Design and Activity of Specific Hypoxia-Inducible Factor-2α (HIF-2α) Inhibitors for the Treatment of Clear Cell Renal Cell Carcinoma: Discovery of Clinical Candidate (S)-3-((2,2-Difluoro-1-hydroxy-7-(methylsulfonyl)-2,3-dihydro-1 H-inden-4-yl)oxy)-5-fluorobenzonitrile (PT2385)

Article abstract of DOI:10.1021/acs.jmedchem.8b01196

HIF-2α, a member of the HIF family of transcription factors, is a key oncogenic driver in cancers such as clear cell renal cell carcinoma (ccRCC). A signature feature of these cancers is the overaccumulation of HIF-2α protein, often by inactivation of the E3 ligase VHL (von Hippel-Lindau). Herein we disclose our structure based drug design (SBDD) approach that culminated in the identification of PT2385, the first HIF-2α antagonist to enter clinical trials. Highlights include the use of a putative n → π?_Ar interaction to guide early analog design, the conformational restriction of an essential hydroxyl moiety, and the remarkable impact of fluorination near the hydroxyl group. Evaluation of select compounds from two structural classes in a sequence of PK/PD, efficacy, PK, and metabolite profiling identified 10i (PT2385, luciferase EC₅₀ = 27 nM) as the clinical candidate. Finally, a retrospective crystallographic analysis describes the structural perturbations necessary for efficient antagonism.

Full text of DOI:10.1021/acs.jmedchem.8b01196

Subscriber access provided by Kaohsiung Medical University
Article
Design and Activity of Specific Hypoxia-Inducible Factor 2# (HIF-2#)
Inhibitors for the Treatment of Clear Cell Renal Cell Carcinoma:
Discovery of Clinical Candidate (S)-3-((2,2-difluoro-1-hydroxy-7-
methylsulfonyl)-2,3-dihydro-1H-inden-4-yl)oxy)-5-fluorobenzonitrile (PT2385)
(
Paul Manning Wehn, James P. Rizzi, Darryl D. Dixon, Jonas Grina, Stephen T. Schlachter, Bin Wang,
Rui Xu, Hanbiao Yang, Xinlin Du, Guangzhou Han, Keshi Wang, Zhaodan Cao, Tzuling Cheng,
Robert M. Czerwinski, Barry S. Goggin, Heli Huang, Megan M. Halfmann, Melissa A. Maddie, Emily
L. Morton, Sarah R. Olive, Huiling Tan, Shanhai Xie, Tai Wong, John A. Josey, and Eli M. Wallace
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01196 • Publication Date (Web): 05 Oct 2018
Downloaded from http://pubs.acs.org on October 6, 2018
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 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Design and Activity of Specific Hypoxia-Inducible Factor 2a (HIF-2a)
Inhibitors for the Treatment of Clear Cell Renal Cell Carcinoma:
Discovery of Clinical Candidate (S)-3-((2,2-difluoro-1-hydroxy-7-
(methylsulfonyl)-2,3-dihydro-1H-inden-4-yl)oxy)-5-fluorobenzonitrile
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
(
PT2385)
†
‡
Paul M. Wehn,* James P. Rizzi, Darryl D. Dixon, Jonas A. Grina, Stephen T. Schlachter, Bin
Wang, Rui Xu, Hanbiao Yang, Xinlin Du, Guangzhou Han, Keshi Wang, Zhaodan Cao, Tzuling
#
⊥
Cheng, Robert M. Czerwinski, Barry S. Goggin, Heli Huang, Megan M. Halfmann, Melissa A.
♦
|
Maddie, Emily L. Morton, Sarah R. Olive, Huiling Tan, Shanhai Xie, Tai Wong, John A. Josey,
Eli M. Wallace
Peloton Therapeutics, Inc., 2330 Inwood Road, Suite 226, Dallas, TX 75235
‡
†#⊥♦|
ABSTRACT
HIF-2a, a member of the HIF family of transcription factors, is a key oncogenic driver in
cancers such as clear cell renal cell carcinoma (ccRCC). A signature feature of these cancers is
the over accumulation of HIF-2a protein, often by inactivation of the E3 ligase VHL (von
Hippel-Lindau). Herein we disclose our structure based drug design (SBDD) approach that
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 103
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
culminated in the identification of PT2385, the first HIF-2a antagonist to enter clinical trials.
Highlights include the use of a putative n ® p* interaction to guide early analog design, the
Ar
conformational restriction of an essential hydroxyl moiety, and the remarkable impact of
fluorination near the hydroxyl group. Evaluation of select compounds from two structural classes
in a sequence of PK/PD, efficacy, PK, and metabolite profiling identified 10i (PT2385,
Luciferase EC = 27 nM) as the clinical candidate. Finally, a retrospective crystallographic
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
0
analysis describes the structural perturbations necessary for efficient antagonism.
INTRODUCTION
1
The hypoxia-inducible factor (HIF) family of transcription factors consists of HIF-1a ,
2
,
3
HIF-2a (EPAS1) and the less characterized HIF-3a . Under normoxia, specific proline
residues on the HIFa proteins are hydroxylated by the oxygen-dependent dioxygenase HIF-
4,5
specific prolyl-hydroxylases (PHDs). These post-translational modifications provide a
mechanism to control transcriptional activity as they initiate the ubiquitin-proteasome
degradation pathway by creating a substrate recognition site for von Hippel-Lindau protein
(pVHL). pVHL is a tumor suppressor and part of an E3 ubiquitin-ligase complex that binds to
the hydroxylated HIFa proteins, targeting them for proteasomal degradation. Under hypoxia, the
PHD enzymes are inactive and, as a result, the HIFa proteins accumulate and translocate to the
nucleus where they dimerize with the constituatively expressed aryl hydrocarbon receptor
nuclear translocator (ARNT, also known as HIF-1b) to form an active transcription factor
complex. Binding of the resulting HIFa-ARNT heterodimers to hypoxia-responsive elements
(HRE) on target genes leads to the expression of hundreds of genes, many of which coordinately
regulate anaerobicꢀmetabolism, angiogenesis, cell proliferation and survival, immune evasion,
ACS Paragon Plus Environment
2

Page 3 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
extracellular matrix remodeling, pH homeostasis, amino acid and nucleotide metabolism, and
6
,7,8
genome stability.
HIFa driven transcriptional activity has been implicated in many diseases including
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
9
10
11
12
cancer , pulmonary arterial hypertension , iron overload and inflammatory bowel disease .
Importantly, in clear cell renal cell carcinoma (ccRCC) HIF-2a activity has been demonstrated
to be a key oncogenic driver. In ccRCC, VHL inactivation is found in more than 90% of
patients, resulting in the accumulation and transcriptional activation of HIF-2α, even under
13
normoxic conditions. Several studies support the pathophysiological role of HIF-2a in ccRCC.
Knockdown of HIF-2a expression in VHL defective ccRCC xenografts impedes tumor formation
1
4
comparable to reintroduction of VHL and overexpression of a stabilized variant of HIF-2a
1
5,16
alone is adequate to overcome pVHL’s tumor suppressive effect
. Significantly, the status of
HIF-1a appears inconsequential in the course of ccRCC: HIF-1a is often deleted or mutated; and
17,18,19
a stabilized variant of HIF-1a does not inhibit tumor suppression by pVHL.
This evidence
implicates HIF-2a as the tumorigenic driver in ccRCC and suggests that HIF-2a antagonists
could provide novel therapeutic intervention for its treatment.
The HIF-2a protein is a member of a larger family of transcription factors that contain a
specific Helix-Loop-Helix-PAS-A-PAS-B fold (HLH-A-B). The pioneering work of Bruick and
Gardner showed that the inner core of the PAS-B domain of HIF-2a possesses a water-filled
2
0
hydrophobic cavity that can bind small molecules. They also demonstrated that binding small
2
1
molecule ligands in this pocket could disrupt the HIF-2a dimer complex with ARNT . The
ligands identified included two independent series; benzoxadiazole 1a and tetrazole 2 (Figure
2
2,23
1
)
. These analogs bound to the inner pocket with a Kd of 90 and 23 nM, respectively, based
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 103
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
on isothermal titration calorimetry analysis (ITC). Protein-inhibitor co-crystal structural
2
4
determination of truncated HIF-2a and ARNT PAS-B domains showed that the structurally
distinct analogs both bound to the inner pocket of HIF-2a. Herein we describe the design and
activity of specific, potent, orally bioavailable HIF-2a inhibitors that culminated in the discovery
of PT2385; the first HIF-2a antagonist to enter clinical development.
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
N N
F
N
F
NO2
HN
N
H
N
Cl
F
NC
O
OH
Me
N
N
O
S
O O
X
Br
Br
F
1
1
a: X = F, ITC K = 90 nM
2: ITC K = 23 nM
10i
(PT2385)
d
d
cLogP = 3.99
cLogP = 4.40
b: X = H, ITC K = 160 nM
d
Figure 1. Literature HIF-2a inhibitors that formed the basis of SBDD efforts to identify PT2385
RESULTS AND DISCUSSION
Initial Inhibitor Design
To begin our own structure-based design effort to discover novel, potent and selective
orally bioavailable HIF-2a antagonists it was necessary to develop a binding assay other than
ITC, which is rather low throughput. We turned, instead, to a competitive binding assay with a
radiolabeled ligand. Compound 1b, a slightly less potent analog (160 nM) of benzoxadiazole 1a,
was labeled on the B-ring with two tritium atoms (Scheme 1) and a scintillation proximity assay
(SPA) with the isolated PAS-B* domain of HIF-2a was developed and utilized to drive SAR.
The HIF-2a PAS-B* domain is the R247E mutant. When combined with the ARNT PAS-B*
ACS Paragon Plus Environment
4

Page 5 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
domain (E362R mutant), enhanced interfacial salt bridge interactions permit facile formation of
2
0
the HIF-2a/ARNT heterodimer and enable crystallization studies where the wild-type fails.
Scheme 1. Generation of radiolabeled HIF-2a inhibitor for the SPA Assay
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
3H2
MeOH/Et N
Cs CO3
DMF, 60 °C
NO2
2
H
N
Cl
NO2
I
Cl
NH2
T
Cl
3
N
N
O
Perkin-Elmer
proprietary
metal catalyst
NO2
I
T
T
T
Cl
N
3
B-Ring
A-Ring
O
3
N
H -1b
2
Although ligand binding is a necessary component to block the function of HIF-2a, it
may not be sufficient. One can envision a scenario where a compound binds to the inner pocket
of HIF-2a but does not sufficiently distort the protein to block dimerization with ARNT. Thus,
functional assays were needed to establish that binding of compounds to the PAS-B domain of
HIF-2a would block dimerization with ARNT and subsequent transcription of the downstream
genes it regulates, thereby acting as true antagonists. For this, two cellular assays were utilized
in the ccRCC cell line 786-O. 786-O is a VHL-mutant cell line with constitutively active HIF-
2a and non-functional HIF-1a, thereby providing a well-defined cellular environment to
evaluate specific HIF-2a activity. The first cell assay was a HIF-2a-driven HRE-dependent
luciferase that provided good throughput for SAR support. The second cell assay was
measurement of vascular endothelial growth factor A (VEGFA) secretion. VEGFA is a well-
established target gene of HIF-2a in ccRCC and provided a physiological readout of cellular
activity.
Analysis of the disclosed HIF-2a antagonists, 1a and 2, showed that the two aromatic
rings were important for binding and occupied approximately the same space. Our starting point
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 103
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
for a structure based design effort focused on nitrobenzoxadiazole 1a. Of the two hit classes it is
less lipophilic with significantly lower molecular weight. Additionally, the highly-polarized
nature of the nitrobenzoxadiazole heterocycle in 1a lead us to our first hypothesis: that the low
electron density of the nitrobenzoxadiazole was an important feature for binding by enhancing a
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
5
potential n ® p*Ar interaction between a tyrosine oxygen lone pair and the benzoxadiazole.
Table 1. Replacements to the benzoxadiazole
NO2
H
N
Cl
X
a
F
X
SPA IC50 (
µ
M)
sp
4
a
SO CF
<0.005
0.96(78)
0.86(106)
-
2
3
4
b
SO CHF
0.016 ± 0.009
0.28 ± 0.11
4.2 ± 1.3
2
2
4
c
SO CH F
2 2
b
4d
SO NHMe
0.65(233)
0.54(70)
2
4
e
CF3
0.28 ± 0.10
12.7 ± 4.6
4f
CH(OH)Me
-0.07(222)
NO2
H
N
Cl
N
O
-
0.49 ± 0.16
-
N
F
1
a
a
s values as found in table 1 from reference 27. The number in parentheses denotes the line where
p
b
the s value can be found. Data for –SO NMe .
p
2
2
ACS Paragon Plus Environment
6

Page 7 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Examination of the published protein co-crystal structure of 1a in HIF-2a PAS-B
2
6
domain revealed an electrostatic contact of 3.32 Å between the phenolic oxygen of Tyr281 and
the centroid of the benzene portion of the benzoxadiazole heterocycle which was well within the
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
25
2
.8–3.8 Å distance reported for n ® p*Ar interactions . A variety of groups replacing the
oxadiazole portion at the 4-position were tested and found to corroborate the initial hypothesis:
2
7
the greater the electron deficiency of the replacement, the better the binding (Table 1). N-(3-
chloro-5-fluorophenyl)-2-nitro-4-((trifluoromethyl)sulfonyl)aniline 4a, with the largest electron
deficiency in this group, significantly improved binding affinity to < 5 nM. Other sulfone
analogs with less fluorination (4b and 4c), hence reduced electron deficiency, showed
intermediate binding affinity between 4a and 1a. In addition, the methyl sulfonamide (4d),
trifluoromethyl (4e) and alcohol (4f) replacements showed reduced affinity. Examination of the
co-crystal structure of 4a (Figure 2) supported the importance of the electrostatic interaction
between Tyr281 and the antagonist (3.46 Å), and also revealed that the trifluoromethylsulfone
moiety can twist out of the plane and fill a relatively small hydrophobic patch in the protein.
This new interaction likely contributed to the high binding affinity of 4a.
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 103
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
Figure 2. (A) X-ray crystal structure of 4a (orange) in complex with the HIF-2a PAS-B*
(white)/ARNT PAS-B* (green) dimer (PDB:6D0B). (B) Close up view of 4a in the HIF-2a
binding pocket with Tyr281 and His293 highlighted (magenta).
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
The increased affinity of 4a afforded the opportunity to replace the metabolically
22
problematic aniline moiety with an aryl ether (Table 2). In contrast to the reported SAR on 1a ,
aryl ether replacement of the aniline was tolerated with only a modest loss in potency (5a).
Evaluation of the SAR of the A-ring 2-position showed that the undesirable nitro moiety could
be replaced. Binding affinity was retained with small electron-withdrawing groups at this
position (5b, 5c, 5d, 5h). Importantly, several of these analogs were active in cells with
micromolar potency in the luciferase assay indicating functional activity. 5h stood out, showing
a nominal shift between the binding and cell assays. Larger, more hydrophilic analogs were
detrimental to binding even though some were electron-withdrawing (compounds 5e, 5f, 5g).
Compound 5i with one charged group was completely inactive. Modifications to examine
lipophilicity included replacement of the trifluoromethylsulfone with a difluoromethylsulfone
(5j) and methylsulfone (5k). While 5j showed a comparable binding affinity to 5b, the cLogP
improved by more than one unit and the shift in the cell based assay was smaller. Despite further
improvement in lipophilicity for 5k, the decreased electron withdrawing potential of the
methylsulfone proved deleterious to binding affinity.
ACS Paragon Plus Environment
8

Page 9 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. SAR for the replacement of the nitro group
X
Cl
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
SO Y
Luciferase EC50
2
F
X
Y
SPA IC50 (
µ
M)
(
µ
M)
cLogP
5
a
NO2
Br
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CHF2
Me
0.032 ± 0.002
0.17 ± 0.07
0.14 ± 0.06
0.034 ± 0.006
52.9 ± 15.3
22.6 ± 12.6
1.2 ± 0.5
-
5.03
6.20
6.05
5.00
5.29
4.01
4.53
5.76
4.88
5.03
4.09
5
b
3.7 ± 1.1
5
c
Cl
3.4 ± 1.3
5
d
CN
11.6 ± 3.4
5
e
CO Me
-
2
5
f
CONH2
-
5
g
CH OH
-
2
5
h
CHF2
0.22 ± 0.04
44.4 ± 7.3
0.24 ± 0.10
5
i
CO H
-
2
5
j
Br
Br
0.18 ± 0.05
4.4 ± 0.4
2.2 ± 0.4
5
k
-
Next, we turned to the 3-position of the A-ring as a variety of crystal structures, both apo
form and those in complex with different ligands, indicated a degree of flexibility for His293 and
Met252 sidechains. Based on this observation we hypothesized that installation of larger and
possibly more hydrophilic substituents could be incorporated into our inhibitors. A series of
tetra-substituted A ring analogs were thus prepared (Table 3). This area of the protein was not as
tolerant as our structural analysis predicted, with many analogs resulting in weakly active
compounds. Introduction of an amino methylene at the 3-position (6a) was poorly tolerated and
acylation of the amino group (6b) ablated the remaining binding affinity. A very important
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 103
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
finding, however, was that benzyl alcohols 6c and 6d were active with improved potency to the
unsubstituted compound 5j. These derivatives have a smaller shift from binding to luciferase
cell activity and have reduced lipophilicity. Interestingly, methylation of the hydroxyl (6e)
dropped the binding affinity two orders of magnitude highlighting the importance of the alcohol
moiety to binding. To get a better understanding of the interactions of the benzyl alcohol,
compound 6c was co-crystalized with HIF-2a and the key interactions are shown in Figure 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
Table 3. SAR for the introduction of hydrophilic substituents
X
Cl
O
Y
SO CHF2
SPA IC50
M)
Luciferase EC50
2
(
µ
(
µ
M)
cLogP
F
X
Y
6
a
Br
Br
Br
Cl
Cl
CH NH
0.70 ± 0.10
18.2 ± 9.4
-
-
3.99
3.84
4.00
3.85
4.68
2
2
6
b
CH NHAc
2
6
c
CH OH
0.055 ± 0.023
0.046 ± 0.018
4.2 ± 2.9
0.33 ± 0.14
0.38 ± 0.07
-
2
6
d
CH OH
2
6
e
CH OMe
2
ACS Paragon Plus Environment
1
0

Page 11 of 103
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. Close up view of X-ray crystal structure of 6c (orange) in complex with the HIF-2a
PAS-B*/ARNT PAS-B* dimer (PDB:6CZW). Tyr281 and His293 are highlighted (magenta).
Dotted lines indicate hydrogen bond interactions and the red sphere denotes a water molecule.
Compound 6c bound to the inner cavity of HIF-2a as expected, however, the hydroxyl
group participated in a unique H-bonding network between Tyr281, a water molecule, and
His293. The other interesting observation was that the difluoromethyl group rotated 180 degrees
away from the lipophilic patch discussed earlier. Initially, we thought that the difluoromethyl
sulfone rotated in order to make potential dipole interactions with the surrounding environment
2
8
much like a hydrogen bond. However, a close analysis of the X-ray for 6c showed potential
contacts were too distant and that the most likely explanation for the differeing orientations of
the sulfones in 4a and 6c is sterics. A consequence of the new sulfone orientation is a slight
elongation (3.55 Å) of the Tyr281 contact to the aromatic system of the A-ring.
ACS Paragon Plus Environment
1
1

Journal of Medicinal Chemistry
Page 12 of 103
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
In addition to providing a lead with more favorable physical properties, the rigid
hydrogen-binding network formed between alcohol 6c and HIF-2a PAS-B domain may
contribute to the observed enhancement in functional activity of the antagonist. Furthermore, the
interactions of the primary alcohol inspired analog designs that constrained this moiety in the
desired binding mode, thereby alleviating the entropic penalty incurred to adopt the preferred
conformation. These designs led to two novel cyclic secondary alcohol scaffolds highlighted in
Scheme 2. Cyclic sulfone derivatives arose from Path A whereas Path B yielded carbocyclic
alcohol analogs. A consequence of the proposed inhibitor designs would be a strong dependence
of potency on alcohol stereochemistry.
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
Scheme 2: Two paths (A and B) to constrain rotation of the benzyl alcohol
Br
OH
Cl
O
Path A
Path B
(
CH )
2 n
S
Br
OH
O
O
F
Cl
O
SO CHF
2
2
(CH2)n
F
Cl
O
OH
6
c
CHF2
S
O O
F
Path A: Development of SAR for the cyclic sulfone Scaffold
Evaluation of the initial 5-membered cyclic sulfone 7a (a 2,3-dihydrobenzo[b]thiophene
1,1-dioxide) proved disappointing with SPA potency 40-fold less than the parent benzyl alcohol
6
c (Table 4). Despite the discouraging start, our familiarity with the SAR of the earlier non-
cyclized compounds suggested that modulation of the benzene by installation of electron
withdrawing groups could increase potency. The first attempt at modulation was to install
ACS Paragon Plus Environment
1
2

Page 13 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
geminal fluorine substitution alpha to the sulfone. Not only would such a modification impart
decreased electron density at the benzene ring, it would be the most isoelectronic with benzyl
alcohol 6c. Gratifyingly, introduction of fluorine increased potency in dramatic fashion (7b).
The magnitude of the increase, >100 fold in luciferase cell assay, suggested that the fluorine
atoms were acting as more than just electron withdrawing groups. Comparing potency of earlier
uncyclized analogs methyl sulfone 5k and difluoromethylsulfone 5j (Table 2), the increase in
potency from fluorination was only 20-fold.
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
ACS Paragon Plus Environment
1
3

Journal of Medicinal Chemistry
Page 14 of 103
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
Table 4. Initial SAR of the cyclic sulfone scaffold
R
OH
Cl
O
X
X
S
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
SPA IC50 (
µ
M)
Luciferase EC50 ( M)
µ
F
R
X
7
a
Br
H
2.4 ± 1.5
7.2 ± 3.8
7b
Br
Br
F
Me
F
<0.005
0.043 ± 0.011
-
7
c
19.6 ± 3.9
0.020 ± 0.005
0.020 ± 0.013
0.049 ± 0.017
1.1 ± 0.7
7d
Cl
0.11 ± 0.03
0.34 ± 0.10
9.2 ± 4.9
-
7
e
Me
CN
H
F
7
f
F
7
g
F
7
h
CF3
CHF2
F
<0.005
0.10 ± 0.03
0.017 ± 0.023
7
i
F
<0.005
F HC
OH
2
F
O
-
-
3.2 ± 1.8
-
S
O O
F
7
j
Recent literature on the impact of alpha fluorine atoms on neighboring alcohols suggest
that in addition to influencing alcohol pK , the fluorine atoms may engage in intramolecular
A
C−F···H−O electrostatic interactions (controversially, hydrogen bonds) that shield and restrict
29
interactions with the larger environment. Quantum chemical calculations of constrained 3- and
ACS Paragon Plus Environment
1
4

Page 15 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
-fluorocyclohexanol derivatives indicate a distinct preference for conformers of the alcohol O-
3
0
H that locate the proton in the vicinity of the fluorine atom. Experimentally, such speculation
1
9
1
is furthered by the observation of F coupling to the alcohol proton in H NMR of these
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
30
constrained fluorocyclohexanol derivatives. In the hydrophobic binding pocket of the HIF-2a
protein, the suggested electrostatic interactions imparted by fluorine could directly influence the
localized solvent shell surrounding the hydroxyl moiety and reduce the desolvation penalty
required for ligand binding and thereby enhance potency. Regardless of the exact contributions
of the fluorine atoms, their positive impact was clear. Supporting electron withdrawing effects
as the explanation for potency enhancement with fluorination, methylation ablated activity (7c)
potentially suggesting additional detrimental steric interactions with the protein.
Exploration of the 2-position of the A-ring demonstrated several substituents were
tolerated with –CHF (7i) affording the best potency (Table 4). The impact of ring expansion to
2
the six-membered cyclic sulfone ring was also examined and determined to be deleterious (7j).
A focused evaluation of the B-ring SAR identified 3-cyano-5-fluorophenyl (8b) as providing the
best balance of cellular activity and lipophilicity (Table 5). The dihalogenated B-rings in 7i and
8a showed excellent celluar activity, but with increased lipophilicity. The pyridine derivatives
8
c and 8d reduced lipophilicity, however, cellular activity suffered. Separation of the
enantiomers of the most promising cyclic sulfone, 8b, confirmed our speculation that binding
and cellular activity should reside in one stereoisomer (Scheme 3). X-ray crystal structure of the
active enantiomer 9a (Figure 4) recapitulated the binding features of benzyl alcohol 6c (Figure
3
). Interestingly, the phenol of Tyr281 achieved a closer contact with the aromatic system of the
A-ring of 9a than 6c, realizing a distance of 3.30 Å to the centroid.
ACS Paragon Plus Environment
1
5

Journal of Medicinal Chemistry
Page 16 of 103
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
Table 5. SAR of the aryloxy moiety
CHF2
OH
O
F
Ar
S
F
SPA IC50
Luciferase
M)
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
Ar
(
µ
M)
EC50 (
µ
cLogP
Cl
F
7i
<0.005
0.017 ± 0.023
0.013 ± 0.002
0.032 ± 0.012
4.89
F
8
a
0.011 ± 0.003
0.014 ± 0.003
4.32
3.61
F
NC
8
b
F
N
N
Cl
8
c
0.35 ± 0.26
0.14 ± 0.07
0.63 ± 0.31
0.18 ± 0.02
3.34
2.17
NC
8d
Scheme 3: Resolution of racemic 8b and testing of both enantiomers
F
F
F
F
F
F
OH
OH
OH
Resolution
NC
O
NC
O
NC
O
F
F
F
S
F
S
F
S
F
ChiralPak (AS-H)
20% EtOH/CO2
O
O
O
O
O
O
F
F
F
1
00 bar
8b
9a (99% ee)
SPA IC50 = <0.005 µM
Luciferase EC50 = 0.010 ± 0.005 µM
9b (99% ee)
SPA IC50 = 1.7 ± 0.6 µM
SPA IC50 = 0.014 ± 0.003 µM
Luciferase EC50 = 0.032 ± 0.012 µM
ACS Paragon Plus Environment
1
6

Page 17 of 103
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 4. Close up view of X-ray crystal structure of 9a (orange) in complex with the HIF-2a
PAS-B*/ARNT PAS-B* dimer (PDB:6D09). Tyr281 and His293 are highlighted (magenta).
Dotted lines indicate hydrogen bond interactions and the red sphere denotes a water molecule.
Path B: Development of SAR for the carbocyclic alcohol scaffold
Additional efforts to constrain the benzyl alcohol of compound 6c involved utilizing 5-
member ring indane derivatives represented in Scheme 2 as Path B, where n=1. The initial
evaluation of the unsubstituted indanol isomer (10a) demonstrated that this scaffold can achieve
significant binding affinity and reasonable functional activity (Table 6). Evaluation of single
enantiomers 10b and 10c showed activity residing in one stereoisomer as observed with the
cyclic sulfone scaffold. Installation of a geminal difluoro group (10d) resulted in a nearly 10-
fold improvement in potency. While not as dramatic as the >100-fold improvement observed in
the cyclic sulfone series, fluorination improved the indanol series to pharmacologically relevant
ACS Paragon Plus Environment
1
7

Journal of Medicinal Chemistry
Page 18 of 103
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
potencies. The differing impact of fluorination between the cyclic sulfone and indanol scaffolds
could reflect the possibility of dual roles in the cyclic sulfone scaffold and a singular effect in the
indanol. In the cyclic sulfone scaffold, fluorination can enhance the electron-withdrawing
capacity of the sulfone and alter the environment of the alcohol through electrostatic interactions
with the adjacent fluorine. By contrast, the indanol (10d) already has fluorination at the sulfone
moiety and the installation of fluorine adjacent to the alcohol primarily impacts the electrostatic
interactions of the alcohol. Comparison of indanol 10a and cyclic sulfone 7a (Table 4) further
illustrates this point. As observed in the cyclic sulfone scaffold, replacement of the chloro group
with cyano (10e) on the B-ring was well tolerated for the indanol series.
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
Unlike the cyclic sulfone series where fluorination of the sulfone proved essential,
imparting both electrostatic interactions with the alcohol and inductive effects on the phenyl ring,
the potency enhancement of the difluoro-indanol allowed independent evaluation of the sulfone
substitution. As expected, increased fluorination of the sulfone increased potency, with the
trifluoromethyl sulfone (10g) exhibiting single-digit nanomolar potency in cells (Table 6).
However, the increased fluorination also significantly increased the lipophilicity. It is important
to highlight that whereas the methyl sulfone (10i) was the least potent in this sub-series, with an
EC of 27 nM in cells, it demonstrated a desirable balance of properties with cLog P < 2.
5
0
Monofluoro indanols were also investigated. The cis-fluorohydrin (10k) retained significant
activity, whereas the trans-fluorohydrin (10l), with enhanced acidity of the alcohol and unable to
engage in intramolecular C−F···H−O interactions, was significantly less potent. Finally, ring
expansion to the six-membered ring (10m) resulted in a substantial loss of binding.
ACS Paragon Plus Environment
1
8

Page 19 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 6. SAR of the indanol scaffold
X
Y
R
O
*
OH
Z
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
S
SPA IC50
M)
Luciferase
EC50 ( M)
O O
(
µ
µ
cLogP
F
R
X
Y
Z
(
±)-10a
R)-10b
S)-10c
±)-10d
±)-10e
S)-10f
S)-10g
S)-10h
Cl
Cl
H
H
H
F
F
F
F
F
F
F
H
F
H
H
H
F
F
F
F
F
F
F
F
H
CHF2
CHF2
CHF2
CHF2
CHF2
CHF2
CF3
<0.005
<0.005
0.18 ± 0.05
0.11 ± 0.01
2.9 ± 0.7
3.78
3.78
3.78
4.06
2.78
2.78
3.95
2.01
1.84
2.37
1.39
1.39
(
(
Cl
0.46 ± 0.08
<0.005
(
Cl
0.020 ± 0.003
0.015 ± 0.001
0.006 ± 0.001
<0.005
(
CN
CN
CN
CN
CN
CN
CN
CN
0.010 ± 0.003
<0.005
(
(
<0.005
(
CH2F
Me
<0.005
0.009 ± 0.003
0.027 ± 0.006
0.088 ± 0.036
0.039 ± 0.021
0.49 ± 0.04
(
S)-10i
0.012 ± 0.005
0.031 ± 0.015
0.041 ± 0.007
0.35 ± 0.03
(
S)-10j
Et
(
1S,2R)-10k
Me
(
1S,2S)-10l
Me
F
S
F
NC
O
OH
Me
-
-
-
-
3.6 ± 1.3
-
2.40
O O
F
(±)-10m
F
F
Cl
O
OH
-
-
-
-
0.060 ± 0.048
0.065 ± 0.031
2.42
CHF2
N
S
O O
(S)-10n
ACS Paragon Plus Environment
1
9

Journal of Medicinal Chemistry
Page 20 of 103
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
X-ray co-crystal structure of 10i (Figure 5) revealed an identical interaction paradigm
with earlier alcohol analogs. Interestingly, the methyl sulfone moiety maintained a similar
orientation as the difluoromethyl sulfone moiety in 6c (Figure 3) and suggested that the original
trifluoromethyl sulfone lead, 4a (Figure 2), adopted the opposite orientation due to unfavorable
contacts in this region. Lastly, Tyr281 maintains a contact with aromatic portion of the A-ring in
10i, albeit elongated relative to cyclic sulfone 9a, at 3.53 Å from the centroid.
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 5. Close up view of X-ray crystal structure of 10i (orange) in complex with the HIF-2a
PAS-B*/ARNT PAS-B* dimer (PDB:5TBM, pink). Tyr281 and His293 are highlighted
(magenta). Dotted lines indicate hydrogen bond interactions and the red sphere denotes a water
molecule.
Pharmacokinetic (PK), Pharmacodynamic (PD), and In Vivo Efficacy of Select Inhibitors
ACS Paragon Plus Environment
2
0

Page 21 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Based on potency and lipophilicity, four indanols and one cyclic sulfone were selected
for further evaluation. Initial evaluation included potency in cellular VEGFA secretion assay,
permeability, mouse microsomal stability, plasma protein binding and mouse pharmacokinetics
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
(Table 7). While 10i was the least potent compound in this group in the VEGFA secretion assay
with an EC50 of 41 nM, the free-fraction adjusted cell EC50 was comparable to 9a, 10f and 10h
due to its lower lipophilicity and significantly decreased plasma protein binding. All five
compounds exhibited good mouse oral bioavailability and low to medium in vivo clearance as
predicted by high cell permeability and good microsomal stability.
ACS Paragon Plus Environment
2
1

Journal of Medicinal Chemistry
Page 22 of 103
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
Table 7. Functional cell activity and mouse PK parameters for select inhibitors
Compound
9a
16 ± 2
89
10f
14 ± 4
94
10h
20 ± 5
89
10i
41 ± 12
74
10n
37 ± 24
94
VEGFA EC (nM)
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
mouse PPB (%)
Free Fraction
145
233
182
158
617
adjusted EC (nM)
50
1
2h plasma
0.22
5.9
0.38
32.2
1.10
26.9
0.46
19.9
0.56
20.1
concentration (µM)
CL mouse
int
(mL/min/kg)
Papp A→B (nm/s)
B→A/A→B ratio
26
0.82
28.4
0.63
27.3
1.04
29.8
0.91
27.9
0.83
a
CL
7
0.6
34.3
7.30
7.4
32.3
9.46
12.6
5.5
29.2
16.43
7.0
58.2
5.44
25
(
mL/min/kg)
AUC0-t
b
6
.1
.5
.5
(
hr*µM)
VSS
a
9
(
L/kg)
t1/2
b
6
7.3
3.3
5.3
(h)
b
%F
110
67
91
110
101
a
3
mg/kg intravenous dose (solution in 20% EtOH, 40% PEG400, 40% water) in male CD1 mice
b
(
n = 3/time point, 27 total). 10 mg/kg oral dose (suspension in 10% EtOH, 30% PEG400, 60%
(0.5% methylcellulose, 0.5% Tween 80 (aq)) in male CD1 mice (n = 3/time point, 24 total).
All analogs were progressed into pharmacokinetic/pharmacodynamic (PK/PD) study in
the 786-O mouse xenograft model at an oral dose of 10 mg/kg on a twice daily schedule.
Endpoints in the 36h PK/PD study were obtained 12h post final dose and included analysis of
HIF-2a-ARNT transcriptional products VEGFA and cyclin D1 (CCND1) in excised tumors by
qPCR and evaluation of circulating levels of human tumor-derived VEGFA in serum by ELISA.
Plasma drug concentrations were determined by LC/MS/MS. Four of the analogs potently
ACS Paragon Plus Environment
2
2

Page 23 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibited HIF-2a-driven gene transcription in tumor as well as completely abolishing circulating
human tumor-derived VEGFA (Figure 6). In contrast, 10n had minimal to no effect on these
endpoints in the experiment. Examination of the drug concentrations showed that the four active
analogs all achieved trough (12 h) levels well-above the free-fraction adjusted VEGFA cell
potency, but that 10n did not.
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
1
1
0
0
.0
.5
.0
.5
.0
2.0
1.5
1.0
0.5
0.0
plasma levels (uM)
CCND1 mRNA
VEGFA mRNA
e
l
9
a
c
10f
10i
i
10h
10n
h
Ve
Figure 6. PK/PD evaluation of five leads in 786-O mouse xenograft (female SCID/beige mice,
n=4) dosed at 10 mg/kg p.o., b.i.d. for three days to achieve steady-state exposure). Tumors were
harvested 12 hours after administration of the final dose and total RNA was isolated from the
tumor samples and used to make single-stranded cDNA. The resulting cDNA was used in
quantitative PCR reactions; mRNA levels were normalized to internal control cyclophilin B
mRNA levels in each sample.
All five analogs were also examined in 786-O mouse xenograft tumor growth study. In
good agreement with the PK/PD, four analogs were efficacious with rapid regression of
ACS Paragon Plus Environment
2
3

Journal of Medicinal Chemistry
Page 24 of 103
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
established tumors, whereas 10n was essentially inactive (Figure 7). Importantly, no morbidity
or mortality was observed with treatment and all the leads showed a negligible impact on body
weight (see Supporting Information).
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
1
000
00
600
00
Vehicle
10n
8
10h
10i
1
9
0f
a
4
200
0
0
10
20
Time after treatment initiation (d)
Figure 7. In vivo efficacy study of five leads in 786-O xenografts (female SCID/beige mice,
n=8). Decrease in tumor volume in SCID mice with established sub-cutaneous 786-O tumors
that were treated orally with twice-daily administration of Hif-2a antagonists (10 mg/kg, p.o,
b.i.d.) for 21 days.
The four active analogs were evaluated in rat pharmacokinetics where 9a and 10i
demonstrated exposure profiles suitable for advancement into higher species, while, in contrast,
the profiles of 10f and 10h were deemed inadequate for progression. Both 9a and 10i performed
well in dog pharmacokinetics (Table 8).
ACS Paragon Plus Environment
2
4

Page 25 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 8. Rat and dog PK parameters for select inhibitors
a
b
a
b
CL
AUC0-t
(hr* M)
VSS
t1/2
b
Species
Compound
%F
(
mL/min/kg)
µ
(L/kg)
(h)
9
a
27.2
5.49
3.5
3.7
36
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
1
0f
57.7
57.6
34.6
0.39
2.14
5.12
5.55
6.15
6.24
5.4
2.9
3.3
6
Rat
1
0h
32
40
10i
9a
9.8
12.6
6.2
3.1
8.6
6.7
72
87
Dog
1
0i
21.5
11.0
a
3
mg/kg intravenous dose (solution in 20% EtOH, 40% PEG400, 40% water) in male Sprague-
b
Dawley rats (n = 3) or male beagles (n = 3). 10 mg/kg oral dose (suspension in 10% EtOH, 30%
PEG400, 60% (0.5% methylcellulose, 0.5% Tween 80 (aq)) in male Sprague-Dawley rats (n = 3)
or male beagles (n = 3).
With two analogs (one indanol and one cyclic sulfone) demonstrating potent efficacy and
favorable pharmacokinetics across multiple species, their profiles were examined more closely
(Table 9). While 9a showed a slight advantage in potency, solubility, and intrinsic clearance over
1
0i, in many other regards the leads were quite comparable. CYP profiling did not show any
liabilities for either lead. Permeability and plasma protein binding were also similar. At 10 µM
concentration, neither lead showed a significant signal in the kinase or safety panels. Finally,
hERG testing, also at 10 µM, suggested limited potential for QT prolongation for both
compounds.
ACS Paragon Plus Environment
2
5

Journal of Medicinal Chemistry
Page 26 of 103
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
Table 9. Profile of Inhibitors 9a and 10i (mouse data omitted see Table 7)
Compound
9a
10i
27 ± 6 nM
41 ± 12 nM
59
Luciferase IC50
10 ± 5 nM
16 ± 2 nM
126
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
VEGFA EC50
thermodynamic solubility
(μmol/L) (pH 7.4)
CL h (mL/min/kg)
< 3
9.1
< 5
< 4
4.7
22.6
< 5
int
CL r (mL/min/kg)
int
CL d (mL/min/kg)
int
CL monkey (mL/min/kg)
5.4
int
1
A2 > 50, 2B6 > 50, 2C8 > 50, 2C9 1A2 > 50, 2B6 > 50, 2C8 > 50, 2C9
CYP(IC , μM)
5
0
1
5.7, 2D6 > 50, 3A4 13.3
23, 2C19 46, 2D6 > 50, 3A4 > 50
PPB h (%)
PPB r (%)
PPB d (%)
PPB monkey (%)
63
85
82
62
79
71
75
74
kinase selectivity (<70% activity
0/40
17.0
0/40
remaining @ 10 µM)
hERG (% inhibition @10 µM)
8.2
Lead Profiling Screen®* (>50%
1
/68
0/68
inhibition @10 µM)
Next, the leads metabolism was examined with a focus on formation of ketone
metabolites as the difluoro ketone species could be highly electrophilic and thus may represent a
reactive metabolite. While the ketone or hydrate of 10i was not observed in any matrix
examined, significant quantities of carboxylic acid 12 (AUC0-24h 12/AUC0-24h 9a = 9.7, 10 mg/kg
PO) were observed in plasma of rats treated with 9a which was shown to arise from facile
hydrolysis of the ketone 11. (Scheme 4). In fact, quantification of ketone 11 in rat plasma was
not done as analysis was complicated by the rapid hydrolysis of the ketone reference standard to
carboxylic acid 12 on standing in DMSO at room temperature. This finding coupled with the
lack of ketone formed from 10i led to the removal of 9a from consideration for advancement.
ACS Paragon Plus Environment
2
6

Page 27 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 4. Identification of ketone and carboxylic acid metabolites of 9a
F
F
F
F
NC
O
OH
Me
NC
O
O
X
Me
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
S
S
O O
O O
F
F
1
0i
Indanone not
detected
F
F
F
F
F
F
O
OH
O
S
NC
O
[O]
NC
O
NC
O
F
F
F
OH
SO CHF2
DMSO reference
undergoes
S
F
2
O
O
O
O
F
F
hydrolysis
F
9
a
11
12
Ketone detected
Carboxylic acid
detected in vivo
SPA IC50 >100 uM
in vivo
Based on potency, selectivity, pharmacokinetics, metabolite profile and efficacy 10i
(PT2385) was selected for further development and has progressed into clinical trials for the
31,32,33
treatment of ccRCC and VHL Disease.
Mechanism of Heterodimer Disruption
3
1
We have reported elsewhere on the mechanism of heterodimer disruption for 10i ,
however, it is instructive to examine crystallographically the evolution of 10i in a retrospective
fashion (Figure 8). Such an analysis yields insights into the residue perturbations that appear to
be essential for efficacious disruption. One of our earliest potent binders, 4a, subtly changed
HIF-2a upon complexation when compared to the apo form (Figure 8, A and B). Only Met252
showed a slight change in disposition to accommodate the nitro group of 4a. Accordingly, 4a
and related analogs showed limited functional effects. A key breakthrough in overcoming
limited functional activity was the discovery of 6c (see Figure 8, C). The benzyl alcohol moiety
engaged a unique hydrogen bonding network replacing the Tyr278 and His293 hydrogen bond
with a water mediated hydrogen bond to Tyr281 (not shown) and a direct hydrogen bond to
ACS Paragon Plus Environment
2
7

Journal of Medicinal Chemistry
Page 28 of 103
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
His293. The consequence of this new hydrogen bond network was that His293 must rotate out
of plane to accommodate interaction with the benzyl alcohol and Tyr278 must shift slightly to
adapt to the water mediated hydrogen bond with Tyr281. Met252 showed little change in
orientation when compared to the structure of 4a. Cyclic sulfone 9a maintained the alcohol
hydrogen bonding interaction with His293, but further displaced Met252 and Tyr278 (Figure 8,
D). Met252 nearly extended over the aromatic portion of the cyclic sulfone and the electron
density for Tyr278 suggested two possible conformations. Interestingly, cyclization to indanol
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
1
0i produced the most compelling changes in binding (Figure 8, E). The steric consequence of
incorporating a new ring with geminal fluorination was profound perturbation of Met252.
Electron density for Met252 indicates that it was splayed in linear fashion above the indane ring
of 10i. To accommodate the new orientation of Met252, Tyr278 shifted completely to the
outside binding mode first observed in 9a. The rotation of His293 deviated slightly from the
structure of 6c and 9a, but maintained the hydrogen bond interaction with the alcohol moiety of
1
0i.
A
B
apo
4a
C
D
E
6
c
9a
10i
ACS Paragon Plus Environment
2
8

Page 29 of 103
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 8. Structural evolution of 10i highlighting the perturbation of Met252, Tyr278, and
His293 by ligand binding. Dotted green lines denote hydrogen bond interactions. (A) X-ray
crystal structure of apo HIF-2a PAS-B*/ARNT PAS-B* dimer (PDB:6D0C). (B) X-ray crystal
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
structure of 4a (PDB:6D0B, pink) in complex with the HIF-2a PAS-B*/ARNT PAS-B* dimer.
(C) X-ray crystal structure of 6c (PDB:6CZW, pink) in complex with the HIF-2a PAS-
B*/ARNT PAS-B* dimer. (D) X-ray crystal structure of 9a (PDB:6D09, pink) in complex with
the HIF-2a PAS-B*/ARNT PAS-B* dimer. (E) X-ray crystal structure of 10i (PDB:5TBM,
pink) in complex with the HIF-2a PAS-B*/ARNT PAS-B* dimer.
The discovery of the role of the alcohol moiety was crucial to developing functional HIF-
2a antagonists. This moiety served to break the hydrogen bond between Tyr278 and His293
and, importantly, reorient Tyr278. It is believed that perturbation of Tyr278 is essential for
functional activity. The structural evolution to 10i showed increasing displacement of Tyr278.
In these later examples, the perturbation of Met252 by the fortuitous incorporation of steric
elements appeared to be important for enhancing the displacement of Tyr278. However, the
exact nature of the small molecule antagonism of HIF-2a that causes the full suite of functional
effects remains elusive. Some of the earlier less active analogs without the structural
perturbations of 10i can elicit functional effects in cell based assays of disruption.
In support of the importance of displacing Tyr278 are data from a resistant ccRCC
patient-derived xenograft (PDX) tumor line where a F466L mutation in ARNT drives resistance
3
1,34
to 10f.
of Tyr278 with Phe466 of ARNT induced by binding of 10f is responsible for functional activity
Figure 9).
The mutant shows resistance to treatment with 10f and suggests that the steric clash
(
ACS Paragon Plus Environment
2
9