Potent and Highly Selective Aldo-Keto Reductase 1C3 (AKR1C3) Inhibitors Act as Chemotherapeutic Potentiators in Acute Myeloid Leukemia and T-Cell Acute Lymphoblastic Leukemia

Article abstract of DOI:10.1021/acs.jmedchem.9b00090

Aldo-keto reductase 1C3 (AKR1C3) catalyzes the synthesis of 9α,11β-prostaglandin (PG) F_2α and PGF_2α prostanoids that sustain the growth of myeloid precursors in the bone marrow. The enzyme is overexpressed in acute myeloid leukemia (

Full text of DOI:10.1021/acs.jmedchem.9b00090

Subscriber access provided by University of Sussex Library
Article
Potent and Highly Selective Aldo–Keto Reductase 1C3 (AKR1C3)
Inhibitors Act as Chemotherapeutic Potentiators in Acute
Myeloid Leukemia and T-Cell Acute Lymphoblastic Leukemia
Kshitij Verma, Tianzhu Zang, Trevor M. Penning, and Paul C. Trippier
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00090 • Publication Date (Web): 05 Mar 2019
Downloaded from http://pubs.acs.org on March 6, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 72
Journal of Medicinal Chemistry
Potent and Highly Selective Aldo–Keto Reductase 1C3 (AKR1C3) Inhibitors Act as Chemotherapeutic
Potentiators in Acute Myeloid Leukemia and T-Cell Acute Lymphoblastic Leukemia
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
b
b
a,c*
Kshitij Verma, Tianzhu Zang, Trevor M. Penning and Paul C. Trippier
a
Department of Pharmaceutical Sciences, Texas Tech University Health Sciences Center, School of Pharmacy,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Amarillo, TX 79106, USA.
b
Center of Excellence in Environmental Toxicology, Department of Systems Pharmacology & Translational
Therapeutics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA.
c
Center for Chemical Biology, Department of Chemistry and Biochemistry, Texas Tech University, Lubbock,
TX 79409, USA.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 72
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
Aldo–keto reductase 1C3 (AKR1C3) catalyzes the synthesis of 9α,11β-prostaglandin (PG) F and PGF
2α
2
α
prostanoids that sustain the growth of myeloid precursors in the bone marrow. The enzyme is overexpressed in
acute myeloid leukemia (AML) and T-cell acute lymphoblastic leukemia (T-ALL). Moreover, AKR1C3 confers
chemotherapeutic resistance to the anthracyclines; first-line agents for the treatment of leukemias. The highly
homologous isoforms AKR1C1 and AKR1C2 inactive 5-dihydroprogesterone and their inhibition would be
undesirable. We report herein, the identification of AKR1C3 inhibitors that demonstrate exquisite isoform
selectivity for AKR1C3 over the other closely related isoforms to the order of >2800-fold. Biological evaluation
of our isoform selective inhibitors revealed a high degree of synergistic drug action in combination with the
clinical leukemia therapeutics daunorubicin and cytarabine in in vitro cellular models of AML and primary
patient-derived T-ALL cells. Our developed compounds exhibited >100-fold dose reduction index that results in
complete resensitization of a daunorubicin-resistant AML cell line to the chemotherapeutic and >100-fold dose
reduction of cytarabine in both AML cell lines and primary T-ALL cells.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
ACS Paragon Plus Environment

Page 3 of 72
Journal of Medicinal Chemistry
INTRODUCTION:
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Acute myeloid leukemia (AML) is a debilitating condition that affects both pediatric and elderly
populations. With an estimated incidence rate of more than 19,520 it is the second most common type of
1
hematologic malignancy. The underlying pathophysiology for AML is marked by a differentiation block and
clonal expansion of immature blast cells in the bone marrow that can be ascribed to various chromosomal
translocations offering a proliferative advantage.2, 3 Clinically, AML is classified into seven distinct subtypes
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(M0-M7) depending on the type of cells affected within the myelocytic lineage during hematopoiesis. The M3
subtype or Acute Promyelocytic Leukemia (APL) stems from a reciprocal chromosomal translocation of
promyelocytic leukemia / retinoic acid receptor alpha (PML / RARα) oncogene that prevents cellular
4
differentiation. Treatment with retinoic acid is a primary therapeutic intervention that is employed along with
5
arsenic trioxide or chemotherapy to induce remission. However, the use of retinoic acid is limited to only the M3
6
subtype, due to adverse effects. Standard treatment of care for the other types of AML relies heavily on the use
of anthracyclines and the antimetabolite cytarabine as first line chemotherapeutics for remission induction as well
as for maintenance therapy. Continuous infusion of cytarabine for 7 days in combination with daunorubicin for 3
days is employed as induction chemotherapy (7+3 regimen) followed by cytarabine administration for
2
maintenance. Since AML predominantly affects the elderly with higher incidence rates among patients >65 years
7
of age, the tolerability of intensive chemotherapy remains a grave concern. Moreover, the disease prognosis
among older patients is poor due to higher relapse rates and unfavorable cytogenetics.8
Acute lymphocytic leukemia (ALL) stems from a malignant transformation of the lymphoid precursors in
the bone marrow and is estimated to affect 5960 individuals per year in the US, predominantly children and
1
adolescents. Approximately 25% of all ALL cases are classified as T-cell ALL (T-ALL) that affects the T
9
lymphocytes of the white blood cell lineage. Owing to chromosomal translocation of transcription factors that
are then placed under the control of strong T-cell receptor gene promoters, immature precursor T-cells gain a
proliferative advantage. The clinical outcome for patients suffering from primary refractory or relapsed T-ALL
remains bleak.10, 11 Chemotherapeutics used to treat AML are commonly used to treat T-ALL and encounter
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 72
1
1
similar resistance profiles in relapse. Development of non-toxic chemical agents that can increase the efficacy
of currently employed chemotherapeutics and prevent the development of resistance represent an attractive
therapeutic strategy to manage leukemic disease progression.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Aldo–keto reductase 1C3 (AKR1C3) is overexpressed in a range of leukemic cell lines spanning various
AML subtypes and in high risk T-ALL.12, 13 By virtue of the enzyme’s prostaglandin F synthase (PGFS) activity,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
AKR1C3 converts the precursor prostanoid PGD to 11β-PGF , that induces proliferation of leukemic blasts.
2
2α
The proliferative activity of 11β-PGF is attributed to the activation of prostaglandin F receptor (FP) and the
2
α
2α
downstream mitogen-activated protein kinase (MAPK) cascade.14 In the absence of AKR1C3, PGD non-
2
1
2,14
enzymatically converts to PGJ and eventually to 15-deoxy-Δ
-prostaglandin J (15Δ-PGJ ). The 15Δ-PGJ
2 2 2
2
serves as a ligand for the peroxisome proliferator-activated receptor-γ (PPARγ) and promotes cellular
15
differentiation. Prior reports have validated the role AKR1C3 plays in inducing myeloid cell proliferation and
demonstrate that upon AKR1C3 upregulation the cells become resistant to differentiation by all-trans retinoic
1
2
acid (ATRA), whereas the loss of AKR1C3 trans-gene expression confers sensitivity to ATRA. Studies have
shown that PGD synergizes with ATRA in HL-60 promyelocytic cells to induce differentiation and apoptosis,
2
while in contrast, incubation with 11β-PGF2α significantly increased cell proliferation.12 Since direct
administration of pro-differentiation prostaniods PGJ and 5Δ-PGJ is severely limited by high toxicity and low
2
2
bioavailability, a superior therapeutic strategy will be to inhibit the activity of AKR1C3 in order to divert the
prostanoid metabolism towards the production of PGJ series prostanoids in order to gain a beneficial therapeutic
2
1
6
outcome. The related AKR1C3 isoforms AKR1C1 and 1C2 share >84% sequence homology to 1C3 and are
responsible for the inactivation of potent androgens (e.g. 5α-DHT) thus the inhibition of these isoforms is
undesirable as it cannot cause androgen insufficency.17, 18 Importantly, AKR1C3 acts as a phase I
biotransformation enzyme, transforming anthracyclines (doxorubicin, daunorubicin and idarubicin) into their less
potent C-13 hydroxy metabolites that results in chemotherapeutic resistance.19, 20 Apart from AML and T-ALL,
21
22
23, 24
endometrium25, 26 and
AKR1C3 is overexpressed in a variety of neoplasms of the breast, colon, prostate,
lungs.27
4
ACS Paragon Plus Environment

Page 5 of 72
Journal of Medicinal Chemistry
Significant efforts have been made to discover and develop different classes of AKR1C3 inhibitors from
1
2
diverse structural classes (steroids,26 flavones,28 jasmonates29 and non-steroidal anti-inflammatory drugs
(NSAID’s))21, 24, 26, 30 (Figure 1). Among steroids, medroxyprogesterone acetate (MPA) and an estrogen lactone
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
EM1404) are potent inhibitors of AKR1C3 but are limited in application due to the lack of isoform-selectivity.
Jasmonic acids suffer from the same fate and lack inhibition potency and selectivity towards AKR1C3. Among
flavonoids, 2-hydroxyflavone exhibits moderate potency while inhibiting AKR1C3 with 157-fold isoform
selectivity. Repurposed NSAIDs possess high inhibitory potency for AKR1C3 and selectivity but have not made
it into the clinic. For example derivatization of the indomethacin scaffold to the des-methyl indomethacin
analogue, modification of naproxen to R-ethyl naproxen and the structural optimization of flufenamic acid to a
meta isomer abrogated cyclooxygenase (COX) inhibition activities while retaining inhibitory potency for
AKR1C3. The N-naphthylamino benzoate analogue was found to be a bifunctional AKR1C3 inhibitor that also
antagonizes androgen receptor (AR) functions.21, 26, 31-35 Inhibitor SN33638 demonstrated only a moderate
inhibition of testosterone production and a modest reduction in cell viabilities evaluated in AKR1C3 dependent
castration-resistant prostate cancer (CRPC) cell lines whereas ASP9521, taken through a phase I/II clinical trial
in metastatic CRPC patients established no observable clinical activity.36-38 Another bifunctional inhibitor GTX-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
60 has been shown to inhibit the AR coactivator functions of AKR1C3 in addition to being an AKR1C3
3
9
inhibitor. Recently reported, a benzoisoxazole derivative designed by applying a scaffold hopping approach to
flufenamic acid displayed a greater isoform selectivity for AKR1C3 inhibition as compared to AKR1C1.
4
0
However, the inhibition potency still remained in mid-nanomolar range (Figure 1). There may be multiple
reasons for AKR1C3 inhibitor failure which has prevented their clinical use.
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 72
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Structures of known AKR1C3 inhibitors.
Even though indomethacin yielded an IC value = 0.1 M and exhibited a 356-fold selectivity for
50
AKR1C3 it displayed a weak antiproliferative effect in AML cells.12, 41 A study combining the pan-AKR1C
inhibitor medroxyprogesterone acetate (MPA) (AKR1C3 IC = 2.7 M, 0.66-fold selectivity for AKR1C3) and
5
0
bezafibrate (a PPAR α/γ agonist) provided an approximate two-fold potentiation of bezafibrate activity;
demonstrating a reduction in cell proliferation with an increased differentiation and apoptosis in AML cells.16
However, further studies on the mechanism of action of MPA attributed this activity to targets other than the
AKR1C family. A recent study reported that the specific AKR1C3 inhibitor 4-methyl(de-dimethylamine)-
tetracycline (4-MDDT) (IC = 0.51 M) does not give the adjuvant effect at concentrations up to 50 M that a
5
0
4
2
pan-AKR1C inhibitor does, thus requiring greater credentialing of the AKR1C3 target in leukemia. Although
inhibitors with nanomolar potency and relatively high selectivity to AKR1C3 (up to 500-fold) over the other
6
ACS Paragon Plus Environment

Page 7 of 72
Journal of Medicinal Chemistry
isoforms have been identified from prior studies, the discovery of new compounds for preclinical development is
still required.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Derivatives of cinnamic acids have been shown to exhibit inhibition activities for AKR isozymes bearing
selectivity towards AKR1C3. Among these compounds, baccharin (Figure 1), a natural product extracted from
honeybee propolis has been shown to potently and selectively inhibit AKR1C3 with an IC of 0.11 µM and a
5
0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
3
fold selectivity of 500 over AKR1C2. This discovery has made baccharin a promising hit to develop a new
series of potent and specific inhibitors against AKR1C3. We adopted this natural product scaffold and have
previously reported the synthesis and a preliminary structure-activity relationship (SAR) for AKR1C3
4
4
inhibition. The dihydrocinnamoyloxy moiety of baccharin is reported as a structural prerequisite for AKR1C3
4
5
inhibition, and other groups have synthesized potent AKR1C3 inhibitors based on this scaffold. However, the
presence of an ester linkage introduces a hydrolytic lability into baccharin making these analogues unsuitable for
drug discovery. Hydrolysis of the ester results in formation of the known phenol drupanin, with complete
4
6
abrogation of AKR1C3 inhibitory potency. Our prior studies have reported the design, synthesis and evaluation
of potent AKR1C3 inhibitors bearing a more stable amide bioisostere (Figure 1). These AKR1C3 inhibitors
exhibit a six-fold potentiation of etoposide and a ten-fold potentiation of daunorubicin in AML cell lines.47
However, the selectivity over AKR1C1 and AKR1C2, while significantly improved above that of MPA (0.66-
fold), remained relatively low (109-fold). Continuing our efforts to identify highly isoform-selective AKR1C3
inhibitors we herein report the discovery of a library of optimized compounds possessing >2800-fold selectivity
for AKR1C3 inhibition, with retention of inhibitory potency in the nanomolar range. Further, we demonstrate that
our highly isoform-selective AKR1C3 inhibitors provide up to 100-fold potentiation of the clinical
chemotherapeutics daunorubicin and cytarabine across a panel of AML cell lines and in primary patient-derived
T-ALL cells.
RESULTS AND DISCUSSION:
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 72
Chemistry. Based on the baccharin structural scaffold, three different classes of analogues were
synthesized and evaluated for AKR1C3 inhibitory potency and selectivity towards the other highly homologous
AKR isoforms (Figure 2). To explore modifications of the dihydrocinnamoyloxy moiety, Class I compounds
possess an ester or amide link to the terminal phenyl B ring. The spacer is of varied length and features a diversity
of substituents on the phenyl ring. Class II compounds have the acrylate side chain moved to the meta-position
and substitution on the terminal phenyl ring is varied. In addition, class II compounds also explore 1,3,5
substituents on the central A-ring. Class III compounds constitute varying substitution patterns on the central A
ring and also explore boronic acid bioisosteres of the cinnamic acid.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. SAR strategy for the design of Baccharin derivatives.
Baccharin (1), compounds 7-7a (class III) and 8b-n (class I) were synthesized using a modified literature
48
procedure. Nucleophilic substitution of commercially available 4-iodophenol (2) with prenyl bromide yielded
8
ACS Paragon Plus Environment

Page 9 of 72
Journal of Medicinal Chemistry
4
9
the alkylated intermediate 3, which was subsequently esterified with an appropriately substituted acid chloride
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
4a-n) to yield ester intermediates (5a-n). Mizoroki-Heck50, 51 coupling of the aryl iodide intermediates (5a-n)
with tert-butyl acrylate or vinyl boronic acid pinacol ester using Pd(OAc) and PPh as a catalyst-ligand complex
2
3
52
produced α,β-unsaturated olefins 6a-n and 7 respectively. Chemoselective hydrolysis of the tert-butyl ester of
6
a-n afforded baccharin (1) and derivatives 8b-n, whereas hydrolysis of the pinacol ester53, 54 afforded boronic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
acid 7a (Scheme 1).
OH
OH
O
i
ii
Cl
R1
4a - 4n: R1 = a - n
I
I
2
3
O
O
O
O
R1
O
R1
O
I
R1
iii
iv
R2
R2
CO H
2
5a - 5n: R1 = a - n
6
7
a - 6n: R1 = a - n; R2 = a
: R1 = a; R2 = b
1: R1 = a
8b - 8n: R1 = b - n
O
R :
1
a =
h =
i =
-CH3
O
v
F
b =
c =
d =
j =
B
HO
OH
F
k =
7a
R : a =
2
O
O
O
l =
b =
B
F
O
OMe
e =
f =
m =
n =
N
OMe
g =
OMe
Reagents and conditions: (i) Prenyl Bromide, NaH, PhMe, 55%; (ii) DMAP, NEt , DCM; (iii) NEt ,
3
3
o
P(Ph) , Pd(OAc) , PhMe, reflux; (iv) SiO , PhMe, reflux; (v) MeOH/H O, 80 C.
3
2
2
2
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 72
4
4
44
Scheme 1: Synthesis of baccharin (1), substituted ester derivatives (8b-g, 8h, 8i-n) of class I and boronic acid
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
analogues (7 and 7a) of class III.
The reaction sequence was then modified where p-iodoaniline (9) was regioselectively brominated using a
5
5
56
modified literature procedure that afforded dihalide 10. Exploiting the greater reactivity of iodide over bromide
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
1
in Mizoroki-Heck couplings, 10 was selectively coupled with tert-butylacrylate yielding the intermediate
aniline 11. Coupling with the corresponding acid chloride gave amide intermediates (12a-e). A palladium-
catalyzed Suzuki-Miyaura57, 58 cross-coupling reaction installed the prenyl side chain that yielded intermediates
(
13a-e) which underwent chemoselective hydrolysis of the tert-butyl ester to afford the final compounds (14a47
and 14b-e) of class I (Scheme 2).
10
ACS Paragon Plus Environment

Page 11 of 72
Journal of Medicinal Chemistry
O
iii
NH2
1
2
3
4
5
6
7
8
9
NH2
Cl
R
NH2
Br
i
Br
ii
I
I
t
CO Bu
2
9
10
11
O
O
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
HN
R
HN
R
HN
R
iv
v
Br
t
t
CO2 Bu
CO H
2
CO Bu
2
12a - 12e: R = a - e
13a - 13e: R = a - e
14a - 14e: R = a - e
R:
a =
b =
c =
-CH3
d =
e =
Reagents and conditions: (i) Br , AcOH, RT, 44%; (ii) tert-butylacrylate,
2
o
NEt , P(Ph) , Pd(OAc) , PhMe, reflux, 64%; (iii) NEt , DMAP, DCM, 40 C;
3
3
2
3
o
(iv) prenylboronic acid pinacol ester, Pd(dppf)Cl , Cs CO , DMF, 90 C; (v)
2
2
3
SiO , PhMe, reflux.
2
47
Scheme 2: Synthesis of amide bioisosteres (14a, 14b-e) of class I derivatives.
Class II compounds in which the substitution pattern on the A-ring of baccharin is altered, in which the ester
chain was moved to the meta-position relative to the acrylate side chain. Commercially available m-coumaric acid
(
15) was protected as the methyl ester and subsequently brominated, with column chromatography providing
access to the desired regioisomer in 35% yield. The brominated intermediate (17) was deprotected under basic
conditions to yield compound 17a which was followed by esterification of the phenol using phenylpropionyl acid
57
chloride to yield ester derivative 18. Suzuki-Miyaura reaction with the appropriately substituted boronic acid
4
7
pinacol ester afforded prenyl and allyl derivatives (19a and 19b) of class II analogues (Scheme 3).
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 72
CO R
CO H
2
2
CO R
2
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ii
iv
Br
Br
OH
m-coumaric acid (15): R = H
O
OH
17: R = Me
17a: R = H
O
i
iii
1
6: R = Me
18
CO H
2
R :
v
1
R1
a =
b =
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
O
B
R1
O
19a: R = a
1
19b: R = b
1
Reagents and conditions: (i) MeOH, H SO , reflux, 93%; (ii) Br , AcOH, RT, 35%;
(iii) NaOH, H O, reflux, 90%; (iv) 3-phenylpropanoyl chloride, DMAP, NEt , DCM, RT,
2
4
2
2
3
o
3
6%; (v) Pd(dppf)Cl , Cs CO , DMF, 90 C.
2 2 3
47
Scheme 3: Synthesis of meta-substituted ester derivatives (19a and 19b) of class II.
5
1
The amide bioisosteres 26a-l, 28 (class II) and 24 (class III) were accessed through Mizoroki-Heck reaction
of commercially available 4-bromo-2-iodoaniline (20) with tert-butyl acrylate or vinyl boronic acid pinacol ester
to afford 21a and 21b respectively. Amide formation with the appropriate acid chloride as previously described
5
7
yielded amides 23a-23j and 24. Subsequent Suzuki-Miyaura cross-coupling afforded intermediates 25a-25j and
4
7
27 which were chemoselectively hydrolyzed to yield final compounds 26a and 26b-26j and 28. Further,
hydrolysis of the tosyl moieties in 26i and 26j, afforded phenols 26k and 26l respectively (Scheme 4).
12
ACS Paragon Plus Environment

Page 13 of 72
Journal of Medicinal Chemistry
R :
2
1
2
3
4
5
6
7
8
9
g =
h =
a =
b =
OMe
F
F
c =
d =
OTs
OTs
i =
j =
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R1
R1
I
i
ii
Br
Br
O
R2COCl
NH2
F
NH2
4a - 4g, 22h - 22j: R2 = a - j
2
1a: R1 = a
20
OMe
21b: R1 = b
e =
f =
O
O
k =
l =
OH
OH
R : a =
1
OMe
b =
B
O
O
B
CO H
2
t
R1
O
CO Bu
2
2
6i - 26j: R2 = i - j
iii
iv
Br
v
HN
R2
HN
R2
HN
R2
26k - 26l: R2 = k - l
O
O
O
2
6a - 26j: R2 = a - j
2
5a - 25j: R2 = a - j
2
2
3a - 23j: R1 = a; R2 = a - j
4: R1 = b; R2 = a
t
CO Bu
CO H
2
2
iv
iii
HO
HO
HN
O
R2
HN
R2
O
B
O
HO
O
27: R2 = a
28: R2
= a
o
o
Reagents and conditions: (i) Pd(OAc) , P(Ph) , NEt , PhMe, 110 C; (ii) DMAP, NEt , DCM, 40 C; (iii) Pd(dppf)Cl ,
2
3
3
3
2
o
Cs CO , DMF, 90 C; (iv) SiO , PhMe, reflux; (v) MeOH, 1M NaOH, H O, reflux.
2
3
2
2
47
Scheme 4: Synthesis of meta-substituted amide derivatives (26a, 26b-l and 28) of class II and boronic ester
derivative 24 of class III.
Class III analogues (33a and 33b) were prepared from the commercially available 4-bromo-2-iodoaniline (29)
as a starting material. After installation of the amide chain by reaction with acid chloride as previously described,
Mizoroki-Heck cross-coupling chemoselectively afforded intermediate 31 due to the increased reactivity of iodide
5
1
57
over bromide for the reaction. Suzuki-Miyaura cross coupling with substituted pinacol boronate followed by
tert-butyl ester deprotection yielded compounds 33a and 33b (Scheme 5).
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 72
O
O
O
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cl
NH2
Br
HN
HN
I
I
ii
tBuO2C
i
Br
Br
31
30
29
O
O
B
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
HN
R
R
O
HN
R
iv
iii
t
HO C
2
BuO C
2
R: a =
32a: R = a
2b: R = b
33a: R = a
33b: R = b
3
OH
b =
o
Reagents and conditions: (i) NEt , DMAP, DCM, 40 C, 69%; (ii) tert-butylacrylate, NEt ,
3
3
o
P(Ph) , Pd(OAc) , PhMe, reflux, 70%; (iii) Pd(dppf)Cl , Cs CO , DMF, 90 C; (iv) SiO , PhMe,
3
2
2
2
3
2
reflux.
Scheme 5: Synthesis of 33a and 33b as amide derivatives of class III.
Further, to modify the side chain substitution pattern of the parent scaffold A ring to the 1,3,5 all meta-
substituted pattern, 3-bromo-5-iodobenzoic acid (34) was used as the common starting material for the synthesis
of amide, ester or retroinverse amide analogues. Conversion of 34 to 3-bromo-5-iodoaniline (35) was achieved
by a modified Curtius rearrangement,59 followed by tert-butyloxycarbonyl (Boc) deprotection. A reaction
sequence of amide synthesis using acid chloride, followed by Pd-catalyzed cross-couplings in the order of
Mizoroki-Heck reaction followed by Suzuki-Miyaura reaction afforded tert-butyl protected intermediates that
were subsequently hydrolyzed to afford compounds 39a and 39b. Conversion of 34 to its acid chloride followed
by esterification or amide formation using phenylethyl alcohol, or substituted primary amine, yielded the ester
and amide intermediates (42, 41a-h) respectively. The reaction sequence of Mizoroki-Heck, Suzuki-Miyaura and
tert-butyl ester hydrolysis was used to obtain compounds 49a-h and 50-52 of class III (Scheme 6).
14
ACS Paragon Plus Environment

Page 15 of 72
Journal of Medicinal Chemistry
t
R3
CO Bu
2
R3
CO H
2
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
R : a =
3
v
b =
HN
HN
OH
O
O
3
8a - 38b: R3 = a - b
3
3
9a: R3 = a
9b: R3 = b
O
B
iv
R3
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Br
I
Br
t
NH2
CO2 Bu
R : a =
1
iii
viii
HN
HN
b =
c =
d =
Br
I
F
vi, vii
35
O
O
COOH
I
36
37
OMe
Br
t
i
Br
CO Bu
Br
I
2
COCl
ii
iii
34
X
R1
R1
R1
OMe
Br
I
O
X
e =
O
X
4
0
43a - 43h: R1 = a - h; X = -NH
44: R1 = a; X = -O
X = -NH2
OH
41a - 41h: R1 = a - h; X = -NH
2: R1 = a; X = -O
-
4
OCF3
R : a =
2
f =
iv
O
B
b =
c =
R2
O
OCF3
t
g =
R2
CO H
R2
CO Bu
2
2
OH
v
R1
R1
h =
O
X
O
X
4
9a - 49h: R1 = a - h; R2 = a; X = -NH
50: R1 = a; R2 = b; X = -NH
1: R1 = a; R2 = c; X = -NH
2: R1 = a; R2 = a; X = -O
45a - 45h: R1 = a - h; R2 = a; X = -NH
46: R1 = a; R2 = b; X = -NH
47: R1 = a; R2 = c; X = -NH
48: R1 = a; R2 = a; X = -O
5
5
N
o
Reagents and Conditions: (i) SOCl , PhMe, reflux; (ii) DMAP, NEt , DCM, 40 C; (iii) tert-butylacrylate, Pd(OAc) , P(Ph) ,
2
3
2
3
o
o
NEt , PhMe, 110 C; (iv) Pd(dppf)Cl , Cs CO , DMF, 90 C; (v) SiO , PhMe, reflux; (vi) DPPA, NEt , tBuOH; (vii) 4M HCl,
3
2
2
3
2
3
o
dioxane, 40%; (viii) 3-phenylpropanoyl chloride, DMAP, NEt , DCM, 40 C, 33%.
3
Scheme 6: Synthesis of 1,3,5-meta-substituted derivatives (39a and 39b, 49a-h and 50-52) of class II.
Ether analogues 59a-b were synthesized using 3-bromo-4-hydroxybenzoic acid (53) as the starting material
which was reacted with benzyl bromide or 3-methoxy benzyl bromide to install the ether side chain. After
activating the carboxylic acid to an acid chloride, a reaction sequence of amide formation followed by Mizoroki-
Heck reaction and tert-butyl ester hydrolysis afforded final compounds (59a and 59b) of class II (Scheme 7).
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 72
Br
Br
Br
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
or
MeO
H
COCl
O
N
COOH
COOR2
iii
iv
i
H2N
Br
OH
OR1
Br
OR1
Br
OR1
53
56a: R = a
54a: R = R = a
1
57a: R = a
57b: R = b
1
2
1
54b: R = R = b
56b: R1 = b
1
2
1
ii
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
55a: R = a; R = H
1 2
55b: R = b; R = H
1
2
H
H
N
R :
1
O
N
O
a =
b =
v
vi
O
t
HO C
OR1
BuO C
OR1
2
2
59a: R = a
5
5
8a: R = a
8b: R = b
1
1
59b: R = b
1
1
o
Reagents and conditions: (i) DIPEA, 150 C; (ii) 1M KOH, MeOH, reflux; (iii) SOCl , PhMe, reflux;
2
o
o
(iv) DMAP, NEt , DCM, 40 C; (v) tert-butylacrylate, Pd(OAc) , DABCO, K CO , DMF, 110 C; (vi)
3
2
2
3
SiO , PhMe, reflux.
2
Scheme 7: Synthesis of 1,3,5-meta-substituted ether derivatives (59a and 59b) of class II
Structure-activity relationship.
Class I analogues: Consistent with previous findings,43, 44, 47 baccharin exhibited an IC of 105 nM for
5
0
AKR1C3 inhibition with 510-fold selectivity over AKR1C2. Removal of the dihydrocinnamoyloxy group to
afford 1a (drupanin), resulted in complete abrogation of AKR1C3 inhibitory potency and selectivity. Replacement
of the dihydrocinnamoyloxy group (compounds 8h-n) resulted in low or sub-micromolar inhibition potency for
AKR1C3, with significant reduction of selectivity. Introduction of an acetoxy group (8h) and a cyclohexylacetoxy
group (8j), decreased both AKR1C3 inhibitory potency and selectivity by four-fold. As compared to 1 an increase
in the carbon spacer in 8i decreased AKR1C3 inhibitory potency and selectivity by two-fold and four-fold
respectively. Excision of the ethyl linker to afford the benzoyloxy moiety (8l) reduced inhibitory potency by
three-fold and isoform selectivity to AKR1C3 by eight-fold. Replacement of the terminal phenyl moiety of 1 with
a bulky napthyl (8k) or simple hexane chain (8n) resulted in decrease of the inhibitory potency by three-fold with
loss of selectivity being twelve-fold (8k) and six-fold (8n) as compared to 1. Replacement of the phenethyl moiety
16
ACS Paragon Plus Environment

Page 17 of 72
Journal of Medicinal Chemistry
in 1 with a pyridine ring (8m) significantly reduced potency and selectivity. Comparison with the phenyl
derivative (8l) indicates the pyridine ring is largely responsible for this drop in potency, potentially due to its
ability to ionize under the assay conditions. Replacement of the ester linkage with an amide linkage (14a-e) led
to less potent inhibitors than their bioisosteric counterparts (1, 8h-n). Differences in potency varied from three-
fold reduction (compare 1 to 14a and 8i to 14c) to an eight-fold reduction in AKR1C3 inhibitory potency (compare
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
8
h to 14b). Likewise, the amide derivatives, with the exception of 14a were also significantly more promiscuous
between AKR1C isoforms with only moderate to minimal selectivity for the AKR1C3 isoform over AKR1C2.
Fortuitously the amide derivative of baccharin (14a) retained potency and high selectivity, IC = 0.420 µM with
5
0
9
3-fold selectivity for AKR1C3 over the closely related AKR1C2 isozyme. Additionally, ligand-lipophilicity
efficiency (LLE) which is a measure of ‘drug-likeliness’ and takes into account the potency and lipophilicity of
compounds was also calculated. Even though the amide bioisosteres (14a-e) are less potent inhibitors of AKR1C3
they exhibit a more favorable LLE profile (except 14b) as compared to the ester derivatives (1, 8i-k) due to their
greater lipophilicity (Table 1).
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 72
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
IC50
IC50
AKR1C2
(μM)
Selectivity
LLE*
Compound
144
R
X
AKR1C3
μM)
1
C2:1C3
(
O
0.105
51
510
1.29
a44
1
-
OH
O
15
107
45
29
44
16
7
1.30
2.76
0.56
0.10
-0.17
(Drupanin)
8
h44
-CH3
0.440
0.233
0.453
0.370
102
124
97
8
i
O
8
j
O
8
k
O
43
8
l
O
0.348
21
60
0.74
8
m
O
O
1.8
14
25
39
7.8
83
93
1.52
0.81
1.89
8
n
0.300
0.420
1
4a
4b
4c
NH
1
-CH3
NH
NH
3.6
30
30
8
2.15
1.19
1
0.866
35
1
4d
4e
NH
NH
2.8
1.8
38
42
14
23
0.51
0.64
1
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10%
5
0
and the mean value is reported.
*
LLE. Ligand-lipophilicity efficiency (calculated as pIC -clogP).
5
0
Table 1. Inhibitory properties of class I analogues on AKR1C3 and AKR1C2.
Introduction of electron withdrawing groups onto the phenyl B ring (8b-d) resulted in compounds with similar
inhibitory profiles to 1 with only marginally decreased selectivity. Introduction of an electron-donating methoxy
18
ACS Paragon Plus Environment

Page 19 of 72
Journal of Medicinal Chemistry
group on the B-ring, gave compounds equipotent with the parent baccharin scaffold at all substitution positions.
However, a clear correlation between selectivity and substitution position was observed. When substitution occurs
at the para (8g) or ortho (8e) positions a two-fold loss of selectivity is apparent. When substitution occurs at the
meta position (8f) selectivity is retained. The derivatives 8e-g display a similar LLE profile as 1. (Table 2).
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
IC50
AKR1C3
IC50
AKR1C2
(μM)
Selectivity
Compound
R
LLE*
1
C2:1C3
(μM)
8
8
8d
8e
8
b
c
2-F
3-F
4-F
2-OMe
3-OMe
4-OMe
0.125
0.156
0.135
0.135
0.126
0.132
45
62
67
37
57
34
360
397
496
274
452
258
1.07
0.97
1.04
1.26
1.29
1.27
f
g
8
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10%
5
0
and the mean value is reported.
*
LLE. Ligand-lipophilicity efficiency (calculated as pIC -clogP).
5
0
Table 2. Inhibitory properties of class I analogues on AKR1C3 and AKR1C2.
Class II analogues: The central A-ring meta-substituted derivatives 19a, 26a and 26h showed enhanced
AKR1C3 inhibitory potency. The meta-ester derivative (19a) provided the best combination of potent inhibition
(IC = 0.088 µM) and selectivity in all of the derivatives discussed thus far, with 261-fold selectivity for AKR1C3
50
(Table 3). The meta-amide derivative (26a) is the most potent AKR1C3 inhibitor identified among all the
baccharin derivatives synthesized to date (IC = 0.066 µM), equipotent with the most active reported AKR1C3
5
0
26
enzyme inhibitors. Selectivity is diminished over the parent scaffold, but the compound still retains a 109-fold
selectivity for AKR1C3. Compounds that have mid-nanomolar potency for AKR1C3 and > 100 fold selectivity
are leads for further development. Moreover, 26a displayed an increase in LLE by two-fold as compared to 1 and
by 1.3-fold in comparison to its isomer 14a. Introduction of rigidity in the side chain amide chain via formation
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 72
of a trans alkene (26h) decreased the inhibitory potency for AKR1C3 (IC = 0.098 µM) as compared to 26a but
5
0
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
resulted in 50% increased selectivity. In comparison with the parent scaffold (1), 26h retained equipotency but
suffered a three-fold loss of selectivity. This data suggests that free rotation on this side chain is essential for high
selectivity. Replacement of the prenyl side chain with a 3-hydroymethylphenyl group (28) decreased inhibitory
potency (five-fold reduction over 1) and selectivity (34-fold loss over 1). Removal of the terminal methyl groups
on the prenyl chain, to form a terminal alkene 19b, also diminished potency and selectivity suggesting that the
methyl groups extend into an open sub-pocket in the active site of AKR1C3 to confer potency and selectivity.
Removal of the prenyl and ester side chains (17a) completely abrogated activity. Consistent with class I
analogues, the amide derivatives (26a, 26h and 28) exhibit more promising physicochemical traits as evidenced
by their LLE values when compared to the ester derivatives (17a and 19a-b) (Table 3).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
IC50
AKR1C3
IC50
AKR1C2
(μM)
Selectivity
1C2:1C3
LLE*
Compound
R1
R2
(μM)
1
7a
-Br
-OH
38
85.3
2
2.26
1.36
19a47
0.088
0.273
0.066
0.098
23
261
19b
44
7.2
14
161
109
143
1.80
2.69
2.11
2
6a47
26h
2
8
0.540
8.2
15
3.09
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10%
5
0
and the mean value is reported.
*
LLE. Ligand-lipophilicity efficiency (calculated as pIC -clogP).
5
0
20
ACS Paragon Plus Environment

Page 21 of 72
Journal of Medicinal Chemistry
Table 3. Inhibitory properties of class II analogues on AKR1C3 and AKR1C2.
1
2
3
4
5
6
7
8
Based on the enhanced potency of 26a we next sought to investigate the effect of substituents on the
terminal B-ring of the parent scaffold. Such derivatives exhibited a range of activities from compounds that were
slightly more potent than 1 (26b-g) to a reduction of approximately 1.5-fold in AKR1C3 inhibition potency (26i-
k). Introduction of fluorine displayed fairly similar potencies, comparable to 26a, with meta substitution (26c)
showing a slight advantage over the ortho and para positions. However, the selectivity decreased by
approximately 1.5-fold (26b-d). Interestingly, methoxy substitution at the para position (26g) increased isoform
selectivity as compared to 26a and all substitution patterns conserved inhibitory potency (26e-g). The inhibition
profiles of tosyl (26i and 26j) and hydroxy (26k and 26l) substituted analogues were comparable to 1 with only
a marginal decrease in potency, however selectivity decreased by 5-10 fold. The LLE value remained similar to
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
6a among all the derivatives except hydroxy substituted compounds (26k and 26l) that displayed an increased
LLE to more than three (a 2.4-fold increase over 1) (Table 4).
IC50
IC50
AKR1C2
Selectivity
Compound
26b
R
AKR1C3
μM)
LLE*
1
C2:1C3
(
(μM)
2-F
3-F
4-F
2-OMe
3-OMe
4-OMe
3-OTs
4-OTs
0.098
0.076
0.091
0.081
0.089
0.094
0.180
0.190
5.8
6
7.8
59
79
86
2.37
2.49
2.41
2.68
2.64
2.62
0.81
0.78
3.03
2
6c
6d
6e
6f
2
2
2
7
86
10
16
7.6
16
112
170
42
84
164
26g
26i
2
6j
2
6k
3-OH
0.140
23
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 72
4
-OH
0.110
14
127
2
6l
3.13
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10%
and the mean value is reported.
5
0
*
LLE. Ligand lipophilicity efficiency (calculated as pIC -clogP).
5
0
Table 4. Inhibitory properties of class II analogues on AKR1C3 and AKR1C2.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Rearrangement of the substituents on the central A ring to adopt a 1,3,5 substitution pattern greatly enhanced
the selectivity for AKR1C3 inhibition over the highly related AKR1C1, 1C2 and 1C4 isoforms, as compared to
1
. With the exception of 39b (Table 5) all compounds possessing this substitution pattern showed inhibitory
potency in the low sub-micromolar range, with the majority exhibiting comparable or greater potency than 1 (39a,
9a-h) (Tables 5 and 6). The meta-amide 39a displayed inhibitory potency similar to 1 with a 1.5-fold increase
4
in selectivity over AKR1C2, while the selectivity over other AKR isoforms also increased. Replacement of the
prenyl side chain with a 3-hydroymethylphenyl group (39b) diminished potency and selectivity over 1C2 by 100-
fold and 21-fold, respectively. When the amide was reteroinverted (49a), we observed isoform selectivity >2857-
fold over AKR1C2, an increase of 5.6-fold as compared to 1. Selectivity over the other isozymes, AKR1C1 and
AKR1C4, increased by 1.5 and seven-fold respectively when compared to 1. Moreover, 49a demonstrated an IC50
of 0.07 µM for AKR1C3, which is a 1.5-fold increase from 1. Additionally, 49a also exhibited a 1.6-fold increase
in LLE as compared to the meta-amide 39a. Replacement of the prenyl side chain with an allyl group (50) resulted
in a seven-fold reduction in isoform selectivity and four-fold reduction in potency in agreement with prior data
(19b). Although compound 50 did exhibit similar selectivity over AKR1C2 with a two-fold reduction in potency
as compared to our hit scaffold (1), replacement of the prenyl chain with a 3-hydroymethylphenyl moiety (51)
reduced AKR1C3 inhibitory potency (three-fold from 1) as well as selectivity (1.7-fold from 1), although to a
lesser extent than 39b and showed an improvement in LLE value (two-fold from 1). The meta-ester analogue (52)
decreased AKR1C3 inhibitory potency by 1.7-fold and selectivity over 1C2 by five-fold from its amide
counterpart (49a), while maintaining comparable selectivity (1C2) to 1 and equipotent inhibition for 1C3,
22
ACS Paragon Plus Environment

Page 23 of 72
Journal of Medicinal Chemistry
although LLE reduction was dramatic (3.4-fold as compared to 49a). The ether analogues (59a-59b) suffered
significant loss (18-46 fold) in inhibition selectivity over 1C2, a decrease in inhibition potency (two – four-fold)
and a slight reduction of LLE (1.01-1.1 fold) when compared to 49a (Table 5).
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
IC50
AKR isoform
Fold Selectivity
Compound
R1
R2
(μM)
LLE*
1C1:
C3
1C2: 1C4:
1
C1
1C2
1C3
1C4
1
1C3
1C3
1
187
51
0.10
21
1870
510
210
1.29
Baccharin
35.5%
inhibition
at 200
3
9a
94
0.13
206
>1538
N.D.
723
34
1584 1.53
N.D. 1.71
μM
1
.2
39b
N.D.
41
N.D.
0
No
39%
inhibition inhibition
4
9a
0.07
0.26
0.33
99
>2857 >2857 1414 2.09
at 200
at 200
μM
μM
5
0
N.D.
N.D.
88
N.D.
N.D.
N.D.
N.D.
400
N.D. 2.45
4
0%
inhibition
at 100
μM
51
>300 N.D. 2.57
4
7.0%
inhibition
at 200
μM
5
2
62
0.12
15
>1666
517
125
0.61
5
9a
N.D.
N.D.
26
18
0.16
0.29
N.D.
N.D.
N.D.
N.D.
163
62
N.D. 2.05
N.D. 1.87
5
9b
N.D.; Not Determined
AKR1C3 IC and fold selectivity over AKR1C2 highlighted in red.
50
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 72
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10% and
the mean value is reported.
5
0
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
*
LLE. Ligand lipophilicity efficiency (calculated as pIC -clogP).
5
0
Table 5. Inhibitory properties of class II analogues on AKR1C1–AKR1C4.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
When substitutions on the B-ring were coupled with homologation of the amide-containing chain, all the
substitutions were generally well tolerated, increasing the selectivity for AKR1C3 inhibition by three – seven-
fold in comparison to 1 and maintaining the inhibition potency equivalent to or greater than 1 (49b-f) (Table 6).
The 4-fluoro (49b) and 4-methoxy (49c) substitutions on the phenethyl side chain exhibited similar inhibitory
potency to 1, increasing the selectivity over 1C2 by three-fold and 2.4-fold respectively, without any significant
change in LLE over 1. Reduction of the carbon spacer length, along with substitution with N,N-dimethyl (49h),
methoxy (49d) or methyl (49g) groups at the 4-position on the benzyl side chain further increased the inhibitory
potency as compared to 1. Compounds 49h and 49d were three-fold more selective than 1, whereas 49g exhibited
the best combination of potency (IC = 0.07 µM) and selectivity (>2800, a seven-fold increase from 1). The 4-
5
0
trifuoromethoxy benzyl containing analogue (49e) exhibited a similar inhibition and selectivity profile as 49b.
However, movement of the -OCF group to the meta position (49f) diminished inhibitory potency by 1.2-fold and
3
selectivity over AKR1C2 by 1.5-fold compared to 49e. With the exception of 49d none of the derivatives
displayed an increase in the LLE value as compared to 49a (Table 6).
24
ACS Paragon Plus Environment

Page 25 of 72
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
IC50
AKR isoform
Selectivity
(μM)
LLE*
Compound n
R
1
C1
1C2
1C3 1C4 1C1:1C3 1C2:1C3 1C4:1C3
No
inhibition
at 200
μM
4
9b
2
2
1
4-F
148
0.10 140
>2000
>1802
>1315
>1802
1480
1236
1144
1500
1400
649
1.80
0.98
2.27
0.99
No
inhibition
at 200
μM
4
9c
4-OMe
4-OMe
136
87
0.11
0.07
72
60
No
inhibition
at 100
μM
4
9d
789
No
inhibition
at 200
μM
49e
1 4-OCF3
165
0.11 151
1360
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 72
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
10.6%
inhibition
at 200
4
9f
1 3-OCF3
129
0.13 133
0.07 145
>1538
992
1023
0.93
μM
No
66.5%
inhibition inhibition
at 200
49g
1
4-Me
>2817
>1493
>2817
1194
2042
880
1.72
1.27
at 200
μM
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
μM
No
inhibition
at 100
μM
4
9h
1 4-NMe2
80
0.06
59
IC values were performed in quadruplicate and technical replicates gave mean + SE. The SE is less than 10%
and the mean value is reported.
LLE. Ligand lipophilicity efficiency (calculated as pIC -clogP).
5
0
*
5
0
Table 6. Inhibitory properties of class II analogues on AKR1C1–AKR1C4.
Class III analogues: Boronic acid bioisosteres of carboxylic acids are expected to demonstrate increased binding
affinity to their molecular targets since they would provide a strong counterion in the oxyanion hole as compared
to ionic interactions with their carboxylic acid counterparts, as well as their ability to form covalent bonds with
amino acid residues.60, 61 Bioisosteric replacement of the carboxylic acid of our parent scaffold with a pinacol
boronate (7) or a boronic acid (7a) decreased AKR1C3 inhibitory potency by 66- and 23-fold respectively
compared to 1, with a concurrent 500-fold decrease in selectivity, in addition to exhibiting a poor LLE profile.
4
4
Removal of the prenyl moiety (24) abrogated bioactivity in agreement with our previous findings. Surprisingly,
when the amide chain was placed ortho relative to the carboxylic acid group (33a) a 1.4-fold increase in potency
was observed compared to 14a with comparable selectivity and a similar LLE as 1. Replacement of the prenyl
moiety in 33a with a 3-hydroymethylphenyl moiety (33b) diminished potency and selectivity by five-fold and
1
2-fold respectively in comparison to 33a without any appreciable impact in LLE (Table 7).
26
ACS Paragon Plus Environment

Page 27 of 72
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
IC50
IC50
Selectivity
1C2:1C3
LLE*
Compound
R1
R2
R3
R4
AKR1C3 AKR1C2
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(μM)
(μM)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
7
H
H
6.6
4.7
0.71
1.7
-2.89
0.26
7
a
4
2.3
3.8
2
-Br
H
16.0
109
6.8
-0.93
3
3a
H
H
0.2
1.0
39
14
163
14
1.26
1.79
3
3b
*
LLE. Ligand lipophilicity efficiency (calculated as pIC -clogP).
5
0
Table 7. Inhibitory properties of class III analogues on AKR1C3 and AKR1C2.
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 72
In this study, the inhibitory potency and selectivity of several analogues of 1 against AKR1C3 were
investigated. Compound 26a showed the most potent inhibition (IC : 66 nM) whereas compound 49a displayed
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
5
0
the highest selectivity for AKR1C3 over AKR1C2 (IC ratio: >2857). Analysis of crystal structures of
5
0
+
AKR1C3.NADP - inhibitor complexes have revealed that the ligand binding site can be dissected into five
+
subsites: the oxyanion site (consisting of catalytic residues Tyr55, His117 and cofactor NADP ), the steroid
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
channel for the binding of steroid ligands, and three sub-pockets (SPs): SP1, SP2 and SP3 to accommodate other
ligands.23, 26 In the AKR1C3.NADP .1 complex model (Figure 3A), the carboxylate group on the cinnamic acid
was predicted to occupy the SP1 pocket and could form hydrogen bonds with the side chain of Ser118, which
contributes to strong binding affinity. In addition, several polar amino acids (e.g. Ser308 and Tyr319) also reside
in the SP1 pocket of AKR1C3 which provide the possibility for ligand-protein interactions. These polar
interactions would be expected to increase the selectivity of 1 for AKR1C3 over the other AKR1C isoforms where
the amino acids in the corresponding positions of the SP1 pocket are Phe118, Leu308 and Phe319.23, 26The
docking model predicted that the benzyl moiety of the 4-dihydrocinnamoyloxy group was located in the SP3
pocket and could form hydrophobic interactions with the side chain of Gln222 or Phe306 to provide high binding
+
43
4
3
affinity. In addition, recent studies have shown that polar substitutions (e.g. hydroxyl and carboxylate group)
on the phenyl ring of the cinnamic acid, or the phenyl ring of the 4-dihydroxycinnamoyloxy group decreased
inhibition for AKR1C343, 45, 62. For example, the displacement of the 4-dihydrocinnamoyloxy group by a hydroxyl
group to form drupanin (1a) resulted in a loss of inhibitory potency (IC : 15 µM) and only a seven-fold selectivity
5
0
for AKR1C3 versus AKR1C2. Increase or decrease of carbon spacer along with replacement of the B-ring in
class I analogues exhibited low or sub-micromolar inhibition activities (8h-8n) but still exhibited >50-fold
isoform selectivity. In order to impart hydrolytic stability to our derivatives, replacement of the ester functional
group with an amide was performed. The corresponding amide bioisosteres displayed diminished potency and
selectivity (14a-14e) suggesting that the increased rigidity of the side chain amide is not tolerated at the para-
position of the A ring. The B-ring substitutions on the dihydrocinnamoyloxy moiety were very well tolerated and
only a marginal reduction of potency and selectivity was observed (8b-8g).
28
ACS Paragon Plus Environment

Page 29 of 72
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Modeling of parent compound (1) and the most selective analogue 49a. A) Model of the
+
+
AKR1C3.NADP .1 complex. AKR1C3 residues (light blue); NADP (pink); and baccharin (green). Dotted line:
+
possible hydrogen bond; OX: oxyanion site; SP: subpocket. The docking model of AKR1C3. NADP .1 complex
was constructed by calculation using the program Glide 5.0. The generated docking structure showing the highest
+
score was selected for further structural analysis. The crystal structure of AKR1C3.NADP complex was chosen
4
3
from RCSB protein data bank (PDB code: 1S2C). Reproduced with permision from reference. B) Quantitative
structure activity-relationship prediction of analogue 49a (gold) with amino acid residues of the AKR1C3 binding
site (grey). Green dotted line: strong predicted hydrogen bond; white dotted line: weak predicted hydrogen bond.
Model constructed with SeeSAR 8.1 (BioSolveIT GmbH) using AKR1C3 protein from RCSB protein data bank
(PDB code: 3UG8).
When the substitution pattern of the dihydrocinnamoyloxy side chain was changed to the meta-position
analogues 19a and 26a) the inhibitory potency increased in comparison to 1. The presence of either a carbonyl
(
or carboxylate group is required to anchor many ligands to the oxyanion site because it can form a strong
hydrogen-bond with Tyr55 and His117 and bring the ligands into close proximity of the nicotinamide head group
of the cofactor.26 Based on the model of baccharin-docked into the AKR1C3-NADP complex structure (Figure
+
4
3
3A), the carbonyl of the 4-dihydrocinnamoyloxy group of 1 was found to be close to the oxyanion site and could
form a hydrogen bond interaction with Tyr55 and His117, which could contribute to the high inhibitory potency.
Thus, it seems plausible that in the meta substituted analogues, the change in the position of the carbonyl group
29
ACS Paragon Plus Environment