Structure-Based Design of Inhibitors with Improved Selectivity for Steroidogenic Cytochrome P450 17A1 over Cytochrome P450 21A2

Article abstract of DOI:10.1021/acs.jmedchem.8b00419

Inhibition of androgen biosynthesis is clinically effective for treating androgen-responsive prostate cancer. Abiraterone is a clinical first-in-class inhibitor of cytochrome P450 17A1 (CYP17A1) required for androgen biosynthesis. However, abiraterone also causes hypertension, hypokalemia, and edema, likely due in part to off-target inhibition of another steroidogenic cytochrome P450, CYP21A2. Abiraterone analogs were designed based on structural evidence that B-ring substituents may favorably interact with polar residues in binding CYP17A1 and sterically clash with residues in the CYP21A2 active site. The best analogs increased selectivity of CYP17A1 inhibition up to 84-fold compared with 6.6-fold for abiraterone. Cocrystallization with CYP17A1 validated the intended new contacts with CYP17A1 active site residues. Docking these analogs into CYP21A2 identified steric clashes that likely underlie decreased binding and CYP21A2 inhibition. Overall, these analogs may offer a clinical advantage in the form of reduced side effects.

Full text of DOI:10.1021/acs.jmedchem.8b00419

Article
pubs.acs.org/jmc
Cite This: J. Med. Chem. XXXX, XXX, XXX−XXX
Structure-Based Design of Inhibitors with Improved Selectivity for
Steroidogenic Cytochrome P450 17A1 over Cytochrome P450 21A2
Charlie Fehl,† Caleb D. Vogt,‡ Rahul Yadav,§ Kelin Li,‡ Emily E. Scott,*,§,∥ and Jeffrey Aube*,‡
́
†Department of Medicinal Chemistry, University of Kansas, Lawrence, Kansas 66047, United States
‡Division of Chemical Biology and Medicinal Chemistry, UNC Eshelman School of Pharmacy, University of North Carolina at
Chapel Hill, Chapel Hill, North Carolina 27599, United States
§Department of Medicinal Chemistry, University of Michigan, Ann Arbor, Michigan 48109, United States
∥Department of Pharmacology, University of Michigan, Ann Arbor, Michigan 48109, United States
S
* Supporting Information
ABSTRACT: Inhibition of androgen biosynthesis is clinically
effective for treating androgen-responsive prostate cancer.
Abiraterone is a clinical first-in-class inhibitor of cytochrome
P450 17A1 (CYP17A1) required for androgen biosynthesis.
However, abiraterone also causes hypertension, hypokalemia,
and edema, likely due in part to off-target inhibition of another
steroidogenic cytochrome P450, CYP21A2. Abiraterone
analogs were designed based on structural evidence that B-
ring substituents may favorably interact with polar residues in
binding CYP17A1 and sterically clash with residues in the
CYP21A2 active site. The best analogs increased selectivity of
CYP17A1 inhibition up to 84-fold compared with 6.6-fold for
abiraterone. Cocrystallization with CYP17A1 validated the intended new contacts with CYP17A1 active site residues. Docking
these analogs into CYP21A2 identified steric clashes that likely underlie decreased binding and CYP21A2 inhibition. Overall,
these analogs may offer a clinical advantage in the form of reduced side effects.
early in clinical trials, including hypertension, hypokalemia, and
edema.11 Notably, many prostate cancer patients are older men
for whom hypertension is frequently a preexisting concern.
These side effects likely arise from abiraterone inhibiting not
only just the CYP17A1 reactions but also inhibiting
cytochrome P450 21A2 (CYP21A2) at clinically relevant
concentrations.12 Like CYP17A1, CYP21A2 hydroxylates
progestogens such as progesterone and 17α-hydroxyprogester-
one, but at C21 instead of C17 (Figure 2B). The resulting
CYP21A2 products are intermediates in the production of the
mineralocorticoid aldosterone and the glucocorticoid cortisol,
thus impacting blood pressure control and potentially stress
and immune responses. Crossover inhibition of CYP21A2 is
not particularly surprising given that the two enzymes have the
same fold, are 64% similar, bind some of the same substrates
and/or products, and have a heme group for abiraterone
pyridine coordination. Reducing abiraterone inhibition of
CYP21A2 would likely reduce side effects without resorting
to concomitant administration of the glucocorticoids predni-
solone or dexamethasone, as is current clinical practice.13
Herein, structures of CYP17A1 and CYP21A2 were used to
direct the design of abiraterone analogs with the goal of
INTRODUCTION
■
In U.S. males, the cancer with the second highest death rate
remains prostate cancer.1 Prostate cancer proliferation is largely
dependent on androgen signaling. Over 80% of tumors respond
at least initially to surgical or chemical androgen deprivation
therapy.2 However, progression to the most aggressive form of
this disease, castration-resistant prostate cancer, is frequently
fatal despite best current therapeutic efforts. A recent addition
to the drug arsenal for treatment of prostate cancer is
abiraterone acetate, which extends overall survival for prostate
cancer patients who had previously failed other treatment
regimens.3−5 Abiraterone acetate is a prodrug with abiraterone
itself (Figure 1) being a tight-binding inhibitor of cytochrome
P450 17A1 (CYP17A1). As CYP17A1 is the only human
enzyme capable of converting C21 progestogens to the C19
androgens that frequently drive prostate cancer proliferation,
this enzyme was an attractive target for systemic ablation of
androgen biosynthesis (Figure 2A). Additionally, abiraterone is
also currently being evaluated for hormone-responsive breast
cancer, as estrogen synthesis is downstream of androgen
synthesis. Other experimental therapeutics being developed in
this class include the abiraterone analog galeterone6,7 and
nonsteroidal orteronel,8 VT-464,9 and BMS-351.10
Targeting CYP17A1 inhibition in this manner is efficacious
Received: March 14, 2018
clinically, but several serious side effects also became apparent
© XXXX American Chemical Society
A
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 3. Comparison of human CYP17A1 (yellow) and CYP21A2
(cyan), both bound to progesterone (sticks). Orientation of the
steroidal core is directed by (1) hydrogen bonding between the C3
oxygen and N202 in helix F for CYP17A1 vs R234 in helix G of
CYP21A2, (2) a shift of the B′ helix outward and away from the active
site in CYP21A2 compared to CYP17A1, and (3) substitution of I
helix G301 in CYP17A1 for I291 in CYP21A2. Heme in black sticks
with iron as central red sphere. Amino acids are labeled with the
CYP17A1 residue given first, followed by the name of the CYP21A2
residue.
Figure 1. Structures of the marketed CYP17A1 inhibitor abiraterone
acetate and related experimental therapeutics.
2A).14 The resulting 17α-hydroxypregnenolone is then very
subtly repositioned14 for a second round of catalysis in the
same active site to cleave the C17−C20 bond, resulting in the
first androgen in the pathway, dehydroepiandrosterone (Figure
2A).15 Both reactions require oxygen binding to the heme iron
while adjacent to the substrate. The steroidal core of
abiraterone (1b) is positioned similarly to these substrates,
but contains a pyridine substituent projecting from C17 such
that its heterocyclic nitrogen forms a coordinate-covalent bond
with the heme iron, thereby occupying the position where
oxygen would need to bind for catalysis to occur.16
X-ray structures of human CYP17A1 with both abiraterone16
and progesterone14 reveal a similar binding mode for their
steroidal cores. While a structure of human CYP21A2 has not
been reported in complex with abiraterone, there is a structure
of the CYP21A2/progesterone complex, and it is likely that the
steroidal core of abiraterone would bind similarly to
progesterone.12 Comparison of these two steroidogenic
enzymes each bound to progesterone align with a Cα root-
mean-square deviation of only 1.9 Å, but reveal some
informative differences to guide analog development. Despite
conservation of the three polar residues lining the active site
(N202/Q198, R239/R234, and D298/D288), progesterone is
oriented differently in the two active sites, corresponding to the
observed sites of metabolism. In CYP17A1, progesterone is
oriented with C17 directed toward the heme, consistent with
oxidation at this site. As a result, the steroidal core is oriented
such that the C3 substituent on the opposite end forms a
hydrogen bond with N202 in the F helix composing part of the
active site roof (Figure 3, yellow). In CYP21A2, the long axis of
the steroid core must adopt a different angle over the heme to
position C21 for hydroxylation. Though N202 in CYP17A1 is
conservatively substituted with Q198, the corresponding Q198
Figure 2. Key biosynthetic pathways involving (A) CYP17A1 and (B)
CYP21A2 in human steroidogenesis. Progesterone is also a substrate
for CYP17A1 17α-hydroxylation (not shown).
retaining effective CYP17A1 inhibition but decreasing
CYP21A2 inhibition. Evaluation of these abiraterone analogs
in steady-state inhibition studies with both enzymes revealed
modifications increasing selectivity by more than 80-fold. By
sparing CYP21A2 activity, these inhibitors could potentially
allow more natural biosynthesis of mineralocorticoids and
glucocorticoids to avoid severe hypertension, hyperkalemia, and
edema, thereby reducing cardiovascular complications without
compromising effective CYP17A1 inhibition in prostate cancer
patients.
RESULTS AND DISCUSSION
■
Structure-Based Rationale for Inhibitor Design.
Abiraterone is a structural analog of the CYP17A1 substrate
pregnenolone (Figures 1 and 2A). CYP17A1 normally binds
either pregnenolone or progesterone with C17 oriented toward
the heme iron (Figure 3, yellow) for 17-hydroxylation (Figure
B
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
side chain in CYP21A2 is directed away from the active site,
and in CYP21A2, the C3 substituent binds instead to R234 in
the G helix, another major helix composing part of the active
site roof (Figure 3, cyan). In addition, the B′ helix composing
an adjacent wall of the active site is moved outward in
CYP21A2 compared to its position in CYP17A1. In
combination with substitution of G301 in CYP17A1 vs I291
in CYP21A2, such changes reapportion the active site volume
available to ligands. With the goal of retaining CYP17A1
inhibition, the basic abiraterone scaffold was retained, with its
N202- and heme-binding features. With the goal of reducing
CYP21A2 inhibition, substituents or modifications of the B ring
were employed. The hypothesis was that these modifications
should occupy available space between R239 and D298 in the
CYP17A1 active site and potentially engage in interactions with
these two residues but be sterically less favorable in the
CYP21A2 active site.
Scheme 2. Analogs Prepared via Epoxidation of Enol Triflate
2
a
Chemistry. All compound syntheses stemmed from enol
triflate 2, a key intermediate in the large-scale production of
abiraterone (1b; Scheme 1).17 Thus, analog 3b, which
a
Reagents and conditions: (a) m-CPBA, CH2Cl2, 0 °C to rt, 18 h,
a
Scheme 1. Synthesis of Abiraterone and Its Derivatives
90%; (b) diethyl(3-pyridyl)borane, Pd(PPh3)2Cl2, NaHCO3, THF,
H2O, 60 °C, 17 h, 36%; (c) K2CO3, MeOH, rt, 17 h, 99%; (d) KCN,
MeOH, μW, 160 °C, 50 min, 63%; (e) NaOH, H2O2, MeOH, H2O, rt,
46 h, 99%.
nucleophile, elimination could occur from the C4 or C6 carbon,
potentially resulting in a mixture of Δ4- and Δ5-constitutionally
isomers. In the case of epoxide-opening using cyanide,
isomerization of the Δ4-double bond (which would form
from trans elimination of the β-oriented nitrile) should form
the more conjugated Δ5 olefin. Accordingly, epoxide 6a was
converted into nitrile 7 as a single isomer with potassium
cyanide in methanol under microwave irradiation, effecting up
to three transformations (including deprotection) in one step.
Nitrile 7 was then hydrolyzed to amide 8 upon treatment with
sodium hydroxide and hydrogen peroxide.
A series of saturated AB-ring analogs of 5α-abiraterone were
also prepared (Scheme 3). Since m-CPBA reacted selectively at
the Δ5-double bond, it was likely that a hydroboration-
oxidation reaction would also preferentially react at this
position. Indeed, enol triflate 2, when treated with 1.05 equiv
of BH3·THF, gave 9 in 28% yield along with 39% recovered
starting material. The stereochemistry of 9 was assigned based
on the coupling pattern of the C−6 methine (i.e., a triplet of
doublets with a J value of 10.5 Hz for the triplet, suggesting two
axial−axial couplings and therefore a β hydrogen atom) and
confirmed by a strong NOE observed at the C-19 Me upon
irradiation of the C-6 methine H (Supporting Information).
The alcohol thus formed was converted to acetate 10a after
Suzuki cross-coupling and subsequently to diol 10b. Alcohol 9
was also transformed into ketone 11 upon treatment with
Dess−Martin periodinane, which was converted to ketone 12a
using the same sequence as 10b. Ketones 12a and 12b were
separately reacted with hydroxylamine hydrochloride to afford
the corresponding oximes 13a and 13b. Finally, treatment of
13a with thionyl chloride formed the ring expanded lactam 14a
via a Beckman rearrangement and 14b following deacylation.
Inhibition of CYP17A1 and CYP21A2 in Vitro by
Abiraterone and Its Analogs. These inhibitors were
evaluated for their ability to differentially inhibit in vitro
progesterone 17α-hydroxylation by CYP17A1 and progester-
one 21-hydroxylation by CYP21A2. Initial experiments without
a
Reagents and conditions: (a) phenylboronic acid, Pd(PPh3)2Cl2,
NaHCO3, THF, H2O, 60 °C, 23 h, 70%; (b) K2CO3, MeOH, rt, 16 h,
62%; (c) diethyl(3-pyridyl)borane, Pd(PPh3)2Cl2, NaHCO3, THF,
H2O, 60 °C, 23 h, 77%; (d) K2CO3, MeOH, rt, 17 h, 93%; (e) Al(O-i-
Pr)3, 2-butanone, toluene, reflux, 20 h, 60%.
substitutes the C17 pyridine ring for a phenyl, was prepared
by modifying established procedures, namely, a Suzuki cross-
coupling reaction followed by deprotection of the resulting
acetate 3a. Abiraterone itself provided the starting material for
enone 4, a key metabolite of the drug, using a previously
reported18,19 Oppenauer oxidation.
The first series of abiraterone analogs maintained the Δ5-
double bond found in the parent compound (Scheme 2). The
enol triflate 2 provided a convenient starting point for
diversification, as the Δ5-double bond selectively reacted with
m-CPBA to afford epoxide 5 in a ca. 2:1 mixture of the 5α,6α-
and 5β,6β-diastereomers, respectively. Following cross-cou-
pling, epoxide 6a was obtained after recrystallization of a ca. 4:1
mixture from MeOH−H2O (36% yield of pure 6a). To
reintroduce the Δ5-double bond, an epoxide ring-opening was
performed under basic conditions followed by elimination of
the resulting tertiary alcohol. Depending on the choice of
C
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Scheme 3. Analogs Prepared via Hydroboration−Oxidation of Enol Triflate 2
a
Reagents and conditions: (a) BH3·THF, THF, 0 °C, 4 h, then H2O followed by NaBO3·H2O, 18 h, 0 °C to rt, 18 h, 28% (and 39% of recovered
starting material); (b) diethyl(3-pyridyl)borane, Pd(PPh3)2Cl2, NaHCO3, THF, H2O, 60 °C, 23 h, 49%; (c) K2CO3, MeOH, rt, 15 h, 97%; (d)
Dess−Martin periodinane, CH2Cl2, 0 °C to rt, 16 h, 89%; (e) diethyl(3-pyridyl)borane, Pd(PPh3)2Cl2, NaHCO3, THF, H2O, 60 °C, 17 h, 82%; (f)
K2CO3, MeOH, rt, 17 h, 88%; (g) NH2OH·HCl, NaOAc, EtOH, H2O, rt, 17 h, 83%; (h) NH2OH·HCl, NaOAc, EtOH, H2O, rt, 24 h, 98%; (i)
SOCl2, THF, 0 °C, 4 h, then H2O, 45%; (j) K2CO3, MeOH, rt, 17 h, 100%.
Table 1. IC50 Values for Inhibition of CYP17A1-Mediated Progesterone 17α-Hydroxylation and CYP21A2-Mediated
Progesterone 21-Hydroxylation by Abiraterone and Its Analogs
entry
inhibitor
modification
CYP17A1 IC50 SE (nM)
CYP21A2 IC50 SE (nM)
CYP21A2 IC50/CYP17A1 IC50
1
abiraterone (1b)
4.94 0.09
28.1 2.3
>20,000
32.4 2.5
77.2 8.9
≥700
6.6
2.7
2
galeterone
3
3b
4
17-phenyl
4
Δ4, 4-oxo
6.13 0.41
7.83 0.92
15.5 2.7
22.8 1.5
7.2 0.32
12.4 0.49
16.5 2.5
230 51
0.70 0.03
42.1 8.5
23.9 3.8
673 90
234 32
440 60
1390 120
3605 1740
0.11
5.4
1.5
29
5
6b
7
5α,6α-epoxide
Δ4, C-6 CN
6
7
8
Δ4, C-6 CONH2
5α-H, 6α-OH
5α-H, 6-oxo
8
10b
12b
13b
14b
33
9
35
10
11
5α-H, 6-oxime
B-ring expanded lactam
84
16
inhibitors with purified P450 enzymes and human NADPH-
cytochrome P450 reductase were used to ensure that the
progesterone substrate was not depleted and to evaluate the
steady-state parameters. Progesterone 17α-hydroxylation by
galeterone (Figure 1), an abiraterone analog previously in
clinical trials, was also evaluated and found to be less selective
at 2.7-fold.
An initial abiraterone analog was constructed to quantitate
the contribution of the pyridine nitrogen coordination to the
heme iron. Interactions with nitrogen-containing heterocycles
are known to contribute substantially to the affinity of
inhibitors for cytochrome P450 enzymes but also contribute
to nonselectivity across P450 enzymes, as is frequently
observed for antifungal azoles. A notable example of this in
the prostate cancer field was the early use of higher doses of the
imidazole-containing antifugal ketoconazole as a treatment for
prostate cancer due to its ability to inhibit CYP17A1 and
thereby decrease systemic androgen production.22 However,
such usage frequently led to hepatotoxicity, potentially related
to broad inhibition of hepatic drug-metabolizing enzymes, and
is no longer recommended. Although contributions of this
interaction to inhibitor affinity are well-known, the effects on
inhibitor selectivity are not as clear.
CYP17A1 resulted in a Km of 6.8
1.5 μM, similar to the
previously reported value of 10.5 1.7 μM.14 Progesterone 21-
hydroxylation by CYP21A2 occurred with a Km of 0.37 0.07
μM, also similar to the previously reported value of 0.21 0.03
μM.20 Thus, progesterone at 6 and 0.3 μM concentrations were
used to determine the IC50 values for CYP17A1 and CYP21A2,
respectively.
Abiraterone has been extensively studied as an inhibitor of
CYP17A1, with reported IC50 values for progesterone 17α-
hydroxylation ranging from 92 4 to 9.4 0.3 nM.21 In the
present study, an IC50 value of 4.94 0.09 nM was determined
for abiraterone inhibition of CYP17A1-mediated progesterone
17α-hydroxylation (Table 1). By comparison, abiraterone
demonstrated an IC50 of 32.4
2.5 nM for inhibition of
CYP21A2-mediated progesterone 21-hydroxylation (Table 1).
Thus, abiraterone itself is 6.6-fold more selective for CYP17A1
inhibition than CYP21A2 inhibition. The selectivity of
Thus, analog 3b was generated in which the C17 pyridine
ring of abiraterone was substituted with a phenyl ring. As
D
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 4. CYP17A1 X-ray structures with abiraterone analogs. Abiraterone analogs bearing (A) a C6 nitrile (7; green, 6CIZ), (B) a C6 amide (8;
cyan, 6CHI), and (C) a C6 oxime (13b; magenta, 6CIR); key residues are shown as sticks. Interactions that fit the criteria for hydrogen bonding are
indicated by dashed lines. Blue mesh represents the |2Fo| − |Fc| electron density maps of the ligands contoured at 1.0σ. The heme is shown as black
sticks with its iron as an orange sphere. Notably, the N202 side chain can be a hydrogen bond donor (A) or acceptor (B,C), as can the C3 hydroxyl.
anticipated, this compound demonstrated a substantial loss in
affinity. The IC50 was increased more than 4000-fold to >20
μM for CYP17A1 and >20-fold to >0.7 μM for CYP21A2
(Table 1, entry 3). This compound is also more selective for
inhibition of CYP21A2 rather than CYP17A1, opposite of the
desired goal. Accordingly, all subsequent abiraterone analogs
retained the pyridine.
observed but was accompanied by a 43-fold increase in the IC50
for CYP21A2 (entry 10). Thus, this analog is ca. 84 times more
selective for inhibition of CYP17A1 over CYP21A2 for their
respective progesterone hydroxylation reactions.
An alternate strategy to incorporate ligand moieties that
might engage with R239 and D298 in the CYP17A1 active site
was to expand and modify the B ring. Thus, the B ring was
replaced with a caprolactam ring (14b). This modification
resulted in poorer potency against both targets (Table 1, entry
11), with ca. 16-fold selectivity for CYP17A1 over CYP21A2.
Accordingly, no further analogs were pursued following this
strategy.
X-ray Structures of Key CYP17A1/Inhibitor Com-
plexes. In order to better understand interactions between
CYP17A1 and some of the abiraterone inhibitor analogs tested
herein, X-ray structures were determined with analogs
incorporating the C6 nitrile 7, the C6 amide 8, and the C6
oxime 13b. Complexes were formed by cocrystallization, and
structures were determined (Supporting Information, Table
S1) as described in the Experimental Section. The resulting
structures revealed the same overall CYP17A1 conformation
and placement of the steroidal core observed for other
CYP17A1 steroidal substrates and inhibitors. All three
abiraterone analogs retained (1) the interaction between the
pyridine nitrogen and the heme iron and (2) a hydrogen bond
between the C3 hydroxyl and N202 (Figure 4).
A 2.6 Å structure of CYP17A1 with the C6 nitrile (7)
revealed that this substituent extended into the space between
D298 and R239 as expected (Figure 4A). Its direct projection
from the B ring displaced a water molecule often found in this
position and places the nitrile N atom to accept hydrogen
bonds from the side chain of R239 (2.9−3.5 Å for the two
different interactions in the four different molecules CYP17A1/
inhibitor complexes of the asymmetric unit). This occurred
without significant reorientation of the R239 side chain
compared to the abiraterone structure. No other significant
changes were observed in the active site.
Progesterone, which contains a Δ4 double bond and a C3
ketone, differs in its interactions between the C3 ketone and
different residues in the active site roof of CYP17A1 (N202 in
the F helix) vs CYP21A2 (R234 in the G helix; see Figure 3
above). In addition, conversion of abiraterone to its Δ4
metabolite leads to increased anticancer activity, in part due
to greater androgen receptor antagonism.19,23 This metabolite,
compound 4, has been reported to inhibit CYP17A1
comparably to abiraterone itself.18 Similarly, little change was
observed in the IC50 value of 4 against CYP17A1, but there was
a 46-fold decrease in the IC50 for CYP21A2 (Table 1, entry 4).
Thus, 4 was 8.8 times more selective for CYP21A2 over
CY17A1 (Table 1), which was again opposite of the design
goal. The IC50 curves for all analogs except 3b are shown in
Figure S1 (Supporting Information).
Analogs 7 and 8 retaining Δ5 unsaturation and modifications
at C6 were tested. These compounds largely maintained the
ability to inhibit CYP17A1, with modest 3- and 4.6-fold
increases in IC50 values for CYP17A1-mediated progesterone
17α-hydroxylation, respectively (Table 1, entries 6 and 7).
However, the nitrile substitution in 7 did not significantly alter
the IC50 for CYP21A2-mediated progesterone 21-hydroxyla-
tion, meaning that the 7 analog is less selective for CYP17A1
than abiraterone itself. In contrast, however, the C6 amide of 8
decreased inhibition of CYP21A2 by 20-fold to yield an IC50
>
600 nM. As a result, the selectivity increased from 6.6-fold for
abiraterone itself to 29-fold with this modification, validating
the idea that C6 substituents could play an important role in
determining selectivity and suggesting that bulkier substitutions
in this part of the molecule might be advantageous.
A second set of analogs lacking A/B unsaturation and
containing modifications at C6 was examined. The epoxide 6b
behaved similarly to abiraterone in all respects (Table 1, entry
5). In contrast, analogs containing a C6α alcohol (10b) and C6
ketone (12b) resulted in markedly increased selectivity, mostly
due to substantially poorer potency at CYP21A2 (entries 8 and
9). Even higher selectivity was observed for oxime 13b. In this
case, a ca. 3-fold increase in the IC50 for CYP17A1 was
A 2.7 Å structure of CYP17A1 bound to the C6 amide analog
(8) demonstrated the amide similarly positioned between the
polar R239 and D298 side chains (Figure 4B). In this instance,
the amide oxygen served as a hydrogen bond acceptor from the
side chain of R239 (2.8−3.4 Å), whereas the amide nitrogen
simultaneously formed a hydrogen bond donor to D298 (3.1−
3.5 Å). Thus, this substituent engages both targeted polar
residues in the CYP17A1 active site. In addition, the amide
E
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
nitrogen is placed to form a weak hydrogen bond (3.4−3.5 Å)
with the backbone carbonyl of A105 in the B′ helix.
Interestingly, the amide is twisted slightly out of conjugation
with the Δ5 olefin, suggesting some induced fit.
A 2.65 Å structure of CYP17A1 cocrystallized with the C6
oxime analog (13b) demonstrated that the oxime is projected
somewhat similarly but above the plane between R239 and
D298 (Figure 4C). As a result, the nitrogen is positioned to
form a weak hydrogen bond (3.3−3.5 Å) with R239, while the
substituent hydroxyl and D298 are similarly spaced for
hydrogen bonding (3.2−3.3 Å). As in the case of the amide
of 8, the substituent hydroxyl of 13b also engaged in an
unanticipated hydrogen bond with the backbone carbonyl of
A105 in the B′ helix, but in this case, the shorter distance
suggests a stronger interaction (2.2−3.0 Å).
All three of these analogs were designed to specifically
complement the CYP17A1 active site. Their IC50 values were
only 1.2- to 4.6-fold higher than the parent abiraterone, and
these structures revealed that they retained key interactions
with the heme iron and N202, while the newly introduced B
ring substituents were easily accommodated without changes in
the CYP17A1 active site. Furthermore, the various substituents
engaged in the intended new interactions with R239 in the G
helix and D298 in the I helix, as well as making an
unanticipated interaction with the backbone of the B′ helix.
Since the B′ helix often has relatively few interactions with the
rest of the P450 protein and has been implicated in
conformational changes associated with ligand entry and exit,
such interactions may lead to reduced inhibitor off rates.
Docking of Abiraterone Analogs into CYP17A1 and
CYP21A2. While the structures above rationalized why several
of these analogs bind and inhibit CYP17A1 similar to the
parent abiraterone, the primary effects of such modifications
were to increase selectivity by reducing inhibition of CYP21A2.
As decreased CYP21A2 affinity is not likely to support
cocrystallization of CYP21A2/inhibitor complexes, abiraterone
and key compounds in this series (7, 6b, 8, 13b, 12b, 10b)
were docked into CYP17A1 and CYP21A2.
Figure 5. Compound 13b (sticks) docked into human CYP17A1
(yellow; from PDB 3RUK, with bound abiraterone) and CYP21A2
(cyan; from PDB 4Y8W, with bound progesterone). Heme in black
sticks with iron as central red sphere. While 13b is accommodated in
CYP17A similarly to abiraterone itself, in CYP21A2·13b, the only pose
that could be docked at all has unfavorable interactions with D107 and
V360. Amino acids are labeled with the CYP17A1 residue given first,
followed by the name of the CYP21A2 residue.
Table 2. Statistics for Docking Abiraterone and Its Analogs
into CYP17A1 and CYP21A2
ΔGbind (kcal/mol)
inhibitor
CYP17A1
CYP21A2
abiraterone (1b)
−42.6
−43.8
−42.6
−46.6
−48.0
−46.6
−49.0
−11.9
11.4
−1.1
−6.4
2.4
8
7
6b
12b
10b
13b
−4.3
8.5
When abiraterone and each of these analogs were docked
into CYP17A1, all analogs retained both (1) the hydrogen
bond between C3 hydroxyl and the CYP17A1 N202 side chain
and (2) pyridine nitrogen coordination to the heme iron, as
exemplified in Figure 5. Compounds 7, 8, and 13b docked into
CYP17A1 reveal very similar conformations and hydrogen
bonding to CYP17A1 residues as shown experimentally in the
X-ray structures of CYP17A1 with these compounds. While the
C6 substituents of 7 and 8 formed hydrogen bonds with either
R239/D298 or both, the C6 substituent of 6b did not make any
contact with R239 and D298 upon docking. Similarly, the C6
hydroxyl substituent in 10b and keto group of 12b did not
hydrogen bond with side chain D298 but did form weak
hydrogen bonds with the side chain of R239. None of these
docked compounds reveal significant unfavorable clashes with
CYP17A1, and the favorable ΔGbind ranges from −42 to −49
kcal/mol, consistent with relatively small differences between
the IC50 values of these analogs for CYP17A1 (Table 2).
Progesterone docked into CYP21A2 very similarly to the X-
ray structure shown in Figure 3, but docking abiraterone
analogs into CYP21A2 revealed a number of key differences.
Abiraterone itself docked in a similar orientation to
progesterone, with the C3 OH hydrogen bonding to R234.
However, docking of 8 into CYP21A2 revealed that R234 and
D288 were involved in hydrogen bonding to the C6 amide
carbonyl and nitrogen, respectively. In this orientation, docked
8 resulted in unfavorable van der Waals energy for adjacent
residues V287 and V360. Overall, the computed ΔGbind was a
highly unfavorable 11.4 kcal/mol for 8, in comparison to −11.9
kcal/mol for abiraterone docked into CYP21A2. All conformers
of compound 13b failed to dock in the CYP21A2 except one.
This single conformer (Figure 5) demonstrated a hydrogen
bond between the C3 hydroxyl of 13b and the side chain of
R234, but no hydrogen bond between the C6 oxime and either
R234 or D288. Instead, the oxime hydrogen bonded with the
side chain of D107, but this side chain also imparts a large
unfavorable van der Waals energy. As a result, the overall
calculated ΔGbind was also highly unfavorable at 8.5 kcal/mol.
Docking 12b into CYP21A2 suggested a hydrogen bond
between its C3 hydroxyl and the side chain of R234 with no
interaction with D288. Both the side chains of D107 and V360
resulted in unfavorable van der Waals interactions. Finally,
docking of compounds 7, 6b, and 10b into CYP21A2 suggested
hydrogen bonds between the C3 hydroxyl and the R234 side
chain. Again, a large unfavorable van der Waals interaction with
V360 is calculated for 10b, which is somewhat reduced for 7
and 6b. The unfavorable van der Waals energy contribution
F
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
tetramethylsilane (TMS).25 Coupling constants are reported in
Hertz (Hz), and peak multiplicities as either a singlet (s), doublet
(d), triplet (t), multiplet (m), or complex. Infrared (IR) spectra were
acquired on a Thermo Scientific Nicolet iS 50 FT-IR Spectrometer.
Melting points were determined on an Optimelt MPA100 instrument
and are uncorrected. Optical rotations were measured with a Rudolph
AUTOPOL IV. High-resolution mass spectrometry (HRMS) data
were collected on a Thermo Electron hybrid ion trap FT-ICR mass
spectrometer equipped with a 7T ICR magnet (LTQ-FT) using either
an electrospray ionization (ESI) or atmospheric-pressure chemical
ionization (APCI) source. Purity data were measured with a Waters
ACQUITY UPLC H-Class System coupled to the aforementioned
mass spectrometer with the ESI source. All samples were prepared in
MeOH at a concentration of 1 mg/mL, and 3 μL of each solution was
introduced at a flow rate of 0.6 mL/min onto a Waters Acquity UPLC
BEH C18 column (2.1 × 50 mm, 1.7 um particle size) at 40 °C. The
solvent gradient was as follows: H2O with 0.1% formic acid ramped
linearly over 9.80 min to 95% MeCN with 0.1% formic acid and held
for 0.40 min. At 10.21 min, the gradient was switched back to H2O and
allowed to re-equilibrate for 1.05 min to prepare for the next sample.
Compound purity was determined on the basis of peak integration
(area under the curve) from the total ion chromatogram (TIC), and
HRMS data provided verification of compound identity. All
compounds used for biological studies had an LC purity of >95%.
General Procedure A for Suzuki Cross-Coupling. A flame-
dried flask was charged with the appropriate enol triflate, either 3-
(diethylboranyl) pyridine or phenylboronic acid, and bis-
(triphenylphosphine)palladium(II) dichloride. After vacuum flushing
with Ar (×3), anhydrous THF was added, followed by a saturated
solution of NaHCO3 that had been sparged with N2 for 15 min prior
to addition. The reaction mixture was heated to 60 °C and stirred
overnight. After cooling to rt, the solvent was removed under a stream
of N2. The resulting black residue was dissolved in CH2Cl2 and washed
with H2O, and the aqueous layer were extracted with CH2Cl2 (×2).
The combined organic layers were dried over MgSO4, filtered, and
concentrated. The crude product was purified by chromatography to
afford the corresponding pyridyl or phenyl compound.
from V360 in CYP21A2 observed for many of the C6
substituted compounds appears to be ameliorated in
CYP17A1 as this position is substituted by A367. Similarly,
D107 in CYP21A2 is G111 in CYP17A1, which may also
support binding to CYP17A1.
CONCLUSIONS
■
Using structure-based design, abiraterone analogs were
generated that had IC50 values similar to the parent inhibitor
for the target enzyme CYP17A1, but increased IC50 values for
the off-target CYP21A2 enzyme by ca. 43-fold. This increased
the 6.6-fold selectivity of abiraterone for CYP17A1 over
CYP21A2 to as high as 84-fold, which was observed for the
C6 oxime 13b. The design strategy was to maintain CYP17A1
potency by adding substituents to the C6 of the steroid
backbone able to engage in attractive hydrogen bonding
interactions with the target but cause steric clashes with the off-
target CYP21A2. These features were confirmed by X-ray
crystallographic studies of three CYP17A1−inhibitor complexes
and consistent with docking studies of the analogs with
CYP21A2.
Although experimental testing with the full panel of human
cytochrome P450 enzymes is beyond the scope of this article,
we have briefly examined the behavior of compound 13b vs
abiraterone with respect to two key drug-metabolizing P450
enzymes. It is known that abiraterone is a substrate of CYP3A4
(formation of the N-oxide) and an inhibitor of the CYP2D6.24
As a result, coadministration of CYP2D6 substrates with a
narrow therapeutic index is not recommended for abiraterone.
Accordingly, we tested CYP2D6 for inhibition by abiraterone
against oxime 13b. While the IC50 for abiraterone was 0.23 μM,
the IC50 for 13b was 8.57 μM, indicating that the potential for
inhibition of CYP2D6 is actually decreased by 37-fold.
Moreover, if abiraterone or its analogs are substrates for
CYP3A4, then they ought to competitively inhibit substrate
metabolism. However, our results show no significant inhibition
of CYP3A4 by either abiraterone or 13b up to concentrations
of 640 nM. Since the Cmax for abiraterone in patients is 90 nM
under the typical dosing regimen, we do not anticipate this
would be clinically significant.
General Procedure B for Acetate Deprotection. A vial was
charged with the appropriate acetate, K2CO3, and MeOH. After
stirring overnight, the reaction was quenched with a saturated solution
of NH4Cl, and the aqueous layer was extracted with CH2Cl2 (×3). The
combined organic layers were dried over MgSO4, filtered, and
concentrated. The crude product was purified by chromatography to
afford the corresponding alcohol.
General Procedure C for Oxime Formation. A vial was charged
with the appropriate ketone, NH2OH·HCl, NaOAc, EtOH, and H2O.
The reaction mixture was stirred vigorously overnight and then
concentrated under a stream on N2. The white solid was dissolved in
CH2Cl2. The organic layer was washed with H2O, and the aqueous
layer was extracted with CH2Cl2. The combined organic layers were
dried over Na2SO4, filtered, and concentrated. The crude product was
purified by chromatography to afford the corresponding oxime.
17-(3-Pyridyl)androsta-5,16-dien-3β-yl Acetate (Abiraterone Ac-
etate, 1a).17,26 Following general procedure A, 2 (0.501 g, 1.08 mmol)
was coupled to 3-(diethylboranyl) pyridine (0.181 g, 1.23 mmol, 1.1
equiv) using bis(triphenylphosphine)palladium(II) dichloride (77.5
mg, 110 μmol, 0.1 equiv) in anhydrous THF (6.5 mL) and a saturated
solution of NaHCO3 (2.2 mL) for 23 h. After workup, the crude
product was purified by chromatography (12 g of silica gel, 0−50%
EtOAc/hexanes) to afford 1a (0.326 g, 0.833 mmol, 77% yield) as a
Overall, reducing the off-target inhibition of CYP21A2 is
expected to reduce the undesirable clinical side effects resulting
from disruption of glucocorticoid and mineralocorticoid
biosynthesis.
EXPERIMENTAL SECTION
■
General Information. All chemicals and solvents were purchased
from commercial sources (Alfa Aesar, Ark Pharm, Oakwood Chemical,
or Sigma-Aldrich) and used as received. Unless stated otherwise,
reactions were performed under ambient conditions and monitored by
thin-layer chromatography (TLC) using Analtech silica gel GHLF
(250 μm) coated glass plates, which were visualized by either
shortwave UV light or cerium ammonium molybdate (CAM) stain.
Normal or reverse phase chromatography was carried out on a
Teledyne Isco Combiflash purification system. Microwave reactions
were performed using a Biotage Initiator Classic with an autosampler.
All nuclear magnetic resonance (NMR) spectra (1H, 13C, and 19F)
were recorded in deuterated solvents (CDCl3 or DMSO-d6) on a
Varian 400 MR NMR spectrometer. Chemical shifts are reported in
parts per million (ppm) and were adjusted using the residual
undeuterated solvents (CDCl3: 7.26 ppm for 1H NMR, 77.2 ppm
1
white, amorphous solid. Rf = 0.6 (50% EtOAc in hexanes); H NMR
(400 MHz, CDCl3) δ 8.61 (d, J = 2.1 Hz, 1H), 8.45 (dd, J = 4.9, 1.6
Hz, 1H), 7.64 (dt, J = 7.9, 1.7 Hz, 1H), 7.21 (dd, J = 7.9, 4.8 Hz, 1H),
5.99 (dd, J = 3.3, 1.8 Hz, 1H), 5.41 (d, J = 5.2 Hz, 1H), 4.68−4.55 (m,
1H), 2.41−2.29 (m, 2H), 2.26 (ddd, J = 15.8, 6.5, 3.3 Hz, 1H), 2.10−
2.04 (m, 2H), 2.03 (s, 3H), 1.91−1.82 (m, 2H), 1.76−1.54 (complex,
7H), 1.53−1.43 (m, 2H), 1.20−1.10 (m, 1H), 1.08 (s, 3H), 1.04 (s,
3H); 13C NMR (101 MHz, CDCl3) δ 170.7, 151.8, 148.1, 148.0,
140.2, 133.8, 133.1, 129.3, 123.1, 122.4, 74.0, 57.6, 50.4, 47.5, 38.3,
37.1, 36.9, 35.4, 31.9, 31.6, 30.6, 27.9, 21.6, 21.0, 19.4, 16.7; IR (neat)
1
for 13C NMR; DMSO-d6: 2.50 ppm for H NMR, 39.5 ppm for 13C
NMR) as an internal reference. Using the unified scale, 19F NMR
spectra were referenced with respect to the 1H frequency of
G
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
17
21
1732 cm−1; mp 142−145 °C, lit. 144−145 °C (hexanes); [α]D
as a white, amorphous solid. Rf = 0.4 (25% EtOAc/hexanes); 1H NMR
(400 MHz, CDCl3) δ 7.40−7.36 (m, 2H), 7.33−7.27 (m, 2H), 7.25−
7.20 (m, 1H), 5.92 (dd, J = 3.3, 1.8 Hz, 1H), 5.42−5.38 (m, 1H),
3.60−3.49 (m, 1H), 2.37−2.19 (m, 3H), 2.14−1.99 (m, 3H), 1.90−
1.81 (m, 2H), 1.81−1.44 (complex, 8H), 1.15−1.02 (m, 2H), 1.08 (s,
3H), 1.07 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 154.9, 141.2,
137.5, 128.2 (2C), 127.3, 126.8 (3C), 121.6, 71.9, 57.8, 50.6, 47.3,
42.5, 37.3, 36.8, 35.5, 31.79, 31.76, 31.7, 30.6, 21.1, 19.5, 16.8; IR
−49.2 (c 1.00, CHCl3); HRMS (ESI) m/z [M + H]+ calcd for
C26H34NO2 392.2584; found 392.2576.
17-(3-Pyridyl)androsta-5,16-dien-3β-ol (Abiraterone, 1b).17,26
Following general procedure B, 1a (60.5 mg, 0.155 mmol) was
reacted with K2CO3 (0.212 g, 1.53 mmol, 10.0 equiv) in MeOH (5.1
mL) for 17 h. After workup, the crude product was purified by
chromatography (4 g of silica gel, 0−60% EtOAc/hexanes) to afford
1b (50.3 mg, 93% yield) as a white, amorphous solid. Rf = 0.5 (60%
EtOAc/hexanes); 1H NMR (400 MHz, CDCl3) δ 8.62 (d, J = 2.1 Hz,
1H), 8.45 (dd, J = 4.8, 1.5 Hz, 1H), 7.64 (dt, J = 7.9, 2.0 Hz, 1H), 7.21
(dd, J = 7.9, 4.8 Hz, 1H), 5.99 (dd, J = 3.4, 1.8 Hz, 1H), 5.41−5.37 (m,
1H), 3.59−3.49 (m, 1H), 2.36−2.20 (m, 3H), 2.12−2.00 (m, 3H),
1.89−1.81 (m, 2H), 1.80−1.43 (complex, 8H), 1.15−1.10 (m, 2H),
1.07 (s, 3H), 1.05 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 151.8,
148.1, 148.0, 141.3, 133.8, 133.1, 129.4, 123.2, 121.5, 71.8, 57.7, 50.5,
47.5, 42.5, 37.3, 36.9, 35.4, 32.0, 31.8, 31.7, 30.6, 21.0, 19.5, 16.7; IR
(neat) 3221, 1446 cm−1; mp 224−227 °C, lit.17 228−229 °C
(toluene); [α]D22 −63.4 (c 1.00, CHCl3); HRMS (ESI) m/z [M + H]+
calcd for C24H32NO 350.2478; found 350.2470; LCMS tR = 4.86 min.
3β-Acetoxyandrosta-5,16-dien-17-yl Trifluoromethanesulfonate
(2).17,27 A flame-dried 250 mL round-bottom flask was charged with
15 (16.0 g, 48.4 mmol), 2,6-lutidine (11.3 mL, 97.0 mmol, 2.0 equiv),
and anhydrous CH2Cl2 (320 mL) under Ar. The reaction mixture was
cooled to −78 °C, and Tf2O (15 mL, 87 mmol, 1.8 equiv) was added
dropwise over 30 min. The reaction mixture was stirred at −78 °C for
2 h and then stored in a −20 °C freezer for 16 h. While the reaction
mixture was still cold, a solution of 1 N HCl (200 mL) was added, and
the aqueous layer was extracted with CH2Cl2 (200 mL). The
combined organic layers were washed with brine (200 mL), dried over
Na2SO4, filtered, and concentrated. The crude product was purified by
chromatography (500 g of silica gel, 25% EtOAc/hexanes) to afford 2
(17.0 g, 36.7 mmol, 76% yield) as an off-white, amorphous solid. An
analytical sample was prepared by recrystallization from MeOH−H2O.
22
(neat) 3258 cm−1; mp 150−154 °C; [α]D −70.1 (c 1.00, CHCl3);
HRMS (APCI) m/z [M − OH]+ calcd for C25H31 331.2420; found
331.2420; LCMS tR = 8.95 min.
17-(3-Pyridyl)androsta-4,16-dien-3-one (4).17,18 Prepared by
adapting a previously reported procedure. A flame-dried 25 mL
round-bottom flask was charged with 1b (0.250 g, 0.715 mmol),
butan-2-one (2.2 mL, 25 mmol, 35 equiv), and anhydrous toluene (7.3
mL) under Ar. The reaction mixture was heated to reflux for 20 h.
After cooling to rt, the reaction mixture was concentrated and diluted
with EtOAc (30 mL). The organic layer was washed with brine (20
mL), and the aqueous layer was extracted with EtOAc (30 mL). The
combined organic layers were dried over Na2SO4, filtered, and
concentrated. The crude product was purified by chromatography (12
g of silica gel, 0−50% EtOAc/hexanes) to afford 4 as a white,
amorphous solid (0.150 g, 0.430 mmol, 60% yield). An analytical
sample was prepared through recrystallization from MeOH−H2O. Rf =
1
0.3 (50% EtOAc/hexanes); H NMR (400 MHz, CDCl3) δ 8.61 (d, J
= 2.3 Hz, 1H), 8.46 (dd, J = 4.7, 1.3 Hz, 1H), 7.63 (dt, J = 7.9, 1.8 Hz,
1H), 7.21 (dd, J = 7.6, 4.8 Hz, 1H), 5.98 (dd, J = 3.3, 1.8 Hz, 1H), 5.75
(s, 1H), 2.53−2.22 (complex, 5H), 2.15−1.98 (m, 3H), 1.97−1.87 (m,
1H), 1.86−1.77 (m, 1H), 1.76−1.64 (m, 2H), 1.63−1.42 (m, 3H),
1.23 (s, 3H), 1.22−0.99 (m, 2H), 1.06 (s, 3H); 13C NMR (101 MHz,
CDCl3) δ 199.5, 171.0, 151.7, 148.2, 148.0, 133.8, 132.9, 129.2, 124.2,
123.2, 56.9, 54.1, 47.4, 38.8, 35.7, 35.2, 34.3, 34.1, 32.9, 31.9, 31.8,
21.0, 17.4, 16.8; IR (neat) 1665, 1610 cm−1; mp 145−147 °C
(MeOH−H2O), lit.17 148−150 °C (Et2O); [α]D +138 (c 1.00,
22
1
CHCl3); HRMS (ESI) m/z [M + H]+ calcd for C24H30NO 348.2322;
found 348.2316; LCMS tR = 4.79 min.
Rf = 0.4 (10% EtOAc in hexanes); H NMR (400 MHz, CDCl3) δ
5.58 (dd, J = 3.3, 1.6 Hz, 1H), 5.39 (d, J = 5.3 Hz, 1H), 4.67−4.53 (m,
1H), 2.39−2.28 (m, 2H), 2.23 (ddd, J = 14.9, 6.2, 3.4 Hz, 1H), 2.03 (s,
3H), 2.01−1.94 (m, 1H), 1.92−1.81 (m, 2H), 1.80−1.74 (m, 1H),
1.72−1.52 (m, 6H), 1.52−1.42 (m, 2H), 1.21−1.05 (m, 2H), 1.05 (s,
3H), 0.99 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 170.6, 159.3,
140.3, 121.9, 118.7 (q, J = 321 Hz), 114.6, 73.8, 54.4, 50.5, 44.8, 38.2,
37.0, 36.9, 32.8, 30.7, 30.1, 28.7, 27.8, 21.6, 20.3, 19.3, 15.2; 19F NMR
(376 MHz, CDCl3) δ −73.6 (s); IR (thin film) 1732, 1629 cm−1; mp
73−76 °C (MeOH−H2O), lit.17 75−76 °C (hexanes); [α]D23 −51.8 (c
1.00, CHCl3); HRMS (APCI) m/z [M − CH3CO2]+ calcd for
C20H36F3O3S 403.1549; found 403.1543.
17-Phenylandrosta-5,16-dien-3β-yl Acetate (3a). Following gen-
eral procedure A, 2 (0.199 g, 0.430 mmol) was coupled to
phenylboronic acid (60.5 mg, 0.496 mmol, 1.1 equiv) using
bis(triphenylphosphine)palladium(II) dichloride (31.1 mg, 44.0
μmol, 0.1 equiv) in anhydrous THF (2.9 mL) and a saturated
solution of NaHCO3 (0.86 mL) for 23 h. After workup, the crude
product was purified by chromatography (12 g of silica gel, 0−5%
EtOAc/hexanes) to afford 3a (0.117 g, 0.301 mmol, 70% yield) as a
3β-Acetoxy-5α,6α-epoxyandrost-16-en-17-yl Trifluoromethane-
sulfonate (5α,6α-5) and 3β-Acetoxy-5β,6β-epoxyandrost-16-en-
17-yl Trifluoromethanesulfonate (5β,6β-5). A 100 mL round-bottom
flask was charged with 2 (1.60 g, 3.46 mmol) and anhydrous CH2Cl2
(35 mL). The reaction mixture was cooled to 0 °C, and m-CPBA (77
wt %, 0.775 g, 3.46 mmol, 1.0 equiv) was added in one portion. The
reaction mixture was allowed to warm to rt and stirred for 18 h.
Afterward, the reaction mixture was diluted with CH2Cl2 (100 mL)
and washed with a saturated solution of NaHCO3 (100 mL). The
aqueous layer was extracted with CH2Cl2 (50 mL). The combined
organic layers were dried over MgSO4, filtered, and concentrated. The
crude product was purified by chromatography (40 g of silica gel, 0−
10% EtOAc/hexanes) to afford an inseparable mixture of 5α,6α-5 and
5β,6β-5 (1.49 g, 3.12 mmol, 90% yield) in a ca. 2:1 ratio (determined
1
by H NMR) as a clear, colorless oil, which solidified into a white,
amorphous solid upon standing. Rf = 0.1 (10% EtOAc/hexanes);
5α,6α-5 (major diastereomer): 1H NMR (400 MHz, CDCl3) δ 5.59−
5.42 (m, 1H), 5.01−4.88 (m, 1H), 2.91 (dd, J = 4.5, 1.3 Hz, 1H),
2.26−1.89 (complex, 5H), 2.00 (d, J = 1.4 Hz, 3H), 1.76−1.22
(complex, 12H), 1.10 (s, 3H), 0.92 (s, 3H); 13C NMR (101 MHz,
CDCl3) δ 170.3, 159.1, 118.7 (q, J = 321 Hz), 114.3, 71.3, 65.4, 58.7,
54.2, 44.8, 42.8, 36.2, 35.4, 32.4, 32.1, 28.5, 28.1, 27.5, 27.3, 21.5, 20.0,
16.0, 15.16; 19F NMR (376 Hz, CDCl3) δ −73.60 (s). 5β,6β-5 (minor
1
white, amorphous solid. Rf = 0.6 (10% EtOAc/hexanes); H NMR
(400 MHz, CDCl3) δ 7.39−7.35 (m, 2H), 7.32−7.27 (m, 2H), 7.25−
7.20 (m, 1H), 5.91 (dd, J = 3.4, 1.8 Hz, 1H), 5.45−5.39 (m, 1H),
4.67−4.56 (m, 1H), 2.39−2.29 (m, 2H), 2.23 (ddd, J = 15.6, 6.5, 3.3
Hz, 1H), 2.13−1.98 (m, 3H), 2.04 (s, 3H), 1.92−1.83 (m, 2H), 1.81−
1.73 (complex, 8H), 1.21−1.10 (m, 1H), 1.09 (s, 3H), 1.06 (s, 3H);
13C NMR (101 MHz, CDCl3) δ 170.7, 154.9, 140.2, 137.5, 128.2
(2C), 127.3, 126.8 (3C), 122.6, 74.1, 57.7, 50.5, 47.7, 38.3, 37.1, 37.0,
35.5, 31.8, 31.7, 30.6, 27.9, 21.6, 21.0, 19.4, 16.8; IR (neat) 1731 cm−1;
mp 141−144 °C; [α]D22 −58.0 (c 1.00, CHCl3); HRMS (APCI) m/z
[M − CH3CO2]+ calcd for C25H31 331.2420; found 331.2415.
17-Phenylandrosta-5,16-dien-3β-ol (3b). Following general pro-
cedure B, 3a (78.1 mg, 0.200 mmol) was reacted with K2CO3 (0.275 g,
1.99 mmol, 10.0 equiv) in MeOH (6.7 mL) for 16 h. After workup, the
crude product was purified by chromatography (4 g of silica gel, 0−
25% EtOAc/hexanes) to afford 3b (43.2 mg, 0.124 mmol, 62% yield)
1
diastereomer, diagnostic peaks only): H NMR (400 MHz, CDCl3) δ
4.82−4.68 (m, 1H), 3.13−3.10 (m, 1H), 2.02 (d, J = 1.5 Hz, 3H),
1.88−1.79 (m, 1H), 1.03 (s, 3H), 0.94 (s, 3H); 13C NMR (101 MHz,
CDCl3) δ 170.7, 159.2, 114.3, 71.2, 63.3, 62.8, 54.0, 51.6, 44.7, 38.1,
36.8, 35.5, 32.8, 31.5, 28.7, 28.0, 21.4, 21.0, 17.1, 15.24; 19F NMR (376
Hz, CDCl3) δ −73.61 (s). 5α,6α-5 and 5β,6β-5: IR (thin film) 1732,
1628 cm−1; HRMS (APCI) m/z [M − CH3CO2]+ calcd for
C20H26F3O4S 419.1498; found 419.1497.
5α,6α-Epoxy-17-(3-pyridyl)androst-16-en-3β-yl acetate (6a) and
5β,6β-Epoxy-17-(3-pyridyl)androst-16-en-3β-yl acetate (5β,6β-6a).
Following general procedure A, 5 (3.00 g, 6.27 mmol) was coupled to
H
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
3-(diethylboranyl)pyridine (1.04 g, 7.10 mmol, 1.1 equiv) using
bis(triphenylphosphine)palladium(II) dichloride (0.440 g, 0.627
mmol, 0.1 equiv) in anhydrous THF (37 mL) and a saturated
solution of NaHCO3 (12.5 mL) for 17 h. After workup, the crude
product was purified by chromatography (80 g of silica gel, 0−50%
EtOAc/hexanes) to afford a mixture of 6a and 5β,6β-6a (1.65 g, 4.06
IR (thin film) 3345, 2209 cm−1; mp 232−235 °C (MeOH−EtOAc,
23
dec); [α]D −69.9 (c 1.00, CHCl3); HRMS (ESI) m/z [M + H]+
calcd for C25H31N2O 375.2431; found 375.2433; LCMS tR = 4.38 min.
3β-Hydroxy-17-(3-pyridyl)androsta-5,16-dien-6-carboxamide (8).
A 25 mL round-bottom flask was charged with 7 (99.9 mg, 0.267
mmol) and MeOH (13.5 mL). Pellets of NaOH (0.485 g, 11.5 mmol,
43.0 equiv) were added. Note: It was important to ensure that the
NaOH is fully dissolved before continuing on to the next step. Then, a
solution of H2O2 (30 wt %, 7.6 mL, 74 mmol, 280 equiv) was added
dropwise over 5 min. Note: During the addition, the reaction mixture
went from clear, colorless to cloudy, white, then back to clear,
colorless. After stirring for 46 h, the reaction mixture was diluted with
H2O (10 mL), and the aqueous layer was extracted with CH2Cl2 (3 ×
10 mL). The combined organic layers were concentrated. The crude
product was purified by chromatography (50 g of C18 silica gel, H2O
containing 0.1 vol % NH4OH to MeCN) to afford 8 (0.104 g, 0.264
mmol, 99% yield) as a white, amorphous solid. Rf = 0.4 (10% MeOH/
1
mmol, 65% yield) in a ca. 4:1 ratio (determined by H NMR) as an
off-white, amorphous solid. Some of this material (0.572 g, 1.40
mmol) was recrystallized from MeOH−H2O to give an analytical
sample of 6a (0.208 g, 0.511 mmol, 36% recovery) as a white,
crystalline solid. Rf = 0.3 (50% EtOAc/hexanes); 6a (major
1
diastereomer): H NMR (400 MHz, CDCl3) δ 8.58 (s, 1H), 8.44
(d, J = 3.9 Hz, 1H), 7.60 (dt, J = 8.0, 2.0 Hz, 1H), 7.19 (dd, J = 8.0, 4.8
Hz, 1H), 5.97−5.93 (m, 1H), 5.01−4.89 (m, 1H), 2.94 (d, J = 4.5 Hz,
1H), 2.31−2.11 (m, 2H), 2.07−1.93 (complex, 4H), 2.02 (s, 3H),
1.90−1.31 (complex, 12H), 1.13 (s, 3H), 0.96 (s, 3H); 13C NMR (101
MHz, CDCl3) δ 170.3, 151.7, 148.1, 148.0, 133.8, 132.9, 129.1, 123.1,
71.4, 65.5, 59.0, 57.4, 47.4, 42.6, 36.2, 35.4, 35.0, 32.1, 31.7, 28.6, 28.5,
27.3, 21.5, 20.6, 16.6, 16.0; IR (neat) 1732 cm−1; mp 197−200 °C
(MeOH−H2O); [α]D22 −52.7 (c 1.00, CHCl3); HRMS (ESI) m/z [M
+ H]+ calcd for C26H34NO3 408.2533; found 408.2547. 5β,6β-6a
1
CH2Cl2); H NMR (400 MHz, CDCl3) δ 8.60 (dd, J = 2.2, 6.0 Hz,
1H), 8.45 (dd, J = 4.8, 1.6 Hz, 1H), 7.63 (dt, J = 7.9, 2.0 Hz, 1H), 7.22
(ddd, J = 7.9, 4.8, 0.7 Hz, 1H), 5.99 (dd, J = 3.3, 1.8 Hz, 1H), 5.74 (s,
1H), 5.53 (s, 1H), 3.62−3.52 (m, 1H), 2.86 (ddd, J = 13.5, 4.7, 2.2 Hz,
1H), 2.38−2.23 (m, 2H), 2.22−2.13 (m, 1H), 2.12−1.76 (complex,
6H), 1.73−1.44 (complex, 7H), 1.19−1.05 (m, 1H), 1.11 (s, 3H), 1.04
(s, 3H); 13C NMR (101 MHz, CDCl3) δ 174.6, 151.6, 148.0, 147.9,
139.8, 133.9, 132.9, 129.3, 129.0, 123.2, 71.1, 57.3, 50.0, 47.4, 37.8,
37.01, 36.99, 35.3, 33.6, 31.9, 31.1, 30.1, 21.0, 19.6, 16.7; IR (thin film)
3321, 3180, 1646, 1600 cm−1; mp 240−245 °C (dec); [α]D22 −32.2 (c
1.00, MeOH); HRMS (ESI) m/z [M + H]+ calcd for C25H33N2O2
393.2537; found 393.2540; LCMS tR = 3.23 min.
3β-Acetoxy-6α-hydroxy-5α-androst-16-en-17-yl Trifluorometha-
nesulfonate (9). A flame-dried 100 mL round-bottom flask was
charged with 2 (3.00 g, 6.49 mmol) and anhydrous THF (32 mL)
under Ar. The reaction mixture was cooled to 0 °C, and a solution of
BH3·THF (1 M in THF, 6.8 mL, 6.80 mmol, 1.05 equiv) was added
dropwise over 10 min. After stirring for 4 h at 0 °C, H2O (11 mL) was
added slowly dropwise. CAUTION: A single drop of H2O results in
rapid gas evolution! Then, NaBO3·4H2O (1.11 g, 7.18 mmol, 1.1
equiv) was added in one portion, and the reaction mixture was allowed
to warm slowly to rt. After stirring for 18 h, the reaction mixture was
filtered, and the precipitate was rinsed with small portions of THF (3
× 20 mL). The combined organic layers were washed with brine (60
mL), dried over Na2SO4, filtered, and concentrated. The crude
product was purified by chromatography (80 g of silica gel, 0−35%
EtOAc/hexanes) to afford 9 (0.871 g, 1.81 mmol, 28% yield) as a
white, foamy solid and recovered 2 (1.17 g, 2.52 mmol, 39%). An
analytical sample was prepared through recrystallization by dissolving
some of the material in Et2O, diluting with hexanes, and allowing slow
evaporation of the solvent mixture. Rf = 0.1 (25% EtOAc/hexanes);
1H NMR (400 MHz, CDCl3) δ 5.57 (dd, J = 3.3, 1.7 Hz, 1H), 4.73−
1
(minor diastereomer, diagnostic peaks only): H NMR (400 MHz,
CDCl3) δ 4.82−4.73 (m, 1H), 3.13 (d, J = 2.6 Hz, 1H), 2.03 (s, 3H),
1.05 (s, 3H), 1.00 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 170.7,
151.7, 150.0, 149.3, 136.7, 133.8, 63.6, 63.5, 62.8, 57.2, 51.5, 51.4, 47.4,
45.3, 38.1, 36.8, 35.4, 32.4, 31.9, 21.7, 17.2, 16.7.
5α,6α-Epoxy-17-(3-pyridyl)androst-16-en-3β-ol (6b). Following
general procedure B, 6a (49.8 mg, 0.122 mmol) was reacted with
K2CO3 (0.170 g, 1.23 mmol, 10.1 equiv) in MeOH (4.2 mL) for 17 h.
After workup, the crude product was purified by chromatography (4 g
of silica gel, 0−5% MeOH/CH2Cl2) to afford 6b (44.4 mg, 0.121
mmol, 99% yield) as a white, amorphous solid. Rf = 0.4 (5% MeOH/
1
CH2Cl2); H NMR (400 MHz, CDCl3) δ 8.58 (dd, J = 2.3, 0.9 Hz,
1H), 8.44 (dd, J = 4.9, 1.6 Hz, 1H), 7.61 (dt, J = 7.9, 2.0 Hz, 1H), 7.20
(ddd, J = 7.9, 4.8, 0.9 Hz, 1H), 5.96 (dd, J = 3.3, 1.8 Hz, 1H), 3.97−
3.87 (m, 1H), 2.95 (d, J = 4.5 Hz, 1H), 2.25 (ddd, J = 15.7, 6.5, 3.3 Hz,
1H), 2.10 (dd, J = 12.7, 11.4 Hz, 1H), 2.05−1.89 (complex, 4H),
1.73−1.58 (complex, 5H), 1.58−1.36 (complex, 6H), 1.36−1.29 (m,
1H), 1.12 (s, 3H), 0.97 (s, 3H); 13C NMR (101 MHz, CDCl3) δ
151.7, 148.0 (2C), 133.8, 133.0, 129.2, 123.1, 68.7, 66.0, 59.2, 57.5,
47.5, 42.8, 40.0, 35.2, 35.0, 32.4, 31.7, 31.2, 28.6, 28.5, 20.7, 16.6, 16.1;
21
IR (neat) 3389, 3180 cm−1; mp 239−242 °C; [α]D −55.6 (c 1.00,
CHCl3); HRMS (ESI) m/z [M + H]+ calcd for C24H32NO2 366.2428;
found 366.2427; LCMS tR = 3.81 min.
3β-Hydroxy-17-(3-pyridyl)androsta-5,16-dien-6-carbonitrile (7).
A 10 mL microwave vial was charged with a ca. 4:1 mixture of 6a
and 5β,6β-6a (0.251 g, 0.547 mmol), KCN (0.399 g, 6.13 mmol, 11.2
equiv), and MeOH (5.0 mL). CAUTION: KCN is highly toxic and
must be handled with care! The reaction mixture was stirred at 160 °C
for 50 min using a microwave reactor (time measured when the
reaction mixture reached the programmed temperature after a ramp
period of ca. 1 min). After cooling to rt, the reaction mixture was
concentrated. The residue was dissolved in CH2Cl2 (25 mL) and
washed with H2O (50 mL). The aqueous layer was extracted with
CH2Cl2 (2 × 25 mL), and the combined organic layers were
concentrated. The crude product was purified by chromatography
(150 g of C18 silica gel, H2O containing 0.1 vol % NH4OH/MeCN)
to afford 7 (0.145 g, 0.388 mmol, 63% yield) as an off-white,
amorphous solid. An analytical sample was prepared through
recrystallization by dissolving some of the material in MeOH, diluting
with EtOAc, and allowing slow evaporation of the solvent mixture. Rf
= 0.7 (10% MeOH/CH2Cl2); 1H NMR (400 MHz, CDCl3) δ 8.60 (s,
1H), 8.47 (d, J = 3.8 Hz, 1H), 7.63 (dt, J = 7.9, 1.8, 1H), 7.22 (dd, J =
7.9, 4.8 Hz, 1H), 6.00 (dd, J = 7.9, 4.8 Hz, 1H), 3.69−3.58 (m, 1H),
3.13 (ddd, J = 13.6, 4.9, 2.2 Hz, 1H), 2.39−2.24 (m, 3H), 2.15−2.01
(m, 3H), 2.00−1.88 (m, 3H), 1.82 (qd, J = 10.9, 5.0 Hz, 1H), 1.73−
1.54 (m, 4H), 1.49 (td, J = 12.2, 4.7 Hz, 1H), 1.20−1.06 (m, 2 H),
1.13 (s, 3H), 1.04 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 159.2,
151.5, 148.2, 148.0, 133.9, 132.8, 129.1, 123.2, 119.1, 106.5, 70.6, 56.9,
49.3, 47.3, 40.5, 38.4, 36.5, 35.1, 33.7, 31.7, 30.8, 29.9, 20.8, 19.7, 16.6;
4.61 (m, 1H), 3.44 (td, J = 10.5, 4.3 Hz, 1H), 2.28−2.16 (m, 2H),
2.05−1.96 (m, 2H), 2.03 (s, 3H), 1.91−1.79 (m, 1H), 1.77−1.54
(complex, 5H), 1.53−1.24 (complex, 5H), 1.16−1.04 (m, 2H), 1.01−
0.90 (m, 1H), 0.96 (s, 3H), 0.87 (s, 3H), 0.85−0.79 (m, 1H); 13C
NMR (101 MHz, CDCl3) δ 170.6, 159.0, 118.5 (q, J = 320 Hz), 114.4,
73.3, 69.1, 54.0, 53.8, 51.8, 44.8, 40.4, 36.7, 36.4, 32.5, 32.3, 28.5, 28.3,
27.1, 21.4, 20.4, 25.2, 13.2; 19F NMR (376 MHz, CDCl3) δ − 73.6 (s);
IR (neat) 3532, 1715, 1625 cm−1; mp 136−139 °C (Et2O−hexanes);
23
[α]D +25.4 (c 1.00, CHCl3); HRMS (APCI) m/z [M − CH3CO2]+
calcd for C22H32F3O6S 421.1655; found 421.1649.
6α-Hydroxy-17-(3-pyridyl)-5α-androst-16-en-3β-yl Acetate (10a).
Following general procedure A, 9 (0.676 g, 1.41 mmol) was coupled to
3-(diethylboranyl)pyridine (0.235 g, 1.60 mmol, 1.1 equiv) using
bis(triphenylphosphine)palladium(II) dichloride (0.100 g, 0.143
mmol, 0.1 equiv) in anhydrous THF (8.3 mL) and a saturated
solution of NaHCO3 (2.8 mL) for 23 h. After workup, the crude
product was purified by chromatography (24 g of silica gel, 0−50%
EtOAc/hexanes) to afford mostly pure 10a (0.465 g), which was
recrystallized from MeOH−H2O to give a white, crystalline solid
(0.282 g, 0.690 mmol, 49% yield). Rf = 0.2 (50% EtOAc/hexanes); 1H
NMR (400 MHz, CDCl3) δ 8.59 (d, J = 1.8 Hz, 1H), 8.44 (dd, J = 4.8,
I
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
1.5 Hz, 1H), 7.62 (dt, J = 7.9, 1.9 Hz, 1H), 7.20 (ddd, J = 7.9, 4.8, 0.7
Hz, 1H), 5.97 (dd, J = 3.2, 1.7 Hz, 1H), 4.73−4.63 (m, 1H), 3.51−
3.42 (m, 1H), 2.30−2.20 (m, 2H), 2.12−1.96 (m, 3H), 2.02 (s, 3H),
1.89−1.80 (m, 1H), 1.80−1.54 (complex, 5H), 1.53−1.27 (complex,
4H), 1.15−0.95 (m, 3H), 0.99 (s, 3H), 0.88 (s, 3H), 0.87−0.77 (m,
1H); 13C NMR (101 MHz, CDCl3) δ 170.7, 151.8, 148.02, 147.99,
133.8, 133.0, 129.2, 123.2, 73.6, 69.4, 57.2, 54.1, 52.0, 47.6, 41.6, 37.0,
36.6, 35.3, 33.0, 31.8, 28.5, 27.3, 21.5, 21.2, 16.8, 13.5; IR (neat) 3216,
Hz, 1H), 7.20 (dd, J = 7.9, 4.8 Hz, 1H), 5.95 (dd, J = 3.3, 1.7 Hz, 1H),
4.70−4.58 (m, 1H), 2.43−2.31 (m, 1H), 2.28 (dd, J = 12.7, 3.0 Hz,
1H), 2.19 (ddd, J = 15.8, 6.7, 3.4 Hz, 1H), 2.13−1.97 (complex, 4H),
2.00 (s, 3H), 1.96−1.89 (m, 1H), 1.88−1.69 (complex, 4H), 1.62−
1.20 (complex, 6H), 0.98 (s, 3H), 0.80 (s, 3H); 13C NMR (101 MHz,
CDCl3) δ 209.9, 170.7, 151.6, 148.2, 148.0, 133.8, 132.7, 128.9, 123.2,
72.8, 57.5, 56.8, 54.1, 47.9, 46.5, 41.1, 36.6, 36.3, 35.0, 31.6, 26.9, 26.6,
21.5, 21.4, 16.7, 13.2; IR (thin film) 1729, 1710 cm−1; mp 159−164 °C
22
1727 cm−1; mp 211−214 °C (MeOH−H2O); [α]D +46.2 (c 1.00,
23
(EtOAc−hexanes); [α]D −6.2 (c 1.00, CHCl3); HRMS (ESI) m/z
CHCl3); HRMS (ESI) m/z [M + H]+ calcd for C26H36NO3 410.2690;
found 410.2693.
[M + H]+ calcd for C26H34NO3 408.2531; found 408.2533.
3β-Hydroxy-17-(3-pyridyl)-5α-androst-16-en-6-one (12b). Fol-
lowing general procedure B, 12a (45.9 mg, 0.113 mmol) was reacted
with K2CO3 (54.9 mg, 0.397 mmol, 3.5 equiv) in MeOH (3.7 mL) for
17 h. After workup, the crude product was purified by chromatography
(4 g of silica gel, 0−5% MeOH/CH2Cl2) to afford 12b (36.3 mg, 99.3
μmol, 88% yield) as a white, amorphous solid. Rf = 0.4 (5% MeOH/
CH2Cl2); 1H NMR (400 MHz, CDCl3) δ 8.61 (s, 1H), 8.47 (d, J = 4.1
Hz, 1H), 7.63 (dt, J = 7.9, 2.0 Hz, 1H), 7.22 (dd, J = 7.8, 4.9 Hz, 1H),
5.98 (dd, J = 3.3, 1.7 Hz, 1H), 3.64−3.54 (m, 1H), 2.47−2.35 (m,
1H), 2.29−2.17 (m, 2H), 2.15−2.00 (complex, 4H), 1.97−1.73
(complex, 5H), 1.71−1.62 (m, 1H), 1.59−1.34 (complex, 5H), 1.32−
1.22 (m, 1H), 1.01 (s, 3H), 0.82 (s, 3H); 13C NMR (101 MHz,
CDCl3) δ 210.4, 151.6, 148.2, 148.0, 133.8, 132.8, 129.0, 123.2, 70.7,
57.6, 57.2, 54.3, 48.0, 46.6, 41.2, 36.6, 36.6, 35.1, 31.6, 30.8, 30.2, 21.6,
17-(3-Pyridyl)-5α-androst-16-en-3β,6α-diol (10b). Following gen-
eral procedure B, 10a (60.4 mg, 0.147 mmol) was reacted with K2CO3
(0.203 g, 1.47 mmol, 9.9 equiv) in MeOH (4.9 mL) for 15 h. After
workup, the crude product was purified by chromatography (4 g of
silica gel, 0−10% MeOH/CH2Cl2) to afford 10b (52.4 mg, 0.143
mmol, 97% yield) as a white, amorphous solid. An analytical sample
was prepared by recrystallization from MeOH−H2O. Rf = 0.6 (10%
1
MeOH/CH2Cl2); H NMR (400 MHz, DMSO-d6) δ 8.58 (s, 1H),
8.43 (d, J = 3.9 Hz, 1H), 7.77−7.71 (m, 1H), 7.33 (dd, J = 7.8, 4.8 Hz,
1H), 6.11−6.07 (m, 1H), 4.44 (d, J = 4.5 Hz, 1H), 4.31 (d, J = 5.6 Hz,
1H), 3.29−3.11 (m, 2H), 2.18 (ddd, J = 15.7, 6.3, 3.2 Hz, 1H), 2.12−
1.94 (m, 3H), 1.89 (dt, J = 12.2, 4.1 Hz, 1H), 1.70−1.46 (complex,
5H), 1.43−1.20 (m, 3H), 1.03−0.82 (complex, 4H), 0.97 (s, 3H), 0.78
(s, 3H), 0.75−0.66 (m, 1H); 13C NMR (101 MHz, DMSO-d6) δ
151.1, 147.8, 147.1, 133.3, 132.1, 128.9, 123.3, 69.6, 67.3, 56.9, 53.7,
51.6, 46.9, 41.5, 37.0, 35.9, 34.7, 32.4, 32.3, 31.3, 31.2, 20.7, 16.4, 13.2;
22
16.8, 13.4; IR (thin film) 3224, 1702 cm−1; mp 207−212 °C; [α]D
−9.0 (c 1.00, CHCl3); HRMS (ESI) m/z [M + H]+ calcd for
C24H32NO2 366.2476; found 366.2422; LCMS: tR = 3.54 min.
6-Hydroxyimino-17-(3-pyridyl)-5α-androst-16-en-3β-yl Acetate
(13a). Following general procedure C, 12a (2.10 g, 5.15 mmol) was
reacted with NH2OH·HCl (2.50 g, 36.1 mmol, 7.0 equiv) and NaOAc
(2.96 g, 36.1 mmol, 7.0 equiv) in EtOH (170 mL) and H2O (1.35 mL)
for 24 h. After workup, the crude product was purified by
chromatography (80 g of silica gel, 0−5% MeOH/CH2Cl2) to afford
13a (2.13 g, 5.04 mmol, 98% yield) as a white, amorphous solid. An
analytical sample was prepared by recrystallization from MeOH−H2O.
22
IR (neat) 3451, 3192 cm−1; mp 225−229 °C (MeOH−H2O); [α]D
+54 (c 0.50, MeOH); HRMS (ESI) m/z [M + H]+ calcd for
C24H34NO2 368.2584; found 368.2583; LCMS tR = 3.45 min.
3β-Acetoxy-6-oxo-5α-androst-16-en-17-yl Trifluoromethanesul-
fonate (11). A 100 mL round-bottom flask was charged with 9
(3.90 g, 8.12 mmol) and anhydrous CH2Cl2 (81 mL). The reaction
mixture was cooled to 0 °C, and Dess−Martin periodinane (6.89 g,
16.2 mmol, 2.0 equiv) was added in one portion. Then, the reaction
mixture was allowed to warm slowly to rt. After stirring for 16 h, a
saturated solution of Na2S2O3 (150 mL) was added, and the reaction
mixture was stirred for 15 min. The aqueous layer was extracted with
CH2Cl2 (2 × 75 mL). The combined organic layers were washed with
a saturated solution of NaHCO3 (200 mL), dried over MgSO4,
filtered, and concentrated. The crude product was purified by
chromatography (80 g of silica gel, 0−25% EtOAc/hexanes) to afford
11 (3.46 g, 7.24 mmol, 89% yield) as a white, amorphous solid. An
analytical sample was prepared through recrystallization by dissolving
some of the material in Et2O, diluting with hexanes, and allowing slow
evaporation of the solvent mixture. Rf = 0.5 (25% EtOAc/hexanes);
1H NMR (400 MHz, CDCl3) δ 5.58 (dd, J = 3.1, 1.5 Hz, 1H), 4.67 (tt,
1
Rf = 0.5 (5% MeOH/CH2Cl2); H NMR (400 MHz, CDCl3) δ 8.61
(d, J = 1.7 Hz, 1H), 8.47 (dd, J = 4.8, 1.4 Hz, 1H), 8.08 (s, 1H), 7.64
(dt, J = 8.0, 1.7 Hz, 1H), 7.22 (dd, J = 7.7, 4.9 Hz, 1H), 5.99 (dd, J =
3.2, 1.7 Hz, 1H), 4.75−4.64 (m, 1H), 3.43 (dd, J = 13.6, 4.3 Hz, 1H),
2.32 (ddd, J = 15.7, 6.4, 3.3 Hz, 1H), 2.20−2.11 (m, 1H), 2.10−1.97
(m, 3H), 2.03 (s, 3H), 1.91−1.69 (complex, 5H), 1.66−1.43
(complex, 4H), 1.42−1.32 (m, 1H), 1.30−1.17 (m, 1H), 1.14−1.06
(m, 1H), 1.01 (s, 3H), 0.84 (s, 3H); 13C NMR (101 MHz, CDCl3) δ
170.8, 159.4, 151.6, 148.1, 148.0, 133.9, 132.9, 129.2, 123.2, 73.4, 57.6,
54.6, 49.7, 47.9, 39.1, 35.9, 35.2, 34.4, 31.8, 29.5, 27.8, 27.2, 21.6, 21.5,
16.9, 12.7; IR (neat) 1727 cm−1; mp 216−220 °C (MeOH−H2O,
22
dec); [α]D −40.0 (c 1.00, CHCl3); HRMS (ESI) m/z [M + H]+
J = 11.5, 4.7 Hz, 1H), 2.37−2.32 (m, 1H), 2.29 (dd, J = 12.7, 3.1 Hz,
1H), 2.20 (ddd, J = 14.9, 6.4, 3.3 Hz, 1H), 2.10−1.92 (complex, 4H),
2.03 (s, 3H), 1.90−1.73 (complex, 5H), 1.61−1.37 (complex, 5H),
1.31 (td, J = 13.5, 4.0 Hz, 1H), 0.97 (s, 3H), 0.80 (s, 3H); 13C NMR
(101 MHz, CDCl3) δ 209.2, 170.7, 158.9, 118.7 (q, J = 321 Hz), 114.6,
72.7, 56.9, 54.3, 54.2, 45.4, 45.3, 41.1, 36.3, 35.9, 32.4, 28.5, 26.9, 26.2,
21.5, 21.0, 15.4, 13.2; 19F NMR (376 MHz, CDCl3) δ −73.6 (s); IR
(thin film) 1713, 1627 cm−1; mp 162−165 °C (Et2O−hexanes);
calcd for C26H35N2O3 423.2642; Found 423.2633.
3β-Hydroxy-17-(3-pyridyl)-5α-androst-16-en-6-one Oxime (13b).
Following general procedure C, 12b (25.0 mg, 68.0 μmol) was reacted
with NH2OH·HCl (33.6 mg, 0.484 mmol, 7.1 equiv) and NaOAc
(39.7 mg, 0.484 mmol, 7.1 equiv) in EtOH (2.3 mL) and H2O (20
μL) for 17 h. After workup, the crude product was purified by
chromatography (4 g of silica gel, 0−5% MeOH/CH2Cl2) to afford
13b (21.5 mg, 56.5 μmol, 83% yield) as a white, amorphous solid. Rf =
23
[α]D −9.9 (c 1.00, CHCl3); HRMS (APCI) m/z [M − CH3CO2]+
1
0.2 (5% MeOH/CH2Cl2); H NMR (400 MHz, DMSO-d6) δ 10.39
calcd for C22H30F3O6S 419.1498; found 419.1492.
(s, 1H), 8.58 (d, J = 4.3 Hz, 1H), 7.75 (d, J = 8.0 Hz, 1H), 7.33 (dd, J
= 7.3, 5.3 Hz, 1H), 6.10 (s, 1H), 4.50 (d, J = 4.1 Hz, 1H), 3.25 (dd, J =
13.1, 3.0 Hz, 1H), 2.23 (d, J = 16.2 Hz, 1H), 2.12−1.88 (m, 3H),
1.82−1.72 (m, 1H), 1.70−1.58 (complex, 5H), 1.46−1.16 (complex,
6H), 1.14−1.01 (m, 2H), 0.96 (s, 3H), 0.69 (s, 3H); 13C NMR (101
MHz, DMSO-d6) δ 156.9, 151.0, 147.8, 147.2, 133.3, 132.0, 128.9,
123.4, 69.3, 56.9, 53.6, 49.0, 47.1, 38.2, 35.6, 34.5, 33.7, 31.9, 31.2,
30.9, 28.8, 21.0, 16.3, 12.3; IR (neat) 3277 cm−1; mp 251−256 °C
(dec); [α]D23 −31 (c 0.50, 1,4-dioxane); HRMS (ESI) m/z [M + H]+
calcd for C24H32N2O2 381.2537; found 381.2534; LCMS tR = 3.54
min.
17-(3-Pyridyl)-6-oxo-5α-androst-16-en-3β-yl Acetate (12a). Fol-
lowing general procedure A, 11 (3.20 g, 6.69 mmol) was coupled to 3-
(diethylboranyl)pyridine (1.11 g, 7.57 mmol, 1.1 equiv) using
bis(triphenylphosphine)palladium(II) dichloride (0.471 g, 0.671
mmol, 0.1 equiv) in anhydrous THF (40 mL) and a saturated
solution of NaHCO3 (13.3 mL) for 17 h. After workup, the crude
product was purified by chromatography (80 g of silica gel, 0−50%
EtOAc/hexanes) to afford 12a (2.25 g, 5.51 mmol, 82% yield) as a
white, amorphous solid. An analytical sample was prepared through
recrystallization by dissolving some of the material in EtOAc, diluting
with hexanes, and allowing slow evaporation of the solvent mixture. Rf
= 0.3 (50% EtOAc/hexanes); 1H NMR (400 MHz, CDCl3) δ 8.58 (d,
J = 2.2 Hz, 1H), 8.44 (dd, J = 4.8, 1.5 Hz, 1H), 7.61 (dt, J = 7.9, 2.0
6-Aza-7-oxo-17-(3-pyridyl)-B-homo-5α-androst-16-en-3β-yl Ace-
tate (14a). A flame-dried 20 mL vial was charged with 13b (0.101 g,
0.239 mmol) and anhydrous THF (5 mL) under Ar. The reaction
J
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
mixture was cooled to 0 °C, and SOCl2 (60.0 μL, 8.22 mmol, 35.0
equiv) was added dropwise over 5 min. After stirring for 4 h at 0 °C,
the reaction mixture was quenched by slow addition of H2O (2 mL).
The reaction mixture was allowed to warm to rt, neutralized with a
saturated solution of NaHCO3 (25 mL), and diluted with CH2Cl2 (10
mL). The organic layer was separated, and the aqueous layer was
extracted with CH2Cl2 (3 × 5 mL). The combined organic layers were
dried over MgSO4, filtered, and concentrated. The crude product was
purified by chromatography (12 g of silica gel, 0−2.5% MeOH/
CH2Cl2) to afford 14a (44.7 mg, 0.106 mmol, 45% yield) as a white,
codon optimization for expression in E. coli and modified to omit the
single, N-terminal transmembrane helix, and a C-terminal histidine tag
was added. Specifically, the 17A1Δ19H construct is as described.16
The CYP21A2dH construct omitted the N-terminal 2−19 residues,
substituted the sequence coding for 20-NWWKLRSLH-28 with
nucleotides coding for AKKTSSKGK, and added codons for a C-
terminal four histidine tag immediately prior to the stop codon. Each
cytochrome P450 gene was inserted into the pCWori+ vector, which
confers carbenicillin resistance. A synthetic, codon-optimized gene of
human full-length NADPH-cytochrome P450 reductase in pET-
29a(+) is as described.31
1
amorphous solid. Rf = 0.6 (10% MeOH/CH2Cl2); H NMR (400
MHz, CDCl3) δ 8.59 (s, 1H), 8.47−8.39 (m, 1H), 7.61 (d, J = 7.9 Hz,
1H), 7.21 (dd, J = 7.9, 4.8 Hz, 1H), 5.99−5.93 (m, 1H), 5.46 (s, 1H),
4.70−4.59 (m, 1H), 3.46 (dt, J = 12.2, 5.1 Hz, 1H), 2.40−2.30 (m,
3H), 2.19 (ddd, J = 15.8, 11.5, 1.8 Hz, 1H), 2.08−1.95 (complex, 6H),
1.95−1.79 (m, 3H), 1.69 (td, J = 11.5, 6.6 Hz, 1H), 1.57−1.32
(complex, 4H), 1.18−1.04 (m, 2H), 1.01 (s, 3H), 0.91 (s, 3H); 13C
NMR (101 MHz, CDCl3) δ 176.0, 170.5, 151.5, 148.2, 147.9, 133.8,
132.6, 128.6, 123.2, 70.9, 59.0, 57.0, 56.7, 47.2, 40.6, 39.0, 35.5, 35.4,
34.7, 33.1, 33.0, 27.1, 22.9, 21.4, 16.5, 12.5; IR (neat) 3231, 1722, 1666
Expression and Purification of CYP17A1, CYP21A2, and
NADPH-cytochrome P450 Reductase Enzymes. E. coli DH5α
cells already harboring the pGro7 plasmid for chaperone expression
(chloramphenicol resistance) were transformed with either the
pCW17A1Δ19H or pCW21A2dH plasmids and grown on lysogeny
broth (LB) agar plates containing chloramphenicol (20 μg/mL) and
carbenicillin (100 μg/mL). A single colony was first grown in 5 mL of
LB media at 37 °C, 250 rpm in the presence of antibiotics until O.D.600
= 0.6. Then 100 μL of this primary culture was inoculated into 20 mL
of LB media and grown at 37 °C, 250 rpm for ca. 12 h, after which 20
mL was used to inoculate 1 L of Terrific Broth (TB) media in a 2.8 L
Fernbach flask. This culture was then grown at 37 °C, 250 rpm until
O.D.600 = 0.5. At this point, chaperone expression was induced by the
addition of L-arabinose (1 g). The heme precursor δ-aminolevulinic
acid (132 mg) also added. The temperature was then reduced to 27
°C, and the culture was allowed to continue growing until the O.D.600
reached 0.8. At this point, the expression of P450 protein was induced
by the addition of 0.8 mM IPTG. Temperature and shaking were
reduced to 25 °C and 200 rpm, respectively. Cultures were harvested
after 48 h, and cells were collected by centrifugation at 8000g for 10
min. Cells were resuspended in 25 mL of buffer (100 mM potassium
phosphate, pH 7.4, 300 mM NaCl, and 20% (v/v) glycerol for
CYP17A1; 100 mM potassium phosphate, pH 6.8, 300 mM NaCl, and
20% glycerol for CYP21A2) per liter of E. coli culture and stored at
−80 °C until purification.
22
cm−1; mp 248−252 °C; [α]D +49.7 (c 1.00, CHCl3); HRMS (ESI)
m/z [M + H]+ Calcd for C26H35N2O3 423.2642; Found 423.2633.
6-Aza-7-oxo-17-(3-pyridyl)-B-homo-5α-androst-16-en-3β-ol
(14b). Following general procedure B, 14a (47.1 mg, 0.111 mmol) was
reacted with K2CO3 (0.154 g, 1.13 mmol, 10.1 equiv) in MeOH (3.7
mL) for 17 h. After workup, the crude product was purified by
chromatography (4 g of silica gel, 0−5% MeOH/CH2Cl2) to afford
14b (42.4 mg, 0.111 mmol, 100%) as a white, amorphous solid. Rf =
1
0.5 (5% MeOH/CH2Cl2); H NMR (400 MHz, CDCl3) δ 8.59 (s,
1H), 8.46 (s, 1H), 7.64 (dt, J = 8.0, 1.9 Hz, 1H), 7.23 (dd, J = 8.0, 4.8
Hz, 1H), 5.98 (dd, J = 3.4, 1.8 Hz, 1H), 5.34 (d, J = 5.3 Hz, 1H),
3.66−3.56 (m, 1H), 3.44−3.35 (m, 1H), 2.37 (d, J = 6.9 Hz, 1H), 2.33
(dd, J = 6.6, 3.4 Hz, 1H), 2.25−2.07 (m, 2H), 2.07−1.89 (complex,
4H), 1.88−1.76 (m, 2H), 1.68 (td, J = 11.5, 6.6 Hz, 1H), 1.59−1.17
(complex, 5H), 1.13−1.04 (m, 2H), 1.02 (s, 3H), 0.91 (s, 3H); 13C
NMR (101 MHz, CDCl3) δ 176.0, 151.5, 147.9, 147.6, 134.1, 132.8,
128.8, 123.3, 68.8, 59.1, 57.3, 56.8, 47.2, 40.7, 39.0, 38.8, 35.7, 35.6,
33.2, 33.0, 31.0, 22.9, 16.6, 12.5; IR (thin film) 3234, 1648 cm−1; mp
CYP17A1 protein was purified as described14 with the following
modifications. A French press was used for cell disruption, and protein
was extracted from the membrane using 2% (v/v) Emulgen-913
(Desert Biologicals) for 60 min. Cell lysate was centrifuged at
100,000g for 1 h to remove membranes, and the resulting P450-
containing supernatant was loaded on a 30 mL Ni-NTA Agarose
(Qiagen) column pre-equilibrated with Ni buffer (100 mM potassium
phosphate, pH 7.4, 300 mM NaCl, 20% (v/v) glycerol, and 0.2% (v/v)
Emulgen-913). This column was subsequently washed with 3 column
volumes (CV) of Ni buffer, followed by 6 CV of Ni buffer
supplemented with 100 mM glycine, and eluted using 4 CV of Ni
buffer supplemented with 100 mM glycine and 80 mM histidine.
Elution fractions were pooled based on absorbance at 417 nm, diluted
5−6-fold in CM buffer (5 mM potassium phosphate, pH 7.4, 20% (v/
v) glycerol, and 100 mM glycine), supplemented with 0.2% (v/v)
Emulgen-913, and loaded onto a 5 mL carboxymethyl sepharose fast-
flow column (GE Healthcare) previously equilibrated with CM buffer.
The column was washed with 20 CV of CM buffer and eluted with
CM elution buffer (50 mM potassium phosphate, pH 7.4, 500 mM
NaCl, 100 mM glycine, and 20% (v/v) glycerol). Fractions were
pooled based on absorbance at 417 nm, concentrated to 4 mL, and
loaded on a size exclusion chromatography column (HiLoad 16/60
Superdex 200) equilibrated with buffer CM elution buffer.
Purification of CYP21A2 involved cell disruption by French press,
followed by membrane protein extraction with 2% (v/v) Emulgen-913
(Desert Biologicals) for 90 min, and ultracentrifugation (100,000g for
1 h) to pellet the membranes. The resulting supernatant containing
CYP21A2 was loaded onto 30 mL of Ni-NTA Agarose resin pre-
equilibrated with Ni buffer (100 mM potassium phosphate, pH 6.8,
300 mM NaCl, 20% (v/v) glycerol, 1 mM BME, and 0.2% (v/v)
Emulgen-913). The column was subsequently washed with 3 CV of Ni
buffer, followed by 6 CV of Ni buffer supplemented with 10 mM
histidine, and eluted using 4 CV of Ni buffer supplemented with 80
mM histidine. Elution fractions were pooled based on absorbance at
22
271−274 °C (dec); [α]D +45 (c 0.95, CHCl3); HRMS (ESI) m/z
[M + H]+ calcd for C24H33N2O2 381.2537; found 381.2544; LCMS tR
= 2.90 min.
17-Oxoandrost-5-en-3β-yl Acetate (15)..28−30 A 20 mL microwave
vial was charged with 17-oxoandrost-5-en-3β-ol (10.0 g, 34.7 mmol),
p-TsOH·H2O (67.7 mg, 0.356 mmol, 0.01 equiv), and Ac2O (13.0 mL,
111 mmol, 4.0 equiv). The suspension was stirred at 80 °C for 12 min
using a microwave reactor (time measured when the reaction mixture
reached the programmed temperature after a ramp period of ca. 1
min). After cooling to rt, the resulting wet solid was dissolved in
CH2Cl2 (200 mL), and a saturated solution of NaHCO3 (200 mL) was
added slowly. The mixture was stirred for 15 h to hydrolyze any
remaining Ac2O. After the organic layer was separated, the aqueous
layer was extracted with CH2Cl2 (3 × 50 mL). The combined organic
layers were washed with brine (100 mL), dried over MgSO4, filtered,
and concentrated to afford 15 (11.3 g, 34.2 mmol, 99% yield) as a
white, amorphous solid, which was used without further purification.
1
Rf = 0.6 (25% EtOAc/hexanes); H NMR (400 MHz, CDCl3) δ 5.40
(d, J = 5.2 Hz, 1H), 4.65−4.55 (m, 1H), 2.46 (dd, J = 19.2, 8.8 Hz,
1H), 2.39−2.25 (m, 2H), 2.16−2.05 (m, 2H), 2.03 (s, 3H), 1.99−1.81
(complex, 4H), 1.72−1.63 (m, 3H), 1.63−1.42 (m, 3H), 1.34−1.24
(m, 2H), 1.20−1.09 (m, 1H), 1.04 (s, 3H), 1.03−0.98 (m, 1H), 0.88
(s, 3H); 13C NMR (101 MHz, CDCl3) δ 221.2, 170.6, 140.1, 122.0,
73.9, 51.8, 50.3, 47.7, 38.2, 37.1, 36.9, 36.0, 31.6, 31.5, 30.9, 27.8, 22.0,
21.6, 20.5, 19.5, 13.7; IR (neat) 1723, 1737 cm−1; mp 169−171 °C,
29
lit.28 167−170 °C; [α]D −4.2 (c 1.00, CHCl3), lit. [α]D −3.1 (c
1.00, CHCl3); HRMS (APCI) m/z [M − CH3CO2]+ calcd for
C19H27O 271.2056; found 271.2054.
23
18
Plasmid Constructs. Recombinantly expressed human cyto-
chrome P450 enzymes CYP17A1, CYP21A2, and human full-length
NADPH cytochrome P450 reductase enzymes were used in this study.
The gene for each P450 protein was constructed synthetically with
K
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
417 nm, diluted 5−6-fold in CM buffer (10 mM potassium phosphate,
pH 6.8, 20% (v/v) glycerol, and 1 mM BME), supplemented with
0.2% (v/v) Emulgen-913, and loaded onto a 5 mL carboxymethyl
sepharose fast-flow column previously equilibrated with CM buffer.
The column was washed with 20 CV of CM buffer and eluted with
CM elution buffer (50 mM potassium phosphate, pH 6.8, 350 mM
NaCl, and 20% (v/v) glycerol). Fractions were pooled based on
absorbance at 417 nm, concentrated to 4 mL, and loaded on a size
exclusion chromatography column equilibrated with buffer CM elution
buffer.
The purity, quality, and quantity of P450 proteins were assessed by
the UV−vis spectrum, SDS-PAGE, and the reduced carbon monoxide
difference spectrum. Concentration of P450 enzyme was determined
from the reduced carbon monoxide difference spectra in the presence
of 2 μM progesterone as a ligand.
Human full-length NADPH-cytochrome P450 reductase (CPR)
enzyme was expressed and purified as described.31 Final reductase
samples were evaluated on SDS-PAGE and by UV−visible spectros-
copy. CPR was quantitated by flavin absorbance of the fully oxidized
protein at 454 nm.
Inhibition of Progesterone Hydroxylation. Abiraterone analogs
were evaluated for their selectivity in inhibiting progesterone
hydroxylation by CYP17A1 (17α-hydroxylation) vs CYP21A2 (21-
hydroxylation). Initial stocks (10 mM) were made of each compound
in DMSO. This stock was then serially diluted with DMSO to yield a
concentration range of 0.3125 nM to 40.960 μM. Compound 3b was
further diluted to 0.0195 nM. Abiraterone itself was used as a reference
inhibitor to assess structural changes altering selectivity in inhibition of
CYP17A1 vs CYP21A2.
Progesterone hydroxylation was evaluated at the Km for each
enzyme under conditions where substrate utilization was <3%.
Progesterone 17α-hydroxylation was evaluated by incubating
CYP17A1 (20 pmol) with CPR (80 pmol) for 20 min at rt before
combination with reaction buffer (50 mM Tris-HCl, pH 7.4, 5 mM
MgCl2) containing 6 μM progesterone. Equal volumes of serial diluted
inhibitor stocks were added to yield a range of inhibitor concentrations
with a constant final DMSO concentration (0.4%) across all reactions.
Samples containing enzyme, substrate, and inhibitor were incubated
for 3 min in a 37 °C water bath, and reactions were initiated by
addition of 1 mM NADPH. Total reaction volume was 500 μL.
Reactions were continued at 37 °C for 10 min and subsequently
stopped by adding 200 mL of trichloroacetic acid (20%). Precipitated
protein was cleared by centrifugation. Progesterone 21-hydroxylase
inhibition assays were accomplished similarly using CYP21A2 (1
pmol) and CPR (4 pmol) with 0.3 μM progesterone substrate and an
8 min reaction time.
All samples were processed using the following protocol to
quantitate hydroxylated progesterone by LC−MS/MS. Portions of
the progesterone 17α-hydroxylation reaction samples (20 μL) and 21-
hydroxylation reaction samples (100 μL) were mixed with 200 μL of
HPLC-grade water in a sample vial, respectively. Deuterated
progesterone-d9 (CDN isotopes) was then added as an internal
standard (12.5 pg). Samples were then extracted by adding 1 mL of
MTBE, vortexing for 4 min, and permitting phase separation for 5 min.
The bottom aqueous layer was removed by pipetting and the upper
organic layer dried using a speedvac. Finally, dried samples were
reconstituted with MeOH/H2O (1:1), and 10 μL was injected on a
Kinetex 50 × 2.1 mm, 3 μm particle size biphenyl column
(Phenomenex) running on an Agilent 1260/1290 binary pump
HPLC. Components were resolved using a gradient elution of
ammonium fluoride (0.2 mM) in H2O (mobile phase A) and 0.2 mM
ammonium fluoride in MeOH (mobile phase B) with 60−100%
mobile phase B gradient. Analytes were further directed into an Agilent
6495 triple quadrupole mass spectrometer using electrospray
ionization in positive ionization mode and analyzed using multiple
reaction monitoring mode.32 Mass transitions 315.2 > 97.0 and 324.1
> 100.1 were used to monitor and quantify progesterone and the
progesterone-d9 internal standard, respectively, while 17α-hydrox-
yprogesterone and 21-hydroxyprogesterone were monitored and
quantified using the 331.2 > 109 mass transition. The peak areas for
specific analytes were determined using the Agilent MassHunter
quantitative analysis program.
Inhibitory concentration (IC50) for compounds were determined
from two independent replicates (n = 2). Raw data was converted into
% maximal activity and fitted with nonlinear regression (inhibitor vs
response) using eq 1 and automatic outlier elimination (ROUT
coefficient, Q = 5%)33 in Prism (GraphPad Software, La Jolla, CA) and
is presented as mean standard error.
(max − min)
Y = min +
1 + 10(logIC50−x)Hill slope
(1)
CYP3A4 and CYP2D6 Inhibition Assays. Abiraterone and
analog 13b (5α-H, 6-oxime) were evaluated for inhibition of
CYP3A4-mediated metabolism of luciferin isopropyl alcohol (Prom-
ega) and CYP2D6-mediated dextromethorphan metabolism, at
substrate concentrations equivalent to their respective Km values.
Like CYP17A1 and CYP21A2, CYP3A4 and CYP2D6 were the
recombinant, N-terminally truncated and C-terminally His-tagged
enzymes that had been highly purified.31 For CYP3A4, 10 pmol of
CYP3A4, 40 pmol of CPR, and 10 pmol of cytochrome b5 were
combined in 100 mM potassium phosphate, pH 7.4. Inhibitors (0.04−
0.64 μM, in EtOH or DMSO) were added, with a final solvent
concentration of 0.4% in all reactions. A 25 μL aliquot of the protein,
substrate, and inhibitor mixture was added into each well of flat-
bottom 96-well plate (Corning Inc.). Using the GloMax Discover plate
reader (Promega), plates were incubated for 3 min at 37 °C, followed
by the addition of 25 μL of NADPH (200 μM) with shaking at 37 °C
for 2 s after each addition to initiate the reaction. Reactions continued
for 30 min at 37 °C and then were terminated, and product was
converted to its luminescent form by the addition of esterase-
containing luciferin detection reagent (50 μL). Finally, the plate was
incubated at 37 °C for 20 min before luminescence was read. Each
data point was generated in triplicate. To evaluate CYP2D6 inhibition,
an assay was performed as described for dextromethorphan conversion
to dextrophan without cytochrome b5,31 employing only minor
changes. Inhibitor stocks (0.020−80 μM) were prepared as described
above for the CYP3A4 assay. Substrate and product were separated
using an isocratic flow of 40% mobile phase A, consisting of 10 mM
potassium phosphate at pH 3.5, and 60% mobile phase B, consisting of
50% MeCN/100% MeOH in a 250/200 v/v ratio. Each sample was
prepared in duplicate. Data from both assays was fit using Prism to
determine the IC50 values as described for the progesterone
hydroxylation assay.
X-ray Crystallography. Several analogs were cocrystallized with
human CYP17A1, and structures were determined using previously
reported methods.16 Briefly, CYP17A1 in 50 mM Tris-HCl, pH 7.4,
20% glycerol, and 100 mM glycine was exchanged with the same buffer
containing 10 μM of the desired inhibitor multiple times. The
saturated CYP17A1 was then concentrated such that the final stock
was ca. 30 mg/mL and 0.5% Emulgen 913. This protein/ligand
complex (1 μL) was mixed 1:1 with precipitant solution consisting of
0.175 M Tris, pH 8.5 containing 30% PEG 3350, 3% glycerol, and
0.250−0.275 M lithium sulfate for the C6 oxime. For the C6 nitrile
and C6 amide analogs, PEG was increased to 35%. Using hanging drop
vapor diffusion, these drops were equilibrated against 750 μL of the
same precipitant solution at 20 °C (C6 oxime) or 4 °C (C6 nitrile and
C6 amide). Rod-shaped crystals grew in ca. 24 h and were
cryoprotected using a 7:3 mixture of precipitant and 80% glycerol.
X-ray diffraction data collected at Stanford Synchrotron Radiation
Lightsource on beamline 12−2 was processed using XDS34 and
Scala.35 Structures were solved by molecular replacement using a
search model of CYP17A1 with galeterone (PDB 3SWZ)16 and
Phaser36 in Phenix.37 Model building and refinement were performed
with COOT38 and phenix.refine,37 respectively. The coordinates for
the three new X-ray crystallographic structures with compound 7, 8,
and 13b have been deposited to the Protein Data Bank under the
identifiers 6CIZ, 6CHI, and 6CIR, respectively. Data collection and
refinement statistics are provided in Table S1 (Supporting
L
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Information). All protein structure figures were generated with
Pymol.39
Computational Docking. Test compounds, with the exception of
3b and 14b, were each docked onto the crystal structures of CYP17A1
and CYP21A2 to analyze the potential binding mode, binding energy,
and favorable or unfavorable interactions. All docking procedures were
the Protein Data Bank under the identifiers 6CIZ, 6CHI, and
6CIR, respectively.
AUTHOR INFORMATION
■
Corresponding Authors
*E-mail: jaube@unc.edu.
*E-mail: scottee@umich.edu.
performed using various modules of Schrodinger Maestro suite
̈
(Schrodinger, LLC, New York, NY, 2017). The X-ray crystal structures
̈
of CYP17A1 with abiraterone (PDB 3RUK) and CYP21A2 with
progesterone (PDB 4Y8W) were retrieved from RCSB PDB database,
and atoms of chain A including heme and the associated ligand were
preserved for docking. The polypeptide structures of CYP17A1 and
CYP21A2 were optimized and prepared for docking using the Maestro
Protein Preparation Wizard to assess bond order and missing
hydrogens, followed by energy minimization using the OPLS3 force
field. Gaps in the protein structures were not corrected as they were far
from the active site. Water molecules were removed from the protein.
The heme iron was assigned an oxidation state of +3. The Maestro
Receptor Grid Generation module was then used to define a 15 × 15
× 15 Å grid from the center of the cocrystallized steroid ligands.
Ligands used in the computational docking study were built by
modifying the abiraterone structure (PDB 3RUK) using the Maestro
3D Build module. The Maestro LigPrep module was then used to
generate conformers of each compound subjected to energy
minimization using the OPLS3 force field protocol. The resulting
compounds were docked into the prepared CYP17A1 and CYP21A2
structures using the Maestro Glide module. Docking was performed
with standard precision with flexible ligand sampling. Ligands were
constrained to within 10 Å of the iron. A total of 5000 initial poses
were generated for each compound. Based on the pose score, the top
400 were selected and subjected to energy minimization using the
OPLS3 force field. Finally, the top ten poses per compound were
generated and ranked according to Glide score, which is an
approximation of binding energy defined by receptor−ligand complex
energies. The top pose was analyzed and presented in Results and
Discussion.40 In addition, the total free energy of binding ΔGbind
(kcal/mol) was determined for these poses using prime molecular
mechanics energies combined with the generalized Born and surface
area continuum solvation.41 Residues within 8 Å from the ligand were
defined as flexible during free energy calculation. This docking
procedure was validated by docking abiraterone and progesterone to
CYP17A1 and CYP21A2, respectively. Abiraterone docked into
CYP17A1 with the highest Glide score was compared to abiraterone
in the X-ray crystallographic structure, while progesterone docked into
CYP21A2 was compared with the crystal structure of CYP21A2 bound
to progesterone.
ORCID
Jeffrey Aube: 0000-0003-1049-5767
́
Present Address
⊥Department of Chemistry, Cambridge University, Cambridge,
England.
Author Contributions
C.F., C.D.V., and R.Y. contributed equally to this project. C.F.
initially designed and produced many of the abiraterone
analogs, undertook preliminary assays, analyzed data, and
participated in crystallization and X-ray structure determina-
tion. C.D.V. resynthesized, characterized, and expanded the
initial set of abiraterone analogs and analyzed data. R.Y.
expressed and purified CYP17A1, CYP21A2, and NADPH-
cytochrome P450 reductase recombinant proteins for assays,
determined all IC50 values, performed the docking calculations,
and analyzed data. K.L. resynthesized particular analogs. J.A.
and E.E.S. conceived the study and analyzed data. The initial
draft of the paper was written by C.D.V., R.Y., E.E.S., and J.A.;
all authors participated in revision.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Patrick Porubsky, who helped develop an initial
GC−MS method for steroid analysis; Linda Blake, who
designed the CYP21A2 expression construct and developed
the original expression and purification protocols; Natasha
DeVore and Elyse Petrunak, who expressed CYP17A1 for
crystallography while mentoring C.F. and worked with C.F. on
crystallization and early structure determination; Patrick O’Day,
who optimized LC−MS/MS methods for the final steroid
analysis method used; and Brandie M. Ehrmann for LC−MS
and MS data acquisition. This work was supported by National
Institute of General Medical Sciences (R01 GM102505, to
E.E.S. and J.A., and R37 GM076343, to E.E.S.), the University
of North Carolina at Chapel Hill, and the University of
MichiganAnn Arbor. The content of this publication is solely
the responsibility of the authors and does not necessarily
represent the official views of NIH.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jmed-
chem.8b00419.
IC50 determination, details of X-ray crystallography, and
copies of 1H and 13C NMR spectra for some
resynthesized and all new compounds (PDF)
ABBREVIATIONS
CYP, cytochrome P450; CV, column volume
■
SMILES strings (CSV)
PDB coordinates of 13b docked into CYP17A1 (PDB)
and CYP21A2 (PDB), 12b docked into CYP17A1
(PDB) and CYP21A2 (PDB), 10b docked into
CYP17A1 (PDB) and CYP21A2 (PDB), 7 docked into
CYP17A1 (PDB) and CYP21A2 (PDB), 8 docked into
CYP17A1 (PDB) and CYP21A2 (PDB), 1b docked into
CYP17A1 (PDB) and CYP21A2 (PDB), and 6b docked
into CYP17A1 (PDB) and CYP21A2 (PDB)
REFERENCES
■
(1) American Cancer Society. Cancer Facts & Figures 2017. https://
www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-
figures/cancer-facts-figures-2017.html (accessed Jan 24, 2018).
(2) Geller, J. Basis for Hormonal Management of Advanced Prostate
Cancer. Cancer 1993, 71, 1039−1045.
(3) Vogiatzi, P.; Claudio, P. P. Efficacy of Abiraterone Acetate in
Post-docetaxel Castration-resistant Prostate Cancer. Expert Rev.
Anticancer Ther. 2010, 10, 1027−1030.
(4) de Bono, J. S.; Logothetis, C. J.; Molina, A.; Fizazi, K.; North, S.;
Chu, L.; Chi, K. N.; Jones, R. J.; Goodman, O. B., Jr.; Saad, F.;
Staffurth, J. N.; Mainwaring, P.; Harland, S.; Flaig, T. W.; Hutson, T.
Accession Codes
The coordinates for the three new X-ray crystallographic
structures with compound 7, 8, and 13b have been deposited to
M
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
E.; Cheng, T.; Patterson, H.; Hainsworth, J. D.; Ryan, C. J.; Sternberg,
C. N.; Ellard, S. L.; Flechon, A.; Saleh, M.; Scholz, M.; Efstathiou, E.;
Zivi, A.; Bianchini, D.; Loriot, Y.; Chieffo, N.; Kheoh, T.; Haqq, C. M.;
Scher, H. I. Abiraterone and Increased Survival in Metastatic Prostate
Cancer. N. Engl. J. Med. 2011, 364, 1995−2005.
(5) Fizazi, K.; Scher, H. I.; Molina, A.; Logothetis, C. J.; Chi, K. N.;
Jones, R. J.; Staffurth, J. N.; North, S.; Vogelzang, N. J.; Saad, F.;
Mainwaring, P.; Harland, S.; Goodman, O. B., Jr.; Sternberg, C. N.; Li,
J. H.; Kheoh, T.; Haqq, C. M.; de Bono, J. S. Abiraterone Acetate for
Treatment of Metastatic Castration-resistant Prostate Cancer: Final
Overall Survival Analysis of the COU-AA-301 Randomised, Double-
Blind, Placebo-controlled Phase 3 Study. Lancet Oncol. 2012, 13, 983−
992.
(19) Li, Z.; Bishop, A. C.; Alyamani, M.; Garcia, J. A.; Dreicer, R.;
Bunch, D.; Liu, J.; Upadhyay, S. K.; Auchus, R. J.; Sharifi, N.
Conversion of Airaterone to D4A Drives Anti-Tumour Activity in
Prostate Cancer. Nature 2015, 523, 347.
(20) Pallan, P. S.; Wang, C.; Lei, L.; Yoshimoto, F. K.; Auchus, R. J.;
Waterman, M. R.; Guengerich, F. P.; Egli, M. Human Cytochrome
P450 21A2, the Major Steroid 21-Hydroxylase: Structure of the
Enzyme/Progesterone Substrate Complex and Rate-Limiting C-H
Bond Cleavage. J. Biol. Chem. 2015, 290, 13128−13143.
(21) Bonomo, S.; Hansen, C. H.; Petrunak, E. M.; Scott, E. E.;
Styrishave, B.; Jorgensen, F. S.; Olsen, L. Promising Tools in Prostate
Cancer Research: Selective Non-Steroidal Cytochrome P450 17A1
Inhibitors. Sci. Rep. 2016, 6, 29468.
(6) Handratta, V. D.; Vasaitis, T. S.; Njar, V. C. O.; Gediya, L. K.;
Kataria, R.; Chopra, P.; Newman, D.; Farquhar, R.; Guo, Z.; Qiu, Y.;
Brodie, A. M. H. Novel C-17-Heteroaryl Steroidal CYP17 Inhibitors/
Antiandrogens: Synthesis, in Vitro Biological Activity, Pharmacoki-
netics, and Antitumor Activity in the LAPC4 Human Prostate Cancer
Xenograft Model. J. Med. Chem. 2005, 48, 2972−2984.
(22) Trachtenberg, J.; Halpern, N.; Pont, A. Ketoconazole: A Novel
and Rapid Treatment for Advanced Prostatic Cancer. J. Urol. 1983,
130, 152−153.
(23) Li, Z.; Alyamani, M.; Li, J.; Rogacki, K.; Abazeed, M.; Upadhyay,
S. K.; Balk, S. P.; Taplin, M.-E.; Auchus, R. J.; Sharifi, N. Redirecting
Abiraterone Metabolism to Fine-Tune Prostate Cancer Anti-Androgen
Therapy. Nature 2016, 533, 547.
(24) Benoist, G. E.; Hendriks, R. J.; Mulders, P. F. A.; Gerritsen, W.
R.; Somford, D. M.; Schalken, J. A.; van Oort, I. M.; Burger, D. M.; van
Erp, N. P. Pharmacokinetic Aspects of the Two Novel Oral Drugs
Used for Metastatic Castration-Resistant Prostate Cancer: Abiraterone
Acetate and Enzalutamide. Clin. Pharmacokinet. 2016, 55, 1369−1380.
(25) Harris, R. K.; Becker, E. D.; Cabral de Menezes, S. M.;
Goodfellow, R.; Granger, P. NMR Nomenclature: Nuclear Spin
Properties and Conventions For Chemical Shifts. IUPAC Recom-
mendations 2001. International Union of Pure and Applied Chemistry.
Physical Chemistry Division. Commission on Molecular Structure and
Spectroscopy. Magn. Reson. Chem. 2002, 40, 489−505.
(26) Madhra, M. K.; Sriram, H. M.; Inamdar, M.; Sharma, M. K.;
Prasad, M.; Joseph, S. Improved Procedure for Preparation of
Abiraterone Acetate. Org. Process Res. Dev. 2014, 18, 555−558.
(27) Bian, X.; Wang, L.; Liu, J.; Wang, C. Green Suzuki Coupling
Reaction for Synthesis of Abiraterone Acetate and its Analogues. J.
Chem. Res. 2016, 40, 289−292.
(28) Bu, M.; Cao, T.; Li, H.; Guo, M.; Yang, B. B.; Zeng, C.; Zhou,
Y.; Zhang, N.; Hu, L. Synthesis And Biological Evaluation of Novel
Steroidal 5α,8α-Epidioxyandrost-6-ene-3β-ol-17-(O-
phenylacetamide)oxime Derivatives as Potential Anticancer Agents.
Bioorg. Med. Chem. Lett. 2017, 27, 3856−3861.
(29) Chaubal, R.; Mujumdar, A. M.; Misar, A.; Deshpande, V. H.;
Deshpande, N. R. Structure-activity Relationship Study of Androstene
Steroids with Respect to Local Anti-inflammatory Activity. Arzneim.
Forsch. 2006, 56, 394−398.
(30) Yu, J.; Zhou, Y.; Lin, Z.; Tong, R. Regioselective and
Stereospecific Copper-Catalyzed Deoxygenation of Epoxides to
Alkenes. Org. Lett. 2016, 18, 4734−4737.
(31) Bart, A. G.; Scott, E. E. Structural and Functional Effects of
Cytochrome b5 Interactions with Human Cytochrome P450 Enzymes.
J. Biol. Chem. 2017, 292, 20818−20833.
(7) Njar, V. C. O.; Brodie, A. M. H. Discovery and Development of
Galeterone (TOK-001 or VN/124−1) for the Treatment of All Stages
of Prostate Cancer. J. Med. Chem. 2015, 58, 2077−2087.
(8) Yamaoka, M.; Hara, T.; Hitaka, T.; Kaku, T.; Takeuchi, T.;
Takahashi, J.; Asahi, S.; Miki, H.; Tasaka, A.; Kusaka, M. Orteronel
(TAK-700), a Novel Non-Steroidal 17,20-Lyase Inhibitor: Effects on
Steroid Synthesis in Human and Monkey Adrenal Cells and Serum
Steroid Levels in Cynomolgus Monkeys. J. Steroid Biochem. Mol. Biol.
2012, 129, 115−128.
(9) Rafferty, S. W.; Eisner, J. R.; Moore, W. R.; Schotzinger, R. J.;
Hoekstra, W. J. Highly-selective 4-(1,2,3-Triazole)-based P450C17A
17,20-Lyase Inhibitors. Bioorg. Med. Chem. Lett. 2014, 24, 2444−2447.
(10) Huang, A.; Jayaraman, L.; Fura, A.; Vite, G. D.; Trainor, G. L.;
Gottardis, M. M.; Spires, T. E.; Spires, V. M.; Rizzo, C. A.; Obermeier,
M. T.; Elzinga, P. A.; Todderud, G.; Fan, Y.; Newitt, J. A.; Beyer, S. M.;
Zhu, Y.; Warrack, B. M.; Goodenough, A. K.; Tebben, A. J.; Doweyko,
A. M.; Gold, D. L.; Balog, A. Discovery of the Selective CYP17A1
Lyase Inhibitor BMS-351 for the Treatment of Prostate Cancer. ACS
Med. Chem. Lett. 2016, 7, 40−45.
(11) Attard, G.; Reid, A. H. M.; Auchus, R. J.; Hughes, B. A.; Cassidy,
A. M.; Thompson, E.; Oommen, N. B.; Folkerd, E.; Dowsett, M.; Arlt,
W.; de Bono, J. S. Clinical and Biochemical Consequences of
CYP17A1 Inhibition with Abiraterone Given with and without
Exogenous Glucocorticoids in Castrate Men with Advanced Prostate
Cancer. J. Clin. Endocrinol. Metab. 2012, 97, 507−516.
(12) Malikova, J.; Brixius-Anderko, S.; Udhane, S. S.; Parween, S.;
Dick, B.; Bernhardt, R.; Pandey, A. V. CYP17A1 Inhibitor Abiraterone,
an Anti-Prostate Cancer Drug, Also Inhibits the 21-Hydroxylase
Activity of CYP21A2. J. Steroid Biochem. Mol. Biol. 2017, 174, 192−
200.
(13) Auchus, R. J.; Yu, M. K.; Nguyen, S.; Mundle, S. D. Use of
Prednisone with Abiraterone Acetate in Metastatic Castration-resistant
Prostate Cancer. Oncologist 2014, 19, 1231−1240.
(14) Petrunak, E. M.; DeVore, N. M.; Porubsky, P. R.; Scott, E. E.
Structures of human Steroidogenic Cytochrome P450 17A1 with
Substrates. J. Biol. Chem. 2014, 289, 32952−32964.
(32) Turcu, A. F.; Rege, J.; Chomic, R.; Liu, J.; Nishimoto, H. K.;
Else, T.; Moraitis, A. G.; Palapattu, G. S.; Rainey, W. E.; Auchus, R. J.
Profiles of 21-Carbon Steroids in 21-Hydroxylase Deficiency. J. Clin.
Endocrinol. Metab. 2015, 100, 2283−2290.
(33) Motulsky, H. J.; Brown, R. E. Detecting Outliers When Fitting
Data with Nonlinear Regression − A New Method Based on Robust
Nonlinear Regression and The False Discovery Rate. BMC Bioinf.
2006, 7, 123.
(15) Yadav, R.; Petrunak, E. M.; Estrada, D. F.; Scott, E. E. Structural
Insights into the Function of Steroidogenic Cytochrome P450 17A1.
Mol. Cell. Endocrinol. 2017, 441, 68−75.
(16) DeVore, N. M.; Scott, E. E. Structures of Cytochrome P450
17A1 with Prostate Cancer Drugs Abiraterone and TOK-001. Nature
2012, 482, 116−119.
(34) Kabsch, W. Xds. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,
(17) Potter, G. A.; Barrie, S. E.; Jarman, M.; Rowlands, M. G. Novel
Steroidal Inhibitors of Human Cytochrome P45017.alpha.-Hydrox-
ylase-C17,20-lyase): Potential Agents for the Treatment of Prostatic
Cancer. J. Med. Chem. 1995, 38, 2463−2471.
66, 125−132.
(35) Collaborative Computational Project, N.. The CCP4 Suite:
Programs for Protein Crystallography. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 1994, 50, 760−763.
(18) Garrido, M.; Peng, H. M.; Yoshimoto, F. K.; Upadhyay, S. K.;
Bratoeff, E.; Auchus, R. J. A-ring Modified Steroidal Azoles Retaining
Similar Potent and Slowly Reversible CYP17A1 Inhibition as
Abiraterone. J. Steroid Biochem. Mol. Biol. 2014, 143, 1−10.
(36) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P. D.; Winn, M.
D.; Storoni, L. C.; Read, R. J. Phaser Crystallographic Software. J. Appl.
Crystallogr. 2007, 40, 658−674.
N
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(37) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis,
I. W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; Grosse-
Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R.
J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H.
PHENIX: A Comprehensive Python-based System for Macro-
molecular Structure Solution. Acta Crystallogr., Sect. D: Biol. Crystallogr.
2010, 66, 213−221.
(38) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Features and
Development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,
66, 486−501.
(39) The AxPyMOL Molecular Graphics Plugin for Microsoft
PowerPoint, 1.0; Schrodinger, LLC, 2010.
(40) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.;
Klicic, J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.;
Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. Glide: A New
Approach For Rapid, Accurate Docking and Scoring. 1. Method and
Assessment of Docking Accuracy. J. Med. Chem. 2004, 47, 1739−1749.
(41) Li, J.; Abel, R.; Zhu, K.; Cao, Y.; Zhao, S.; Friesner, R. A. The
VSGB 2.0 Model: A Next Generation Energy Model for High
Resolution Protein Structure Modeling. Proteins: Struct., Funct., Genet.
2011, 79, 2794−2812.
O
DOI: 10.1021/acs.jmedchem.8b00419
J. Med. Chem. XXXX, XXX, XXX−XXX