NMR analyses on N-hydroxymethylated nucleobases-implications for formaldehyde toxicity and nucleic acid demethylases

Article abstract of DOI:10.1039/c8ob00734a

Formaldehyde is produced in cells by enzyme-catalysed demethylation reactions, including those occurring on N-methylated nucleic acids. Formaldehyde reacts with nucleobases to form N-hydroxymethylated adducts that may contribute to its toxicity/carcinogenicity when added exogenously, but the chemistry of these reactions has been incompletely defined. We report NMR studies on the reactions of formaldehyde with canonical/modified nucleobases. The results reveal that hydroxymethyl hemiaminals on endocyclic nitrogens, as observed with thymidine and uridine monophosphates, are faster to form than equivalent hemiaminals on exocyclic nitrogens; however, the exocyclic adducts, as formed with adenine, guanine and cytosine, are more stable in solution. Nucleic acid demethylase (FTO)-catalysed hydroxylation of (6-methyl)adenosine results in (6-hydroxymethyl)adenosine as the major observed product; by contrast no evidence for a stable 3-hydroxymethyl adduct was accrued with FTO-catalysed oxidation of (3-methyl)thymidine. Collectively, our results imply N-hydroxymethyled adducts of nucleic acid bases, formed either by reactions with formaldehyde or via demethylase catalysis, have substantially different stabilities, with some being sufficiently stable to have functional roles in disease or the regulation of nucleic acid/nucleobase activity.

Full text of DOI:10.1039/c8ob00734a

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
NMR analyses on N-hydroxymethylated
nucleobases – implications for formaldehyde
toxicity and nucleic acid demethylases†
Cite this: DOI: 10.1039/c8ob00734a
a
S. Shishodia, a D. Zhang,a A. H. El-Sagheer, a,c T. Brown,
T. D. W. Claridge, a C. J. Schofield a and R. J. Hopkinson
*
a,b
Formaldehyde is produced in cells by enzyme-catalysed demethylation reactions, including those occur-
ring on N-methylated nucleic acids. Formaldehyde reacts with nucleobases to form
N-hydroxymethylated adducts that may contribute to its toxicity/carcinogenicity when added exogen-
ously, but the chemistry of these reactions has been incompletely defined. We report NMR studies on the
reactions of formaldehyde with canonical/modified nucleobases. The results reveal that hydroxymethyl
hemiaminals on endocyclic nitrogens, as observed with thymidine and uridine monophosphates, are
faster to form than equivalent hemiaminals on exocyclic nitrogens; however, the exocyclic adducts, as
formed with adenine, guanine and cytosine, are more stable in solution. Nucleic acid demethylase (FTO)-
catalysed hydroxylation of (6-methyl)adenosine results in (6-hydroxymethyl)adenosine as the major
observed product; by contrast no evidence for a stable 3-hydroxymethyl adduct was accrued with FTO-
catalysed oxidation of (3-methyl)thymidine. Collectively, our results imply N-hydroxymethyled adducts of
nucleic acid bases, formed either by reactions with formaldehyde or via demethylase catalysis, have sub-
stantially different stabilities, with some being sufficiently stable to have functional roles in disease or the
regulation of nucleic acid/nucleobase activity.
Received 28th March 2018,
Accepted 29th March 2018
DOI: 10.1039/c8ob00734a
rsc.li/obc
produced by the action of N-methyl group demethylases acting
Introduction
in both DNA repair and chromatin regulation.17 Endogenous
Formaldehyde, the simplest aldehyde, exists principally in its HCHO levels are difficult to quantify, although reports esti-
hydrated form in water and is toxic and carcinogenic to mate concentrations in the mid to high micromolar
humans above threshold levels.1 Acute exposure to HCHO can range;12,15,18–21 there is also potential for relatively high
lead to pulmonary oedema,2 renal failure3 and inflammation,4 localised HCHO concentrations, e.g. in cell nuclei, where
while chronic HCHO exposure correlates with increased N-methyl demethylases are present. The physiological impli-
incidence of cancer.5 HCHO toxicity/carcinogenicity is likely a cations of endogenous HCHO production are poorly appreci-
consequence of its reactions with biomolecules; HCHO is ated, although is emerging evidence that it is a carbon source
reported to cross-link DNA,6 while reactions with some pro- for C1 metabolism.15 Cellular and mouse studies have also
teins are proposed to induce sensory responses7,8 and protein revealed a correlation between endogenous HCHO and toxicity
aggregation.9,10
in Fanconi Anaemia.22,23 HCHO is also proposed as a prebiotic
In addition to exogenous sources, e.g. from plastics,1 precursor.24,25
food11,12 and cosmetics,13,14 HCHO is produced endogenously
Defining the chemistry underpinning HCHO-driven toxi-
during folate oxidation15 and by the action of enzymes includ- city/carcinogenicity is challenging due to the likely reversible
ing semicarbazide-sensitive amine oxidase.16 HCHO is also nature of many HCHO-biomolecule adducts.26,27 Hence it is
important to define in detail the nature of HCHO adducts with
isolated components, including nucleobases. Pioneering work
aChemistry Research Laboratory, 12 Mansfield Road, Oxford, OX1 3TA, UK
by McGhee and von Hippel defined hemiaminal adducts pro-
bLeicester Institute of Structural and Chemical Biology and Department of Chemistry, duced by reaction of the canonical nucleobases with HCHO
University of Leicester, Henry Wellcome Building, Lancaster Road, Leicester, LE1
and revealed the potential of HCHO adducts to modulate base-
pairing (hydroxymethylated adenine can base-pair with
7RH, UK. E-mail: richard.hopkinson@leicester.ac.uk
cChemistry Branch, Department of Science and Mathematics, Faculty of Petroleum
thymine, but in a less stable manner than unmodified
and Mining Engineering, Suez University, 43721 Suez, Egypt
adenine).28,29 The exocyclic amino groups of adenine, guanine
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c8ob00734a
and cytosine react to give hemiaminal adducts, as do the endo-
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 1 Formation of N-hydroxymethyl adducts on nucleic acids can occur via both enzymatic and non-enzymatic pathways. The nucleophilic nitro-
gens of nucleic acid bases react in non-enzyme-catalysed reactions with electrophiles such as formaldehyde (forming N-hydroxymethyl adducts)
and methylating agents. N-Methylation is also catalysed at some sites by methyltransferases. Although not reported, there is potential for enzyme-
catalysed reactions of nucleobases and formaldehyde. Many N-methylated nucleobases are substrates for 2-oxoglutarate (2OG)-, oxygen-, and
ferrous iron-dependent nucleic acid demethylases, which catalyse demethylation via an oxidative mechanism involving formation of
N-hydroxymethyl intermediates, which in some cases could be considered as products. As shown in the present work, there are considerable vari-
ations in the stabilities of the hydroxymethyl adducts.
cyclic imino groups of thymine and uracil. These findings are the demethylated product and HCHO. These observations
consistent with mass spectrometric (MS) studies on oligo- reveal the potential for site-specific accumulation of
nucleotides, which identified nucleobase N-methylation on N-hydroxymethyl adducts in nucleic acids. The combined
treatment with HCHO and NaCNBH3.30,31 Cross-linking studies suggest different hemiaminal adducts predominate
between nucleobases after treatment of DNA with HCHO has under different conditions, and imply that the outcomes of
also been observed;32,33 recent studies by Swenberg and co- nucleic acid N-methyl group oxidation may be more complex
workers have supplied evidence for cross-linked adducts than previously perceived.
between guanine and nucleophilic amino acids, involving
aminal and hemithioaminal linkages.34–36 Interestingly, many
of the observed nucleobase adducts are structurally analogous
Results and discussion
HCHO reactions with canonical nucleotides
to the proposed hemiaminal intermediates produced during
enzyme-catalysed N-methyl group demethylation (Fig. 1).17
We are interested in understanding how the physiological Initially, we incubated excess HCHO (53-fold) with the five
effects of HCHO arise from its reactions, and in testing the canonical DNA/RNA nucleotide monophosphates deoxyadeno-
hypothesis that nucleic acid-HCHO adducts have functional sine monophosphate (dAMP), deoxycytidine monophosphate
roles in biology and disease. We report studies on the reactions (dCMP), deoxyguanosine monophosphate (dGMP), thymidine
of HCHO with both canonical and non-canonical nucleotides/ monophosphate (TMP), and uridine monophosphate (UMP, all
nucleosides, some which have emerged since the work of at 2.4 mM) at pD 6, pD 7.5 and pD 9, monitoring reaction over
McGhee and von Hippel,28,37 using NMR to determine their 5–8 hours by H NMR (Fig. 2A, S1–S3, S5 and S6†). A sample
1
relative rates of formation and stability. To highlight the rele- simultaneously containing all five nucleotides (each at 2.2 mM)
vance of the results to biology, we also profiled the reactions of and HCHO (58 equivalents) was also analysed (800 minutes at
N-methylated nucleosides with the human nucleic acid de- pD 7.5) to robustly compare their relative reaction rates (Fig. 2B).
methylase FTO. Studies on FTO catalysis reveal that the stabi-
As reported by McGhee and von Hippel, we observed reac-
lities of the hemiaminal products depend on the position of tions between HCHO and all five nucleotides (Scheme 2 and
N-methyl group on the substrate; the major product of FTO- Fig. 2). With TMP, the major product was observed at the first
catalysed oxidation of (6-methyl)adenosine was the hemiam- time-point (7 minutes after mixing, Fig. 2A and Fig. S1†) and
inal, whereas oxidation of (3-methyl)thymidine revealed only was present at constant concentration during the analysis
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 2 Canonical nucleotides react with HCHO at either exo- or endocyclic nitrogens and at different rates to form N-hydroxymethyl adducts. (A)
1H NMR time course analyses of incubations of canonical nucleotides with HCHO. Five canonical nucleotide monophosphates (dAMP, dCMP, dGMP,
TMP, and UMP) were incubated with 53 eq. of HCHO in D2O at pD 7.5, and the reactions were monitored over time using 1H NMR (final nucleotide
concentration = 2.4 mM). The pyrimidines TMP and UMP react via their endocyclic N-3 nitrogens to form 3hmTMP and 3hmUMP, respectively (blue).
dAMP, dCMP and dGMP react via their exocyclic N-6, N-4 and N-2 nitrogens respectively to form 6hmdAMP, 4hmdCMP, and 2hmdGMP as major
products (blue). Under the tested conditions, evidence for low-level formation of the dihydroxymethylated adducts 6dhmdAMP and 2dhmdGMP
was also accrued (green). (B) 1H NMR time course analysis of a reaction mixture containing dAMP, dCMP, dGMP, TMP and UMP (all at 2.2 mM) and
HCHO (58 eq.) in D2O at pD 7.5. The fastest adducts to form were 3hmTMP and 3hmUMP, formed via reactions of HCHO with TMP and UMP
respectively. The other observed adducts (6hmdAMP, 4hmdCMP and 2hmdGMP), formed via reactions of HCHO with exocyclic nitrogens on dAMP,
dCMP and dGMP respectively, were slower to form, but showed time-dependent increases in concentration. Of these three products, 4hmdCMP
was fastest to form, followed by 2hmdGMP and 6hmdAMP respectively. Only the major adducts for each nucleotide are shown. (C) 1H NMR time
course analyses of incubations of dAMP, dCMP and dGMP with HCHO, after two weeks reaction and subsequent dilution with D2O (25-fold). In all
cases, the concentrations of the N-hydroxymethyl adducts decreased in a time-dependent manner after dilution, revealing non-immediate adjust-
ments to new equilibrium positions. Of the three nucleotides, degradation of adducts derived from dAMP (6hmdAMP and 6dhmdAMP) were the
slowest to degrade, revealing their relative stabilities compared to the other adducts. Upon dilution, 3hmTMP and 3hmUMP immediately decreased
in concentration, indicating a fast dynamic equilibrium between the adducts and free nucleotides (and HCHO).
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
(0.4 mM at pD 7.5 over 6 hours, Fig. 2A). The product was 360 minutes at pD 9, formation rate = 0.117 µM s−1
,
characterised as (3-hydroxymethyl)thymidine monophosphate Fig. S5C†). 1H–13C-HMBC analyses revealed correlations
(3hmTMP), on the basis of 1H and 13C NMR assignments, indi- between this resonance and both the C-2 carbon (at δC
cating that HCHO preferentially reacts on the N-3 nitrogen of 154 ppm) and to itself (at δC 72 ppm); the species was therefore
TMP. While this endocyclic product is rapidly produced, its (provisionally) assigned as the N2-dihydroxymethylated adduct
concentration (0.4 mM at pD 7.5) was low relative to unreacted (2,2-dihydroxymethyl)deoxyguanosine
monophosphate
TMP (2 mM at pD 7.5). Formation of 3hmTMP was too fast to (2dhmdGMP). As with dCMP, no significant changes in adduct
determine its initial formation rates at pDs 7.5 and 9; however, formation rates were observed across the 3 pD’s tested (for-
the reaction was slower at pD 6 (3hmTMP formation rate = mation rates for 2hmdGMP = 0.1 µM s−1 and 0.117 µM s−1 at
0.3 µM s−1). On dilution of the reaction mixtures with D2O pDs 6 and 7.5 respectively, formation rates for 2dhmdGMP
(2- to 25-fold), the concentration of 3hmTMP rapidly decreased were too low to be accurately determined at pDs 6 and 7.5,
from the first time-point to reform TMP, implying 3hmTMP is Fig. 2A and Fig. S5B/C†). On dilution of the reaction mixtures
in dynamic equilibrium with TMP. Rapid fragmentation of with D2O (2- to 25-fold), both adducts degraded to free HCHO
3hmTMP to TMP was observed on addition of the HCHO and dGMP; the degradation rates of 2hmdGMP and
‘scavenger’ dimedone.39 Identical behaviour (within our limits 2dhmdGMP (0.033 µM s−1 and 0.003 µM s−1 respectively at pD
of detection) was observed with mixtures of HCHO and the 7.5 after 25-fold dilution) were comparable to those observed
RNA nucleotide uridine monosphosphate (UMP), also forming for 4hmdCMP and 4dhmdCMP (Fig. 2C and Fig. S4E/F†).
the endocyclic adduct (3-hydroxymethyl)uridine monophos-
dAMP preferentially reacted with HCHO via its exocyclic N-6
nitrogen, but reacted more slowly than the other canonical
nucleotides (Fig. 2A/B and Fig. S6†). One major and one minor
phate (3hmUMP, formation rate at pD 6 = 0.783 µM s−1
,
Fig. 2A and Fig. S2†).
The other canonical nucleotides reacted with HCHO via product were detected; the major product, (6-hydroxymethyl)
their exocyclic nitrogens (Scheme 1). dCMP preferentially deoxyadenosine monophosphate (6hmdAMP), was observed to
reacted with HCHO via its exocyclic N-4 nitrogen forming increase in concentration throughout the analyses, reaching a
(4-hydroxymethyl)deoxycytidine monophosphate (4hmdCMP) maximum of 1.2 mM at pD 9 (formation rates = 0.067 µM s−1
,
as the major product (Fig. 2A and Fig. S3†). This product, 0.05 µM s−1 and 0.05 µM s−1 at pDs 6, 7.5, and 9 respectively,
which was observed as two conformational isomers (Fig. S3†), Fig. 2A and Fig. S6B/C†). The minor product was provisionally
formed more slowly than 3hmTMP or 3hmUMP (formation assigned as N6-dihydroxymethylated (6,6-dihydroxymethyl)
rate at pD 7.5 = 0.2 µM s−1, Fig. 2B), but reached a higher deoxyadenosine monophosphate (6dhmdAMP) on the basis of
maximal concentration (1.3 mM) after 350 minutes at pD 7.5 1H–13C-HMBC analyses, which revealed correlations between
(Fig. 2A). No significant differences in the reaction profile was the hemiaminal resonance (at δH 5.35 ppm and δC 72 ppm)
observed at pDs 6, 7.5 and 9 (formation rates = 0.117 µM s−1 and both the C-6 carbon (at δC 154 ppm) and itself. Formation
and 0.133 µM s−1 at pDs 6 and 9 respectively, Fig. 2A and rates for this adduct were too low to be accurately determined
Fig. S3B/C†). Two additional 1H-resonances were observed at (<0.003 µM s−1). Analyses with adenosine AMP, ADP and ATP
low levels (at δH 4.97 ppm, δH 6.35 ppm and δH 8.16 ppm at pD revealed similar reactions to those of dAMP, suggesting that
7.5), which were assigned as the N-hydroxymethyl and aro- the outcomes of reactions with HCHO are not significantly
matic protons of (4,4-dihydroxymethyl)deoxycytidine mono- affected under the tested conditions by the presence of a 2′
phosphate (4dhmdCMP). This assignment, as well as that of alcohol or additional phosphate groups (Fig. S7†).
4hmdCMP, was supported by MS analyses on the reaction Interestingly, the dAMP adducts were apparently more stable
mixture at pD 7.5, which revealed the presence of species with than the adducts formed from the other deoxynucleotides, as
masses expected for dCMP (308 Da, [M + H]+), a monohydroxy- evidenced by D2O dilution studies; under the conditions
methylated adduct (338 Da, [M + H]+) and a dihydroxymethyl- tested, 6hmdAMP was present in high concentrations for over
ated adduct (366 Da, [M − H]−) respectively. On D2O dilution 10 hours (degradation rate = 0.003 µM s−1 at pD 7.5 after
(2- to 25-fold), both 4hmdCMP and 4dhmdCMP degraded to 25-fold dilution, Fig. 2C and Fig. S4C/D†).
free HCHO and dCMP, demonstrating dynamic equilibrium;
NMR analyses were then undertaken on mixtures contain-
however, the rates of degradation (0.133 µM s−1 for 4hmdCMP, ing the canonical nucleotides and 4 equivalents of HCHO
0.003 µM s−1 for 4dhmdCMP after 25-fold dilution) were sig- (Fig. S8†). Formation of the dihydroxymethylated adducts
nificantly slower than the near-instantaneous degradation (4dhmdCMP, 2dhmdGMP and 6dhmdAMP) was not observed;
observed for 3hmTMP (Fig. 2C and Fig. S4A/B†).
however, the monohydroxymethylated adducts 3hmTMP,
dGMP reacted with HCHO to form the exocyclic adduct 3hmUMP, 4hmdCMP, 2hmdGMP and 6hmdAMP were
(2-hydroxymethyl)deoxyguanosine monophosphate (2hmdGMP) observed, suggesting these adducts might be more biologically
as the major product (Fig. 2A/B and Fig. S5†). The emergence relevant (note: physiological HCHO concentrations, especially
of a 1H-resonance at δH 5.07 ppm, likely corresponding to a at the sub-cellular level, are currently unclear). While the for-
hemiaminal, implied formation of a second adduct, although mation rates for 4hmdCMP, 2hmdGMP and 6hmdAMP were
the maximal concentration of this species (0.06 mM after markedly slower than in the mixtures with excess HCHO
360 minutes at pD 9, formation rate = 0.005 µM s−1, Fig. S5C†) (Fig. S8†), analyses of the mixtures containing TMP and UMP
was significantly lower than that of 2hmdGMP (1.2 mM after revealed rapid equilibration (i.e. equilibrium was reached prior
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Paper
Scheme 1 Summary of N-hydroxymethyl adducts observed upon treatment of nucleotides with HCHO. Five canonical nucleotides were tested for
reaction with HCHO; dAMP, dCMP, dGMP, TMP and UMP. Seven modified nucleotides were also tested; 1 mA, m6A, 6mG, 5hmdC, 5mdC, I, and Ψ.
Reactions were observed with all nucleotides (note: reaction with 1 mA was too low-level for NMR characterisation). The rates of reaction, however,
vary significantly.
to the first NMR analysis), although the concentrations of hour at 10 °C (Fig. S9†). As expected, the formation rates for all
3hmTMP and 3hmUMP were low (both 0.04 mM).
adducts were slower at 10 °C than at 25 °C; formation of
Temperature dependence studies on the reactions of HCHO 3hmTMP was sufficiently slow to enable a provisional for-
with the canonical nucleotides were then carried out. mation rate to be determined (0.617 µM s−1). Formation of
Formation of adducts from mixtures of excess HCHO with the 3hmUMP was largely complete before the first NMR analysis.
five canonical nucleotides (2.4 mM) was monitored over one Analysis of mixtures of HCHO with dAMP, dCMP and dGMP
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
over one hour at 15, 20, 25, and 37 °C revealed a near-linear the samples with 5mdC, 5hmdC, m6A, 6mG, I and Ψ. Of these
dependence on adduct formation rates with temperature nucleotides, m6A reacted least efficiently with HCHO
(Fig. S10†). For the N-monohydroxymethylated adducts, for- (Fig. S12†); formation of the only observed product, (6-hydroxy-
mation of 6hmdAMP was slowest, followed by formation of methyl-6-methyl)adenosine (6hm6mA), was slower than
2hmdGMP and 4hmdCMP respectively (Fig. S10†). A similar 6hmdAMP formation (formation rate of 6hm6 mA = 0.033
general trend was observed for the dihydroxymethylated µM s−1). Both 5mdC and 5hmdC react with HCHO via their
adducts, although their low levels in many of the samples pre- exocyclic N-4 nitrogens (as for dCMP) to form N4-hydroxy-
cluded accurate determination of formation rates.
methylated adducts as major products (4-hydroxymethyl-5-
1-Dimensional Exchange Spectroscopy (1D-EXSY) was then methyl)deoxycytidine (4hm5mdC) and (4-hydroxymethyl-5-
carried out on the reactions to investigate exchange rates hydroxymethyl)deoxycytidine (4hm5hmdC), respectively (Fig. 3
between the nucleotides and their HCHO-derived adducts. and Fig. S13†). Low-level resonances were also observed but
EXSY, which utilises the same pulse sequences used to identify were too weak for assignment. A competition experiment,
nuclear Overhauser effect (nOe) correlations, can be applied to where cytidine (C), 5mdC and 5hmdC were treated together
identify correlations between species in chemical exchange; with HCHO, revealed that the formation rate for 4hm5mdC
the exchange-peak intensity growth rates at different mixing was faster than that observed for 4hm5hmdC and (4-hydroxy-
times are proportional to the exchange rates.40–43 The methyl)cytidine (4hmC); with the exception of 3hmTMP and
intensities of the N-hydroxymethyl 1H resonances of the 3hmUMP, 4hm5mdC was the fastest to form of all the
nucleotide-HCHO adducts (δH 5–6 ppm) were determined after observed adducts (formation rate = 0.233 µM s−1, Fig. 3C).
pre-irradiation of the 1H-resonance of hydrated HCHO (δH 6mG reacted with HCHO in a similar manner to dGMP,
4.74 ppm at 25 °C). No N-hydroxymethyl resonances were forming (2-hydroxymethyl-6-methyl)guanosine (2hm6mG,
observed with dAMP or dGMP at 25 °C or 45 °C (dCMP was not formation rate = 0.05 µM s−1, Fig. S14†). Other low-level reso-
studied as the chemical shifts of the 1H resonances for nances were also observed in the spectra but were too weak to
4hmdCMP and 4dhmdCMP were too close to the HCHO fully assign; they may reflect low level formation of dihydroxy-
1H-resonance).
methylated adducts.
By contrast, 1D-EXSY analyses with TMP and UMP revealed
exchange between 3hmTMP/TMP and 3hmUMP/UMP, respect-
ively (Fig. S11†), implying the establishment of fast dynamic
equilibria. Initial formation rates for 3hmTMP and 3hmUMP
were calculated by measuring the intensities of the
N-hydroxymethyl 1H-resonances at different mixing times
(0.01–1.5 s, for details see ESI†). The results reveal that for-
mation of 3hmTMP and 3hmUMP is significantly faster than
for the adducts of dAMP, dGMP and dCMP (initial formation
rates at 25 °C: 65 µM s−1 and 97 µM s−1, respectively). Plotting
the second order rate constants for adduct formation as a
function of temperature gave activation energies of 103.7 kJ
mol−1 and 86.6 kJ mol−1 for 3hmTMP and 3hmUMP, respect-
ively (Fig. S11†).
Overall, the NMR studies with canonical nucleotides reveal
formation of distinct N-hydroxymethylated adducts with
different formation rates and stabilities. With the exception of
TMP and UMP, NMR-observable HCHO reactions occur on the
exocyclic nitrogens, with the adducts of dAMP being most
stable; TMP and UMP react fastest but their adducts are in
dynamic exchange with the free nucleotides.
Fig. 3 HCHO reacts with modified cytosine nucleosides to give
N-hydroxymethyl adducts. (A) 1H NMR spectrum of 4hm5hmdC, formed
after reaction of 5hmdC with HCHO. 1H-resonances for 4hm5hmdC are
highlighted (note: unreacted 5hmdC is also present in the mixture – its
1H-resonances overlap with those of 4hm5hmdC). 1H-resonances
marked with # are derived from HCHO. (B) 1H NMR time course analysis
Reactions of HCHO with modified nucleosides
The reactions of HCHO with non-canonical bases, i.e.
(5-methyl)deoxycytidine (5mdC), (5-hydroxymethyl)deoxycyti-
dine (5hmdC), (1-methyl)adenosine (1 mA), (6-methyl)adeno-
sine (m6A), (6-methyl)guanosine (6mG), inosine (I), and of a reaction mixture containing 5hmdC (2.4 mM) and HCHO (53 eq.) in
D2O at pD 7.5. Time-dependent formation of 4hm5hmdC is observed,
pseudouridine (Ψ) were then analysed. As with the canonical
nucleotides, mixtures containing the nucleoside with a 53-fold
excess of HCHO in D2O at pD 7.5, were analysed over
with concomitant decrease in 5hmdC. (C) 1H NMR time course analysis
of a reaction mixture containing cytidine (C), 5mdC and 5hmdC (all
2.4 mM) and HCHO (53 eq.) at pD 7.5. d5mC reacted mostly efficiently
600–800 min. No/trace reaction was observed between HCHO
with HCHO, forming 4hm5mdC (green), followed by 5hmdC (forming
and 1 mA; however, NMR-observable products were detected in 4hm5hmdC, maroon) and C, forming 4hmC (blue).
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Paper
Inosine (I), the N-6 deaminated product of adenosine, (2D-1H–13C-HSQC). No discrete resonances that could be
arises by spontaneous hydrolysis and by action of deaminase assigned to HCHO-derived adducts with the oligonucleotide
enzymes.44 The presence of I in DNA is considered pre-muta- were identified; it could be that such species are distributed
genic and it has functional roles in mRNA and tRNA.44 HCHO across different sites, making their detection difficult.
reacted with I via its endocyclic N-1 nitrogen, forming However, the intensity of the 13C-resonance of hydrated
(1-hydroxymethyl)inosine (1hmI); as observed with UMP and 13C-labelled HCHO (δC 82 ppm, Fig. S16†) was significantly
TMP, reaction was fast, resulting in establishment of a lower in the presence of oligonucleotide, implying sequestra-
dynamic equilibrium before the first NMR analysis (Fig. S15†). tion of HCHO by the oligonucleotide.
Ψ, one of most abundant modified nucleosides in RNA,45,46
We then analysed the formation and stability of the
also reacted with HCHO via either of its endocyclic N-1 and N-hydroxymethylated nucleotide adducts formed via nucleic
N-3 nitrogens, forming both mono-N-hydroxymethylated var- acid demethylase (FTO) catalysis (Fig. 5 and Fig. S17–S20†).
iants ((1-hydroxymethyl)pseudouridine, 1hmΨ, and (3-hydroxy- FTO was initially reported as
a demethylase acting on
methyl)pseudouridine 3hmΨ), as well as the dihydroxymethyl- N-methylated DNA,48 and subsequently reported to act on
ated adduct (1,3-dihydroxymethyl)pseudouridine (1hm3hmΨ, N-methylated RNA.49 FTO catalyses demethylation of
Fig. 4). Formation of all adducts appeared to reach equilibrium (3-methyl)thymine and (6-methyl)adenine in mRNA. Given our
before the first NMR analysis (within 5 min reaction), but their observation that 3hmTMP and 6hmdAMP have different for-
concentrations (0.4 mM for 1hmΨ and 3hmΨ, 0.1 mM for mation rates and stabilities, we proposed that the properties of
1hm3hmΨ) were low relative to unreacted Ψ (1.5 mM, Fig. 4B).
the two FTO-derived products should differ.
FTO-catalysed oxidation of (3-methyl)thymidine (3mT) was
first analysed; samples containing 3mT 13C-labelled on its
3-methyl group (400 µM), 2OG (5 mM), sodium ascorbate
(1 mM), ferrous iron (20 µM) and FTO (20 µM, diluted from a
200 µM stock in HEPES/Tris buffer pH 7.5) were prepared in
ammonium formate buffer (D2O, pH 7.5), then immediately
transferred to a tube for analysis by 1H NMR (700 MHz) and
1-dimensional heteronuclear single quantum correlation NMR
(1D-1H–13C-HSQC). 1D-1H–13C-HSQC analysis allows for
enhanced detection of 1H resonances attached to 13C-labels,
and was employed to accentuate resonances derived from
the 13C methyl group. FTO-catalysed demethylation of 3mT
was observed, as evidenced by loss of the N-methyl group
1H-resonance (doublet at δH 3.3 ppm, JCH = 145 Hz), and for-
Analyses of the N-hydroxymethylated adducts in
oligonucleotides
Studies were then carried out on N-hydroxymethyl adducts in
oligonucleotides. A mixture containing a synthetic Dickerson-
Drew-type
B-DNA
oligonucleotide47
(CGCGAATTCGCG,
0.2 mg mL−1) and 13C-labelled HCHO (0.4 mM) in ammonium
formate buffer pH 7.5, and was monitored by 13C NMR and
2-dimensional heteronuclear single quantum correlation NMR
mation of 13C-labelled HCHO (doublet at δH 4.8 ppm, JCH
=
165 Hz) using 1D-1H–13C-HSQC (Fig. S17†). No evidence for
the presence of 3hmT was accrued, suggesting that either the
proposed hydroxymethylated product is unstable (consistent
with our HCHO incubation experiments with TMP, Fig. 2 and
Fig. S1†), or, less likely, that the demethylation in this case
does not proceed via initial hydroxylation. Repeating the
experiment in the presence of the non-reactive 2OG analogues
N-oxalylglycine and 2,4-pyridinedicarboxylic acid38 inhibited
catalysis (90% and 80% inhibition respectively).
Studies with FTO and m6A (13C-labelled on its 6-methyl
group) under identical conditions over 19 hours also revealed
reaction. A competition experiment, with 3mT and m6A both
present in the reaction, revealed that 3mT is marginally
favoured as a substrate, although similar levels of reactivity
with both were observed (Fig. S18†). By contrast with the 3 mT
results, the major FTO-catalysed product with m6A during the
first 6 hours was clearly the HCHO adduct, i.e. (6-hydroxy-
methyl)adenosine (6hmA); this slowly degraded to adenosine
and HCHO (Fig. 5 and Fig. S19†). After the first time point
(one hour) no substantial further formation of succinate was
Fig. 4 HCHO reacts with Ψ to give three hydroxymethylated adducts.
(A) 1H NMR spectrum of a reaction mixture of Ψ (2.4 mM) and HCHO (53
eq.) in D2O at pD 7.5. 1H-resonances for 1hmΨ, 3hmΨ, and 1hm3hmΨ
are highlighted. (B) 1H NMR time course analysis of the same mixture.
Formation of the three adducts is completed before the first NMR time
point (7 minutes after mixing).
1
observed (using H NMR, Fig. S20†), suggesting that FTO cata-
lysis was concluded and that formation of HCHO is a conse-
quence of 6hmA fragmentation. Similar results were observed
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 5 NMR Analyses of FTO-catalysed oxidation of m6A and 3mT. (A) 1-Dimensional 1H–13C heteronuclear single quantum correlation
(1D-1H–13C-HSQC, 13C-coupled) spectra showing FTO-catalysed oxidation of m6A. A sample containing m6A (400 µM), which contained a 13C-
label on the 6-methyl group, was incubated with 2OG (5 mM), ascorbate (1 mM), ferrous iron (20 µM) and FTO (20 µM) in deuterated ammonium
formate buffer in D2O at pH 7.5; the reaction was monitored by 1H and 1D-1H–13C-HSQC NMR (over 19 hours, 310 K). Using 1D-1H-13C-HSQC, 1H-
resonances from the 13C-labelled 6-methyl group are accentuated, enabling sensitive observation of intermediates/products (1H-resonances
attached to 13C-labelled carbon appear as doublets due to 1H–13C coupling). After one hour, the major 13C-labelled product in the reaction was
6hmA; only trace amounts of hydrated 13C-labelled HCHO (H13CHO) were observed (middle). Formation of 6hmA coincided with decreased m6A
and formation of succinate, validating FTO-catalysed hydroxylation of m6A to 6hmA. Over time, 6hmA levels decreased with concomitant H13CHO
formation (top); no significant increase in succinate was observed after one hour, indicating termination of catalysis. Evidence for formation of 13C-
labelled methylamine was accrued in the 1D-1H–13C-HSQC spectrum after 19 hours (asterisks), likely as a consequence of (6-methyl)adenosine dea-
mination; the adduct of residual Tris and
H
13CHO was also observed.67 Resonances corresponding to residual HEPES are hashed. (B)
1D-1H–13C-HSQC time course analysis of FTO-catalysed oxidation of m6A. Levels of 6hmA were highest at the first time point (one hour) before
degrading to form H13CHO. However, under the tested conditions, 6hmA was more prevalent than H13CHO for 6 hours after FTO addition. (C)
1D-1H–13C-HSQC time course analysis of FTO-catalysed oxidation of 3mT. 13C-labelled 3mT (labelled on the 3-methyl group, 400 µM), 2OG
(5 mM), ascorbate (1 mM), ferrous iron (20 µM) and FTO (20 µM) were mixed in deuterated ammonium formate buffer in D2O at pD 7.5, and the reac-
tion was monitored by 1H and 1D-1H–13C-HSQC NMR over 8 hours at 310 K. Throughout the analysis, the major 13C-labelled product was hydrated
H13CHO; no evidence for the 3hmT species was accrued. (D) MS time course analysis of FTO-catalysed oxidation of m6A-containing RNA oligo-
nucleotide (AUUGUGG-m6A-CUGCAGC). The oligonucleotide (1 µM) was incubated with 2OG (10 µM), ascorbate (100 µM), ferrous iron (10 µM) and
FTO (100 nM) in 50 mM Tris buffer in H2O at pH 7.5, and the reaction was monitored by MS over 30 min at 298 K. Formation of 6hmA was observed;
however, no evidence for the demethylated oligonucleotide was accrued over the time course.
in experiments with an RNA oligonucleotide containing m6A; lation, but this may be due to the different experimental con-
MS analysis revealed FTO-catalysed hydroxylation of the oligo- ditions used).
nucleotide, which appeared stable over the course of the
Overall, the NMR and MS observations correlate with the
experiments (0.5 hours, Fig. 5 and Fig. S21†). This observation relative stabilities of 3hmTMP and 6hmdAMP in the HCHO
is consistent with previous studies on FTO catalysis with m6A incubation experiments, and imply that demethylase catalysis
in RNA, where MS analyses provided evidence for formation of has the potential to result in accumulation of N-hydroxymethyl
N6-hydroxymethylated and N6-formylated adenine in FTO- adducts in nucleic acids. Further, the results suggest that, at
treated m6A-containing oligonucleotides, as well as in RNA least in vitro, the primary product of FTO catalysis with
isolated from HeLa cells50 (note: we did not observe N6-formy- (6-methyl)adenine is the (6-hydroxymethyl)adenine adduct.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Paper
While clear reactions between HCHO and nucleic acids to
give specific products have been observed, the physiological
Conclusions
Our studies on the reactions of HCHO with nucleobases reveal relevance of such reactions is unknown. Given that HCHO reacts
the formation of hemiaminal adducts via reactions on both with free nucleotides and with oligonucleotides, and that HCHO
exocyclic and endocyclic nitrogens, consistent with previous is continually produced in cells, it seems probable that intra-
work (Scheme 2).28,29,35,36 Incubation of HCHO with TMP and cellular HCHO-DNA/RNA reactions occur, potentially with func-
UMP resulted in rapid reaction with the endocyclic nitrogen to tional consequences. In addition, it is established that changes
form 3-hydroxymethylated derivatives that are in fast dynamic in the composition of the cellular nucleotide pool affect DNA/
equilibrium with the free nucleotide (Scheme 1 and Fig. 2, RNA biosynthesis;54,55 it is thus possible that reactions with
Fig. S1 and S2†). Similar reactivity was observed with inosine HCHO, either on DNA/RNA or free nucleotides, regulate nucleo-
(I), forming (1-hydroxymethyl)inosine (1hmI, Fig. S15†), and tide reprocessing. The observation of HCHO reactions on endo-
with pseudouridine (Ψ), which strikingly formed three pro- cyclic nitrogens, forming hemiaminal products in fast dynamic
ducts comprising both 1- and 3-hydroxymethylated adducts, as equilibrium with the free bases, implies that the hydrogen
well as the 1,3-dihydroxymethylated adduct (Fig. 4). The reac- bonding capabilities of thymine, uracil, inosine and pseudoura-
tive nitrogens in these nucleotides are proposed to predomi- cil (for duplex formation) could be regulated by HCHO in a con-
nantly exist as amide N–H groups in their canonical tauto- centration-dependent manner. In contrast, reactions on the exo-
meric forms (as depicted in the figures).51 The observed reac- cyclic nitrogens of cytosine, adenine and guanine bases do not
tion rates for TMP and UMP with HCHO were faster at pD 9 necessarily block hydrogen bonding interactions (at least in
over pD 7.5, consistent with the pKa values for cyclic amide N–H duplex DNA), and therefore, these bases could act as ‘reservoirs’
bonds (7–10).52 The observed fast fragmentation of the hydroxy- for HCHO, accumulating HCHO over time.
methyl adducts formed from reactions at the endocyclic nitro-
gens is also consistent with reported kinetic studies.53
It is important to consider that within biological contexts
the stabilities of the observed adducts may be altered. We
By contrast to the reactions of HCHO with TMP, UMP, I and accrued no direct evidence for HCHO-derived methylene cross-
Ψ, those with dCMP, dGMP and dAMP, as well with the modi- links, though such species have been proposed in DNA, RNA
fied nucleosides 5mdC, 5hmdC, m6A and 6mG, occur on the and between free nucleobases after prolonged incubations;56,57
exocyclic nitrogens (Scheme 1 and Fig. 2, 3, Fig. S3, S5–S7 and we observed low levels of unassigned products. For cross-
S12–S14†). Whilst slower to form than the adducts with TMP linking to occur, it is likely that formation of an imine/
and UMP (correlating with the higher pKa values of the exocyc- iminium intermediate is required, which is then ‘trapped’ by a
lic amines),52 these adducts are more stable. In all cases, nucleophile, e.g. a nucleic acid amine.27,58,59 Such imine/
adduct formation appeared slightly faster at pD 9 and slower iminium species were not observed in our NMR analyses with
at pD 6, relative to pD 7.5 (Fig. S3, S5 and S6†), suggesting the nucleotides/nucleosides, implying production of such inter-
possibility of base catalysis during adduct formation and/or an mediates might be limiting, at least when HCHO concen-
altering of the dynamicity between the carbonyl and hydrated trations are low. However, in the case of cellular DNA, where
forms of HCHO. Within limits of detection, no evidence for nucleic acid bases are proximal and conformationally con-
reactions on endocyclic nitrogens was observed with these strained, it is possible cross-linking is more efficient, either
nucleotides/nucleosides; this suggests that such reactions due to immediate trapping of an imine/iminium ion, or by
either do not occur or that these potential products are promoting alternative cross-linking mechanisms. While recent
unstable. The observations imply that nucleobase hemiaminal work has shown aminal cross-links between two adenosine
adducts are stable enough to have functional roles in biology, molecules are susceptible to rapid hydrolysis, particularly in
potentially relating to their different stabilities and/or rates of the presence of anthranilates and phosphanilates, the stability
formation and breakdown.
of such species in DNA/RNA is unknown.57 It is of interest that
Scheme 2 Summary of observed reactions between nucleotides and HCHO. R = H or CH3, R’ = H, CH3 or CH2OH.
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Scheme 3 FTO-catalysed oxidation of 3mT and m6A give differently stable N-hydroxymethyl products. Treatment of 3mT with FTO leads to for-
mation of 3hmT, which rapidly degrades to the free nucleotide (blue) and HCHO. However, FTO-catalysed hydroxylation of m6A forms 6hmA (red),
which only very slowly fragments (over hours). KDM-catalysed demethylation of Nε-methylated lysine residues (dashed box) leads to an unstable
hemiaminal that rapidly fragments to the give the demethylated lysine (blue) and formaldehyde (cf. 3mT demethylation).
histone lysine side chains are proposed to react with (5-formyl) methyl)adenine has biologically-relevant functional roles. In
cytosine bases via imine formation.60–61
this regard, it will be interesting to determine whether oxyge-
The effects of the HCHO ‘adducts’ on the cellular functions nase-catalysed hydroxylation of (6-methyl)adenine may be
of nucleobases are yet to be determined, as is how their pres- ‘partnered’ with enzyme-catalysed fragmentation of (6-hydroxy-
ence relates to the toxic effects of HCHO. However, the roles of methyl)adenine. It is also possible that the reactions of HCHO
some modifications to nucleobases are well-established, e.g. with nucleobases are enzyme-catalysed.
that of (5-methyl)cytosine in the regulation of transcription,
Finally, it is likely that the full extent of the reversible reac-
and the roles of others are emerging, e.g. that of (5-hydroxy- tions between HCHO, and other electrophilic small molecules,
methyl)cytosine in transcription and of (6-methyl)adenine in with nucleophiles on nucleic acids, proteins, and biological
the regulation of RNA stability/activity.62,63 As well as modulat- small-molecules is very far from being completely defined. In
ing interactions between nucleic acids, and between nucleic contrast to the observation of hemiaminal adducts with FTO,
acids and proteins, by formation of covalent cross-links, adducts have not been observed in studies with Nε-methyl-
HCHO adducts have the potential to modulate such inter- lysine demethylases.39,66 However, work with small molecules
actions by altering non-covalent interactions, e.g. by hydrogen- reveals the potential for formation of stable HCHO-derived
bonding interactions or aqueous solvation. They also have the adducts with alkylamines with appropriate stabilisation.67 As
potential to directly alter the chemical stability of nucleic acids suggested by the work described here and previously on
and their oligomers (e.g. duplex/quadruplex stability), either nucleic acids and some small molecules (e.g. glutathione),68
directly as shown for (5-methyl)cytosine and (5-hydroxymethyl) there is scope for immense complexity, raising questions
cytosine,47,64 or indirectly by affecting the activity of nucleic about how such interactions may affect and/or regulate the
acid modifying enzymes such as APOBEC (apolipoprotein B canonical genetic code.
mRNA editing enzyme, catalytic polypeptide-like) and related
enzymes.65 Whether or not such effects are relevant in cells
requires further investigation.
Conflicts of interest
The studies with FTO suggest enzyme-catalysed oxidation of
N-methylated nucleic acids might be a significant source of There are no conflicts to declare.
N-hydroxymethyl adducts in cells, as exemplified by (6-hydroxy-
methyl)adenine. Oxygenase-catalysed N-methyl demethylation
has the potential to deliver hemiaminals at specific sites
in vivo. Further, the results imply that hydroxylation, rather
Acknowledgements
than demethylation, should perhaps be considered as the We thank the Biotechnology and Biological Sciences Research
major biochemical function of FTO catalysis on (6-methyl) Council (BB/L000121/1 and sLoLa grant BB/J001694/2: extend-
adenine (Scheme 3), revealing the possibility that (6-hydroxy- ing the boundaries of nucleic acid chemistry), the Wellcome
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Paper
Trust (091857/7/10/7) and Cancer Research UK (C8717/A18245) 23 L. B. Pontel, I. V. Rosado, G. Burgos-Barragan,
for funding the work. RJH acknowledges a William R Miller
Junior Research Fellowship, St Edmund Hall, Oxford. SS is
funded by the Felix Scholarship. We thank Jules Camp for pre-
liminary adduct characterisation by NMR.
J. I. Garaycoechea, R. Yu, M. J. Arends, G. Chandrasekaran,
V. Broecker, W. Wei, L. Liu, J. A. Swenberg, G. P. Crossan
and Ketan J. Patel, Mol. Cell, 2015, 60, 177–188.
24 R. Saladino, C. Crestini, G. Costanzo and E. DiMauro, Curr.
Org. Chem., 2004, 8, 1425–1443.
25 A. L. Weber, Origins Life Evol. Biospheres, 2000, 30, 33–43.
26 R. G. Kallen and W. P. Jencks, J. Biol. Chem., 1966, 241,
5864–5878.
References
1 H. P. Agency, http://www.gov.uk, 2014.
2 N. P. I. Service, TOXBASE®, 2015.
27 E. H. Cordes and W. P. Jencks, J. Am. Chem. Soc., 1962, 84,
832–837.
3 M. İnci, İ. Zararsız, M. Davarcı and S. Görür, 28 J. D. McGhee and P. H. Von Hippel, Biochemistry, 1975, 14,
Turk. J. Neurol., 2013, 39, 48–52. 1281–1296.
4 L. F. Lima, G. L. Murta, A. C. B. Bandeira, C. R. Nardeli, 29 J. D. McGhee and P. H. Von Hippel, Biochemistry, 1975, 14,
W. G. Lima and F. S. Bezerra, Ann. Anat., 2015, 202, 45–51. 1297–1303.
5 I. A. f. R. o. Cancer, IARC Monographs on the Evaluation OF 30 K. Lu, S. Craft, J. Nakamura, B. C. Moeller and
Carcinogenic Risks to Humans, 2006, p. 88. J. A. Swenberg, Chem. Res. Toxicol., 2012, 25(3), 664–675.
6 N. Y. Tretyakova, A. Groehler and S. Ji, Acc. Chem. Res., 31 M. Wang, G. Cheng, S. Balbo, S. G. Carmella, P. W. Villalta
2015, 48, 1631–1644. and S. S. Hecht, Cancer Res., 2009, 69, 7170–7174.
7 C. R. McNamara, J. Mandel-Brehm, D. M. Bautista, 32 Y. F. M. Chaw, L. E. Crane, P. Lange and R. Shapiro,
J. Siemens, K. L. Deranian, M. Zhao, N. J. Hayward, Biochemistry, 1980, 19, 5525–5531.
J. A. Chong, D. Julius, M. M. Moran and C. M. Fanger, Proc. 33 H. Huang, M. S. Solomon and P. B. Hopkins, J. Am. Chem.
Natl. Acad. Sci. U. S. A., 2007, 104, 13525–13530. Soc., 1992, 114, 9240–9241.
8 Y. Chen, P. Kanju, Q. Fang, S. H. Lee, P. K. Parekh, W. Lee, 34 K. Lu, G. Boysen, L. Gao, L. B. Collins and J. A. Swenberg,
C. Moore, D. Brenner, R. W. Gereau Iv, F. Wang and
W. Liedtke, Pain, 2014, 155, 2662–2672.
9 S. Tao, C. M. Woodrow, H. Xintian, H. Yingge and
H. Rongqiao, Curr. Top. Med. Chem., 2016, 16, 472–484.
10 C. L. Nie, X. S. Wang, Y. Liu, S. Perrett and R. Q. He, BMC
Neurosci., 2007, 8, 9–9.
Chem. Res. Toxicol., 2008, 21, 1586–1593.
35 K. Lu, W. Ye, A. Gold, L. M. Ball and J. A. Swenberg, J. Am.
Chem. Soc., 2009, 131, 3414–3415.
36 K. Lu, W. Ye, L. Zhou, L. B. Collins, X. Chen, A. Gold,
L. M. Ball and J. A. Swenberg, J. Am. Chem. Soc., 2010, 132,
3388–3399.
11 W. H. Organization, Concise International Chemical 37 D.-H. Lee, S.-G. Jin, S. Cai, Y. Chen, G. P. Pfeifer and
Assessment Document 40, 2002.
T. R. O’Connor, J. Biol. Chem., 2005, 280, 39448–39459.
38 W. Aik, M. Demetriades, M. K. K. Hamdan, E. A. L. Bagg,
K. K. Yeoh, C. Lejeune, Z. Zhang, M. A. McDonough and
C. J. Schofield, J. Med. Chem., 2013, 56, 3680–3688.
12 A. European Food Safety, EFSA J., 2014, 12, 3550.
13 A. C. De Groot, I. R. White, M.-A. Flyvholm, G. Lensen and
P.-J. Coenraads, Contact Dermatitis, 2010, 62, 2–17.
14 A. De Groot, I. R. White, M.-A. Flyvholm, G. Lensen and 39 R. J. Hopkinson, R. B. Hamed, N. R. Rose, T. D. Claridge
P.-J. Coenraads, Contact Dermatitis, 2010, 62, 18–31. and C. J. Schofield, ChemBioChem, 2010, 11, 506–510.
15 G. Burgos-Barragan, N. Wit, J. Meiser, F. A. Dingler, 40 J. Jeener, B. H. Meier, P. Bachmann and R. R. Ernst,
M. Pietzke, L. Mulderrig, L. B. Pontel, I. V. Rosado, J. Chem. Phys., 1979, 71, 4546–4553.
T. F. Brewer, R. L. Cordell, P. S. Monks, C. J. Chang, 41 C. L. Perrin and T. J. Dwyer, Chem. Rev., 1990, 90, 935–967.
A. Vazquez and K. J. Patel, Nature, 2017, 548, 549.
16 P. J. Boor, M. B. Trent, G. A. Lyles, M. Tao and
G. A. S. Ansari, Toxicology, 1992, 73, 251–258.
42 R. J. Hopkinson, I. K. Leung, T. J. Smart, N. R. Rose,
T. D. Claridge and C. J. Schofield, PLoS One, 2015, DOI:
10.1371/journal.pone.0145085.
17 L. J. Walport, R. J. Hopkinson and C. J. Schofield, Curr. 43 Modern NMR Spectroscopy in Education, American Chemical
Opin. Chem. Biol., 2012, 16, 525–534. Society, 2007, ch. 13, vol. 969, pp. 176–189.
18 H. d. A. Heck, M. Cassanova-Schmitz, P. B. Dodd, 44 I. Alseth, B. Dalhus and M. Bjørås, Curr. Opin. Genet. Dev.,
E. N. Schachter, T. J. Witek and T. Tosun, AIHA J., 1985, 46, 1–3. 2014, 26, 116–123.
19 Z. Tong, J. Zhang, W. Luo, W. Wang, F. Li, H. Li, H. Luo, 45 T. M. Carlile, M. F. Rojas-Duran, B. Zinshteyn, H. Shin,
J. Lu, J. Zhou, Y. Wan and R. He, Neurobiol. Aging, 2011, 32,
31–41.
K. M. Bartoli and W. V. Gilbert, Nature, 2014, 515, 143–146.
46 J. Ge and Y.-T. Yu, Trends Biochem. Sci., 2013, 38, 210–218.
20 Z. Tong, C. Han, W. Luo, X. Wang, H. Li, H. Luo, J. Zhou, 47 L. Lercher, M. A. McDonough, A. H. El-Sagheer,
J. Qi and R. He, Age, 2013, 35, 583–596.
21 H. d’A. Heck, E. L. White and M. Casanova-Schmitz, Biol.
Mass Spectrom., 1982, 9, 347–353.
22 I. V. Rosado, F. Langevin, G. P. Crossan, M. Takata and
K. J. Patel, Nat. Struct. Mol. Biol., 2011, 18, 1432–1434.
A. Thalhammer, S. Kriaucionis, T. Brown and
C. J. Schofield, Chem. Commun., 2014, 50, 1794–1796.
48 T. Gerken, C. A. Girard, Y.-C. L. Tung, C. J. Webby,
V. Saudek, K. S. Hewitson, G. S. H. Yeo, M. A. McDonough,
S. Cunliffe, L. A. McNeill, J. Galvanovskis, P. Rorsman,
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
P. Robins, X. Prieur, A. P. Coll, M. Ma, Z. Jovanovic, 58 M. Y. Feldman, in Progress in Nucleic Acid Research and
I. S. Farooqi, B. Sedgwick, I. Barroso, T. Lindahl,
C. P. Ponting, F. M. Ashcroft, S. O’Rahilly and
C. J. Schofield, Science, 2007, 318, 1469–1472.
Molecular Biology, ed. J. N. Davidson and W. E. Cohn,
Academic Press, 1973, vol. 13, pp. 1–49.
59 E. A. Hoffman, B. L. Frey, L. M. Smith and D. T. Auble,
J. Biol. Chem., 2015, 290, 26404–26411.
49 G. Jia, Y. Fu, X. Zhao, Q. Dai, G. Zheng, Y. Yang, C. Yi,
T. Lindahl, T. Pan, Y.-G. Yang and C. He, Nat. Chem. Biol., 60 F. Li, Y. Zhang, J. Bai, M. M. Greenberg, Z. Xi and C. Zhou,
2011, 7, 885–887. J. Am. Chem. Soc., 2017, 139, 10617–10620.
50 Y. Fu, G. Jia, X. Pang, R. N. Wang, X. Wang, C. J. Li, 61 S. Ji, H. Shao, Q. Han, C. L. Seiler and N. Y. Tretyakova,
S. Smemo, Q. Dai, K. A. Bailey, M. A. Nobrega, K.-L. Han,
Q. Cui and C. He, Nat. Commun., 2013, 4, 1798.
51 E. Westhof, FEBS Lett., 2014, 588, 2464–2469.
Angew. Chem., Int. Ed., 2017, 56, 14130–14134.
62 L. I. Kroeze, B. A. van der Reijden and J. H. Jansen,
Biochim. Biophys. Acta, Rev. Cancer, 2015, 1855, 144–154.
52 B. Thapa and H. B. Schlegel, J. Phys. Chem. A, 2015, 119, 63 P. J. Batista, Genomics, Proteomics Bioinf., 2017, 15, 154–163.
5134–5144.
64 A. Thalhammer, A. S. Hansen, A. H. El-Sagheer, T. Brown
and C. J. Schofield, Chem. Commun., 2011, 47, 5325–5327.
65 S. A. Roberts, M. S. Lawrence, L. J. Klimczak, S. A. Grimm,
D. Fargo, P. Stojanov, A. Kiezun, G. V. Kryukov, S. L. Carter,
G. Saksena, S. Harris, R. R. Shah, M. A. Resnick, G. Getz
and D. A. Gordenin, Nat. Genet., 2013, 45, 970–976.
53 J. Ugelstad and J. de Jonge, Recl. Trav. Chim. Pays-Bas, 1957,
76, 919–945.
54 M. Zauri, G. Berridge, M.-L. Thézénas, K. M. Pugh,
R. Goldin, B. M. Kessler and S. Kriaucionis, Nature, 2015,
524, 114–118.
55 G. Phear and M. Meuth, Mutat. Res., Fundam. Mol. Mech. 66 R. J. Hopkinson, L. J. Walport, M. Münzel, N. R. Rose,
Mutagen., 1989, 214, 201–206.
56 H. Huang and P. B. Hopkins, J. Am. Chem. Soc., 1993, 115,
9402–9408.
57 S. Karmakar, E. M. Harcourt, D. S. Hewings, F. Scherer,
A. F. Lovejoy, D. M. Kurtz, T. Ehrenschwender,
T. J. Smart, A. Kawamura, T. D. W. Claridge and
C. J. Schofield, Angew. Chem., Int. Ed., 2013, 52, 7709–7713.
67 I. K. Leung, T. J. Krojer, G. T. Kochan, L. Henry, F. von
Delft, T. D. Claridge, U. Oppermann, M. A. McDonough
and C. J. Schofield, Chem. Biol., 2010, 17, 1316–1324.
L. J. Barandun, C. Roost, A. A. Alizadeh and E. T. Kool, Nat. 68 R. J. Hopkinson, P. S. Barlow, C. J. Schofield and
Chem., 2015, 7, 752–758.
T. D. Claridge, Org. Biomol. Chem., 2010, 8, 4915–4920.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2018