Potent and selective triazole-based inhibitors of the hypoxia-inducible factor prolyl-hydroxylases with activity in the murine brain

Article abstract of DOI:10.1371/journal.pone.0132004

As part of the cellular adaptation to limiting oxygen availability in animals, the expression of a large set of genes is activated by the upregulation of the hypoxia-inducible transcription factors (HIFs). Therapeutic activation of the natural human hypoxic response can be achieved by the inhibition of the hypoxia sensors for the HIF system, i.e. the HIF prolylhydroxylases (PHDs). Here, we report studies on tricyclic triazole-containing compounds as potent and selective PHD inhibitors which compete with the 2-oxoglutarate co-substrate. One compound (IOX4) induces HIFα in cells and in wildtype mice with marked induction in the brain tissue, revealing that it is useful for studies aimed at validating the upregulation of HIF for treatment of cerebral diseases including stroke.

Full text of DOI:10.1371/journal.pone.0132004

RESEARCH ARTICLE
Potent and Selective Triazole-Based
Inhibitors of the Hypoxia-Inducible Factor
Prolyl-Hydroxylases with Activity in the
Murine Brain
Mun Chiang Chan1,2, Onur Atasoylu1¤a, Emma Hodson2, Anthony Tumber3, Ivanhoe K.
H. Leung1¤b, Rasheduzzaman Chowdhury1, Verónica Gómez-Pérez1¤c
,
Marina Demetriades1, Anna M. Rydzik1, James Holt-Martyn1, Ya-Min Tian2,
Tammie Bishop2, Timothy D. W. Claridge1, Akane Kawamura1, Christopher W. Pugh2,
Peter J. Ratcliffe2, Christopher J. Schofield1*
1 Chemistry Research Laboratory, Department of Chemistry, University of Oxford, Oxford, United Kingdom,
2 Centre for Cellular and Molecular Physiology, University of Oxford, Oxford, United Kingdom, 3 Target
Discovery Institute, University of Oxford, Oxford, United Kingdom
¤a Current Address: School of Pharmacy, University of East Anglia, Norwich, Norfolk, United Kingdom
¤b Current Address: School of Chemical Sciences, University of Auckland, Auckland, New Zealand
¤c Current Address: Instituto de Parasitología y Biomedicina "López-Neyra", IPBLN-CSIC, Parque
Tecnológico de Ciencias de la Salud, Armilla, Granada, Spain
OPEN ACCESS
Citation: Chan MC, Atasoylu O, Hodson E, Tumber
A, Leung IKH, Chowdhury R, et al. (2015) Potent and
Selective Triazole-Based Inhibitors of the Hypoxia-
Inducible Factor Prolyl-Hydroxylases with Activity in
the Murine Brain. PLoS ONE 10(7): e0132004.
doi:10.1371/journal.pone.0132004
* christopher.schofield@chem.ox.ac.uk
Abstract
As part of the cellular adaptation to limiting oxygen availability in animals, the expression of
a large set of genes is activated by the upregulation of the hypoxia-inducible transcription
factors (HIFs). Therapeutic activation of the natural human hypoxic response can be
achieved by the inhibition of the hypoxia sensors for the HIF system, i.e. the HIF prolyl-
hydroxylases (PHDs). Here, we report studies on tricyclic triazole-containing compounds as
potent and selective PHD inhibitors which compete with the 2-oxoglutarate co-substrate.
One compound (IOX4) induces HIFα in cells and in wildtype mice with marked induction in
the brain tissue, revealing that it is useful for studies aimed at validating the upregulation of
HIF for treatment of cerebral diseases including stroke.
Editor: Sonia Rocha, University of Dundee, UNITED
KINGDOM
Received: April 2, 2015
Accepted: June 9, 2015
Published: July 6, 2015
Copyright: © 2015 Chan et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information files.
Introduction
Funding: This work was funded by the Wellcome
Trust, http://www.wellcome.ac.uk/, grant 091857/Z.10/
Z to CJS, and the British Heart Foundation, https://
www.bhf.org.uk/, grant PG/12/33/29546 to CJS. The
funders had no role in study design, data collection
and analysis, decision to publish, or preparation of
the manuscript.
In metazoans, the α/β heterodimeric hypoxia-inducible factor (HIF) complex activates the
expression of hundreds of target genes in response to hypoxia, including those involved in cell
growth, apoptosis, energy metabolism and angiogenesis [1]. Prolyl hydroxylation of human
HIFα in its C- and N-terminal oxygen dependent degradation domains (CODD and NODD),
as catalyzed by three HIF prolyl hydroxylases (PHD1-3, Fig 1A), leads to subsequent HIFα
polyubiquitination by the von Hippel-Lindau (VHL) protein complex and proteasomal degra-
dation [2,3,4,5,6] HIFα is also regulated via asparaginyl hydroxylation as catalyzed by factor
Competing Interests: The authors have declared
that no competing interests exist.
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
1 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
inhibiting HIF (FIH), a modification which blocks the recruitment of the transcriptional co-
activators CBP/p300 to HIFα, so causing reduced HIF transcriptional activity [7,8]. The activi-
ties of both the PHDs and FIH are suppressed by limiting oxygen, resulting in HIFα stabiliza-
tion and activation of the HIF system.
Various compounds have been reported as PHD inhibitors in the academic and patent liter-
ature [10,11,12]. Most of these compounds likely bind to the active site iron and compete with
2-oxoglutarate (2OG). Many bind in the pocket occupied by the 2OG (CH2)2COOH side
chain, and as a consequence contain a carboxylic acid, which is often undesirable from a phar-
macological perspective. We are interested in identifying potent and selective PHD inhibitors
that do not contain a carboxylic acid, including ones that may be useful in treatments for ische-
mic stroke [13,14]. Recently, a set of tricyclic PHD inhibitor with linked pyridine-carboxylate,
Fig 1. Hydroxylation of HIFα and the chemical structures of IOX4 and other PHD inhibitors used in this study. (a) Prolyl-hydroxylation (as catalyzed
by the PHDs) of HIFα. (b) Structures of the dihydropyrazoles (1 and IOX4) in comparison to structures of 2-oxoglutarate (2OG), N-oxalylglycine (NOG) (a
catalytically inactive analogue of 2OG), dimethyloxalylglycine (DMOG) (a cell-permeable ester derivative of NOG) and IOX2 [9]. Chemical structures of
previously reported PHD inhibitors (compound 2, bicyclic isoquinolinyl inhibitor IOX3 and bicyclic naphthalenylsulfone hydroxythiazole BNS) used in this
study are also shown.
doi:10.1371/journal.pone.0132004.g001
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
2 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
dihydropyrazole and triazole rings, such as 1 and IOX4 (Fig 1B) has been reported with one
compound being in clinical development [15,16]. This series has the potential for more than
one mode of iron chelation and 2OG competition; we were particularly interested in the possi-
bility that the triazole (an established carboxylate mimic), rather than the pyridine-carboxylate
(an established 2OG mimic) might occupy the 2OG (CH2)2COOH binding pocket. To test this
proposal, we prepared and studied both the acid (1) and tertiary butyl ester (IOX4) of the ‘tri-
cyclic series’, both reported as PHD inhibitors, but of uncharacterized selectivity and mecha-
nism of action [15,16].
Materials and Methods
Organic synthesis procedures
Ethyl 2-(1H-1,2,3-triazol-1-yl)acetate and Ethyl 2-(2H-1,2,3-triazol-2-yl)acetate [15].
To a solution of 1,2,3-triazole (5.42 g, 78.5 mmol) in acetone (80 mL) at room temperature was
added ethylbromoacetate (13.07 g, 80 mmol) and Na2CO3 (11.4 g, 108 mmol). The reaction
mixture was heated at 35°C for five days, filtered and then concentrated in vacuo. Flash column
chromatography using a gradient of ethyl acetate-hexanes (1–10%) as elutant afforded ethyl 2-
(2H-1,2,3-triazol-2-yl) acetate (1.66 g, 14%) and ethyl 2-(1H-1,2,3-triazol-1-yl) acetate (3.56 g,
30%). Spectroscopic data are consistent with those reported [15]. Ethyl 2-(1H-1,2,3-triazol-
1-yl) acetate, 1H NMR (400 MHz, CDCl3) δ 7.69 (s, 2H), 5.24 (s, 2H), 4.24 (q, J = 7.15 Hz,
2H), 1.28 (t, J = 7.15 Hz, 3H). Ethyl 2-(2H-1,2,3-triazol-2-yl) acetate 1H NMR (400 MHz,
CDCl3) δ 7.75 (s, 1H), 7.71 (s, 1H), 5.19 (s, 2H), 4.26 (q, J = 7.15 Hz, 2H), 1.29 (t, J = 7.15 Hz,
3H).
Ethyl 3-(dimethylamino)-2-(1H-1,2,3-triazol-1-yl)acrylate and methyl 3-(dimethyla-
mino)-2-(1H-1,2,3-triazol-1-yl)acrylate [16]. A mixture of ethyl 2-(1H-1,2,3-triazol-1-yl)
acetate (336 mg, 2.16 mmol) and DMFꢀdimethylamine (1.021 g, 6.94 mmol) was placed in a
sealed tube and heated at 100°C for 24 hours. The reaction mixture was concentrated in vacuo
to yield a mixture (1:3, respectively) of the ethyl 3-(dimethylamino)-2-(1H-1,2,3-triazol-1-yl)
acrylate and methyl 3-(dimethylamino)-2-(1H-1,2,3-triazol-1-yl)acrylate (210 mg, 55%). The
mixture was used in the next step without further purification. Spectroscopic data are consis-
tent with those reported [16]. Ethyl 3-(dimethylamino)-2-(1H-1,2,3-triazol-1-yl)acrylate, 1H
NMR (400 MHz, CDCl3) δ 7.73 (s, 1H), 7.61 (s, 1 H), 7.60 (s, 1H), 4.11 (q, J = 7.0, 2H), 3.06
(br s, 3H), 2.23 (br s, 3H), 1.16 (t, J = 7.0). Methyl 3-(dimethylamino)-2-(1H-1,2,3-triazol-1-yl)
acrylate, 1H NMR (400 MHz, CDCl3) δ 7.73 (s, 1H), 7.61 (s, 1 H), 7.60 (s, 1H), 3.63 (s, 3H)
3.06 (br s, 3H), 2.23 (br s, 3H).
t-Butyl 6-(5-oxo-4-(1H-1,2,3-triazol-1-yl)-2,5-dihydro-1H-pyrazol-1-yl)nicotinate
(IOX4) [16]. To a solution of t-butyl-6-hydrazinylnicotinate (1.84 g, 8.8 mmol) in ethanol
(16 ml) at room temperature was added the mixture of ethyl and methyl 3-(dimethylamino)-2-
(1H-1,2,3-triazol-1-yl)acrylate (1.68 g, 8 mmol assumed) and CF3CO2H (30 μL). The reaction
mixture was stirred in a sealed tube at 80°C for 12 h, then allowed to cool to room temperature.
4M HCl in dioxane (300 μL) was added and the resultant mixture was cooled with an ice-water
bath to form a pale yellow paste. The precipitate was washed with ethanol (25 mL). Purification
with reverse phase flash column chromatography using a gradient of acetonitrile and water as
elutant afforded IOX4 as a pale yellow solid (761 mg, 29%). Melting point: 194°C. 1H NMR
(400 MHz, DMSO-d6) δ 8.93 (s, 1H), 8.53 (s, 1H), 8.45–8.44 (m, 3H), 7.91 (d, J = 1.0 Hz, 1H),
1.57 (s, 9H); 13C NMR (101 MHz, DMSO-d6) δ 163.2, 148.8, 140.3, 133.3, 124.9, 124.4, 111.5,
81.8, 27.8; Rf = 0.45 (DCM: MeOH: NEt3(95: 5: 1)); IR: 3133.11(m) (N-H), 1713.96 (s)
(C = O), 1673.59 (s) (C = O), 1611.76 (m) (N-H), 1220.37 (s) (C-O, ester); m/z (MS, ES+)
296.055 (100%, MNa+).
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
3 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
6-(5-oxo-4-(1H-1,2,3-triazol-1-yl)-2,5-dihydro-1H-pyrazol-1-yl)pyridine-3-carbocylic
acid hydrochloride [16]. To a solution of t-butyl 6-(5-oxo-4-(1H-1,2,3-triazol-1-yl)-
2,5-dihydro-1H-pyrazol-1-yl)nicotinate (50 mg, 0.15 mmol) in CH2Cl2 (0.5 mL) was added
CF3CO2H (0.5 mL). The reaction mixture was stirred at room temperature for 1 h and concen-
trated in vacuo. The residue was suspended in aqueous HCl (1M, 2 mL) and lyophilized. Melt-
ing point: ꢁ316.6°C (decomposed). 1H NMR (500 MHz, D2O) δ ppm 8.84 (s, 1H), 8.29 (d,
J = 2.21 Hz, 1H), 8.27 (d, J = 2.05 Hz, 1H), 8.10–8.11 (m, 1H), 7.84–7.84 (m, 1H), 7.79–7.80
(m, 1H). 13C NMR (126 MHz, D2O) δ 173.0, 168.3, 156.8, 151.9, 149.2, 139.6, 137.4, 133.4,
129.5, 126.5, 115.4; IR: 3194.40 (C-OH) (m), 1627 (s) (C = O), 1594.03 (m) (N-H), 1411.93 (s)
(aromatic); m/z (MS, ES+) 351.11 (100%, MNa+).
In vitro hydroxylation assays
Inhibition assays for PHD2 (AlphaScreen) [9], the KDMs (AlphaScreen) [17], BBOX [18] and
FTO [19] were carried out as previously described. N-Terminally biotinylated 19-residue
human HIF1α CODD peptide (DLDLEMLAPYIPMDDDFQL) used in the PHD2 inhibition
assays was from GL Biochem (Shanghai). IC50 values were calculated using nonlinear regres-
sion with normalized dose-response fit using GraphPad Prism 5.
FIH liquid chromatography—mass spectrometry assay. 20 ml FIH enzyme (100 nM) in
assay buffer (50 mM Tris.Cl pH 7.8, 50 mM NaCl) was pre-incubated for 15 minutes in the
presence of compound and the enzyme reaction was initiated by the addition of 20 ml substrate
(200 mM L-ascorbic Acid, 20 mM Fe2+, 10 mM consensus sequence ankyrin repeat domain
peptide [20,21] [and 20 mM 2-oxoglutarate). After 15 minutes, the enzyme reaction was
stopped by the addition of 4 ml 10% formic acid and the reaction mixture transferred to a Rap-
idFire RF360 high throughput sampling robot. Samples were aspirated under vacuum onto a
C4 Solid Phase Extraction (SPE) cartridge. After an aqueous wash step (0.1% formic acid in
water) to remove non-volatile buffer components from the C4 SPE, peptide was eluted in an
organic wash step (85% acetonitrile in water, 0.1% formic acid) onto an Agilent 6530 Q-TOF.
Ion chromatogram data was extracted for the non-hydroxylated peptide substrate and the
hydroxylated peptide product and peak area data for extracted ion chromatograms were inte-
grated using RapidFire Integrator software. Percentage conversion of substrate to product was
calculated in Microsoft Excel and IC50 values were calculated using Graphpad Prism 5.
Cell culture and compound treatment
Human cell lines (RCC4, HeLa, Hep3B, MCF-7 and U2OS) were cultured in DMEM (D6546-
500ML; Sigma Aldrich) each supplemented with 10% fetal bovine serum (F7524-500ML;
Sigma Aldrich), 2 mM L-glutamine (G7513- 100ML; Sigma Aldrich), 50 units/ml of penicillin,
and 50 μg/ml of streptomycin (P0781-100ML; Sigma Aldrich). The RCC4 cell line was a gift
from C.H.C.M. Buys (University of Groningen, [2]), the HeLa and Hep3B cell lines were from
the European Collection of Cell Cultures (ECACC) [2], the MCF-7 cell line was from the
American Type Culture Collection (ATCC) and the U2OS cell line was a gift from S.Galey
(ICRF Clare Hall Laboratories, United Kingdom). Cells were treated either with DMSO (con-
trol) or compounds (dissolved in DMSO) and added directly into the cell culture medium to
the desired final concentration for 4–5 h as previously described [9,22].
Immunoblotting for cell extracts
Cells were extracted using urea/SDS buffer (6.7 M urea, 10 mM Tris-HCl pH 6.8, 10% glycerol
and 1% SDS) and processed for immunoblotting as described [22]. The following primary anti-
bodies were used for immunoblotting: mouse monoclonal HIF1α antibody clone 54 (610958,
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
4 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
BD Transduction Laboratories, 1:1000), rabbit polyclonal HIF1α hydroxy-Pro402 antibody
(07–1585, Millipore, 1:1000), rabbit monoclonal HIF1α hydroxy-Pro564 antibody clone
D43B5 (3434S, Cell Signaling, 1:500), mouse monoclonal HIF1α hydroxy-Asn803 antibody (a
kind gift from Dr M. K. Lee, Republic of Korea [23], 1:4000), mouse monoclonal PHD2 anti-
body clone 76a [24] (1:10), mouse monoclonal FIH antibody clone 162c [25] (1:200) and β-
actin/HRP (clone AC15, Abcam). HRP-conjugated swine polyclonal anti-rabbit IgG (P0399,
Dako), and goat polyclonal anti-mouse IgG (P0447, Dako) were used as secondary antibodies.
Quantitative HIF1α meso scale discovery immunoassay
Cells were seeded into 96-well cell culture plates (150 μl culture volume) at approximately 60–
70% confluency at least 12 h prior to the compound treatment. Half of the culture medium
(75 μl/well) were removed and 75 μl/well of medium containing the desired concentrations of
compounds were added into cells for 5 h. The cell medium was then removed and the cells
were washed once with phosphate buffered saline (PBS), followed by the addition of lysis buffer
(20 mM Tris pH 7.4, 450 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.5% Triton X-100, 1x com-
plete protease inhibitor cocktail and 100 μM DFO). Plates were then left at 4°C with shaking.
Cell extracts (25 μl/well) were added into each well of the MSD Bare Standard Bind 96-well
Multi Array plates (L15XA, Meso Scale Discovery) which have been pre-coated for 16 h at 4°C
with HIF1α mouse monoclonal antibody (NB100-105, Novus Biologicals, 50 ng/well) and pre-
blocked for 1 h at room temperature with 5% skimmed milk. Plates were incubated for 1 h with
cell extracts at room temperature with shaking, before being washed thrice with 200 μl/well
Tris Wash Buffer (50 mM Tris pH 7.6, 0.5% Tween-20 and 150 mM NaCl) and incubated with
25 μl/well of secondary HIF1α rabbit monoclonal antibody clone EP1215Y (ab51608, Abcam,
1:250) for 1 h at room temperature with shaking. Plates were then washed thrice again with
Tris Wash Buffer, followed by incubation with 25 μl/well of MSD SULFO-TAG labeled goat
anti-rabbit polyclonal antibody (R32AB-1, Meso Scale Discovery, 25 ng/well). Plates were
washed thrice with Tris Wash Buffer, added with 150 μl/well 2X Read Buffer (R92TC, Meso
Scale Discovery) before being read with an MSD Sector S600 plate reader (Meso Scale Discov-
ery). Signals generated were normalized to “lysis buffer-only” controls.
Experiments in mice
All animal experiments were performed on male C57BL/6 mice aged 3 months (purchased
from Harlan Laboratories, Blackthorn) housed in specific pathogen free conditions with free
access to water and standard chow. Mice were injected intraperitonially with the inhibitor (dis-
solved in Hanks’ buffered saline solution, pH 7.0, 5% DMSO final) or vehicle at the indicated
doses for the indicated time as previously described [26]. Tissues from treated mice were har-
vested and ‘snap frozen’ with liquid nitrogen before being stored at -80°C. For immunoblot-
ting, a portion of the tissues was removed, weighed and homogenised in urea/SDS buffer
before being analysed by immunoblotting with HIF1α rabbit polyclonal antibody (10006421,
Cayman Chemical, 1:1000) or HIF2α rabbit polyclonal antibody (PM9 [27], 1:1000). For RNA
extraction, tissues were ‘snap frozen’ in liquid nitrogen immediately after harvesting, then
ground to a fine powder using a cryogenic tissue pulveriser, keeping samples on dry ice at all
times and processed for total RNA extraction using the mirVana miRNA isolation kit
(AM1560, Life Technologies) according to the manufacturer’s protocol for processing frozen
tissues. All animal procedures were compliant with the UK Home Office Animals (Scientific
Procedures) Act 1986 and Local Ethical Review Procedures (University of Oxford Medical Sci-
ences Division Ethical Review Committee).
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
5 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
Real-time quantitative PCR analyses
Genomic DNA were removed from total RNA preparation using the TURBO DNA-free kit
(AM1907, Life Technologies) according to manufacturer’s protocol. Total RNA preparations
were reverse-transcribed to cDNA using the High Capacity cDNA kit (4374966, Life Technolo-
gies) according to manufacturer’s protocol. SYBR Green-based qRT-PCR were then performed
on the synthesised cDNA using Fast SYBR Green Master Mix (4385612, Life Technologies) on
an Applied Biosystem StepOnePlus thermocycler (Life Technologies). Relative quantifications
were calculated using the ΔΔCt method normalised to β-actin. The following primers were
used: Actb_fwd, CAGCCTTCCTTCTTGGGTATGG; Actb_rev, GAGGTCTTTACGGATGT
CAACG; Adm_fwd, CACCCTGATGTTATTGGGTTCA; Adm_rev, TTAGCGCCCACTT
ATTCCACT; Epo_fwd,; Epo_rev,; Il6_fwd, TGCTGGTGACAACCACGGCC; Il6_rev, GT
ACTCCAGAAGACCAGAGG; Ldh1_fwd, aggctccccagaacaagatt; Ldh1_rev, TCTCGCCCTT
GAGTTTGTCT; Vegfa_fwd, CCACGTCAGAGAGCAACATCA; Vegfa_rev, TCATTCTCT
CTATGTGCTGGCTTT.
Crystallography
Crystals of PHD2 (residues 181–426) in complex with Mn(II)/1 were grown in sitting drops at
293K using vapor diffusion methods at a protein to reservoir ratio of 1:1. PHD2 protein solu-
tion contained 32.0 mg/mL protein (in 50 mM TrisꢀHCl pH 7.5), 2.8 mM MnCl2 and 2.8 mM
1. The reservoir solution contained 1.6 M sodium citrate/citric acid pH 6.5 (Hampton Research
Crystal Screen-II). Mn(II) was used as a replacement for Fe(II) during crystallisation to prevent
metal oxidation and consequent formation of heterogenous sample. The crystals were cryopro-
tected using well solution that was diluted to 30% v/v glycerol and flash frozen in liquid nitro-
gen. Data were collected from a single crystal at 100K at the Diamond MX beamline I04-1
(0.9200 Å) equipped with a Pilatus 2M detector. The data were indexed, integrated and scaled
using MOSFLM and SCALA [28]. The structure was determined by molecular replacement
using PHASER [29] and PDB ID: 4BQY [9] as the search model. Iterative rounds of model
building (COOT [30]) and refinement (CNS [31] and PHENIX [32]) were performed until the
R and Rfree converged.
Modelling
A model predicting the binding mode for PHD2 in complex with Mn(II) and IOX4 was gener-
ated using the PHD2.Mn(II).1 structure as a template (PDB ID: 5A3U). Parameter and topol-
ogy files for IOX4 were generated using PRODRG [33]. The model was conjugate energy
minimized using CNS (version 1.3) [31] without applying external energy terms.
NMR experiments
For water relaxation NMR assays [34,35], apo-PHD2 (50 μM) was used. Solutions were buff-
ered using 50 mM Tris-D11 (pH 7.5) in 12.5% H2O and 87.5% D2O and 125 mM NaCl. The
final concentration of Mn(II) was 50 μM (final volume 160 μL). The water relaxation measure-
ments rely on paramagnetic relaxation enhancement. In our studies, paramagnetic Mn(II) was
used as a substitute for the Fe(II) in order to enhance the longitudinal relaxation rate of the
bulk water and to stop uncoupled turnover. When an inhibitor is bound to PHD2-Mn(II), it
displaces bound water molecules away from the paramagnetic Mn(II) at the PHD2 active site.
This leads to a net increase in the bulk water relaxation rate [34,35]. Experiments were
recorded using a Bruker AVII 500 MHz instrument equipped with a 5 mm inverse TXI probe,
and 3 mm MATCH tubes were used throughout. Saturation recovery (90°x−G1−90°y−G2−90°x
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
6 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
−G3−τ−acq) experiments were performed with 1 scan with a relaxation delay of at least 5 times
T1 between transients. The gradient pulses were achieved using 1 ms Sinebell gradient pulse
(G1 = 40%; G2 = 27.1%; G3 = 15%). The receiver gain was set to minimum value (rg = 1) to pre-
vent receiver overload. Typically, 10−16 delay points varied between 100 ms and 60 s were
used. T1 values were obtained using the Bruker T1/T2 relaxation option and peak area was
used for curve fitting. The titrant (typically *0.2 μL) was added using a 1 μL plunger-in-needle
syringe (SGE), and sample mixing was conducted using a 250 μL gas tight syringe (SGE). Bind-
ing constants were obtained by nonlinear curve fitting using OriginPro 8.0 (OriginLab) with
the equation previously described [36].
For 2OG displacement assays [37], selective 1H-13C 1D-HSQC experiments were conducted
at 700 MHz using a Bruker Avance III spectrometer equipped with an inverse TCI cryoprobe
optimized for 1H observation. The CLIP-HSQC sequence was used (without 13C decoupling).
Typical experimental parameters were as follows: acquisition time 0.58 s, relaxation delay 2 s,
number of transients 256−1600. The 1JCH was set to 160 Hz. For the selective version of the
experiment, a 6.8 ms Q3 180 degree pulse was used, and selective irradiation was applied at the
appropriate [13C] chemical shift. Three millimeter MATCH tubes with a 160 μL final sample
volume were used. Solutions were buffered using Tris-D11 50 mM (pH 7.5) dissolved in 90%
H2O and 10% D2O. Assays were conducted at 298 K in solutions typically containing 50 μM
apo-PHD2, 400 μM Zn(II), 50 μM 1,2,3,4-[13C]-2OG or [13C]-labeled CODD (uniformly
[13C]-labeled at proline-564) and 400 μM competitor (except unlabeled CODD competitor
used at 800 μM). Selective irradiation was applied at 30.5 ppm for [13C]-2OG and 24.25 ppm
for [13C]-labeled CODD. Percentage displacement was calculated according to equation:
ðI ꢂ I0Þ = ðIblank ꢂ I0Þ
where I0 is the intensity of the reporter in the presence of protein but without inhibitor, I is the
intensity of the reporter in the presence of protein and inhibitor, and Iblank is the intensity of
the reporter without protein or inhibitor.
Results
Validation of IOX4 as a potent and selective inhibitor of PHD2 in vitro
Using an antibody-based (AlphaScreen) in vitro hydroxylation assay for PHD2 catalysis [9],
both 1 and IOX4 were found to potently inhibit PHD2 with IC50 values of 4.8 nM and 1.6 nM,
respectively (Table 1 and S1 Fig). In comparison to previously identified PHD inhibitor IOX2
(Fig 1B, IOX2 IC50 = 22 nM) [9], both 1 and IOX4 are at least 4-fold more potent in vitro.
Given that the PHDs are part of the human 2OG-dependent oxygenase family which com-
prises ~60 members, we investigated the selectivity of 1 and IOX4 by screening against a panel
of human 2OG oxygenases, including the human Jumonji C (JmjC)-domain containing his-
tone demethylases (KDMs), γ-butyrobetaine hydroxylase (BBOX) and the fat mass and obesity
associated protein (FTO). The KDM assays utilised analogous AlphaScreen methodology as
the PHD2 CODD AlphaScreen assay (with enzyme concentrations within the range of 0.2 to
25 nM) [17], the BBOX assay employed a fluoride-release / detection based fluorescence assay
[18], the FTO assay is based on liquid chromatography-mass spectrometry (LC-MS) analysis
[19], and the FIH assay is a high-throughput mass spectrometry based assay (Agilent RapidFire
LC-MS). Using the IC50 values determined from each in vitro assay as an approximate measure
of selectivity, 1 and IOX4 are at least 875-fold more selective for PHD2 over all other tested
enzymes (Table 1). In comparison, IOX2 displays approximately 400-fold selectivity for
PHD2 over the same panel. Note that given the similarity of the catalytic domains of PHD1
and PHD3 to that of PHD2, it is likely that 1 and IOX4 also potently inhibits PHD1 and PHD3
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
7 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
Table 1. Selectivity profiling of the dihydropyrazoles 1 and IOX4 against a panel of human 2OG-dependent dioxygenases.
IOX2
NOG
1
IOX4
PHD2
0.022 μM [9]
>100 μM [38]
100 μM [38]
100 μM [38]
>100 μM [38]
>100 μM [38]
52 μM [38]
159 μM [38]
9.5 μM
11.2 μM [9]
0.17 μM [38]
0.2 μM [38]
0.6 μM [38]
0.3 μM [38]
0.14 μM [38]
15.4 μM [38]
25 μM [38]
0.39 μM [18]
2.3 μM
0.0048 μM
>20 μM
>20 μM
>20 μM
>20 μM
>20 μM
20 μM
0.0016 μM
>20 μM
>20 μM
>20 μM
>20 μM
1.4 μM
JMJD1A (KDM3A)
JMJD2A (KDM4A)
JMJD2C (KDM4C)
JMJD2E (KDM4E)
JMJD3 (KDM6B)
FBXL11 (KDM2A)
JARID1C (KDM5C)
BBOX1
20 μM
>100 μM
>50 μM
> 100 μM
37 μM
>100 μM
73 μM
FIH
> 100 μM
23.9 μM
49 μM
FTO
11 μM
4.4 μM [19]
The IC50 values obtained reveal the selectivity of dihydropyrazoles 1 and IOX4 for PHD2 in comparison with IOX2 and NOG. Assays were carried out as
previously reported [9,18,19,38].
PHD2: HIF-prolyl hydroxylase-2, JMJD1A (KDM3A): Lysine-specific demethylase 3A, JMJD2A (KDM4A): Lysine-specific demethylase 4A, JMJD2C
(KDM4C): Lysine-specific demethylase 4C, JMJD2E (KDM4E): Lysine-specific demethylase 4E, JMJD3 (KDM6B): Lysine-specific demethylase 6B,
FBXL11 (KDM2A): Lysine-specific demethylase 2A, JARID1C (KDM5C): Lysine-specific demethylase 5C, BBOX: γ-butyrobetaine hydroxylase, FIH: factor
inhibiting HIF, FTO: fat mass and obesity associated protein.
doi:10.1371/journal.pone.0132004.t001
(as supported by cell based work–see below). Although the panel is incomplete, the results sug-
gest that 1 and IOX4 are highly selective PHD inhibitors, at least over the 2OG-dependent
dioxygenases tested.
IOX4 inhibits PHD2 via binding to the 2OG binding site
To obtain insights into the mode of PHD inhibition by the dihydropyrazoles, we made
attempts to crystallize the PHD2 catalytic domain (residues 181–426) in complex with 1 or
IOX4; a crystal structure of PHD2.Mn(II).1 was obtained. In the PHD2.Mn(II).1 complex, the
inhibitor coordinates to Mn(II) in a bidentate manner via the nitrogen atoms of its pyridine
and pyrazolone rings (Fig 2A). In contrast to previously observed binding mode for other
PHD inhibitors occupying the 2OG binding site such as IOX3 (Fig 2B) [9], 2 (Fig 2E) [9] and
NOG (Fig 2H) [39], the pyridine-based chain of 1 extends towards the entrance of the active
site and is positioned to make electrostatic interactions with Arg-322 (which is known to be
involved in substrate binding) [39]. The 2OG C-5 carboxylate binding site of PHD2 which is
reported to bind a ligand carboxylate in other PHD2.inhibitor complex structures [40,41,42],is
occupied by the triazole ring of 1. Based on modelling studies (Fig 2D), IOX4, for which we
have not obtained a crystal structure, is predicted to bind in an analogous manner to 1, with its
tert-butyl group protruding towards the entrance of the active site.
To further study the mechanism of actions of 1 and IOX4 in solution, we employed a com-
petition-based nuclear magnetic resonance (NMR) method using labeled 2OG ([13C]-2OG) or
labeled HIF1α peptide fragment corresponding to the C-terminal oxygen-dependent degrada-
tion domain ([13C]-CODD) as reporter ligands [37]. The results reveal that the previously
described inhibitor IOX2 displaces 2OG from PHD2 (S3A Fig), consistent with its predicted
binding mode [9]. However, [13C]-CODD was not displaced by IOX2 within limits of detec-
tion, suggesting that HIF1α CODD can still bind to PHD2 in the presence of IOX2, likely in a
non-productive orientation that avoids the steric clash between the IOX2 benzyl group with
the hydroxylated CODD prolyl-residue). Previous crystallographic studies have shown that the
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
8 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
Fig 2. Comparison of the binding modes of PHD inhibitors. Views from crystal structures of PHD2 complexed with 1 (a), IOX3 (b), 2 (e) and NOG (h)
Compound 1 coordinates the active site metal in a bidentate manner via the nitrogens of its pyridine (trans to His374 Nε2) and pyrazolone (trans to the
Asp315 Oδ1) rings as shown in a. A model of IOX4 binding based on that of 1 (d) and the overlay of a and d (g) are shown for comparison. This coordination
mode enables 1 to competitively inhibit PHD2 with respect to 2OG (as observed with the other inhibitors described here); the triazole ring of 1 is located in the
2OG C-5 carboxylate binding site whilst the carboxylate side chain of 1 makes electrostatic interaction with another arginine, R322 (1 carboxylate O–NH1
R322, 2.9 Å) that is located at the entrance of the active site; R322 is directly involved in substrate binding (P564/HIF1α CODD O–NH1 R322/PHD2, 2.6 Å;
P564/CODD O–NH1 R322/PHD2, 2.8 Å) [39]. Compare a, b and c for differences in binding modes between 1 and IOX3; a, e and f for differences between 1
and 2; a, h and i for differences between 1 and NOG. PDB ID: 4BQX (PHD2.IOX3) [9], 4BQW (PHD2.IOX2) [9]; 3HQR (PHD2.NOG.CODD) [39].
doi:10.1371/journal.pone.0132004.g002
HIF substrate can still bind even when the 2OG active site is occupied by some, but not all,
inhibitors [39]. Studies with IOX4 gave similar results to that observed with IOX2, with
CODD (but not 2OG) remained bound to PHD2.IOX4 complex, suggesting that IOX4 com-
petes with and displaces 2OG at the active site of PHD2 (S3A Fig). In contrast, a different type
of PHD2 inhibitor, i.e. the bicyclic naphthalenylsulfone hydroxythiazole, BNS (see Fig 1B),
was found to displace both 2OG and CODD (S3A Fig), consistent with previous observations
[35]. Although we did not study NODD inhibition, given that the same active site is responsible
for the NODD and CODD hydroxylation [39,40], 1 and IOX4 are likely to inhibit NODD
hydroxylation in a similar manner to that for CODD, as subsequently shown by cellular studies
(see below). Together, the crystallographic and NMR-based studies suggest that the dihydro-
pyrazoles 1 and IOX4 inhibit PHD2 via binding to the 2OG binding site, but do not displace
the HIFα substrate of PHD2.
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
9 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
IOX4 inhibits HIF prolyl hydroxylase activity in cells
We then tested the cellular inhibition of 1 and IOX4 using a von Hippel-Lindau (VHL)-defec-
tive human renal cell carcinoma (RCC4) cell line, in which HIFα proteins are constitutively
stabilized [22]. In this cell line, the extent of the differential HIF1α prolyl hydroxylation at the
N-terminal and C-terminal oxygen-dependent degradation domains (HyPro402 and
HyPro564, respectively), as well as HIF1α asparaginyl hydroxylation at the C-terminal transac-
tivation domain (HyAsn803) can be interrogated by immunoblotting using hydroxylation spe-
cific antibodies [22,23]. Dimethyloxalylglycine, DMOG (a cell-permeable ester of the generic
2OG-dependent dioxygenase inhibitor, NOG) was used as a control. The results show that
both 1 and IOX4 block prolyl-hydroxylation of HIF1α at both NODD (HyPro402) and CODD
(HyPro564) without markedly affecting the levels of PHD2, consistent with their proposed
mode of action via the displacement of 2OG (Fig 3A and 3B). IOX4 was clearly more potent
in inhibiting HIF1α prolyl-hydroxylation than IOX2 (Fig 3B); some apparent inhibition of
HIF1α asparaginyl-hydroxylation (HyAsn803) was observed at 50 μM of IOX4. In VHL-com-
petent HeLa cells, both 1 and IOX4 induced HIF1α, with IOX4 being substantially more
potent (activity being observed at ꢁ 1 μM, Fig 3C). Induction of HIF1α levels were observed in
MCF-7, Hep3B and U2OS cells treated with IOX2 and IOX4, with IOX4 apparently being
markedly more potent than IOX2 (Fig 3D–3F). These cellular results are consistent with the
in vitro data indicating that IOX4 is a substantially more potent PHD inhibitor than IOX2.
To quantify the cellular efficiency of PHD inhibition, we developed an electrochemilumi-
nescence-based assay for the quantification of HIF1α levels using the Meso Scale Discovery
(MSD, http://www.mesoscale.com) technology (HIF1α MSD assay), which is based on the cap-
ture and detection of HIF1α protein using two specific HIF1α antibodies. The technology of
this immunoassay is similar to that of the sandwich enzyme-linked immunosorbent assay
(ELISA), but utilises electrochemiluminescence-based signal generation and detection [43] (as
opposed to the colorimetric-based detection employed in ELISA). The linearity of this assay
detection signal was demonstrated using lysates from Hep3B cells treated for 24 h with iron-
chelating deferoxamine, DFO to induce HIF1α, with low signal detected in untreated Hep3B
cell lysates (S4 Fig). Using this assay, we determined the half-maximal effective concentration
(EC50) for HIF1α induction in three human cell lines (MCF-7, Hep3B and U2OS) after 5 h
treatment with IOX2 or IOX4. Consistent with the immunoblotting observations, IOX4
(HIF1α EC50 values = 11.7 μM, 11.1 μM and 5.6 μM in MCF-7, Hep3B and U2OS, respectively)
is more potent than IOX2 (HIF1μ EC50 values = 114 μM, 86 μM and 49.5 μM in MCF-7,
Hep3B and U2OS, respectively), in all three cell lines at a suitable concentration in cells), with
IOX4 being more potent than IOX2 in all the tested cell lines.
HIFα proteins are upregulated in tissues of mice treated with IOX4
To explore their utility in a mammalian animal model, IOX2 and IOX4 were tested for their
ability to induce HIFα in mice. The inhibitors or vehicle controls were injected intraperitonially
into wild type C57BL/6 mice and sacrificed after the indicated treatment time before harvesting
their tissues to be analysed for HIFα levels by immunoblotting. The results reveal HIF1α is
induced in mouse liver after 1 h of treatment of IOX2, which persisted even after 2.5 h of treat-
ment albeit at a lower level than the shorter treatment (S5 Fig). In contrast, treatment with
another PHD inhibitor FG2216/IOX3 (see S1B Fig, previously used in mice studies) [26,44] at
an equivalent molar dose to that of IOX2 led to a lower and shorter induction of HIF1α (S5
Fig). To compare the potencies of IOX2, IOX4 and DMOG, mice were treated with equimolar
concentrations of each inhibitor for 1 h. Immunoblot analyses across multiple tissues reveal
induction of HIF1α and HIF2α by all three inhibitors, with IOX2 displaying the strongest
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
10 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
11 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
Fig 3. Cellular inhibition of HIF prolyl-hydroxylases by IOX4 leads to HIFα induction. (a-b) Immunoblots showing selective inhibition of the HIF1α prolyl-,
over asparaginyl-hydroxylation in HIFα-stablized RCC4 cells by 1, IOX2 and IOX4. (c) Immunoblots showing the dose-dependent upregulation of HIF1α in HeLa
cells by 1 and IOX4. (d-f) Immunoblots highlighting the dose-dependent induction of HIF1α in MCF-7 (d), Hep3B (e) and U2OS (f) cells by IOX2 and IOX4. Note
the higher potency of IOX4 compared to IOX2, and the lack of inhibition of FIH-catalyzed HIF1α asparaginyl-hydroxylation at concentrations in which maximal
HIF1α induction was observed. (g) The dose-dependent upregulation of HIF1α in MCF-7, Hep3B and U2OS cells by IOX2 and IOX4 as measured using a
quantitative HIF1α immunoassay. Each data point represents the average signal standard deviation, n = 3. HyPro402: hydroxyproline402; HyPro564:
hydroxyproline564; HyAsn803: hydroxyasparagine803; l.e.: long exposure. See Materials and Methods for details (including antibodies used).
doi:10.1371/journal.pone.0132004.g003
induction followed by IOX4 and DMOG in the liver, kidney and heart (Fig 4A). Importantly,
both HIF1α and HIF2α were induced in the brain by IOX4 but not IOX2 or DMOG, suggest-
ing that IOX4 may better penetrate the blood-brain barrier than the latter two inhibitors.
Dose-dependent induction of both HIF1α and HIF2α in the liver and brain was observed after
IOX4 treatment (Fig 4B and 4C). These observations reveal both IOX2 and IOX4 as active
PHD inhibitors in mice; however, induction of HIF1α and HIF2α protein levels in the brain
were only observed with IOX4.
To investigate whether the induction of HIFα in the mouse brain has an effect on HIF target
genes, we performed quantitative real-time PCR (qRT-PCR) analyses on HIF target genes
(including those involved in angiogenesis, i.e. Vegfa, Adm and Il6). The analyses reveal that the
mRNA levels of some of the HIF target genes tested (Epo, Vegfa and Adm) were induced in a
dose-dependent manner after 1 h treatment with IOX4, whereas others (Bnip3, Ldha and Il6)
Fig 4. IOX4 induces HIFα in mice. (a) Immunoblots showing HIF1α and HIF2α induction in various mouse tissues (liver, brain, kidney, heart) after 1 h
treatment at equimolar concentrations of IOX2 (37.7 mg/kg), IOX4 (35 mg/kg) or dimethyl N-oxalylglycine DMOG (75 mg/kg). (b-c) Immunoblot showing
dose-dependent induction of HIF1α and HIF2α in the mouse liver (b) and in the mouse brain (c) after 1 h treatment by various doses of IOX4 (17.5 to 70 mg/
kg) in comparison to vehicle control and DMOG (160 mg/kg). n.s.: non-specific; l.e.: long exposure.
doi:10.1371/journal.pone.0132004.g004
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
12 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
remain unaffected (S6 Fig). These results indicate that the HIFα proteins induced by IOX4 are
transcriptionally active and could lead to the induction of a subset of HIF target genes.
Discussion
The combined results reveal IOX4 as a highly potent and selective inhibitor of human PHD2,
As shown by studies with mice, IOX4 is useful for in vivo work; it will then be useful for inves-
tigations on the suitability of the PHDs as targets for cerebral diseases such as stroke. The
combined kinetic and biophysical analyses reveal that IOX4 (and IOX2) compete with 2OG
for binding to PHD2; the triazole rings of the inhibitors bind in the pocket occupied by the
CH2CH2COOH side chain of 2OG. The results thus reveal the potential of non-acid containing
PHD inhibitors—an important finding given the potential of HIFα upregulation mediated by
PHD inhibition in the treatment of stroke, as acids do not often permeate the blood-brain bar-
rier efficiently.
Notably, the NMR results show that IOX2 and IOX4 bind to PHD2 such that they do not,
displace the HIF1α substrate from the active site (though they must alter its binding mode /
strength of binding). Thus, PHD inhibitors that compete with 2OG may be classified into
those that still enable the formation of stable PHD.HIFα complexes (such as N-oxalylglycine,
IOX2 and IOX4), and those that strongly displace HIFα from the PHDs (such as BNS [35]).
The physiological consequences of these different modes of inhibition are unclear, but may
become apparent with the clinical use of the different categories of PHD inhibitor. IOX2 and
IOX4 are highly selective PHD inhibitors–this is important given the roles of other 2OG oxy-
genases; FIH has multiple substrates other than HIFα [45,46,47,48,49,50,51] and the Jumonji-
C histone demethylases (KDMs) are generally important in the regulation of gene expression
[52].
Studies in human cell lines support the efficacy of IOX4 as a potent and selective PHD
inhibitor, with a maximal level of HIFα induction being observed at low μM concentrations
across all the cell lines tested. Although we did not test IOX4 with PHD1 and PHD3, the potent
inhibition of NODD and CODD in cell lines containing all three PHDs [24] implies that IOX4
also inhibits PHD1 and PHD3, as anticipated given the similarity of the catalytic domain of the
three human PHDs [41]. The use of selective antibodies [22] illustrates the selectivity (within
limits of detection) of the inhibitors for the PHDs over FIH. It is notable that the concentra-
tions of PHD inhibitors required to maximally induce HIFα vary across the different cell lines.
Aside from other considerations (e.g., uptake, export, metabolism), this observation may, in
part, reflect the different levels of the PHD isoforms in different cell types [24]. Given that the
quantitative control of HIF target genes (e.g., EPO and VEGF) is of clinical importance in
terms of PHD inhibitors use, the effective concentrations and the duration of treatment of any
PHD inhibitor are therefore important considerations.
The induction of HIFα in mice treated with IOX4 reflects the in vivo utility of this inhibitor.
When compared with the results for IOX2, lower HIFα induction by IOX4 was observed in
the liver, heart and kidney; however markedly higher induction by IOX4 was observed in the
brain. As with the variations in the results obtained with different cell lines, the reasons for
these organ-specific effects are unclear; in part they likely reflect differences in the transport
(uptake/export) and/or metabolism of these inhibitors in the different tissues, but they may
also reflect differences in the context dependent regulation of the HIF system. Importantly,
IOX4 is effective at inducing HIFα in the mouse brain–to our knowledge, this is the first selec-
tive PHD inhibitor that has been reported to do so in wildtype, uncompromised mice.
Although activity in the brain may be an undesirable property of PHD inhibitors aimed at
treating anaemia, inhibition of the PHDs is proposed as being protective in ischaemic and
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
13 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
hemorrhagic stroke models where it is aimed at inducing blood vessel formation [13,14]. HIF
target genes such as Epo, Vegfa and Adm were induced in the mouse brain in a dose-dependent
manner after treatment with IOX4. However, not all HIF target genes were affected (within
limits of detection), as highlighted by the lack of induction of Bnip3, Il6 and Ldha. The differing
effects in the brain of the PHD inhibitors tested could in part be due to the more hydrophobic
nature of IOX4 compared to IOX2, resulting in an improved blood-brain barrier penetration
for IOX4, though other reasons cannot be ruled out. Note also that there is likely considerable
scope for improving the blood brain barrier permeating ability of IOX4 in terms of lipophili-
city and polar surface area [53,54]. The identification of IOX4 as a selective PHD inhibitor that
induces HIFα and its target genes in the brain may thus be a productive step towards validating
the PHDs as therapeutic targets for stroke.
Supporting Information
S1 Table. Crystallographic data processing and refinement statistics.
(DOCX)
S1 Fig. Inhibition of PHD2 activity in vitro by IOX4. (a) Dose-dependent inhibition of
PHD2 activity by the dihydropyrazoles 1 (IC50 = 4.8 nM) and IOX4 (IC50 = 1.6 nM) in com-
parison to IOX2 (IC50 = 22 nM), as determined by the AlphaScreen assay. Each datapoint rep-
resents the average standard deviation, nꢁ3.
(TIF)
S2 Fig. View of the tPHD2.Mn(II). 1 crystal structure showing the Fo-Fc OMIT map (con-
toured to 3σ) for 1.
(TIF)
S3 Fig. NMR studies on the displacement of 2OG and CODD by IOX4. (a) Percentage dis-
placement of [13C]-2OG and [13C]-CODD using an NMR based assay by IOX2 and IOX4.
BNS, unlabeled CODD and unlabeled NODD were used as controls. Errors shown represent
the standard deviation from the means of three separate measurements. (b) NMR spectra for
the displacement of [13C]-2OG and/or [13C]-CODD by the PHD inhibitors tested. The posi-
tions of the labeled carbon atoms in [13C]-2OG are as indicated. [13C]-CODD is uniformly
labeled at every carbon atom of the highlighted proline residue. (c) KD determination for com-
pounds 2 and IOX2 by a water relaxation-based method. CODD: HIF1α C-terminal oxygen-
dependent degradation domain, NODD: HIF1α N-terminal oxygen-dependent degradation
domain. See Materials and Methods for assay details.
(TIF)
S4 Fig. The detection of HIF1α using the quantitative HIF1α immunoassay based on the
Meso Scale Discovery assay technology. Lysates from Hep3B cells DFO (24 h treatment)
were serially diluted and assayed for the presence of HIF1α using the MSD assay. Each data
point represents the average signal standard deviation, n = 2.
(TIF)
S5 Fig. Induction of HIFα by IOX2 and IOX3 in mice. (a) Induction of HIF1α in the mouse
liver by IOX2 (37.7 mg/kg) in comparison to vehicle control, DMOG (320 mg/kg) and IOX3
(30 mg/kg).
(TIF)
S6 Fig. Induction of HIFα target genes by IOX4 in mouse brain tissues. Some of the HIF
target genes (Epo, Vegfa and Adm) were induced by IOX4 in a dose-dependent manner in the
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
14 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
mouse brain tissues after 1 h of treatment. However, other HIF target genes (Bnip3, Ldha and
Il6) were not markedly affected.
(TIF)
S7 Fig. 13C-NMR and 1H-NMR for t-Butyl 6-(5-oxo-4-(1H-1,2,3-triazol-1-yl)-2,5-dihydro-
1H-pyrazol-1-yl)nicotinate.
(TIF)
S8 Fig. Synthetic scheme for IOX4. See Materials and Methods for details of synthesis.
(TIF)
Author Contributions
Conceived and designed the experiments: MCC EH AT IKHL RC TB TDWC AK CWP PJR
CJS. Performed the experiments: MCC OA EH AT IKHL RC VG MD AR JH TB. Analyzed the
data: MCC OA EH AT IKHL RC VG MD AR JH YT TB TDWC AK CWP PJR CJS. Wrote the
paper: MCC CJS.
References
1. Schofield CJ, Ratcliffe PJ (2004) Oxygen sensing by HIF hydroxylases. Nat Rev Mol Cell Biol 5: 343–
354. PMID: 15122348
2. Maxwell PH, Wiesener MS, Chang GW, Clifford SC, Vaux EC, Cockman ME, et al. (1999) The tumour
suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature
399: 271–275. PMID: 10353251
3. Ivan M, Kondo K, Yang H, Kim W, Valiando J, Ohh M, et al. (2001) HIFalpha targeted for VHL-mediated
destruction by proline hydroxylation: implications for O2 sensing. Science 292: 464–468. PMID:
11292862
4. Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, et al. (2001) Targeting of HIF-alpha
to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 292:
468–472. PMID: 11292861
5. Epstein AC, Gleadle JM, McNeill LA, Hewitson KS, O'Rourke J, Mole DR, et al. (2001) C. elegans EGL-
9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hydroxylation.
Cell 107: 43–54. PMID: 11595184
6. Bruick RK, McKnight SL (2001) A conserved family of prolyl-4-hydroxylases that modify HIF. Science
294: 1337–1340. PMID: 11598268
7. Lando D, Peet DJ, Whelan DA, Gorman JJ, Whitelaw ML (2002) Asparagine hydroxylation of the HIF
transactivation domain a hypoxic switch. Science 295: 858–861. PMID: 11823643
8. Hewitson KS, McNeill LA, Riordan MV, Tian YM, Bullock AN, Welford RW, et al. (2002) Hypoxia-induc-
ible factor (HIF) asparagine hydroxylase is identical to factor inhibiting HIF (FIH) and is related to the
cupin structural family. J Biol Chem 277: 26351–26355. PMID: 12042299
9. Chowdhury R, Candela-Lena JI, Chan MC, Greenald DJ, Yeoh KK, Tian YM et al. (2013) Selective
small molecule probes for the hypoxia inducible factor (HIF) prolyl hydroxylases. ACS Chem Biol 8:
1488–1496. doi: 10.1021/cb400088q PMID: 23683440
10. Eltzschig HK, Bratton DL, Colgan SP (2014) Targeting hypoxia signalling for the treatment of ischaemic
and inflammatory diseases. Nat Rev Drug Discov 13: 852–869. doi: 10.1038/nrd4422 PMID:
25359381
11. Rabinowitz MH (2013) Inhibition of hypoxia-inducible factor prolyl hydroxylase domain oxygen sensors:
tricking the body into mounting orchestrated survival and repair responses. J Med Chem 56: 9369–
9402. doi: 10.1021/jm400386j PMID: 23977883
12. Yan L, Colandrea VJ, Hale JJ (2010) Prolyl hydroxylase domain-containing protein inhibitors as stabiliz-
ers of hypoxia-inducible factor: small molecule-based therapeutics for anemia. Expert Opin Ther Pat
20: 1219–1245. doi: 10.1517/13543776.2010.510836 PMID: 20698812
13. Karuppagounder SS, Ratan RR (2012) Hypoxia-inducible factor prolyl hydroxylase inhibition: robust
new target or another big bust for stroke therapeutics? J Cereb Blood Flow Metab 32: 1347–1361. doi:
10.1038/jcbfm.2012.28 PMID: 22415525
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
15 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
14. Harten SK, Ashcroft M, Maxwell PH (2010) Prolyl hydroxylase domain inhibitors: a route to HIF activa-
tion and neuroprotection. Antioxid Redox Signal 12: 459–480. doi: 10.1089/ars.2009.2870 PMID:
19737089
15. Bigge CF, Bridges AJ, Casimiro-Garcia A, Fakhoury SA, Lee HT, Reed JE, et al. (2004) Oral antidia-
betic agents. 2003018553, PCT Int Appl.
16. Thede K, Flamme I, Oehme F, Ergueden JK, Stoll F, Schuhmacher J, et al. (2008) Substituted dihydro-
pyrazolones for treating cardiovascular and haematological diseases. 2008067871, PCT Int Appl.
17. Rose NR, Woon EC, Tumber A, Walport LJ, Chowdhury R, Li XS, et al. (2012) Plant growth regulator
daminozide is a selective inhibitor of human KDM2/7 histone demethylases. J Med Chem 55: 6639–
6643. doi: 10.1021/jm300677j PMID: 22724510
18. Rydzik AM, Leung IK, Kochan GT, Thalhammer A, Oppermann U, Claridge TD, et al. (2012) Develop-
ment and application of a fluoride-detection-based fluorescence assay for gamma-butyrobetaine
hydroxylase. Chembiochem 13: 1559–1563. doi: 10.1002/cbic.201200256 PMID: 22730246
19. Aik W, Demetriades M, Hamdan MK, Bagg EA, Yeoh KK, Lejeune C, et al. (2013) Structural basis for
inhibition of the fat mass and obesity associated protein (FTO). J Med Chem 56: 3680–3688. doi: 10.
1021/jm400193d PMID: 23547775
20. Hardy AP, Prokes I, Kelly L, Campbell ID, Schofield CJ (2009) Asparaginyl beta-hydroxylation of pro-
teins containing ankyrin repeat domains influences their stability and function. J Mol Biol 392: 994–
1006. doi: 10.1016/j.jmb.2009.07.070 PMID: 19646994
21. Kelly L, McDonough MA, Coleman ML, Ratcliffe PJ, Schofield CJ (2009) Asparagine beta-hydroxyl-
ation stabilizes the ankyrin repeat domain fold. Mol Biosyst 5: 52–58. doi: 10.1039/b815271c PMID:
19081931
22. Tian YM, Yeoh KK, Lee MK, Eriksson T, Kessler BM, Kramer HB, et al. (2011) Differential sensitivity of
hypoxia inducible factor hydroxylation sites to hypoxia and hydroxylase inhibitors. J Biol Chem 286:
13041–13051. doi: 10.1074/jbc.M110.211110 PMID: 21335549
23. Lee SH, Jeong Hee M, Eun Ah C, Ryu SE, Myung Kyu L (2008) Monoclonal antibody-based screening
assay for factor inhibiting hypoxia-inducible factor inhibitors. J Biomol Screen 13: 494–503. doi: 10.
1177/1087057108318800 PMID: 18566480
24. Appelhoff RJ, Tian YM, Raval RR, Turley H, Harris AL, Pugh CW, et al. (2004) Differential function of
the prolyl hydroxylases PHD1, PHD2, and PHD3 in the regulation of hypoxia-inducible factor. J Biol
Chem 279: 38458–38465. PMID: 15247232
25. Stolze IP, Tian YM, Appelhoff RJ, Turley H, Wykoff CC, Gleadle JM, et al. (2004) Genetic analysis of
the role of the asparaginyl hydroxylase factor inhibiting hypoxia-inducible factor (FIH) in regulating hyp-
oxia-inducible factor (HIF) transcriptional target genes [corrected]. J Biol Chem 279: 42719–42725.
PMID: 15302861
26. Bishop T, Talbot NP, Turner PJ, Nicholls LG, Pascual A, Hodson EJ, et al. (2013) Carotid body hyper-
plasia and enhanced ventilatory responses to hypoxia in mice with heterozygous deficiency of PHD2. J
Physiol 591: 3565–3577. doi: 10.1113/jphysiol.2012.247254 PMID: 23690557
27. Lau KW, Tian YM, Raval RR, Ratcliffe PJ, Pugh CW (2007) Target gene selectivity of hypoxia-inducible
factor-alpha in renal cancer cells is conveyed by post-DNA-binding mechanisms. Br J Cancer 96:
1284–1292. PMID: 17387348
28. Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans PR, et al. (2011) Overview of the
CCP4 suite and current developments. Acta Crystallogr D Biol Crystallogr 67: 235–242. doi: 10.1107/
S0907444910045749 PMID: 21460441
29. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ. (2007) Phaser crystal-
lographic software. J Appl Crystallogr 40: 658–674. PMID: 19461840
30. Emsley P, Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol
Crystallogr 60: 2126–2132. PMID: 15572765
31. Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, et al. (1998) Crystal-
lography & NMR system: A new software suite for macromolecular structure determination. Acta Crys-
tallogr D Biol Crystallogr 54: 905–921. PMID: 9757107
32. Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, et al. (2010) PHENIX: a compre-
hensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystal-
logr 66: 213–221. doi: 10.1107/S0907444909052925 PMID: 20124702
33. Schuttelkopf AW, van Aalten DM (2004) PRODRG: a tool for high-throughput crystallography of pro-
tein-ligand complexes. Acta Crystallogr D Biol Crystallogr 60: 1355–1363. PMID: 15272157
34. Bertini I, Fragai M, Luchinat C, Talluri E (2008) Water-based ligand screening for paramagnetic metallo-
proteins. Angew Chem Int Ed Engl 47: 4533–4537. doi: 10.1002/anie.200800327 PMID: 18465767
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
16 / 17

Potent and Selective HIF Prolyl-Hydroxylase Inhibitors
35. Leung IK, Flashman E, Yeoh KK, Schofield CJ, Claridge TD (2010) Using NMR solvent water relaxation
to investigate metalloenzyme-ligand binding interactions. J Med Chem 53: 867–875. doi: 10.1021/
jm901537q PMID: 20025281
36. Morton CJ, Pugh DJ, Brown EL, Kahmann JD, Renzoni DA, Campbell ID (1996) Solution structure and
peptide binding of the SH3 domain from human Fyn. Structure 4: 705–714. PMID: 8805554
37. Leung IK, Demetriades M, Hardy AP, Lejeune C, Smart TJ, Szöllössi A, et al. (2013) Reporter ligand
NMR screening method for 2-oxoglutarate oxygenase inhibitors. J Med Chem 56: 547–555. doi: 10.
1021/jm301583m PMID: 23234607
38. Hopkinson RJ, Tumber A, Yapp C, Chowdhury R, Aik W, Che KH, et al. (2013) 5-Carboxy-8-hydroxy-
quinoline is a broad spectrum 2-oxoglutarate oxygenase inhibitor which causes iron translocation.
Chem Sci 4: 3110–3117.
39. Chowdhury R, McDonough MA, Mecinovic J, Loenarz C, Flashman E, Hewitson KS, et al. (2009) Struc-
tural basis for binding of hypoxia-inducible factor to the oxygen-sensing prolyl hydroxylases. Structure
17: 981–989. doi: 10.1016/j.str.2009.06.002 PMID: 19604478
40. Aik W, McDonough MA, Thalhammer A, Chowdhury R, Schofield CJ (2012) Role of the jelly-roll fold in
substrate binding by 2-oxoglutarate oxygenases. Curr Opin Struct Biol 22: 691–700. doi: 10.1016/j.sbi.
2012.10.001 PMID: 23142576
41. McDonough MA, Li V, Flashman E, Chowdhury R, Mohr C, Liénard BM, et al. (2006) Cellular oxygen
sensing: Crystal structure of hypoxia-inducible factor prolyl hydroxylase (PHD2). Proc Natl Acad Sci
USA 103: 9814–9819. PMID: 16782814
42. Rose NR, McDonough MA, King ON, Kawamura A, Schofield CJ (2011) Inhibition of 2-oxoglutarate
dependent oxygenases. Chem Soc Rev 40: 4364–4397. doi: 10.1039/c0cs00203h PMID: 21390379
43. Leland JK, Powell MJ (1990) Electrogenerated chemiluminescence: An oxidative-reduction type ECL
reaction sequence using tripropylamine. J Electrochem Soc 137: 3127–3131.
44. Hsieh MM, Linde NS, Wynter A, Metzger M, Wong C, Langsetmo I, et al. (2007) HIF prolyl hydroxylase
inhibition results in endogenous erythropoietin induction, erythrocytosis, and modest fetal hemoglobin
expression in rhesus macaques. Blood 110: 2140–2147. PMID: 17557894
45. Cockman ME, Lancaster DE, Stolze IP, Hewitson KS, McDonough MA, Coleman ML, et al. (2006)
Posttranslational hydroxylation of ankyrin repeats in IkappaB proteins by the hypoxia-inducible factor
(HIF) asparaginyl hydroxylase, factor inhibiting HIF (FIH). Proc Natl Acad Sci USA 103: 14767–14772.
PMID: 17003112
46. Coleman ML, McDonough MA, Hewitson KS, Coles C, Mecinovic J, Edelmann M, et al. (2007) Aspara-
ginyl hydroxylation of the Notch ankyrin repeat domain by factor inhibiting hypoxia-inducible factor. J
Biol Chem 282: 24027–24038. PMID: 17573339
47. Ferguson JE 3rd, Wu Y, Smith K, Charles P, Powers K, Wang H, et al. (2007) ASB4 is a hydroxylation
substrate of FIH and promotes vascular differentiation via an oxygen-dependent mechanism. Mol Cell
Biol 27: 6407–6419. PMID: 17636018
48. Cockman ME, Webb JD, Kramer HB, Kessler BM, Ratcliffe PJ (2009) Proteomics-based identification
of novel factor inhibiting hypoxia-inducible factor (FIH) substrates indicates widespread asparaginyl
hydroxylation of ankyrin repeat domain-containing proteins. Mol Cell Proteomics 8: 535–546. doi: 10.
1074/mcp.M800340-MCP200 PMID: 18936059
49. Webb JD, Muranyi A, Pugh CW, Ratcliffe PJ, Coleman ML (2009) MYPT1, the targeting subunit of
smooth-muscle myosin phosphatase, is a substrate for the asparaginyl hydroxylase factor inhibiting
hypoxia-inducible factor (FIH). Biochem J 420: 327–333. doi: 10.1042/BJ20081905 PMID: 19245366
50. Zheng X, Linke S, Dias JM, Gradin K, Wallis TP, Hamilton BR, et al. (2008) Interaction with factor inhib-
iting HIF-1 defines an additional mode of cross-coupling between the Notch and hypoxia signaling path-
ways. Proc Natl Acad Sci USA 105: 3368–3373. doi: 10.1073/pnas.0711591105 PMID: 18299578
51. Yang M, Ge W, Chowdhury R, Claridge TD, Kramer HB, Schmierer B, et al. (2011) Asparagine and
aspartate hydroxylation of the cytoskeletal ankyrin family is catalyzed by factor-inhibiting hypoxia-
inducible factor. J Biol Chem 286: 7648–7660. doi: 10.1074/jbc.M110.193540 PMID: 21177872
52. Klose RJ, Kallin EM, Zhang Y (2006) JmjC-domain-containing proteins and histone demethylation. Nat
Rev Genet 7: 715–727. PMID: 16983801
53. van de Waterbeemd H, Camenisch G, Folkers G, Chretien JR, Raevsky OA (1998) Estimation of
blood-brain barrier crossing of drugs using molecular size and shape, and H-bonding descriptors. J
Drug Target 6: 151–165. PMID: 9886238
54. Clark DE (2011) What has polar surface area ever done for drug discovery? Future Med Chem 3: 469–
484. doi: 10.4155/fmc.11.1 PMID: 21452982
PLOS ONE | DOI:10.1371/journal.pone.0132004 July 6, 2015
17 / 17