Monitoring the activity of 2-oxoglutarate dependent histone demethylases by NMR spectroscopy: Direct observation of formaldehyde

Article abstract of DOI:10.1002/cbic.200900713

(Chemical Equation Presented) Formaldetected: Application of ¹H- and 1D ¹³C HSQC NMR spectroscopy methods for monitoring the activity of a 2-oxoglutarate dependent histone N^ε-emethyllysine demethylase (JMJD2E) enabled the

Full text of DOI:10.1002/cbic.200900713

DOI: 10.1002/cbic.200900713
Monitoring the Activity of 2-Oxoglutarate Dependent Histone
Demethylases by NMR Spectroscopy: Direct Observation of Formaldehyde
[
a]
Richard J. Hopkinson, Refaat B. Hamed, Nathan R. Rose, Timothy D. W. Claridge, and Christopher J. Schofield*
Ferrous iron and 2-oxoglutarate (2OG) dependent oxygenases
are a diverse superfamily, with members involved in many im-
portant biological processes, including oxygen sensing, epige-
netic regulation, and collagen, antibiotic, and fatty acid biosyn-
[
1,2]
thesis.
The 2OG oxygenase histone demethylase (HDM) sub-
e
family catalyses the demethylation of N -methylated lysine resi-
dues in the N-terminal tails of histones. Methylated histone tail
lysines are involved in the establishment of different chromatin
[3]
states, and contribute to both gene silencing and activation.
Various HDM subfamilies have been implicated in disease
states, with the JMJD2 HDMs being linked to prostate and
[
4]
oesophageal cancers.
e
N -Methyllysine demethylation is proposed to occur via hy-
e
droxylation of the N -methylated lysine (with concomitant oxi-
dation of the cosubstrate 2OG, and decarboxylation to give
carbon dioxide and succinate), followed by fragmentation of
e
the N -hydroxymethyllysine, to give formaldehyde and the de-
[
5]
II
methylated lysine (Scheme 1). The Fe /2OG-dependent oxy-
genases can be challenging to study in detail, with three sub-
strates (“prime” substrate, 2OG and oxygen) and at least three
products (in the case of the histone demethylases, four: deme-
thylated peptide, succinate, formaldehyde and carbon dioxide).
NMR spectroscopy is a potentially useful technique for study-
ing 2OG oxygenase reactions in a single assay mixture. Previ-
ous reports regarding quantification of formaldehyde released
from biocatalysed reactions or formaldehyde levels in biologi-
cal systems have been based either on its oxidation to formic
[
6]
acid, or on its derivatisation with reagents, such as dimedone
[
7–10]
Scheme 1. Proposed mechanism for JMJD2-catalysed demethylation. Oxida-
tive decarboxylation of 2OG generates a Fe =O species, which reacts with
the methylated lysine residue. Hydroxymethyllysine then fragments to give
formaldehyde and the demethylated product. The demethylated carbon is
highlighted in bold; R=H or CH3.
or ampicillin.
To our knowledge, formaldehyde detection
IV
by NMR spectroscopy in enzyme-catalysed reactions has not
been reported. Here, we report the use of NMR spectroscopy
for monitoring N-demethylation by JMJD2E (a HDM that is
sufficiently active for kinetic studies) by monitoring 2OG con-
e
e
version to succinate, demethylation of N -trimethyl and N -di-
methyllysine residues, and both direct and indirect detection
of formaldehyde production in demethylation reactions.
Gly-Lys) in order to allow sufficient substrate recognition by
1
Initially, the JMJD2E catalysed demethylation reaction was
investigated by monitoring the demethylation of octapeptide
fragments of the histone H3 N-terminal tail (residues Ala7 to
the enzyme whilst reducing signal overlap in the H NMR spec-
tra. A standardised demethylation reaction protocol was devel-
oped, firstly by screening a variety of buffers suitable for
e
1
Lys14, N -methylated at residue Lys9). A peptide length of
H NMR spectroscopy (nondeuterated potassium phosphate
eight amino acids was selected (Ala-Arg-lys(Me )-Ser-Thr-Gly-
and ammonium formate buffers, both at 50 mm, pH 7.5) and
then by optimising concentrations of reagents to allow NMR
detection. JMJD2E only displayed sufficient activity in ammoni-
um formate buffer, and this was selected for further work. l-As-
corbate has been previously shown to increase activity of
3
+
+
+
[
a] R. J. Hopkinson, R. B. Hamed, N. R. Rose, Dr. T. D. W. Claridge,
Prof. C. J. Schofield
Department of Chemistry, University of Oxford
1
2 Mansfield Road, Oxford, OX1 3TA (UK)
II
some Fe /2OG-dependent oxygenases and was therefore in-
Fax: (+44)1865-285002
[11]
II
cluded in the assay mixture. The presence of Fe ions at
100 mm did not cause any noticeable loss of resolution. The
demethylation reactions with tri- and dimethylated peptides
(K9me3 and K9me2, respectively) under optimised conditions
E-mail: christopher.schofield@chem.ox.ac.uk
+
[
] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.200900713.
5
06
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 506 – 510

1
were then monitored by H NMR spectroscopy (700 MHz) over
0 min, at 75 s intervals, and experiments were carried out
preferred to the dimethylated form, and imply that this selec-
tivity arises, predominantly, from a difference in KM.
3
with varying the concentration of peptide substrate, with an
excess of 2OG. The relative concentrations of peptide substrate
and demethylated products were calculated at each time point
The NMR spectroscopy method was then used to investigate
how succinate and demethylated peptide product concentra-
tions varied with time. K9me3 and K9me2 were incubated with
JMJD2E by using the described protocol and were analysed at
75 s intervals over a period of 15 min. The concentrations of
demethylated products (summed in the case of K9me3 deme-
thylation) and succinate were then compared at each time
point. In both cases, concentrations grew in a linear relation-
ship relative to each other over all time points (Figure 1C) with
an excess of succinate to demethylated peptide, indicating
that 2OG turnover is largely, but not absolutely, coupled to
substrate demethylation. It is possible that the extent of cou-
pling is condition or substrate dependent, as observed for
e
1
by integration of the corresponding N -CH3 H resonances, and
normalising each value for the number of protons and initial
substrate concentration. Kinetic parameters were then calculat-
ed by analysing the initial rates of demethylation at different
substrate concentrations. The KM for the Lys9 trimethylated
peptide K9me3 (203Æ79 mm) was found to be lower than that
for the analogous dimethylated peptide K9me2 (282Æ36 mm).
À1
However, the V
(0.180Æ0.018 and 0.196Æ0.008 mms for
max
K9me3 and K9me2, respectively) and k values (0.018Æ0.002
cat
À1
and 0.020Æ0.001 s for K9me3 and K9me2, respectively) for
[1]
both substrates had higher similarity. The k values that we
other 2OG oxygenases. The proportion of “uncoupled” turn-
over remained constant for each substrate over time (24.8Æ
1.3 and 16.2Æ0.5% for K9me3 and K9me2, respectively, rela-
tive to demethylated product concentration), implying that it
did not arise at a time point prior to the first NMR analysis
(150 s after mixing).
cat
obtained for JMJD2E are substantially higher than those previ-
[
6]
ously reported for JMJD2A/D, although this might reflect dif-
ferences in assay conditions. Our results are consistent with
[
6,12]
prior studies
conducted with a formaldehyde dehydrogen-
ase-coupled assay, which indicate that trimethylated H3K9 is
1
Figure 1. Monitoring JMJD2E-catalysed demethylation over time. A) H NMR spectra of a time course monitoring the JMJD2E-catalysed demethylation of
K9me3. Spectra are shown at 300 s intervals (acquisition time for each spectrum was 75 s). The first acquisition was started 150 s after mixing the assay
components. The initial peptide concentration was 1 mm. B) Concentrations of peptide substrate, 2OG, demethylated products and succinate during a time
course monitoring the JMJD2E-catalysed demethylation of K9me3. C) Demethylated peptide product concentration plotted against succinate concentration
at time points during JMJD2E-catalysed demethylation of K9me3 and K9me2. The initial peptide concentrations were 0.75 mm.
ChemBioChem 2010, 11, 506 – 510
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
507

The NMR assay was then used to investigate the effect of l-
ascorbate on 2OG oxidation/decarboxylation and peptide de-
methylation. The level of substrate-uncoupled 2OG turnover is
II
known to vary significantly between members of the Fe /2OG-
[
1]
dependent oxygenase family, and in some cases to vary in
[
11]
the presence of ascorbate. Our experiments were carried out
e
with K9me3, K9me2, N -monomethylated K9me1, unmethylat-
ed K9me0, and in the absence of any peptide substrate. Exper-
iments were carried out with saturating concentrations of
peptide and 2OG (JMJD2E:K9me0/1/2/3:2OG, 10 mm:750 mm:
5
mm); initial rates were measured for peptide demethylation
(where applicable) and 2OG conversion to succinate, in the
presence (1 mm) or in the absence of l-ascorbate (Figure 2). In
the presence of K9me2/me3 peptides, ascorbate stimulated ac-
tivity of both peptide demethylation and 2OG turnover. Ascor-
bate also accelerated 2OG turnover in assays with K9me1. In
the absence of peptide and in the presence of K9me0, the dif-
ferences with and without ascorbate were within overall exper-
imental error. The levels of stimulation of K9me3 and K9me2
demethylation were significant, but less than observed for
some other 2OG oxygenases (e.g., collagen prolyl 4-hydroxy-
Figure 2. Rates of peptide demethylation (where applicable) and 2OG turn-
over (succinate formation) in the presence (+Asc) and absence (ÀAsc) of
ascorbate. Error bars are displayed as standard deviations.
To directly detect free formaldehyde formed by demethyla-
13
13
tion of methylated lysine, C-labelled K9me2 (K9 Cme2) was
synthesised and its reaction with JMJD2E was analysed by 1D
[
11]
lase). While there is evidence to suggest that longer peptide
fragments of K9-monomethylated H3 are substrates for
JMJD2E (N. R. Rose, unpublished data) we did not observe
demethylation of K9me1 by our NMR assay, and the extent of
13
C heteronuclear single quantum coherence (HSQC) NMR
spectroscopy (Figure 3). The HSQC experiment benefits from
the greater sensitivity of proton observation and also provides
1
2
OG turnover was not above that observed in the absence of
peptide.
We then investigated the de-
selectivity by editing the 1D H spectrum to retain responses
tection of formaldehyde by NMR
spectroscopy. Formaldehyde was
not unambiguously observed in
1
the H NMR spectra, probably
due partly to its low concentra-
tion relative to the other re-
agents in solution, and also be-
cause of overlap of the signal
with the solvent resonance
(HDO). Thus, dimedone (5,5-di-
methyl-cyclohexane-1,3-dione)
was added to the reaction mix-
ture to capture formaldehyde
1
and allow its detection by
H
NMR spectroscopy. Dimedone
reacts with formaldehyde
Scheme S2 in the Supporting In-
(
formation) to form two different
adducts, both of which have dis-
tinctive proton NMR chemical
shifts (Figure S8 in the Support-
ing Information). Both of these
adducts were observed for form-
aldehyde produced by JMJD2E
1
3
1
catalysis (Figure S9 in the Sup- Figure 3. Direct observation of formaldehyde production by JMJD2E catalysis. A)–D) 1D C HSQC, and E) H NMR
1
3
1
13
spectra show production of C-formaldehyde ( JCH =168 Hz) from the demethylation of K9 Cme2. A) and B) De-
methylation and C-formaldehyde production at 25 and 48C, respectively. C) Reaction mixture enriched with au-
thentic C-formaldehyde. D) Authentic C-formaldehyde. E) H NMR spectrum of JMJD2E-catalysed demethylation
of K9 Cme2. Note: The small coupling observed in the K9 Cme2 methyl resonance arises from a three-bond H–
porting Information) with the
mono” adduct (2-hydroxyme-
1
3
“
1
3
13
1
thyldimedone) being the major
product observed.
13
13
1
1
3
3
C coupling to the adjacent methyl carbon ( JCH =4 Hz).
5
08
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 506 – 510

1
3
only from those protons bound directly to C. Thus, only spe-
(97.5%/2.5%, w/v) and purified to >95% purity by HPLC. Details
13
13
e
on the synthesis of C-labelled N -dimethyllysine and full charac-
cies containing the C label originating from the labelled
methyl groups were observed, with all other resonances being
1
terisation of peptides by H NMR spectroscopy are described in the
13
Supporting Information.
eliminated (down to the level of natural C abundance, 1.1%).
13
Owing to the removal of C decoupling in the 1D HSQC se-
quence employed, each resonance displays a distinctive dou-
NMR spectroscopy: NMR analyses employed a Bruker AVIII 700
spectrometer equipped with an inverse TCI cryoprobe optimised
1
1
13
1
for H observation and installed with TOPSPIN 2 software. Chemical
blet structure arising from the one-bond H– C coupling ( J ).
CH
13
shifts are reported in ppm relative to D O (d =4.72 ppm); the deu-
2 H
C-Formaldehyde was observed in its hydrated form CH (OH)
2
2
terium signal was used as an internal lock signal and the HDO
signal was reduced by presaturation where necessary. The spec-
trometer conditions were optimised by using a control sample
with all the components of the reaction except JMJD2E. The ex-
periments were performed on an identical sample; the enzyme
was added to the assay mixture directly prior to transfer to a 2 mm
NMR tube. The NMR spectroscopy tube was centrifuged for a few
seconds in a hand centrifuge. The sample was introduced to the
magnet and data acquisition was started after a brief optimisation
1
(
d =4.72 ppm, J =168 Hz) as expected under these solution
H
CH
conditions (Figure 3, spectra A, B and E). Varying the tempera-
ture at which the NMR spectroscopy experiments were carried
out caused a relative shift in the HDO resonance, allowing
13
both resonances for the C-formaldehyde to be observed
Figure 3, spectrum B). A 2D HSQC experiment performed on
the final reaction mixture (Figure S10 in the Supporting Infor-
(
13
mation) gave the C shift of the formaldehyde as 82 ppm,
which was consistent with an authentic sample. Enriching the
(total time lapse between adding the enzyme and the start of data
acquisition was 150 s). The time course data were collected by
using an automated routine. Twenty four analyses were performed
on each sample, each accumulating 16 transients corresponding to
75 s of total acquisition time and providing a single spectrum. The
delay time between analyses was 0 s. The sample temperature was
maintained at 298 K throughout the run. Data were processed by
using automated routines and spectra integrated with absolute
intensity scaling to monitor changes in intensity of signals of inter-
13
reaction mixture with an authentic sample of C-formaldehyde
confirmed that the chemical shifts observed in the enzymatic
reaction mixture corresponded to formaldehyde (Figure 3,
spectrum C). The putative hydroxymethyllysine intermediate
(
Scheme 1) was not observed, implying that this intermediate
is either enzyme-bound or too short-lived to be observed on
the NMR timescales used here.
1
est. H NMR spectra of the substrate peptides gave signal intensi-
Overall, in situ NMR analyses of the JMJD2E reaction with a
model substrate have provided information on the stoichiome-
try of its reactions and kinetic data. Formaldehyde was detect-
ed as a reaction product, both through its derivatisation with
dimedone, and as free formaldehyde in solution. To our knowl-
edge, this is the first reported instance of direct detection by
NMR spectroscopy of enzymatically produced formaldehyde.
The NMR methods described could be useful in investigating
ties consistent with their predicted relative values.
1
3
13
1
D
C HSQC NMR spectroscopy: Demethylation of the C-la-
13
belled peptide (K9 Cme2) was followed by using a gradient-select-
ed 1D C HSQC method in which the standard 2D HSQC sequence
13
1
3
was modified by removal of both the variable t period and C de-
1
coupling during data acquisition. The 1/2J delays were optimised
CH
1
for J of 145 Hz. For each 1D experiment, eight transients were
CH
accumulated corresponding to 39 s of total acquisition time. The
experiments were performed at two different temperatures (298
II
the stoichiometry and mechanisms of other Fe /2OG-depen-
13
dent oxygenases; when appropriate spectrometers are avail-
able they provide a useful alternative to assays based on la-
belled 2OG, or chromatography, which have been widely used
in the field. In the field of histone modifying and related en-
zymes, functional assignments are commonly made by mass
spectrometric analyses, which can be difficult and, at the bio-
chemical level, are complicated by redundancy issues. We be-
lieve that, whenever possible, NMR spectroscopy should be
and 277 K) in order to clearly see the two C-coupled signals for
formaldehyde and avoid the interference from the HDO signal.
Acknowledgements
We thank the Wellcome Trust, BBSRC, the Commonwealth Schol-
arship Commision, and the Ministry of Higher Education, Egypt,
for funding the work. Help with peptide characterisation by Phil-
ippa Barlow is gratefully acknowledged. We also thank Prof. Udo
Oppermann and Dr. Stanley Ng for encouragement.
[13]
used in such assignments, as reported recently for JMJD6.
Experimental Section
Keywords: enzyme catalysis
·
formaldehyde
·
histone
Expression and purification of the histone demethylase,
JMJD2E: The catalytic domain of human JMJD2E (residues 1–337)
demethylases · NMR spectroscopy · transcriptional regulation
was produced as an N-terminally His -tagged protein in E. coli, and
6
[1] R. P. Hausinger, Crit. Rev. Biochem. Mol. Biol. 2004, 39, 21–68.
[2] C. Loenarz, C. J. Schofield, Nat. Chem. Biol. 2008, 4, 152–156.
[3] R. J. Klose, Y. Zhang, Nat. Rev. Mol. Cell Biol. 2007, 8, 307–318.
[4] K. Agger, J. Christensen, P. A. Cloos, K. Helin, Curr. Opin. Genet. Dev.
purified by Ni-affinity and size-exclusion chromatography, and
À1
stored at a concentration of 60 mgmL in HEPES (50 mm), NaCl
[14]
(
500 mm), pH 7.5, as reported.
2
008, 18, 159–168.
Synthesis of peptides: Synthesis of peptide substrates was carried
out by using standard Fmoc-based solid phase peptide synthesis
[
5] S. Ng, K. Kavanagh, M. McDonough, D. Butler, E. Pilka, B. Lienard, J.
Bray, P. Savitsky, O. Gileadi, F. von Delft, Nature 2007, 448, 87–91.
6] J. F. Couture, E. Collazo, P. A. Ortiz-Tello, J. S. Brunzelle, R. C. Trievel, Nat.
Struct. Mol. Biol. 2007, 14, 689–695.
(
SPPS) with a CSPep336X peptide synthesiser (CSBio, California,
[
USA). Peptides were synthesised on a PL-AMS (aminomethylpolys-
tyrene) resin (Polymer Labs) by using a Rink amide linker, and
[7] ꢁ. Sꢂrdi, E. Tyihꢂk, Biomed. Chromatogr. 1994, 8, 313–314.
cleaved from the resin by using CF COOH/triisopropylsilane
[8] E. Mꢂday, E. Tyihꢂk, ꢁ. Szçke, Plant Growth Regul. 2000, 30, 105–110.
3
ChemBioChem 2010, 11, 506 – 510
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
509

[
9] W. Luo, H. Li, Y. Zhang, C. Y. W. Ang, J. Chromatogr. B: Biomed. Sci. Appl.
A. Lengeling, M. Mann, N. J. Proudfoot, C. J. Schofield, A. Bçttger, Sci-
ence 2009, 325, 90–93.
2001, 753, 253–257.
[
[
[
10] Y. Shi, F. Lan, C. Matson, P. Mulligan, J. R. Whetstine, P. A. Cole, R. A.
Casero, Y. Shi, Cell 2004, 119, 941–953.
11] R. Myllyla, K. Majamaa, V. Gunzler, H. Hanauske-Abel, K. Kivirikko, J. Biol.
Chem. 1984, 259, 5403–5405.
12] M. Sakurai, N. Rose, L. Schultz, A. Quinn, A. Jadhav, S. Ng, U. Opper-
mann, C. Schofield, A. Simeonov, Mol. BioSyst. 2010, DOI: 10.1039/
b9192993f.
[14] N. R. Rose, S. S. Ng, J. Mecinovic, B. M. Lienard, S. H. Bello, Z. Sun, M. A.
McDonough, U. Oppermann, C. J. Schofield, J. Med. Chem. 2008, 51,
7053–7056.
[13] C. J. Webby, A. Wolf, N. Gromak, M. Dreger, H. Kramer, B. Kessler, M. L.
Received: November 21, 2009
Nielsen, C. Schmitz, D. S. Butler, J. R. Yates III, C. M. Delahunty, P. Hahn,
Published online on January 21, 2010
5
10
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 506 – 510