Synthesis of 17β-hydroxysteroid dehydrogenase type 10 steroidal inhibitors: Selectivity, metabolic stability and enhanced potency

Article abstract of DOI:10.1016/j.ejmech.2020.112909

17beta-Hydroxysteroid dehydrogenase type 10 (17β-HSD10) is the only mitochondrial member of 17β-HSD family. This enzyme can oxidize estradiol (E2) into estrone (E1), thus reducing concentration of this neuroprotective steroid. Since 17β-HSD10 possesses properties that suggest a possible role in Alzheimer's disease, its inhibition appears to be a therapeutic strategy. After we identified the androsterone (ADT) derivative 1 as a first steroidal inhibitor of 17β-HSD10, new analogs were synthesized to increase the metabolic stability, to improve the selectivity of inhibition over 17β-HSD3 and to optimize the inhibitory potency. From six D-ring derivatives of 1 (17-C[dbnd]O), two compounds (17β-H/17α-OH and 17β-OH/17α-C[tbnd]CH) were more metabolically stable and did not inhibit the 17β-HSD3. Moreover, solid phase synthesis was used to extend the molecular diversity on the 3β-piperazinylmethyl group of the steroid base core. Eight over 120 new derivatives were more potent inhibitors than 1 for the transformation of E2 to E1, with the 4-(4-trifluoromethyl-3-methoxybenzyl)piperazin-1-ylmethyl-ADT (D-3,7) being 16 times more potent (IC₅₀ = 0.14 μM). Finally, D-ring modification of D-3,7 provided 17β-OH/17α-C[tbnd]CH derivative 25 and 17β-H/17α-OH derivative 26, which were more potent inhibitor than 1 (1.8 and 2.4 times, respectively).

Full text of DOI:10.1016/j.ejmech.2020.112909

Journal Pre-proof
Synthesis of 17β-hydroxysteroid dehydrogenase type 10 steroidal inhibitors:
selectivity, metabolic stability and enhanced potency
Sophie Boutin, René Maltais, Jenny Roy, Donald Poirier
PII:
S0223-5234(20)30881-3
DOI:
https://doi.org/10.1016/j.ejmech.2020.112909
EJMECH 112909
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 23 June 2020
Revised Date: 28 September 2020
Accepted Date: 3 October 2020
Please cite this article as: S. Boutin, R. Maltais, J. Roy, D. Poirier, Synthesis of 17β-hydroxysteroid
dehydrogenase type 10 steroidal inhibitors: selectivity, metabolic stability and enhanced potency,
European Journal of Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112909.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2020 Elsevier Masson SAS. All rights reserved.

Eur. J. Med. Chem. (Third revised version)
Synthesis of 17
β-hydroxysteroid dehydrogenase type 10 steroidal inhibitors:
selectivity, metabolic stability and enhanced potency
Sophie Boutina,b, René Maltaisa, Jenny Roya and Donald Poiriera,b,*
a Laboratory of Medicinal Chemistry, Endocrinology and Nephrology Unit, CHU de Québec – Research
Center, Québec, QC, Canada
b Department of Molecular Medicine, Faculty of Medicine, Université Laval, Québec, QC, Canada
(*) Corresponding Author
Donald Poirier
Laboratory of Medicinal Chemistry
CHU de Québec – Research Center (CHUL, T4-42)
2705 Laurier Boulevard
Québec, QC, G1V 4G2, Canada
Tel.: 1-418-654-2296; Fax: 1-418-654-2298; E-mail: Donald.poirier@crchul.ulaval.ca

Abstract:
17beta-Hydroxysteroid dehydrogenase type 10 (17β-HSD10) is the only mitochondrial
member of 17β-HSD family. This enzyme can oxidize estradiol (E2) into estrone (E1), thus
reducing concentration of this neuroprotective steroid. Since 17β-HSD10 possesses properties
that suggest a possible role in Alzheimer’s disease, its inhibition appears to be a therapeutic
strategy. After we identified the androsterone (ADT) derivative 1 as a first steroidal inhibitor of
17β-HSD10, new analogs were synthesized to increase the metabolic stability, to improve the
selectivity of inhibition over 17β-HSD3 and to optimize the inhibitory potency. From six D-ring
derivatives of 1 (17-C=O), two (17β-H/17α-OH and 17β-OH/17α-C≡CH) were more
metabolically stable and did not inhibit the 17β-HSD3. Moreover, solid phase synthesis was used
to extend the molecular diversity on the 3β-piperazinylmethyl group of the steroid base core.
Eight over 120 new derivatives were more potent inhibitors than 1 for the transformation of E2
to E1, with the 4-(4-trifluoromethyl-3-methoxybenzyl)piperazin-1-ylmethyl-ADT (D-3,7) being
16 times more potent (IC50 = 0.14 μM). Finally, D-ring modification of D-3,7 provided 17β-
OH/17α-C≡CH derivative 25 and 17β-H/17α-OH derivative 26, which were more potent
inhibitor than 1 (1.8 and 2.4 times, respectively).
Keywords: 17β-HSD10, Alzheimer disease, enzyme, inhibitor, steroid, solid-phase synthesis.

1. Introduction
Alzheimer’s disease (AD) is a neurodegenerative disorder and the most common cause of
dementia [1]. About 50 million people are living with dementia worldwide and it is estimated
that this prevalence will almost double every 20 years [2]. AD cause the gradual loss of
important neuronal functions as language, motor functions, but especially memory. These are
clinical symptoms, but neuropathologically, AD is characterized by the brain atrophy and the
abundance of extracellular amyloid beta (Aβ) plaques [1]. Human 17β-hydroxysteroid
dehydrogenase (17β-HSD) is a family of oxidoreductases that catalyzes the interconversions of
ketones and alcohols. These enzymes transform functional groups at position 17 of the steroid
backbone depending on their substrate specificity [3]. Some 17β-HSDs have thus key roles in up
or down regulation of androgens, estrogens and neurosteroids, making them interesting
therapeutic targets in many diseases [3-5].
Mitochondrial 17β-HSD type 10 (17β-HSD10), a member of 17β-HSD family, possesses
properties that suggest a possible role in AD [6,7]. One of these properties is its capacity to
transform estradiol (E2) into estrone (E1). In fact, E2 is one substrate among others of 17β-
HSD10 (Km = 14-43 μM, Fig. 1) [7-10]. The E2 concentration in brains of AD patients being
reduced, the inhibition of 17β-HSD10 could help to restore E2 levels and promote its
neuroprotective effects. 17β-HSD10 is also known for the formation of a high affinity neurotoxic
complex with Aβ-42, the peptide responsible of the formation of Aβ plaque in the brain. In
mitochondria, Aβ-42 can bind to 17β-HSD10, disrupts the oxidoreductase function of the
enzyme and causes dysfunction and neuronal cell death [11]. Furthermore, several studies
showed that inhibiting this neurotoxic complex formation is a promising therapeutic approach
[6,7, 12-14]. It is also important to note that higher concentrations of 17β-HSD10 are found in
the brain of AD patients [7,15]. Consequently, the development of 17β-HSD10 inhibitors offer
the opportunity to validate the role of 17β-HSD10 in AD and provide a new therapeutic approach
against this disease.
O
OH
17β-HSD10
+
NADH
+
NAD+
HO
HO
17β-Estradiol (E2)
Estrone (E1)

Fig. 1. Oxidation of the 17β-OH of estradiol catalysed by 17β
-HSD10 and cofactor NAD+ to form
estrone, a 17-ketosteroid, and NADH.
There are few known 17β-HSD10 inhibitors
[
16-18
]
. One of them is AG18051, an
. Benzothiazole and
frentizole derivatives were also synthesized and evaluated as potent inhibitors of 17β-HSD10
20-28 . None of these inhibitor, including AG18051, were however based on a steroid scaffold.
In fact, our research group reported compound 1 (Fig. 2) as the first steroidal inhibitor of 17β-
HSD10 29,30 . This 5α-androstane-3α,17β-diol derivative inhibited (IC50 = 0.55 μM) the
irreversible inhibitor with a pyrimidine core identified by Kissinger et al [19
]
[
]
[
]
transformation of E2 to E1 by HEK-293[17β-HSD10] cells, but it also inhibited 17β-HSD3, an
enzyme that catalyzes the reduction of 4-androstene-3,17-dione (4-dione) into testosterone (T).
Since a low level of androstane (T or 5α-dihydrotestosterone) is associated with an increased risk
of AD
[31], and considering the expression of 17β-HSD3 in hippocampus [32], the
type 10/type 3 selectivity of compound 1 needed to be improved. Furthermore, a first round of
metabolic stability assay also showed that compound 1 was weakly stable in the conditions of the
assay using human liver microsomes, highlighting a potential issue toward its translation into a
drug.
Three different strategies (Fig. 2) were tested to improve the biological activities of 17β-
HSD10 steroidal inhibitor 1. In the first one, D-ring derivatives of 1 (modifications at C16 or
C17) were synthesized using classical chemistry in solution to increase its metabolic stability and
type 10/type 3 selectivity. In the second strategy, libraries of amide, sulfonamide, urea, thiourea
and amine derivatives of 1 (side chain modifications) were generated by parallel solid phase-
synthesis to increase its inhibitor potency and selectivity. Finally, in the third strategy, two
hybrid compounds were synthesized to combine the results of strategies 1 and 2.

Fig. 2. Three different strategies (#1-3) behind the optimization of lead compound 1. 1) Derivatization of
compound 1 to improve the metabolic stability and selectivity for 17 -HSD10 over 17 -HSD3 (D-ring
derivatives 2-7, Schemes 1 and 2); 2) Side chain modification of compound 1 to optimize the inhibitory
potency for 17 -HSD10 (libraries A-D members, 120 compounds represented by 8-11, Scheme 3); and 3)
β
β
β
Two hybrid inhibitors, compounds 25 and 26, combining best elements from steps 1 and 2 (Scheme 4).
Selected positions 3, 16, 17, 18 and 19 as well as key A- and D-rings of steroid scaffold were
represented.
2. Results
2.1 Chemical synthesis of D-ring derivatives 2-7 (first strategy)
The lead compound 1 was first synthesized from epi-androsterone (epi-ADT; 12) or
dihydrotestosterone (DHT; 15a) using the conditions previously published [33], and then used as
starting material for the synthesis of 2-4 (Scheme 1). The C-17 carbonyl of 1 underwent a
stereoselective reduction using sodium borohydride to give 17β-OH derivative 2. The compound
3 (17β-OH/17α-CH3) was synthesized through a Grignard reaction with methyl magnesium
iodide. Compound 4b (17β-OH/17α-C≡CH) was synthesized in two steps by adding
lithium(trimethylsilyl)ethylinide giving intermediate 4a, which underwent hydrolysis with
K2CO3 in methanol to afford 4b [34]. As previously reported [35], the steric hindrance of the
axial methyl-18 is responsible for the stereoselectivity of these three reactions involving the C-17
carbonyl and providing 17β-OH configuration in 2, 3 and 4a.

Scheme 1. D-ring steroid derivatives 2 (17
Reagents and conditions: (a) NaBH4, MeOH, 0°C (4 h); (b) CH3MgI (3.0 M in Et2O), toluene, 80°C (4 h)
to rt (16 h); (c) TMS-C CH, CH3Li, Et2O, THF, 0°C to rt (16 h); (d) K2CO3 5%, MeOH, rt (17 h).
β-OH), 3 (17β-OH/17α-CH3) and 4b (17β-OH/17α-C≡CH).
≡
For the synthesis of 17α-OH derivative 5 (Scheme 2), we first attempted to start from 2
by performing a Mitsunobu reaction followed by a hydrolysis of the intermediate ester 36].
[
However, the first step, the inversion of the C-17β alcohol by reacting with 4-nitrobenzoic acid,
triphenylphosphine and diethyl azodicarboxylate (DEAD) did not lead to the corresponding 17α-
ester. One hypothesis is the presence of two tertiary amines in the piperazine derivative 2, which
could promote the formation of a salt in these conditions and thus prevent any further reaction.
After this unsuccessful attempt, compound 5 was obtained in four steps from commercially
available 5α-dihydrotestosterone (DHT, 15a) (Scheme 2). The chiral center at position 17 (17β-
OH) of 15a was first inversed in two steps (Mitsunobu and hydrolysis reactions) to give the
corresponding 17α-alcohol 15c. The ketone at C-3 was then transformed into the oxirane 16
using the conditions of the Johnson-Corey-Chaykovsky reaction [37,38] and this oxirane was
regioselectively opened by the secondary amine 18 to afford 5. The building block 18, not
commercially available, was obtained by refluxing 3-methoxybenzyl chloride in ethanol (17)
with piperazine in excess.

Scheme 2. Synthesis of D-ring steroid derivatives 5 (17α-OH), 6 (17-CF2) and 7 (17-C=O/16-C(CH3)2).
Reagents and conditions: (a) 4-Nitrobenzoic acid, PPh3, DEAD, toluene, 80°C (18 h); (b) KOH 10%,
MeOH, rt (6 h); (c) (CH3)3SOI, NaH, DMSO, 50°C (16 h); (d) 1-(3-Methoxybenzyl)-piperazine (18),
EtOH, 60°C (24 h); (e) Piperazine, DCM, rt (19 h).
C-17-difluoro derivative 6 was obtained from 19 (Scheme 2), which was synthesized
previously in five steps from epi-ADT (12) as reported in literature [39]. Briefly, the 3β-alcohol
of 12 was first protected as an acetate and this intermediate reacted with DAST reagent, allowing
the addition of two fluorides on the C-17 ketone. Then, the ester at C-3 was hydrolyzed, the
corresponding alcohol oxidized, and the corresponding ketone transformed into the oxirane 19.
Finally, 19 reacted with piperazine derivative 18 to afford 6. C-16 dimethyl derivative 7 was
generated from 20, which was also synthesized from epi-ADT (12) in five steps as reported in
literature [39]. Briefly, after protecting the alcohol of 12 as an acetate derivative, the
dimethylation at C-16 was afforded using sodium hydride and methyl iodide in excess. The
deprotection of the alcohol and its oxidation to a ketone provided the oxirane 20, which then
reacted with 18 to afford the final compound 7. A direct dimethylation of 1 to obtain 7 was not
attempted since the methyl iodide used in the methylation step would have react with the
piperazine nitrogens, producing a salt.
2.2 Chemical synthesis of libraries A, B, C and D (second strategy)

To generate libraries A to D members, represented by compounds 8-11, the androstane
derivative 22 must first be synthetized from epi-ADT (12) and, then, loaded on the glycerol
polymer-bound (Scheme 3). Briefly, and as previously published
[
40
]
, the ketone at position 17
-TSA
of epi-ADT was protected as an acetal via a transacetalization with ethylene glycol and
p
using a Dean-Stark apparatus. The 3β-OH was then oxidized using tetrapropylammonium
perruthenate (TPAP), N-methylmorpholine N-oxide (NMO) and molecular sieves, as desiccant,
and this ketone was transformed into the oxirane 21 using the Johnson-Corey-Chaykovsky
conditions
position 3 to give a piperazino derivative, which was protected as a Fmoc derivative using Fmoc-
-succinimide and NaHCO3 as base. Finally, the acetal at position 17 was hydrolysed under
[37, 38]. An excess of piperazine was used to open regioselectively the oxirane 21 at
O
acidic conditions to afford 22 in 72% yield for three steps. The protection of the secondary amine
of the piperazino derivative was necessary since the acidic conditions needed for the
acetalization could favorize the intermolecular formation of an imine between the piperazine and
the 17-ketone, forming a dimer, and thus preventing the loading of the steroid on the solid
support (resin).
Scheme 3. Parallel solid-phase synthesis of amide (A-1 and A-2), sulfonamide (B-1), urea and
thiourea (C-1) and amine (D-1, D-2 and D-3) libraries of steroid derivatives 8-11. Reagents and
conditions: (a) Ethylene glycol, p-TSA, toluene, Dean-Stark apparatus, reflux (24 h); (b) TPAP, NMO,

DCM, molecular sieves, 0°C (2 h); (c) (CH3)3SOI, NaH, DMSO, 50°C (16 h); (d) Piperazine, EtOH,
reflux; (e) Fmoc-O-succinimide, NaHCO3, THF/H2O (5:1), rt (16h); (f) HCl/MeOH (1.0 M):DCM
(75:25), rt (16 h); (g) Glycerol polymer-bound, trimethyl-orthoformate, p-TSA (10%), toluene, rt (17 h);
(h) Piperidine/DMF (2:8), rt (3.5 h); (i) Carboxylic acid, HBTU, DIPEA, DMF, rt (3 h); (j) HCl / MeOH /
DCM (25:75), rt (20 h); (k) Sulfonyl chloride, TEA, DCM, rt (3 h); (l) Isocyanate or isothiocyanate,
DCM, rt (5 h); (m) Aldehyde, AcOH (1%), NaBH3CN, DMF, rt (5 h).
Compound 22 was added to a suspension of glycerol polymer bound (loading capacity of
1.00 mmol/g) in presence of p-TSA to provoke the acetalization. Tri-methyl orthoformate was
added to scavenge water produced as a side product during the formation of 23a. The Fmoc of
resin 23a was cleaved in presence of piperidine, a weak base commonly used for this type of
reaction, to give resin 23b. The global loading of 22 after Fmoc hydrolysis (resin 23b) was
established to 0.41 mmol/g from weight difference. The glycerol polymer-bound was selected
since the ketal functionality generated is stable under basic conditions used to produce, from 23b
in parallel synthesis, the library members represented by compounds 8 - 11. Advantageously, the
cleavage of the steroids from resins is easily done by a hydrolysis under weak acidic conditions,
thus giving the desired ketosteroids in good yields (libraries A-D; Supplementary data).
Amide derivatives 8 (libraries A-1 and A-2) were synthetized by reacting equal portions
of resin 23b with diverse carboxylic acids (21 for each library), which were previously activated
with HBTU. The final step of this solid-phase synthesis required an acid cleavage (HCl
generated in situ) to provide the free androstane derivatives 8 (A-1,1 to A-1,21 and A-2,1 to A-
2,21). For the amide library A-1, the HPLC purity of all 21 synthetized compounds ranged from
71.4% to 99.4%, with an average purity of 95.2 ± 6.1% (Table 1). For the amide library A-2, 20
derivatives of 21 were selected and the HPLC purity ranged from 76.8% to 99.7%, with an
average purity of 95.6 ± 5.6%.
Table 1. Assessment of solid-phase chemical synthesis of libraries A1 to D3
Library
Number
Family Compound
Number of
Number of
Range of
Average of
Purity
Building
Blocks
21
Selected
Purity (%)b
Compoundsa
(%)b,c
A
A
B
C
D
D
D
1
2
1
1
1
2
3
Amides 8
Amides 8
Sulfonamides 9
Ureas/Thioureas 10
Amines 11
21
20
16
14
16
17
16
71.4 – 99.4
76.8 – 99.7
65.8 – 99.3
68.0 – 98.5
60.0 – 95.7
65.0 – 98.3
72.0 – 97.2
95.2 ± 6.1
95.6 ± 5.6
87.2 ± 11.7
89.6 ± 10.4
85.2 ± 9.6
87.8 ± 8.3
92.3 ± 7.1
21
21
21
21
21
21
Amines 11
Amines 11

a Compounds with NMR and HPLC purity over 60%
b For selected compounds
c Reported as purity ± SD
To obtain sulfonamide derivatives 9 (library B-1), equal portions of resin 23b were
placed in presence of a solution of each sulfonyl chloride to allow a nucleophilic substitution.
The free sulfonamide derivatives 9 (B-1,1 to B-1,21) were obtained after the acid cleavage. From
this third library (B-1), five compounds were eliminated because of their low NMR or HPLC
purity. Thus, 16 sulfonamide derivatives, with a range of HPLC purity between 65.8% and
99.3% and an average purity of 87.2 ± 11.7%, were selected to undergo biological assays.
To produce urea and thiourea derivatives 10 (library C-1), equal portions of resin 23b
were reacted with a solution of each appropriate isocyanate or isothiocyanate under basic
conditions. After final cleavage, compounds 10 (C-1,1 to C-1,21) were obtained. A smaller
number of compounds were selected from this urea and thiourea library since the purity of some
members was not high enough. From library C-1, 14 urea and thiourea derivatives were selected
for biological assays. These compounds had a range of HPLC purity between 68.0% and 98.5%
with an average purity of 89.6 ± 10.4%. The difference of reactivity of the building blocks may
explain why more compounds were not selected. In fact, when the resins were cleaved, a mixture
of the desired compound 10 and the non-derived amine (3β-piperazinyl-methyl-ADT) were
obtained in some cases. Since we always used the same reaction conditions, indifferently of the
building blocks, maybe more time would have favorized completion of the reaction.
Amine derivatives 11 (libraries D-1, D-2 and D-3) were synthetized by a reductive
amination. Equal portions of resin 23b were treated in presence of each aldehyde under low
acidic conditions to promote the formation of an iminium. Then, a solution of sodium
cyanoborohydride was added to reduce the iminium into the corresponding tertiary amine. After
cleavage of the acetal link, amines 11 (D-1,1 to D-1,21, D-2,1 to D-2,21 and D-3,1 to D-3,21)
were successfully obtained. For library D-1, 15 amine derivatives were selected, with a range of
HPLC purity between 60.0% and 95.7%, and an average purity of 85.2 ± 9.6%. For library D-2,
17 amine derivatives were selected, with a range of HPLC purity between 65.0% and 98.3%, and
an average purity of 87.8 ± 8.3%. Finally, for library D-3, 16 amine derivatives were selected,
with a range of HPLC purity between 72.0% and 97.2%, and an average purity of 92.3 ± 7.1%.

2.3 Chemical synthesis of hybrid inhibitors 25 and 26 with dual modifications (third strategy)
The most potent inhibitor resulting from libraries A-D, compound D-3,7, was modified at
position C17 to generate two D-ring derivatives (compounds 25 and 26). Compound 25 was
synthesized from D-3,7 (Scheme 4) following the two steps procedure used for the synthesis of
4b (Scheme 1). Briefly, lithium(trimethylsilyl)ethylinide was added to ketone D-3,7 to give the
intermediate 24, which underwent hydrolysis with K2CO3 in methanol to afford 25 (17β-
OH/17α-CCH). The second hybrid, compound 26, was obtained from 16 also in two steps. First,
the 17α-OH-oxirane 16 was opened regioselectively with piperazine to generate a secondary
amine and, next, it was reacted with 4-trifluoromethyl-3-methoxybenzaldehyde and sodium
cyanoborohydride to afford 26.
Scheme 4. Synthesis of compounds 25 and 26. Reagents and conditions: (a) TMS-C≡CH, CH3Li, Et2O,
THF, 0°C to rt (16 h); (b) K2CO3 5%, MeOH, rt (17 h); (c) piperazine, EtOH, 60°C (24 h); (d) 4-
trifluoromethyl-3-methoxybenzaldehyde, AcOH (1%), NaBH3CN, DMF, rt (5 h).
2.4 Biological assessment of D-ring derivatives 2-7
After chemical synthesis of D-ring derivatives 2-7, those resulting from the first strategy,
their potency to inhibit the transformation of E2 into E1 by17β-HSD10, their selectivity of
inhibition (17β-HSD10 vs 17β-HSD3) and their metabolic stability were assessed and compared
to the lead compound 1. The oxidation of E2 to E1 was evaluated using stably transfected HEK-
293[17β-HSD10] cells in culture. For each assay, transfected cells were incubated for 40 h with

the inhibitor (0.3 μM and 3.0 μM) and the substrate E2 (1.0 μM, E2/[14C]-E2 in proportions 9:1).
The steroids E2 and E1 were then extracted from the culture medium with Et2O and separated by
thin layer chromatography (TLC). The radioactivity of the substrate (E2) and the metabolite (E1)
was quantified to calculate the percentage of transformation and then the percentage of inhibition
(Table 2).
Table 2. Biological results for D-ring derivatives 2-7
Inhibition (%)
of E2 to E1 by
Inhibition (%)
of 4-dione to T by
Metabolic stability –
Remaining
Quantity (%)c
1.0 μM
Cpda Derivatization
HEK-293[17β-HSD10] cellsb
LNCaP[17β-HSD3] cellsb
0.3 μM
40.7 ± 4.4
8.5 ± 3.3
20.5 ± 1.3
22.4 ± 8.4
13.5 ± 1.0
5.5 ± 2.0
10.8 ± 2.0
3.0 μM
0.3 μM
30.2 ± 8.9d
5.0 ± 0.6
2.0 ± 0.2
2.0 ± 1.8
5.2 ± 2.6
1.1 ± 0.4
12.4 ± 1.9
3.0 μM
84.4 ± 9.3d
48.7 ± 2.4
20.6 ± 0.7
4.0 ± 1.2
3.8 ± 0.5
-0.5 ± 2.9
51.9 ± 1.9
C17-ketone (C=O)
17β-OH/17α-H
70.4 ± 1.3
42.4 ± 4.4
41.9 ± 8.4
45.8 ± 6.2
36.9 ± 1.8
40.8 ± 1.5
27.9 ± 2.8
28.0 ± 12.0
67.2 ± 5.2
21.8 ± 4.2
45.5 ± 6.8
70.6 ± 3.6
1
2
17β-OH/17α-CH3
17β-OH/17α-CCH
3
4b
17β-H/17α-OH
17-di-F
16,16-di-CH3
5
6
7
34.1 ± 2.1
54.5 ± 6.0
a See schemes for the full chemical structure of 1-7.
b Average results of one experiment performed in triplicate ± SEM.
c Average results of two experiments performed in duplicate ± SEM.
d Average results of five experiments performed in triplicate ± SEM (see table S1 in Supporting Information).
At a concentration of 3.0 μM, the secondary 17β-alcohol 2 inhibited 42.4% of E2
transformation into E1, but the corresponding ketone 1 was more potent with 70.4% of
inhibition. This result confirms that a ketone at C-17 is better than a 17β-hydroxy to inhibit 17β-
HSD10 and is in accord with an observation from our previous screening study [29]. Tertiary
alcohols 3 and 4b were synthesized and tested considering that the addition of a methyl or an
acetylene group at position 17α is a known strategy to increase the metabolic stability of steroids
[
41]. Their inhibitory potencies, 41.9% and 45.8%, respectively for 3 and 4b at a concentration
of 3.0 μM, were similar to that of 2 (42.4%). With 36.9% of inhibition, the 17α-OH derivative 5
was a less potent inhibitor than ketone 1. In fact, the similar inhibitory potencies of 2 and 5
suggest that the stereochemistry of the alcohol at C17 has no impact on the inhibitory potency.
Compound 6 was synthesized since a CF2 group is a potential bioisostere of a ketone
also, taking into account that the C-F bond is more resistant to metabolization than
corresponding C-O bond 43 . This 17-CF2 derivative inhibits E2 into E1 transformation (40.8%)
[42] and,
[
]
but it is less potent than the ketone 1 (70.4%) when tested at 3.0 μM. The last D-ring derivative,
compound 7, is a 17-ketone derivative modified by the addition of two methyl groups at C16 to
avoid the hydroxylation at this position. We also expected that steric hindrance could prevent the
reduction of the ketone. In fact, derivative 7 was the less potent inhibitor, inhibiting only 27.9%

of the E2 to E1 transformation at 3.0 μM. All the above results showed that D-ring derivatives 2-
7 inhibit the transformation of E2 to E1 in HEK-293[17β-HSD10] cells, but they are less potent
inhibitors than lead compound 1.
To test the selectivity of inhibition, derivatives 2-7 were assessed as 17β-HSD3 inhibitors
using LNCaP cells overexpressing 17β-HSD3 (LNCaP[17β-HSD3]) as source of enzyme. After
1 h of incubation, the radiolabeled substrate 4-dione and its metabolite T were extracted with
Et2O and separated by TLC. After quantification of the radioactivity, the percentage of
transformation and then the percentage of inhibition were calculated (Table 2). In this assay, the
lead compound 1 inhibited 17β-HSD3 at both concentrations tested. Derivatives 2 (17β-OH/17α-
H), 3 (17β-OH/17α-CH3) and 7 (17-C=O/16,16-di-CH3) inhibited 17β-HSD3 less than
compound 1 (17-C=O), but still prevent the formation of T and therefore they are not selective
inhibitors of 17β-HSD10. However, derivatives 4b (17β-OH/17α-C≡CH), 5 (17β-H/17α-OH)
and 6 (17,17-di-F) did not inhibit significantly the enzymatic activity of 17β-HSD3 at the two
concentrations tested (0.3 and 3.0 μM). Therefore, these three compounds are selective inhibitors
of 17β-HSD10 at the tested concentrations.
Finally, to test the metabolic stability of derivatives 2-7, the compounds were treated 1 h
with a microsomal preparation of human liver in the presence of NADPH as cofactor. In this
assay, a compound of interest is submitted to metabolism related enzyme reactions, and the
remaining compound is measured at the end of the incubation time, and expressed in %. Thus the
more the compound tested is present after this treatment (higher % of remaining compound), the
more the compound is considered stable (less degraded by the enzymes present in the liver).
Under these artificial conditions, which are more severe than those existing in the liver, the
remaining quantities of derivatives 3 (21.8%) and 6 (34.1%) were not significantly different to
the value of 1 (28.0%), but the other four derivatives were all more metabolically stable (45.5-
70.6%). Compounds 2 and 5 were the most stable, with 67.2% and 70.6% respectively, whereas
4b and 7 were slightly less stable with 45.5% and 54.5% of remaining compound, respectively.
In fact, the tertiary alcohol 4b cannot be oxidized, thus preventing phase-I reaction. Also, the
steric hindrance of 4b (17α-ethynyl) and 7 (two methyls at C-16) seems to stabilize these
androstane derivatives, by making the D-ring less available to phase-I reactions. Compounds 2
and 5, two possible metabolites of 1 after carbonyl reduction are probably more stable than the
17-ketone of 1 in this biological assay using the cofactor NADPH (for the reduction) over NAD+
(for the oxidation). In summary only one reaction (oxidation at C17) can transform 2 and 5

whereas two reactions (reduction at C17 and hydroxylation at C16) can decrease the quantity of
1.
2.5 Biological assessment of compounds 8-11 (libraries A-D members)
Libraries of androsterone (ADT) derivatives resulting from the second strategy (side
chain modification) were synthesized to optimize the inhibitory potency of lead compound 1 for
the transformation of E2 into E1 in HEK-293[17β-HSD10] cells in culture and to potentially
improve the selectivity over 17β-HSD3. In our design, we kept the 3β-piperazinyl-methyl-ADT
nucleus as base core, but greatly expanded the molecular diversity (aryl group) present on the
piperazine nucleus in order to extend our structure-activity relationship (SAR) study. Five
functional groups to link these aryl building blocks (amide, sulfonamide, urea, thiourea and
amine) were also used to extend the molecular diversity. From the 150 compounds synthesized
(libraries A-1 to D-3, 120 were selected for a first round of screening tests at two concentrations
of 0.3 μM and 3.0 μM (Supplementary data; Table S1).
The inhibitions of E2 into E1 transformation provoked by the new compounds were
compared to the lead compound 1, and only the compounds producing a similar or better
inhibition were retained. For the amide library A-1, although compounds A-1,5, A-1,10, A-1,20
and A-1,21 similarly inhibited the transformation of E2 to E1 by 17β-HSD10 as 1, none were
more potent at 0.3 or 3.0 μM. From the second amide library A-2, however, five amide
derivatives were similarly or more potent than 1 at 0.3 μM (namely A-2,5, A-2,10, A-2,12, A-
2,13 and A-2,20), but were all more potent at 3.0 μM. These compounds were thus selected for
the next round of assay (to obtain their IC50 values). From the sulfonamide library B-1, two
compounds B-1,3 and B-1,19 were more potent than 1 at 3.0 μM, but only B-1,3 was a stronger
inhibitor at 0.3 μM, and therefore selected. In library C-1, representing urea and thiourea
derivatives, compounds C-1,16 and C-1,12 were more potent inhibitor than 1 at the two
concentrations tested and thus selected. From the amine library D-1, any compound inhibited E2
into E1 transformation at sufficient level to justify their selection. For the amine library D-2,
compound D-2,12 was more potent at 0.3 μM than 1 and D-2,4 produced a similar inhibitory
potency. Both compounds D-2,4 and D-2,12 were however more potent than 1 at 3.0 μM and
therefore they were retained for additional assays. Finally, the last library D-3 was designed
based on these two compounds. The first derivatives were isomers of D-2,4; the position of the
methyl or the methoxy group being changed. Three out of four of these derivatives had similar

inhibitory potency of 17β-HSD10 as 1, and only one (D-3,4) was more potent at both
concentrations tested. Derivatives with a trifluoromethyl group, which increased the
hydrophobicity
[44], were also synthesized for comparison with the ones with methyl
substituent. Interestingly, the trifluoromethyl analog of D-2,4, compound D-3,7, was the most
potent 17β-HSD10 inhibitor from this library and all the others.
Compounds from libraries A-1 to D-3 were also screened against 17β-HSD3 to determine
their selectivity of inhibition (17β-HSD10 over 17β-HSD3). LNCaP[17β-HSD3] cells were
incubated to determine if the transformation of 4-dione into T was affected by each library
member (Table S1, Supplementary data). From the data expressed as percentages of inhibition,
almost all compounds inhibited 17β-HSD3 activity at 0.3 μM (between 10% and 70%) and more
at 3.0 μM (between 60% and 100%). Thus, a selectivity of inhibition for 17β-HSD10 over 17β-
HSD3 cannot be obtained by modifying the capping group added on the piperazine nucleus. In
fact, the two enzymes probably have a similar binding pocket, mainly hydrophobic, where the
side chain at position C-3β can produce key favorable interactions affecting their inhibitory
potency, but not their selectivity of inhibition. However, as previously exemplified by D-ring
derivatives (Table 1), the area surrounding D-ring seems more promising to obtain a selectivity
of inhibitor action.
Selected library members from the screening study reported above are structurally closely
related to lead compound 1 and they were reported in Table 3. For a better comparison they have
been tested again in HEK-293[17β-HSD10] cells using a range of concentrations to obtain
inhibition curves, which were used to calculate IC50 values. Eight of the twelve derivatives had
lower IC50 values than 1 in their respective assay (Table 3), but the amine derivatives D-2,4 and
D-3,7 were the best inhibitors. Thus, D-2,4 has a 5.4 folds better inhibition than lead compound
1 (IC50 = 0.42 μM and 2.26 ± 0.36 μM, respectively). The 3-methoxy-4-benzyl derivative D-2,4
differs from 1 only by the presence of a methyl group in the para position on the benzyl
substituent. Overall, the 4-trifluoromethyl-3-methoxybenzyl derivative D-3,7 was the most
potent inhibitor of all derivatives with an IC50 value of 0.14 μM (16 fold better than 1).
Interestingly, the addition of a hydrophobic groups on the benzyl part of the androstane
derivative improves the inhibitory potency of some compounds (Table 3). As example, the
difference between 1, D-2,4 and D-3,7 is related to the nature of the groups on the para position
of the phenyl, respectively being a hydrogen, a methyl and a trifluoromethyl group. The ClogP,

one of the criteria of Lipinski rule of 5 [45,46], allows to estimate the lipophilicity of a molecule
and the estimated ClogP values of 1, D-2,4 and D-3,7 are respectively 4.4, 4.7 and 5.2. These
estimated values show a correlation between the hydrophobicity and potency of the three
inhibitors (Supplementary data; Fig. S8A). Thus, we hypothesized that hydrophobic interactions
between the enzyme and the side chain of the steroidal inhibitor enhanced inhibitory potency
against 17β-HSD10. However, even if derivatives A-2,13 and C-1,12 were more hydrophobic
than D-3,7 they were not more potent. In fact, compound A-2,13 possesses a voluminous rigid
side chain, and the phenyl moiety of C-1,12 is distant by a thiourea bond from the piperazine
which is longer than the methylene group compound 1 (Ar-CH2-NH-C=S-piperazine-ADT for
C-1,12 vs Ar-CH2-piperazine-ADT for 1). The rigidity and the length of the side chain seem to
prevent the formation of optimal hydrophobic interactions in that enzyme region, and thus, these
derivatives do not provide a marked inhibition improvement. In fact, the plot for Log P versus
IC50 values for all compounds in Table 3 showed no correlation (Supplementary data; Fig. S8B).
Finally, the presence of a methoxy group, a hydrogen accepting group, on the aryl ring seems to
be helpful to inhibition, potentially by producing a hydrogen bonding with the enzyme.
Therefore, we can hypothesize that 17β-HSD10 possesses a pocket where the 3β-side chain of
the steroid inhibitors is well positioned to form beneficial hydrophobic and hydrogen bond
interactions.
In addition to 17β-HSD10 inhibition, penetration across the blood brain barrier (BBB) is also an
important attribute to consider. Two different prediction scores (CNS MPO and CNS ACD) were
then used to estimate the BBB penetration of our inhibitors (Table 3). The CNS multiparameter
optimization (MPO) score was developed by Wager et al.
[47] while the ACD score was
generated from ACD/Percepta 14.0.0 software 48 , both scores use different physicochemical
[
]
properties such as hydrogen bond donors, topological polar surface area, pKa, ClogP and ClogD.
On a scale from 0-6, it was observed that marketed central nervous system (CNS) drugs
displayed a CNS MPO score of ≤ 2 (1%), between 2 and 4 (25%) and ≥ 4 (74%) [47]. With CNS
MPO scores ranging from 1.9 to 3.8, our best inhibitors fall into the second category. For the
ACD score, which is expressed as a negative value, a molecule is considered to have a good
penetrating BBB power when its score is ≥ -3. Most inhibitors reported in Table 3 have a CNS
ACD score that is higher than -3 (-2.43 to -2.98) or close to -3 (-3.10 to -3.71). It is interesting to
note that the same trend is observed between the two predictive scores. It is also important to
mention that compound 1, with a CNS ACD score of -2.54, demonstrated a very good ability to

cross the BBB when assessed in mice using the in situ cerebral perfusion technique (see Fig. S1
in 30 ). We can therefore conclude that with scores similar to or close to that of 17β-HSD10
[
]
steroid inhibitor 1, the new inhibitors reported in Table 3 should penetrate the BBB.
Table 3. Biological results for a selection of compounds from libraries A-D
Y
Cpd
Family
X
IC50 (µM)a Potency
(E2 to E1)
ClogPc
CNS
CNS
(N-Aryl
Group)
(fold)b
MPOd
ACDe
Amine
Amide
CH2
---
4.43
4.99
3.1
-2.54
1
2.26 ±
0.36f
2.85
C=O
0.79
2.8
-3.53
A-2,5
Amide
Amide
C=O
C=O
2.05
2.30
1.10
0.98
4.66
3.85
3.0
3.4
-2.82
-2.98
A-2,10
A-2,12
Amide
Amide
C=O
C=O
2.60
1.66
1.57
0.87
1.36
1.44
5.30
3.81
2.97
2.8
3.4
3.8
-3.66
-2.55
-2.43
A-2,13
A-2,20
B-1,3
Sulfonamide SO2
Thiourea
Thiourea
Amine
NH-
C=S-
NH-
C=S
CH2
1.41
1.07
0.42
1.60
2.11
5.38
5.54
4.25
4.72
2.5
1.9
3.0
-3.29
-3.10
-2.60
C-1,12
C-1,16
D-2,4
Amine
Amine
Amine
CH2
CH2
CH2
0.97
3.00
0.14
2.33
0.75
16.1
4.58
4.72
5.25
3.0
3.0
2.8
-2.82
-2.60
-2.95
D-2,12
D-3,4
D-3,7
Amine
Amine
CH2
CH2
1.26
0.95
1.8
2.4
5.79
5.54
2.5
2.5
-3.71
-3.20
25
26
a
For the transformation of E2 to E1 by HEK-293[17β-HSD10] cells.
Fold = IC50 (lead compound 1) / IC50 (compound).
b
c Values estimated by ACD/Percepta software [48].
d Values calculated according to Wager et al [47].

e Values estimated by ACD/Percepta [48].
f Average results of four experiments (2.69, 3.03, 1.75 and 1.57 µM) ± SEM.
2.6 Assessment of hybrid compounds 25 and 26
Up to this point, we obtained from the strategies 1 and 2 new steroidal inhibitors of 17β-
HSD10 that were either more metabolically stable than 1, selective for this enzyme over 17β-
HSD3 (from D-ring derivatives 2-7) or better inhibitors than 1 for the E2 into E1 transformation
in HEK-293[17β-HSD10] (from libraries A-D). To obtain an inhibitor more metabolically stable
and more potent than lead compound 1, and with a selective action for 17β-HSD10 over 17β-
HSD3, we synthetize two hybrid compounds (25 and 26) of the potent inhibitor D-3,7 and D-ring
derivatives 4b or 5 in a third optimization strategy. The inhibition activities of these new
derivatives, 25 (17β-OH/17α-C≡CH) and 26 (17β-H/17α-OH) (Scheme 4), were assessed in
HEK-293[17β-HSD10] by measuring their inhibition potency of the E2 into E1 transformation,
in the same conditions as previously described. Complete inhibition curves for 25 and 26 were
obtained (Fig. 3) and their IC50 values were calculated as 1.26 and 0.95 μM, respectively. In fact,
25 and 26 were found 1.8 and 2.4 times more potent inhibitors of 17β-HSD10 than lead
compound 1, but in the same time, the modifications of D-ring of D-3,7 were found to reduce
their inhibitory potency by 7-9 times (Table 3). This loss of activity is however not detrimental
since comparative activities to lead compound 1 were obtained, and most importantly, an
enhanced metabolic stability and selectivity over 17β-HSD3 is expected based on results
obtained with corresponding derivatives 4b and 5.

Fig. 3. Inhibition of E2 into E1 transformation by HEK-293[17
β-HSD10] intact cells. The calculated IC50
are 1.57, 1.26 and 0.95 M for compounds 1, 25 and 26, respectively. Error bars are smaller than the
μ
symbol for some concentration points.
3. Discussion
Optimization of 17β-HSD10 lead inhibitor 1 has been addressed by using three
successive SAR strategies. We first performed six D-ring modifications to improve metabolic
stability and/or selectivity of action over 17β-HSD3 by synthesizing compounds 2-7 using
classic chemistry in solution. Secondly, we prepared 120 3β-piperazinyl-methyl-ADT derivatives
(libraries A-D members represented by general structures 8-11) by parallel solid-phase synthesis
in order to increase the level of inhibition of E2 into E1 transformation by17β-HSD10. Finally,
we combined the best D-ring modifications to the best inhibitors obtained from libraries A-D by
synthesizing compounds 25 and 26. Steroid derivatives generated by classic chemistry
13
(compounds 2-7) were characterized by IR, 1H NMR, C NMR and MS while library members
generated by solid-phase synthesis (amides 8, sulfonamide 9, ureas 10 and amines 11) were
1
characterized by H NMR and MS. Obtained in very small quantities and in amorphous forms
because having a long hydrophobic side chain at position C3, it was not possible to determine the
melting point of these 128 steroid derivatives, but their percentages of purity were however
determined by HPLC and 1H NMR spectra provided (Supplementary data).
Interesting observations emerged from those three optimization strategies. In the first one,
six derivatives aiming D-ring modifications were synthesized and were found to inhibit E2 into
E1 transformation by 17β-HSD10, but they were slightly less potent than 1. However, three of
these compounds, 4b (17β-OH/17α-C≡CH), 5 (17β-H/17α-OH) and 6 (17,17-di-F), did not
inhibit the 17β-HSD3 whereas two, 4a and 5, were metabolically more stable than 1. Therefore,
we had at this stage new steroidal inhibitors of 17β-HSD10, selective over 17β-HSD3 and more
metabolically stable than lead compound 1.
In the second strategy, we were interested to increase the 17β-HSD10 inhibition level of
compound 1 by exploring effect of modifications at the C3 side chain. In that end, the 3β-
piperazinyl-methyl-ADT nucleus as base core was kept intact but the aryl group present on the
piperazine nucleus was modified to extend our SAR study. We thus synthesized new compounds
by solid-phase synthesis, but only twelve compounds from the 120 shown a better 17β-HSD10
inhibition than compound 1 and their IC50 values were determined. Thus, from the four families

of side chain tested (amides, sulfonamides, ureas and amines), the amine derivatives D-2,4 and
D-3,7 showed the highest inhibitions. Indeed, D-2,4 was 5.4 folds better than lead compound 1
(IC50 = 0.42 μM and 2.26 μM, respectively). This molecule differs from 1 only by the addition of
a methyl group at the para position on the benzyl substituent. Even better, when the methyl
group was replaced by corresponding 4-trifluoromethyl group (D-3,7) an additional gain of
inhibition was obtained with an IC50 value of 0.14 μM (16-fold better than 1).
In a final step of optimization (third strategy), we decided to combine the two best D-ring
modifications to the best inhibitor of libraries A-D (D-3,7) leading to hybrid compounds 25
(17β-OH/17α-C≡CH) and 26 (17β-H/17α-OH). Overall, these dual modifications provided more
potent inhibitors than lead inhibitor 1 (IC50 of 1.26, 0.95 and 2.26 μM, respectively), but the
metabolic stability and selectivity of inhibition for 17β-HSD10 over 17β-HSD3 remain to be
addressed for compounds 25 and 26.
4. Conclusion
We synthesized 128 compounds (3α-hydroxy-5α-androstane derivatives with D-ring
modification or/and N-substituted 3β-piperazinylmethyl side chain) and tested their ability to
inhibit the transformation (oxidation) of E2 to E1 by 17β-HSD10, a mitochondrial enzyme
suspected to play a role in AD. Two D-ring modifications (17β-OH/17α-CCH and 17β-H/17α-
OH) made it possible to increase the metabolic stability of lead compound 1 while making these
inhibitors selective for 17β-HSD10 over the 17β-HSD3. The SAR study also made it possible to
obtain trifluorinated compound D-3,7 which is 16 times more active as an inhibitor of 17β-
HSD10 (IC50 = 0.14 µM) than lead compound 1. Finally, the combination of two SAR strategies
generated compounds 25 and 26, two promising candidates for future in vivo studies.
5. Experimental section
5.1 Chemistry
5.1.1 General
Chemical reagents were purchased from Sigma-Aldrich (Saint-Louis, MI, USA), Matrix
Innovation (Québec, QC, Canada), Alfa Aesar (Wood Hill, MA, USA) and AstaTech (Bristol,
PA, USA), Enamine Building Blocks (Cincinnati, OH, USA), Platte Valley Scientifics
(Gothenburg, NE, USA), Aldlab Chemicals (Woburn, MA, USA) and LabNetwork Compounds
(Cambridge, MA, USA). The glycerol polymer bound with a loading of 1.0 mmol/g was supplied

by Sigma-Aldrich. Anhydrous dichloromethane (DCM), diethyl ether (Et2O),
dimethylformamide (DMF), ethanol (EtOH), dimethylsulfoxide (DMSO), and tetrahydrofuran
(THF) were obtained from Sigma-Aldrich. Ethyl acetate (EtOAc), hexanes and methanol
(MeOH) were purchased from Fisher Scientific (Montréal, QC, Canada) and were used as
received. The loading of steroid 22 was performed in peptide synthesis vessels with frit equipped
for vaccum filtration (ChemGlass Inc.; Vineland, NJ, USA). The steps from i to m (Scheme 3)
were performed with an AAPPTec Solutions automated organic synthesizer (Louisville, KY,
USA) using a solid phase reaction block (96 wells). Thin layer chromatography (TLC) was
performed on 0.20 mm silica gel 60 F254 plates (E. Merck; Darmstadt, Germany). Flash column
chromatography were performed with 230-400 mesh ASTM silica gel 60 supplied by SiliCycle,
(Québec, QC, Canada).
Infrared (IR) spectra were recorded with a Horizon MB 3000 ABB FTIR spectrometer
(ABB, Québec, QC, Canada) and only characteristic bands were reported in cm-1. Nuclear
magnetic resonance (NMR) spectra were recorded at 400 MHz for 1H and 100.6 MHz for 13C on
a Bruker Avance 400 digital spectrometer (Billerica, MA, USA). The chemical shifts (δ) are
1
expressed in ppm and referenced to chloroform (7.26 and 77.0 ppm) for H and 13C NMR,
1
respectively. H NMR signals were reported as s (singlet), d (doublet), t (triplet), q (quartet), m
(multiplet) and broad (br). High performance liquid chromatograph (HPLC) purities of final
compounds released from solid support were determined with a Shimadzu apparatus (Kyoto,
Japan) using a Shimadzu SPD-M20A Photodiode array detector, a Alltima HP C18 reverse-
phase column (250 mm x 4.6 mm, 5 μm) and a solvent gradient of MeOH-H2O with 0.1% acetic
acid (AcOH). The wavelength of the UV detector was selected between 190-265 nm. Low
resolution mass spectra (LRMS) were recorded on a Shimadzu apparatus equipped with APCI
(atmospheric pressure chemical ionization). High-resolution mass spectra (HRMS) were
provided by Pierre Audet from the Department of Chemistry at Université Laval (Québec, QC,
Canada).
5.1.2 Chemical synthesis of D-ring derivatives
5.1.2.1 Synthesis of 2
To a solution of compound 1 [33] (0.150 g, 0.30 mmol) in MeOH (10 mL) was added
sodium borohydride (0.056 g, 1.49 mmol) at 0°C. The mixture was stirred 4 h under inert
atmosphere and the resulting solution was evaporated, water added and extracted three times

with EtOAc. The organic layers were washed with water and brine, dried with MgSO4, filtered
and evaporated over reduced pressure. The crude compound was purified by flash column
chromatography using a gradient of hexanes/EtOAc (1:1 to 8:2) to give compound 2.
(3α,5α,17β)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}androstan-17-ol (2):
Amorphous solid (151 mg, 99 %). IR (ATR) υ: 3472 (OH). 1H NMR (CDCl3) δ: 0.72 and 0.73 (2
s, 6H, 19-CH3 and 18-CH3), 0.70-1.79 (m, unassigned CH and CH2), 1.98-2.07 ( br m, 1H), 2.25
(s, 2H, NCH2-COH), 2.46 (br s, 4H, 2 x NCH2), 2.64 (br s, 4H, 2 x NCH2), 3.47 (s, 2H, NCH2-
Ar), 3.62 (t, 1H, J = 8.6Hz, 17α-CH), 3.80 (s, 3H, OCH3), 6.79 (dd, J = 7.8, 1.9 Hz, CH of Ar),
6.87-6.89 (m, 2H, 2 x CH of Ar), 7.24 (t, 1H, J = 8.1 Hz, CH of Ar). 13C NMR (CDCl3) δ: 11.1,
11.2, 20.5, 23.3, 28.5, 30.5, 31.5, 32.7, 33.9, 35.5, 35.8, 36.7, 39.7, 40.8, 42.9, 51.0, 53.4 (2C),
54.2, 55.2, 55.7 (2C), 62.9, 68.9, 70.2, 81.9, 112.4, 114.6, 121.5, 129.1, 139.6, 159.5. HRMS for
C32H51N2O3 [M+H]+: calculated 511.3894, found 511.3897. HPLC purity: 98.3% (RT = 11.8
min, MeOH/H2O/formic acid (55:45:0.1), Luna C18 column).
5.1.2.2 Synthesis of 3
To a solution of compound 1 (0.100 g, 0.20 mmol) in dry toluene (3 mL) was added
methyl magnesium iodide, 3.0 M in diethyl ether (0.66 mL, 1.97 mmol) under inert atmosphere.
The mixture was stirred 4 h at reflux, and then at room temperature (rt) overnight. The resulting
solution was cooled at 0°C, and an aqueous solution of NH4Cl (2.6 M) was added dropwise. The
organic solvents were evaporated and the aqueous phase was extracted eight times with DCM.
The organic layers were combined and washed with water, dried with MgSO4, filtered and
evaporated over reduced pressure. The crude compound was purified by flash column
chromatography using DCM/MeOH (96:4) to give compound 3.
(3
α,5
α,17
α,17
β)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}androstan-17-
1
methyl-17-ol (3): Amorphous white solid (102 mg, 98%). IR (ATR) υ: 3425 (OH). H NMR
(CDCl3) δ: 0.74 (s, 3H, 19-CH3), 0.84 (s, 3H, 18-CH3), 1.20 (s, 3H, 17α-CH3) 0.70-1.81 (m,
unassigned CH and CH2), 2.25 (s, 2H, N-CH2-COH), 2.46 (br s, 4H, 2 x NCH2), 2.65 (br s, 4H, 2
x NCH2), 3.47 (s, 2H, NCH2-Ar), 3.80 (s, 3H, OCH3), 6.79 (dd, J = 7.4, 1.9 Hz, 1H, CH of Ar),
6.87-6.90 (m, 2H, 2 x CH of Ar), 7.21 (t, J = 8.0 Hz, 1H, CH of Ar). 13C NMR (CDCl3) δ: 11.2,
14.0, 20.6, 23.2, 25.8, 28.6, 31.7 (2C), 32.7, 33.9, 35.8, 36.4, 39.0, 39.7, 40.8, 45.5, 50.7, 53.4,
54.1, 55.2, 55.7, 62.9, 68.9, 70.2, 81.7, 112.4, 114.6, 121.5, 129.1, 139.7, 159.6. HRMS for
C33H53N2O3 [M+H]+: calculated 525.4051, found 525.4060. HPLC purity: 95.9% (RT = 11.3
min, MeOH/H2O/formic acid (55:45:0.1), Luna C18 column).

5.1.2.3 Synthesis of 4a and 4b
5.1.2.3.1 Synthesis of 4a
To a solution of trimethylsilylacetylene (0.438 g, 4.46 mmol) in anhydrous diethyl ether
(25 mL) was added dropwise MeLi (1.6 M in THF, 2.23 mL, 3.57 mmol) at 0°C. The mixture
was stirred at rt for 1 h and cooled again at 0°C. The compound 1 (0.150 g, 0.30 mmol) was
added in anhydrous THF (25 mL) and the solution was stirred at rt for 15 h. The reaction mixture
was stopped by addition of ice/water and the crude compound was extracted with EtOAc,
washed with brine, dried with MgSO4, filtered and evaporated. The mixture was purified by flash
column chromatography using a gradient of hexanes/EtOAc (6:4 to 5:5) to give 4a.
(3α,5α,17α,17β)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}androstan-17-
(trimethylsilyl)ethynyl-17-ol (4a): Amorphous solid (144 mg, 80%). IR (ATR) υ: 3425 (OH),
1
841 (SiMe3). H NMR (CDCl3) δ: 0.17 (s, 9H, Si(CH3)3), 0.74 (s, 3H, 19-CH3), 0.81 (s, 3H, 18-
CH3), 0.70-1.70 (m, unassigned CH and CH2), 1.88-1.96 (m, 1H, 16α-CH2), 2.19-2.24 (m, 1H),
2.26 (s, 2H, NCH2-COH), 2.47 (br s, 4H, 2 x NCH2), 2.66 (br s, 4H, 2 x NCH2), 3.48 (s, 2H,
NCH2-Ar), 3.80 (s, 3H, OCH3), 6.79 (dd, J = 7.3, 2.0 Hz, 1H, CH of Ar), 6.87-6.90 (m, 2H, 2 x
13
CH of Ar), 7.24 (t, J = 8.0 Hz, 1H, CH of Ar). C NMR (CDCl3) δ: 0.05 (3C), 11.2, 12.9, 20.6,
23.2, 28.5, 29.7, 31.5, 32.8 (2C), 34.0, 35.7, 36.2, 38.9, 39.8, 40.8, 47.0, 50.5, 53.4, 53.8, 55.2,
55.7 (2C), 62.9, 68.9, 70.2, 80.1, 89.9, 109.5, 112.4, 114.7, 121.5, 129.1, 139.6, 159.6. LRMS for
C37H59N2O3Si [M+H+]: 608.7.
5.1.2.3.2 Synthesis of 4b
To a solution of K2CO3 (5% in MeOH, 10 mL) was added 4a (0.120 g, 0.20 mmol) and
the mixture was stirred at room temperature overnight. The reaction was quenched with water,
extracted three times with DCM, washed with brine and filtered and evaporated under reduce
pressure. The crude solid was purified by flash column chromatography using a gradient of
hexanes/EtOAc (4:6 to 2:8) to give 4b.
(3α,5α,17α,17β)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}androstan-17-
ethynyl-17-ol (4b): Amorphous white solid (92 mg, 87%). IR (ATR) υ: 3502 and 3387 (OH),
1
3263 (C
≡
C-H). H NMR (CDCl3) δ: 0.74 (s, 3H, 19-CH3), 0.83 (s, 3H, 18-CH3), 0.75-1.71 (m,
unassigned CH and CH2), 1.91-1.99 (m, 1H, 16α-CH2), 2.25 (s, 2H, NCH2-COH), 2.22-2.30 (m,
1H), 2.46 (br s, 4H, 2 x NCH2), 2.57 (s, 1H, C CH), 2.64 (br s, 4H, 2 x NCH2), 3.47 (s, 2H,
≡
NCH2-Ar), 3.80 (s, 3H, OCH3), 6.79 (dd, J = 7.7, 1.9 Hz, 1H, CH of Ar), 6.88 (m, 2H, 2 x CH of
13
Ar), 6.87-6.90 (m, 2H, 2 x CH of Ar), 7.22 (t, J = 8.0 Hz, 1H, CH of Ar). C NMR (CDCl3) δ:

11.2, 12.8, 20.6, 23.1, 28.5, 31.5, 32.7 (2C), 33.9, 35.8, 36.1, 38.9, 39.7, 40.7, 46.9, 50.4, 53.4
(2C), 53.7, 55.2, 55.7 (2C), 62.9, 68.9, 70.2, 73.8, 79.9, 87.6, 112.4, 114.7, 121.5, 129.1, 139.7,
159.6. HRMS for C34H51N2O3 [M+H]+: calculated 535.3894, found 535.3898. HPLC purity:
99.5% (RT = 11.3 min, MeOH/H2O/formic acid (55:45:0.1), Luna C18 column).
5.1.2.4 Synthesis of 5
5.1.2.4.1 Synthesis of 15b
To a solution of 4-nitrobenzoic acid (0.920 g, 5.51 mmol) in toluene (20 mL) was added,
under argon atmosphere and over ice bath, triphenylphosphine (1.370 g, 5.23 mmol) and diethyl
azodicarboxylate (DEAD) (0.7 mL, 4.46 mmol). The mixture was stirred for 30 min, and 5α-
dihydrotestosterone (15a) (0.400 g, 1.38 mmol) was added. The resulting solution was stirred at
80
chromatography using a gradient of hexanes/EtOAc (95:5 to 85:15) to give 15b.
-androstan-(17 -(4-nitrobenzoate))-3-one (15b): Amorphous white solid (325 mg, 53 %).
°C for 18 h. The solvent was evaporated and the crude compound purified by flash column
5
α
α
IR (ATR) υ: 1713 (C=O). 1H NMR (CDCl3) δ: 0.85 (s, 3H, 19-CH3), 1.03 (s, 3H, 18-CH3), 0.78-
2.43 (m, unassigned CH and CH2), 5.08 (d, J = 6.1 Hz, 1H, 17β-CH), 8.19 (d, J = 8.8 Hz, 2H, 2 x
13
CH of Ar), 8.30 (d, J = 8.8 Hz, 2H, 2 x CH of Ar). C NMR (CDCl3) δ: 11.5, 16.7, 20.8, 24.8,
28.8, 30.1, 32.0, 32.1, 35.6, 35.8, 38.1, 38.5, 44.6, 45.2, 46.6, 50.4, 53.5, 83.7, 123.5 (2C), 130.4,
130.6 (2C), 136.1, 150.4, 164.2, 211.8. LRMS for C26H34NO5 [M+H]+: 440.25.
5.1.2.4.2 Synthesis of 15c
Compound 15b (0.315 g, 0.72 mmol) was solubilized in a solution of 10% KOH in
MeOH (30 mL). The solution was stirred at rt for 6 h. The solvent was evaporated and water was
added. The mixture was extracted with EtOAc (three times), washed with brine, dried over
MgSO4, filtered and evaporated under reduced pressure. The crude solid was purified by flash
column chromatography using hexanes/EtOAc (7:3) to give 15c.
5
α-androstan-17
α-ol-3-one (15c): Amorphous white solid (146 mg, 70%). IR (ATR) υ: 3441
1
(OH), 1697 (C=O). H NMR (CDCl3) δ: 0.68 (s, 3H, 19-CH3), 1.02 (s, 3H, 18-CH3), 0.73-2.43
(m, unassigned CH and CH2), 3.74 (d, J = 5.0 Hz, 1H, 17β-CH). 13C NMR (CDCl3) δ: 11.5, 17.0,
20.9, 24.6, 28.9, 31.4, 32.0, 32.3, 35.6, 35.7, 38.1, 38.6, 44.6, 45.3, 46.6, 48.5, 53.6, 79.9, 212.1.
LRMS for C19H31O2 [M+H]+: 291.25.
5.1.2.4.3 Synthesis of 16
To a solution of trimethylsulfoxonium iodide (0.212 g, 0.96 mmol) in dry DMSO (7 mL)
over an ice bath and under argon atmosphere was slowly added sodium hydride 60% in oil

(0.038 mg, 0.95 mmol). The solution was stirred at rt for 1 h before adding compound 15c (0.135
g, 0.46 mmol) dissolved in dry THF (4 mL). The reaction mixture was stirred overnight, then
poured in water (500 mL), and extracted with EtOAc (three times). The combined organic layers
were washed with water, brine, dried over MgSO4, filtered and evaporated under reduced
pressure. The crude compound was purified by flash column chromatography using
hexanes/EtOAc (8:2) to give 16.
17
α
-Hydroxy-spiro-3(
R
)-oxirane-5α-androstane (16): Amorphous white solid (124 mg, 87%).
1
IR (ATR) υ: 3455 (OH). H NMR (CDCl3) δ: 0.66 (s, 3H, 19-CH3), 0.85 (s, 3H, 18-CH3), 0.78-
1.78 (m, unassigned CH and CH2), 1.87 (t, J = 13.6 Hz, 1H), 2.04 (td, J = 14.1, 4.7 Hz, 1H),
2.10-2.18 (m, 1H), 2.61 (s, 2H, CH2(O)C), 3.72 (d, J = 5.9 Hz, 1H, 17β-CH). 13C NMR (CDCl3)
δ: 11.3, 17.0, 20.5, 24.6, 28.5, 29.2, 31.4, 32.1, 32.3, 35.5, 35.7, 35.9, 35.8, 43.6, 45.3, 48.7, 53.5,
53.8, 58.6, 80.0. LRMS for C20H33O2 [M+H]+: 291.25.
5.1.2.4.4 Synthesis of building block 18
To a solution of piperazine (1.65 g, 19.20 mmol) in DCM (50 mL) was added 3-
methoxybenzylchloride (17) (0.300 g, 1.92 mmol) at 0°C. The solution was next stirred at rt for
19 h. After completion, water was added, and the solution was extracted with DCM (four times),
washed with water, brine, dried over MgSO4, filtered and evaporated under reduced pressure.
The mixture was purified by flash column chromatography using DCM/MeOH/Et3N (94:5:1) to
give the building block 18.
1-(3-methoxybenzyl)-piperazine (18): Gummy brownish solid (398 mg, 99%). IR (ATR) υ:
3225 (NH). 1H NMR (CDCl3) δ: 2.42 (br s, 4 H, 2 x NCH2), 2.88-2.90 (br m, 5H, 2 x NCH2 and
NH), 3.45 (s, 2H, NCH2-Ar), 3.78 (s, 3H, OCH3), 6.78 (dd, J = 7.1, 1.5 Hz, 1H, CH of Ar), 6.88-
6.91 (m, 2H, 2 x CH of Ar), 7.22 (t, J = 8.0 Hz, 1H, CH of Ar). 13C NMR (CDCl3) δ: 45.8, 54.0
(2C), 55.2, 63.5 (2C), 112.4, 114.6, 121.5, 129.1, 139.7, 159.6. LRMS for C12H19N2O [M+H]+:
207.2.
5.1.2.4.5 Synthesis of 5
To a solution of compound 16 (0.040 g, 0.13 mmol) in EtOH (10 mL) was added the
piperazine derivative 18 (0.136 g, 0.66 mmol). The resulting mixture was stirred at 60°C for 24 h
and the solvent evaporated. The crude compound was purified by flash column chromatography
using DCM:MeOH (97:3) to give 5.
(3
α,5α,17α)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}androstan-17-ol (5):
Amorphous brownish solid (59 mg, 85%). IR (ATR) υ: 3448 (OH). 1H NMR (CDCl3) δ: 0.64 (s,

3H, 19-CH3), 0.73 (s, 3H, 18-CH3), 0.77-1.80 (m, unassigned CH and CH2), 2.04-2.17 (m, 1H),
2.25 (s, 2H, NCH2-COH), 2.46 (br s, 4H, 2 x NCH2), 2.64 (br s, 4H, 2 x NCH2), 3.47 (s, 2H,
NCH2-Ar), 3.71 (d, J = 5.9 Hz, 1H, 17α-CH), 3.80 (s, 3H, OCH3), 6.79 (dd, J = 7.4, 1.9 Hz, 1H,
CH of Ar), 6.87-6.90 (m, 2H, 2 x CH of Ar), 7.22 (t, J = 8.1 Hz, 1H, CH of Ar). 13C NMR
(CDCl3) δ: 11.2, 17.1, 20.4, 24.6, 28.7, 31.5, 32.3, 32.8 (2C), 34.0, 35.7, 35.8, 39.8, 40.7, 45.3,
48.7, 53.4 (2C), 53.9, 55.2, 55.7 (2C), 62.9, 69.0, 70.2, 80.0, 112.4, 114.6, 121.5, 129.1, 139.7,
159.6. HRMS for C32H51N2O3 [M+H]+: calculated 511.3894, found 511.3900. HPLC purity:
98.5% (RT = 15.3 min, MeOH/H2O/formic acid (55:45:0.1), Luna C18 column).
5.1.2.5 Synthesis of 6
To a solution of compound 19
0.05 mmol) in EtOH (3 mL) was added the piperazine derivative 18 (0.054 g, 0.26 mmol) and
the resulting mixture was stirred at 60 C for 23 h. The solvent was evaporated and the crude
compound was purified by flash column chromatography using DCM:MeOH (98:2) to give 6.
(3 ,5 ,17 ,17 )-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}-17-
[39; compound D in Supporting Information] (0.017 g,
°
α
α
α
β
difluoroandrostane (6): Amorphous brownish solid (18 mg, 65%). IR (ATR) υ: 3490 (OH),
1
1311 (C-F). H NMR (CDCl3) δ: 0.74 (s, 3H, 19-CH3), 0.85 (s, 3H, 18-CH3), 0.75-1.72 (m,
unassigned CH and CH2), 1.97-2.19 (m, 2H), 2.26 (s, 2H, NCH2-COH), 2.46 (br s, 4H, 2 x
NCH2), 2.65 (br s, 4H, 2 x NCH2), 3.47 (s, 2H, NCH2-Ar, 2H), 3.80 (s, 3H, OCH3), 6.79 (dd,
J = 7.5, 1.9 Hz, 1H, CH of Ar), 6.87-6.90 (m, 2H, 2 x CH of Ar), 7.22 (t, J = 8.0 Hz, 1H, CH of
Ar). 13C NMR (CDCl3) δ: 11.2, 13.4 (t, JCCCF = 4.5 Hz), 19.9, 22.2 (22.3), 28.4, 28.8 (28.9), 31.0,
32.7, 32.9 (t, JCCF = 25.2 Hz), 33.9, 35.4, 35.8, 39.7, 40.7, 45.3 (t, JCCF = 20.1 Hz), 49.3 (49.4),
53.4, 53.7 (2C), 55.2, 55.7 (2C), 62.9, 68.9, 70.1, 112.4, 114.7, 121.5, 129.1, 132.7 (dd, JCF
=
259.1, 250.8 Hz, 17-CF2), 139.7, 159.6. HRMS for C32H49F2N2O2 [M+H]+: calculated 531.3757,
found 531.3763. HPLC purity: 94.9% (RT = 18.9 min, MeOH/H2O/formic acid (40:60:0.1),
Luna C18 column).
5.1.2.6 Synthesis of 7
To a solution of compound 20
[
39
]
(0.007 g, 0.02 mmol) in EtOH (10 mL) was added
C for
piperazine derivative 18 (0.022 g, 0.11 mmol) and the resulting mixture was stirred at 60
°
23 h. The solvent was evaporated and the crude compound was purified by flash column
chromatography using a gradient of DCM:MeOH (98:2 to 96:4) and to give 7.

(3α,5α)-3-Hydroxy-3-{[4-(3-methoxybenzyl)piperazin-1-yl]methyl}-16,16-dimethyl-
androstan-17-one (7): Amorphous brownish solid (9 mg, 81%). IR (ATR) υ: 3450 (OH), 1728
1
(C=O). H NMR (CDCl3) δ: 0.75 (s, 3H, 19-CH3), 0.87 (s, 3H, 18-CH3), 1.02 (s, 3H, 16-CH3),
1.16 (s, 3H, 16-CH3), 0.80-1.82 (m, unassigned CH and CH2), 2.26 (s, 2H, NCH2-COH), 2.46 (br
s, 4H, 2 x NCH2), 2.65 (br s, 4H, 2 x NCH2), 3.47 (s, 2H, NCH2-Ar), 3.80 (s, 3H, OCH3), 6.79
(dd, J = 7.4, 1.9 Hz, 1H, CH of Ar), 6.87-6.90 (m, 2H, 2 x CH of Ar), 7.22 (t, J = 8.0 Hz, 1H,
13
CH of Ar). C NMR (CDCl3) δ: 11.2, 14.4, 20.1, 25.9, 27.3, 28.4, 31.0, 32.3, 32.7, 33.8, 34.6,
35.9, 37.9, 39.7, 40.8, 45.1, 48.1, 48.9, 53.4 (2C), 54.4, 55.2, 55.8 (2C), 62.9, 68.9, 70.1, 112.4,
114.7, 121.5, 129.1, 139.7, 159.6, 225.7. HRMS for C34H53N2O3 [M+H]+: calculated
537.537.4051, found 537.4055. HPLC purity: 93.7% (RT = 15.7 min, MeOH/H2O/formic acid
(70:30:0.1), Luna C18 column).
5.1.3 Solid phase synthesis of androstane derivatives
5.1.3.1 Synthesis of 22 (Fmoc protection of piperazine and deprotection of 17-ketone)
To a solution of oxirane 21 [40] (7.5 g, 21 mmol) in EtOH (500 mL) was added
piperazine (7.4 g, 85 mmol) and the resulting mixture was stirred at 60°C for 6 h. The solvent
was evaporated and water was added. The mixture was extracted with EtOAc and the organic
phase washed with brine, dried over Na2SO4, filtered and evaporated under reduced pressure to
give 8.7 g of the piperazino derivative. To a solution of this crude compound in THF (650 mL)
was added NaHCO3 (1.0 M, 350 mL) and Fmoc-O-succinimide (8.39 g, 24.9 mmol). The
resulting solution was stirred at rt for 16 h and then quenched with water and extracted with
EtOAc. The organic phase was washed with brine, dried with Na2SO4 and evaporated to give
12.3 g of crude compound. To a solution of this crude compound in DCM was added HCl 1.0 M
in MeOH (25:75). The reaction was stirred at rt for 16 h. The crude solution was washed with
saturated aqueous NaHCO3, filtered on a phase separator (Biotage, Uppsala, Sweden) and
washed with DCM. The crude solid was purified by flash chromatography using a gradient of
hexanes/EtOAc (7:3 to 1:1) to give 22.
(3α,5α)-3-Hydroxy-3-{[4-((9H-fluoren-9-yl)methoxy)carbonyl)piperazin-1-yl]methyl}-
16,16-dimethyl-androstan-17-one (22): Amorphous white solid (9.0 g, 72%). IR (ATR) υ: 3455
(OH), 1736 (C=O of Fmoc), 1705 (17-C=O). 1H NMR (CDCl3) δ: 0.77 (s, 3H, 19-CH3), 0.86 (s,
3H, 18-CH3), 0.80-1.97 (m, unassigned CH and CH2), 2.02-2.12 (m, 1H, 16α-CH), 2.29 (s, 2H,
NCH2-COH), 2.43 (dd, J = 19.2, 8.7 Hz, 1H, 16β-CH), 2.56 (br s, 4H, 2 x NCH2), 3.47 (br s,
4H, 2 x NCH2), 4.24 (t, J = 6.8 Hz, 1H, CHCH2O), 4.44 (d, J = 6.8 Hz, 2H, CHCH2O), 7.37 (t, J

= 7.4 Hz, 2H, 2 x CH of Ar), 7.40 (t, J = 7.5 Hz, 2H, 2 x CH of Ar), 7.56 (d, J = 7.4 Hz, 2H, 2 x
13
CH of Ar), 7.76 (d, J = 7.5 Hz, 2H, 2 x CH of Ar). C NMR (CDCl3) δ: 11.2, 13.8, 20.2, 21.8,
28.4, 30.8, 31.6, 32.5, 33.8, 35.1, 35.8, 35.9, 39.4, 40.7, 44.1, 47.4, 47.8, 51.5, 54.2, 55.4, 67.2,
69.2, 70.5, 120.0 (2C), 124.9 (2C), 127.0 (2C), 127.7 (2C), 141.3 (2C), 144.0 (2C), 155.1, 221.3.
5.1.3.2 Coupling of steroid (synthesis of resin 23a)
To a suspension of glycerol polymer bound resin (10.0 g; loading of 1.0 mmol/g) in dry
toluene was added ketone 22 in dry toluene (9.0 g, 14.7 mmol), trimethyl orthoformate (16.1 mL,
147.2 mmol) and p-TSA (3.8 g, 20.2 mmol). The solution was stirred at rt for 17 h, washed three
times successively by DCM, MeOH and DCM. The resin was dried overnight under vacuum to
give 14.8 g of 23a.
5.1.3.3 Fmoc deprotection of piperazine (synthesis of resin 23b)
To a suspension of resin 23a (14.3 g) in DMF and in a peptide flask was added a solution
of piperidine in DMF (20%, 250 mL) and the resulting mixture was stirred at rt for 3.5 h. The
resin was filtered, washed five times successively with DCM, MeOH and DMF, and dried
overnight under vacuum to afford 11.3 g of resin 23b (loading of 0.41 mmol/g by weight
difference).
5.1.3.4 Synthesis of amide derivatives 8 (Libraries A1 and A2)
5.1.3.4.1 Amide bound formation
Portions of resin 23b (50 mg; loading 0.41 mmol/g) were placed in 4 mL reactor wells of
an automated synthesizer reaction block (96 well format, AAPPTec). A solution of HBTU (0.5
M, 18.9 g in 100 mL DMF) was added (3 mL) to the well and the suspension was shaken for 2
min. Then, a solution of DIPEA (1.0 M, 17.4 mL in 82.6 mL DMF) was added (1 mL) to each
well followed by the addition of a solution of the appropriate carboxylic acid (0.5 M), solubilized
in DMF (2 mL). The suspension was vortexed at 600 rpm for 3 h. The well was then filtered
using the vacuum system and resin washed with DCM (3 mL), then EtOH (3 mL) three times.
5.1.3.4.2 Cleavage of amide derivatives
To the coupled amide resin in 4 mL reactor wells was added a solution of acetyl chloride-
MeOH-DCM (1:9:30) (2.5 mL) and the resulting suspension was vortexed at 600 rpm for a total
of 20 h. The resin was then filtered, washed with MeOH:DCM (1:1) (1.5 mL) and the filtrate
neutralized with a saturated aqueous solution of NaHCO3 (2 mL). The biphasic solutions were
filtered with a phase separator syringe (Biotage) and each organic solution was evaporated and
dried under reduced pressure. The amide derivatives 8 (two libraries, A-1 and A-2, generated in
1
parallel) were weighted and characterized by H NMR (Table 3). Only compounds with