The clinically used iron chelator deferasirox is an inhibitor of epigenetic jumonjic domain-containing histone demethylases

Article abstract of DOI:10.1021/acschembio.9b00289

Fe(II)- A nd 2-oxoglutarate (2OG)-dependent JumonjiC domain-containing histone demethylases (JmjC KDMs) are "epigenetic eraser" enzymes involved in the regulation of gene expression and are emerging drug targets in oncology. We screened a set of clinically used iron chelators and report that they potently inhibit JMJD2A (KDM4A) in vitro. Mode of action investigations revealed that one compound, deferasirox, is a bona fide active site-binding inhibitor as shown by kinetic and spectroscopic studies. Synthesis of derivatives with improved cell permeability resulted in significant upregulation of histone trimethylation and potent cancer cell growth inhibition. Deferasirox was also found to inhibit human 2OG-dependent hypoxia inducible factor prolyl hydroxylase activity. Therapeutic effects of clinically used deferasirox may thus involve transcriptional regulation through 2OG oxygenase inhibition. Deferasirox might provide a useful starting point for the development of novel anticancer drugs targeting 2OG oxygenases and a valuable tool compound for investigations of KDM function.

Full text of DOI:10.1021/acschembio.9b00289

Subscriber access provided by UNIVERSITY OF TOLEDO LIBRARIES
Article
The Clinically Used Iron Chelator Deferasirox is an Inhibitor of
Epigenetic JumonjiC Domain-Containing Histone Demethylases
Martin Roatsch, Inga Hoffmann, Martine I. Abboud, Rebecca L. Hancock, Hanna Tarhonskaya, Kuo-Feng
Hsu, Sarah E. Wilkins, Tzu-Lan Yeh, Kerstin Lippl, Kerstin Serrer, Isabelle Moneke, Theresa D. Ahrens, Dina
Robaa, Sandra Wenzler, Nicolas P. F. Barthes, Henriette Franz, Wolfgang Sippl, Silke Lassmann, Sven
Diederichs, Erik Schleicher, Christopher J. Schofield, Akane Kawamura, Roland Schüle, and Manfred Jung
ACS Chem. Biol., Just Accepted Manuscript • DOI: 10.1021/acschembio.9b00289 • Publication Date (Web): 09 Jul 2019
Downloaded from pubs.acs.org on July 10, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

ACS Chemical Biology
Page 2 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The Clinically Used Iron Chelator Deferasirox is an
Inhibitor of Epigenetic JumonjiC Domain-
Containing Histone Demethylases
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a,
a
b
b
Martin Roatsch, † Inga Hoffmann, Martine I. Abboud, Rebecca L. Hancock,
b
b
b
b
b
Hanna Tarhonskaya, Kuo-Feng Hsu, Sarah E. Wilkins, Tzu-Lan Yeh, Kerstin Lippl,
c
d
e
f
a
Kerstin Serrer, Isabelle Moneke, Theresa D. Ahrens, Dina Robaa, Sandra Wenzler, Nicolas
a
g
f
e
d,h
P. F. Barthes, Henriette Franz, Wolfgang Sippl, Silke Lassmann, Sven Diederichs,
c
b
b
g
a*
Erik Schleicher, Christopher J. Schofield, Akane Kawamura, Roland Schüle, Manfred Jung
a. Institute of Pharmaceutical Sciences, Albert-Ludwigs-Universität Freiburg, Albertstraße
5, 79104 Freiburg i.Br., Germany
2
b. Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford,
OX1 3TA, United Kingdom
c. Institute of Physical Chemistry, Albert-Ludwigs-Universität Freiburg, Albertstraße 21,
7
9104 Freiburg i.Br., Germany
d. Division of Cancer Research, Department of Thoracic Surgery, Medical Center –
University of Freiburg, Faculty of Medicine, University of Freiburg, German Cancer
Consortium (DKTK) – Partner Site Freiburg, Breisacher Straße 115, 79106 Freiburg
i.Br., Germany
ACS Paragon Plus Environment
1

Page 3 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
e. Institute for Surgical Pathology, Medical Center and Faculty of Medicine – University of
Freiburg, Breisacher Straße 115a, 79106 Freiburg i.Br., Germany
f. Institute of Pharmacy, Martin-Luther-University Halle-Wittenberg, Wolfgang-
Langenbeck-Straße 4, 06120 Halle (Saale), Germany
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
g. Central Clinical Research, Medical Center and Faculty of Medicine – University of
Freiburg, Breisacher Straße 66, 79106 Freiburg i.Br., Germany
h. Division of RNA Biology & Cancer, German Cancer Research Center (DKFZ), Im
Neuenheimer Feld 280, 69120 Heidelberg, Germany
ACS Paragon Plus Environment
2

ACS Chemical Biology
Page 4 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT. Fe(II)- and 2-oxoglutarate (2OG)-dependent JumonjiC domain-containing
histone demethylases (JmjC KDMs) are ‘epigenetic eraser’ enzymes involved in the regulation
of gene expression and are emerging drug targets in oncology. We screened a set of clinically
used iron chelators and report that they potently inhibit JMJD2A (KDM4A) in vitro. Mode of
action investigations revealed that one compound, deferasirox, is a bona fide active site-binding
inhibitor as shown by kinetic and spectroscopic studies. Synthesis of derivatives with improved
cell permeability resulted in significant upregulation of histone trimethylation and potent cancer
cell growth inhibition. Deferasirox was also found to inhibit human 2OG-dependent hypoxia
inducible factor prolyl hydroxylase activity. Therapeutic effects of clinically used deferasirox
may thus involve transcriptional regulation through 2OG oxygenase inhibition. Deferasirox
might provide a useful starting point for the development of novel anticancer drugs targeting
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
OG oxygenases and a valuable tool compound for investigations of KDM function.
KEYWORDS
epigenetics • histone demethylase • inhibitor • off-target • clinically used drugs • deferasirox •
Jumonji • hypoxia • 2-oxoglutarate oxygenase
ACS Paragon Plus Environment
3

Page 5 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
INTRODUCTION. Posttranslational modifications of histone proteins, such as lysine
methylation, are a vital mechanism of epigenetic gene regulation in cells.1, 2 These ‘marks’ are
deposited by histone methyltransferases (HMTs) and removed by lysine demethylases (KDMs),
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
of which two mechanistically different classes are known. While LSD1 (KDM1A) demethylates
histones in an FAD-dependent reaction, the much larger class of JumonjiC domain-containing
histone demethylases (JmjC KDMs), discovered in 2006, are Fe(II)- and 2-oxoglutarate (2OG)-
dependent oxygenases. Demethylation proceeds with an oxidative mechanism via formation of
an unstable hemiaminal intermediate, which decomposes to formaldehyde and the demethylated
lysine residue (Figure 1a).3-5
6
Overexpression of JmjC demethylase genes is linked to diseases, in particular cancer, with the
H3K9me
3
and H3K36me
3
demethylase JMJD2A (KDM4A), a member of the KDM4 subfamily
(subtypes A-E), being a prominent example. Elevated levels of JMJD2A have been detected in
human malignancies including breast,7, 8 prostate, bladder, colorectal, and lung cancers. In
breast and prostate hormone-dependent cancers, JMJD2A is implicated as a co-activator of the
estrogen and androgen receptors, respectively.7-9 Conversely, in malignant urothelium, lowered
9
10
11
12
13
JMJD2A levels correlate with poor prognosis. The contrasting roles of JMJD2A in human
1
4
disease have recently been discussed. As JumonjiC KDMs use molecular oxygen as oxidizing
agent for the demethylation reaction, a role in cellular oxygen sensing has also been discussed,
e. g. for the subtype KDM6A.15
JmjC histone demethylases have emerged as promising drug targets,1, 16-20 with inhibitor
discovery programs being reported by us21-27 and others,28-35 as recently reviewed.17, 19, 36-40 The
majority of reported JmjC KDM inhibitors function by active site metal chelation and
competitive displacement of the co-substrate 2OG.36-39 Very recently, covalent KDM5
ACS Paragon Plus Environment
4

ACS Chemical Biology
Page 6 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
inhibitors41, 42 and inhibitors of both the lysyl- and arginyl-demethylase activities have also
43
been reported.
We were, therefore, interested in investigating the effect of clinically used iron chelating drugs
on KDM activity. These compounds are used to treat patients with excess levels of iron ions in
their blood (hemochromatosis). Currently, three such molecules are in clinical use, namely
deferoxamine 1, deferiprone 2, and deferasirox 3a (Figure 1b).44, 45 For our tests, we used two
established in vitro screening assays with orthogonal read-outs as previously reported,21-23
namely the enzyme-coupled formaldehyde dehydrogenase (FDH)-coupled fluorescence intensity
assay and the commercial antibody-based TR-FRET LANCEUltra assay.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R1
N+
R1
N+
R
1
A)
R2
R2
H
3
C
HO H C
R2
2
+HN
JmjC
Fe
2
+
O
O
N
O
H
N
H
N
H
N
Histone
N
Histone
Histone
Histone
Histone
N
Histone
H
H
O
H
O
O
O
O
HCHO
O
O
O
O
O
O
O
R , R2 = H or CH
1
O
2
CO
2
3
B)
R
1
OH
N
O
O
H
N
N
H
N
N
O
OH
O
OH
N
R2
N
N
R2
OH
NH2
O
N
O
OH HO
1
deferoxamine
2 deferiprone
R1 = COOH, R2 = H
3
a deferasirox
®
®
(Exjade , Jadenu , ICL670)
Figure 1. a) Iron(II)- and 2-oxoglutarate-dependent demethylation reaction catalyzed by JmjC
KDMs. b) Chemical structures of investigated clinically used iron chelators 1-3a.
ACS Paragon Plus Environment
5

Page 7 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
RESULTS AND DISCUSSION. Intriguingly, all three compounds 1-3a showed potent
inhibition of demethylation of H3K9me peptides by JMJD2A in these in vitro assays with IC50
3
values between 3 and 17 µM (Table 1). The potency of inhibition is in a comparable range to that
of the well-known JmjC inhibitor JIB-04,33, 46 which was used as reference compound. Inhibition
of JMJD2A and the closely related subtype JMJD2B by 1 and 3 was confirmed using a mass
spectrometry (MS) assay (Supporting Table 1). In a very recent publication, data was presented
implicated that deferiprone 2 and derivatives thereof inhibited some KDMs with comparable
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
47
potency. In contrast to our findings, the authors report no inhibition of the KDM4 subfamily,
but of selected members of other subfamilies. A different in vitro assay was used than in our set-
up, which can potentially explain the observed differences as KDM inhibition is highly sensitive
2+
to co-factor Fe and co-substrate 2OG concentrations. We aimed to study to what extent the
observed inhibition by metal chelators was bona fide inhibition of the enzyme by active site
binding of the molecules or non-specific inhibition arising from sequestration of iron ions in the
2+
assay buffer or extraction of metal ions from the protein, i.e. either Fe from the active site or of
2
+
48
2+
the structural Zn present in some JmjC KDM subtypes. To investigate this, we varied the Fe
concentration in all assays (Supporting Table 2), but the results were inconclusive.
ACS Paragon Plus Environment
6

ACS Chemical Biology
Page 8 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. In vitro inhibition of JMJD2A by iron chelators, the reference JmjC
inhibitor JIB-04 and control compounds.[a] n. d. – not determined. n. i. – no
inhibition.
FDH assay
IC50 / µM
3.22 ± 0.30
LANCEUltra assay
IC50 / µM
Compound R1
R2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
--
--
3.33 ± 0.48
2
--
--
17.4 ± 0.8
7.37 ± 1.58
6.17 ± 1.43
7.77 ± 1.32
14.0 ± 3.1
5.51 ± 0.37
5.49 ± 1.59
17.6
3.87 ± 0.27
3a
3b
COOH
H
H
4.76 ± 0.23
H
5.27 ± 1.00
3c
3d
3e
3f
COOMe
COOH
COOH
COOH
--
H
4.12 ± 0.13
Cl
OMe
F
5.12 ± 0.05
3.66 ± 0.11
4.20 ± 0.22
JIB-04
--
4.12 ± 0.39
1
2a
2b
n. d.
67% inhib. @ 100 µM
n. i. @ 100 µM
1
n. d.
[
a] Data shown are mean ± s. d. of at least two independent experiments.
To further investigate the mechanism of inhibition, we performed enzyme kinetic analyses.
The results showed that only inhibition by deferasirox 3a was, at least in part, competitive with
the co-substrate 2OG (Supporting Figure 1), indicating that inhibition can stem from active site
binding. With different mechanisms of inhibition likely occurring simultaneously, interpretation
2
+
of the kinetic data is further complicated by background Fe chelation as JmjC KDMs do not
2+
efficiently bind 2OG in the absence of Fe . By contrast, inhibition by deferoxamine 1 and
deferiprone 2 was not competitive with 2OG in our experiments (Supporting Figure 2),
2
+
consistent with inhibition occurring only by Fe chelation in solution. The recent report of KDM
inhibition by deferiprone 2 also included docking poses of 2 in the active sites of some KDMs.47
ACS Paragon Plus Environment
7

Page 9 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
Our results highlight, however, that this does not necessarily also imply a 2OG-competitive
inhibition mechanism as evidenced by our kinetic data, again underlining the complexity of
inhibition with regard to iron chelation.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We then adapted the LANCEUltra assay system for the related subtypes JARID1A (KDM5A)
and JMJD3 (KDM6B) and their respective substrates H3K4me
3
and H3K27me for selectivity
3
studies. We observed potent inhibition of these enzymes as well (Table 2), implying general
inhibition of the JmjC KDMs by the entire set of clinically used iron chelators. The lack of
selectivity for specific KDMs may be a challenge in the further development of these inhibitors
for therapeutic use. On the other hand, many JmjC KDM inhibitors published to date, including
reference inhibitor JIB-04, exhibit no or only limited selectivity for certain subfamilies, while
selective inhibition of individual enzyme subtypes remains challenging.36
Table 2. In vitro inhibition of other KDM subtypes in LANCEUltra assays
by iron chelators and reference JMJ inhibitor JIB-04.[a]
JMJD2A
Compound (KDM4A)
JARID1A
(KDM5A)
JMJD3 (KDM6B)
IC50 / µM
IC50 / µM
IC50 / µM
1
3.33 ± 0.48
4.76 ± 0.23
4.12 ± 0.39
4.62 ± 0.27
2.36 ± 0.23
3.95 ± 0.65
4.60 ± 0.65
3a
5.00 ± 0.62
7.20 ± 0.39
JIB-04
[a] Data shown are mean ± s. d. of at least two independent experiments.
Further support for the hypothesis that deferasirox 3a is a direct protein-binding inhibitor of
JMJD2A comes from spectroscopic analyses. Pseudo-modulated field-sweep echo electron
paramagnetic resonance (EPR) spectra of JMJD2A with Fe3+ bound (substituting for the
ACS Paragon Plus Environment
8

ACS Chemical Biology
Page 10 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2+
catalytically active Fe ) are shown in Figure 2 (green curve); the spectrum shows a pronounced
EPR signature between 400 and 650 mT with a number of maxima and minima as is typical for
nitrogen-complexed Fe3 .
+ 49-51
The presence of the EPR-visible Fe stems from aerial oxidation
3+
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2+
of the catalytically active Fe . Upon addition of 3a, additional signals appeared, three of which
are located between 100 and 400 mT, and one broad positive band, which is centered at around
8
70 mT. Clear dose dependency was observed at the marker band located at around 870 mT
(
orange curves in Figure 2 and Supporting Figure 3).
Figure 2. Pseudo-modulated field-sweep echo EPR spectra of JMJD2A in complex with Fe3+
without 3a (green curve) and increasing concentrations (25, 50, 100, 150, 200 µM) of inhibitor
orange curves). The complex of 3a and Fe3+ in solution results in a profoundly different
(
spectrum (blue curve).
3
+
The EPR experiments were also performed on a sample without enzyme, but with Fe and 3a
at a molar ratio of ~1:3. This resulted in significantly different band patterns and signal maxima
(
blue curve), in particular at the magnetic-field positions around 550 mT and 900 mT. The
ACS Paragon Plus Environment
9

Page 11 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
3+
combined results can be rationalized by invoking formation of two different Fe complexes: one
with 3a bound to iron within the active site if enzyme is present and one with iron complexed by
3
+
3a in solution. The formation of a complex of 3a and Fe within the active site of JMJD2A is
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
supported by additional pulsed ESEEM experiments (Supporting Figure 4).
The binding of deferoxamine 1 and deferasirox 3a to JMJD2A was further studied by NMR
spectroscopy using recombinant protein. The addition of JMJD2A complexed with Zn2+ to a
sample of 3a led to significant line broadening and signal reduction of 3a peaks indicating that
3
a binds to JMJD2A (Figure 3a-b). This observation is consistent with wLOGSY analyses
whereby 3a shares the same NOE sign with JMJD2A in a sample containing 3a and JMJD2A
(Figure 3c), indicating the formation of a 3a:JMJD2A complex. By contrast, the addition of
JMJD2A to a sample of 1 led to little line broadening of the deferoxamine peaks suggesting that
1
binds only relatively weakly to JMJD2A, consistent with wLOGSY experiments indicating
only a weak binding event (Supporting Figure 5). Thus, under our experimental conditions, 3a
binds JMJD2A significantly more tightly than 1.
ACS Paragon Plus Environment
1
0

ACS Chemical Biology
Page 12 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 3. NMR analyses of deferasirox binding to JMJD2A (KDM4A) by a) H excitation
1
sculpting suppression NMR analysis, b) H CPMG NMR analysis, and c) wLOGSY NMR
analysis. Blue trace: 200 µM 3a, red trace: 200 µM 3a and 10 µM JMJD2A in buffer containing
2
+
5
0 µM Zn .
As co-crystallization of deferasirox with JMJD2A proved unsuccessful, we further investigated
the binding mode of this inhibitor by molecular docking. The suggested binding of 3a at the
active site of KDMs is corroborated by plausible binding models in JMJD2A, JARID1A, and
JMJD3 (Figure 4 and Supporting Figure 6). The binding poses show that 3a can be
2
+
accommodated in the active site, chelating the Fe ion in a tridentate manner via its triazole-N
and both phenol-O atoms in JMJD2A and JMJD3, and in a bidentate manner via its triazole-N
and one phenol-O in JARID1A. Additionally, viable H-bonding and hydrophobic interactions
with residues of the protein binding pocket are predicted. In JMJD2A (KDM4A), 3a is predicted
to complex the central metal ion via one nitrogen and two phenolic oxygen atoms. One of the
ACS Paragon Plus Environment
1
1

Page 13 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
phenolic rings is embedded in a hydrophobic region via --stacking interactions with F185,
while the phenolic-OH is able to form H-bond interactions with the histidine residue H276
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Molecular docking pose of 3a in the active site of JMJD2A (KDM4A), showing that
3
a can be accommodated in the active site of this enzyme. The ligand is shown in yellow, side
chains of the protein residues involved in the interaction as white sticks, water molecules and the
2+
Fe ion as red and teal spheres, respectively. Predicted hydrogen bonds and interactions with the
metal ion are depicted as yellow dashed lines.
Based on this binding model, we envisaged optimization of potency of deferasirox 3a by the
synthesis of analogs with phenol ring modifications aimed at binding to the conserved lysine
residue K206 in the JMJD2A active site. The synthetic route employed (Scheme 1) followed that
of reported syntheses of this structural class and the commercial deferasirox process.52-56 This
efficient two-step protocol started with condensation of appropriately substituted salicylamides 4
ACS Paragon Plus Environment
1
2

ACS Chemical Biology
Page 14 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and salicylic acids 5 to benzoxazinones 6 followed by condensation with the corresponding
phenylhydrazines 7 to construct the central triazole ring. In the case of fluoro derivative 3f,
synthesis proceeded through the open bis(salicyl)imide form 8, which could similarly be
condensed with phenylhydrazine 7a.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Synthetic route to substituted derivatives of deferasirox 3a and control compounds
1
2a/b.
The substituted derivatives 3b-f were tested in the in vitro assays against JMJD2A (Table 1)
and found to be active in a concentration range similar to 3a. In order to investigate the relevance
of the tridentate iron-binding motif composed of triazole-N and two phenol-O atoms in
deferasirox 3a for potent inhibition, we designed two structural analogues 12a and 12b
ACS Paragon Plus Environment
1
3

Page 15 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
(
Scheme 1), where one or both of the phenol-O atoms have been replaced. These could
synthetically be accessed by condensing picolinaldehyde 9 with phenylhydrazine 7b and
subsequent iodine-catalyzed oxidative coupling of the formed hydrazone intermediate 10 with an
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
57
appropriate benzylamine 11. These compounds were also tested in the LANCEUltra in vitro
demethylation assay (Table 1), revealing that 12a with one phenol group retained some, albeit
drastically reduced, inhibitory potency, while 12b was completely inactive.
Encouraged by the finding that deferasirox 3a is a JMJD2A in vitro inhibitor, we investigated
its biological effects in a cell culture model using KYSE-150 esophageal cancer cells, known to
3
0
be sensitive to KDM inhibition, as well as in HL-60 leukemia cells. Deferasirox 3a showed
potent antiproliferative effects on both cell lines in an MTS assay with GI50 values in the single-
digit micromolar range (Table 3). Importantly, the less polar uncharged derivatives 3b and 3c
showed a 7-fold improvement relative to 3a in KYSE-150 cells. Further, 3d with lipophilic Cl
substituents was still 4-fold more potent than 3a despite the presence of a charged carboxylate.
While HL-60 cells are generally less susceptible to growth inhibition than KYSE-150 cells, the
observed trend in potency was the same. The differences in potency likely reflect, at least in part,
improved cell permeability of the more hydrophobic derivatives 3b-d. It cannot be ruled out that
the ester compound 3c is, at least in part, hydrolyzed by esterases in cells to the free carboxylic
acid 3a and, thus, rather functions as a prodrug of 3a. However, as both 3a and 3c are virtually
equipotent in vitro, the biological effect would be the same, once the compound has crossed the
cell membrane.
Pleasingly, the control compound 12b with no inhibitory activity (Table 1) also exhibited
virtually no antiproliferative effect (Table 3).
ACS Paragon Plus Environment
1
4

ACS Chemical Biology
Page 16 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 3. Antiproliferative effect of deferasirox-based inhibitors 3a-e and control
compound 12b on KYSE-150 esophageal cancer and HL-60 leukemia cells.[a]
n. i. – no inhibition.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
KYSE-150 cells
GI50 / µM
HL-60 cells
GI50 / µM
5.5 ± 0.4
1.7 ± 0.3
1.7 ± 0.2
5.7 ± 1.3
7.4 ± 6.8
n. i.
Entry
3
3
3
3
3
1
a
3.27 ± 0.62
b
c
0.45 ± 0.06
0.46 ± 0.06
d
e
0.85 ± 0.21
2.24 ± 0.06
2b
17% inhib. @ 50 µM
[
a] Data are mean ± s. d. of triplicate experiments.
We then extended our analysis of the cellular effects of 3a and its methyl ester 3c to a broader
panel of esophageal cancer cell lines, where we observed a potent growth inhibitory effect and
consistently higher potency for the less polar compound 3c (Supporting Figure 7).
Moreover, we extended our analysis to a panel of 18 lung adenocarcinoma cell lines and three
non-transformed lung-derived cell lines (BEAS-2B, IMR-90, WI-38), which were treated either
with 3a or 3c, the reference JmjC KDM inhibitor JIB-0433, 46 or one of two inhibitors each of the
histone demethylase LSD1 (KDM1A) and of histone deacetylases (HDACs). The full results of
the screening of an epigenetic inhibitor library against the lung cancer cell line panel will be
reported in due course elsewhere. For structures of the used inhibitors, refer to Supporting
Figure 8a. For the majority of tumor cell lines, 3a and 3c led to loss of cell viability, with 3c
being significantly more potent at the same concentration (t-test: p = 0.015, Supporting
ACS Paragon Plus Environment
1
5

Page 17 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
Figure 8b). Importantly, 3c did not lead to loss of cell viability in all cell lines, arguing against
non-specific cytotoxicity. This was especially apparent in two out of the three non-transformed
cell lines: BEAS-2B and IMR-90. Comparative analysis revealed that the cellular response of 3c
on the entire panel of cell lines correlated best with that of parent compound 3a and second best
with the well-known JmjC KDM inhibitor JIB-04. This resemblance in biological activity again
points to the specific action of 3a and 3c as bona fide JmjC KDM inhibitors within cells. On the
other hand, there is little to no correlation to the biological activity profile of LSD1 or HDAC
inhibitors (Figure 5).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Heatmap of Pearson correlation coefficients for pairwise comparisons of inhibitor
effects on cell viability on the panel of 18 tested lung adenocarcinoma cell lines. Chemical
structures of and literature references to the used HDAC (KK16, Vorinostat) and LSD1 (AW84,
JSF81) inhibitors can be found in Supporting Figure 8a.
Further support for on-target cellular KDM inhibition was observed in cells treated with 3a as
evidenced by a shift to overall increased masses of histones H3.1 and H3.2. This was shown by
quantitative analysis of LC-MS data of histone proteins extracted from HEK293T cells treated
with 3a for 24 h, revealing a shift towards higher modification states (Supporting Figures 9-12).
The corresponding immunoblots of histone extracts also manifested an increase in the levels of
H3K4me
3
, H3K9me
3
, H3K27me
3
, and H3K36me methylation states upon treatment with 3a,
3
with no obvious changes in the levels of acetyl lysine (K-ac) (Figure 6a and Supporting Figure
ACS Paragon Plus Environment
1
6

ACS Chemical Biology
Page 18 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
3). This observation indicates an overall increase in histone H3 methylation, and methylation-
specific action of 3a rather than unspecific effects on chromatin (e.g. via changes in acetylation).
An increase in histone H3K9 trimethylation was also observed by immunofluorescence
microscopy upon treatment of KYSE-150 cancer cells with 3a and its synthetic derivatives 3b-e
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
Figure 6b and Supporting Figure 14). Importantly, for the less polar derivatives 3b and 3c,
lower concentrations were required to reach the same apparent level of increased trimethylation.
Dose-dependent studies manifested an EC50 value of 40.5 µM for 3a, 3.3 µM for 3b, and 10 µM
for 3c, consistent with the improved antiproliferative effects of 3b/c on cells (Table 3), likely
caused by the enhanced cell permeability relative to 3a. The potency of our compounds to
upregulate histone trimethylation in the lower micromolar range is remarkable considering that
other iron chelators such as deferiprone 2 and its derivatives only showed similar effects at
concentrations of 100 or even up to 250 µM.47
ACS Paragon Plus Environment
1
7

Page 19 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. Treatment of cells with 3a causes histone hypermethylation. a) Representative
immunoblots of histone extracts from HEK293T cells treated with DMSO or 3a for 24 h. b)
Immunofluorescence microscopy of KYSE-150 cells treated with DMSO or 3a for 72 h stained
for H3K9 trimethylation.
Moreover, deferasirox and its analogues (3a-c) increased the H3K9me staining in U2OS cells
3
ectopically expressing JMJD2A in a concentration-dependent manner, suggesting that these
compounds may quite directly inhibit demethylation by JMJD2A in cells. However, a dose-
dependent increase in H3K9me is also observed in cells overexpressing the catalytically inactive
3
H188A variant of JMJD2A at higher concentrations (in particular 3c), suggesting that these
compounds may affect hypermethylation by also affecting endogenous KDMs or via other
mechanisms. This is not unexpected as demonstrated by the promiscuity of KDM inhibition by
3
a in vitro (Table 2). Interestingly, consistent with the data obtained for the cell proliferation
assays (Table 3), the cellular inhibition of JMJD2A was enhanced by an order of magnitude for
the non-polar derivatives 3b and 3c relative to the more polar compounds (Figure 7 and
Supporting Figure 15, Supporting Table 3).
ACS Paragon Plus Environment
1
8

ACS Chemical Biology
Page 20 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Quantitative immunofluorescence analysis of H3K9me
3
staining. U2OS cells
overexpressing FLAG-KDM4A (wild type, WT) or catalytically inactive variant (H188A) were
dosed with varying concentrations of 3a, 3b, or 3c for 24 h. H3K9me fluorescence intensities of
3
KDM4A WT are shown as solid lines and H188A as dotted lines. On the intensity axis, the lower
dashed grid line indicates the fluorescence intensity for control 1% DMSO dosed WT cells,
whereas the upper dashed line indicates the control 1% DMSO dosed H188A cells. Shown is
average fluorescence intensity (n > 50 cells ± s. e. m.).
We then investigated the generality of our results for human 2OG oxygenase inhibition (there
are 60-70 such human enzymes). Because of its role in the regulation of erythropoietin (EPO),
we studied the effects of 1 and 3a on catalysis by the prolyl hydroxylase domain 2 (PHD2)
enzyme, which is a 2OG oxygenase catalyzing hydroxylation of the hypoxia inducible factor-α
subunits (HIF-α), using a peptide fragment of the C-terminal oxygen dependent degradation
domain (CODD) of HIF-1α.58 The results were similar to those observed for inhibition of
JMJD2A. Thus, using a low concentration of Fe2+ in these inhibition assays, 1 and 3a were
moderately potent PHD2 inhibitors (2.4 ± 0.2 μM and 2.4 ± 0.2 μM, respectively), but at higher
ACS Paragon Plus Environment
1
9

Page 21 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
2+
Fe concentration, the IC50 increased to 53.1 ± 12.5 μM with 3a (Supporting Figure 16). NMR
studies58, 59 implied deferoxamine 1 inhibits substantially by Fe2+ sequestration in solution,
whereas deferasirox 3a inhibits by a mixture of protein binding and Fe2+ sequestration in
solution. Interestingly, although protein binding has been observed under the NMR conditions,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
a did not (efficiently) displace 2OG from PHD2.Zn complex or NOG from KDM4A.Fe
complex, suggesting complexity in the inhibition modes (Supporting Figure 17). Note that metal
ions are required for 2OG binding and the NMR results are compatible with the kinetic analyses
due to the different conditions (i.e. the interaction of 1/3a with the enzymes may be different
under catalytic and non-catalytic conditions, due to conformational changes during catalysis).
6
0
Consistent with prior reports, both 1 and 3a upregulated HIF-α levels in cells and inhibited
ODD prolyl hydroxylation (Supporting Figure 18). Some histone-modifying enzymes are HIF
target genes, thus, PHD inhibition by 1, 3a, and related compounds may account for some of the
complexity in our cellular studies.
CONCLUSION. In summary, we have shown that the small molecule iron chelator
deferasirox 3a can function as a bona fide active site-binding small molecule inhibitor of
Jumonji-type histone KDMs as evidenced by kinetic and extensive spectroscopic methods, both
with isolated enzyme and in cancer cell lines. In contrast to deferoxamine 1, which substantially
inhibits by complexing Fe2+ in solution, as observed with JMJD2A and PHD2, 3a has the
potential to alter epigenetic regulation and gene expression by globally increasing histone
methylation. The precise mode(s) of binding of 3a to the JmjC KDMs, the PHDs, and potentially
other 2OG oxygenases, requires further investigation. Our results imply more than one inhibition
mode is likely involved. Deferasirox 3a was developed to chelate iron ions in patient blood, so it
ACS Paragon Plus Environment
2
0

ACS Chemical Biology
Page 22 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
is highly soluble in water. Through simple chemical modifications, we have improved the
biological effects of its core structure on intracellular targets, i.e. the JmjC KDMs. Deferasirox
could, thus, be a starting point for the development of novel anticancer drugs based on a
clinically proven lead structure and represents a valuable test compound to investigate the
cellular effects of pan-KDM inhibition in vitro and, possibly, in vivo.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Indeed, there is already some evidence for a beneficial in vivo effect when cancer cells or
patients have been treated with 3a.56, 61-65 Treatment with 1 or 3a has been reported to inhibit
proliferation of esophageal cancer cells in cell culture and a mouse xenograft model and
increased susceptibility to standard chemotherapy. The mechanism of action, however, was
elusive.61 Similar effects were also observed for 3a in leukemia cell lines, independent of
63
64
exogenously supplied iron, and gastric cancer cells. In terms of clinical use, it was found that
the mortality of patients with myelodysplastic syndromes receiving frequent blood transfusions
was significantly reduced when they were additionally treated with 3a.65, 66 Therefore, the use of
iron chelators like 3a has been proposed as a novel avenue in antitumor treatment.67-69
Our results suggest that the observed anticancer potential may be due, at least in part, to 2OG
oxygenase inhibition and the consequent changes in epigenetic gene regulation of cancer cells.
As much pharmacological and clinical safety data for this compound class are already available,
3
a is an attractive starting point for further development. Our results, however, also raise
concerns because 3a is currently in clinical use for an entirely unrelated disorder (iron overload
disease) and the potential epigenetic effects of this compound due to the modulation of histone
methylation and hence, transcription, might have been overlooked so far.
ACS Paragon Plus Environment
2
1

Page 23 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
EXPERIMENTAL SECTION
Test compounds
Clinically used iron chelators were from commercial sources and used for testing without
further purification. Deferoxamine mesylate 1 was from Sigma-Aldrich (catalog: D9533),
deferiprone 2 was from Acros Organics (catalog: 278740050), and deferasirox 3a was obtained
from Selleckchem (catalog: S1712) for the initial in vitro screening and from AK Scientific
(catalog: V0697) for the mass spectrometric analyses. Later, it was re-synthesized in house (see
below) and the results confirmed those obtained with the commercial sample. Derivatives 3b-f
were synthesized in house (see below).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
All screening compounds 1, 2, 3a-f, 12a/b were tested for known classes of assay interference
compounds using the publicly available online tool “False Positive Remover”
(
www.cbligand.org/PAINS). None of the tested compounds were flagged as PAINS. Moreover,
all compounds reported in this study were characterized in at least two in vitro assays with
different readout techniques to provide orthogonal proof of their specific activities.
Enzyme inhibition assays were performed as previously reported.21-23
JMJD2A (KDM4A) FDH assay: The formaldehyde dehydrogenase (FDH) enzyme-coupled
demethylase activity assay was performed in a total volume of 20 µL on white OptiPlate-384
microtiter plates (PerkinElmer) using a 50 mM HEPES buffer at pH = 7.50 containing 0.01%
−
1
(v/v) Tween-20. A solution of 0.10 mgꞏmL (2.4 µM) KDM4A 1-359 was pre-incubated with
compound solutions of varying concentration (0 – 400 µM) in DMSO at room temperature for
1
0 minutes. A substrate solution containing 100 µM ascorbic acid, 10 µM FeSO
, 0.001 UꞏµL−1
4
+
FDH, 500 µM NAD , 50 µM 2-oxoglutarate, and 35 µM of H3K9me
3
7-14 substrate peptide
ARK(me
3
)-STGGK-NH
2
(PSL Peptide Specialty Laboratories) was added (final concentrations).
ACS Paragon Plus Environment
2
2

ACS Chemical Biology
Page 24 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Final DMSO concentration was 2% (v/v) in all wells. Fluorescence intensity of the forming
product NADH was measured at ex = 330 nm and em = 460 nm on a POLARstar Optima
microplate reader (BMG Labtech) immediately after addition (t = 0) and after one hour of
incubation on a horizontal shaker at 37°C. Values were blank-corrected and the difference in
intensity at t = 1 h and t = 0 was taken as a measurement of enzyme activity. Activity in % is in
comparison to compound-free DMSO control and no-substrate negative control. Inhibition
curves were analyzed by sigmoidal curve fitting using GraphPad Prism 4.00 and IC50 values
calculated from the fit parameters as mean ± s. d. from two independent experiments.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
JMJD2A (KDM4A) LANCE assay: The commercial antibody-based LANCEUltra
demethylase activity assay (PerkinElmer) was performed in a total volume of 10 µL on white
OptiPlate-384 microtiter plates (PerkinElmer) using a 50 mM HEPES buffer at pH = 7.50
containing 0.01% (v/v) Tween-20 and 0.01% (w/v) BSA. A solution of 60 nM JMJD2A
(KDM4A) 1-359 was pre-incubated with compound solutions of varying concentration in DMSO
at room temperature for 10 minutes. A substrate solution containing 100 µM ascorbic acid, 5 µM
FeSO
4
, 1 µM 2-oxoglutarate, and 400 nM of biotinylated H3K9me
3
1-21 substrate peptide
ARTKQTARK(me
3
)-STGGKAPRKQLA-GGK(biotin) (BPS Bioscience) was added (final
concentrations). Final DMSO concentration was 5% (v/v) in all wells. Plates were incubated on a
horizontal shaker at room temperature for 45 minutes. Reactions were stopped by addition of
10 µL of detection mix containing 2 nM europium-labeled anti-H3K9me
2
LANCE antibody
(
PerkinElmer), 50 nM ULight-streptavidin dye (PerkinElmer), and 1 mM EDTA in 1X LANCE
detection buffer (PerkinElmer) (final concentrations). Plates were again incubated on a
horizontal shaker at room temperature for 60 minutes. FRET intensity was measured on a
PerkinElmer EnVision 2102 multilabel plate reader at ex = 340 nm and em = 665 nm with a
ACS Paragon Plus Environment
2
3

Page 25 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
delay of 100 µs. Values were blank-corrected and activity in % is in comparison to compound-
free DMSO control and no-enzyme negative control. Inhibition curves were analyzed by
sigmoidal curve fitting using GraphPad Prism 4.00 (GraphPad Software) and IC50 values
calculated from the fit parameters as mean ± s. d. from two independent experiments.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
JARID1A (KDM5A) LANCE assay: The antibody-based LANCEUltra activity assay for
JARID1A (KDM5A) was performed essentially as described for JMJD2A (KDM4A) with the
following modifications: a solution of 25 nM full-length JARID1A 1-1090 (BPS BioScience)
was used with 100 nM of biotinylated H3K4me
QTARKSTGGKAPRKQLA-GGK(biotin) (AnaSpec). The detection mix contained the
appropriate europium-labeled anti-H3K4me /me LANCE antibody (PerkinElmer).
3
1-21 substrate peptide ARTK(me
3
)-
2
1
JMJD3 (KDM6B) LANCE assay: The antibody-based LANCEUltra activity assay for JMJD3
(
KDM6B) was performed essentially as described for JMJD2A (KDM4A) with the following
modifications: a solution of 50 nM catalytic domain JMJD3 1043-end (BPS BioScience) was
used with 400 nM of biotinylated H3K27me 21-44 substrate peptide ATKAARK(me )-
3
3
SAPATGGVKKPHRYRPG-GK(biotin) (PSL Peptide Specialty Laboratories). Incubation time
for the enzymatic reaction was 120 min and the detection mix contained the appropriate
europium-labeled anti-H3K27me LANCE antibody (PerkinElmer).
2
JMJD2A (KDM4A) 2-oxoglutarate competition assay: In order to assess the competitivity of
enzyme inhibition by test compounds to 2-oxoglutarate, the LANCEUltra assay was performed
as described above with varying concentrations of 2-oxoglutarate (0 – 5.0 µM) and inhibitor.
Each combination was tested in duplicate. The blank-corrected LANCE signal was compared to
a pre-established calibration curve to determine the amount of demethylated product formed and,
thus, the reaction velocity over the incubation time. Curve fitting for compounds that were not
ACS Paragon Plus Environment
2
4

ACS Chemical Biology
Page 26 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
competitive with regard to 2-oxoglutarate (Supporting Figure 2) was performed using the
Michaelis-Menten equation:
vmax [2-oxoglutarate]
v 
K  [2-oxoglutarate]
M
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EPR spectroscopy
All Q-band electron paramagnetic resonance (EPR) measurements were performed on a
Bruker ELEXSYS E580 spectrometer equipped using a Bruker EN 5107D2 1.6 mm dielectric
resonator. Temperature (4 K) was controlled by a liquid helium cryostat (Oxford CF935O) and
an ITC temperature controller (Oxford ITC4). Field-sweep echo detected EPR spectra were
recorded at a magnetic field range of 0 – 1400 mT using a /2 pulse of 16 ns and a  value of
32 ns at 6 dB microwave power. All obtained spectra were pseudo-modulated (14 mT
modulation amplitude) after subtraction of a linear background function.
Three-pulse ESEEM data were collected using a /2 –  – /2 –  – /2 pulse sequence (/2 =
1
6 ns;  = 100 ns) together with a four-step phase cycling. Primary ESEEM data were further
processed by subtraction of an exponential decay background function, processed by a Hamming
window, zero-filled to 1500 points, and Fourier-transformed.
All samples contained 100 µM FeSO . Enzyme-containing samples contained 65 µM of the
4
catalytic domain of JMJD2A residues 1-359. Deferasirox 3a concentrations were variable from
0, 25, 50, 100, 150, to 200 µM. For the sample without enzyme, [3a] was 200 µM.
In contrast to enzyme activity or kinetic experiments, these samples did not contain any
ascorbic acid as reducing agent so that a partial oxidation from Fe2+ to Fe3+ was possible.
Therefore, iron(III) species are observed in all EPR spectra.
ACS Paragon Plus Environment
2
5

Page 27 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
NMR spectroscopy
All nuclear magnetic resonance (NMR) spectra were recorded with a Bruker AVIII 600 MHz
NMR spectrometer equipped with a BB-F/1H Prodigy N cryoprobe using 3 mm diameter tubes
Norell). Data were processed with Bruker 3.1 software.
Assay mixtures were buffered in 50 mM Tris-d11, pH 7.5, in 10% D
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
(
2
O and 90% H O
2
2
+
2+
containing 50 µM Zn (except for KDM4A experiments, the enzyme was in its Fe -metallated
state).
1H excitation sculpting suppression NMR: Spectra were typically obtained using 256 scans
and a relaxation delay of 1 s. A 2 ms Sinc pulse was used for water suppression. Prior to Fourier
transformation, data were multiplied with an exponential function with 2 Hz line broadening.58
1H CPMG NMR: Typical experimental parameters for Carr-Purcell-Meiboom-Gill (CPMG)
NMR spectroscopy were: total echo time: 40 ms; relaxation delay: 2 s; and number of transients:
2
64. The PROJECT-CPMG sequence (90° x – [– 180° y – – 90° y – – 180° y – ] – acq)
n
was applied. Water suppression was achieved by presaturation. Prior to Fourier transformation,
the data were multiplied with an exponential function with 3 Hz line broadening.58
wLOGSY NMR experiments: water-Ligand Observed Gradient SpectroscopY (wLOGSY)
experiments were conducted using the pulse sequence described by Dalvit et al.70 Typical
experimental parameters were as follows: mixing time: 1 s; relaxation delay: 2 s; and number of
transients: 256. Solvent excitation was achieved using a 16 ms 180° selective rectangular shape
pulse with 1000 points (Squa100.1000) set at the H
2
O frequency. Water suppression was
achieved by a 2 ms Sinc pulse (Sinc1.1000) at the H
2
O frequency.58
ACS Paragon Plus Environment
2
6

ACS Chemical Biology
Page 28 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Immunoblotting
Histones were prepared as described above by acid extraction from HEK293T cells.
Approximately 0.2 µg of extracted proteins were separated by SDS-PAGE and analyzed by
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
immunoblotting with antibodies directed against H3 (abcam, ab10799), H3K9me
ab8898), H3K36me (abcam, ab9050), H3K79me (abcam, ab2621), acetyl-lysine (abcam,
ab61257), and H3K27me (Merck Millipore, 07-449) to investigate changes in specific histone
3
(abcam,
3
3
3
modifications. Immunoblots were imaged using fluorescent secondary antibodies and Odyssey
infrared imaging technology (LI-COR Biosciences), enabling quantification of histone marks by
infrared fluorescent signals, which are directly proportional to the amount of antigen.
Immunofluorescence microscopy in KYSE-150 cells
4
,000 cells per well were cultured overnight in sterile ViewPlate-96 black culture plates
(PerkinElmer) at 37°C and 5% CO . Compounds were added to a final concentration of 1.0 –
00 µM in 0.5% DMSO and incubated for 72 h at 37°C and 5% CO . The following work was
2
1
2
performed at room temperature. Cells were washed with phosphate-buffered saline (PBS), fixed
for 20 minutes with 4% formaldehyde in PBS, and washed with PBS. After cells were
permeabilized with 0.5% Triton X-100 in PBS for 8 minutes, non-specific binding sites were
blocked with 3% FCS in PBS for 30 minutes. Cells were incubated with the 1:500 diluted
primary antibody rabbit anti-H3K9me (abcam, ab8898) in 3% FCS in PBS overnight. After
3
washing three times with PBS, the secondary antibody anti-rabbit Alexa Fluor® 488 (Life
Technologies, A-11008) was added in a 1:500 dilution in 3% FCS in PBS and incubated for 1 h
in the dark. After a threefold washing step with PBS, cells were stained with DAPI (10 µg∙mL-1
ACS Paragon Plus Environment
2
7

Page 29 of 54
ACS Chemical Biology
1
2
3
4
5
6
7
8
9
in PBS) for 5 minutes. Image acquisition was conducted in PBS, after a threefold washing step
with PBS, using an Olympus Scan^R Screening station.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Immunofluorescence microscopy in U2OS cells with wild-type and mutant KDM4A
U2OS cells were maintained in 10% FBS, 1 mM glutamine in DMEM at 37°C and 5% CO .
2
Cells were seeded at 500 cells per well in black CellCarrier-96 plates (PerkinElmer) and
incubated overnight. Cells were transfected with pCDNA3-FLAG-KDM4A (WT) or catalytically
7
1
inactive control pCDNA3-FLAG-KDM4A H188A (MUT) using Lipofectamine 3000 (Thermo)
in OPTIMEM. After 4 h, the transfection reagent was replaced with media supplemented with
inhibitors (1% (v/v) DMSO final) and incubated for further 24 h. Cells were washed with PBS,
26
fixed and permeabilized, and stained as previously described. Cells were stained for FLAG
(
mouse monoclonal, Sigma, F1804), H3K9me (rabbit polyclonal, abcam, ab8898) and DAPI
3
(Sigma, D9564) and probed using fluorescence-conjugated secondary antibodies (goat anti-rabbit
Alexa Fluor 488, Life Technologies, A11034 and goat anti-mouse Alexa Fluor 594, Life
Technologies, A11032). Image acquisition was performed using an Operetta high content
imaging system (PerkinElmer), and images were analyzed using Harmony software
(
PerkinElmer). Cell nuclei were automatically identified by DAPI staining. H3K9me levels of
3
cells overexpressing KDM4A WT/MUT were analyzed.
CHEMISTRY
General Methods. Starting materials, reagents, and analytical-grade solvents were obtained
from commercial sources and used without further purification. Solvents for preparative
chromatography were generally purified by distillation prior to use. Reaction progress was
ACS Paragon Plus Environment
2
8