Utilization of an in vivo reporter for high throughput identification of branched small molecule regulators of hypoxic adaptation.

Abstract

Small molecules inhibiting hypoxia inducible factor (HIF) prolyl hydroxylases (PHDs) are the focus of drug development efforts directed toward the treatment of ischemia and metabolic imbalance. A cell-based reporter produced by fusing HIF-1 alpha oxygen degradable domain (ODD) to luciferase was shown to work as a capture assay monitoring stability of the overexpressed luciferase-labeled HIF PHD substrate under conditions more physiological than in vitro test tubes. High throughput screening identified novel catechol and oxyquinoline pharmacophores with a "branching motif" immediately adjacent to a Fe-binding motif that fits selectively into the HIF PHD active site in in silico models. In accord with their structure-activity relationship in the primary screen, the best "hits" stabilize HIF1 alpha, upregulate known HIF target genes in a human neuronal line, and exert neuroprotective effects in established model of oxidative stress in cortical neurons.

Full text

Page 1

Chemistry & Biology
Article
Utilization of an In Vivo Reporter for High
Throughput Identification of Branched
Small Molecule Regulators of Hypoxic Adaptation
Natalya A. Smirnova,1Ilay Rakhman,1Natalia Moroz,1Manuela Basso,1Jimmy Payappilly,1Sergey Kazakov,2
Francisco Hernandez-Guzman,3Irina N. Gaisina,4Alan P. Kozikowski,4Rajiv R. Ratan,1,* and Irina G. Gazaryan1,*
1Burke Medical Research Institute, Department of Neurology and Neuroscience, Weill Medical College of Cornell University,
785 Mamaroneck Ave, White Plains, NY 10605, USA
2Department of Chemistry and Physical Sciences, Pace University, 861 Bedford Road, Pleasantville, NY 10570, USA
3Accelrys Inc., 10188 Telesis Ct. Suite 100, San Diego, CA 92121, USA
4Department of Medicinal Chemistry and Pharmacognosy, University of Illinois at Chicago, 833 South Wood Street, Chicago, IL 60612, USA
*Correspondence: igazarya@burke.org (I.G.G.), rratan@burke.org (R.R.R.)
DOI 10.1016/j.chembiol.2010.03.008
SUMMARY
Small molecules inhibiting hypoxia inducible factor
(HIF) prolyl hydroxylases (PHDs) are the focus of
drug development efforts directed toward the treat-
ment of ischemia and metabolic imbalance. A cell-
based reporter produced by fusing HIF-1a oxygen
degradable domain (ODD) to luciferase was shown
to work as a capture assay monitoring stability of
the overexpressed luciferase-labeled HIF PHD sub-
strate under conditions more physiological than
in vitro test tubes. High throughput screening identi-
fied novel catechol and oxyquinoline pharmaco-
phores with a ‘‘branching motif’’ immediately adja-
cent to a Fe-binding motif that fits selectively into
the HIF PHD active site in in silico models. In accord
with their structure-activity relationship in the pri-
mary screen, the best ‘‘hits’’ stabilize HIF1a, upregu-
lateknown HIFtargetgenesinahuman neuronalline,
and exert neuroprotective effects in established
model of oxidative stress in cortical neurons.
INTRODUCTION
Hypoxia is a common etiology of cell injury in human disease,
including stroke, myocardial infarction, and solid tumors. Over
the past two decades, cell adaptation to hypoxia has emerged
as a well-defined active process. Each cell of a multicellular
organism can respond to hypoxia by building up hypoxia induc-
ible factor (HIF), a ubiquitous transcription factor capable of acti-
vating a battery of genes including genes involved in glucose
uptake and metabolism, extracellular pH control, angiogenesis,
erythropoiesis, mitogenesis, and apoptosis. The discovery of
HIF opened new horizons for the treatment of ischemia and
cancer: upregulation of HIF levels has been shown to be benefi-
cial for ischemic diseases, stem cell proliferation (Zhang et al.,
2006), and transplantation (Liu et al., 2009), whereas downregu-
lation of elevated HIF, a marker of most aggressive cancers,
represents a new approach for cancer treatment.
HIF consists of two subunits, HIF-1a and HIF-1b; HIF-1a is
rapidly degraded under normoxic conditions, whereas HIF-1b
is stable (Wang et al., 1995; Wang and Semenza, 1995). HIF
levels are regulated primarily by posttranslational modification
of conserved proline residues. Hydroxylation of Pro564 and/or
402 residues in HIF-1a is a prerequisite for its interaction with
the von Hippel-Lindau (VHL) protein yielding a complex that pro-
vides HIF ubiquitinylation and subsequent proteasomal degra-
dation (Kaelin, 2005). Hydroxylation of Pro564 occurs prior to
Pro402 (Chan et al., 2005), though some experiments contradict
this finding (Villar et al., 2007). Hydroxylation of HIF-1a Asn803
blocks its interaction with transcriptional proactivator p300
(Lando et al., 2002). In both cases HIF hydroxylation is executed
by a-KG dependent non-heme iron dioxygenases, HIF prolyl-4-
hydroxylase (PHD1-3 isozymes) and asparaginyl hydroxylase
(or the so-called FIH, factor inhibiting HIF) (Hirota and Semenza,
2005).
HIF1alsoupregulatesanumberofprodeathproteins,andthus
HIF1 upregulation can be either prodeath or prosurvival. How-
ever, recent evidence (Siddiq et al., 2005; Knowles et al., 2004;
Baranova et al., 2007) strongly suggests that PHDs and FIH
are important targets for medical intervention for a number of
conditions, including chronic anemia and stroke. PHD inhibitors
abrogate the ability of HIF1-mediated transactivation of BNIP3
and PUMA to potentiate oxidative death in normoxia (Aminova
et al., 2008). Although new targets for intervention in the HIF
pathway are constantly emerging, the latter observation justifies
the search for PHD inhibitors rather than for other types of HIF
activators. New substrates have been recently identified for
PHD1 (e.g., Rpb1, large subunit of RNA polymerase II [Mikhay-
lova et al., 2008]) responsible for the fundamental enzymatic
activity of the complex, synthesizing all cellular mRNAs) and
PHD3 (e.g., b2-adrenergetic receptor [Xie et al., 2009], whose
sustained downregulation is associated with heart failure and
asthma) placing HIF PHDs into the focus of drug development
efforts. Despite characterization of HIF PHDs as a potential
target for anti-ischemic therapy, few high throughput screening
(HTS) results for PHD inhibitors are publically available.
Inthiswork,wedeveloped a novel application foranapproach
elegantly validated by Kaelin and Livingston’s group for visuali-
zation of HIF stabilization in transgenic mice (Safran et al.,
2006) and used it for the purposes of HTS. The reporter system
380 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 2

consists of the HIF-1a gene fragment encoding the oxygen
degradable domain (ODD) containing the key proline residue
followed by luciferase gene (luc). The regulation of luciferase
protein stability in this reporter system is the same as the phys-
iological activation of HIF: hydroxylation of oxygen-degradable
domain (ODD, which contains 530-653 amino acids [aa] of
HIF1-a) results in recognition of the ODD-luc fusion protein by
VHL followed by its ubiquitinylation and proteasomal degrada-
tion (Figure 1A), and as we present below, the approach proved
to be productive for HTS purposes.
We performed a cell-based HTS of 85,000 compounds for HIF
protein stabilizers to identify those working as specific HIF PHD
inhibitors. The most intriguing finding from the primary screen of
85,000 compounds was the group of branched 8-hydroxyquino-
line derivatives whose binding mode into the active site of PHD2
resembles that of the HIF peptide. The biological effects of the
newly identified hits—i.e., HIF1 protein stabilization, induction
of HIF1-regulated genes such as vascular endothelial growth
factor (VEGF), lactate dehydrogenase (LDHA), and phospho-
glycerate kinase 1 (PGK1), neuroprotection in homocysteic acid
(HCA) cellular model of oxidative stress—are in good agreement
with their activation effects in the reporter assay.
RESULTS
Development and Optimization of the ODD-luc
Reporter System
The reporter cell lines constitutively expressing ODD-luc (human
neuroblastoma, SH-SY5Y) were stable for more than 1 year
without significant change in their response to canonical PHD
inhibitors such as deferoxamine (DFO), dihydroxybenzoate
(DHB),dimethyloxalylglycine(DMOG),andciclopirox(Figure1B).
The dependence of the ODD-luc reporter signal on inhibitor
concentration has a sigmoid shape (Figure 1B and Figure S1
available online), which is characterized by maximum activation,
IC50anda‘‘concentration lag,’’whichlikely reflectsthepresence
Figure 1. Mechanism of Reporter Activation and Response to Canonical HIF PHD Inhibitors
(A) Schematic presentation of reporter performance showing key steps/potential sites of inhibition.
(B) Reporter response to canonical HIF PHD inhibitors: ciclopirox, DFO, DMOG, and DHB.
All values are presented as mean ± SEM. Calculation of activation parameters from the titration curve in shown in Figure S1.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 381

Page 3

of iron in the media, which can bind drug and delay or impede its
intracellular effects. The concentration of iron in the serum used
for neuroblastoma cultivation is 1.5 mM.
In order to verify the specificity of luciferase changes as an
assay for PHD activity, several control lines were developed:
the control line expressing ODD-luc with proline 564 and 567
mutated to Ala generates luciferase fusion that cannot be
degraded and experimentally identifies a ceiling level of ODD-
luc protein attainable in these cells. The background signal for
the wild-type HIF ODD-luc line (PYIP) corresponds to approxi-
mately 5%–6% of the ODD-luc levels in the control line AYIA
(double mutant P564A/P567A line) (Figure 2). Treatment with
10 mM ciclopirox results in a 10-fold increase of a background
signal forthe wild-typeODD-luc reporter(PYIP line),i.e.,reaches
almost 50% of the threshold value (Figure 2). These particular
conditions are ideal for HTS because they promote the selection
of both weaker and more potent inhibitors than ciclopirox.
The neuroblastoma cell line expresses all three PHD isoforms
(Figure S2). In the second control line, as an initial test of fidelity
of our constructs for reporting endogenous regulation, we
mutated Tyr565 to Ala, which has been previously shown to
decrease the affinity of HIF for PHD2 (Landazuri et al., 2006;
Bruick and McKnight, 2001; Jaakkola et al., 2001). As expected,
line PAIP shows a 3-4-fold higher steady-state level of ODD-luc
than the wild-type line (Figure 2). Of note, the third control, muta-
tion of Pro567 to Ala, which has been shown to influence recog-
nition oftheHIFODDbyPHD3(Landazurietal.,2006),haslesser
effect on ODD-luc levels (Figure 2). The fact that the reporter
system is sensitive to single-point mutations surrounding
Pro564 region in accord with previously published observations
(Landazuri et al., 2006; Bruick and McKnight, 2001; Jaakkola
et al., 2001) provides evidence that the rate-limiting step is
controlled by the PHD catalyzed reaction.
The HIF ODD-luciferase reporter system is controlled by the
rate of PHD-catalyzed reaction, and from an enzyme kinetics
point of view it is a ‘‘capture assay’’ monitored by the consump-
tion of a substrate, the heterologously expressed HIF ODD-lucif-
erase fusion protein. In the kinetic regimen, i.e., monitoring the
time course of luminescence changes (Figure 3A), the ODD-luc
reporter system permits quantitative characterization of pro-
moter capacity (Ko, rate of fusion protein generation), enzyme
activity, and inhibition constant determination.
The rate of fusion accumulation equals to the rate of its
production (Ko) minus the rate of rate-limiting step, controlled
by HIF PHDs, which obeys Michaelis-Menten kinetics, as
follows:
v=d½ODDluc?=dt=K0? k1½PHD?½ODDluc?=fKmð1+½I?=KiÞ
+½ODDluc?g
where Kmis the inhibition constant for a competitive inhibitor, k1
is the rate coefficient, and [PHD] and [ODD-luc] are the concen-
trations of the enzyme and substrate, respectively.
Thebackgroundluminescencesignalcalibratedwithrecombi-
nantluciferase allowsusto estimatethesteady-stateconcentra-
tion of the ODD-luc fusion protein. Under the conditions used
the steady-state value of 60 rlu (relative light units) corresponds
to 1 pg luciferase protein; dividing this number by the total cell
volume taken as a single cell volume (233 m3) multiplied by
30,000 cells/well density (number of cells in a 96-well dish), we
get the ODD-luc fusion protein steady-state concentration equal
to 2.3 nM, which is way below all reported Kmvalues for HIF1
(Tuckerman et al., 2004; Koivunen et al., 2006; Hewitson et al.,
2007; Dao et al., 2009). Therefore, we work under nonsaturating
conditions with respect to HIF substrate, i.e., optimal conditions
for selecting inhibitors competitive against HIF substrate. More-
over, as compared with the in vitro assay, which uses a 19 amino
acid peptide fragment surrounding the hydroxylated proline
(P564), our ODD-luciferase construct contains 123 amino acid
acids, and thus more closely emulates the behavior of native
HIF. We can consider the initial concentration of fusion much
lower than Kmand ignore it in the rate equation:
(1)
v=d½ODDluc?=dt=K0? k1½PHD?½ODDluc?=Kmð1+½I?=KiÞ:
(2)
Knowing the capacity of the promoter, we can determine the
inhibition constant, but not the inhibition type, from the initial
rates of signal accumulation at varied fixed concentrations of
potential inhibitor. The capacity of promoter K0can be deter-
mined under the conditions of total inhibition of PHD activity by
means of complete iron deprivation achieved in the presence
of high concentrations of ciclopirox, i.e., when the increase in
the ciclopirox concentration gives no further increase in the
rate of luciferase signal growth (Figure 3B). The intracellular
enzyme activity (k1[PHD]/Km) can be also determined by dividing
the rate of fusion protein accumulation by the steady-state
concentration of the fusion protein determined directly from
one and the same experiment in luciferase units, without recal-
culation for the cellular volume, and corresponds to 0.05 min-1.
The linear plot of 1/(K0– v) versus the inhibitor concentration
Figure 2. Effect of Mutations Adjacent to Pro564 on the Reporter
Response to 10 mM Ciclopirox upon 3 hr Incubation
PHD1 and PHD2 are supposed to be the major HIF PHD isoforms present in
neuroblastoma cell line as judged by PCR (Figure S2). All values are presented
as mean ± SEM.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
382 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 4

gives the value of the apparent inhibition constant as the inter-
cept on x axis (Figure 3C):
1=ðK0? vÞ=Kmð1+½I?=KiÞ=k1½PHD?½ODDluc?0:
(3)
The apparent inhibition constant determined for DHB and
DMOG is 0.52 mM and 1.3 mM, respectively, which is in agree-
ment with previous observations on their biological effects
exerted in the millimolar range (Philipp et al., 2006; Asikainen
et al., 2005; Lomb et al., 2009) and IC50reported for PHD2
in vitro assay (Cho et al., 2005).
The response of the ODD-luc reporter to canonical HIF PHD
inhibitors, and the increased stability of single-point mutant
reporters in accord with the predictions, provided confidence
that this system could be utilized for screening for novel small
molecule HIF PHD inhibitors; hence, we then set out to optimize
conditions for high throughput screening (see Experimental
Procedures).
A Primary Screen Identified 160 Validated Hits
The screen of an 85,000 compound library (including those from
Spectrum, ChemDiv, AMRI, ChemBridge. Prestwick, and Cerep)
resulted in 295 hits, among which 160 were confirmed in follow
up testing. Of note, no established proteasomal inhibitors were
identified in the screen. Hits were then classified into 10 struc-
tural clusters, 6 of which are shown in Figure 4A.
Hydrazides and Hydrazones
Among this chemical group (55 hits of 76 hydrazides and hydra-
zones tested) are well-established HIF activators (Almstead
etal.,2002)thatformatight2:16-cordinatedFecomplexinsolu-
tion but not within the enzyme. It is likely that global cellular iron
deprivation emerging from these compounds results in reporter
activation comparable or even higher than that for ciclopirox.
Interestingly, we identified a group of edaravone hydrazone
derivatives that can provide only two iron ligands, such as
(Z)-1,3-diphenyl-4-(thiazol-2-yl-hydrazono)-1H-pyrazol-5(4H)-
one (I) (Figure 4A) and (Z)-4-(benzo[d]thiazol-2-yl-hydrazono)-1,
3-diphenyl-1H-pyrazol-5(4H)-one, and unlike other members of
the class that bind iron outside the enzyme, these may coordi-
nate iron inside the enzyme. However, the level of reporter acti-
vationforthetwoironligandedaravonederivativesislower(86%
and 70%, respectively) than for hydrazones providing three iron
ligands (above 100%). Of note, edaravone has been used in
Japan to treat humans with stroke (Kikuchi et al., 2009a,
2009b). However, edaravone itself does not provide iron ligands
and shows no reporter activation.
Hydralazine, an FDA-approved antihypertensive agent, is
known to be a HIF stabilizer (Knowles et al., 2004; Michels
et al., 2009) and an activator of a hypoxia response element
(HRE)-luc reporter (Ratan et al., 2005). Our data also implicate
a hydralazine analog, (2-hydrazinyl-4,6-diphenylpyrimidine) (II)
(Figure 4A) as a potent HIF stabilizer (64% or 6.7-fold activation).
The presence of a terminal amino group and free rotation of
phenyl rings (ortho-hydroxy group in one ring abolishes activa-
tion effect) are required for the activation effect observed. The
compound exhibits no iron chelation properties in solution.
Dibenzoylmethanes
We identified two compounds which exhibit a similar docking
mode to DBM, another established HIF activator (Mabjeesh
Figure 3. Determination of Apparent HIF PHD Inhibition Constants
from Time Course of Reporter Activation
(A and B)Original kinetic curves for DHB (A; all values are presentedas mean ±
SEM) and ciclopirox (B).
(C) Linear plot to calculate the apparent inhibition constants using Equation 3.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 383

Page 5

et al., 2003). The sulfur-containing analog of DBM (III) (Figure 4A)
resulted in a 5-fold (or 41% of ciclopirox) reporter activation.
Thiadiazoles
Thiadiazole compounds (four hits out of eight compounds
tested) provide modest activation (3-5-fold activation or 25%–
40% of the internal standard, Figure 4A, IV) and have a number
of potential iron ligands that may support two modes of the com-
pounddockingintothePHD2activesiteinplaceofaKG.Ineither
case, the predicted interference from‘‘bulky’’attachments atthe
amide end is in agreement with experimental observations:
2-amino-N-(5-(4-bromophenyl)-[1,3,4]thiadiazol-2-yl)-3-methyl-
pentanamide and 2-amino-N-(5-(4-methoxy-phenyl)-[1,3,4]thia-
diazol-2-yl)-3-phenylpropanamide show no reporter activation
at all.
Isoquinolinotriazolylthiols
This class of compounds likely stabilized ODD-luc in our assay
via their iron chelation in solution. Iron chelation by isoquinolino-
triazolylthiols (V) (Figure 4A) depends on their ability to provide
two iron ligands and depends on the rotation restriction between
isoquinoline and triazole rings. Activation drops by 3-fold with
substitutions at position 4 of the thiazole ring and with bulky
attachments like meta-, and para-substituted phenyl rings or
benzyl substituent containing a methyl branch. Two of eight
hits (5-isoquinolin-1-yl-4-(2-trifluoromethyl-phenyl)-4H-[1,2,4]
triazole-3-thioland4-isopropyl-5-isoquinolin-1-yl-4H-[1,2,4]tria-
zole-3-thiol (V) (Figure 4A) show reporter activation close to that
Figure 4. New Hit Groups Identified in HTS
(A)Chemicalstructuresofhits:(I),Edaravon-type hit:(Z)-1,
3-diphenyl-4-(thiazol-2-yl-hydrazono)-1H-pyrazol-5(4H)-
one (86%); (II), Hydralazine type hit: 2-hydrazinyl-4,6-
diphenylpyrimidine (64%); (III), Dibenzoylmethane group
hit: 1-phenyl-3-(1,3-thiazol-2-yl)thiourea (41%); (IV), Thia-
diazole group hit: N-[5-(3-bromophenyl)-[1,3,4]thiadiazol-
2-yl]pyrrolidine-2-carboxamide (35%); (V), Triazole group
hit: 4-isopropyl-5-isoquinolin-1-yl-4H-[1,2,4]triazole-3-
thiol (90%); (VI), Catechol group hit: (E)-5-(3,4-dihydroxy-
benzylidene)-3-phenyl-2-thioxothiazolidin-4-one
docking of branched catechol hits is shown in Figure S3.
(B) Schematic presentation of docking mode of hydroxyl-
ated HIF peptide into PHD2.
(C) Docking of best hits, compounds 7 and 8, into PHD2.
(38%);
for our internal standard, ciclopirox. The activa-
tion effects were consistent with the predicted
substitution effects on iron chelation ability,
therefore this group is likely to act through iron
chelation rather than iron coordination in the
active center.
Flavonoids are HIF activators (Wilson and
Poellinger, 2002; Jeon et al., 2007), but their
mechanism of action remained obscure and
no structure-activity relationships (SAR) have
been described for this class of compounds.
The 85,000 compound library had 80 flavones
(20 of which were hits), 90 isoflavones (7 of
which were hits), and 16 flavanones (6 of which
were hits). The SAR studies for flavone/isofla-
vone/flavanone family have been completed
and structural requirements for optimal docking
have been formulated. In particular, the absence of substitutions
in the phenyl ring for 3-hydroxyflavone derivatives is a must for
optimal docking and reporter activation, while the presence of
hydroxy-ormethoxy-groupsinphenylringof5-hydroxyflavones
(as well as isolfavones and flavanones) is also a requirement for
optimal docking and reporter activation.
Chalcones, precursors for flavones, were also hits in our
screen (the best one is 20b-dihydroxychalcone, showing 7.5-
fold activation or 71%). All provide at least two iron ligands but
are unlikely to be useful as biological tools or drugs because
they show rather high toxicity compared with flavones. Of the
active coumarines (2 active out of 36 tested), both contain
astrong ironbinding motif, vicinalhydroxyls, and arewell-known
for forming tight iron complexes. Their size allows binding inside
thePHDactivesite,whereascoumarineswithbulkyattachments
show no reporter activation.
Catechols
The catechol (3,4-dihydroxyphenyl) moiety was found in 100
compounds tested (excluding flavones and coumarins), how-
ever,only 8amongthemwerehits.Ethyl-3,4-dihydroxybenzoate
is a known PHD inhibitor, which activated the reporter at con-
centrations 5-fold higher (>50 mM, see Figure 1B) than the stan-
dard screening concentration 10 mM. L-Dopa, but not D-dopa,
carbidopa, dopamine, and other analogs tested, was a modest
hit (21% or 3-fold reporter activation, IC50= 15 mM). This finding
is of potential interest given the recent data on increased
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
384 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 6

protective effects for Sinemet (combination of L-dopa and
carbidopa) in Parkinson’s disease. Thioxothiazolidinones (VI)
(Figure 4A) exhibited rather modest reporter activation (35%–
45% or 4-5-fold). Most intriguing was the identification of four
hits containing a branching motif that in in vitro models, screens
the entrance to the active site (two of them are shown in
Figure S3).
Oxyquinolines (22 hits out of 66 compounds tested) could be
classified into four groups, and among those one could be
furtherdividedintothreesubgroups(VIIA-D,Figure4A).Oxyqui-
noline derivatives are established inhibitors of HIF prolyl hydrox-
ylase. They appear to act by providing two ligands for iron
binding and thus inhibit PHD2 in vitro with an IC50above 3 mM
(Warshakoon et al., 2006d). Compounds from group A, such
as chloroacetoquinoline, exhibit reporter activation comparable
to 8-hydroxyquinoline itself (4-5-fold reporter activation or
35%–45%), and its halogenated derivatives such as iodoquinol,
broxyquinoline, clioquinol (Choi et al., 2006), and chloroxine.
Clioquinol is of particular interest given its established salutary
effects in models of Alzheimer’s disease and Huntington’s dis-
ease and its serious consideration for late-stage human trials
(Kaur et al., 2003; Adlard et al., 2008). For the oxyquinolines,
the reporter signal drops with the increased size of the R group
(see Table S1 available online). By contrast, we found that
oxyquinoline derivatives containing a branched substitution at
position 7 (VII D, Figure 4A), as well as their conformationally
constraint analogs (group C) with reasonably short but flexible
linkers, are the best compounds that result in reporter activation
comparable or superior to ciclopirox (see Table S1). For optimal
HIF activation, either halogen or NO2group at position 4 (R3) is
favorable, whereas any substitution at position 2 (group B) is
not tolerated. The binding mode of hydroxylated HIF peptide
(Figure 4B) gives a strikingly similar position of Tyr565 against
aKG-binding plane to that of hydroxy-phenyl ring against the
oxyquinoline plane in our best hit, compound 8 (Figure 4C),
and this pushed us to study the effects of branched oxyquino-
lines in detail.
Structure-Activity Studies for Selected
Branched Oxyquinolines
To explore a variety of our hits further, we had to develop a ratio-
nale to discriminate between specific inhibition of PHD and
nonspecific iron chelation. We assume that if there is no specific
interaction between the enzyme and iron-binding inhibitors, the
enzyme inhibition constants (or in our case, IC50) will change in
parallel with the iron-binding constants. Specific inhibitors, i.e.,
those coordinating iron directly at the PHD active site, should
deviate (‘‘pop-up’’) from the group of iron chelators with the
same affinity and exhibit better inhibition constants (IC50) than
those that simply bind iron. We determined the apparent iron
binding constants in solution for a dozen of hits of interest
(Table 1). In addition, we determined the rate constant for asso-
ciation from the kinetics of calcein displacement from its
complexwithiron(ka),whichcharacterizeshowfastoxyquinoline
binds iron (Table 1). The apparent iron binding constant KFefor
the compounds studied varies more than one order of magni-
tude, from 0.08 to 2.0 mM, in parallel with the changes in the
association rate constant (from 20 to 250 M-1s-1), while the
dissociation rate constant is barely affected.
Wecandividethestudiedoxyquinolinesintothreegroupswith
respect to their iron binding ability: first, those close to that of
ciclopirox (compounds 1,2,4,7,10,13), second, similar to oxyqui-
noline (6,8,11,12), and third, very poor iron chelators (3,5,9)
showing very poor reporter activation. The best inhibitors are
found in the first two groups. Table 1 clearly points to five com-
pounds as reporter activators better than ciclopirox: compound
8 belongs to D2 group, while all others (1,4,6,7) belong to D3
group. The comparison of iron binding and reporter activation
parameters (Table 1) shows no direct correlation between chela-
tion ability of oxyquinolines and IC50for PHDs; the obvious
requirements for good activation are the absence of 2-methyl
group in oxyquinoline (R4) and dioxol group in the phenyl ring
R1. The absence of a linker, i.e., immediately attached branched
motif to the position 7 of oxyquinoline (10, Table 1), is good for
iron chelation, but not for reporter activation. We conclude that
structural determinants, but not iron binding constants, play
a major role in reporter activation.
Biological Effects of Best Branched Oxyquinoline Hits
To verify that reporter activation corresponds to the biological
effects exerted by branched oxyquinolines, we selected com-
pounds 7 and 8 as positive controls, and oxyquinoline and
compound 10 as negative controls. The reduced efficacy of
compound 10 in the reporter assay is not related to poorer cell
membrane permeability: indeed, compound 10 activates the
reporter much faster than compound 7 (see kinetics of reporter
activation for both compounds in Figure S4). As expected, the
best hits (7 and 8), but not our negative controls (5 mM), sig-
nificantly stabilized HIF1a protein (Figure 5A) in the SH-SY5Y
human neuroblastoma cell line 3 hr after addition of the com-
pounds to the bathing media. Accordingly, HIF1-regulated
genes such as Epo, VEGF, LDHA, and PGK1 (Figure 5B) were
also induced by the small molecules that stabilized HIF-1a,
although it is interesting to note that compound 7 and 8 acti-
vated distinct patterns of HIF-dependent gene expression,
suggesting that these compounds may affect distinct HIF PHD
isoforms.
Our prior studies on HIF PHD inhibition in glutathione deple-
tion model of oxidative stress in cortical neurons revealed that
inhibition of HIF PHD1 by gene silencing or with canonical PHD
inhibitors such as DFO (iron chelator), DHB, and DMOG (aKG
mimics) was sufficient to prevent cell death independent of
HIF (Siddiq et al., 2009). The hits identified in HTS with HIF1
ODD-luc reporter are unlikely to be PHD-isoform specific, there-
fore, one may expect that they will target PHD1 as well and exert
neuroprotection in the above model. As expected, compounds
7 and 8, but not 10, protect cortical neurons stimulated to die
by depletion of the versatile antioxidant glutathione (Figure 6).
Oxyquinolineandcompound10arenotonlylesspotentHIFacti-
vators, but also 20-40 times less potent as neuroprotectants
(Figure 6). Of note, compounds 7 and 8 possess an IC50for neu-
roprotection an order of magnitude lower (0.25 mM) than those
required for reporter and HIF stabilization (IC502.0-2.5 mM).
Thus, neuroprotection is exhibited in the nanomolar range only
bythe best hits,and, asone would predict fromthe reporter acti-
vation (Table 1), compound 7 is more potent than compound 8.
The inhibition of other enzymes of this class with the resolved
crystal structure is ruled out due to the structural constraints
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 385

Page 7

Table 1. Comparison of Reporter Activation Parameters and Iron-Binding Properties of Branched Oxyquinolines (Structural
Subgroups D1-D3 as Depicted in Figure 4)
Compound Group
and NumberCompound Structure
Maximum
Activation (-fold)IC50, mM ‘‘Lag,’’ mMKFe,mMka, M-1s-1
Ciclopirox8.5 ± 0.5 4.5 ± 0.52 ± 0.50.08 ± 0.02300 ± 50
Oxyquinoline4.5 ± 0.310 ± 14 ± 0.50.20 ± 0.05110 ± 20
D1, #10
5.0 ± 0.410 ± 11.8 ± 0.20.08 ± 0.02250 ± 50
D2, #9
5.5 ± 0.510 ± 15 ± 0.2 0.78 ± 0.1830 ± 5
D2, #12
6.0 ± 0.5 4.5 ± 0.53 ± 0.40.20 ± 0.05 n.d.a
D2, #5 5.2 ± 0.3 7 ± 0.25 ± 0.51.50 ± 0.3020 ± 3
D2, #8
7.0 ± 0.22 ± 0.3 0.6 ± 0.10.19 ± 0.04110 ± 20
D2, #3
2.0 ± 0.312 ± 0.510 ± 0.50.80 ± 0.20 20 ± 3
D2, #13
6.0 ± 0.27 ± 0.2 5 ± 0.50.10 ± 0.02 n.d.
D2, #2
4.5 ± 0.512 ± 17 ± 0.5 0.10 ± 0.02n.d.
D3, #1
7.0 ± 0.22.6 ± 0.2 1 ± 0.1 0.11 ± 0.02 240 ± 40
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
386 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 8

(Figure S5). These results support our conclusion that structural
effects play a major role in reporter activation (and HIF stabiliza-
tion) rather than common iron binding potency of the studied
oxyquinolines. These studies are the first to use chemical tools
to validate a role for HIF PHD inhibition and not iron chelation
per se in stabilizing HIF and protecting neurons from oxidative
death.
DISCUSSION
Two primary modes of screening for HIF activators have been
well described: a recombinant enzyme-based screen for PHD2
inhibitors (used by Fibrogen [Ivan et al., 2002], Amgen [Tegley
et al., 2008], Procter and Gamble Pharmaceuticals Inc. [Warsha-
koon et al., 2006a, 2006b, 2006c, 2006d], and other teams [Nan-
gaku et al., 2007]); and a cell-based screen using HRE-luciferase
reporter construct used by a number of labs including our own
(Ratan et al., Priority: 21.10.2005).
High throughputscreening for PHD inhibitors using anenzyme
assay is a challenge both in terms of the enzyme source and
the assay format. The enzymatic activity and stability of purified
PHD is very low, and enzyme assays suitable for high content
screening require large quantities of recombinant enzyme sup-
plementedwithiron.Oneofchallengesinthesearchforselective
HIF PHD inhibitors or other regulators of HIF stability is to
discriminate between nonspecific iron chelation in solution
and specific iron chelation inside the active center of the PHD
enzyme. The apparent iron binding constants for the 7-branched
oxyquinolinesidentifiedherein(Table1)aremuchlowerthanIC50
reported for 7-linear 8-hydroxyquinoline derivatives (3-10 mM) in
the PHD2 in vitro assay (Warshakoon et al., 2006d), which may
reflect the use of excess iron in the in vitro assay mixture as
Table 1. Continued
Compound Group
and NumberCompound Structure
Maximum
Activation (-fold)IC50, mM‘‘Lag,’’ mMKFe,mMka, M-1s-1
D3, #4 6.8 ± 0.22.1 ± 0.20.6 ± 0.2 0.10 ± 0.02210 ± 40
D3, #7
7.0 ± 0.22.2 ± 0.2 1 ± 0.10.10 ± 0.02 180 ± 35
D3, #6
7.2 ± 0.34 ± 0.1 1.8 ± 0.20.20 ± 0.0495 ± 15
D3, #11
4.0 ± 0.312 ± 18 ± 10.15 ± 0.03100 ± 20
an.d. = not determined.
Figure 5. Upregulation of HIF1a and HIF-
Regulated Human Genes
Upregulation of HIF1a (A) and HIF-regulated
human genes (e.g. EPO, VEGF, PGK1, LDHA) (B)
upon 3 hr treatment of neuroblastoma cells with
5 mM inhibitor (7 and 8 as positive hits, oxyquino-
line and 10 as negative hits, control with no inhib-
itor added). All values are presented as mean ±
SEM. All compounds used have comparable cell
permeability as judged by kinetics of reporter acti-
vation shown in Figure S4.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 387

Page 9

comparedwithourcell-based assay.Giventhenonphysiological
conditions under which screening for inhibitors occurs with
recombinant PHD2, it is not surprising that the IC50value deter-
mined in the enzyme in vitro assay did not correlate with the IC50
for VEGF activation reported in the same study (Warshakoon
et al., 2006d). Another limitation is the use of a 19-mer HIF
peptide, whose affinity for the HIF PHDs is orders of magnitude
lesser than the full-length protein. So far only Amgen team used
recombinant HIF protein in HTS of their internal collection,
although again they had been varying the concentration of
aKG, not HIF, when determining the inhibition constant for their
best hit (Tegley et al., 2008). Current studies are underway to
establish the affinity of our ODD-luciferase construct (containing
128 amino acids from HIF-1a) for the HIF PHDs. A negative
consequence of the test tube strategy is the assay format is
more likely to yield inhibitors competitive with respect to aKG
than those competing with HIF itself. The recently reported
crystal structure of PHD2 with a 17-mer HIF peptide (Chowdhury
et al., 2009) shows no active-site water displacement, which
appears to be a mandatory requirement for the initiation of
the catalytic cycle (see p.277 in Solomon et al., 2000, and Price
et al., 2005). Given these biases, it is not surprising that all PHD
inhibitors developed using the recombinant enzyme explored
only the aKG-binding motif inside PHD2 active site and had
a carboxyl group interacting with Arg-383 in addition to a clearly
defined iron-binding motif (Tegley et al., 2008; Warshakoon
et al., 2006a, 2006b, 2006c, 2006d; Ivan et al., 2002).
The cell-based assay with HRE-luc reporter system, a pro-
moter-reporter construct that contained 68 bp of a known
hypoxia and HIF-1 regulated gene, enolase, containing a wild-
typeHRE(50-RCTGT-30),isa widelyusedapproachforscreening
of HIF activators with diverse mechanisms of action (Semenza
et al., 1996). A reporter system is based on transfected immor-
talized hippocampal neuroblast cell line (HT22) and allows
screening for a broad spectrum of compounds that include acti-
Figure 6. Neuroprotection Effects of Best
BranchedOxyquinoline
Comparison with a Poor Hit from the Same
Group (10) in Oxidative Stress (HCA) Model
(A) Concentration titration for (7,8 and 10).
(B) Photographs of viability test for 0.6 mM for (8)
and (10). All values are presented as mean ±
SEM. See Figure S5 for discussion of specificity
of compounds effects.
Hits(7,8) in
vators of HIF transcription, activators of
HIF binding to HRE, and effectors of HIF
proteinstability(PHDinhibitors,pVHL,and
proteasomeinhibitors).Themanualscreen
ofSpectrumlibraryperformedinthislabo-
ratory using HRE-luc/HT22 line took half
a year and resulted in 43 hits. However,
in our hands, the cell line’s response to
positive controls decreases after seven
passages, makingthe systemnot suitable
for a robotic HTS on 384-well plates.
Taking into account the low specific
activity of recombinant enzymes, and
the inadequacy of interpretation of the inhibition constant
generated using different types of enzyme in vitro assays, we
developed a cell-based reporter system for HTS of ODD-luc
stability, a variant of the cell-based ‘‘capture’’ assay, and
accomplished a screen of 85,000 structurally diverse com-
pounds in less than a month. We identified a novel, previously
unknown structural motif in the group of catechol-type and oxy-
quinoline hits that suggests specific recognition by PHD among
other aKG-dependent Fe-dioxygenases. Comparison of crystal
structures of PHD2 (Figure S5A) and FIH (Figure S5) shows the
difference in access to the active sites: PHD2 allows sliding of
the aKG mimic into the active site, leaving the branched portion
outside, whereas FIH does not. Analysis of docking modes for
best hits from our HTS into the available crystal structures of
aKG-dependent Fe-dioxygenases, e.g., FIH (Figure S5C), HIF
PHD2 (Figure 4C), and jumonji histone demethylase (Fig-
ure S5D), demonstrates that newly identified branched motifs
provide specificity for HIF PHD by exploring the active site
entrance which significantly differs from those of the other
enzymes of this class. In addition, these branched inhibitors
did not fit into the active center of lipoxygenase-12 (Figures
S5E-S5G), the enzyme directly implicated into the survival
mechanism in glutathione-depletion HCA model (Ratan et al.,
2002). Although future studies will examine these modeling
predictions experimentally, the presence of the newly identified
branched motif appears to increase the likelihood of the inhib-
itor specificity for HIF PHD. For the future, it is the branched
motif that holds the promise to discriminate between different
PHD isoforms: construction of novel mutant reporter lines that
are currently in progress in this laboratory will not only allow
us to answer questions regarding the specificity of PHD iso-
forms for regulating HIF under conditions close to physiolog-
ical, but also can provide a mechanism for intelligent design
of novel inhibitors that may discriminate between different
PHD forms.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
388 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 10

SIGNIFICANCE
Brain ischemia underlies many nervous system disorders
triggeringacascadeofeventsthatinduceacuteanddelayed
changes resulting in disability and cognitive decline. Over
the past decade, cell adaptation to hypoxia has emerged as
an active process. Although the panoply of mechanisms
involved in hypoxic preconditioning are incompletely under-
stood, the discovery of HIF opened new horizons for the
treatment of ischemia. Recent evidence strongly suggests
that HIF PHDs and FIH are important targets for medical
intervention: small molecules that inhibit HIF PHDs are the
focus ofdrug development efforts directed toward the treat-
ment of ischemia in many organs, including the muscle,
heart,andbrain.Thevalidationofthereporterconstructper-
formed in this work shows the cell-based system capacity
both for substrate specificity studies and for quantitative
inhibitionstudies:(1)themutationalanalysispresentsanovel
approach to study sequence-specificity of PHDs, and (2)
quantitativetreatmentofreporteractivationkineticspermits
direct calculation of an apparent inhibition constant instead
of IC50. The results presented here are the first comprehen-
sive report on HTS for HIF activators/HIF PHD inhibitors
under the conditions most closely resembling physiological
ones. We found a novel, previously undescribed branching
motif adjacent to iron-binding ligands in the group of cate-
chol-type and oxyquinoline hits that warrants specific
recognition by PHD among other aKG dependent Fe-dioxy-
genases due to the structural constraints. The biological
effects of newly identified branched hits were in accord
with their rating in HTS. Altogether these findings validate
a novel HTS for small molecules that can modulate hypoxic
adaptation.
EXPERIMENTAL PROCEDURES
Materials
Cell Lines
Human neuroblastoma SH-SY5Y cells were transfected with 1 mg pcDNA3-
ODDLUC8 or mutant variants of this plasmid by using Lipofectamine? 2000
(Invitrogen). Transfected cell were grown in the presence of 500 mg/ml Genet-
icin (GIBCO-Invitrogen) on DMEM/F12+GlitaMAX (Dulbecco’s modified Eagle
medium Nutrient Mixture F-12 (Ham)(1:1) 1X, GIBCO 10565) medium.
Reporter Plasmid Construction and Mutagenesis
The ODDLUC encoded plasmid pcDNA3-ODDLUC8 was constructed as
describedpreviously(Safranetal.,2006).Thisplasmidwasusedasatemplate
to introduce amino acid substitutions into the PYIP (564-567 aa) region of HIF-
ODD fragment that was suggested to determine enzyme-specific interaction
of HIF-1a to three isoforms of HIF PHDs (Safran et al., 2006). The plasmids
pPAIP, pPYIA, and pAYIA were obtained from pcDNA3-ODDLUC8 using
QuikChange Multi Site-Directed Mutagenesis kit (Agilent Technology) to intro-
duce the corresponding mutations: Tyr565Ala, Pro567Ala, and Pro564Ala/
Pro567Ala. pAYIA was used as a control line.
Methods
HTS Optimization and SAR Analysis
The assay was optimized for HTS format to provide Z values above 0.7.
SH-SY5Y-ODD-luc cells were plated into 384 well, white, flat-bottom plates
at 7000 cell/well in 30 ml serum and incubated overnight at 37?C, 5% CO2.
The next day compounds were added to a final concentration of 10 mM, plates
were incubated for 3 hr at 37?C, and luciferase activity was measured using
SteadyGlo? reagent (Promega). Each plate had two internal standards, ciclo-
pirox (100%) and DMSO (0%). The reporter activation (%) was calculated as
a ratio (L–LDMSO)/(Lciclopirox-LDMSO). Hits were defined as those greater than
25%. HTS of 85,000 compounds was performed at Rockefeller HTS Resource
Center. A total of 295 hits from the initial screen have been tested in quadru-
plicate, and 160 were confirmed. Classification into ten structural clusters
has been done manually; 25 hits were singletons.
Extended SAR Analysis
Oxyquinolines were purchased from ChemDiv (San Diego, CA) and tested in
96-format plates with varied concentrations of an inhibitor (0.05-15 mM). Cells
wereplatedatthedensityof30,000cellperwellusingaWellMatemultichannel
dispenser from Matrix (Thermo Fisher Scientific) and grown overnight on
DMEM/F12+GlitaMAX (100 ml per well). Then the inhibitor was added, and
the plates were incubated for a fixed time interval; the medium was removed,
cells were lysed, and luciferase activity was measured on a luminometer pla-
tereader Lmax11384(Molecular Devices) with BrightGlo? reagent (Promega).
The reporter activation was normalized to the background luminescence.
Kinetics of reporter activation were measured by adding varied fixed
concentrations of an inhibitor at different time points followed by simultaneous
celllysisandactivitymeasurementinthewhole96-wellplate;thisassayformat
minimizesexperimentalerrororiginatedfromthewell-knowninstabilityoflucif-
erase reagent.
HIF Immunoblot
After 3 hr of 5 mM drug treatment, cells were scraped in iced cold PBS and
centrifugedat1,000xgper5min.Thepelletwasusedfornuclearextractprep-
aration with the NE-PER Nuclear and Cytoplasmic Extraction kit (Pierce). After
SDS-PAGE followed by transfer to a nitrocellulose membrane, the latter was
incubated overnight at 4?C with primary polyclonal antibody against HIF-1a
(Upstate) and monoclonal antibody against b-actin (Sigma) (dilution 1:250
and 1:5000, respectively, in Odyssey Blocking Buffer). Secondary fluorophore
conjugatedOdysseyIRDye-680andIRD-800antibodies(LI-CORBiosciences)
were added at 1:20,000 in Odyssey Blocking Buffer, and incubated for 1h at
RT. Immunoreactive proteins were detected using Odyssey IR-imaging
system (LI-COR Biosciences).
Real-Time Polymerase Chain Reaction
Total RNA was isolated from SH-SY5Y cells by using NucleoSpin RNAII kit
(Macherey-Nagel) and was used for cDNA synthesis by SuperScript III First-
Strand Synthesis System for RT-PCR (Invitrogen). Quantitative real-time
PCR analyses of human PHD1,2,3, LDHA, PKG1 and EPO were performed
by using the corresponding primers and probe set from Applied Biosystems
on the ABI 7500 Fast Real Time PCR TaqMan system (Applied Biosystems).
GAPDH was used for normalization.
Cell Death and Viability Assays
Primary neuronal cultures were prepared from the forebrains of Sprague-
Dawley rat embryos (E17) and plated on 96-well plates at a 106cells/ml
density. After 16 hr, cells were rinsed with warm phosphate-buffered saline
and then placed in minimum essential medium (Life Technologies) containing
5 mM HCA in the presence of oxyquinolines (0.25-2 mM). Cells were incubated
for 24 hr or longer to see 90% cell death in HCA-treated controls. Viability
was assessed by the MTT (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium
bromide) assay (Mosmann, 1983).
Iron-Binding Properties of Oxyquinolines
The iron chelation ability is determined by displacement of calcein from its
complex with iron monitored by fluorescence (excitation 490 nm, emission
523 nm, cutoff 515 nm) on a Spectramax M5eplatereader (Molecular Devices).
The apparent binding constant for calcein (ca. 50 nM) was determined from Fe
titration curve for 1 mM calcein in 5 mM Tris-HCl buffer (pH 7.5). The ratio
between the iron binding constant for calcein and a particular compound
KQ/KCawas estimated by fitting the titration curve into the dependence of
[Fe]oversus Y, where Y = [Ca-Fe]/[Ca] is a ratio of calcein-bound Fe to free
(fluorescent) calcein:
½Fe?o=KCaY +½Ca?oY=ðY +1Þ+½Q?oY=ðY +KQ=KCaÞ:
(4)
The association rate constant was determined as the second order rate
constant for calcein displacement from its complex with iron (1 mM:1 mM)
uponadditionofanoxyquinoline(5-20mM)calculatedfromtheslopeofalinear
plot of the initial rate of calcein release versus the concentration of oxyquino-
lineadded.Allexperimentswereperformedintriplicatesormore.Allvaluesare
presented as mean and standard error of the mean (± SEM).
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 389

Page 11

Computer Modeling
Docking experiments were performed using the CDOCKER algorithm
(Wu et al., 2003) as implemented in the Discovery Studio 2.5 software suite
(Accelrys, San Diego, CA), followed by force-field minimization and binding
energy calculations using the PHD2 crystal structure with the bound inhibitor
(2G19.pdb) as the starting template structure. Preparation of the receptor
was done by running a protein check and identifying all the elements of the
structure. It was noted that there were amino acids missing on the N terminus
and C terminus, however these were not in close proximity to the binding site
and thereforethere was no need to add them to the structure. Force-field mini-
mization was carried out using the molecular mechanics algorithm CHARMm
(Brooks et al., 1983) as implemented in Discovery Studio 2.5.
SUPPLEMENTAL INFORMATION
Supplemental Informationincludesfivefigures and onetable andcan befound
with this article online at doi:10.1016/j.chembiol.2010.03.008.
ACKNOWLEDGMENTS
We thank Ronald Realubit, BS, for his help with HTS, and Dmitry Hushpulian,
PhD, for introducing into various docking programs and help with alignment
and docking. This work was funded by the Winifred Masterson Burke Relief
Foundation, the Adelson Foundation for Neurorehabilitation and Repair, and
NYS DOH Center of Research Excellence # CO19772. I.N.G was supported
by NIH grant #1R43CA133985-01.
Received: December 18, 2009
Revised: February 26, 2010
Accepted: March 9, 2010
Published: April 22, 2010
REFERENCES
Adlard, P.A., Cherny, R.A., Finkelstein, D.I., Gautier, E., Robb, E., Cortes, M.,
Volitakis, I., Liu, X., Smith, J.P., Perez, K., et al. (2008). Rapid restoration of
cognition in Alzheimer’s transgenic mice with 8-hydroxy quinoline analogs is
associated with decreased interstitial Abeta. Neuron 59, 43–55.
Almstead, J.-I.K., Izzo, N.J., Jones, D.R., and Kawamoto, R.M. (2002) Medic-
inal uses of hydrazones issued. US Patent 6660737.
Aminova, L., Siddiq, A., and Ratan, R.R. (2008). Antioxidants, HIF prolyl
hydroxylase inhibitors or short interfering RNAs to BNIP3 or PUMA, can
preventprodeath effect ofthe transcriptional activator, HIF-1alpha,in amouse
hippocampal neuronal line. Antioxid. Redox Signal 10, 1989–1998.
Asikainen,T.M.,Ahmad,A.,Schneider,B.K.,Ho,W.B.,Arend,M.,Brenner,M.,
Gunzler, V., and White, C.W. (2005). Stimulation of HIF-1alpha, HIF-2alpha,
and VEGF by prolyl 4-hydroxylase inhibition in human lung endothelial and
epithelial cells. Free Radic. Biol. Med. 38, 1002–1013.
Baranova, O., Miranda, L.F., Pichiule, P., Dragatsis, I., Johnson, R.S., and
Chavez, J.C. (2007). Neuron-specific inactivation of the hypoxia inducible
factor 1 alpha increases brain injury in a mouse model of transient focal cere-
bral ischemia. J. Neurosci. 27, 6320–6332.
Brooks, B.R., Bruccoleri, R.E., Olafson, B.D., States, D.J., Swaminathan, S.,
and Karplus, M. (1983). CHARMM: A program for macromolecular energy
minimization and dynamics calculations. J. Comput. Chem. 4, 187–217.
Bruick,R.K.,andMcKnight,S.L.(2001).Aconservedfamilyofprolyl-4-hydrox-
ylases that modify HIF. Science 294, 1337–1340.
Chan, D.A., Sutphin, P.D., Yen, S.E., and Giaccia, A.J. (2005). Coordinate
regulation of the oxygen-dependent degradation domains of hypoxia-induc-
ible factor 1 alpha. Mol. Cell. Biol. 25, 6415–6426.
Cho, H., Park, H., and Yang, E.G. (2005). A fluorescence polarization-based
interaction assay for hypoxia-inducible factor prolyl hydroxylases. Biochem.
Biophys. Res. Commun. 337, 275–280.
Choi, S.M., Choi, K.O., Park, Y.K., Cho, H., Yang, E.G., and Park, H. (2006).
Clioquinol, a Cu(II)/Zn(II) chelator, inhibits both ubiquitination and asparagine
hydroxylation of hypoxia-inducible factor-1alpha, leading to expression of
vascular endothelial growth factor and erythropoietin in normoxic cells. J.
Biol. Chem. 281, 34056–34063.
Chowdhury, R., McDonough, M.A., Mecinovic, J., Loenarz, C., Flashman, E.,
Hewitson, K.S., Domene, C., and Schofield, C.J. (2009). Structural basis for
binding of hypoxia-inducible factor tothe oxygen-sensing prolyl hydroxylases.
Structure 17, 981–989.
Dao, J.H., Kurzeja, R.J., Morachis, J.M., Veith, H., Lewis, J., Yu, V., Tegley,
C.M., and Tagari, P. (2009). Kinetic characterization and identification of
a novel inhibitor of hypoxia-inducible factor prolyl hydroxylase 2 using
a time-resolved fluorescence resonance energy transfer-based assay tech-
nology. Anal. Biochem. 384, 213–223.
Hewitson, K.S., Schofield, C.J., and Ratcliffe, P.J. (2007). Hypoxia-inducible
factor prolyl-hydroxylase: purification and assays of PHD2. Methods Enzymol.
435, 25–42.
Hirota, K., and Semenza, G.L. (2005). Regulation of hypoxia-inducible factor 1
by prolyl and asparaginyl hydroxylases. Biochem. Biophys. Res. Commun.
338, 610–616.
Ivan, M., Haberberger, T., Gervasi, D.C., et al. (2002). Biochemical purification
and pharmacological inhibition of a mammalian prolyl hydroxylase acting on
hypoxia-inducible factor. Proc. Natl. Acad. Sci. USA 99, 13459–13464.
Jaakkola, P., Mole, D.R., Tian, Y.M., et al. (2001). Targeting of HIF-alpha to the
von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxyl-
ation. Science 292, 468–472.
Jeon, H., Kim, H., Choi, D., Kim, D., Park, S.Y., Kim, Y.J., Kim, Y.M., and Jung,
Y.(2007). Quercetin activates anangiogenicpathway,hypoxiainducible factor
(HIF)-1-vascular endothelial growth factor, by inhibiting HIF-prolyl hydroxy-
lase: a structural analysis of quercetin for inhibiting HIF-prolyl hydroxylase.
Mol. Pharmacol. 71, 1676–1684.
Kaelin, W.G., Jr. (2005).The von Hippel-Lindau protein, HIF hydroxylation,and
oxygen sensing. Biochem. Biophys. Res. Commun. 338, 627–638.
Kaur, D., Yantiri, F., Rajagopalan, S., et al. (2003). Genetic or pharmacological
ironchelationpreventsMPTP-inducedneurotoxicityinvivo:anoveltherapyfor
Parkinson’s disease. Neuron 37, 899–909.
Kikuchi, K., Kawahara,K., Tancharoen, S., etal. (2009a). The free radicalscav-
enger edaravone rescues rats from cerebral infarction by attenuating the
release of high-mobility group box-1 in neuronal cells. J. Pharmacol. Exp.
Ther. 329, 865–874.
Kikuchi, K., Tancharoen, S., Matsuda, F., et al. (2009b). Edaravone attenuates
cerebral ischemic injury by suppressing aquaporin-4. Biochem. Biophys. Res.
Commun. 390, 1121–1125.
Knowles, H.J., Tian, Y.M., Mole, D.R., and Harris, A.L. (2004). Novel mecha-
nism of action for hydralazine: induction of hypoxia-inducible factor-1alpha,
vascular endothelial growth factor, and angiogenesis by inhibition of prolyl
hydroxylases. Circ. Res. 95, 162–169.
Koivunen, P., Hirsila, M., Kivirikko, K.I., and Myllyharju, J. (2006). The length of
peptide substrates has a marked effect on hydroxylation by the hypoxia-
inducible factor prolyl 4-hydroxylases. J. Biol. Chem. 281, 28712–28720.
Landazuri, M.O., Vara-Vega, A., Viton, M., Cuevas, Y., and del Peso, L. (2006).
Analysis of HIF-prolyl hydroxylases binding to substrates. Biochem. Biophys.
Res. Commun. 351, 313–320.
Lando, D., Peet, D.J., Whelan, D.A., Gorman, J.J., and Whitelaw, M.L. (2002).
Asparagine hydroxylation of the HIF transactivation domain a hypoxic switch.
Science 295, 858–861.
Liu, X.B., Wang, J.A., Ogle, M.E., and Wei, L. (2009). Prolyl hydroxylase inhib-
itor dimethyloxalylglycine enhances mesenchymal stem cell survival. J. Cell.
Biochem. 106, 903–911.
Lomb, D.J., Desouza, L.A., Franklin, J.L., and Freeman, R.S. (2009). Prolyl
hydroxylase inhibitors depend on extracellular glucose and hypoxia-inducible
factor (HIF)-2alpha to inhibit cell death caused by nerve growth factor (NGF)
deprivation: evidence that HIF-2alpha has a role in NGF-promoted survival
of sympathetic neurons. Mol. Pharmacol. 75, 1198–1209.
Mabjeesh, N.J., Willard, M.T., Harris, W.B., Sun, H.Y., Wang, R., Zhong, H.,
Umbreit, J.N., and Simons, J.W. (2003). Dibenzoylmethane, a natural dietary
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
390 Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved

Page 12

compound, induces HIF-1 alpha and increases expression of VEGF. Biochem.
Biophys. Res. Commun. 303, 279–286.
Michels, C., Dorai, T., Chander, P., Choudhury, M., and Grasso, M. (2009).
Hypoxic pre-conditioning in a rat renal ischemia model: an evaluation of the
use of hydralazine. World J. Urol.
Mikhaylova, O., Ignacak, M.L., Barankiewicz, T.J., Harbaugh, S.V., Yi, Y.,
Maxwell, P.H., Schneider, M., Van Geyte, K., Carmeliet, P., Revelo, M.P.,
et al. (2008). The von Hippel-Lindau tumor suppressor protein and Egl-9-
Type proline hydroxylases regulate the large subunit of RNA polymerase II in
response to oxidative stress. Mol. Cell. Biol. 28, 2701–2717.
Mosmann, T. (1983). Rapid colorimetric assay for cellular growth and survival:
application to proliferation and cytotoxicity assays. J. Immunol. Methods 65,
55–63.
Nangaku, M., Izuhara, Y., Takizawa, S., Yamashita, T., Fujii-Kuriyama, Y.,
Ohneda, O., Yamamoto, M., van Ypersele de Strihou, C., Hirayama, N., and
Miyata, T. (2007). A novel class of prolyl hydroxylase inhibitors induces angio-
genesis and exerts organ protection against ischemia. Arterioscler. Thromb.
Vasc. Biol. 27, 2548–2554.
Philipp, S., Cui, L., Ludolph, B., Kelm, M., Schulz, R., Cohen, M.V., and
Downey, J.M. (2006). Desferoxamine and ethyl-3,4-dihydroxybenzoate
protect myocardium by activating NOS and generating mitochondrial ROS.
Am. J. Physiol. Heart Circ. Physiol. 290, H450–H457.
Price, J.C., Barr, E.W., Hoffart, L.M., Krebs, C., and Bollinger, J.M., Jr. (2005).
Kinetic dissection of the catalytic mechanism of taurine:alpha-ketoglutarate
dioxygenase (TauD) from Escherichia coli. Biochemistry 44, 8138–8147.
Ratan, R.R., Siddiq, A., and Chavez, J. (2005) Compounds for enhancing
hypoxia inducible factor activity and methods of use. WO/2007/048004;
60/729,059.
Ratan, R.R., Ryu, H., Lee, J., Mwidau, A., and Neve, R.L. (2002). In vitro model
of oxidative stress in cortical neurons. Methods Enzymol. 352, 183–190.
Safran, M., Kim, W.Y., O’Connell, F., Flippin, L., Gunzler, V., Horner, J.W.,
Depinho, R.A., and Kaelin, W.G., Jr. (2006). Mouse model for noninvasive
imaging of HIF prolyl hydroxylase activity: assessment of an oral agent that
stimulateserythropoietinproduction.Proc.Natl.Acad.Sci.USA103,105–110.
Semenza, G.L., Jiang, B.H., Leung, S.W., Passantino, R., Concordet, J.P.,
Maire, P., and Giallongo, A. (1996). Hypoxia response elements in the aldolase
A, enolase 1, and lactate dehydrogenase A gene promoters contain essential
binding sites for hypoxia-inducible factor 1. J. Biol. Chem. 271, 32529–32537.
Siddiq, A., Ayoub, I.A., Chavez, J.C., Aminova, L., Shah, S., LaManna, J.C.,
Patton, S.M., Connor, J.R., Cherny, R.A., Volitakis, I., et al. (2005). Hypoxia-
inducible factor prolyl 4-hydroxylase inhibition. A target for neuroprotection
in the central nervous system. J. Biol. Chem. 280, 41732–41743.
Siddiq, A., Aminova, L.R., Troy, C.M., Suh, K., Messer, Z., Semenza, G.L., and
Ratan, R.R. (2009). Selective inhibition of hypoxia-inducible factor (HIF) prolyl-
hydroxylase 1 mediates neuroprotection against normoxic oxidative death via
HIF- and CREB-independent pathways. J. Neurosci. 29, 8828–8838.
Solomon, E.I., Brunold, T.C., Davis, M.I., Kemsley, J.N., Lee, S.K., Lehnert, N.,
Neese, F., Skulan, A.J., Yang, Y.S., and Zhou, J. (2000). Geometric and elec-
tronic structure/function correlations in non-heme iron enzymes. Chem. Rev.
100, 235–350.
Tegley, C.M., Viswanadhan, V.N., Biswas, K., Frohn, M.J., Peterkin, T.A.,
Chang, C., Bu ¨rli, R.W., Dao, J.H., Veith, H., Rogers, N., et al. (2008). Discovery
of novel hydroxy-thiazoles as HIF-alpha prolyl hydroxylase inhibitors: SAR,
synthesis, and modeling evaluation. Bioorg. Med. Chem. Lett. 18, 3925–3928.
Tuckerman, J.R., Zhao, Y., Hewitson, K.S., Tian, Y.M., Pugh, C.W., Ratcliffe,
P.J., and Mole, D.R. (2004). Determination and comparison of specific activity
of the HIF-prolyl hydroxylases. FEBS Lett. 576, 145–150.
Villar, D., Vara-Vega, A., Landazuri, M.O., and Del Peso, L. (2007). Identifica-
tion of a region on hypoxia-inducible-factor prolyl 4-hydroxylases that deter-
mines their specificity for the oxygen degradation domains. Biochem. J.
408, 231–240.
Wang, G.L., and Semenza, G.L. (1995). Purification and characterization of
hypoxia-inducible factor 1. J. Biol. Chem. 270, 1230–1237.
Wang, G.L., Jiang, B.H., Rue, E.A., and Semenza, G.L. (1995). Hypoxia-induc-
ible factor 1 is a basic-helix-loop-helix-PAS heterodimer regulated by cellular
O2 tension. Proc. Natl. Acad. Sci. USA 92, 5510–5514.
Warshakoon, N.C., Wu, S., Boyer, A., Kawamoto, R., Renock, S., Xu, K.,
Pokross, M., Evdokimov, A.G., Zhou, S., et al. (2006a). Design and synthesis
of a series of novel pyrazolopyridines as HIF-1alpha prolyl hydroxylase inhib-
itors. Bioorg. Med. Chem. Lett. 16, 5687–5690.
Warshakoon, N.C., Wu, S., Boyer, A., Kawamoto, R., Sheville, J., Bhatt, R.T.,
Renock, S., Xu, K., Pokross, M., Zhou, S., et al. (2006b). Design and synthesis
of substituted pyridine derivatives as HIF-1alpha prolyl hydroxylase inhibitors.
Bioorg. Med. Chem. Lett. 16, 5616–5620.
Warshakoon, N.C., Wu, S., Boyer, A., Kawamoto, R., Sheville, J., Renock, S.,
Xu, K., Pokross, M., Evdokimov, A.G., Walter, R., and Mekel, M. (2006c).
A novel series of imidazo[1,2-a]pyridine derivatives as HIF-1alpha prolyl
hydroxylase inhibitors. Bioorg. Med. Chem. Lett. 16, 5598–5601.
Warshakoon, N.C., Wu, S., Boyer, A., Kawamoto, R., Sheville, J., Renock, S.,
Xu, K., Pokross, M., Zhou, S., Winter, C., et al. (2006d). Structure-based
design, synthesis, and SAR evaluation of a new series of 8-hydroxyquinolines
as HIF-1alpha prolyl hydroxylase inhibitors. Bioorg. Med. Chem. Lett. 16,
5517–5522.
Wilson, W.J., and Poellinger, L. (2002). The dietary flavonoid quercetin modu-
lates HIF-1 alpha activity in endothelial cells. Biochem. Biophys. Res. Com-
mun. 293, 446–450.
Wu,G.,Robertson, D.H., Brooks,C.L.,3rd,andVieth, M.(2003).Detailed anal-
ysis of grid-based molecular docking: A case study of CDOCKER-A
CHARMm-based MD docking algorithm. J. Comput. Chem. 24, 1549–1562.
Xie, L., Xiao, K., Whalen, E.J., Forrester, M.T., Freeman, R.S., Fong, G., Gygi,
S.P., Lefkowitz, R.J., and Stamler, J.S. (2009). Oxygen-regulated2-adrenergic
receptor hydroxylation by EGLN3 and ubiquitylation by pVHL. Sci. Signal 2,
1–10.
Zhang, C.P., Zhu, L.L., Zhao, T., Zhao, H., Huang, X., Ma, X., Wang, H., and
Fan, M. (2006). Characteristics of neural stem cells expanded in lowered
oxygen and the potential role of hypoxia-inducible factor-1Alpha. Neurosig-
nals 15, 259–265.
Chemistry & Biology
Novel Branched Oxyquinolines as HIF Activators
Chemistry & Biology 17, 380–391, April 23, 2010 ª2010 Elsevier Ltd All rights reserved 391