Enzyme kinetic studies of histone demethylases KDM4C and KDM6A: Towards understanding selectivity of inhibitors targeting oncogenic histone demethylases

Article abstract of DOI:10.1016/j.febslet.2011.05.023

To investigate ligand selectivity between the oncogenic KDM4C and tumor repressor protein KDM6A histone demethylases, KDM4C and KDM6A were enzymatically characterized, and subsequently, four compounds were tested for inhibitory effects. 2,4-dicarboxypyrid

Full text of DOI:10.1016/j.febslet.2011.05.023

FEBS Letters 585 (2011) 1951–1956
journal homepage: www.FEBSLetters.org
Enzyme kinetic studies of histone demethylases KDM4C and KDM6A: Towards
understanding selectivity of inhibitors targeting oncogenic histone demethylases
Jan B.L. Kristensen a, Anders L. Nielsen a,b, Lars Jørgensen a, Line H. Kristensen a, Charlotte Helgstrand a,
Lina Juknaite a, Jesper L. Kristensen a, Jette S. Kastrup a, Rasmus P. Clausen a, Lars Olsen a, Michael Gajhede a,
⇑
a Department of Medicinal Chemistry, Faculty of Pharmaceutical Sciences, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen, Denmark
b Novo Nordisk A/S, Biopharm Chemistry, Novo Nordisk Park 1, DK-2760 Måløv, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
To investigate ligand selectivity between the oncogenic KDM4C and tumor repressor protein KDM6A
histone demethylases, KDM4C and KDM6A were enzymatically characterized, and subsequently,
four compounds were tested for inhibitory effects. 2,4-dicarboxypyridine and (R)-N-oxalyl-O-ben-
zyltyrosine (3) are both known to bind to a close KDM4C homolog and 3 binds in the part of the cav-
ity that accommodates the side chain in position 11 of histone 3. The inhibition measurements
showed significant selectivity between KDM4C and KDM6A. This demonstrates that despite very
similar active site topologies, selectivity between Jumonji family histone demethylases can be
obtained even with small molecule ligands.
Received 11 March 2011
Accepted 4 May 2011
Available online 12 May 2011
Edited by Christian Griesinger
Keywords:
Histone lysine demethylases
Oncogenic
Ó 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
Tumor repressor
Enzyme kinetics
Ligand selectivity
1. Introduction
to be an oncogene, thus making the development of inhibitors an
important issue. A number of inhibitors against KDM4C and the
The organization of chromatin plays an important role in gene
regulation. The state of chromatin is determined by covalent modi-
fications of the DNA itself or of the histones that organizes DNA in
the nucleosomes. One such modification is methylation of lysines
in the N-terminal regions of in particular histones H3 and H4 [1].
These methylations were thought to be irreversible until the discov-
ery of lysine specific demethylase LSD1 [2]. Shortly after, another
family of histone lysine-specific demethylases (HDMs) was discov-
ered [3]. These enzymes are dioxygenases containing a Jumonji C
(JmjC) domain with an active site containing Fe(II) and the co-factor
closely related KDM4A and KDM4E have recently been disclosed
[6–9], and the binding mode of a number of them have been estab-
lished, since 3D structures are available of the catalytic cores (cc) of
KDM4A [10], KDM4C (PDB code 2XML, unpublished) and KDM4D
(PDB code 3DXT, unpublished).
In 2007, the histone 3 lysine 27 (H3K27) specific HDM denoted
KDM6A was discovered, and shown to be important for the regula-
tion of HOX genes that are central during cellular differentiation
[11]. Results accumulated during the later years suggest that
KDM6A acts as a tumor repressor [12–15]. Thus, in terms of the po-
tential of JmjC HDM inhibitors as anti-cancer agents, it appears cru-
cial that inhibitors targeting KDM4C do not affect KDM6A function
and activity. The 3D structure of KDM6A is currently unknown, how-
ever, the structure of the cc of the closely related KDM6B has re-
cently been deposited in PDB (PDB code 2XXZ, unpublished). This
enables structural comparison as a tool to identify differences in
the active site architecture and, through this comparison, to obtain
inhibitors that display selectivity between KDM4C and KDM6A.
Using the oncogenic KDM4C and tumor repressor protein
KDM6A as model system, we show that selective inhibition of
KDM4C by small molecule inhibitors is achievable. Furthermore,
we demonstrate that inhibition of KDM4C by targeting of the
specific part of the active site accommodating position 11 of the his-
tone 3 peptide substrate, based on a strategy described elsewhere
a-ketoglutaric acid.
In 2006, the discovery of the HDM KDM4C, a histone 3 lysine 9
(H3K9) demethylase was published. It was shown that inhibition
of this HDM decreases tumor cell proliferation [4], and the gene
had previously been found to be up regulated in cell lines derived
from esophageal squamous carcinomas [5]. This suggested KDM4C
Abbreviations: cc, catalytic core; cc4C, catalytic core of KDM4C; cc6A, catalytic
core of KDM6A; 2,4-DCP, 2,4-dicarboxypyridine; FDH, formaldehyde dehydroge-
nase; H3K9, histone 3 lysine 9; H3K27, histone 3 lysine 27; HDM, histone lysine-
specific demethylase; IMAC, immobilized metal affinity chromatography; JmjC,
Jumonji C; 1, (R)-N-oxalyl-tyrosine; 2, (R)-O-benzyltyrosine; 3, (R)-N-oxalyl-O-
benzyltyrosine; NOG, N-oxalylglycine; SEC, size exclusion chromatography
⇑
Corresponding author. Fax: +45 35 33 60 41.
E-mail address: mig@farma.ku.dk (M. Gajhede).
0014-5793/$36.00 Ó 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.febslet.2011.05.023

1952
J.B.L. Kristensen et al. / FEBS Letters 585 (2011) 1951–1956
4-hydroxy cinnamic acid and polystyrene NBS™ treated 384 wells
plates used for FDH assay were from Sigma–Aldrich, Denmark. Sol-
vents for MALDI-TOF were of HPLC-grade and from Merck, Den-
mark. EDTA free protease inhibitor cocktail tablets used in
protein purification were from Roche. Histone tail H3K9me3 and
H3K27me3 synthetic peptide substrates were purchased from Pep-
tide 2.0, USA or generously donated by Novo Nordisk A/S, Den-
mark. The plasmid coding for cc4C (1–349) was generously
donated by the Biotech Research and Innovation Center (BRIC),
Denmark. The pOPIN vector F was generously donated by the Ox-
ford Protein Production Facility, Great Britain. All reagents and sol-
vents for the synthesis of (R)-N-oxalyl-tyrosine (1) and (R)-N-
oxalyl-O-benzyltyrosine (3) were purchased from Sigma–Aldrich,
Denmark or VWR, Denmark and used without further purification.
NMR spectra were recorded on a 300 MHz Varian spectrometer at
room temperature (r.t.). Elemental analysis was performed at the
University of Vienna, Austria. Melting points were determined in
an open capillary tube and are uncorrected.
Fig. 1. Synthesis of compounds 1–3 from (R)-tyrosine. (a) SOCl2, CH3OH, reflux, 4 h;
(b) monomethyl oxalylchloride, toluene, reflux, 4 h; (c) 1 M NaOH, r.t., 2 h, then ion-
exchange; (d) CuSO4, NaOH, benzylbromide, H2O/CH3OH, r.t., 16 h, and then
neutralization.
[8], can lead to KDM4 subtype-selective inhibitors. Moreover, since
the kinetic characterizations of these proteins have not been re-
ported, we present the kinetic characterization of the catalytic cores
of KDM4C (cc4C) and KDM6A (cc6A). Based on the kinetic character-
izations, we identify H3 peptide substrates suitable for screening
purposes and utilize these to identify KDM4C selective inhibitors.
2.2. Cloning, expression and purification of cc4C and cc6A
The detailed procedures are supplied in Supplementary infor-
mation. In summary, the N-terminal His-tagged proteins were ex-
pressed in Escherichia coli using auto-induction [16] and purified
by immobilized metal affinity chromatography (IMAC) and size
exclusion chromatography (SEC).
2. Materials and methods
2.3. Synthesis of compounds 1–3
2.1. General
Compounds 1 and 3 were both synthesized starting from (R)-
tyrosine (Fig. 1). O-Benzylation of (R)-tyrosine to produce (R)-O-
benzyltyrosine (2) was done similar to the procedure of Otake
and Izawa [17]. The N-oxalyl amino acids, 1 and 3, were produced
All chemicals used for buffers, co-factors, lysozyme, DNAse,
formaldehyde dehydrogenase (FDH) from Pseudomonas putida,
N-oxalylglycine (NOG), 2,4-dicarboxypyridine (2,4-DCP), a-cyano-
Fig. 2. Enzyme kinetic plots of 25
l
M cc6A for synthetic 24- and 15-meric H3K27me3 substrates and 2.5
lM cc4C for 8-meric H3K9me3 substrate. Initial velocity (V0)
(l
M formed NADH minÀ1) as a function of concentration of peptide substrate in
l
M. (A) The kinetic profile of the 24-meric peptide substrate at cc6A. (B) The kinetic profile of
the 15-meric peptide substrate at cc6A. (C) The kinetic profile of the 8-meric peptide substrate at cc4C. Corresponding Hill plots have been inserted to illustrate the negative
cooperativity observed for the 24-meric HeK27me3 substrate. Error bars represent one standard deviation (S.D.).

J.B.L. Kristensen et al. / FEBS Letters 585 (2011) 1951–1956
1953
using a method described elsewhere [18]. The detailed procedure
2.6. GRID analysis
is stated in Supplementary information.
The GRID program version 22 [20] was used to characterize the
binding pockets of cc4A and cc6B using the C3 probe.
2.4. Enzyme kinetics and inhibition studies
For determination of enzyme kinetics and inhibition studies of
cc4C and cc6A, a FDH coupled assay was set up, essentially as de-
scribed elsewhere [19]. Experiments were carried out in 384 wells,
low volume, flat bottomed solid black NBS™ surface treated poly-
styrene plates (Corning) and on the same batch of protein. Assay
3. Results and discussion
3.1. The ccs of KDM4C and KDM6A
Approximately 2 mg of purified cc4C (1–349) and cc6A (940–
1401) was obtained pr. liter expression media. Initial protein pur-
ity after IMAC purifications were approximately 60–75%. The pro-
tein purity after SEC was higher than 90% as assessed by SDS–PAGE
(Supplementary Fig. S2).
buffer consisted of 50 mM HEPES, pH 7.5, 50 mM NaCl, 500 lM
a
-ketoglutaric acid, 0.5–5 mM ascorbate, 2 mM NAD+, 0.05–0.5 U
FDH and 50–500 lM Fe(II). Assays were carried out in triplicate
measurements in 10% DMSO at 37 °C and the total reaction volume
was 25 l. For the determination of kinetics parameters, 2.5
cc4C was incubated with 8-meric H3(7–14)K9me3 peptide
substrate ARK(me3)STGGK (7.8–500 M) and 25 M cc6A was
incubated with 24-meric H3(21–44)K27me3 ATKAARK(me3)SAPA-
TGGVKKPHRYRPG (5–2000 M), 15-meric H3(20–34)K27me3
LATKAARK(me3)SAPATGG (5–2000 M) and 9-meric H3(23–
31)K27me3 KAARK(me3)SAPA (5–2000 M) substrates. For
inhibition studies, 5 M cc4C and 12.5 M cc6A were incubated
for 30 min with various concentrations of ligands before adding
100 M of H3K9me3 in the case of cc4C and 500 M H3K27me3
l
lM
3.2. Enzyme kinetics of cc4C and cc6A
l
l
To identify histone 3 peptide substrates suitable for screening
purposes, a series of H3K27me3 peptides were tested. The results
show that different cc6A peptide substrates resulted in different ki-
netic profiles. The 24-meric H3K27me3 substrate exhibited a nega-
tive cooperative kinetic profile with a Hill slope of 0.64, while the 15-
meric substrate exhibited Michaelis–Menten kinetics (Fig. 2). Turn-
over of a shorter (9-meric) peptide was not detected by FDH assay or
by MALDI-TOF. Based on the kinetic studies, the 15-meric peptide
substrate was chosen for inhibition studies. With regard to cc4C,
only an 8-meric peptide was tested, since this substrate has fre-
quently been used in cc4A, cc4C and cc4E inhibition studies [7,8].
This substrate also exhibited Michaelis–Menten kinetics (Fig. 2).
l
l
l
l
l
l
l
in the case of cc6A. The formation of NADH, measured as
an increase in relative fluorescence units, were monitored with a
Safire2™ microplate reader (Tecan) in 30 s cycles over
period of minimum 15 min (up to 24 h) and converted to
M NADH formed minÀ1 using a NADH standard curve (Supple-
a
l
mentary Fig. S1). Excitation wavelength and emission wavelength
was 355 nm and 460 nm, respectively. For inhibition studies, the
Various concentrations of Fe(II) (50–500 lM) and FDH (0.05–
0.5 U) did not alter the turnover of peptide substrates. The kinetic
parameters Kprime/Km, kcat and kcat/Km of cc6Aand cc4C for thevarious
peptide substrates are summarized in Table 1. The difference in Km/
Kprime and Kcat indicates that the 15-meric substrate may have a low-
er binding affinity, but a higher turnover. However, comparison of
the substrate specificity constants (kcat/Km values) reveals that
cc6A demethylates the 24-meric substrate approximately 2.5-fold
more efficiently than the 15-meric substrate. Also, the kinetics re-
vealed that KDM4C is a more efficient enzyme than KDM6A, when
comparing the substrate specificity constants. General enzymatic
activity was confirmed by MALDI-TOF MS. The H3K27me3 cc6A sub-
strates were primarily demethylated to H3K27me2, but in case of
the 24-meric substrate also to trace amounts of the monomethylat-
ed species (Supplementary Fig. S3). This correlates with previous
findings for full length KDM6A incubated with core histones
[11,21]. An 8-meric H3K9me3 cc4C substrate was primarily
demethylated to H3K9me2 but also to the monomethylated species
(Supplementary Fig. S3). This observation is in agreement with pre-
vious findings for the full length protein, in which a 40-meric
synthetic substrate and core histones were used [4].
8-meric cc4C (100 lM) and 15-meric cc6A (500 lM) substrates
were used. Reaction mixtures of assay buffer without substrate
were used as negative controls. For curve fitting and data analysis,
GraphPad PrismÒ 5.0 was used.
2.5. MALDI-TOF
The enzymatic reaction was carried out as described and
quenched by addition of 50% (v/v) of 1% TFA. The reaction mixture
was diluted 1:9 with a saturated solution of a-cyano-4-hydroxycin-
namic acid in 65% acetonitrile, in MilliQ water containing 0.1% TFA.
One milliliter was spotted onto the target and air dried for 15 min at
r.t. The MALDI-TOF MS was carried out on an Ultraflex TOF/TOF (Bru-
ker) operated in positive ion mode with an ion source voltage of
25 kV, a lens voltage of 7.5 kV and a reflector voltage of 26.3 kV.
The system was run in deflection mode with a mass suppression of
500 Da. The data analysis was carried out using the FlexAnalysis
software (Bruker). Baseline subtraction and smoothing of the curves
was applied.
Table 1
Kprime/Km, kcat and kcat/Km values of cc6A and cc4C for various H3K27me3 and H3K9me3 peptide substrates. In addition, IC50 values of four inhibitors are given. One S.D. is reported
for each value. Note that for the 24-meric peptide substrate, a Kprime value is listed instead of a Km value, due to negative cooperativity. N.D: Not detectable by FDH-coupled assay
or MALDI-TOF MS.
Substrate
Enzyme
Kprime/Km
(lM)
kcat (minÀ1
)
kcat/Km (minÀ1
l
MÀ1 Â 103)
H3(21–44)K27me3 24-meric
H3(20–34)K27me3 15-meric
H3(23–31)K27me3 9-meric
H3(7–14)K9me3 8-meric
cc6A
cc6A
cc6A
cc4C
139.5 15.1
495.4 34.7
N.D
0.19 0.04
0.29 0.01
N.D
1.36 0.29
0.58 0.05
N.D
23.9 3.9
0.37 0.06
15.5 3.5
Enzyme
Ligands
NOG
2,4-DCP
1
3
cc6A
cc4C
IC50
IC50
(
(
l
l
M)
M)
531 14
681 14
177 12
2.4 0.1
ꢀ1000
ꢀ625
281 15
135 16

1954
J.B.L. Kristensen et al. / FEBS Letters 585 (2011) 1951–1956
Fig. 3. Inhibition data of compound 3 at cc4C and representative MALDI-TOF MS spectra. (A) IC50 curve of 3. (B–E) MALDI-TOF MS spectra of 100
incubated with: (B) reaction buffer (negative control); (C) 5 M cc4C; (D) 5 M cc4C and 78 M 3; (E) 5 M cc4C and 1.25 mM 3. All intensities are in relative units. The
additional top at approximately 854 in (C–E) is a me2, Na adduct. Error bars represent one S.D.
lM 8-meric peptide
l
l
l
l
3.3. Inhibition studies on cc4C and cc6A
KDM4 selective inhibitor, when compared to factor-inhibiting hy-
poxia-inducible factor, another JmjC containing dioxygenase
For inhibitionstudies, the four compounds, NOG, 2,4-DCP, 1 and 3
were tested (Table 1). In the case of cc4C, all four tested ligands have
inhibitory effect, in agreement with previous findings for cc4A, cc4C
[8,22]. Similarly, we observed no inhibition of cc6A at 625 lM,
indicating selectivity of 3 towards KDM4C compared to KDM6A.
For compound 3, an IC50 curve and representative MALDI-TOF
MS spectra for the inhibition of KDM4C are shown in Fig. 3. Com-
parison of the structures of cc4A (in complex with 3 (PDB code
2WWJ), NOG and an 8-meric H3K9me3 substrate (PDB code
2OQ6)) and cc6B (PDB code 2XXZ), shows that the amino acids sur-
rounding 3 are different in cc4A (and cc4C) compared to cc6B (and
cc6A). For example, in the crystal structure of cc4A (Fig. 4A), 3
makes favorable van der Waals interactions with the surrounding
amino acids, with the benzyloxy phenyl group of 3 occupying posi-
tion 11 of the substrate [8]. In the cc6B structure, 3 does not fit well
into the pocket (Fig. 4B), when superimposing it to cc4A. Moreover,
in cc4A, a weak hydrogen bond between the oxygen atom of 3 and
the backbone amide nitrogen of Ala186 was reported [8]. This
interaction is not possible in cc6B and presumably also not in
cc6A, because Ala186 (cc4A) corresponds to Pro1385 (cc6B).
Taken together, we have shown that selective inhibition of cc4C,
and cc4E [7–9]. NOG, the inactive amide analog of
a-ketoglutaric
acid, inhibited both cc4C and cc6A with comparable IC50 values, sug-
gesting the same binding mode of NOG to cc4C and cc6A. In contrast,
the IC50 of 2,4-DCP was almost a factor 75 higher for cc6A (177
lM)
than for cc4C (2.4 M), demonstrating subtype selectivity. This dif-
l
ference in affinity may be explained by structural comparison of the
crystal structures of cc4A with 2,4-DCP bound [7] and cc6B (PDB
code 2XXZ, unpublished), a close homolog to cc6A (84% sequence
similarity of the cc region [21]). In cc4A (and cc4C), Lys241 interacts
with the carboxylic group in position 2 of 2,4-DCP (Fig. 4C). In cc6A
and cc6B, Lys241 corresponds to Asp1440. Furthermore, this residue
is positioned away from the cc, due to differences in a backbone loop
(Fig. 4C), causing a loss of interaction that most likely causes a signif-
icant decrease in affinity, as observed.
Compounds 1 and 3 inhibited cc4C in the
lM range (281 and
135 M, respectively). It has previously been shown that 3 is a
l
in the low lM range, by small molecule ligands is achievable.

J.B.L. Kristensen et al. / FEBS Letters 585 (2011) 1951–1956
1955
Fig. 4. Structural comparison of cc4A and cc6B. (A) Crystal structure of cc4A (PDB code 2WWJ [8]) in complex with 3 (cyan C atoms). (B) Crystal structure of cc6B (PDB code
2XXZ) superimposed with 2WWJ. (C) Ribbon representation of the crystal structure of cc4A in complex with 2,4-DCP (cyan C atoms) superimposed with 2XXZ. Gray surface
shows isosurface at À0.1 kcal/mol of the C3 probe, indicating favorable vdW interactions.
[4] Cloos, P.A.C., Christensen, J., Agger, K., Maiolica, A., Rappsilber, J., Antal, T.,
However, optimization of the scaffolds is necessary for the devel-
opment of new inhibitors exhibiting high affinity and selectivity
towards the KDM4 HDM subtypes. Such inhibitors may provide a
new strategy for future cancer treatment.
Hansen, K.H. and Helin, K. (2006) The putative oncogene GASC1 demethylates
tri- and dimethylated lysine 9 on histone H3. Nature 442, 307–311.
[5] Yang, Z.Q. et al. (2000) Identification of a novel gene, GASC1, within an
amplicon at 9p23-24 frequently detected in esophageal cancer cell lines.
Cancer Res. 60, 4735–4739.
[6] King, O.N.F. et al. (2010) Quantitative high-throughput screening identifies 8-
hydroxyquinolines as cell-active histone demethylase inhibitors. PLoS ONE 5,
e15535.
Acknowledgements
[7] Rose, N.R. et al. (2008) Inhibitor scaffolds for 2-oxoglutarate-dependent
histone lysine demethylases. J. Med. Chem. 51, 7053–7056.
[8] Rose, N.R. et al. (2010) Selective inhibitors of the JMJD2 histone demethylases:
combined nondenaturing mass spectrometric screening and crystallographic
approaches. J. Med. Chem. 53, 1810–1818.
[9] Hamada, S. et al. (2010) Design, synthesis, enzyme-inhibitory activity, and
effect on human cancer cells of a novel series of jumonji domain-containing
protein 2 histone demethylase inhibitors. J. Med. Chem. 53, 5629–5638.
[10] Chen, Z. et al. (2006) Structural insights into histone demethylation by JMJD2
family members. Cell 125, 691–702.
The authors wish to thank the Biotech Research and Innovation
Center (BRIC) for the cc4C construct, Oxford Protein Production
Facility for pOPIN vector F, and Novo Nordisk A/S for conducting
MALDI-TOF MS experiments and donation of synthetic peptide
substrates. The Lundbeck Foundation is acknowledged for financial
support to J.B.L.K. The University of Copenhagen programme of
excellence on epigenetics is acknowledged for financial support.
[11] Agger, K. et al. (2007) UTX and JMJD3 are histone H3K27 demethylases
involved in HOX gene regulation and development. Nature 449,
731–734.
Appendix A. Supplementary data
[12] Dalgliesh, G.L. et al. (2010) Systematic sequencing of renal carcinoma reveals
inactivation of histone modifying genes. Nature 463, 360–363.
[13] Terashima, M., Ishimura, A., Yoshida, M., Suzuki, Y., Sugano, S. and Suzuki, T.
(2010) The tumor suppressor Rb and its related Rbl2 genes are regulated by
Utx histone demethylase. Biochem. Biophys. Res. Commun. 399, 238–244.
[14] Tsai, M.-C., Wang, J.K. and Chang, H.Y. (2010) Tumor suppression by the
histone demethylase UTX. Cell Cycle (Georgetown, TX) 9, 2043–2044.
[15] van Haaften, G. et al. (2009) Somatic mutations of the histone H3K27
demethylase gene UTX in human cancer. Nat. Genet. 41, 521–523.
[16] Studier, F.W. (2005) Protein production by auto-induction in high density
shaking cultures. Protein Expr. Purif. 41, 207–234.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.febslet.2011.05.023.
References
[1] Mosammaparast, N. and Shi, Y. (2010) Reversal of histone methylation:
biochemical and molecular mechanisms of histone demethylases. Annu. Rev.
Biochem. 79, 155–179.
[2] Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J.R., Cole, P.A., Casero, R.A.
and Shi, Y. (2004) Histone demethylation mediated by the nuclear amine
oxidase homolog LSD1. Cell 119, 941–953.
[17] Otake, Y. and Izawa, K. (2005) Process for producing aminoepoxide. Patent US
2005/0137408 A1, Ajinomoto Co. Inc., Japan.
[3] Tsukada, Y., Fang, J., Erdjument-bromage, H., Warren, M.E., Borchers, C.H.,
Tempst, P. and Zhang, Y. (2006) Histone demethylation by a family of JmjC
domain-containing proteins. Nature 439, 811–816.
[18] Cunliffe, C.J., Franklin, T.J., Hales, N.J. and Hill, G.B. (1992) Novel inhibitors of
prolyl 4-hydroxylase. 3. Inhibition by the substrate analogue N-oxaloglycine
and its derivatives. J. Med. Chem. 35, 2652–2658.

1956
J.B.L. Kristensen et al. / FEBS Letters 585 (2011) 1951–1956
[19] Couture, J.F., Collazo, E., Ortiz-Tello, P.A., Brunzelle, J.S. and Trievel, R.C. (2007)
Specificity and mechanism of JMJD2A, trimethyllysine-specific histone
demethylase. Nat. Struct. Mol. Biol. 14, 689–695.
[20] Goodford, P.J. (1985) A computational procedure for determining energetically
favorable binding sites on biologically important macromolecules. J. Med.
Chem. 28, 849–857.
[21] Hong, S., Cho, Y.-W., Yu, L.-R., Yu, H., Veenstra, T.D. and Ge, K. (2007)
Identification of JmjC domain-containing UTX and JMJD3 as histone H3 lysine
27 demethylases. Proc. Natl. Acad. Sci. USA 104, 18439–18444.
[22] Shi, Y. and Whetstine, J.R. (2007) Dynamic regulation of histone lysine
methylation by demethylases. Mol. Cell 25, 1–14.
a