Isoform-Selective and Stereoselective Inhibition of Hypoxia Inducible Factor-2

Article abstract of DOI:10.1021/acs.jmedchem.5b00529

Hypoxia inducible factor (HIF) transcription factors reside at the center of signaling pathways used by mammalian cells to sense and respond to low oxygen levels. While essential to maintain oxygen homeostasis, misregulation of HIF protein activity correl

Full text of DOI:10.1021/acs.jmedchem.5b00529

Article
pubs.acs.org/jmc
Isoform-Selective and Stereoselective Inhibition of Hypoxia Inducible
Factor‑2
Thomas H. Scheuermann,† Daniel Stroud,‡ Christopher E. Sleet,‡ Liela Bayeh,‡ Cameron Shokri,‡
Hanzhi Wang,‡ Charles G. Caldwell,‡ Jamie Longgood,‡ John B. MacMillan,*,‡ Richard K. Bruick,*,‡
Kevin H. Gardner,*,†,§,∥ and Uttam K. Tambar*,‡
†Department of Biophysics, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390,
United States
‡Department of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390,
United States
§Structural Biology Initiative, CUNY Advanced Science Research Center, New York, New York 10031, United States
∥Department of Chemistry and Biochemistry, City College of New York, New York, New York 10031, United States
S
* Supporting Information
ABSTRACT: Hypoxia inducible factor (HIF) transcription
factors reside at the center of signaling pathways used by
mammalian cells to sense and respond to low oxygen levels.
While essential to maintain oxygen homeostasis, misregulation
of HIF protein activity correlates with tumor development and
metastasis. To provide artificial routes to target misregulated
HIF activity, we identified small molecule antagonists of the
HIF-2 transcription factor that bind an internal cavity within
the C-terminal PAS domain of the HIF-2α subunit. Here we
describe a new class of chiral small molecule ligands that
provide the highest affinity binding, the most effective,
isoform-selective inhibition of HIF-2 in cells, and trigger the largest protein conformation changes reported to date. The
current results further illuminate the molecular mechanism of HIF-2 antagonism and suggest additional routes to develop higher
affinity and potency HIF-2 antagonists.
recruit coactivators, as revealed by combinations of biochemical
and biophysical analyses.5 Strikingly, crystal structures of the
HIF-1α and HIF-2α C-terminal PAS domains (PAS-B) contain
unusually large internal cavities, suggesting the potential for
ligand binding and regulation via these sites. While the
identities and functional roles of endogenous ligands that
might naturally regulate HIFs in this way are unclear, we and
others have identified artificial small molecules that bind within
PAS-B internal cavities.6−8
Our own efforts targeting the HIF-2 transcription factor
identified small molecules that bind within the HIF-2α PAS-B
internal cavity in vitro and antagonize HIF-2 activity in cell-
based assays (Figure 1).7,8 Achiral compounds 1−2 were
INTRODUCTION
■
Multicellular organisms rely upon conserved biochemical
pathways to monitor and maintain appropriate cellular oxygen
levels. This hypoxia response pathway converges upon a family
of hypoxia inducible factors (HIF-1, HIF-2, and HIF-3) which
accumulate under low oxygen conditions (hypoxia) in the cell
nucleus to upregulate the transcription of hundreds of genes
that help the cell adapt to this environmental stress.1,2 This
pathway plays important roles in normal physiological
processes such as embryonic development, metabolism, oxygen
transport, and angiogenesis.3 It unfortunately also supports the
growth and metastasis of solid tumors, the progression of which
might otherwise be constrained by oxygen deprivation born of
unregulated cell proliferation.4
HIFs are heterodimeric transcription factors consisting of a
common constitutive subunit called the aryl hydrocarbon
receptor nuclear translocator (ARNT, or HIF-β) and one of
three HIF-α subunits. Each subunit consists of a bHLH domain
for DNA binding, two Per-ARNT-Sim (PAS) domains, and C-
terminal elements that confer oxygen-dependent degradation
and mediate the recruitment of transcriptional coactivator
proteins. The PAS domains provide important intermolecular
contacts that stabilize the HIF heterodimer and can themselves
Figure 1. Achiral ligands of HIF-2α.
Received: April 6, 2015
Published: July 30, 2015
© 2015 American Chemical Society
5930
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
developed from a fragment-based drug discovery lead, and
achiral compound 3 was derived from an initial lead compound
identified from high-throughput screening.
By reviewing additional lead compounds from the high
throughput screen (HTS), we identified a series of tetrazolo-
tetrahydropyrimidine structures with several intriguing charac-
teristics (Figure 2, 4).9 Notably, these chiral compounds
conditions to yield the desired trans-chalcones 6 (Scheme 1).
Similarly, a series of 3-bromo chalcones were synthesized
through the aldol condensation of aldehydes 7 and
acetophenone 8a under basic protic conditions (method B).
For chalcones that could not be synthesized efficiently by the
aldol condensation strategy, we employed dimethyl methyl-
phosphonate and stabilized phosphorus ylide reagents (Scheme
2). The Horner−Wadsworth−Emmons olefination of alde-
hydes 7 and dimethyl methylphosphonate furnished 3-bromo-
5-fluoro chalcones 6w−x (method C). Modified Wittig
olefination between phosphonium ylide 12 and aldehydes 7
was employed for the synthesis of the remaining chalcones
(method D).
We assembled the core structure of the HIF-2α inhibitors by
coupling the synthesized chalcones 6 with 5-aminotetrazole 5
under thermal conditions (Scheme 3).11 The cyclization step
presumably proceeded through a stepwise mechanism, with
initial 1,4-addition of the tetrazolic nitrogen, followed by the 5-
exo-dig cyclization of the amino group and the carbonyl carbon
with subsequent loss of water. The resulting enamines ( )-13
were then subjected to NaBH4 mediated reduction. Interest-
ingly, the reduction products ( )-4 were formed with complete
selectivity for the syn diastereomer.
Biophysical and Biological Characterization. To obtain
the affinities of the racemic tetrazolo-tetrahydropyrimidines
binding to the HIF-2α PAS-B domain, we used isothermal
titration calorimetry (ITC) to measure thermodynamic
parameters of the interaction (Table 1). In addition to
dissociation constants (KD), we also obtained enthalpies of
binding (ΔH) and estimates of the fraction of ligands
incompetent to bind (incA), which is defined in the Methods
section. Clear trends of activity were identified from early SAR
studies around the initial hit ( )-14 from the high throughput
screen. The presence of a 3-bromo substituent in the phenyl B-
ring was necessary for activity, as removal of the bromide
abrogated binding (Figure 4, ( )-15) and incorporation of a
different halide slightly lowered the complex affinity (( )-17).
In addition, an aromatic ring was slightly preferred over an
aliphatic system for the C-ring (( )-14 vs ( )-16).
For most of the remaining inhibitors in this study, we
focused on 3-bromophenyl for the B-ring and differentially
substituted phenyl rings for the C-ring. Ortho-halide sub-
stitution in the C-ring in some cases provided a modest 2-fold
reduction of KD compared to ( )-14 (( )-18, ( )-19,
( )-20, and ( )-21), with the most active compound in this
series exhibiting a KD value of 327 nM (( )-19). Once again,
incorporation of a 3-iodo substituent in the B-ring instead of a
3-bromo group substantially lowered activity (( )-21 vs
( )-22), negating the positive contributions of C-ring ortho-
halide substituents.
Figure 2. New class of chiral ligands of HIF-2α.
contain two stereocenters, in contrast to our previously
described achiral scaffolds (Figure 1). Given the achirality of
the initial lead compounds 1−3, which is typical for most PAS
domain binding ligands,10 the topologies of chiral ligands
immediately raised questions about the nature and affinity of
their interactions with the HIF-2α PAS-B target. Coupled with
the removal of an obligatory nitro group and the accompanying
potential for metabolic liabilities from the prior series, we opted
to further characterize these chiral compounds for their ability
to bind HIF-2α PAS-B and antagonize HIF-2.
Here we describe the discovery of HIF-2α PAS-B chiral
ligands that possess the general structure depicted in Figure 2,
which has resulted in the highest affinity isoform-selective
ligand of HIF-2α PAS-B reported to date. Although hundreds
of compounds were synthesized and evaluated by our
laboratories, here we discuss a subset of inhibitors. By
combining ligand-binding information from isothermal titration
calorimetry (ITC) and functional readouts from an Al-
phaScreen luminescence proximity assay,8 we demonstrate
the stereoselective binding to the HIF-2α PAS-B domain and
disruption of HIF-2α/ARNT PAS-B heterodimerization by one
enantiomeric series of this chiral scaffold. Crystallographic data
is presented to account for this observed enantioselectivity in
inhibition and to suggest the superiority of this chiral scaffold
over other scaffolds as a starting point for discovering
chemotherapeutic agents that function by inhibiting the HIF-
2 transcription factor.
RESULTS
■
Chemistry. We envisaged that inhibitors of the general
structure 4 could be assembled from 5-aminotetrazole 5 and
chalcones 6 (Figure 3). This convergent strategy allowed for
meta-Halide substitution in the C-ring proved to be even
more effective (( )-23, ( )-24, and ( )-25), as installation of
a 3-bromo substituent in the C-ring (( )-24) resulted in the
first tetrazolo-tetrahydropyrimidine inhibitor with a KD value
less than 100 nM (KD = 70 nM).
Figure 3. Retrosynthetic analysis for chiral HIF-2α ligands.
To combine the advantageous effects of ortho- and meta-
substitution in the C-ring, we synthesized a series of inhibitors
with ortho−meta-disubstitution (( )-26, ( )-27, and ( )-28).
In this series, the 2-fluoro-3-chlorophenyl analogue ( )-28 was
the strongest binder (KD = 38 nM). Although ortho−ortho−
meta-trisubstitution of the C-ring was also effective (( )-29,
( )-30, ( )-31, ( )-32, ( )-33, ( )-34, and ( )-35), we
the efficient synthesis of a diverse series of compounds. These
inhibitors contain an A-ring tetrazole, a B-ring, and a C-ring
adorning the periphery of the molecular scaffold.
Chalcones 6 were generated by four distinct approaches
(Schemes 1 and 2). In method A, aldehydes 7 and
acetophenones 8 were condensed under basic aprotic
5931
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
Scheme 1. Synthesis of Chalcones by Aldol Condensation (Methods A and B)
did not observe any additional improvement over ortho−meta-
disubstitution of the C-ring.
compound preparations (present within the calorimeter cell)
saturated at much lower concentrations of injected protein than
would be expected from a single site binding model that
assumes that both compound enantiomers bind with similar
affinity. To best fit these ITC data, it was necessary to model a
large incompetent fraction (incA) for the compound (see
Methods section), suggesting substantially different binding
affinities between the two enantiomers.
To further characterize these complexes, 10 racemic
tetrazolo-tetrahydropyrimidines with KD values ranging from
28−1410 nM were subjected to preparatory chiral HPLC
separation of enantiomers (Figure 5). Racemic tetrazolo-
tetrahydropyrimidines were synthesized efficiently and eco-
nomically in our laboratories, and the enantiomers were
separated by Lotus Separations, LLC, for $960 per racemic
mixture. To facilitate efficient scale-up of the enantiopure
tetrazolo-tetrahydropyrimidines for future studies, we are
currently developing a catalytic enantioselective synthesis of
these chiral ligands. The separated enantiomers were then
utilized in an AlphaScreen luminescence proximity assay to test
for stereoselective disruption of an engineered high affinity
heterodimer of the HIF-2α PAS-B and ARNT PAS-B domains
Finally, we modified the B-ring by incorporating an
additional meta-substituent (( )-36 and ( )-37). The
combination of meta-substitution in the C-ring and meta−
meta-disubstitution in the B-ring resulted in the most potent
small molecule binder of the HIF-2α PAS-B domain reported
to date (Figure 4, ( )-37, KD = 28 nM). Notably, this B-ring
meta−meta-disubstitution was also essential for high affinity
binding with the achiral compound series represented by 3.8
All measurements were collected by injection of 40 μM HIF-
2α PAS-B into 4 μM compound racemic mixtures. The incA
value reports the fraction of binding incompetent compound
modeled in order to best fit the binding isotherm.
The binding of racemic tetrazolo-tetrahydropyrimidine
inhibitors to the HIF-2α PAS-B domain with this high level
of potency (KD < 100 nM) presumably requires a specific
orientation of the chiral small molecules in the buried cavity
within the core of this domain. Therefore, we reasoned that the
enantiomers of the chiral tetrazolo-tetrahydropyrimidines
would exhibit different binding properties. Consistent with
this expectation, all of our ITC results obtained from racemic
5932
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
confirm the trend toward much more negative ΔH values
observed in the single ITC measurements of tetrazolo-
tetrahydropyrimidine inhibitor complexes, triplicate ITC data
were collected from the 37 complexes under higher reactant
concentrations (Figure 4). Triplicate isotherms collected in a
different, phosphate buffer background provided very similar
results (Supporting Information Figure S2), demonstrating that
the more negative enthalpy changes observed for the present
compound series may be attributed to protein/ligand complex-
ation and is not an artifact derived from buffer ionization
effects.12
Scheme 2. Synthesis of Chalcones by Olefination with
Phosphorous Reagents (Methods C and D)
Additional calorimetry data establish that the current series of
compounds bind specifically within the HIF-2α PAS-B internal
cavity, which we validated by examining the binding affinities of
these compounds to HIF-α PAS-B domains with smaller
internal cavities. One line of support was provided by using the
engineered S304M mutant HIF-2α PAS-B domain,8 which
substitutes a small polar amino side chain in the central cavity
with a large hydrophobic side chain. This single amino acid
change at the presumed ligand binding site attenuated the
affinity of the (S,R)-37 complex to approximately KD = 2 μM,
roughly a 40-fold diminution (Supporting Information Figure
S2C,E). An additional line of support was provided by
examining compound binding to the HIF-1α PAS-B domain,
a paralogue which has high sequence and structural similarities
with HIF-2α but contains larger amino acid side chains at
several cavity-lining positions.6 The much smaller HIF-1α
cavity appears incompatible with (S,R)-37 binding, as we
observed no evidence of complex formation within the limits of
protein and ligand solubility (Supporting Information Figure
S2D,F).
We obtained confirmation of the binding mode of these
chiral compounds within HIF-2α PAS-B by obtaining a crystal
structure of the (S,R)-24−HIF-2 PAS-B* ternary complex
(Figure 6A,B; Supporting Information Table S1). This
structure demonstrates the general features of previous HIF-2
PAS-B* crystal structures while presenting a new set of
protein/ligand interactions. The HIF-2α PAS-B and ARNT
PAS-B domains adopt similar overall conformations as
observed in the ligand-free structure (0.44 and 0.24 Å Cα
rmsd, respectively to PDB 3F1P).7 These domains also utilize
the same heterodimer interface demonstrated by NMR-driven
modeling methods13 and crystal structures (0.66 Å Cα rmsd
against the HIF-2 PAS-B* heterodimer from 3F1P).7,8
Additionally, compound (S,R)-24 binds within the HIF-2α
PAS-B internal cavity (Figure 6B), utilizing the same site as
nitro-aniline 2 and benzoxadiazole 3 (Figure 6C).7,8 The
assignment of absolute stereochemistry to the active enantio-
meric series of compounds was confirmed by this crystal
structure of the (S,R)-24-HIF-2 PAS-B* ternary complex, and
the assignment of the absolute stereochemistry of the inactive
enantiomeric series of compounds was confirmed by analysis of
crystals of (S,R)-17 that were suitable for X-ray crystallography
(see Supporting Information).
(HIF-2 PAS-B* heterodimer)8 and ITC to detect any
stereoselectivity in binding to the HIF-2α PAS-B domain
(Table 2). Interestingly, in all cases, one enantiomer was a
strong binder of the HIF-2α PAS-B domain and correspond-
ingly potent disruptor of HIF-2 PAS-B* heterodimerization,
whereas the other enantiomer was inactive (Supporting
Information Figure S1). Notably, ITC measurements from
purified, AlphaScreen-active enantiomers demonstrated greatly
reduced incompetent fractions compared to their correspond-
ing racemic mixtures (Figure 4). Together, these data strongly
indicate that only one enantiomer present in the racemic
mixtures forms high affinity complexes and does so with the 1:1
stoichiometry observed for previously described small molecule
complexes with HIF-2α PAS-B.7,8
These ITC data also report ΔH values between −10 to −20
kcal/mol, significantly more negative than those reported for
previously characterized HIF-2α PAS-B small molecule
ligands.7,8 The first reported small molecules targeting this
protein domain were identified by an NMR-screen of a drug
fragment library that were optimized to low micromolar
affinities with ΔH values between −8 to −10 kcal/mol (e.g.,
1−2).7 The previously described scaffold (represented by 3),
which was identified by the same HTS screen that identified 14,
provided complexes with slightly more negative binding
enthalpies in the range of ΔH = −9 to −13 kcal/mol.8 To
A superposition of 2−, 3−, and 24−HIF-2 PAS-B* ternary
complexes reveals a common pharmacophore (Figure 6D).
Ligand B-rings demonstrate similar binding modes and bind
within the hydrophobic portion of the HIF-2α PAS-B cavity.
While the trifluoromethylbenzene of 2 and the benzoxadiazole
ring of 3 adopt similar binding poses and lay roughly in the
same plane, the C-ring of (S,R)-24 adopts a perpendicular
orientation and places its meta-bromine proximal to the HIF-2α
PAS-B β-sheet. The A-ring of (S,R)-24 represents the most
5933
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
Scheme 3. Synthesis of Chiral HIF-2α Inhibitors
interesting chemical contributions to the common pharmaco-
phore, with the tetrazole of (S,R)-24 occupying the nitro-
binding site of 2 and 3 and the pyrimidine group from (S,R)-24
superposing with the amine-containing linker from 2 and 3.
Notably, the two chemical groups from 2 and 3 that superpose
with the 14 A-ring were refractory to medicinal chemistry
approaches and were for those two scaffolds apparently critical
for HIF-2α PAS-B binding.
The current compound series adopts and improves upon the
molecular interactions observed in previously characterized
complexes. (S,R)-24 generally complements the internal cavity
present in the ligand-free protein structure and retains key
features of the B-ring of achiral 3, such as a large halogen
substitution proximal to the Eα secondary structure and a π-
bond shared with the Y307 hydroxyl (Figure 6E; Supporting
Information Figure S4A,B). Where the amine linker and A-ring
nitro-group of 2 and 3 provide weak electrostatic interactions
with the H248 imidazole side chain, the (S,R)-24 pyrimidine
secondary amine is H-bonded to internally bound water (W1)
that in turn forms an H-bond with the H248 imidazole. The
(S,R)-24 complex is further stabilized by an additional water-
mediated (W2) H-bond between H293 and an A-ring tetrazole
nitrogen. Lastly, the (S,R)-24 C-ring is rotated 90° relative to
the other scaffolds and places its meta-bromine proximal to the
Gβ protein secondary structure, a portion of the protein
internal cavity not well-accessed by other ligands, where it is
well placed to form a halogen bond with the V302 backbone
carbonyl (Figure 6E; Supporting Information Figure S4C,D).
Additionally, (S,R)-24-binding is associated with larger
protein conformation changes relative to previously charac-
terized ligands. Crystal structures of complexes with scaffolds
represented by 2 and 3 are highly similar to the unliganded
structure (see Supporting Information Figure S4A,B), with
ligand-induced changes to the binding site limited to the
rotamerization of the M252 side chain and small magnitude
perturbations of the β-sheet protein heterodimerization inter-
face.8 In contrast, the (S,R)-24 complex demonstrates wide-
spread binding site conformation differences (Supporting
Information Figure S4C). Most prominently: (A) the H248
side chain imidazole is displaced by 0.6 Å, positioning it to
participate in the W1-mediated hydrogen bond with the A-ring,
(B) the H293 Cβ position is displaced by 0.6 Å and its side
chain rotated 90° to facilitate a second water-mediated H-bond
(W2) to the ligand A-ring, and (C) the V302 side chain is
displaced by 1.6 Å at the Cβ position and rotates nearly 180° to
accommodate the C-ring meta-bromine (Figure 6E, see also
Supporting Information Figure S4D). These conformation
changes together reshape and resize the internal ligand binding
site, providing an expanded internal cavity bearing 40−65%
5934
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
Table 1. ITC-derived Thermodynamic Parameters for Racemic HIF-2α PAS-B Binding Compounds
compd
B-ring
3-BrPh
C-ring
ITC KD (nM)
ΔH (kcal/mol)
−16.2
incA
(
)-14
)-15
)-16
)-17
)-18
)-19
)-20
)-21
)-22
)-23
)-24
)-25
)-26
)-27
)-28
)-29
)-30
)-31
)-32
)-33
)-34
)-35
)-36
)-37
Ph
883
>2000
1170
1410
766
327
400
361
837
116
70
0.65
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
Ph
Ph
3-BrPh
3-IPh
i-Bu
Ph
−14.8
−15.5
−10.7
−16.2
−15.5
−15.3
−15.7
−18.6
−18.8
−20.4
−12.9
−18.4
−16.3
−15.5
−20.3
−16.8
−16.9
−18.2
−16.9
−16.6
−15.9
−20.2
0.55
0.66
0.5
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-IPh
2-IPh
2-FPh
0.52
0.49
0.56
0.65
0.51
0.56
0.55
0.67
0.63
0.49
0.54
0.9
2-BrPh
2-ClPh
2-ClPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-BrPh
3-Br-5-FPh
3-Br-5-FPh
3-ClPh
3-BrPh
3-IPh
103
159
126
38
2-F-5-IPh
2-Cl-5-BrPh
2-F-3-ClPh
2,3,6-F3Ph
2-Cl-3-Br-6-FPh
2,6-F2-3-ClPh
2,5-Cl2-6-FPh
2,6-F2-3-BrPh
2,5-Br2-6-FPh
2-Cl-5-Br-6-FPh
Ph
393
411
85
0.61
0.55
0.61
0.74
0.74
0.53
0.78
194
49
240
85
216
28
3-BrPh
complexes with ethylene glycol, 2 and 3 demonstrate few HIF-
2α PAS-B backbone conformation differences relative to the
ligand-free structure on the 2.5 Å scale; such changes are widely
observed in the (S,R)-24 ternary complex. Most dramatic in
this analysis are the ligand-induced (A) displacement of the Eα
helix away from the Fα helix and the β-sheet, (B) displacement
of the N-terminus of the Gβ strand away from the N-terminus
of Fα, Eα and the GH loop, and (C) contraction of the middle
of Gβ toward the C-terminus of Fα. The latter two DDM-
identified effects correlate with C-ring perturbation of V302,
repositioning its Cα 1.2 Å away from the ligand binding site.
The HIF-2α PAS-B sheet also has several additional Cα sites
with ≥0.5 Å differences from the ligand-free state, which are
either present at or are near the periphery of the protein
heterodimerization interface (Figure 7).
Given these improvements in chemical composition, ligand
affinity, and induced conformational changes, we also observed
a corresponding efficacy in the inhibition of endogenous HIF-2
in cultured cells. As shown in Figure 8A, administration of the
active enantiomer (S,R)-37 to hypoxic Hep3B cells antagonized
mRNA accumulation from the EPO gene, a HIF-2 selective
target.14 This effect was dose-dependent, with an IC50 less than
1 μM. The inactive enantiomer (R,S)-37 had no effect on EPO
mRNA levels. Consistent with the observed in vitro selectivity,
(S,R)-37 peak 2 had no effect on the expression of PGK1, a
HIF-1 selective target gene (Figure 8B). To date, this
compound series represents the most effective HIF-2 selective
antagonist in cells.7,8
Figure 4. ITC of racemic and enantiopure compound preparations.
Left and center are isotherms collected from ( )-15 and ( )-37. At
right are triplicate data collected from enantiopure (S,R)-37. KD and
ΔH values are reported as the best fit value followed by the associated
1σ error intervals in parentheses. Incompetent fractions for compound
(incA) and protein (incB) reactants reflect the difference between
reported reactant concentration and the concentration of binding
competent material that best fits the observed data.
larger volume than cavities observed in other HIF-2 PAS-B*
crystal structures (Supporting Information Figure S5).
Notably, ligand-induced conformational changes caused by
binding (S,R)-24 appear to propagate beyond the protein side
chains forming the ligand binding site and are readily observed
in the protein backbone (Supporting Information Figure S4D).
Alignment-independent information comparing ligand-induced
protein backbone conformation changes were obtained from
difference distance matrix (DDM) plots of Cα positions
(Supporting Information Figure S6). While HIF-2 PAS-B*
DISCUSSION
■
The present work describes the most effective small molecule
antagonists of HIF-2 reported to date. Early examples of
compounds targeting this protein were identified from an
NMR-based ligand binding assay of a drug fragment library and
provided low-micromolar dissociation constant ligands that
5935
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
Figure 5. Structures of selected enantiopure HIF-2 inhibitors.
Table 2. AlphaScreen IC50 and ITC KD Values for Enantiopure Compounds
compd
AlphaScreen IC50 (WT HIF-2α)
AlphaScreen IC50 (S304M control)
ITC KD (nM)
ITC incA/B
ITC ΔH kcal/mol
14 peak 1
14 peak 2
17 peak 1
17 peak 2
20 peak 1
20 peak 2
21 peak 1
21 peak 2
24 peak 1
24 peak 2
25 peak 1
25 peak 2
27 peak 1
27 peak 2
33 peak 1
33 peak 2
36 peak 1
36 peak 2
37 peak 1
37 peak 2
248 nM
>30000 nM
>30000 nM
1000 nM
28300 nM
314 nM
>30000 nM
>30000 nM
>30000 nM
>30000 nM
>30000 nM
18900 μM
>30000 nM
15900 nM
>30000 nM
>30000 nM
>30000 nM
>30000 nM
23900 nM
>30000 nM
3800 μM
487 nM
≫2000 nM
≫2000 nM
1239 nM
≫2000 nM
360 nM
B: 0.24
N/A
−18.1
N/A
N/A
N/A
A: 0.02
N/A
−13.2
N/A
A: 0.23
N/A
−15.1
N/A
>30000 nM
195 nM
≫2000 nM
294 nM
A: 0.03
B: 0.08
N/A
−15.6
−20.3
N/A
112 nM
63 nM
>30000 nM
108 nM
≫2000 nM
109 nM
B: 0.03
N/A
−19.8
N/A
>30000 nM
87 nM
≫2000 nM
125 nM
B: 0.01
N/A
−17.2
N/A
>30000 nM
32 nM
≫2000 nM
34 nM
B: 0.09
N/A
−20.8
N/A
22300 nM
>30000 nM
97 nM
>30000 nM
>30000 nM
10100 nM
>30000 nM
14500 μM
≫2000 nM
≫2000 nM
165 nM
N/A
N/A
B: 0.18
N/A
−19.1
N/A
>30000 nM
43 nM
≫2000 nM
23 nM
B: 0.14
−21.6
5936
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
Figure 6. Structural features of the (S,R)-24−HIF-2α PAS-B binding. (A) The ternary complex of (S,R)-24 (spheres) with the HIF-2α PAS-B*
(gray) and ARNT PAS-B* (blue) domains (PDB: 4XT2) presents the general features of previous HIF-2α PAS-B* crystal structures, including the
β-sheet protein heterodimerization interface and the antiparallel orientation of the two PAS domains. (B) (S,R)-24 binds within the HIF-2α PAS-B
internal ligand binding site (4XT2). Locations of HIF-2α PAS-B secondary structure elements are labeled. (C) The chemical structures
representative of the three scaffolds of known HIF-2α PAS-B ligands, with rings A, B, and C indicated for each. (D) Protein backbone alignment of 2
(black, 3H7W), 3 (gray, 4GHI) and (S,R)-24 (ball and stick, 4XT2) complexes reveal a common pharmacophore. (E) Electronic interactions
stabilize the (S,R)-24−HIF-2α PAS-B complex (4XT2). Water 3 superposes with one of the eight water molecules present in the apo protein
structure and does not obviously contribute to (S,R)-24 binding.
based functional assay monitoring for compound-based
disruption of the mutation-stabilized, high affinity HIF-2
PAS-B* heterodimer.8 This latter series of compounds offered
proof-of-concept that this important drug target could be
artificially regulated by drug-like small molecule ligands. A
comparison of apo and ligand bound cocrystal structures
illustrates the incomplete shape complementation and weak
electrostatic interactions at play for the previously described
complexes.
Figure 7. (S,R)-24-induced conformational changes extend to the
The same HTS screen identified an additional HIF-2α PAS-
HIF-2 PAS-B* heterodimer interface. (A) Red spheres plotted on the
B* ligand, 14, which presented intriguing opportunities and
was submitted to medicinal chemistry optimization. Where the
first two generations of HIF-2α PAS-B antagonists consisted of
planar moieties connected by one or two atom linkers, 14
possessed two stereocenters and lacked any linker connecting
its three rings. These features suggested an opportunity to
explore new binding modes that might access previously
underutilized portions of the internal cavity and, given the
limited conformational flexibility of this series, perhaps reshape
the cavity and distort the HIF-2 PAS-B* protein heterodime-
rization interface. Starting from the roughly KD = 1 μM HTS
hit, this new scaffold was optimized to provide the highest
known affinities for the target HIF-2α PAS-B domain.7,8
ligand-free protein (3F1P) denote HIF-2α PAS-B* Cα positions
displaced by 0.5 Å or greater in the presence and absence of (S,R)-24.
Particularly large conformation differences are denoted by blue
asterisks for three sites in the Gβ strand: 1.1, 1.2, and 0.8 Å, from
left to right. (B) These β-sheet backbone conformational changes are
plotted onto the ligand-free HIF-2α PAS-B* heterodimer, highlighting
their presence near the HIF-2α PAS-B* interface (light blue surface,
HIF-2α PAS-B sites within 5 Å of ARNT PAS-B (blue).
weakly antagonized heterodimerization of the isolated, wild-
type HIF-2α and ARNT PAS-B domains.7 Compounds that
disrupted endogenous HIF-2 heterodimerization and antago-
nized its transcription factor activity were identified from an
HTS screen of a much larger library using an AlphaScreen-
5937
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
domain, analogous to other PAS domains which sense
environmental changes through changes in the occupancy or
configuration of cofactors.17 Cocrystal structures with the 2 and
3 compound families showed moderate compound-induced
changes along the HIF-2α PAS-B β-sheet, perhaps due to
strong lattice contacts that inhibit such changes within the
crystal. Interestingly, in contrast to several of our previous HIF-
2α/ligand structures,7,8 which we were able to solve by
cocrystallizing protein and compound, this strategy repeatedly
failed for the current compound family. As such, we had to
instead resort to soaking (S,R)-24 into preformed, ligand-free
HIF-2 PAS-B* crystals, suggesting a potential correlation
between this observation and the (S,R)-24 complex having
the most profound compound-induced HIF-2α PAS-B
conformational changes observed to date. This correlation
further suggests that we may observe even larger ligand-induced
conformation changes in protein/ligand complexes crystallized
with different space groups or by using solution approaches that
are free of the limitations imposed by crystalline lattices.
CONCLUSION
■
In conclusion, we have developed chiral ligands of the HIF-2α
PAS-B domain that bind with higher potency and induce larger
protein conformation changes than previously reported for
achiral HIF-2α ligands, underscoring the utility of broad HTS-
based approaches to identify diverse drug scaffolds in targeting
protein binding sites. The favorable characteristics of these new
ligands make them the most effective HIF-2 selective
antagonists in cells reported to date,7,8 laying the foundation
for studies of their in vivo antagonism of HIF-2 in animal
models for cancer is currently under investigation.
Figure 8. Only one enantiomer is able to selectively antagonize
endogenous HIF-2 in cultured cells. (A) RT-PCR of treated Hep3B
cells reveals that only the (S,R)-37 enantiomer can antagonize
expression of the HIF-2 selective target genes EPO. Neither
enantiomer antagonizes expression of the HIF-1 selective target
PGK1 (B) or the gene encoding the HIF-2α subunit (C). The RT-
PCR data for each gene is the mean of three values determined from
three independently harvested sets and the error bars represent SD.
METHODS
■
Chemistry. Experimental procedures and characterization data of
new compounds are provided in the Supporting Information.
Protein Preparation. Proteins were expressed in Escherichia coli
and purified by affinity and size exclusion chromatographies as
previously described.1 Briefly, all proteins were expressed as affinity
fusions, with the affinity domain typically removed by a second round
of affinity chromatography following TEV protease treatment. Wild-
type and S304M HIF-2α PAS-B domains (240−350) for ITC
experiments were expressed as His6-Gβ1 fusions, as were the HIF-
2α PAS-B R247E and ARNT PAS-B E362R (“B*”) mutants that
provide a higher affinity protein heterodimer that readily crystallizes.
The HIF-1α PAS-B (238−349) was expressed as a GST-fusion. For
the AlphaScreen assay, the affinity tag of GST-HIF-2α PAS-B* was
retained but was removed from the expressed His6-Gβ1-ARNT PAS-
B*FLAG fusion protein.
A crystal structure of one analogue in a ternary complex with
the HIF-2 PAS-B* protein heterodimer finds both previously
observed and new interaction modes. As detailed above, the
(S,R)-24 B-ring adopts a similar conformation to that of bound
3, placing a large halogen proximal to the HIF-2α PAS-B Eα
secondary structure. Despite having distinctively different
chemical structures, comparison of various ligand-bound states
revealed a common pharmacophore for the (S,R)-24 A-ring,
which replaces the nitro group and amine linkers that were
indispensable for high affinity binding by the scaffolds
represented by 2 and 3. Replacement of this nitro group
represents an important step toward a clinically relevant small
molecule candidate, given the metabolic liabilities typically
associated with this functional group.15 Additionally, the A-ring
closely contacted with several internally bound water
molecules, two of which appear to mediate H-bonds to
cavity-lining protein side chains. An important role for these
electrostatic contacts is supported by ITC measurements of the
(S,R)-37 complex, which demonstrates substantially larger
enthalpy changes relative to 2 or 3.7,8 Given previous difficulty
encountered in engineering strong electrostatic interactions
into 2 and 3, these results support the strategy described by
Freire and colleagues wherein drug discovery efforts might
employ calorimetry to identify HTS-hits with favorable
enthalpies, laying the foundation for optimization of entropic
considerations by improved hydrophobic contacts.16
AlphaScreen. Experimental procedures are based on the assay
reported previously by our groups.8
RT-PCR. Experimental procedures are based on the assay reported
previously by our groups.8
ITC. HIF-2α PAS-B domain was dialyzed overnight against the
experimental buffer (typically 25 mM Tris pH 7.5, 20 mM NaCl),
which was also used to prepare compound solutions by dilution from
concentrated DMSO stocks. An equivalent final concentration of
DMSO was added to the protein solution prior to data collection and
did not exceed 0.1%. Initial ITC tests of racemic compounds were
collected as 34, 6 μL injections of 40 μM protein into 1.4 mL of 4 μM
compound (2 μM of each enantiomer) with a VP-ITC instrument.
Complexes that provided weakly exothermic thermograms were
recollected with 21, 12 μL injections under similar conditions.
Enantiopure compound preparations that demonstrated activity in the
AlphaScreen assay were tested with 34, 6 μL injections of 40 μM
protein into 2 μM compound, while inactive enantiomers were tested
with 22, 12 μL injections. To better characterize thermodynamics
Lastly, these results support an allosteric mechanism by
which HIF-2α PAS-B internal compound binding disrupts
important intermolecular contacts with the ARNT PAS-B
5938
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
parameters of the 37 complex, triplicate isotherms in Tris and
phosphate buffers (10 mM sodium phosphate, 20 mM NaCl) were
collected as 34, 8 μL injections of 50 μM protein into 5 μM single
enantiomer 37 solution. NITPIC 1.1.018 was used to determine the
injection peak baselines, integrate areas of injection heats, prepare an
isotherm, and estimate initial parameter values for subsequent analysis
with SEDPHAT 10.58d.19 In contrast with software that models an
“N” value (often discussed as a stoichiometry parameter), SEDPHAT
instead models an “incompetent fraction” for cell (here, incA) or
syringe (incB) material that accounts for discrepancies between
intended concentrations and the actual concentration of available,
binding competent material. For ideally behaved complexes with a 1:1
stoichiometry, SEDPHAT incompetent fractions will refine to a value
of 0.0, where other software would provide an N = 1.0 result. Were the
protein and racemic compounds ideally behaved and without any
concentration error, ITC of the racemic compounds would provide an
incompetent A fraction of 0.5 (i.e., 50% of the material in the cell is
not active for protein binding). ITC data plots were subsequently
prepared with GUSSI 1.0.8d.20
ASSOCIATED CONTENT
* Supporting Information
■
S
General procedures, compound characterization, assay proce-
dures, data tables, and biophysical characterization (PDF);
crystallographic information file (CIF); molecular formula
strings (CSV). The Supporting Information is available free of
charge on the ACS Publications website at DOI: 10.1021/
acs.jmedchem.5b00529.
Accession Codes
Atomic coordinates and structure factors of the 24-HIF-2 PAS-
B* ternary complex have been deposited to the PDB databank
and assigned the code 4XT2.
AUTHOR INFORMATION
Corresponding Authors
*For J.B.M.: phone, 214- 648-8653; fax, 214-648-8856; E-mail,
John.Macmillan@utsouthwestern.edu.
*For R.K.B.: phone, 214-648-6477; fax, 214-648-0320; E-mail,
Richard.Bruick@utsouthwestern.edu.
■
Crystallography. Ligand-free crystals were grown at 20 °C from
hanging drops composed of 2 μL of precipitant solution (100 mM
BisTris pH, 6.5, 25% PEG3350) and 2 μL of protein solution (500 μM
HIF-2 PAS-B* heterodimer, 50 mM Tris pH 8.0, 20 mM NaCl, 5 mM
DTT). Ligand was introduced by addition of 2 μL of compound
solution (1 mM 24 in 30% PEG3350) to drops containing promising
crystalline material and soaked for 3 days at 20 °C. Prior to flash
freezing in liquid nitrogen, crystals were soaked in precipitant solution
supplemented with 25% PEG400. Data were collected from the
Advanced Photon Source (APS, beamline ID-19), which were indexed,
scaled, and reduced with HKL2000_704k.21 A molecular replacement
solution was determined with Phaser,22 using the PAS-B* domains
from the ligand-free structure (3F1P) as a search model and refined
with Phenix 1.8.2.23 The PRODRG2 server24 was used to create initial
ligand coordinates, which were further processed with REEL and
eLBOW25 to create parameter files for use with Phenix and Coot.26
While sharing a common space group, the 24-soaked crystal
demonstrated a larger a unit cell and asymmetric unit (ASU) than
those observed in previous HIF-2 PAS-B* crystals. Where the ASU of
previously described apo- and ligand cocrystal structures consisted of a
single pair of HIF-2α and ARNT PAS-B* domains, the ASU of the
present structure contains two copies of HIF-2α and ARNT PAS-B*
domains. One pair of HIF-2α and ARNT PAS-B* domains in the ASU
demonstrates the previously described HIF-2 PAS-B* heterodimer
interface, while the remaining two PAS-B* domains form similar
contacts with PAS-B domains located in a neighboring ASU. Water
molecule labels W1, W2, and W3 in Figure 6E correspond to water
molecule numbers 515, 519, and 552 associated with chain C and
water molecules 524 and 505 from chain A, which lacks interpretable
density for a third water W3 in its cavity. The model demonstrates
good stereochemical quality (1.24, 98th percentile molprobity score),
with the single Ramachandran violation present in an extended loop
(HIF-2α PAS-B N331) that is represented by relatively poor electron
density.
*For K.H.G.: phone, 212-413-3220; E-mail, Kevin.Gardner@
asrc.cuny.edu.
*For U.K.T.: phone, 214-648-0580; fax, 214-648-8856; E-mail,
Uttam.Tambar@utsouthwestern.edu.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Grants from the National Institutes of Health (NIH P01
CA95471) and the Cancer Prevention and Research Institute of
Texas (RP100846, RP130513) to K.H.G. and R.K.B. are
gratefully acknowledged. This investigation was conducted in a
facility constructed with support from the Research Facilities
Improvement Program (grant no. C06 RR 15437-01) from the
National Center for Research Resources, US National Institutes
of Health. R.K.B. is the Michael L. Rosenberg Scholar in
Medical Research and was supported by a Career Award in the
Biomedical Sciences from the Burroughs Wellcome Fund
(1003437) and the Welch Foundation (I-1568). Financial
support for U.K.T. was provided by the W. W. Caruth, Jr.
Endowed Scholarship, the Welch Foundation (I-1748), and the
Sloan Research Fellowship. This work benefitted from the
excellent persons and facilities available from the synthetic
chemistry, high throughput screening, structural biology, and
biophysics core facilities at UT Southwestern. Yingli Duan is
acknowledged for experimental assistance. Results shown in
this report are derived from work performed at Argonne
National Laboratory, Structural Biology Center at the
Advanced Photon Source. Argonne is operated by UChicago
Argonne, LLC, for the U.S. Department of Energy, Office of
Biological and Environmental Research under contract DE-
AC02-06CH11357.
Difference density matrix (DDM) comparison of HIF-2α PAS-B*
structures were calculated with DDMP27 and required protein
structures lacking individual amino acids modeled with multiple
conformers. For this analysis, lower-occupancy conformers were
deleted from both the ligand-free (3F1P) structure and the 24-ternary
complex (and the most occupied conformer retained). Both single
conformer models were rerefined with Phenix prior to DDM analysis.
Similarly, in Figure 7, the backbone conformation of HIF-2α PAS-B in
complex with 24 was compared against the refined, single conformer-
only 3F1P coordinates. Internal binding site surfaces were rendered
and cavity volumes calculated with SURFNET,28 using a 1.4 Å radius
probe.
ABBREVIATIONS USED
■
ARNT, aryl hydrocarbon receptor nuclear translocator; DDM,
difference distance matrix; EPO, erythropoietin; HIF, hypoxia-
inducible factor; HTS, high throughput screening; inc,
incompetent fraction; ITC, isothermal titration calorimetery;
mRNA, messenger ribonucleic acid; NMR, nuclear magnetic
resonance; PAS, Per-ARNT-Sim; PDB, Protein Data Bank;
PGK1, phosphoglycerate kinase 1; RT-PCR, reverse tran-
scriptase−polymerase chain reaction; SAR, structure−activity
relationship
5939
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
(11) (a) Desenko, S. M.; Gladkov, E. S.; Komykhov, S. A.; Shishkin,
O. V.; Orlov, V. D. Partially hydrogenated aryl substituted tetrazolo-
[1,5-a]pyrimidines. Chem. Heterocycl. Compd. 2001, 37, 747−754.
(b) Zeng, L.-Y.; Cai, C. Iodine Catalyzed One-Pot Multicomponent
Synthesis of a Library of Compounds Containing Tetrazolo[1,5-
a]pyrimidine Core. J. Comb. Chem. 2010, 12, 35−40.
(12) (a) Baker, B. M.; Murphy, K. P. Evaluation of linked
protonation effects in protein binding reactions using isothermal
titration calorimetry. Biophys. J. 1996, 71, 2049−2055. (b) Armstrong,
K. M.; Baker, B. M. A Comprehensive Calorimetric Investigation of an
Entropically Driven T Cell Receptor-Peptide/Major Histocompati-
bility Complex Interaction. Biophys. J. 2007, 93, 597−609.
(c) Bernhard, S. A. Ionization constants and heats of tris-
(hydroxymethyl)aminomethane and phosphate buffers. J. Biol. Chem.
1956, 218, 961−969.
(13) Card, P. B.; Erbel, P. J. A.; Gardner, K. H. Structural Basis of
ARNT PAS-B Dimerization: Use of a Common Beta-sheet Interface
for Hetero- and Homodimerization. J. Mol. Biol. 2005, 353, 664−677.
(14) (a) Dioum, E. M.; Chen, R.; Alexander, M. S.; Zhang, Q.; Hogg,
R. T.; Gerard, R. D.; Garcia, J. A. Regulation of Hypoxia-Inducible
Factor 2α Signaling by the Stress-Responsive Deacetylase Sirtuin 1.
Science 2009, 324, 1289−1293. (b) Scortegagna, M.; Ding, K.; Zhang,
Q.; Oktay, Y.; Bennett, M. J.; Bennett, M.; Shelton, J. M.; Richardson,
J. A.; Moe, O.; Garcia, J. A. HIF-2α regulates murine hematopoietic
development in an erythropoietin-dependent manner. Blood 2004,
105, 3133−3140.
(15) (a) Meanwell, N. A. Synopsis of Some Recent Tactical
Application of Bioisosteres in Drug Design. J. Med. Chem. 2011, 54,
2529−2591. (b) Patani, G. A.; LaVoie, E. J. Bioisosterism: A Rational
Approach in Drug Design. Chem. Rev. 1996, 96, 3147−3176.
(16) Freire, E. Do enthalpy and entropy distinguish first in class from
best in class? Drug Discovery Today 2008, 13, 869−874.
REFERENCES
■
(1) Semenza, G. L. Hypoxia-inducible factors: mediators of cancer
progression and targets for cancer therapy. Trends Pharmacol. Sci.
2012, 33, 207−214.
(2) (a) Xia, X.; Lemieux, M. E.; Li, W.; Carroll, J. S.; Brown, M.; Liu,
X. S.; Kung, A. L. Integrative analysis of HIF binding and
transactivation reveals its role in maintaining histone methylation
homeostasis. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 4260−4265.
(b) Schodel, J.; Oikonomopoulos, S.; Ragoussis, J.; Pugh, C. W.;
Ratcliffe, P. J.; Mole, D. R. High-resolution genome-wide mapping of
HIF-binding sites by ChIP-seq. Blood 2011, 117, e207−e217.
(3) (a) Majmundar, A. J.; Wong, W. J.; Simon, M. C. Hypoxia-
Inducible Factors and the Response to Hypoxic Stress. Mol. Cell 2010,
40, 294−309. (b) Greer, S. N.; Metcalf, J. L.; Wang, Y.; Ohh, M. The
updated biology of hypoxia-inducible factor. EMBO J. 2012, 31, 2448−
2460.
(4) (a) Kondo, K.; Kim, W. Y.; Lechpammer, M.; Kaelin, W. G., Jr.
Inhibition of HIF2α Is Sufficient to Suppress pVHL-Defective Tumor
Growth. PLoS Biol. 2003, 1, e83. (b) Zagzag, D.; Zhong, H.; Scalzitti, J.
M.; Laughner, E.; Simons, J. W.; Semenza, G. L. Expression of
hypoxia-inducible factor 1α in brain tumors: association with
angiogenesis, invasion, and progression. Cancer (N. Y.) 2000, 88,
2606−2618.
(5) (a) Erbel, P. J. A.; Card, P. B.; Karakuzu, O.; Bruick, R. K.;
Gardner, K. H. Structural basis for PAS domain heterodimerization in
the basic helix-loop-helix-PAS transcription factor hypoxia-inducible
factor. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 15504−15509.
(b) Yang, J.; Zhang, L.; Erbel, P. J. A.; Gardner, K. H.; Ding, K.;
Garcia, J. A.; Bruick, R. K. Functions of the Per/ARNT/Sim Domains
of the Hypoxia-inducible Factor. J. Biol. Chem. 2005, 280, 36047−
36054. (c) Guo, Y.; Scheuermann, T. H.; Partch, C. L.; Tomchick, D.
R.; Gardner, K. H. Coiled-coil Coactivators Play a Structural Role
Mediating Interactions in Hypoxia-inducible Factor Heterodimeriza-
tion. J. Biol. Chem. 2015, 290, 7707−7721.
(6) Cardoso, R.; Love, R.; Nilsson, C. L.; Bergqvist, S.; Nowlin, D.;
Yan, J.; Liu, K. K. C.; Zhu, J.; Chen, P.; Deng, Y.-L.; Dyson, H. J.;
Greig, M. J.; Brooun, A. Identification of Cys255 in HIF-1α as a novel
site for development of covalent inhibitors of HIF-1α/ARNT PasB
domain protein−protein interaction. Protein Sci. 2012, 21, 1885−1896.
(7) (a) Scheuermann, T. H.; Tomchick, D. R.; Machius, M.; Guo, Y.;
Bruick, R. K.; Gardner, K. H. Artificial ligand binding within the
HIF2α PAS-B domain of the HIF2 transcription factor. Proc. Natl.
Acad. Sci. U. S. A. 2009, 106, 450−455. (b) Key, J.; Scheuermann, T.
H.; Anderson, P. C.; Daggett, V.; Gardner, K. H. Principles of Ligand
Binding within a Completely Buried Cavity in HIF2α PAS-B. J. Am.
Chem. Soc. 2009, 131, 17647−17654.
(8) (a) Scheuermann, T. H.; Li, Q.; Ma, H.-W.; Key, J.; Zhang, L.;
Chen, R.; Garcia, J. A.; Naidoo, J.; Longgood, J.; Frantz, D. E.;
Tambar, U. K.; Gardner, K. H.; Bruick, R. K. Allosteric inhibition of
hypoxia inducible factor-2 with small molecules. Nat. Chem. Biol. 2013,
9, 271−276. (b) Rogers, J. L.; Bayeh, L.; Scheuermann, T. H.;
Longgood, J.; Key, J.; Naidoo, J.; Melito, L.; Shokri, C.; Frantz, D. E.;
Bruick, R. K.; Gardner, K. H.; MacMillan, J. B.; Tambar, U. K.
Development of Inhibitors of the PAS-B Domain of the HIF-2α
Transcription Factor. J. Med. Chem. 2013, 56, 1739−1747.
(9) For recent discussions of the activity of similar tetrazolo-
tetrahydropyrimidines in other biological contexts, see: (a) Cuconati,
A.; Block, T. M.; Xu, X. Tetrazolopyridine compounds as inhibitors of
secretion of hepatitis B virus antigens for the treatment of a chronic
hepatitis virus. WO2007131168A2, 2007. (b) Dougherty, A. M.; Guo,
H.; Westby, G.; Liu, Y.; Simsek, E.; Guo, J.-T.; Mehta, A.; Norton, P.;
Gu, B.; Block, T.; Cuconati, A. A Substituted Tetrahydro-Tetrazolo-
Pyrimidine Is a Specific and Novel Inhibitor of Hepatitis B Virus
Surface Antigen Secretion. Antimicrob. Agents Chemother. 2007, 51,
4427−4437.
(17) Rivera-Cancel, G.; Ko, W.-h.; Tomchick, D. R.; Correa, F.;
Gardner, K. H. Full-length structure of a monomeric histidine kinase
reveals basis for sensory regulation. Proc. Natl. Acad. Sci. U. S. A. 2014,
111, 17839−17844.
(18) (a) Keller, S.; Vargas, C.; Zhao, H.; Piszczek, G.; Brautigam, C.
A.; Schuck, P. High-Precision Isothermal Titration Calorimetry with
Automated Peak-Shape Analysis. Anal. Chem. 2012, 84, 5066−5073.
(b) Scheuermann, T. H.; Brautigam, C. A. High-precision, automated
integration of multiple isothermal titration calorimetric thermograms:
New features of NITPIC. Methods 2015, 76, 87−98.
(19) (a) Houtman, J. C. D.; Brown, P. H.; Bowden, B.; Yamaguchi,
H.; Appella, E.; Samelson, L. E.; Schuck, P. Studying multisite binary
and ternary protein interactions by global analysis of isothermal
titration calorimetry data in SEDPHAT: Application to adaptor
protein complexes in cell signaling. Protein Sci. 2007, 16, 30−42.
(b) Brautigam, C. A.; Deka, R. K.; Schuck, P.; Tomchick, D. R.;
Norgard, M. V. Structural and Thermodynamic Characterization of the
Interaction between Two Periplasmic Treponema pallidum Lip-
oproteins that are Components of a TPR-Protein-Associated TRAP
Transporter (TPAT). J. Mol. Biol. 2012, 420, 70−86.
(20) Brautigam, C. A. Calculations and Publication-Quality
Illustrations for Analytical Ultracentrifugation Data. Methods Enzymol.
2015, 10.1016/bs.mie.2015.05.001.
(21) Otwinowski, Z.; Minor, W. Processing of X-ray Diffraction Data
Collected in Oscillation Mode. In Methods in Enzymology, Vol. 276:
Macromolecular Crystallography, Part A; Carter, C.W., Jr., Sweet, R. M.,
Eds.; Academic Press: New York, 1997; pp 307−326,
(22) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P. D.; Winn, M.
D.; Storoni, L. C.; Read, R. J. Phaser crystallographic software. J. Appl.
Crystallogr. 2007, 40, 658−674.
́
(23) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis,
I. W.; Echols, N.; Headd, J. J.; Hung, L.-W.; Kapral, G. J.; Grosse-
Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R.
J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H.
PHENIX: a comprehensive Python-based system for macromolecular
structure solution. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66,
213−221.
(10) Henry, J. T.; Crosson, S. Ligand-Binding PAS Domains in a
Genomic, Cellular, and Structural Context. Annu. Rev. Microbiol. 2011,
65, 261−286.
5940
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941

Journal of Medicinal Chemistry
Article
(24) Schuttelkopf, A. W.; van Aalten, D. M. F. PRODRG: a tool for
high-throughput crystallography of protein-ligand complexes. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 1355−1363.
(25) Moriarty, N. W.; Grosse-Kunstleve, R. W.; Adams, P. D.
electronic Ligand Builder and Optimization Workbench (eLBOW): a
tool for ligand coordinate and restraint generation. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 2009, 65 (10), 1074−1080.
(26) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Features and
development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,
66, 486−501.
(27) Flemming, P. J. DDMP; Yale University: New Haven, CT, 2005;
http://www.csb.yale.edu/userguides/datamanip/ddmp/.
(28) Laskowski, R. A. SURFNET: A program for visualizing
molecular surfaces, cavities, and intermolecular interactions. J. Mol.
Graphics 1995, 13, 323−330.
5941
DOI: 10.1021/acs.jmedchem.5b00529
J. Med. Chem. 2015, 58, 5930−5941