A hydantoin isoform of cyclic N6-threonylcarbamoyladenosine (ct6A) is present in tRNAs

Article abstract of DOI:10.1093/nar/gkw1189

N⁶-Threonylcarbamoyladenosine (t⁶A) and its derivatives are universally conserved modified nucleosides found at position 37, 3 adjacent to the anticodon in tRNAs responsible for ANN codons. These modifications have pleiotropic functions of tRNAs in decoding and protein synthesis. In certain species of bacteria, fungi, plants and protists, t⁶A is further modified to the cyclic t⁶A (ct⁶A) via dehydration catalyzed by TcdA. This additional modification is involved in efficient decoding of tRNA^Lys. Previous work indicated that the chemical structure of ct⁶A is a cyclic active ester with an oxazolone ring. In this study, we solved the crystal structure of chemically synthesized ct⁶A nucleoside. Unexpectedly, we found that the ct⁶A adopted a hydantoin isoform rather than an oxazolone isoform, and further showed that the hydantoin isoform of ct⁶A was actually present in Escherichia coli tRNAs. In addition, we observed that hydantoin ct⁶A is susceptible to epimerization under mild alkaline conditions, warning us to avoid conventional deacylation of tRNAs. A hallmark structural feature of this isoform is the twisted arrangement of the hydantoin and adenine rings. Functional roles of ct⁶A37 in tRNAs should be reconsidered.

Full text of DOI:10.1093/nar/gkw1189

Published online 2 December 2016
Nucleic Acids Research, 2017, Vol. 45, No. 4 2137–2149
doi: 10.1093/nar/gkw1189
A hydantoin isoform of cyclic
N6-threonylcarbamoyladenosine (ct6A) is present in
tRNAs
Michal Matuszewski1, Jakub Wojciechowski2, Kenjyo Miyauchi3, Zofia Gdaniec4, Wojciech
M. Wolf2, Tsutomu Suzuki3,* and Elzbieta Sochacka1,*
1Institute of Organic Chemistry, Faculty of Chemistry, Lodz University of Technology, Zeromskiego 116, 90-924 Lodz,
Poland, 2Institute of General and Ecological Chemistry, Faculty of Chemistry, Lodz University of Technology,
Zeromskiego 116, 90-924 Lodz, Poland, 3Department of Chemistry and Biotechnology, University of Tokyo, 7-3-1
Hongo, Bunkyo-ku, Tokyo 113-8656, Japan and 4Institute of Bioorganic Chemistry, Polish Academy of Sciences,
Noskowskiego 12/14, 61-704 Poznan, Poland
Received June 22, 2016; Revised November 07, 2016; Editorial Decision November 10, 2016; Accepted November 25, 2016
ABSTRACT
ter transcription. To date, more than 100 species of modi-
fied nucleosides have been identified in tRNAs from all do-
mains of life (1). These modifications help to ensure proper
tRNA function by stabilizing tertiary structures and mod-
ulating decoding properties (2–6). A wide variety of chem-
ical modifications are found in the tRNA anticodon loops,
especially at the first letter of the anticodon (position 34),
the so-called ‘wobble’ position. Modifications at this posi-
tion play critical roles in modulating decoding capabilities
(4,6,7). On the other hand, hypermodified purine bases are
also found at position 37, 3ꢀ adjacent to the anticodon; these
modifications contribute to decoding processes by ensuring
reading-frame maintenance via stabilization of the codon–
anticodon interaction on the ribosome (8,9).
N6-Threonylcarbamoyladenosine (t6A) and its deriva-
tives are universally conserved modified nucleosides
found at position 37, 3ꢀ adjacent to the anticodon in
tRNAs responsible for ANN codons. These modifica-
tions have pleiotropic functions of tRNAs in decod-
ing and protein synthesis. In certain species of bacte-
ria, fungi, plants and protists, t6A is further modified
to the cyclic t6A (ct6A) via dehydration catalyzed by
TcdA. This additional modification is involved in ef-
ficient decoding of tRNALys. Previous work indicated
that the chemical structure of ct6A is a cyclic ac-
tive ester with an oxazolone ring. In this study, we
solved the crystal structure of chemically synthe-
sized ct6A nucleoside. Unexpectedly, we found that
the ct6A adopted a hydantoin isoform rather than an
oxazolone isoform, and further showed that the hy-
dantoin isoform of ct6A was actually present in Es-
cherichia coli tRNAs. In addition, we observed that
hydantoin ct6A is susceptible to epimerization under
mild alkaline conditions, warning us to avoid con-
ventional deacylation of tRNAs. A hallmark structural
feature of this isoform is the twisted arrangement of
the hydantoin and adenine rings. Functional roles of
ct6A37 in tRNAs should be reconsidered.
N6-Threonylcarbamoyladenosine (t6A) and its deriva-
tives are modifications found at position 37 of tRNAs re-
sponsible for decoding A-starting codons (ANN) in all do-
mains of life (Figure 1A) (8,10). Extensive in vitro and in
vivo studies of the biological functions of t6A revealed that
this modification has pleiotropic functions in translation,
including aminoacylation of tRNAs (11), tRNA binding to
the A-site codon (12), efficient translocation (13), reading-
frame maintenance (14) and prevention of leaky scanning
of initiation codons and read-through of stop codons (15).
Structural studies revealed that t6A enhances anticodon–
codon base pairing by cross-strand base stacking of the t6A
base with the first adenine base of the codon, and stabilizes
the anticodon stem–loop structure by stacking with A38
and preventing base-pairing with U33 (16–19).
In 2013, a cyclic form of t6A (ct6A) with a molecular mass
of 394 Da (Figure 1A) was discovered in bacteria, fungi,
protists, and plants (20). Because ct6A is formed via dehy-
dration of t6A, it is easily hydrolyzed to t6A during handling
of tRNA and preparation of nucleosides under mild alka-
INTRODUCTION
Transfer (t) RNA is an adaptor molecule that links the
codon in messenger (m) RNA to its corresponding amino
acid during protein synthesis. tRNAs contain a number of
modified nucleosides that are introduced enzymatically af-
*To whom correspondence should be addressed. Tel: +48 42 631 31 55; Fax +48 42 636 55 30; Email: elzbieta.sochacka@p.lodz.pl
Correspondence may also be addressed to Tsutomu Suzuki. Tel: +81 3 5841 8752; Fax: +81 3 5841 0550; Email: ts@chembio.t.u-tokyo.ac.jp
ꢁ
C
The Author(s) 2016. Published by Oxford University Press on behalf of Nucleic Acids Research.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/), which
permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact
journals.permissions@oup.com

2138 Nucleic Acids Research, 2017, Vol. 45, No. 4
O
OH
15
A
14
OH
11
NH
15
15
14
O
13
12
O
14
12
13
10
11
13
OH
N 12
H
N
10
10
11
H
O
N
O
OH
HN
N
6
6
6
O
7N
7N
7N
5
N1
5
4
5
4
N1
2
N1
8
8
8
2
2
HO
4
HO
9
9
9
HO
N
N
N
N
N
N
3
3
3
O
O
O
OH OH
OH OH
OH OH
t6A
ct6A
ct6A
(oxazolone isoform)
(hydantoin isoform)
B
C
1.0
0.8
0.6
0.4
0.2
0
0.8
0.6
0.4
0.2
0
ct6A
ct6A
t6A
t6A
5
10
15
20
250
300
350
Time (min)
Wavelength (nm)
Figure 1. Chemical synthesis of ct6A. (A) Chemical structures of t6A, oxazolone isoform of cyclic t6A, and hydantoin isoform of cyclic t6A. (B) HPLC
elution profiles of chemically synthesized t6A and ct6A. (C) UV spectra of chemically synthesized t6A and ct6A in H2O; for ct6A λmin = 233, λ
= 269;
max
ε = 18 900 dm3·mol−1·cm−1
.
line conditions (21). Consequently, t6A was discovered >40
years (10) before the detection of ct6A. When total nucleo-
sides are prepared under neutral conditions in the shortest
possible period of time, ct6A can be clearly detected, and
very little t6A is observed in Escherichia coli tRNAs (20),
demonstrating that ct6A is a bona fide modification in tR-
NAs from species including E. coli and yeast. According to
MS and NMR analyses, the chemical structure of ct6A was
determined to be a cyclic active ester with an oxazolone ring
(Figure 1A).
tRNALys, and possibly for other tRNAs containing ct6A.
The crystal structures of TcdA in complex with AMP or
ATP have been solved (22,23), providing insight into the
molecular basis of ct6A formation.
The discovery of ct6A encouraged us to characterize the
structural and biochemical contributions of this modifica-
tion to tRNA functions. For this purpose, we chemically
synthesized the 5ꢀ,3ꢀ,2ꢀ-O-acetylated derivative of ct6A,
which was stable under acidic and mild basic conditions
(24). This advance made it feasible to synthesize model
RNA oligonucleotides containing ct6A for use in further
structural and biochemical studies.
ct6A is synthesized by tRNA threonylcarbamoyladeno-
sine dehydratase A (TcdA), which catalyzes ATP-dependent
dehydration of t6A to form ct6A (20). CsdA and CsdE are
additional factors required for efficient ct6A formation (20).
The yeast homologs of tcdA, TCD1 and TCD2, are re-
quired for respiratory cell growth, indicating the physiolog-
ical importance of ct6A (20). Reporter assays using a ꢀtcdA
strain revealed that ct6A is involved in efficient decoding of
In this study, we elaborated simple synthesis of the non-
sugar-protected ct6A nucleoside using carbodiimide chem-
istry, and synthesized ct6A nucleoside at multi-milligram
scale. The resultant ct6A was successfully crystallized, and
the tertiary structure of the ct6A nucleoside was deter-
mined by X-ray crystallography. Contrary to our expecta-

Nucleic Acids Research, 2017, Vol. 45, No. 4 2139
tion, the crystal structure of ct6A contained a hydantoin
isoform rather than the previously determined oxazolone
isoform (Figure 1A). These two isoforms are not differenti-
ated by mass spectrometric analysis and simple NMR stud-
ies. The hydantoin structure of synthetic ct6A in solution
was further confirmed by detailed analysis of 15N NMR
data and the presence of characteristic absorption bands
in IR spectrum. Very surprising result of these structural
studies prompted us to investigate whether the hydantoin
isoform of ct6A is actually present in natural tRNAs. Care-
ful LC/MS co-injection analysis revealed the identity of the
synthetic hydantoin ct6A with the nucleoside in E. coli tR-
NAs. In addition, we found that hydantoin ct6A is suscepti-
ble to epimerization under mild alkaline conditions, warn-
ing us to avoid conventional deacylation procedure of tR-
NAs. Based on the structural features of this isoform, the
structure–function relationship of ct6A37 in tRNAs should
be reconsidered.
give sugar-protected t6A, which was de-protected by treat-
ment with ammonium-saturated methanol (10 M), yielding
the t6A nucleoside as the ammonium salt. The t6A ammo-
nium salt was then purified by column chromatography on
silica gel 60 (230–400 mesh, Sigma-Aldrich). Before cycliza-
tion, the t6A ammonium salt was converted to the nucleo-
side with a free carboxylate by cation ion exchange chro-
matography on Amberlite™ IR120 (H+ form). The purity
and structure of the t6A nucleoside were unambiguously
confirmed by analytical HPLC (Figure 1B), UV spectrum
1
(Figure 1C), IR (Supplementary Figure S1A), H NMR
(Supplementary Figure S2A) and 13C NMR (Supplemen-
tary Figure S3A), and the results were in full agreement with
earlier publications (25–28).
t6A nucleoside bearing D-allo-threonine residue was syn-
thesized according the same procedure as described above
for L-t6A isomer. The spectral data of D-allo-t6A fully cor-
related with those reported previously (29), and they are
shown in Supplementary Figure S4A (IR), Supplementary
Figure S5A (1H NMR) and Supplementary Figure S6A
(13C NMR).
MATERIALS AND METHODS
General information
A Waters HPLC (515) system using XTerra® Waters col-
umn (MS C8, 5 ␮m, 4.6 × 150 mm, 100 A) was used for an-
Chemical synthesis of ct6A
˚
t6A nucleoside in the free carboxylic form (300 mg, 0.72
mmol) was dissolved in 10 ml of water and mixed with
1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide immobi-
lized on a polymer support (EDC-P) (10 equiv, 5.14 g, 7.2
mmol, loading 1.4 mmol/g), followed by vigorous stirring
at room temperature for 1 h. After consumption of t6A nu-
cleoside was checked by TLC analysis (n-BuOH/H2O =
85/15, v/v), the reaction was quenched by filtering off the
polymer-bounded EDC, followed by three careful washes
of the polymer bed with H2O (10 ml each time). The filtrate
was evaporated to dryness under reduced pressure, and the
crude product was subjected to column chromatography on
silica gel with a linear gradient of H2O (0–3%) in n-BuOH to
purify ct6A nucleoside (140 mg, yield 49%). ct6A nucleoside
was obtained with similar yield (153 mg, yield 51%) using
dimethylformamide (DMF) as a solvent. The purity of ct6A
and its different lipophilicity relative to t6A was confirmed
by analytical HPLC (Figure 1B), UV spectroscopy (Fig-
alytical high-performance liquid column chromatography
(HPLC). Chromatography was performed at room temper-
ature, at a flow rate of 1 ml/min, in a gradient of acetoni-
trile (Solvent B) in 0.1% AcOH/water (Solvent A), as fol-
lows: 0% B to 30% B in 0–30 min, and 30% B to 50% B in
30–35 min. The elution profile was monitored by UV ab-
sorption at 254 nm using a Waters 996 photodiode array
detector. High-resolution mass spectrometry with electro-
spray ionization was conducted on a Maldi SYNAPT G2-
S HDMS (Waters). UV–vis spectra were obtained on an
O-2800 Hitachi UV Digilab1 spectrophotometer. IR data
were recorded on an FT-IR ALPHA (Bruker) equipped
with a Platinum ATR QuickSnap™ module. NMR spec-
tra were recorded in DMSO-d6 on Bruker Avance II Plus
700 and Bruker DPX-250 spectrometers. Chemical shifts
are given as ppm values relative to DMSO, used as an in-
ternal reference (2.50 ppm for 1H NMR and 39.51 ppm for
13C NMR). Signals were assigned based on 1H–1H COSY,
1
1
1H–13C HSQC, H–13C HMBC and 13C DEPT 135 spec-
ure 1C), IR (Supplementary Figure S1B), H NMR (Sup-
plementary Figure S2B), 13C NMR (Supplementary Figure
S3B), and high-resolution MS (Supplementary Figure S7).
Cyclization of D-allo-t6A was performed with the same
procedure as described above to generate D-allo-ct6A with
slightly lower yield (40%). The structure of D-allo-ct6A was
tra. Signals shapes and multiplicities are indicated as fol-
lows: br s = broad singlet, s = singlet, d = doublet, dd =
double doublets, ddd = double double doublets, t = triplet,
q = quartet, m = multiplet. Scalar coupling constants J are
given in Hertz (Hz). 15N chemical shifts were obtained from
analysis of 1H–15N HSQC and 1H–15N HMBC spectra, and
are given relative to liquid NH3. 1D NOE and 2D NOESY
spectra were recorded with a mixing time of 300 ms.
1
confirmed by IR (Supplementary Figure S4B), H NMR
(Supplementary Figure S5B) and 13C NMR (Supplemen-
tary Figure S6B). HPLC co-injection analysis of D-allo-
ct6A and L-ct6A mixture showed that both isomers can be
clearly separated (Supplementary Figure S8).
Chemical synthesis of t6A
t6A nucleoside was obtained in good yield according to
the procedure reported by Chheda et al. (25–27). In this
method, the exo-amine group of sugar-acetylated adeno-
sine was functionalized with an ethoxycarbonyl group us-
ing ethylchloroformate as the electrophile, affording a reac-
tive N6-carbamate derivative. The ethoxy group of this com-
pound was subsequently substituted with L-threonine to
Single crystal X-ray diffraction analysis
Colorless crystals of ct6A (0.30 × 0.10 × 0.05 mm) were
obtained from aqueous ethanol (1:1) solution. Single crys-
tal X-ray diffraction data were collected on a Bruker Smart
APEX2 diffractometer at 100 K, using an Incoatec Micro-
˚
focus Source I␮S Cu-K␣ (λ = 1.54178 A) as the source of ra-

2140 Nucleic Acids Research, 2017, Vol. 45, No. 4
diation. Integration of the data yielded 14 087 reflections to
a 2θmax angle of 142◦, of which 3384 were independent (Rint
= 0.040), and 3152 reflections with I ≥ 2σ (I). The crystal
structure was solved by the intrinsic phasing method. For-
mula: C15H18N6O7·H2O, Mw = 412.37, Z = 4, crystal sys-
tem: orthorhombic, space group: P212121, a = 6.355(1), b =
solution (typically 40 ␮l) containing 1 ␮g/␮l total RNA,
20 mM trimethylamine-acetate (pH 5.3), nuclease P1 (0.1
units for 40 ␮g of RNA), and BAP (0.16 units for 40 ␮g of
RNA) was incubated at 37◦C for 1 h.
Total nucleosides were analyzed using a Q Exactive
hybrid Quadrupole-Orbitrap mass spectrometer (Thermo
Fisher Scientific) equipped with an ESI source and Ul-
timate 3000 liquid chromatography system (Dionex). For
HILIC/ESI-MS (31) was performed using a ZIC-cHILIC
column (3 ␮m particle size, 2.1 × 150 mm, Merck Milli-
pore). The mobile phase consisted of 5 mM ammonium ac-
etate (pH 5.3) (solvent A) and acetonitrile (solvent B). Total
nucleosides (2 ␮g) or synthetic ct6A (5 pmol) dissolved in
90% acetonitrile were injected and chromatographed with
a flow rate of 100 ␮l/min in a multistep linear gradient:
90–40% B from 0 to 30 min, 40% B for 10 min, followed
by initialization to 90% B. Proton adducts of nucleosides
were scanned in positive polarity mode over an m/z range
of 110–700. ODS/ESI-MS (32) was performed on a Sun-
shell C18 column (2.6 ␮m particle size, 2.1 × 150 mm,
ChromaNik Technologies) with the same solvent system de-
scribed above. The gradient program was as follows: 0–40%
B from 0 to 30 min, 40% B for 5 min, followed by initializa-
tion to 0% B, at a flow rate of 75 ␮l/min. Total nucleosides (4
␮g) or synthetic ct6A (10 pmol) dissolved in LC/MS grade
ultrapure water (Wako) were injected. Positively charged
nucleosides were scanned over an m/z range of 200–500.
For higher-energy collision-induced dissociation (CID) in
ODS/ESI-MS, the following gradient program was used: 0–
15% B from 0 to 30 min with a curved gradient, 15–60% B
from 30 to 35 min, 60% B for 10 min, followed by initial-
ization to 0% B, at a flow rate of 75 ␮l/min. Bases liberated
from nucleosides by in-source fragmentation (20 eV) were
scanned over an m/z range of 240–400, and further scanned
by data-dependent HCD with a normalized collision energy
of 30. Total nucleosides (12 ␮g) or synthetic ct6A (400 pmol)
dissolved in LC/MS grade ultrapure water were subjected
to this analysis.
◦
˚
8.173(1), c = 33.492(3) A, α = β = γ = 90 , V = 1739.4(2)
3
A , ρcalcd = 1.57 g cm−3, μ = 1.11 mm−1. Semi-empirical
˚
absorption correction based on multiple scanned equiva-
lent reflections (0.636 < T < 0.754) was applied. The final
anisotropic full-matrix least-squares refinement on F2 with
342 variables converged at Robs = 0.041, Rall = 0.045, wR
(F2) = 0.099 with residual electron density ꢀρmax = 0.23
eA−3 (ꢀρmin = -0.25 eA−3). Positions of all hydrogen atoms
were located on a difference Fourier map, and their coordi-
nates and isotropic displacement parameters were refined
without restriction, with the exception of hydrogen atoms
in water molecules, whose parameters were restrained to
the oxygen atom. The correct absolute configuration was
provided using a starting material with known configura-
tion. Data collection, reduction, and absorption correction
were performed using theAPEX2 (Version 2014.9, Bruker
AXS), SAINT-PLUS (Version 8.34A, Bruker AXS), and
SADABS (Version 2014/4, Bruker AXS) programs, respec-
tively. Structure solution and refinement were performed us-
ing the SHELXTL suite (Version 2014/6, Bruker AXS). The
CDCC 2015 software package (CSD system 2015, Version
5.36, Conquest, Mercury, Mogul) (30) was utilized for sub-
sequent molecular geometry analysis and visualizations.
˚
˚
Quantum chemistry computations
Ab initio energy scans around N1-C6-N6-C13 torsion an-
gle for ct6A were performed using the Gaussian09 pro-
gram with the DFT methodology (Gaussian 09, Revision
C.01, Gaussian, Inc., Wallingford CT, 2010). The Truhlar
exchange-correlation energy functional M062x and aug-cc-
pvdz Dunning’s basis set were applied to calculate the wave
functions. Initial atomic coordinates were those determined
for the crystal. For rigid energy scans, a 10◦ step was set, and
then for every conformer an EM062x was computed with-
out molecular geometry optimization. Differences in ener-
gies (ꢀE) compared to EM062x for the initial atomic coordi-
nates were calculated, and their highest values are defined
as energies of the rotation barrier. For relaxed energy scans,
the geometric parameter under investigation was set to a
starting value and then iteratively increased by 20◦. For each
scan step, the optimised molecular geometry EM062x was cal-
culated. Conformational energy (ꢀE) was calculated as the
difference between the particular conformer EM062x and the
lowest value of EM062x determined during the energy scan.
Epimerization of ct6A
For synthetic nucleosides, 4 nmol of L-ct6A or D-allo-ct6A
was dissolved in 400 ␮l of 100 mM sodium borate buffer
(pH 9.0) and incubated at 37◦C for 5 min, followed by
adding 10 ␮l of 3 M NaOAc (pH 5.2) to stop the reaction.
The nucleosides were desalted by using Oasis HLB cartridge
(3 ml, 30 mg, Waters), dried in vacuo and dissolved in water.
10 pmol of each sample was subjected to ODS/ESI-MS as
described above.
For natural ct6A, 160 ␮g of E. coli total RNA in 400 ␮l
of 100 mM sodium borate buffer (pH 9.0) was incubated at
37◦C for 15 min, followed by adding 40 ␮l of 3 M NaOAc
(pH 5.2) to stop the reaction and subjected to ethanol pre-
cipitation to recover total RNA. Nucleoside preparation
was carried out by one-step acidic digestion (31) as de-
scribed above. Four microgram of total nucleoside was sub-
jected to ODS/ESI-MS. For co-injection analysis, 10 pmol
of D-allo-ct6A was mixed.
Nucleoside analysis by mass spectrometry
Total RNA was extracted from E. coli strain BW25113
cultured in LB broth for 18 h, using the acidic phenol
method as reported previously (20). Total nucleosides were
prepared by digesting E. coli total RNA with nuclease P1
(Wako Pure Chemical Industries) and bacterial alkaline
phosphatase (BAP) (E. coli strain C75, TAKARA BIO
INC.) under acidic conditions (31). For these reactions, a
For amine adduct formation, 4 nmol of L-ct6A or D-allo-
ct6A, or 160 ␮g of E. coli total RNA was dissolved in 400 ␮l
of 100 mM Tris–HCl (pH 8.5) and incubated at 37◦C for 3 h.

Nucleic Acids Research, 2017, Vol. 45, No. 4 2141
Detailed structure of hydantoin ct6A
Further preparation was performed as the same procedure
described above.
The bond lengths and valency angles in the adenosine
and hydantoin moieties of ct6A (Supplementary Tables S3
and S4) are consistent with the respective values observed
in crystal structures of similar chemical compounds de-
posited in the Cambridge Structural Database (CSD) (Sup-
plementary Table S5) (30). The only significant difference
in the ct6A nucleoside is the C6–N6 bond, which is longer
RESULTS
Chemical synthesis of ct6A
As reported previously (24), 5ꢀ,3ꢀ,2ꢀ-O-acetylated derivative
of L-ct6A was synthesized by dehydration of sugar protected
L-t6A with acetic anhydride, while for L-ct6A preparation
carbodiimide chemistry was applied (20). Here, for multi-
milligram scale synthesis, we used polymer-bound EDC as
the activating agent for the carboxyl group of t6A. After the
reaction, the polymer-bound EDC and its urea form prod-
uct were removed by filtration, and the product was sepa-
rated by silica gel column chromatography. The purity of
the products was confirmed by HPLC analysis (Figure 1B).
Yields of L-ct6A nucleoside synthesized in water and DMF
as solvents were 49% and 51%, respectively. The resultant
L-ct6A nucleoside was characterized by UV (Figure 1C),
IR (Supplementary Figure S1B), 1H NMR (Supplementary
Figure S2B), 13C NMR (Supplementary Figure S3B) and
MS (Supplementary Figure S7); the results were identical
to those reported previously (20).
˚
(1.407(4) A) than the corresponding bond in the structure
6
˚
of t A nucleoside (1.3784(3) A). This difference can be ex-
plained by repulsive interactions between the carbonyl oxy-
gens of the ct6A hydantoin ring and the N1 and N7 nitro-
gens of the adenine moiety (Figure 1A). Similarly, elongated
˚
C6–N6 bond (1.413 A) was found in the crystal structure
of adenyl-N6-tetramethylsuccinimide (CSD-PULQIX), a
modified adenosine with a cyclic imide containing the N6
nitrogen (33).
The ureido functionality in the ct6A hydantoin ring ex-
hibits a significant electron density delocalization (Supple-
mentary Figure S9). The elongated C10–N6 bond (1.428(4)
˚
A) accompanied by the short C10–N11 bond (1.337(5)
˚
A) (Supplementary Table S3) indicate a possibility of tau-
tomeric conversion to the enol form with protonation at
O10, which is additionally stabilized by hydrogen bonding
with a nearby water molecule. Similarly, in the t6A structure,
the ureido group is also influenced by electron density delo-
calization as indicated by relatively long C10–N6 (1.4089(3)
We also synthesized diastereoisomer of ct6A with D-allo-
threonine (C␣ epimer) to examine possible epimerization of
L-ct6A under mild alkaline conditions. First, we obtained
D-allo-t6A nucleoside that was subjected to the cyclization
with the same procedure as for L-ct6A nucleoside. The resul-
tant D-allo-ct6A nucleoside was characterized by IR (Sup-
˚
˚
A) and short C10–N11 ((1.3216(3) A)) bonds (34,35).
The relative orientation of the hydantoin and adenine
rings in the ct6A crystal structure is described by the ␩(N1–
C6–N6–C13) dihedral angle. It adopts value of -52.7(5)◦,
clearly indicating twisting arrangement of those rings (Fig-
ure 2A and Table 1). It is stabilized by the repulsive interac-
tions between carbonyl oxygens of the hydantoin ring and
the N1 and N7 nitrogens of the adenine moiety (Supple-
mentary Figure S9). This repulsion restricts free rotation
around C6–N6 bond. Indeed, an ab initio energy rigid scan
over the ␩ angle revealed two rotation barriers around η =
0◦ (ꢀE = 23.0 kcal) and η = 180◦ (ꢀE = 17.7 kcal) aris-
ing from the plane arrangement of these two rings (Figure
3). The dihedral angle η = −52.7(5)◦ observed in the crys-
tal is close to the most stable conformation determined by
the rigid energy scans (Figure 3). Similar rotation barriers
were also observed by the relaxed energy scan (Supplemen-
tary Figure S10). The corresponding twisted arrangement
of two rings was found in the crystal structure of adenyl-
N6-tetramethylsuccinimide (CSD-PULQIX) with a corre-
sponding dihedral angle η = −43.3◦ (33).
1
plementary Figure S4B), H NMR (Supplementary Fig-
ure S5B) and 13C NMR (Supplementary Figure S6B). L-
ct6A and D-allo-ct6A nucleosides were clearly separated by
HPLC analysis (Supplementary Figure S8).
Hydantoin isoform of ct6A revealed by crystal structure
The ct6A nucleoside crystallized as the monohydrate in an
orthorhombic crystal system of space group P212121. Ad-
ditional X-ray crystallography data are summarized in Sup-
plementary Tables S1–S5. Strikingly, the crystal structure
showed that the N6-substituent of ct6A adopts the hydan-
toin form (Figure 2A), rather than the previously assigned
oxazolone form (Figure 1A). This observation prompted us
to check whether the isomerization occurs during the crys-
tallization. However, careful analysis of ct6A nucleoside by
TLC and HPLC before and after crystallization (data not
shown) revealed that no isomerization occurred during this
process.
In this structure, the hydrogen bonds between molecules
play a vital role in crystal packing (Supplementary Table
S6). In particular, the N3 nitrogen forms a strong inter-
molecular bond with 5ꢀ-OH in the ribose, stabilizing the
syn conformation of the ct6A base (Figure 2A). The water
molecule localized in the crystal links O3ꢀ of the ribose and
O10 in the hydantoin ring from another molecule. These in-
teractions enable formation of unique crystal packing with
an infinite layer of sugar rings perpendicular to the [001] di-
rection (Figure 2B).
The ribose of ct6A adopts the C2ꢀ-endo pucker confor-
mation (Figure 2A) with the values of endocyclic torsion
angles ν0 − ν4 shown in Table 1. The pseudorotation phase
angle and puckering amplitude (36,37) are P = 160.9(3)◦
and τm = 37.2(2)◦, respectively. Because the torsion angle
about the C4ꢀ–O4ꢀ bond is only 0.4◦, the conformation of
2
the ribose moiety is almost exactly E. Similar sugar ring
puckering was also observed in the t6A crystal structure,
with P = 157.8◦ and τm = 31.8◦ (35).
Conformation of the 5ꢀ-hydroxymethyl group of the ct6A
nucleoside is described by the dihedral angle ␥ and ␥(–) (37)
and their values (Table 1) define a gauche(+) conformation

2142 Nucleic Acids Research, 2017, Vol. 45, No. 4
Figure 2. Crystal structure of chemically synthesized ct6A. (A) Crystal structure of the hydantoin isoform of ct6A. Displacement ellipsoids were drawn at
the 50% probability level. Hydrogen atoms are represented by circles with an arbitrary radius. (B) Crystal packing of ct6A. Hydrogen bonds are indicated
by dashed blue lines.
5ꢀOH of sugar moiety and N3 nitrogen of heterobase. This
type of interaction has not been found in crystal structure
of the t6A nucleoside which adopts the anti conformation
around the N-glycosidic bond (35).
Table 1. Selected dihedral angles (◦) of ct6A and t6A
Dihedral angles
ct6A
t6A
N6 substituent arrangement
N1–C6–N6–C13 (␩)
− 52.7(5)
− 1.54(5)
Nucleoside conformation
C4ꢀ–O4ꢀ–C1ꢀC2ꢀ (ν0)
O4ꢀ–C1ꢀ–C2ꢀ–C3ꢀ (ν1)
C1ꢀ–C2ꢀ–C3ꢀ–C4ꢀ (ν2)
C2ꢀ–C3ꢀ–C4ꢀ–O4ꢀ (ν3)
C3ꢀ–C4ꢀ–O4ꢀ–C1ꢀ (ν4)
O5ꢀ–C5ꢀ–C4ꢀ–C3ꢀ (␥)
O5ꢀ–C5ꢀ–C4ꢀ–O4ꢀ (␥-)
C4–N9–C1ꢀ–O4ꢀ (␹)
− 23.1(4)
35.9(3)
− 34.1(3)
22.0(4)
0.4(4)
48.4(4)
− 71.0(4)
51.0(4)
− 21.31(5)
Spectroscopic characterization of ct6A
31.72(5)
− 29.49(5)
17.99(5)
1.95(5)
57.23(5)
The ct6A nucleoside was further characterized by NMR
1
spectroscopy. All H and 13C resonances of ct6A were un-
1
ambiguously assigned by a combination of H,1H-COSY,
1H,1H-NOESY, 1H,13C-HSQC and 1H,13C-HMBC experi-
ments, all of which are identical to our published data (20).
However, these NMR analyses did not provide direct evi-
dence allowing differentiation of the two isoforms of ct6A.
To obtain a signature of the hydantoin isoform of ct6A, we
further recorded 1H,15N-HSQC and 1H,15N-HMBC spec-
tra for the ct6A nucleoside to determine 15N chemical shifts
(Table 2 and Supplementary Figure S11). The chemical
shifts of N6 (163 ppm) and N11 (87 ppm) of ct6A were in
− 63.76(5)
− 153.41(5)
around the exocyclic C4ꢀ–C5ꢀ bond. Similar conformation
was also found in the t6A crystal structure (35).
The glycosidic dihedral angle χ = 51.0(4)◦ (Table 1)
clearly indicates the syn conformation of the ct6A base. It
is stabilized by the intramolecular hydrogen bond between

Nucleic Acids Research, 2017, Vol. 45, No. 4 2143
Figure 3. Ab initio energy calculation for the rigid scan over the dihedral angle η. The rigid energy scan over the dihedral angle η(N1–C6–N6–C13) was
performed at 10◦ steps using the hybrid functional M062x and aug-cc-pvdz Dunning’s basis set. Initial atomic coordinates were obtained from the crystal
structure. ꢀE is calculated as the EM062x difference between the coordinates of each conformer and the initial coordinates. Structural models of ct6A are
depicted at the indicated η angles.
Table 2. 15N chemical shifts of ct6A and t6A in DMSO
Then, we conducted 1D NOE experiments to examine
orientation of the ct6A base relative to the ribose (41,42).
When H8 was irradiated, strong NOE effect was observed
to H2ꢀ rather than to H1ꢀ (Supplementary Figure S13 and
Supplementary Table S7), suggesting that ct6A predomi-
nantly takes anti conformation in solution. Supporting this
finding, a strong cross peak of H8–H2ꢀ, and a weak cross
peak of H8–H1ꢀ were also observed in 2D NOESY spec-
trum in D2O (Supplementary Figure S14). All of these re-
sults confirmed the anti conformation of ct6A. This ob-
servation is not consistent with that found in the crystal
structure where the ct6A base adopts the syn conformation.
The relative orientation of the hydantoin and adenine rings
could not be provided by NMR data analyses as there are
no vicinal coupling constants available for the Karplus-like
relationship.
ct6A
t6A
N-1
N-3
N-7
N-9
N-6
N-11
275
254
241
171
163
87
233
236
239
172
116
93
15N chemical shifts (ppm) relative to liquid NH3.
the range characteristic to urea-type nitrogens, consistent
with the analogous nitrogens in the N3-phenyl hydantoin
structure (38), supporting the idea that the hydantoin iso-
form actually exists in solution. In the case of the oxazolone
isoform, the C=N–Ar type nitrogen would appear at much
lower magnetic field (39).
We next obtained IR spectra of t6A (Supplementary Fig-
ure S1A) and ct6A (Supplementary Figure S1B). The char-
acteristic ureido carbonyl absorption at 1680 cm−1 in t6A
disappeared in the spectrum of ct6A, and was replaced by
two absorption bands at 1780 and 1730 cm−1. These absorp-
tions are in the same range as the hydantoin ring of 3-purin-
6-yl-hydantoins (40), further supporting the idea that the
synthetic ct6A adopts the hydantoin isoform.
Presence of hydantoin ct6A in E. coli tRNAs
To determine whether the hydantoin isoform of ct6A is ac-
tually present in cellular tRNAs, we conducted LC/MS
analyses to compare synthetic and natural ct6A. In reverse-
phase column chromatography (ODS) coupled with mass
spectrometry, the synthetic ct6A and natural ct6A in to-
tal nucleosides in E. coli tRNAs eluted at similar retention
times, ∼24 min (Figure 4A). When both specimens were co-
injected, we observed a single peak of mass chromatogram
at m/z 394 (Figure 4A). In addition, we conducted the same
MS analysis using hydrophilic interaction chromatography
(HILIC). The synthetic and natural ct6A co-eluted at the
same retention time as well (Figure 4B).
Next, we performed higher-energy collision-induced dis-
sociation (CID) analysis of the base-related ion (BH2+) of
both synthetic and natural ct6A. Product ions generated
from the BH2+ of ct6A (m/z 263) were almost identical
for both specimens (Figure 5A and B). These product ions
could be assigned in the chemical structure of the hydantoin
isoform of the ct6A base (Figure 5C).
Solution structure of hydantoin ct6A
Conformation of hydantoin ct6A in solution was deter-
mined using NMR spectroscopy. The conformation of the
ribose was inferred by vicinal spin–spin coupling constants
1
determined by H NMR spectra in D2O (Supplementary
Figure S12). At room temperature, the C2ꢀ-endo and C3ꢀ-
endo conformers of ct6A were equally populated in solu-
tion, while in the crystal it was C2ꢀ-endo sugar pucker (Fig-
ure 2A). Regarding exocyclic C4ꢀ–C5ꢀ bond, we found that
gauche (+) is the preferred conformation (∼62%) of ct6A
in solution which is consistent with the result of the crystal
structure.

2144 Nucleic Acids Research, 2017, Vol. 45, No. 4
A
B
Synthetic ct6A (10 pmol)
Synthetic ct6A (5 pmol)
100
100
0
UV254
UV254
0
-100
100
-100
100
ct6A
(m/z 395)
ct6A
(m/z 395)
18.33
24.15
50
0
50
0
5
10
15
20
25
30
0
10
20
30
40
Time (min)
Time (min)
E. coli tRNA (4 g)
E. coli tRNA (2 g)
G
G
100
50
100
50
UV254
UV254
A
A
C
C
U
U
0
100
0
100
ct6A
ct6A
(m/z 395)
23.76
18.46
(m/z 395)
50
0
50
0
0
10
20
30
40
5
10
15
20
25
30
Time (min)
Time (min)
Coinjection
Coinjection
G
G
100
50
100
50
UV254
UV254
A
A
C
C
U
U
0
0
24.80
100
100
18.57
ct6A
(m/z 395)
ct6A
50
0
50
0
(m/z 395)
5
10
15
20
25
30
0
10
20
30
40
Time (min)
Time (min)
Figure 4. LC/MS co-injection analyses of synthetic and natural ct6A nucleosides. LC/MS analyses of synthetic ct6A and E. coli total nucleosides by
reverse-phase chromatography using octadecylsilyl resin (ODS, A), and by hydrophilic interaction chromatography (HILIC, B). UV traces (254 nm) and
mass chromatograms (m/z 395) of synthetic ct6A (top), natural ct6A in E. coli total RNA (middle), and co-injected natural and synthetic ct6A (bottom).
ct6A peaks in the UV trace are indicated by arrows.
Epimerization of hydantoin ct6A under mild alkaline condi-
tions
Tris-adduct of t6A appeared (20). These findings prompted
us to speculate that ␣-carbon of threonine residue in hy-
dantoin ring is epimerized under mild alkaline treatment
(Supplementary Figure S15). To confirm this speculation,
ct6A epimer (D-allo-ct6A) was synthesized chemically, and
compared to L-ct6A treated under mild alkaline condition
and/or incubated with Tris buffer. As shown in Figure 6A,
the synthetic L-ct6A gave a single peak in LC/MS. In ad-
dition, L-t6A, a hydrolyzed product of L-ct6A, was also de-
tected. When L-ct6A was incubated with a buffer adjusted
at pH 9 for 5 min, second peak of ct6A that eluted slower
When total RNA was digested into nucleosides by conven-
tional condition (pH 8.2), most of ct6A was hydrolyzed and
converted to t6A as reported (20). In this condition, we no-
ticed appearance of second peak of t6A that elutes faster
than the original peak of t6A in LC/MS chromatogram
(20). In addition, rest of ct6A also split and gave second
peak that elutes slower than the original ct6A peak (20).
When ct6A was incubated with Tris buffer, two peaks of

Nucleic Acids Research, 2017, Vol. 45, No. 4 2145
A
C
162.042
OH
219
100
90
80
70
60
50
40
30
20
10
0
H
74
N
162
263.090
N
N
O 235
O
174.078
219.063
136
74.061
N
136.062
191.105
N
235.094
H+
50
100
150
m/z
200
250
N
H
B
162.041
100
90
80
70
60
50
40
30
20
10
0
263.088
174.077
219.062
74.061
136.062
191.104
235.093
250
15m0 /z
50
100
200
Figure 5. CID spectra of the ct6A base. CID spectra of the ct6A base (m/z 263) derived from synthetic ct6A nucleoside (A) and E. coli total nucleosides
(B). (C) Product ions are assigned in the chemical structure of the hydantoin isoform of the ct6A base.
than the L-ct6A peak clearly appeared, while second peak
of t6A that eluted faster than the L-t6A peak was also de-
tected. Judging from the retention times of the synthetic D-
allo-ct6A and its hydrolysate, D-allo-t6A, these second peaks
were found to be D-allo-ct6A and D-allo-t6A, respectively
(Figures 6A and Supplementary Figure S15). Furthermore,
when the synthetic D-allo-ct6A was incubated under mild al-
kaline conditions (pH 9 for 5 min), L-ct6A and L-t6A clearly
appeared. To confirm these findings, L-ct6A treated with
mild alkaline was coinjected with D-allo-ct6A which was
treated or untreated with mild alkaline (Supplementary Fig-
ure S16). Second peaks of L-ct6A appeared under mild al-
kaline conditions coeluted with D-allo-ct6A and D-allo-t6A,
respectively.
pH 8.5, followed by epimer formation of t6A and its amine-
adducts.
DISCUSSION
The crystal structure of the ct6A nucleoside clearly demon-
strates that ct6A adopts a hydantoin isoform (Figure 2A),
rather than the previously predicted oxazolone isoform
(20). We also confirmed that the hydantoin ct6A is actu-
ally present in E. coli tRNAs (Figure 4). As shown here,
it was impossible to differentiate the two isoforms by a se-
ries of NMR and MS analyses, explaining the reason why
the chemical structure of ct6A was once assigned to be the
oxazolone isoform. For instance, the product ions of CID
spectrum of ct6A base (Figure 5A and B) could be assigned
to the hydantoin isoform as well as to oxazolone isoform
(Supplementary Figure S17) (20). Additionally, a pattern of
Next we examined epimerization of natural ct6A in E.
coli tRNAs. As reported previously (20), when E. coli to-
tal RNA was incubated with a buffer at pH 9 for 15 min, we
detected D-allo-ct6A along with D-allo-t6A (Figure 6B). D-
allo-ct6A in E. coli total RNA was confirmed by coinjection
with synthetic specimen. Moreover, we examined epimer-
ization of t6A upon adduct formation with amine. When
the synthetic or natural ct6A was incubated with Tris buffer
at pH 8.5, both L-type and D-allo-type Tris adducts of t6A
were generated (Figure 6C).
1
the scalar couplings observed in H–1H-COSY spectra of
both isoforms is very similar. In 1H–1H-COSY spectrum of
ct6A, we observed cross peaks between NH11–H12, H12–
H14, H14–OH14 and H14–H15 attributed to the threonine
side chain (Supplementary Figure S18), however, these cou-
plings are present in both hydantoin and oxazolone iso-
forms. No coupling was found between amino acid side
chain and adenine base. In this study, however, we have
found three spectroscopic features implying the presence of
a hydantoin group in ct6A. First, the unique chemical shift
Collectively, these findings demonstrate that epimeriza-
tion of ct6A takes place under alkaline conditions, even at

2146 Nucleic Acids Research, 2017, Vol. 45, No. 4
L-ct6A
D-allo-ct6A
A
Mild alkaline
(pH 9 for 5min)
-
-
+
+
100
50
L
D-allo
D-allo
L
ct6A
D-allo
L
(m/z 395)
0
100
L
L
D-allo
D-allo
t6A
50
0
L
(m/z 413)
D-allo
30
35
40
35
40
35
40
35
40
Time (min)
E. coli total RNA
B
Mild alkaline
(pH 9 for 15min)
-
+
+
(added with D-allo-ct6A)
100
50
UV trace
0
100
L
L
D-allo
L
ct6A
50
(m/z 395)
D-allo
0
100
L
L
L
t6A
50
0
(m/z 413)
D-allo
D-allo
35
40
40
35
40
35
Time (min)
L-ct6A
E. coli RNA
C
+
+
L-ct6A
D-allo-ct6A
E. coli RNA
D-allo-ct6A
D-allo-ct6A
100
50
0
L
L
L
D-allo
L
D-allo
Tris-adduct
(m/z 516)
L
D-allo
D-allo
D-allo
35
40
45 35
40
45 35
40
45 35
40
45 35
40
45
Time (min)
Figure 6. Epimerization of ct6A. (A) Epimerization of synthetic L-ct6A and D-allo-ct6A under mild alkaline condition. Each nucleoside was treated with
(+) or without (−) mild alkaline (pH 9 for 5 min), and subjected to ODS/ESI-MS analysis to detect ct6A (m/z 395, upper panels) and t6A (m/z 413, lower
panels). A peak corresponding to each epimer is indicated in the chromatographs. (B) Epimerization of natural ct6A in E. coli total RNA under mild
alkaline condition. E. coli total RNA was treated with (+) or without (−) mild alkaline (pH 9 for 15 min), digested into nucleoside under acidic condition,
and subjected to ODS/ESI-MS analysis to detect UV trace at 254 nm (top panels), ct6A (m/z 395, middle panels) and t6A (m/z 413, bottom panels). (C)
Epimerization of amine adduct of t6A. Synthetic ct6A epimers and E. coli total RNA were incubated with Tris buffer (pH 8.5) for 3 h. Nucleosides were
analyzed by ODS/ESI-MS to detect Tris-adduct of t6A (m/z 516).

Nucleic Acids Research, 2017, Vol. 45, No. 4 2147
of N6 atom (163 ppm) of ct6A in 15N NMR implies the ex-
istence of hydantoin structure (Table 2). Second, the chem-
ical shift of NH11 proton (9.72 ppm) of t6A moved toward
high magnetic field by 1 ppm (8.72 ppm) in ct6A (Supple-
mentary Figure S2), supporting the fact that hydrogen bond
of NH11–N1 is impossible in the hydantoin isoform (Figure
1A). Third, two characteristic absorption bands at 1780 and
1730 cm−1 in the IR spectrum of ct6A nucleoside (Supple-
mentary Figure S1) also indicate carbonyl bond stretching
in the hydantoin ring.
age of primary amines and mild alkaline conditions should
be avoided to handle and prepare tRNAs bearing ct6A.
As shown in the crystal structure of ct6A nucleoside, the
hydantoin ring adopts a twisted position with a torsion an-
gle of −52.7◦ against the adenine base (Figure 2A). From
the ab initio energy calculation of hydantoin ring rotation,
we identified four energy minimum rotamers with different
␩ dihedral angles (127.3◦, 67.3◦, −52.7◦ and −122.7◦) be-
tween the hydantoin and adenine rings (Figure 3). They all
have ‘twisted conformation’ due to repulsive interaction be-
tween carbonyl oxygens of the hydantoin ring and the N1
and N7 nitrogens of the adenine moiety. In fact, actual con-
formation of ct6A in the crystal structure adopts one of
these energy minimum rotamers. Although we do not have
any experimental evidence to determine the orientation of
the rings in solution, it is quite natural to assume that ct6A
adopts ‘twisted conformation’ in solution, because repul-
sive interaction between the two rings should come from
very basic chemical nature of ct6A. Strong rotational bar-
riers (23 and 17.7 kcal) of the planar arrangement of the
two rings of ct6A (Figure 3) restrict free rotation of C6–N6
bond. However, given that these barriers can be overcome
by heat energy at normal growth temperatures, the hydan-
toin ring might be able to rotate to some extent under phys-
iological conditions.
TcdA catalyzes ATP-dependent dehydration of t6A to
form ct6A on tRNA (20). Given that TcdA is a ubiquitin-
activating E1-like protein with ATPase activity, it is likely
that it first adenylates the carboxyl group of t6A to form
an activated ester intermediate, and subsequently allows the
N6 nitrogen of t6A to attack the C13 carbon to cyclize
the side chain with release of AMP. This reaction mech-
anism resembles the chemical synthesis of ct6A using a
carbodiimide-type activating reagent such as EDC. Intrigu-
ingly, a previous study reported hydantoin ring formation of
the t6A base using N,N’-dicyclohexylcarbodiimide (DCC)
(40). The activated ester intermediate of t6A with EDC or
DCC is analogous to the adenylate intermediate of t6A gen-
erated by TcdA.
ct6A is easily hydrolyzed to convert to t6A under mild al-
kaline conditions (20). In addition, primary amines such as
tris(hydroxymethyl)aminomethane or ethanolamine react
efficiently with ct6A to generate amides of t6A (20,43,44).
The hydantoin ct6A (Figure 1A) has two carbonyl carbons,
C10 and C13, both of which could potentially be attacked
by a water molecule or an amine. According to the gen-
eral hydantoin chemistry, the non-ureido carbonyl carbon
of a hydantoin compound is more susceptible to the re-
action with nucleophiles, yielding the hydantoic acid and
its derivatives (45–47). Thus, in the case of ct6A, C13 car-
bonyl carbon is naturally targeted by hydrolysis (20,40) and
amine adduct formation (20,43,44), whereas the ureido C10
is comparatively non-reactive. In addition, we here showed
that hydantoin ct6A is susceptible to epimerization under
mild alkaline conditions, indicating that H12 in hydantoin
ring is acidic enough for such process (48,49). It was shown
that hydantoin derivatives are racemized as the result of tau-
tomeric change (50). Just for 5 min incubation of L-ct6A at
pH 9, 20–30% of L-ct6A is epimerized to yield D-allo-ct6A,
and then it is hydrolyzed to form D-allo-t6A. The epimer-
ization of ct6A and t6A also takes place at pH 8.2 under
conventional nucleoside preparation (20). For tRNA prepa-
ration used for in vitro translation or other biochemical ex-
periments, aminoacyl-moieties attached to 3ꢀ termini of aa-
tRNAs isolated from the cell are frequently removed by
mild alkaline treatment. According to the original meth-
ods, tRNA fractions extracted from the cell are incubated
with 500 mM Tris-HCL (pH 8.8) (51) or 200 mM glycine
buffer (pH 10.3) (52) for their deacylation. In these condi-
tions, ct6A in tRNAs should be epimerized and converted to
L-t6A and D-allo-t6A as well as Tris- and glycine-adducts of
t6A, respectively. If D-allo-ct6A and D-allo-t6A have differ-
ent activity in protein synthesis from their L-isomers, trans-
lational activity of tRNAs having ct6A should be reconsid-
ered carefully. From these findings, we have learned that us-
As observed in the crystal structure of the t6A37-
containing anticodon stem–loop recognizing the AAG
codon at the ribosomal A site (16), the N1–NH11 hydrogen
bond extends a planar adenine base that stabilizes codon–
anticodon pairing as well as the anticodon stem–loop struc-
ture. One possible structural model of the oxazolone ct6A
(20) indicated that the oxazolone ring favors being fixed at
the planar position with the adenine base to extend its ␲-
conjugated system, enabling a strong stacking interaction
with the codon–anticodon helix. This model provides a po-
tential molecular basis for the contribution of ct6A to recog-
nition of the adenine base of the ANN codon. However, the
hydantoin ring of ct6A cannot adopt a planar position with
the adenine base due to strong coulombic repulsion between
the two rings. If there is no conformational restriction in
the decoding center, the hydantoin ct6A may adopt one of
the four twisted conformations on the ribosome. Taking ac-
count of the twisted conformation and low probability of
␲–␲ stacking of the hydantoin ring, it is an enigmatic issue
how the twisted hydantoin ring stabilizes codon–anticodon
interaction at the ribosomal A-site. Structural studies of 70S
ribosome complexed with tRNA with ct6A37 and its cog-
nate codon will be required to reveal the actual conforma-
tion of the hydantoin ring at the decoding center.
We here confirmed that the hydantoin isoform of ct6A is
present in E. coli tRNAs. Presumably, this isoform is present
in tRNAs from other organisms. However, we cannot rule
out the possibility that the oxazolone isoform of ct6A may
serve as an intermediate of the hydantoin isoform during
biogenesis of ct6A.
AVAILABILITY
Crystallographic data (excluding structure factors) for the
structures reported herein, have been deposited with the
Cambridge Crystallographic Data Centre under accession

2148 Nucleic Acids Research, 2017, Vol. 45, No. 4
14. El Yacoubi,B., Hatin,I., Deutsch,C., Kahveci,T., Rousset,J.P.,
Iwata-Reuyl,D., Murzin,A.G. and de Crecy-Lagard,V. (2011) A role
for the universal Kae1/Qri7/YgjD (COG0533) family in tRNA
modification. EMBO J., 30, 882–893.
15. Lin,C.A., Ellis,S.R. and True,H.L. (2010) The Sua5 protein is
essential for normal translational regulation in yeast. Mol. Cell. Biol.,
30, 354–363.
number CCDC 1458975. Copies of the data can be obtained
free of charge at http://www.ccdc.cam.ac.uk/.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
16. Murphy,F.V., Ramakrishnan,V., Malkiewicz,A. and Agris,P.F. (2004)
The role of modifications in codon discrimination by
ACKNOWLEDGEMENTS
tRNA(Lys)UUU. Nat. Struct. Mol. Biol., 11, 1186–1191.
17. Stuart,J.W., Gdaniec,Z., Guenther,R., Marszalek,M., Sochacka,E.,
Malkiewicz,A. and Agris,P.F. (2000) Functional anticodon
architecture of human tRNALys3 includes disruption of intraloop
hydrogen bonding by the naturally occurring amino acid
modification, t6A. Biochemistry, 39, 13396–13404.
We are grateful to all members of the Sochacka and Suzuki
laboratories for their technical support and many insightful
discussions.
18. Durant,P.C., Bajji,A.C., Sundaram,M., Kumar,R.K. and Davis,D.R.
(2005) Structural effects of hypermodified nucleosides in the
Escherichia coli and human tRNALys anticodon loop: the effect of
nucleosides s2U, mcm5U, mcm5s2U, mnm5s2U, t6A, and ms2t6A.
Biochemistry, 44, 8078–8089.
19. Lescrinier,E., Nauwelaerts,K., Zanier,K., Poesen,K., Sattler,M. and
Herdewijn,P. (2006) The naturally occurring N6-threonyl adenine in
anticodon loop of Schizosaccharomyces pombe tRNAi causes
formation of a unique U-turn motif. Nucleic Acids Res., 34,
2878–2886.
FUNDING
National Science Centre in Poland [UMO-2014/13/N/ST5
/01591 to M.M.]; Grants-in-Aid for Scientific Research on
Priority Areas from the Ministry of Education, Science,
Sports and Culture of Japan (to T.S.). Funding for open ac-
cess charge: Grants-in-Aid for Scientific Research on Prior-
ity Areas from the Ministry of Education, Science, Sports,
and Culture of Japan.
20. Miyauchi,K., Kimura,S. and Suzuki,T. (2013) A cyclic form of
N6-threonylcarbamoyladenosine as a widely distributed tRNA
hypermodification. Nat. Chem. Biol., 9, 105–111.
Conflict of interest statement. None declared.
21. Crain,P.F. (1990) Preparation and enzymatic hydrolysis of DNA and
RNA for mass spectrometry. Methods Enzymol., 193, 782–790.
22. Kim,S., Lee,H. and Park,S. (2015) The structure of Escherichia coli
TcdA (also known as CsdL) reveals a novel topology and provides
insight into the tRNA binding surface required for
N(6)-threonylcarbamoyladenosine dehydratase activity. J. Mol. Biol.,
427, 3074–3085.
23. Lopez-Estepa,M., Arda,A., Savko,M., Round,A., Shepard,W.E.,
Bruix,M., Coll,M., Fernandez,F.J., Jimenez-Barbero,J. and
Vega,M.C. (2015) The crystal structure and small-angle X-ray
analysis of CsdL/TcdA reveal a new tRNA binding motif in the
MoeB/E1 superfamily. PLoS One, 10, e0118606.
REFERENCES
1. Machnicka,M.A., Milanowska,K., Osman Oglou,O., Purta,E.,
Kurkowska,M., Olchowik,A., Januszewski,W., Kalinowski,S.,
Dunin-Horkawicz,S., Rother,K.M. et al. (2013) MODOMICS: a
database of RNA modification pathways–2013 update. Nucleic Acids
Res., 41, D262–D267.
2. Bjork,G. (1995) In: Soll,D and RajBhandary,UL (eds). tRNA:
Structure, Biosynthesis, and Function. American Society for
Microbiology, Washington, D.C., pp. 165–205.
3. Grosjean,H. (2009) In: Grosjean,H (ed). DNA and RNA Modification
Enzymes: Structure, Mechanism, Function and Evolution. Landes
Bioscience, Austin, pp. 1–18.
4. Suzuki,T. (2005) In: Grosjean,H (ed). Fine-Tuning of RNA Functions
by Modification and Editing. Springer-Verlag Berlin and Heidelberg
GmbH & Co. KG, Vol. 12, pp. 23–69.
5. Helm,M. and Alfonzo,J.D. (2014) Posttranscriptional RNA
modifications: playing metabolic games in a cell’s chemical Legoland.
Chem. Biol., 21, 174–185.
6. Agris,P.F., Vendeix,F.A. and Graham,W.D. (2007) tRNA’s wobble
decoding of the genome: 40 years of modification. J. Mol. Biol., 366,
1–13.
7. Yokoyama,S. and Nishimura,S. (1995) In: Soll,D and
RajBhandary,UL (eds). tRNA: Structure, Biosynthesis, and Function.
American Society for Microbiology, Washington, D.C., pp. 207–224.
8. El Yacoubi,B., Bailly,M. and de Cre´cy-Lagard,V. (2012) Biosynthesis
and function of posttranscriptional modifications of transfer RNAs.
Annu. Rev. Genet., 46, 69–95.
9. Gustilo,E.M., Vendeix,F.A. and Agris,P.F. (2008) tRNA’s
modifications bring order to gene expression. Curr. Opin. Microbiol.,
11, 134–140.
10. Schweizer,M.P., Chheda,G.B., Baczynskyj,L. and Hall,R.H. (1969)
Aminoacyl nucleosides. VII. N-(Purin-6-ylcarbamoyl)threonine. A
new component of transfer ribonucleic acid. Biochemistry, 8,
3283–3289.
11. Niimi,T., Nureki,O., Yokogawa,T., Hayashi,N., Nishikawa,K.,
Watanabe,K. and Yokoyama,S. (1994) Recognition of the anticodon
loop of tRNA(1)(Ile) by isoleucyl-transfer-RNA synthetase from
Escherichia coli. Nucleos. Nucleot. Nucl., 13, 1231–1237.
12. Yarian,C., Townsend,H., Czestkowski,W., Sochacka,E.,
Malkiewicz,A.J., Guenther,R., Miskiewicz,A. and Agris,P.F. (2002)
Accurate translation of the genetic code depends on tRNA modified
nucleosides. J. Biol. Chem., 277, 16391–16395.
24. Matuszewski,M. and Sochacka,E. (2014) Stability studies on the
newly discovered cyclic form of tRNA
N(6)-threonylcarbamoyladenosine (ct(6)A). Bioorg. Med. Chem.
Lett., 24, 2703–2706.
25. Chheda,G. and Hong,C.I. (1971) Synthesis of naturally occurring
6-ureidopurines and their nucleosides. J. Med. Chem., 14, 748–753.
26. Hong,C.I. and Chheda,G.B. (1972) In: Towsend,LB and Tipson,RS
(eds). Nucleic Acid Chemistry. John Wiley & Sons, NY, pp. 661–664.
27. Hong,C.I., Chheda,G.B., Dutta,S.P., O’Grady-Curtis,A. and
Tritsch,G.L. (1973) Synthesis and biological activity of analogs of
naturally occurring 6-ureidopurines and their nucleosides. J. Med.
Chem., 16, 139–147.
28. Adamiak,R.W. and Wiewio´rowski,M. (1975) The modified
nucleosides of tRNAs. Synthesis and spectra of some natural
ureidonucleosides. Bull. Acad. Pol. Sci., Ser. Sci. Chim., 23, 241–253.
29. Martin,D. and Schlimme,E. (1994) Preparation of ureidonucleosides
of the threonine isomers. Z. Naturforsch. C, 49, 834–842.
30. Allen,F.H. (2002) The Cambridge Structural Database: a quarter of a
million crystal structures and rising. Acta Crystallogr., Sect. B:
Struct. Sci., 58, 380–388.
31. Sakaguchi,Y., Miyauchi,K., Kang,B.I. and Suzuki,T. (2015)
Nucleoside analysis by hydrophilic interaction liquid chromatography
coupled with mass spectrometry. Methods Enzymol., 560, 19–28.
32. Suzuki,T., Ikeuchi,Y., Noma,A., Suzuki,T. and Sakaguchi,Y. (2007)
Mass spectrometric identification and characterization of
RNA-modifying enzymes. Methods Enzymol., 425, 211–229.
33. Arico,J.W., Calhoun,A.K., Salandria,K.J. and McLaughlin,L.W.
(2010) Tetramethylsuccinimide as a directing/protecting group in
purine glycosylations. Org. Lett., 12, 120–122.
34. Parthasarathy,R., Ohrt,J.M. and Chheda,G.B. (1974) Conformation
and possible role of hypermodified nucleosides adjacent to 3’-end of
anticodon in tRNA: N-(purin-6-ylcarbamoyl)-L-threonine riboside.
Biochem. Biophys. Res. Commun., 60, 211–218.
13. Phelps,S.S., Malkiewicz,A., Agris,P.F. and Joseph,S. (2004) Modified
nucleotides in tRNA(Lys) and tRNA(Val) are important for
translocation. J. Mol. Biol., 338, 439–444.

Nucleic Acids Research, 2017, Vol. 45, No. 4 2149
35. Parthasarathy,R., Ohrt,J.M. and Chheda,G.B. (1977) Modified
nucleosides and conformation of anticodon loops: crystal structure of
t6A and g6A. Biochemistry, 16, 4999–5008.
36. Altona,C. and Sundaralingam,M. (1972) Conformational analysis of
the sugar ring in nucleosides and nucleotides. A new description using
the concept of pseudorotation. J. Am. Chem. Soc., 94, 8205–8212.
37. Saenger,W. (1984) Principles of Nucleic Acids Structure. Springer:
NY.
38. Buchman,R. and Komoroski,R.A. (1980) A carbon-13 and
nitrogen-15 NMR study of some nitrogen heterocycles. J. Heterocyc.
Chem., 17, 1089–1092.
39. Westerman,P.W., Botto,R.E. and Roberts,J.D. (1978) Substituent and
medium effects on Nitrogen-15 shieldings of compounds with >C=N
bonds (imines, oximes and phenylhydrazones). J. Org. Chem., 43,
2590–2596.
43. Nishimura,S. (1972) Minor components in transfer RNA: their
characterization, location, and function. Prog. Nucleic Acid Res. Mol.
Biol., 12, 49–85.
44. Kasai,H., Murao,K., Nishimura,S., Liehr,J.G., Crain,P.F. and
Mccloskey,J.A. (1976) Structure determination of a modified
nucleoside isolated from Escherichia-coli transfer ribonucleic-acid -
N-(N-((9-beta-D-ribofuranosylpurin-6-Yl)carbamoyl)threonyl)2-
amido-2-hydroxymethylpropane-1,3-diol. Eur. J. Biochem., 69,
435–444.
45. Blagoeva,I.B., Pojarlieff,I.G. and Dimitrov,V.S. (1978)
Alkaline-hydrolysis of hydantoin, 3-methylhydantoin, and
1-acetyl-3-methylurea - effect of ring size on cleavage of acylureas. J.
Chem. Soc., Perkin Trans. 2, 887–892.
46. Meusel,M. and Gutschow,M. (2004) Recent developments in
hydantoin chemistry – a review. Org. Prep. Proced. Int., 36, 391–443.
47. Lopez,C.A. and Trigo,G.G. (1985) The chemistry of hydantoins. Adv.
Heterocycl. Chem., 38, 177–228.
40. Hong,C.I. and Chheda,G.B. (1975) Purinylhydantoins. Facile
conversion of the naturally occurring
N-(purin-6-ylcarbamoyl)-L-amino acids into 3-purin-6-ylhydantoins
and 3-cyclohexyl-1-(purin-6-ylcarbamoyl)hydantoins. J. Med. Chem.,
18, 79–84.
48. Dudley,K.H. and Bius,D.L. (1976) Buffer catalysis of the
racemization reaction of some 5-phenylhydantoins and its relation to
the in vivo metabolism of ethotoin. Drug Metab. Dispos., 4, 340–348.
49. Lazarus,R.A. (1990) Chemical racemization of 5-benzylhydantoin. J.
Org. Chem., 55, 4755–4757.
50. Dakin,H.D. (1910) The catalytic racemization of optically active
hydantoin derivatives and of related substances as the result of
tautomeric change. Am. Chem. J., 44, 48–60.
51. Von Ehrenstein,G. and Lipmann,F. (1961) Experiments on
hemoglobin biosynthesis. Proc. Natl. Acad. Sci. U.S.A., 47, 941–950.
52. Zubay,G. (1962) The isolation and fractionation of soluble
ribonucleic acid. J. Mol. Biol., 4, 347–356.
41. Rosemeyer,H., Toth,G., Golankiewicz,B., Kazimierczuk,Z.,
Bourgeois,W., Kretschmer,U., Muth,H.P. and Seela,F. (1990) Syn-anti
conformational analysis of regular and modified nucleosides by 1D
1H NOE difference spectroscopy: a simple graphical method based
on conformationally rigid molecules. J. Org. Chem., 55, 5784–5790.
42. Lynch,S.R. and Tinoco,I. Jr (1998) The structure of the L3 loop from
the hepatitis delta virus ribozyme: a syn cytidine. Nucleic Acids Res.,
26, 980–987.