arom); IR 1740, 1657, 1499
cmϪ1; MS m/z 480 (Mϩ), 465, 421, 318, 302, 274. Elemen-
tal analysis calculated for C29H40O4N2: theoretical C ϭ
72.47%, H ϭ 8.39%, N ϭ 5.83%; found C ϭ 72.42%, H ϭ
8.35%, N ϭ 5.80%.
N-[(R)-phenylalanyl methyl ester]-3-oxo-4-aza-5␣-an-
1
drostane-17␤-carboxamide (6) mp 193°C; H NMR ␦ 0.51
(3 H, s, H-18), 0.89 (3 H, s, H-19), 2.40 (2 H, m, H-2),
2.9–3.6 (2 H, m, H-5 and CHPh), 3.17 (1 H, dd, J ϭ 5.6, -14
Hz, CH-Ph), 3.73 (3 H, s, OCH3), 4.93 (1 H, m, CH (D)),
5.48 (1 H, br s, 4-NH), 5.65 (1 H, d, J ϭ 8.1 Hz, NH-CH),
7.13 (2 H, m, arom), 7.27 (3 H, m, arom); IR 1740, 1657,
1498 cmϪ1; MS m/z. 480 (Mϩ), 421, 318, 302, 274. Ele-
mental analysis calculated for C29H40O4N2: theoretical C ϭ
72.47%, H ϭ 8.39%, N ϭ 5.83%; found C ϭ 72.44%, H ϭ
8.34%, N ϭ 5.79%. N-Glycinyl-3-oxo-4-aza-5␣-andro-
1
stane-17␤-carboxamide (9) mp Ͼ 280°C; H NMR ␦ 0.69
Scheme 1.

806
P. Grisenti et al. / Steroids 66 (2001) 803–810
5.9 (1 H, s, H-4); IR 1735, 1670, 1503 cmϪ1. Elemental
analysis calculated for C28H43O4N: theoretical C ϭ
73.49%, H ϭ 9.47%, N ϭ 3.06%; found C ϭ 73.47%, H ϭ
9.45%, N ϭ 3.01%.
and the solution was slowly heated to reach 180°C, and kept
at this temperature for 20 min. The reaction mixture was
then cooled to room temperature and acidified with 3N HCl.
The precipitate was recovered by filtration, dried under
vacuum at 80°C and used without further purification (4.25
g) in the next step. For analytical purposes a sample has
been purified by silica gel column chromatography (1:80
w/w; CH2Cl2/CH3OH ϭ 8:2 v/v as eluant); MS m/z 457
(Mϩ), 442, 412, 401, 384, 299, 271. Elemental analysis
calculated for C27H43O3N3: theoretical C ϭ 70.86%, H ϭ
9.47%, N ϭ 9.18%; found C ϭ 70.80%, H ϭ 9.41%, N ϭ
9.12%
2.4. Preparation of secosteroid 15
N-[(S)-leucyl ethyl ester]-androst-4-en-3-oxo-17␤-car-
boxamide (14) (3.1 g, 6.78 mmol) was dissolved in t-
butanol (41.34 ml) and an aqueous solution of Na2CO3 (1.1
g in 6.7 ml) was added. The reaction mixture was heated at
80°C and a hot aqueous solution (42 ml) of KMnO4 (80 mg)
and NaIO4 (10.5 g) was added under stirring, during 45 min.
After 1 h at reflux the mixture was cooled at room temper-
ature, kept under stirring for 2 h, filtered through a celite
pad, concentrated under vacuum to near 1/3 of initial vol-
ume. The solution obtained was cooled at 10°C and care-
fully acidified under vigorous stirring to pH ϭ 2 with 6 N
HCl. The precipitate was recovered by filtration, dissolved
in CH2Cl2 and the solution was dried on Na2SO4. After
evaporation under vacuum pure 15 (3 g, 93% yield) was
obtained: 1H NMR (60 MHz) ␦ 0.75 (3 H, s, H-18), 0.9–2.8
(complex), 5.4 (2 H, m, exchangeable); IR 1731, 1702.5,
1673, 1505 cmϪ1; MS m/z 459, 432, 404, 372, 318. Ele-
mental analysis calculated for C27H43O6N: theoretical C ϭ
67.90%, H ϭ 9.07%, N ϭ 2.93%; found C ϭ 67.85%, H ϭ
9.11%, N ϭ 2.90%.
2.4.2. N-[(S)-leucyl methylamide]-4-methyl-3-oxo-4-aza-
5␣-androstane-17␤-carboxamide (8)
The crude compound 16 (1.4 g, 3.06 mmol) was dis-
solved in acetic acid (20 ml) and hydrogenated on 5% Pt/C
(1.4 g) at 3 atm, 75°C during 6 h. The solvent was evapo-
rated under vacuum after filtration through a celite pad to
give a residue which was dissolved in CH2Cl2 and washed
with an aqueous solution of NaHCO3 to neutrality. The
solution was then dried on Na2SO4, and after filtration
evaporated to dryness affording a crude. After purification
by silica gel column chromatography (1:100 ϭ w/w,
CH2Cl2/CH3OH ϭ 95:5 v/v as eluant), pure 8 was obtained
1
(1.2 g, 2.61 mM, 85% yield), mp 225–226°C; H NMR ␦
0.61 (3 H, s, H-18), 0.84 (3 H, s, H-19), 0.92 (3 H, d, J ϭ
5.9 Hz, CH3CH), 0.94 (3 H, d, J ϭ 6.3 Hz, CH3CH), 2.40
(2 H, m, H-2), 2.74 (3 H, d, J ϭ 4.9 Hz, CH3NH), 2.89 (3
H, s, NCH3), 3.01 (1 H, dd, J ϭ 3.5, 12.6 Hz, H-5), 4.45
(1H, m, CH (L)), 5.90 (1 H, d, J ϭ 8 Hz, NH-CH), 6.65 (1
H, m, NH-CH3); IR 1669 (sh), 1654, 1622, 1502 cmϪ1; MS
m/z 460 (Mϩ1), 459 (Mϩ), 444, 429, 401, 331, 316. Ele-
mental analysis calculated for C27H45O3N3: theoretical C ϭ
70.55%, H ϭ 9.87%, N ϭ 9.14%; found C ϭ 70.60%, H ϭ
9.82%, N ϭ 9.10%.
2.4.1. N-[(S)-leucyl methylamide]-4-methyl-3-oxo-4-aza-4-
androstene-17␤-carboxamide (16)
Compound 15 (5 g, 10.47 mmol) was dissolved under
stirring in ethylene glycol (50 ml), at room temperature.
After cooling to Ϫ10°C methylamine (26 ml) was added
Analogously was prepared N-[(R)-leucyl methylamide]-
4-methyl-3-oxo-4-aza-5␣-androstane-17␤-carboxamide (7)
1
mp 150°C; H NMR ␦ 0.68 (3 H, s, H-18), 0.87 (3 H, s,
H-19), 0.92 (3 H, d, J ϭ 5.9 Hz, CH3CH), 0.94 (3 H, d, J ϭ
6.3 Hz, CH3CH), 2.42 (2 H, m, H-2), 2.75 (3 H, 17␤ d, J ϭ
4.2 Hz, CH3NH), 2.90 (3 H, s, N-CH3), 3.01 (1 H, dd, J ϭ
3.5, 12.6 Hz, H-5), 4.50 (1 H, m, CH (D)), 5.97 (1 H, d, J ϭ
8.4, NH-CH), 6.69 (1 H, broad q, NH-CH3); IR 1663, 1622,
1505 cmϪ1; MS m/z 460 (Mϩ1), 459 (Mϩ), 429, 444, 401,
316. Elemental analysis calculated for C27H45O3N3: theo-
retical C ϭ 70.55%, H ϭ 9.87%, N ϭ 9.14%; found C ϭ
70.58%, H ϭ 9.85%, N ϭ 9.11%.
3. Biological activity
3.1. Pharmacological testing
At the moment the unique compound of this class of
molecules entered into therapy is represented by Finasteride
Scheme 2.

P. Grisenti et al. / Steroids 66 (2001) 803–810
807
prostate, freshly dissected from an adult animal (Crl:
CD®RR Charles River Italia S.p.A., 200–300 g of body
weight), containing about 100 ␮g of protein (measured
according the method of Bradford [21]) are incubated with
increasing amounts of the test substance (initially in the
range of 10Ϫ5–10Ϫ8 M; lower concentrations have been
included for some very active products in subsequent tests)
or without any inhibitor. Each point is evaluated in dupli-
cate. The incubation medium is a Krebs-Ringer buffer so-
lution (250 ␮l-pH 7.1) containing a NADPH generating
system (NADPH, disodium salt (Boehringer Mannheim)
3.32 ϫ 10Ϫ3 M; glucose 6-phosphate, disodium salt (Boehr-
inger Mannheim) 11.76 ϫ 10Ϫ2 M and glucose 6-phosphate
dehydrogenase from yeast grade 1 (Boehringer Mannheim),
3.5 ϫ 10Ϫ2 U.I. or 1 mg/ml) and [14C]testosterone 3.2 ϫ
10Ϫ6 M (specific activity Ϸ 56.9 mCi/mM, Amersham
England). The incubation is carried out at 37°C in a
Dubnoff metabolic shaker under a stream of O2/CO2 98:2
(v:v), for 2 h. At the end of the incubation the reaction is
stopped by freezing the samples to Ϫ20°C. Tritium labeled
DHT and 3␣-diol (about 5000 dpm each) are added to each
sample in order to evaluate the recoveries. The metabolites
formed are extracted twice with diethyl ether, non-radio-
active steroids (40 ␮g/100 ml) are added to each sample in
order to aid visualization of the steroid on the TLC plates.
Samples are then separated on a thin layer silica gel plate
(Merck 60 F254, DC) eluting three times with a mixture of
dichloromethane-diethyl ether (11:1 v/v). DHT and 3␣-diol
spots are identified with iodine vapors, scraped off and the
radioactivity counted in a Packard 1600 CA liquid scintil-
lation spectrometer. Quench corrected dpm of the isotope
are obtained by a calibration standard curve. The identifi-
cation of the metabolites has been performed, when the
Fig. 2. Finasteride.
(Fig. 2) [19]; for this reason we have utilized this molecule
as reference standard for the pharmacological tests. The
inhibitory effect of the compounds synthesized as above on
the 5AR activity has been tested in the homogenate of rat
prostate at neutral pH, comparing the molecules’ IC50s
(Table 1) with that of Finasteride (prepared according pub-
lished procedure [3]).
In these assay conditions the activity measured is mostly
that of type 1 isozyme which has a wide range of pH
optimum (6–8.5), while type 2 isozyme has a narrow acidic
pH optimum (around 5.0) and it is almost inactive at pH 7.
Metabolic studies at acidic pH have not been effected since
it is almost impossible to have a clear separation of the two
enzymatic activities changing the incubation medium pH, at
least in the rat prostate. The two recombinant enzymatic
isoforms expressed in a suitable yeast expression system
[20] will be used in future experiments.
The 5␣-reductase activity has been assayed radioenzy-
matically as routinely done in our laboratory. The transfor-
mation of testosterone into its main 5␣-reduced metabolites
(dihydrotestosterone or DHT and 5␣-androstane-3␣,17␤-
diol or 3␣-diol) has been evaluated after the in vitro incu-
bation of [14C]testosterone with an aliquot of normal adult
rat prostate fresh homogenate. Briefly the method is the
following: aliquots of an homogenate of the rat ventral
3
method was validated by recrystallization to constant H/
14C ratio. The 5␣-reductase activity is expressed as pg of
steroids (DHTϩ3␣-diol) formed in the incubation time per mg
of protein. The enzymatic activity in all the assays has been
normalized to the finasteride, which was included in each
experiment. The intraassay variation was 10–30% and the
Table 1
5␣-Reductase in vitro inhibitory activities; values are expressed in relation to finasteride
Compound
Configurationa
(a) group
(b) group
Relative activity
1
2
3
4
5
6
7
8
S
R
R
S
S
R
R
S
-
-CH2Ph
-CONHCH3
-COOEt
-CH2CH(CH3)2
-CH2Ph
-COOCH3
-CONHCH3
-CH2CH(CH3)2
-H
-CONHCH3
-CH2Ph
-CH2CH(CH3)2
-COOEt
-COOCH3
-CH2Ph
-CH2CH(CH3)2
-CONHCH3
-H
0.00
5.25
0.60
3.84
0.15
3.88
1.60
1.51
0.00
16.60
56.60
1.00
9
10
11
R
S
-
-Ph
-CH3
-CH3
-Ph
Finasteride
a Related to the stereogenic center at C-22.

808
P. Grisenti et al. / Steroids 66 (2001) 803–810
interassay variation 30–50%. The curves obtained are statisti-
cally evaluated and IC50 calculated by the program ALLFIT
developed by D. Rodbard and colleagues, NIH, USA in the
Macintosh version prepared by Vincenzo Guardabasso and
Giovanni Angeli, Biomathematics and Applied Informatic Re-
search Unit, Consorzio Mario Negri Sud.
4. Results and discussion
The data reported in Table 1 show that inhibition ob-
tained in vitro with this series of derivatives is significantly
different for the two epimers of compounds 1–6 and 10–11,
while the glycine derivative (9) shows no inhibitory activity
and compounds 7 and 8 exhibit the same activity. Although
our screening is restricted to a small number of compounds,
the results obtained on the couples of epimers (1–6 and
10–11) show that the configuration of the stereogenic center
in the 17␤-side chain can be related to different levels of
pharmacological activity. The above data indicate that the
inhibition of 5AR is influenced not only by the nature of the
group bound to the 17␤-carbamoyl group but also by its
stereochemistry. The diastereoselectivity should therefore
be taken into account in designing molecules with a stereo-
center in 17␤-side chain, in order to optimize the interaction
with the binding site of 5AR. As compounds 2, 4, 6, 11
show a higher relative activity compared to 1, 3, 5 and 10,
the former can be defined as eutomers (more biologically
active stereoisomers) and the latter as distomers (less active
isomers), according to Lehmann et al. [22]. The above
different biologic effect would not be detectable if a mixture
of both stereoisomeric forms had been used. On the basis of
the obtained results we have attempted to develop a coher-
ent model to elucidate some conformational aspects of the
binding site of 5AR. In our model, the side-chain dependent
portion of the binding site of 5AR which would accommo-
date the 17␤ chain of 4-azasteroids, consists of 2 cavities, A
Fig. 4. Optimized geometries for the more active 4-azasteroid epimers 2, 4,
6 and 11 in Table 1. Both 7 and 8 diastereoisomers have been reported,
being of nearly equal activity. Molecules have been projected with the
larger group downward and the smaller upward, according to the scheme
of Fig. 3.
and B, in which the (a) and (b) substituents at the stereo-
center in position 22 are to be lodged (Fig. 3).
Tables 2, 3 and 4 show a number of chemico-physical
parameters for substituents (a) and (b) for all the molecules
tested. We were hoping to be able to relate the inhibitory
effect of a molecule in an epimeric pair (e.g. 1 and 2, 3 and
4) to some of the chemico-physical parameters we had
considered. The parameters were calculated on optimized
molecular structures. The MOLGEN-3 software [23] was
used for both the optimization of the structures and the
evaluation of the properties.
Fig. 4 shows the optimized molecular structures for more
active epimers 2, 4, 6, 11 and for the nearly equally active
couple of epimers 7 and 8. Table 2 shows values of log P
Table 2
Lipophilicity of substituents at chiral C-22: calculated molar refractivity
(MR) and logP for tested compounds
Compound
Relative
activity
logP(a)
logP(b)
MR(a)
MR(b)
1
2
3
4
5
6
7
8
10
11
0
1.88
Ϫ0.35
0.66
1.44
1.88
0.32
Ϫ0.35
1.44
1.69
0.21
Ϫ0.35
1.88
1.44
0.66
0.32
1.88
1.44
Ϫ0.35
0.21
1.69
30.99
13.67
16.51
20.67
30.99
11.71
13.67
20.67
24.88
6.84
13.67
30.99
20.67
16.51
11.71
30.99
20.67
13.67
6.84
5.25
0.60
3.84
0.15
3.88
1.60
1.51
16.6
56.6
24.88
Fig. 3.

P. Grisenti et al. / Steroids 66 (2001) 803–810
809
Table 3
lengths, there is no preferential spatial arrangement of (a)
and (b) in the A and B cavities, respectively. Table 1 shows
that the A cavity can accept substituents as long as a phenyl
group (dH 5.44 Å, compound 10, relative activity 16.6), but
not as a benzyl group (dH 6.15 Å, compounds 1 and 5,
relative activities 0 and 0.15, respectively). In Table 4 are
listed other spatial parameters (volumes and areas accessi-
ble to the solvent), also showing a direct correlation with the
inhibitory activity of the members of the epimeric pairs. A
good correlation also appears to exist between the activity
and the degree of compactness (Wiener and Randic topo-
logical indices [18]). For each pair, the more active epimer
is the one that possesses the more extended, less compact,
(b) substituent, in agreement with the previously discussed
correlation to other parameters. Of the two indices, the
Randic index is the more suitable, as it appears to have
definite values for the greatest majority of the substituents.
In conclusion, our model’s predictions, as borne out by
the data obtained, are the following:
Spatial parameters of substituents at chiral C-22: v(a) or v(b)
calculated van der Waals volumes and dH(a) or
(Å) between the chiral carbon atom and the farther H of the groups (a)
or (b) of tested compounds
ϭ
ϭ calculated distance
(b)
Compound
Relative
activity
v(b)
v(a)
dH(b)
dH(a)
1
2
3
4
5
6
7
8
10
11
0
34.4
56.5
36.3
41.3
33.3
57.0
41.9
33.1
14.2
48.2
57.1
34.3
42.4
39.3
57.1
32.9
33.8
42.1
47.8
14.9
4.21
6.16
4.80
5.68
4.38
6.17
4.81
4.21
2.19
5.45
6.14
4.56
5.71
4.78
6.15
4.38
4.57
4.80
5.44
2.20
2.25
0.60
3.84
0.15
3.88
1.60
1.51
16.6
56.6
(partition coefficient in the system n-octanol/water) and
molar refractivity [24] of the molecules studied. These val-
ues are measures of the lipophilicity of substituent groups
(a) and (b). The data show that there is no correlation
between inhibitory activity and lipophilicity. Table 3 lists
spatial parameters v and dH (van der Waals volumes [25]
and the distance between the stereocenter and the furthest
hydrogen in the substituents (a) and (b)). Substituent (b) is
longer than (a) between 20% and 150% for pairs 1 and 2, 3
and 4, 5 and 6, 10 and 11. While in the case of van der
Waals volumes v there is not a direct correlation with the
inhibitory activity levels, such correlation exists in the case
of dH values. This suggests that what is important for a
significant inhibitory effect is not the right ‘bulk,’ but the
proper ‘length’: to be a good inhibitor, a molecule must
have a ‘long’ (b) substituent, able to fit into the B cavity
(Fig. 3). In the opposite configuration, (b ‘short’ and a
‘long’), the molecule would not fit well in the binding site,
and the inhibitory effect would be much lower or not exis-
tent. This is borne out by the results obtained for pair 7 and
8, where similar dH values correspond to very close inhib-
itory activities. As the substituents are of very similar
the binding site of 5AR (type 1 isozyme) consists of two
cavities which are meant to accommodate the two sub-
stituents bound to C-22 stereocenter, at the ␣ position of
the ␤ carbamoyl group in the 4-azasteroid molecules
(Fig. 1)
these cavities have different sizes, but their lipophilic
character is not well defined. Cavity A is smaller than
cavity B, and the two are arranged as shown in Fig. 3.
The size and shape of cavity A are such as to accommo-
date a phenyl group (5.4 Å long), but not a benzyl group
(6.15 Å long)
if the arrangement around the C-22 stereocenter is such
that the ‘short’ (a) group fits in the ‘small’ A cavity (and,
conversely, the long (b) group fits in the ‘large’ B cav-
ity), the interaction between the inhibitor and the enzyme
is strong enough to produce an inhibitory effect
for the other epimer of the pair, where the spatial orien-
tation is the opposite, the ‘long’ (a) group cannot fit well
into the ‘small’ A cavity. The enzyme-inhibitor interac-
Table 4
Other spatial parameters of substituents at chiral C-22: Vsol (a) or (b) and Ssol (a) or (b) ϭ calculated volumes (Å3) and areas (Å2) accessible to the solvent.
Calculated Wiener and Randic indices as measures of molecular compactness
Compound
Relative
activity
Vsol (b)
Vsol (a)
Ssol (b)
Ssol (a)
Wiener (b)
Wiener (a)
Randic (b)
Randic (a)
1
2
3
4
5
6
7
8
10
11
0
14.1
19.6
12.3
14.3
13.5
19.8
14.3
12.6
6.00
19.0
21.2
14.4
15.9
12.6
20.1
13.5
13.5
14.0
18.6
6.40
30.1
41.1
29.8
31.3
28.5
42.0
33.5
27.0
13.8
39.1
44.5
29.1
35.1
29.4
43.9
28.1
29.1
32.8
38.8
14.9
u.d.
10
9
20
10
u.d.
9
10
u.d.
u.d.
u.d.
10
20
9
u.d.
10
10
9
1.914
3.394
1.732
2.414
1.914
3.394
1.732
1.914
u.d.
3.394
1.914
2.414
1.732
3.394
1.914
1.914
1.732
3
2.25
0.60
3.84
0.15
3.88
1.60
1.51
16.6
56.6
u.d.
u.d.
3
u.d.

810
P. Grisenti et al. / Steroids 66 (2001) 803–810
[11] Jones CD, Audia JE, Lawhorn DE, McQuaid LA, Neubauer BL, Pyke
tion and the resulting relative activity are therefore re-
duced
when the (a) and (b) substituents are of equal length the
corresponding level of inhibition is almost the same:
either group can fit equally well into cavity A or B
AJ, Pennington PA, Stamm NB, Toomey RE, Hirsch KS. Nonsteroi-
dal inhibitors of human type I steroid 5␣-reductase. J Med Chem
1993;36:21–423.
[12] Frye SV, Haffner CD, Maloney PR, Mook RA Jr, Dorsey GF Jr,
Hiner RN, Cribbs CM, Wheeler TN, Ray JA, Andrews RC, Batchelor
KW, Bramson HN, Stuart JD, Schweiken SL, van Arnold J, Croom S,
Bichett DM, Moss ML, Tian G, Unwalla RJ, Lee FW, Tippin TK,
James MK, Grizzle MK, Long JE, Schuster SV. 6-Azasteroids: struc-
ture-activity relationships for inhibition of type 1 and 2 human 5␣-
reductase and human adrenal 3␤-hydroxy-⌬5-steroid dehydrogenase/
3-keto-⌬5-steroid isomerase. J Med Chem 1994;37:2352–60.
[13] Thigpen AE, Russell DW. Four-amino acid segment in steroid 5␣-
reductase 1 confers sensitivity to finasteride, a competitive inhibitor.
J Biol Chem 1992;267:8577–83.
Further research is needed, aimed primarily at obtaining
a better definition of the ‘large’ cavity, and at identifying
clearly the possible stereospecific effect of the spatial fea-
tures of the substituents on isozyme 2 as well as on isozyme
1.
Acknowledgments
[14] Tian G, Stuart JD, Moss ML, Domanico PL, Bramson HN, Patel IR,
Kadwell SH, Overton LK, Kost TA, Mook RA Jr, Frye SV, Batchelor
KW, Wiseman JS. 17␤-(N-tert-Butylcarbamoyl)-4-aza-5␣-androstan-
1-en-3-one is an active site-directed slow time-dependent inhibitor of
human steroid 5␣-reductase 1. Biochemistry 1994;33:2291–6.
[15] Steinberg NG, Salzmann TN, Rasmusson GH, Patel GF. 17␤-N-
substituted adamantyl/norbornanyl carbamoyl-4-aza-5␣-androst-1-
en-3-ones and androstan-3-ones. European Patent Application No 0
547 1991:687.
We thank Ministero dell’Universita` e Ricerca Scientifica
(MURST) for financial support and Dr. M.L. Sartirana for
helpful discussion and linguistic revision.
References
[16] Bakshi RK, Patel GF, Rasmusson GH, Baginsky WF, Cimis G,
Ellsworth K, Chang B, Bull H, Tolman RL, Harris GS. 4,7␤-Dimeth-
yl-4-azacholestan-3-one (MK-386) and related 4-azasteroids as selec-
tive inhibitors of human type I 5␣-reductase. J Med Chem 1994;37:
3871–4.
[17] Panzeri A, Nesi M, Di Salle E. 17␤-Substituted 4-aza-5␣-androstan-
3-one derivatives. International Patent Application No.WO 94/ 1994:
03476.
[18] Bogdanov B, Nikolic´ S, Trinastic´ N. On the three dimensional Wiener
number. J Math Chem 1989;3:299–309.
[19] Kenny B, Ballard S, Blagg J, Fox D. Pharmacological options in the
treatment of benign prostatic hyperplasia. J Med Chem 1997;40:
1293–315.
[20] Poletti A, Celotti F, Motta M, Martini L. Characterisation of rat 5␣-
reductase type 1 and type 2 expressed in Saccharomyces cerevisiae.
Biochem J 1996;314:1047–52.
[21] Bradford MM. A rapid and sensitive method for the quantitation of
microgram quantities of protein utilizing the principle of protein-dye
binding. Anal Biochem 1976;72:248–54.
[22] Lehmann FPA, Rodrigues de Miranda JF, Arie¨ns EJ. Stereoselectivity
and affinity in molecular pharmacology. Prog Drug Res 1976;20:101–
42.
[1] Saunders FJ. Biology of the prostate, and related tissue. EP Vollmer:
Washington DC; US Gov. Print. Off., pp. 139–59.
[2] Rasmusson GH, Reynolds GF, Utne T, Jobson RB, Primka RL,
Berman C, Brooks JR. Azasteroids as inhibitors of rat prostatic 5␣-
reductase. J Med Chem 1984;27:1690–701.
[3] Rasmusson GH, Reynolds GF, Steinberg NG, Walton E, Patel GF,
Liang T, Cascieri MA, Cheung AH, Brooks JR, Berman C. Azaster-
oids: structure-activity relationships for inhibition of 5␣-reductase,
and of androgen receptor binding. J Med Chem 1986;29:2298–315.
[4] Holt DA, Levy MA, Ladd DL, Oh H-J, Erb JM, Heaslip JI, Brandt M,
Metcalf BW. Steroidal A ring aryl carboxylic acids: a new class of
steroid 5␣-reductase inhibitors. J Med Chem 1990;33:937–42.
[5] Anderson KM, Liao S. Selective retention of dihydrotestosterone by
prosthetic nuclei. Nature (London) 1968;219:277–9.
[6] Bruchovsky N, Wilson JD. The conversion of testosterone to 5␣-
androstan-17␤-ol-3-one by rat prostate in vivo and in vitro. J Biol
Chem 1968;243:2012–21.
[7] Metcalf BW, Levy MA, Holt DA. Inhibitors of steroid 5␣- reductase
in benign prostatic hyperplasia, male pattern baldness and acne. TiPS
1989;10:491–5.
[8] Russell DW, Wilson JD. Steroid 5␣-reductase: two genes/two en-
zymes. Ann Rev Biochem 1994;63:25–61.
[23] Baricic P, Mackov M. MOLGEN- 1993:3.
[24] Ghose AK, Pritchett A, Crippen GM. Atomic physicochemical pa-
rameters for three dimensional structure directed quantitative struc-
ture-activity relationships III: modeling hydrophobic interactions.
J Comput Chem 1988;9:80–90.
[25] Gavezzotti A. The calculation of molecular volumes and the use of
volume analysis in the investigation of structured media and of
solid-state organic reactivity. J Am Chem Soc 1983;105:5220–5.
[9] Normington K, Russell DW. Tissue distribution and kinetic charac-
teristics of rat steroid 5␣-reductase isozymes. J Biol Chem 1992;267:
19548–54.
[10] Holt DA, Yamashita DS, Konialian-Beck AL, Luengo JI, Abell AD,
Bergsma DJ, Brandt M, Levy MA. Benzophenone- and indolecar-
boxylic acids. Potent type-2 specific inhibitors of human steroid 5␣-
reductase. J Med Chem 1995;38:13–5.