Opened-ring adducts of 5-methylcytosine and 1,5-dimethylcytosine with amines and water and evidence for an opened-ring hydrate of 2′- deoxycytidine

Article abstract of DOI:10.1111/j.1751-1097.2011.00936.x

A variety of nucleic acid components and related compounds undergo photoreaction with water to form so-called photohydrates (e.g. uracil forms 6-hydroxy-5,6-dihydrouracil). However, the corresponding hydrates of 5-methylcytosine (a minor nucleobase in eukaryotic DNA) and related compounds have not been characterized. We report the preparation of opened-ring forms of such products for 5-methylcytosine (m5C) and 1,5-dimethylcytosine (DMC). This was accomplished via thermal reaction of ring-opened amine adducts (e.g. N-carbamoyl-3-amino-2-methylacrylamidine (IVa) or N-(N′-methylcarbamoyl)- 3-amino-2-methylacrylamidine (IVb)) produced by photo-induced reactions of m5C with ammonia or methylamine. When these adducts were treated with dilute trifluoroacetic acid, the amino group at the 3-position was replaced with a hydroxyl group; with IVa, N-carbamoyl-3-hydroxy-2-methylacrylamidine (Va) was formed, while reaction of IVb led to N-(N′-methylcarbamoyl)-3-hydroxy-2- methylacrylamidine (Vb). These compounds are ring-opened isomers of 5,6-dihydro-6-hydroxy-5-methylcytosine (Ia and IIa) and 5,6-dihydro-6-hydroxy-1, 5-dimethylcytosine (Ib and IIb). Compounds Va and Vb each undergo thermal ring closure reactions to form two unstable compounds with chemical and UV spectral properties expected for Ia and IIa (or Ib and IIb). The latter compounds have been identified as minor products in UV-irradiated aqueous solutions of m5C and DMC. Evidence is also presented that the 2′-deoxycytidine photohydrates coexist with an opened-ring form, possibly similar in nature to Vb. The nucleobase 5-methylcytosine (m5C) reacts photochemically with ammonia to form an opened-ring adduct (IVa); similar reactions occur with 1,5-dimethylcytosine and with methylamine. Subjection of such adducts to hydrolysis in dilute acid (e.g. 0.1% trifluoracetic acid) produces opened-ring hydrates of m5C (Va) and DMC. Photoreaction of the nucleoside 2′-deoxycytidine in water results in formation of closed-ring hydrates that equilibrate thermally with a compound that has the UV spectral properties expected for an adduct similar in nature to Va.

Full text of DOI:10.1111/j.1751-1097.2011.00936.x

Photochemistry and Photobiology, 2011, 87: 818–832
Opened-Ring Adducts of 5-Methylcytosine and 1,5-Dimethylcytosine
with Amines and Water and Evidence for an Opened-Ring Hydrate
of 2¢-Deoxycytidine
Martin D. Shetlar* and Janet Chung§
Department of Pharmaceutical Chemistry, School of Pharmacy, University of California, San Francisco, CA
Received 21 December 2010, accepted 15 April 2011, DOI: 10.1111/j.1751-1097.2011.00936.x
molecules (e.g. with water to form so called ‘‘photohydrates’’)
or with reactive groups in neighboring complexed proteins
ABSTRACT
A variety of nucleic acid components and related compounds
(e.g. with the e-amino moieties of lysine contained in histones).
undergo photoreaction with water to form so-called ‘‘photohy-
drates’’ (e.g. uracil forms 6-hydroxy-5,6-dihydrouracil). How-
ever, the corresponding hydrates of 5-methylcytosine (a minor
nucleobase in eukaryotic DNA) and related compounds have not
been characterized. We report the preparation of opened-ring
forms of such products for 5-methylcytosine (m5C) and
1,5-dimethylcytosine (DMC). This was accomplished via
thermal reaction of ring-opened amine adducts (e.g. N-carbam-
oyl-3-amino-2-methylacrylamidine (IVa) or N-(N¢-methylcar-
bamoyl)-3-amino-2-methylacrylamidine (IVb)) produced by
photo-induced reactions of m5C with ammonia or methylamine.
When these adducts were treated with dilute trifluoroacetic acid,
the amino group at the 3-position was replaced with a hydroxyl
group; with IVa, N-carbamoyl-3-hydroxy-2-methylacrylami-
dine (Va) was formed, while reaction of IVb led to N-(N¢-
methylcarbamoyl)-3-hydroxy-2-methylacrylamidine (Vb). These
compounds are ring-opened isomers of 5,6-dihydro-6-hydroxy-5-
methylcytosine (Ia and IIa) and 5,6-dihydro-6-hydroxy-1,5-
dimethylcytosine (Ib and IIb). Compounds Va and Vb each
undergo thermal ring closure reactions to form two unstable
compounds with chemical and UV spectral properties expected
for Ia and IIa (or Ib and IIb). The latter compounds have been
identified as minor products in UV-irradiated aqueous solutions
of m5C and DMC. Evidence is also presented that the
2¢-deoxycytidine photohydrates coexist with an opened-ring
form, possibly similar in nature to Vb.
This latter type of reaction leads to DNA-protein cross-
linking. Progress in understanding the photochemistry occur-
ring within the cell has drawn heavily from the results of a
large number of studies on model systems, including those on
photochemistry of the component pyrimidine nucleobases and
nucleosides. These studies have been extensively reviewed; for
older reviews on cyclobutane dimers, (6-4) adducts and
hydrates see, for example, Fisher and Johns (2,3), Wang (4),
Cadet and Vigny (5), Davies (6) and Ruzsicska and Lemaire
(7). More recent reviews, including within their purview the
photochemistry of DNA at the cellular level, are given by
Ravanat et al. (8) and Cadet et al. (9,10). Reviews of photo-
induced nucleic acid–protein cross-linking and the chemical
nature of the adducts responsible are given by Shetlar (11),
Saito and Sugiyama (12) and Meisenheimer and Koch (13).
Among the earliest studies of the photochemistry of
pyrimidine nucleobases and related compounds, in which the
product was isolated and characterized, was the reaction
of 1,3-dimethyluracil with water to form the ‘‘photohydrate’’
5,6-dihydro-1,3-dimethyl-6-hydroxyuracil (14). Since that time,
there have been a large number of studies of the photohydra-
tion reactions of uracil, cytosine, thymine and related com-
pounds, including nucleosides (for comprehensive reviews
covering much of this literature, see Fisher and Johns [2] and
Cadet and Vigny [5]).
In addition to thymine and cytosine, 5-methylcytosine
(m5C) is also found in eukaryotic DNA as a minor, but
important component base; it is involved in gene regulation
and other chromosomal transactions. (For an overview of the
roles of m5C in eukaryotic DNA function and of the
chromosomal machinery associated with its presence, see
Robertson [15].) Interestingly, the isolation and characteriza-
tion of hydrates similar in structure to Ia and Ib (Scheme 1)
have not been reported in the chemical literature for m5C or
for related compounds (e.g. nucleosides such as 5-methyl-
2¢-deoxycytidine [m5dC]) or derivatives of m5C (e.g. 1,5-
dimethylcytosine [DMC]). Based on indirect evidence, it has
been proposed that such species exist transiently as interme-
diates in observed chemical processes. For example, it has been
suggested (16) that unstable m5C hydrate (e.g. Ia) may be an
intermediate in a deamination reaction leading to thymine that
occurs when m5C is irradiated at a concentration of 0.1 mM in
INTRODUCTION
UV-irradiation of living cells leads to DNA photoproducts
that can act as precursors to mutagenesis, carcinogenesis and
cell death (1). A number of these products involve photore-
action of pyrimidine nucleobases within DNA to form
compounds of various types. For example, the pyrimidine
nucleobases can react with one another to form cyclobutane
dimers and so-called (6-4) adducts. In solution, reactions of
nucleobases within DNA can occur with surrounding solvent
*Corresponding author email: shetlar@cgl.ucsf.edu (Martin D. Shetlar)
§Current address: Janet Chung, Molecular and Cell Biology, Beckman Research
Institute, City of Hope, 1500 East Duarte Road, Duarte, CA, USA.
ꢀ 2011 The Authors
Photochemistry and Photobiology ꢀ 2011 The American Society of Photobiology 0031-8655/11
818

Photochemistry and Photobiology, 2011, 87 819
unbuffered aqueous solution. Vairapandi and Duker (17), in a
comment in passing, also indicated that irradiation of m5C as
a free base, using unspecified reaction conditions, resulted in
formation of thymine hydrates; they likewise suggested that
such hydrates resulted from deamination of an unstable
intermediate m5C hydrate.
upon irradiation of m5C (and Cyt) in the presence of the lysine
analog Na-acetyl-L-lysine ([21], fig. 1); ring closure of such
adducts produces the type of photoexchange adducts first
observed by Saito and co-workers (22) in the analogous
thymine-lysine system.
Reaction of m5C with methylamine (MA) leads to N-(N¢-
methylcarbamoyl)-3-amino-2-methylacrylamidine (IVb) (23).
Similarly, reaction with methanol leads to N-carbomethoxy-3-
amino-2-methylacrylamidine (24). In the case of the latter
compound, the hydroxy analog, namely N-carbomethoxy-3-
hydroxy-2-methylacrylamidine has also been isolated and
characterized (25). (For the numbering system in this type of
opened-ring compound, see IVb in Scheme 1.) The existence of
such compounds suggests that compounds with structures
similar to IVa and IVb, but with a hydroxyl group replacing
the amine functionality (as shown in Va and Vb in Scheme 2),
might be stable enough for characterization if they could be
prepared.
NH2
NH2
NH2
Me
Me
H
OH
Me
HN
N
N
H
H
NH2
O
O
OH
H
N
R
N
R
IIa: R = H
IIb: R = Me
Ia: R = H
Ib: R = Me
III
NH2
NH2
NH2
1
Me
Me
Me
2
N
N
N
3
NH2
NHMe
O
O
O
NH2
NHMe
IVc
NHMe
NH2
NH2
IVb
IVa
Me
N
Scheme 1.
It has been suggested that m5C hydrates are formed in a
OH
O
NHR
polynucletide context. Vairapandi and Duker (17) indicated
that irradiation of poly (dG-m5dC): poly (dG-m5dC),
(a double stranded polydeoxyribonucleotide containing alter-
nating deoxyguanine and m5dC residues) produced m5C
hydrates. The evidence for this conclusion was that treatment
of the irradiated polynucleotide with Escherichia coli endonu-
clease III released a mixture of the cis and trans isomers of the
thymine hydrate. It was suggested that these two hydrates
resulted from deamination of the cis and trans isomers of the
corresponding hydrates of m5C (Ia and IIa) within the milieu
of the polynucleotide and that these precursors were produced
by photochemical reaction of m5C residues, contained in the
polynucleotide, with water. However, Zuo et al. (18) examined
the photochemical reactions in the same polydeoxyribonucle-
otide and found no evidence for release of thymine hydrates
upon treatment with E. coli endonuclease III. The reasons for
these differing results are unclear.
Va: R = H
Vb: R = Me
NH2
Me
HN
OH
VI
Scheme 2.
Detailed studies of the photochemical reactions of m5C in
water have revealed that the predominant product in this
system is very different in nature than the type of hydrates
depicted as Ia and IIa. An early spectroscopic study (19)
showed that irradiation of m5C in aqueous solution leads to
rapid production of a compound with an absorption maxi-
mum at k = 285 nm. In later work (20), the compound
responsible for the observed spectroscopic behavior was
isolated and characterized; it was identified as 3-amino-
2-methylacrylamidine (III). Similar types of product were
shown to form in the corresponding photoreactions of m5dC
with water. Compounds with properties corresponding to
those expected of Ia and ⁄ or IIa or their nucleoside analogs
were not identified in these studies.
In the following, we describe the results of experiments that
prove this supposition correct. Using authentic samples of
such opened-ring water adducts, it was then determined that,
in fact, detectable amounts of such hydrates were not produced
when m5C or DMC are irradiated in water. However, in other
experiments, we found that standing of solutions of Va and Vb
at )20ꢁC slowly produced unstable compounds with chemical
and UV spectroscopic properties expected of closed-ring water
adducts. Below, we describe experiments indicating these latter
compounds are formed as minor products when m5C and
DMC are irradiated in water. We also describe results of work
indicating that an opened-ring isomer of the photohydrates of
2¢-deoxycytidine (dCyd) coexists with parent hydrates; it may
have a structure similar to those depicted in Va and Vb.
Furthermore, we present results of some experiments suggest-
When m5C and related compounds are irradiated in the
presence of amines, ring-opened adducts are formed. Indeed,
such opened-ring compounds can be detected spectroscopically

Stock solutions of aqueous DMC (20 mM) were similarly used for
820 Martin D. Shetlar and Janet Chung
ing that hydrates of cytosine (Cyt) and 1-methylcytosine
(1MC) can undergo ring-opening reactions.
preparation of solutions containing DMC (1 mM) and either ammonia
(100 mM) or methylamine (100 mM). As above, the reactants were
dissolved in a mixture of acetonitrile ⁄ water (70% ⁄ 30% vol ⁄ vol) and
the pH values of the solutions to be irradiated were near 7.
MATERIALS AND METHODS
Product preparation, isolation and purification in the m5C-ammonia
system. In a typical run, 160 mL of solution 1 mM in m5C and 100 mM
in ammonia (pH 7.2) (contained in 70% ⁄ 30% vol ⁄ vol CH3CN ⁄ H2O)
was irradiated for 4 h as described above. The UV spectrum of a
sample of the photoreaction mixture, after five-fold dilution in water,
showed greatly enhanced absorbance centered at 305 nm. One
milliliter samples of unirradiated solution and the irradiated solution
were reduced to dryness by rotatory evaporation and redissolved in
distilled water. The initial analysis of the m5C-NH3 system was carried
out on Column B using the following gradient: 0 min: 0% MeOH;
1 min: 0% MeOH; 4 min: 25% MeOH; 7 min: 25% MeOH; 7 min
15 s: 0% MeOH; 12 min: 0% MeOH. The aqueous eluent was 100 mM
NaCl; peak detection was at 274 nm. Under these conditions, the
parent m5C peak elutes at 3.5 min, while III in the irradiated sample
elutes at 2.3 min and the desired putative adduct, termed PA, elutes as
the major peak at 4.5 min; no other product peaks, aside from very
minor amounts of the ureidoacrylonitriles of m5C (27), were noted in
the HPLC elution profile. (The identification of the peak correspond-
ing to m5C-ammonia adduct was based on its absorption maximum at
k = 305 nm, while identification of III was based on the chromato-
graphic behavior and UV spectroscopic properties of an authentic
sample prepared as described in Celewicz and Shetlar [20].) It was
also determined that about 65% of parent m5C had been converted
to product, based on areas of the parent peaks after 0 and 4 h
irradiation.
After reducing the remaining irradiated solution to dryness via
rotatory evaporation, the residue was redissolved in 1.5 mL of distilled
water to give a cloudy solution. This solution was filtered and a trial
run was made on Column A using a 20 lL injection with 98% (10 mM
sodium phosphate buffer, pH 7.5) ⁄ 2% MeOH flowing at 4 mL min)1
as eluent. Under these conditions, III eluted at 5.9 min, m5C eluted at
8.9 min and the desired PA eluted as a broad peak at 9.6 min. The
concentrated reaction mixture was then partitioned in 0.5 mL batches
on Column A using the same elution conditions. With the greater
injection volume, the m5C peak eluted as a sharp peak centered at
8.8 min, within the envelope of the broad peak corresponding to PA,
which eluted between 7.2 and 10.5 min. The two materials eluting in
this latter time interval were collected separately and the fraction
containing PA was purified multiple times to eliminate the sharp peak
corresponding to m5C. (A variety of chromatographic separation
conditions were explored in attempts to find conditions to cleanly
separate m5C from PA when 0.5 mL volumes of concentrated reaction
mixture were injected. However, while several worked well with small
injection volumes, none gave clean separations with the larger volume.)
Prior to further studies, salt was removed from PA. The collected
purified PA was reduced to dryness via rotatory evaporation. The
residue was then shaken with 10 mL of methanol and the resulting
suspension was placed in a 15 mL plastic centrifuge tube; this was
repeated with a 5 mL portion of MeOH. After spinning insoluble salt
down to a pellet, the methanolic supernatant containing PA was
removed and stored in solution in a freezer at )20ꢁC. Samples of PA
were removed as needed for spectroscopic and chemical studies.
Product preparation, isolation and purification in the m5C-
methylamine system and the DMC-methylamine system. Since the
protocols for the preparation, isolation and purification of the m5C-
MA adduct (previously shown to be IVb [23]) and the DMC-MA
adduct are similar to those given for the m5C-NH3 system, they are
provided in Appendix S1, rather than in the body of this paper.
Studies on the conversion of PA to PH, a putative opened-ring isomer
of 5,6-dihydro-6-hydroxy-5-methylcytosine. In spectrophotometric
studies of the behavior of PA when dissolved in 0.1% trifluoroacetic
acid (TFA), we found that the characteristic spectrum of this
compound (kmax = 305 nm) disappeared over a period of 16 min
and was replaced by a low absorbance band with a maximum at about
276 nm. During this time, the absorbance at 306 nm decreased from
2.11 to 0.05; the absorbance at 275 nm after 16 min was about 0.2.
When PA was treated with 0.01% TFA, the rate of disappearance of
parent compound was much smaller; after 48 min of incubation, the
absorbance due to PA was 0.46. When PA was treated with methanolic
General aspects. 5-Methylcytosine hydrochloride, 1,5-dimethylcyto-
sine, cytosine, 1-methylcytosine and 2¢-deoxycytidine were purchased
from Sigma (St. Louis, MO), while 5-methyl-2¢-deoxycytidine was
from R. I. Chemical (Orange, CA). Methylamine was from Aldrich
(Milwaukee, WI). Ammonium hydroxide and HPLC solvents were
from Fisher (Fair Lawn, NJ); NMR solvents were from Aldrich
(Milwaukee, WI). Preparative separations were done on a Shiseido
Capcell UG120 10 · 250 mm reverse phase column (5 lm particle size
[Yokohama, Japan]); in the following, this column is termed Column A.
Analytical HPLC used a Capcell UG120 4.6 · 150 mm reverse phase
column (5 lm particle size); hereafter we call this Column B. The HPLC
system used was a Rainin binary gradient pumping system (Emeryville,
CA) coupled to a Hewlett-Packard 1040A diode array HPLC detector
(Palo Alto, CA). Prior to injection of HPLC samples, they were
subjected to spin filtration, using Costar Spin-X micro-centrifuge filter
tubes containing a 0.2 lm nylon filter (Corning Incorporated, Corning,
NY). UV spectra were run on a Hewlett-Packard 8452A diode array
spectrometer or using the ‘‘on the fly’’ spectral capture capability of the
Hewlett-Packard diode array HPLC detector.
NMR spectra were run at 600 MHz on a Varian INOVA NMR
spectrometer (Palo Alto, CA). Electrospray ionization (ESI) mass
spectra were run on either a Waters Micromass ZQ4000 instrument
(Beverly, MA) or a Sciex API300 triple quadrupole electrospray
instrument (Toronto, Canada). High resolution mass spectra were run
on a Thermo Electron Corp. LTQ Orbitrap XL (Waltham, MA).
Irradiation methods. Preparative irradiations were done in the cold
with light centered at 254 nm, which was provided by unfiltered
Spectronics BLE-1T155 15 watt lamps (Spectronics, Westbury, NY)
housed in Spectroline XX-15A lamp holders. The solutions to be
irradiated were placed in 187 mL cylindrical quartz vessels (47.6 cm
length · 2.6 cm OD) from Southern New England Ultraviolet Com-
pany (Branford, CT). These vessels fit snugly between the two lamps in
a Spectroline XX-15A lamp holder placed in a ‘‘lamps up’’ orientation
and photoreactions were carried out on 160 mL of solution with the
irradiation vessels placed in this position; one end of the lamp housing
was elevated at about a 10ꢁ angle to the horizontal to avoid leakage of
the solution out the top of the stoppered vessels. During irradiation,
the reaction vessels were stoppered with appropriate glass taper seal
stoppers. All irradiations were done under air in a cold room at about
5ꢁC.
Some irradiations were carried out at 254 nm in a Southern New
England Ultraviolet Company RP-100 reactor. In this case, the
solution, contained in a 187 mL quartz tube, was placed at the center
of the reactor; during irradiation the reaction vessel was inserted into a
cylindrical Vycor shield.
Preparation of m5C-ammonia, m5C-methylamine and DMC-
methylamine solutions for irradiation. Commercially available m5C is
usually in the form of the hydrochloride salt. However, we conducted
our irradiations at pH 7 or slightly above. We prepared 20 mM stock
solutions of m5CÆHCl for use in making up solutions for irradiation.
This solution was kept refrigerated when not in use. (The concentra-
tion was verified by running the UV spectrum of a diluted sample in
phosphate buffer [10 mM, pH 7.5] [e273 of m5C is 6230 (26)].) Using
this stock solution, 500 mL portions of the solution to be irradiated
were made up at a time. Twenty five milliliter of stock m5CÆHCl was
diluted with 50 mL of 1 M NH3 or 50 mL of 1 M MA, each of which
had been adjusted to have a pH of 7.0. The pH of the resulting
solution (with an initial pH between 3.5 and 4) was then adjusted with
1 M NaOH until it was near pH 7. The volume of this solution was
brought to 150 mL in
a graduate cylinder and additional pH
adjustment was done by adding small amounts of 1 M NaOH until
the pH was above 7. The solution was then diluted with 350 mL of
HPLC grade acetonitrile; the measured pH of this final solution,
containing both water and acetonitrile, was above 7 in all cases. (Use
of this mixed solvent medium significantly increases the yield of the
desired amine adducts relative to the yield of 3-amino-2-acrylamidine
(III), which is the dominant product when m5C is irradiated in pure
water [20].)

Photochemistry and Photobiology, 2011, 87 821
0.1% TFA (prepared by diluting concentrated TFA in methanol), the
rate of disappearance was very slow; over a period of 8 min, about
1.6% of the absorbance at 306 nm had disappeared. Disappearance of
PA upon incubation in either 1 M ammonium formate (pH = 7) or
100 mM phosphate buffer (pH 7.3) was likewise very slow. These
results suggest that (i) the rate of this reaction is enhanced by
increasing the concentration of H3O+ and (ii) water is required in
order for the reaction to proceed at appreciable rates.
We examined the HPLC of PA that had been incubated in 0.1%
aqueous TFA for various times at room temperature and then diluted
three-fold with 100 mM sodium phosphate buffer (pH 7.3). For this
analysis, we used Column B with the following gradient: 0 min, 0%
MeOH; 0.75 min, 0% MeOH; 4 min, 15% MeOH; 5 min, 15%
MeOH; 5 min 15 s, 0% MeOH; 9 min, 0% MeOH. The aqueous
eluent was 10 mM Na2SO4 and the rate of flow was 2 mL min)1. Under
these conditions, the parent PA elutes at 3.1 min. The HPLC of
samples taken after various times of incubation (1.5, 13.5, 24 min)
indicated that the area of the peak associated with PA decreased with
RESULTS AND DISCUSSION
Introduction
Study of the implications of an observation in our laboratory,
concerning the spectroscopic behavior of the m5C-methyl-
amine adduct IVb when it is treated gently with dilute acid, has
provided a route to compounds that can be regarded as
opened-ring isomers of 5,6-dihydro-6-hydroxy-5-methylcyto-
sine (having cis and trans forms, depicted in Scheme 1 as
Ia and IIa) and 5,6-dihydro-1,5-dimethyl-6-hydroxycytosine
(Ib and IIb). Our original observation was spectroscopic in
nature and indicated that a new product was gradually formed
upon incubation of IVb with 0.1% TFA. When IVb when
dissolved in 0.1% TFA, the characteristic spectrum of this
compound (kmax = 305 nm) disappeared over a period of a
few min and was replaced by a low absorbance band with a
weak maximum at about 276 nm. Neutralization led to
recovery of absorbance at long wavelengths, however, the kmax
was shifted to the red, as compared with that of IVb, suggesting
that a new compound was present. (More detail about these
observations will be given later.) In the following, we describe
the studies that led to preparation and characterization of ring-
opened isomeric forms of the m5C and DMC hydrates.
increasing incubation time and that the area of
a new peak,
corresponding to a compound of unknown structure eluting at
6.3 min with kmax = 313 nm (termed PH in the following discussion),
correspondingly increased. After 24 min, the peak associated with PA
had almost completely disappeared from the HPLC chromatogram
and PH dominated the HPLC trace. (Note that this reaction cannot be
followed cleanly by UV spectroscopy, as both PA and PH absorb
strongly in the same region of the spectrum.)
Preparation, isolation and purification of PH for spectroscopic
characterization. For purposes of preparation of sufficient amounts
of the dominant product PH for mass spectral and NMR studies, we
used the following protocol. About 5 lmol of PA was taken to dryness
by rotatory evaporation at 40ꢁC. To the residual film in the
evaporation flask was added 500 lL of distilled water and then
1 mL of 0.1%TFA; the mixture was incubated at ambient room
temperature. The reaction was followed on Column B by HPLC using
90% (100 mM NaCl) ⁄ 10% MeOH flowing at 2 mL min)1. Samples
were diluted with equivalent amounts of 100 mM phosphate buffer, pH
7.3, prior to injection. Under these conditions, PA eluted at 1.3 min
and PH at 3.7 min. While most of the parent PA had been consumed
after 19 min, the reaction mixture was allowed to stand for 75 min,
after which very little parent reactant remained. The reaction mixture
was then taken to dryness by rotatory evaporation; the residue was
then dissolved in 0.5 mL of distilled water and filtered. The resulting
material was chromatographed on column B, using 76% (100 mM
NaCl) ⁄ 24% MeOH flowing at 2 mL min)1 as eluent. Prior to
injection, 100 lL portions of sample were diluted with 100 lL of
100 mM sodium phosphate buffer (pH 7.3). Peak PH, eluting between
1.7 and 2.2 min, was collected. Using rotatory evaporation at 40ꢁC,
the collected material was evaporated to dryness and then extracted
with 5 mL of MeOH. After centrifugation and decanting of the
methanolic extract, analytical HPLC of the extract on Column B, after
dilution with 100 mM sodium phosphate (pH 7.3) and using 90%
(100 mM NaCl) ⁄ 10% MeOH flowing at 2 mL min)1 for elution,
indicated that the product was quite pure. The amount of PH isolated
was estimated to be about 1.6 lmol via UV spectroscopy of an
aqueous dilute solution; this corresponds to an isolated yield of 32%.
In this calculation, it was assumed that the emax for IVb (39050) is a
good approximation for the corresponding emax of the product (see
section A3 in Appendix S1). No attempts were made to optimize the
reaction and workup conditions with a view of achieving higher yields
of isolated PH.
Characterization of PA isolated from the m5C-ammonia system
The UV spectrum of PA in distilled water showed that
kmax = 305 nm and kmin = 254 nm; the ratio A305 ⁄ A254 was
measured to be about 60. The ESI mass spectrum of PA
displayed peaks at 143.1 [M + H+] and 165.1 [M + Na+],
indicating that this product contained both m5C and NH3. The
high resolution ESI mass spectrum was consistent with proton-
ated PA having the molecular formula C5H11N4O (calculated:
143.0927; observed:143.0923). These mass spectralobservations
are consistent with PA being identified as IVa (Scheme 1).
The NMR spectral data, contained in Table 1, are also
consistent with identification of PA as IVa (N-carbamoyl-
3-amino-2-methylacrylamidine). Note that the structure IVa is
displayed in the Z conformation. The actual conformation has
not been definitively established. Results of quantum mechan-
ical calculations (Table 2) indicate that this conformation and
the alternative E conformation (displayed for esthetic reasons in
structures IVb and IVc in Scheme 1) have comparable energies
and, thus, do not supply definitive guidance on this question.
When IVa is heated in 0.1 M HCl for 15 min at 100ꢁC, it is
completely converted into m5C (major product) and Thy
(minor product). The HPLC analysis demonstrating this
(carried out after dilution of the heated sample with two
volumes of 100 mM phosphate buffer, pH 7.3) was done on
Column B using 100 mM Na2SO4 as eluent flowing at
2 mL min)1. Under these conditions, m5C eluted at 2.5 min
and Thy eluted at 4.4 min. This is consistent with analogous
observations, reported in (23), that found that IVb, the
corresponding methylamine adduct, is converted into DMC
and 1-methylthymine after similar treatment.
For the purposes of NMR study, an appropriate volume of
methanolic abstract was rotatory evaporated to dryness at ambient
temperature using a water aspirator as a pump; the resulting film was
used for preparation of NMR samples. NMR samples were prepared
within a couple of hours of having their spectra run; samples in D2O
were kept on ice until shortly before they were placed in the
spectrometer probe.
Conversion of IVb to PJ, a putative opened-ring isomer of 5,6-dihydro-
6-hydroxy-1,5-dimethylcytosine and its isolation and purification.
Hydrolysis studies analogous to those done for PA were done for IVb
as well. The results of this study, as well as the protocols for isolation,
purification and sample preparation for PJ, the main product of this
reaction, are similar to those discussed above for the acid hydrolysis
product of PA; these matters are discussed in Appendix S1.
Product characterization in the m5C-methylamine system
The opened-ring conjugate, formed in the reaction between
m5C and methylamine, has been previously isolated and

822 Martin D. Shetlar and Janet Chung
Table 1. Proton NMR chemical shift data for the photoproducts produced in the reactions of m5C with ammonia (IVa) and methylamine (IVb)
and of DMC with methylamine (IVc) and the opened-ring hydrates Va and Vb.
IVa*
IVb†
IVc
Va
Vb
C2CH3
C3H
C3NH2
C3NHCH3
CONH2
1.71(3)
8.17(t,1, J = 11.4)
8.05(b, 0.5)
1.71(3)
8.13 (t,1, J = 11.4)
8.04 (b,0.4)
1.71 (3)
8.37(d,1, J = 14)
1.74 (3)
8.35 (1)
1.73 (3)
8.32 (1)
8.00 (b,m,1)
8.81(b,0.8)
CONHCH3
C3NHCH3
CONHCH3
8.82 (b,0.8)
8.69 (b,0.6)
3.01(d,3, J = 4.4)
2.68 (d,3, J = 4.4)
2.76 (3)
2.68 (d,3, J = 3.6)
*Roman numerals correspond to the structures given in Schemes 1 and 2, while m denotes a multiplet of complex structure and b indicates a
broadened peak.
†Spectra of IVa, IVb and IVc were run in d6-DMSO, while the spectra of Va and Vb were run in D2O. The chemical shift values (d) are relative to
tetramethylsilane (TMS) in d6-DMSO or TSP (3-(trimethylsilyl)propionic-2,2,3,3-d4 acid, sodium salt) in D2O and are given in ppm; J values are in
Hz. Integrated values are rounded to nearest integer, except when such values deviate significantly from that expected (for some of the broad NH
proton resonances); then the actual measured integrated intensity is given. In all cases run in d6-DMSO, addition of a small amount of D2O to the
NMR sample resulted in disappearance or, in same cases, considerable loss in size of peaks assigned as hydrogens attached to nitrogen; under these
conditions, spin-spin splittings due to coupling to such hydrogens also disappeared.
characterized as having the structure IVb (23). The protocols
for its preparation, isolation and purification are given in
Appendix S1. (It should be noted that the overall protocol
given in Appendix S1 is a modified and improved version of
that given in [23].) Consistent with, but differing very slightly
from previously reported data (23), the UV spectrum of PB in
Table 2. Calculated aqueous energies of various compounds refer-
enced in the text.
Compound
Energy† (hydrated), kJ mol)1
Ia (trans)
Ib (cis)
IIa (trans)
IIb (trans)
IVa (Z)
IVa (E)
IVb (Z)
IVb (E)
Va (Z)
Va (E)
Vb (Z)
Vb (E)
VIIa
VIIb
VIIc
VIId
VIIIa
VIIIb
IX
)1340854.58
)1444042.93
)1340850.76
)1444052.71
)1288676.17
)1288672.79
)1391880.28
)1391875.75
)1340840.70
)1340811.85
)1443999.47
)1443995.64
)1289879.59
)1289874.07
)1289873.97
)1289814.88
)1342028.69
)1341975.42
)1340826.81
distilled water showed kmax = 305 and 228 nm and kmin
=
256 nm. The ESI mass spectrum of PB displayed peaks at 157.1
[M + H+] and 179.1 [M + Na+], indicating that this prod-
uct contained both m5C and methylamine. The high resolution
ESI mass spectra confirmed that protonated PB has a
molecular formula of C6H13N4O (calculated: 157.1084; ob-
served: 157.1080). The NMR spectra run in DMSO and
contained in Table 1, along with the UV and mass spectral
data given above, are all consistent with the previous identi-
fication of PB as N-(N¢-methylcarbamoyl-3-amino-2-methylac-
rylamidine (IVb in Scheme 1). Note that the peak at d = 8.13,
corresponding to C3H moiety is a triplet. This verifies that a
NH2 group is also attached to C3. The presence of a doublet,
in the signal corresponding to the CONHCH3 protons,
provides confirmation that these protons are incorporated
into the amide group. The chemical shift values of the
nonexchangeable protons given in Table 1 for IVb are near
the values of those found previously (23) where the NMR of
this compound was run in D2O.
†These values given are those evaluated at 0 K using Spartan 2010 for
the Macintosh. The methods used for these calculations were as
follows. A set of conformers was generated via molecular mechanics
using the molecular mechanics force field implementation of Spartan.
This was followed by optimization of the equilibrium geometry of each
of the conformers generated, using semi-empirical calculations
employing a PM3 model. After eliminating duplicate conformations
using the alignment tool built into Spartan, an optimized energy
As was the case for IVa, based on the data at our disposal,
we are unable to definitively assign this compound as either the
E form (shown in Scheme 1) or Z form. However, as discussed
below, results of quantum chemical calculations (Table 2)
suggest that the Z form is somewhat more stable than the E
form.
calculation in water was made using
a 3-21G basis set. After
elimination of high-energy conformations, the remaining conforma-
tions were subjected to an energy optimization in water using Hartree-
Fock calculations with the 6-31G* basis set. In a final stage, the set of
values obtained via this calculation were subjected to a final optimi-
zation in water utilizing B3LYP density functional calculations, again
using the 6-31G* basis set. The energies of the lowest energy
conformation of each molecular species are those tabulated above.
Spartan 2010 calculates the energy of solvated species by applying the
SM8 model of A. V. Marenich, R. M. Olsen, C. P. Kelly, C. J. Cramer
and D. G. Truhlar (2007) Self-consistent reaction field model for
aqueous and nonaqueous solutions based on accurate polarized partial
charges. J. Chem. Theory Comput., 3, 2011–2033, which adjusts wave
functions to account for solvent effects.
Product characterization in the DMC-methylamine system
As for the m5C-MA system, the detailed protocols for
preparation, isolation and purification of PC, the m5C-
methylamine adduct, are given in Appendix S1. The UV
spectrum of PC in distilled water showed that kmax = 319 nm;
there was a very broad kmin centered at about 262 nm. The ESI
mass spectrum of PC displayed peaks at 171.2 [M + H+] and

Photochemistry and Photobiology, 2011, 87 823
193.1 [M + Na+], indicating this product contained the
elements of both DMC and methylamine. The high resolution
ESI mass spectra of this compound indicated that protonated
using a protocol similar to that used for the preparation of Va
(Materials and Methods). Details of this preparation are given
in Appendix S1.
PC has
a
molecular formula of C7H15N4O (calculated:
The UV absorption spectrum displayed a maximum at
313 nm and a minimum at 257 nm; the ratio of A313 ⁄ A257 is
about 7.8. Molecular mass peaks, determined by ESI mass
spectrometry, were observed at 158.1 [M + H+] and 180.1
[M + Na+]). The high resolution ESI mass spectra indicated
that protonated PJ has a molecular formula of C6H12N3O2
(calculated: 158.0924; observed: 158.0921); this value is that
expected for a species with structure Vb (Scheme 2). Exam-
ination of the proton NMR spectrum (see Table 1) also
provides data consistent with the structure of PJ being Vb.
171.1240; observed: 171.1232); this is consistent with identifi-
cation of PC as IVc. The NMR spectral data, contained in
Table 1, along with the UV and mass spectral data, are also
consistent with identification of PC as N-(N¢-methylcarba-
moyl)-3-methylamino-2-methylacrylamidine (IVc). Note that
the peak at d = 8.37, corresponding to C3H moiety is a
doublet. This verifies that a hydrogen is contained in the amine
group that is also attached to C3. The presence of a doublet, in
the signal corresponding to the CONHCH3 protons, likewise
provides confirmation that these protons are incorporated into
the amide group.
The NMR spectrum contains
a proton resonance at
d = 8.32. This value is quite close to the values for the
C3H resonances in IVa, IVb and IVc, in which an NH2 or
NHCH3 group is likewise attached to C3. This indicates that
this resonance should also be assigned to C3H and supports
the supposition that PJ contains a vinyl group as exhibited by
Vb. There are two methyl group resonances, one at d = 1.73
and the other at d = 2.76. The chemical shift of the first is
near the values for the C2CH3 group observed for IVb, its
amino analog (d = 1.71), indicating that this methyl group is
attached to the vinyl linkage; the lack of spin-spin splitting in
this peak indicates that there is not a nonexchangeable
proton attached to the carbon adjacent to the methyl group.
The chemical shift of the second methyl resonance is
consistent with this group being attached to nitrogen;
however, this alone does not rule out the possibility that
the attachment site of the NHCH3 group could be at C3. A
chemical argument discounts this. Consider the stability of
the alternative structure to Vb, with the NHCH3 moiety at
C3 and an OH group attached to the carbonyl group (i.e. an
N-substituted carbamic acid). As discussed in (20), this type
of compound would be expected to be labile in aqueous
solution, decomposing to form 3-methylamino-2-methylacry-
lamidine and carbon dioxide. Indeed, the analogous com-
pounds, in which NHCH3 at C3 is replaced with either an
amino group or with an NH-2¢-deoxyribosyl group, cannot
be isolated. They undergo rapid reaction to form 3-amino-2-
methylacrylamidine and 3-(2-erythro-D-pentopyranos-1-yl)a-
mino-2-methylacrylamidine respectively (20). An additional
argument, based on results obtained by treatment of a crude
reaction mixture from photoreaction of DMC and ammonia
with TFA (see below), lends additional evidence supporting
the conclusion that the NHCH3 moiety is positioned as
shown in Vb.
Yields of IVa and IVb
The % yields of IVa and IVb, based on the amount of parent
consumed, were estimated to be about 58% (IVa) and 59%
(IVb). A detailed description of the protocols used for these
measurements will be found in Appendix S1.
The product PH, formed via acid treatment of IVa, is
N-carbamoyl-3-hydroxy-2-methylacrylamidine (Va)
The UV spectrum of purified PH in water displays, in addition
to the maximum at 313 nm, an absorption minimum at
k = 257 nm; the ratio A313 ⁄ A257 is about 9.9. ESI mass
spectrometry shows [M + H+] at 144.1, corresponding to the
molecular mass expected for a compound with the structure
shown as Va in Scheme 2. Comparing this structure to that
given in Scheme 1 for IVa, the parent compound in the acid
induced reaction, it can be seen that the NH2 at C3 in IVa has
been replaced by an OH in Va. A smaller peak occurs at 127.1,
corresponding to the [M + H+] expected for thymine, while
another yet smaller peak is seen at 126.1, which corresponds to
the [M + H+] value predicted for m5C. The high resolution
ESI mass spectra indicated that protonated PH has the
expected molecular formula of C5H10N3O2 (calculated:
144.0764; observed: 144.0768). The proton NMR spectrum
(Table 1), run in D2O, is also consistent with PH being
assigned the structure given by Va; it displays the expected
singlet methyl resonance at 1.74 ppm and a singlet C3H
resonance at 8.35 ppm. Both of these resonances are quite
close to those seen in the NMR spectrum of IVa, the amino
analog of Va. In addition to the resonances corresponding to
Va, there are weak resonances at d = 1.95 and 7.35; these
resonances have the same d values as those observed by
Sulkowska (28) for the C5CH3 and C6H protons of m5C in
D2O. A chemical argument, similar to that outlined below in
the section containing discussion of the DMC hydrate, can be
used to rule out an alternative structure to Va, in which an
amino group is attached to C3 and a hydroxyl group is
attached to the terminal nitrogen.
NMR spectra of Va and Vb in d6-DMSO
When the NMR properties of either Va or Vb are studied in
d6-DMSO (using TMS as standard), rather than D2O, the
complexity of the resulting spectra is greater than in D2O. In
the case of Va, two peaks appear at d = 1.65 and 1.74 with an
integrated area ratio of 3.14. The latter d value is close to that
seen for the methyl group when Va is run in D2O. In addition
there is a pair of peaks with d = 8.54 and 8.68 with an area
ratio of 3.06. One possibility is that these two pairs of
resonances correspond to C2 methyl and C3H groups in two
different structural isomers of Va. One reasonable interpreta-
tion is that they correspond to the C2 methyl and C3H
The product PJ, isolated by acid hydrolysis of IVb, is
N-(N¢-methylcarbamoyl)-3-hydroxy-2-methylacrylamidine (Vb)
The compound PJ, which is putatively an opened-ring form of
5,6-dihydro-1,5-dimethyl-6-hydroxycytosine, was prepared by

824 Martin D. Shetlar and Janet Chung
resonances in the E and Z structural isomers of Va. Similarly,
when Vb is run in d6-DMSO, there appears to be two sets of
peaks relating to two forms of the same compound. One set
displays d = 1.65 (s, 3), 2.58 (broad, 2.3) and 8.51 (s, 1) and
the second has d = 1.74 (s, 0.83), 2.63 (broad doublet, 0.97)
and 8.67 (s, 0.24). (Integrated areas for the various peaks for
Vb are referenced to a value of 3.0 for the d = 1.65 peak.)
Again, one possible interpretation of this data is that these two
sets of spectra correspond to E and Z structural isomers of Vb.
The NMR spectra of Va and Vb in d6-DMSO both display
peaks that disappear when a drop of D2O is added to the
sample. These likely correspond to the expected exchangeable
protons in these compounds (i.e. protons attached to nitrogen
or oxygen).
formation of Va. A portion of concentrated crude reaction
mixture, resulting from photoreaction of DMC with ammonia,
was made 0.05% in TFA and incubated at ambient room
temperature. After varying times, samples were withdrawn and
neutralized by mixing with an equal portion of 100 mM sodium
phosphate buffer (pH 7.3). The HPLC of 10 lL of the
resulting mixture was then run on Column B using 90%
(100 mM NaCl) ⁄ 10% MeOH flowing at 2 mL min)1 as eluent.
Under these conditions, parent DMC eluted at 2.0 min, the
putative DMC-NH3 adduct at 2.2 min and Va eluted at
3.7 min. After 60 min, about 75% of the parent adduct, as
measured by area, had disappeared and a new peak at 3.7 min
appeared, containing material with kmax at 313 nm. The
identity of the HPLC retention behavior and the UV spectro-
scopic properties of the material eluting at 3.7 min with the
corresponding properties of Va confirms that the TFA-
mediated hydrolysis of the putative DMC-NH3 adduct leads
to Va.
Incubation of N-(N¢-methylcarbamoyl)-3-methylamino-2-
methylacrylamidine (IVc) with 0.05% TFA
We also examined the effects of incubation of IVc with TFA.
A portion of aqueous purified IVc was diluted with an equal
amount of 0.1% TFA and allowed to stand at room
temperature for 6 min. After addition of an equal volume
of sodium phosphate buffer (100 mM, pH 7.3), a sample was
injected on Column B using 76% (100 mM NaCl) ⁄ 24%
MeOH flowing at a rate of 2 mL min)1 as eluent. The
product peak eluted at 3.4 min. Injection of a sample of IVb
that had been simultaneously treated in the same manner
yielded a peak that eluted after 3.2 min. The UV spectra of
the dominant product contained in each of the two samples
were identical. Since the product from the incubation of IVb
was identified as Vb, it can be concluded that IVc hydrolyzes
to the same product.
In a discussion above, chemical reasoning was used to reach
the conclusion that TFA-mediated hydrolysis of IVb results in
loss of the alkylamino group attached to C3. Although the
results of these experiments described immediately above are
not definitive, as they were not done with purified adduct, they
do provide suggestive supporting evidence for this conclusion;
hydrolysis at the alternative position, next to the carbonyl
group, would leave an NCH3 group at C3 and the resulting
product would not be Va.
Effect of treatment of 3-aminoacrylamidine (III) with TFA
The results of an earlier unpublished study in our laboratory
on the effects of dilute acid on the spectrophotometric
properties of III (20) indicated (in retrospect) that this
compound behaves similarly to the m5C-amine adducts. In
view of this similarity in behavior, we conducted further
studies to see if indeed the chemistry occurring is related. The
results of these studies suggest that III, as well as its nucleoside
analog (20), undergo reaction to form the compound VI or its
Z isomer (or a mixture of these two isomers). However, since
the problematic HPLC properties of the putative VI produced
preclude isolating sufficient material for NMR, this identifica-
tion must be regarded as tentative. The details of these studies
on acid hydrolysis of III and its nucleoside analog are
contained in Appendix S1.
Preliminary results from study of the TFA-mediated hydrolysis
of a photoreacted 1,5-dimethylcytosine-ammonia system
When 1 mM DMC is irradiated in the presence of 100 mM
NH3 at pH 7 in 70% acetonitrile ⁄ 30% water, UV spectro-
scopic and HPLC studies, conducted as described above,
indicated that an analogous opened-ring adduct was formed
in good yield. After 10 h irradiation at 254 nm, about 64%
conversion of parent to products was achieved, as evidenced
by calculations involving peak areas for the parent compound
in unirradiated and irradiated solution. (Column B was used
for the analysis, utilizing 94% [100 mM NaCl] ⁄ 6% MeOH
flowing at 2 mL min)1. Under these conditions, DMC elutes
at 2.3 min and the desired adduct elutes at 2.8 min.) However,
DMC elutes as a much broader peak than m5C on this
column. When preparative runs were attempted on Column A
using the same solvent system, it was found that the peak
corresponding to parent DMC completely enveloped the
adduct peak and separation of these two compounds was not
possible. Indeed, we were not able to find suitable preparative
separation conditions for this system utilizing the various
column and eluent systems available to us. Isolation and
characterization of DMC-NH3 adduct, putatively analogous
to IVa, but with NHCH3 attached to C3, remains to be
accomplished.
The hydrates Va and Vb are not formed when m5C and DMC
are irradiated in water
To determine if Va is produced in significant amounts when
the parent compound m5C is irradiated in distilled water, we
irradiated a 1 mM solution of this compound (pH = 7.6) at
room temperature for 2 h at 254 nm (68% conversion to
products) and immediately analyzed it via HPLC. Using
Column B with 76% (100 mM NaCl) ⁄ 24% MeOH flowing at
2 mL min)1, authentic Va elutes at 1.7 min. We found no
evidence for the presence of this compound in the reaction
mixture. A similar study with 1 mM DMC in distilled water at
pH 6.9 (41% conversion to products), using the same HPLC
conditions (Vb elutes at 3.0 min) yielded similar results,
suggesting that this compound is not a significant photoprod-
uct under these reaction conditions.
Although we were unable to isolate a pure DMC-NH3
adduct, we were able to obtain evidence strongly suggesting
that hydrolysis of the putative ammonia adduct results in

Photochemistry and Photobiology, 2011, 87 825
putative hydrate peak eluting at 0.85 min. The t1 ⁄ 2 for
disappearance of Va was about 5 min at this incubation
temperature. The putative closed-ring hydrates also displayed
instability at 40ꢁC; about 22% of the summed area for the
hydrates disappeared over the 20 min incubation interval after
the Va initially present had disappeared.
Ib and IIb may be intermediates in the thermal decomposition of
Vb to DMC, while Ia and Ib may similarly be intermediates in
the decomposition of Va
During storage of Va and Vb in methanolic solution at )20ꢁC,
decomposition of these compounds occurs slowly to form the
parent compounds m5C and DMC respectively. In each case,
however, there were also small amounts of other compounds
present with UV spectra reminiscent in profile to that of the
photohydrates of 2¢-deoxycytidine (see below). It can be
hypothesized that these compounds are Ia and ⁄ or IIa in the
case of the decomposition of Va and Ib and ⁄ or IIb in the
corresponding decomposition of Vb. One test of this hypoth-
esis would be to determine if, in fact, these other compounds
have Va or Vb as a direct precursor and whether these
compounds are direct precursors of the m5C or DMC also
seen in the solutions stored at )20ꢁC. It might be expected that
ring closure of an opened-ring hydrate (e.g. Vb fi Ib and ⁄ or
IIb) would precede elimination of water to form the final
pyrimidine product. We carried out such experiments by
following the thermal decomposition of Va and Vb as a
function of time of incubation at 40ꢁC in aqueous media. In
the case of Vb, we rotatory evaporated 100 lL of the
methanolic solution of Vb to dryness and took up the resulting
residue in 300 lL of 100 mM NaCl. The resulting solution was
incubated in a water bath maintained at 40ꢁC and 30 lL
samples were removed after successive time intervals and
subjected to HPLC on Column B using 100 mM NaCl and
MeOH flowing at 2 mL min)1 as gradient components. The
gradient (Gradient M) was as follows: 0 min: 0% MeOH;
2 min: 0% MeOH; 5 min: 25% MeOH; 6.75 min: 25%
MeOH; 7 min: 0% MeOH; 10 min: 0% MeOH. The peak
areas for the peaks of interest in each HPLC chromatogram
were measured at 313 nm (Vb), 243 nm (kmax for putative Ib
and IIb) and 279 nm (DMC). These peak areas were plotted as
a function of incubation time. We observed that (1) Vb (elution
time [te]: 7.3 min) disappeared via 1st order kinetics
(t1 ⁄ 2 = 10 min) over a 80 min time interval, (2) the sum of
the two peak areas corresponding to the putative Ib and IIb
(te: 1.5 min [kmax = 243 nm] and 2.0 min [kmax = 245 nm] or
vice versa) at first increased and then steadily decreased and (3)
the area corresponding to DMC (te: 5.1 min) correspondingly
increased with time at 40ꢁC. These results indicate that the
reaction in the incubated system can be represented by the
consecutive reaction sequence Vb fi (putative Ib, IIb) fi
DMC. While more detailed structural characterization of the
two unstable intermediate products must be done before they
can be definitively identified, these results are consistent with
the interpretation that they can be identified as Ib and IIb
formed via ring closure reactions of Vb. (It can be noted that
over the final 30 min period of incubation, after most of Vb
had disappeared, ca 24% of the summed peak areas for
putative Ib and IIb were converted to DMC. Thus, preparation
and maintenance of pure samples of these compounds for
NMR studies could require special care.)
Putative closed-ring hydrates are present in freshly irradiated
aqueous m5C and DMC solutions
As noted in the Introduction, closed-ring photohydrates of
m5C (Ia and ⁄ or Ib) have not been directly observed as
products, although indirect evidence suggesting formation of
these compounds has been presented. We have used the HPLC
properties of the putative closed-ring hydrates, described in the
previous section, to answer the following question. Are
products with the properties of putative hydrates formed
when either m5C or, alternatively, DMC is irradiated in
aqueous solution?
We irradiated 10 mL samples of m5C and DMC, each 1 mM
in concentration, for 32 min in a cold room (about 4ꢁC) at
254 nm; the samples were placed next to an unfiltered
Spectronics BLE-1T155 15 watt lamp. The samples were then
analyzed using Gradient M, described above. In the case of the
DMC system (5% conversion to product), three product peaks
were observed, in addition to small late eluting peaks
corresponding to the ureidoacrylonitriles of DMC (27). The
main product eluted at about 5.2 min (just prior to the DMC
peak eluting at 5.4 min) and had an absorption spectrum
similar in profile (kmax = 299 nm) to that expected for an
aminoacrylamidine (20). Two minor products eluted at 1.5 and
2.0 min, with the area of peak at 2.0 min, measured at 243 nm,
being about four times as large as the peak eluting at 1.5 min.
The HPLC retention times and UV absorption spectra
corresponding to these two peaks were identical to those of
the putative Ib and IIb described above. The summed area of
the two peaks, measured at 243 nm, was about 19% of the
area of the putative acrylamidine peak, measured at 299 nm.
Based on these observations, it can be stated that compounds
with the properties expected of Ib and IIb are indeed formed
when DMC is irradiated aqueous solution. An additional
experiment showed that, after heating for 10 min at 100ꢁC, the
peaks at 1.5 and 2.0 min disappeared from the HPLC
chromatogram. This indicates that the corresponding com-
pounds are unstable to elevated temperature (as, in general,
are photohydrates of pyrimidine bases [2]) and provides
additional support for the idea that the two putative hydrates
can, in actuality, be identified as Ib and IIb.
A similar analysis was conducted for the irradiated m5C
system. In this case, three product peaks eluting prior to m5C
were observed, two of them corresponding to the previously
observed isomeric forms of the aminoacrylamidine (III) of
m5C (20). (As for the DMC system, small peaks corresponding
to ureidoacrylonitriles [27] were also seen.) The two aminoac-
rylamidine peaks eluted at 1.7 (minor) and 1.8 min (major),
while parent m5C eluted at 2.2 min. The third product peak
eluted at 0.88 min and had a UV spectrum identical to that of
the dominant putative hydrate (te = 0.85 min) described
above for the system in which Va was incubated at 40ꢁC.
The peak area, measured at 233 nm, was about 0.9% of the
sums of the areas of the aminoacrylamidine peaks, measured
Analogous kinetic behavior was observed when Va was
similarly incubated at 40ꢁC. Using Gradient M (see above), the
dominant putative closed-ring hydrate (kmax = 233 nm)
eluted at 0.85 min, m5C at 2.3 min and Va at 6.0 min; a
minor putative closed-ring hydrate eluted at about 1.3 min and
had an area about 11-fold less than that of the dominant

826 Martin D. Shetlar and Janet Chung
at 285 nm. These results are suggestive that either putative Ia
or, alternatively, Ib is indeed produced upon irradiation of
m5C in aqueous solution, albeit with a very low yield. This
latter result is perhaps not surprising; the quantum yield for
the photohydration reaction of Thy (also containing a methyl
group at C5) at pH 6 is 3 · 10)6 ([2], p.180).
The results of an additional HPLC experiment, in which the
irradiated m5C solution was heated at 100ꢁC for 10 min prior
to injection, showed that the peak at 0.88 min disappeared
from the chromatogram, consistent with this peak being a
photohydrate with a structure given by either Ia or IIa.
solutions of Va upon standing, either in the cold or upon being
heated (see above).
In the first reaction in Scheme 3, IVa becomes partially
protonated in the presence of dilute acid (e.g. 0.1%TFA,
pH = 1.9) to form the iminium cation VIIa; as will be
discussed below, IVa and VIIa are postulated to coexist in a
pre-equilibrium. Iminium cations often have a pKa in the range
between 3 and 4 (29), while evidence has been recently
presented that an iminium cation with a pKa in the neighbor-
hood of 2.5 is involved in the ring closure of thymine-ammonia
adducts to form thymine hydrates (30).
On the mechanism of acid-promoted transformation of amine
adducts of m5C to opened-ring hydrates of m5C, then
closed-ring hydrates and finally m5C
+
NH2
NH3
NH2
H
Me
Me
Me
N
N
N
+
The experimental results discussed above provide several hints
about the pathway by which amine adducts of m5C are
converted to the corresponding water adducts in the presence
of dilute acid. In the following, we propose a partial
mechanism (shown in Scheme 3) consistent with what we have
observed for this reaction. For the sake of simplifying the
discussion, we will focus on the specific case of conversion of
IVa, the m5C-ammonia adduct to Va, the ring-opened water
adduct of m5C. However, the discussion can likely be
generalized to similar reactions of the various other amine
adducts of m5C and DMC.
O
NH2
+
O
O
NH2
NH2
NH2
NH3
NH2
VIIb
VIIc
VIId
+
NH3
NH2
Me
H
Me
H
N
O
N
O
O
O
NH2
NH2
H
H
NH2
NH2
NH2
VIIIb
Me
H
Me
H
Me
H
IX
N
N
N
+
H2O
HA
A-
+
-
Scheme 4.
NH3
+
O
NH2
NH2
O
NH2
H
NH2
H
NH2
O
O
IVa
VIIIa
VIIa
There are other possible structures resulting from
pH 7
N-protonation of IVa, resulting, for example, in the cations
shown in VIIb, VIIc and VIId in Scheme 4; however, the
calculated solvated energies of these species (Table 2) suggest
that VIIb and VIIc are somewhat less stable energetically than
VIIa in aqueous solution, while VIId is much less stable; these
results predict that VIIa will be the dominant protonated form.
Other structures analogous to VIIa, but containing an
unprotonated imine (HC=NH) moiety and a protonation site
on another nitrogen, are much less stable than VIIa; the
solvated energies for two such cases are E = )1289838.86 kJ
(NH2CONH+CNH2CHMeCHNH) and E = )1289802.46 kJ
(NH2CONCNH3+CHMeCHNH).
NH2
NH2
NH2
Me
Me
H
Me
N
N
N
H
OH
NH2
O
O
H
O
N
H
OH
N
H
Va (Z)
m5C
IIa
Scheme 3.
In constructing this mechanism, we used the calculated
The next step in Scheme 3 is an irreversible hydrolysis
reaction of VIIa to form VIIIa, resulting in loss of ammonia
and gain of water. This step likely involves addition of water to
the iminium ion functionality to form a protonated amino-
carbinol species, which then undergoes loss of ammonia to
form VIIIa. (It can be noted in Table 2 that VIIIa is calculated
to be considerably more stable than the alternative VIIIb, in
which the exocyclic amino group is protonated.) It is this
conversion of iminium moiety to carbonyl that gradually
depletes the concentration of VIIa, as well as that of IVa
existing in pre-equilibrium with VIIa. Experimentally, this is
manifested by the disappearance (over a period of 16 min at
room temperature) of the absorption spectrum characteristic
solvated energies of various species as an approximate guide in
selecting appropriate structures of the various intermediates
shown in Scheme 3. These energies were evaluated by quantum
mechanical calculations using Spartan 10 for the Macintosh.
Details are given in Table 2. (While values of the Gibbs free
energy at 298 K for the various solvated species listed in
Table 2 would provide a more rigorous guide for constructing
this mechanism, calculated values for this thermodynamic
quantity are not readily evaluated in aqueous solution.) In the
latter part of the mechanism, we show steps containing the
observed products, in which Va transforms to a mixture of Ia
and IIa and finally m5C, as is observed experimentally in

Photochemistry and Photobiology, 2011, 87 827
of opened-ring adduct IVa that is seen after it is placed in 0.1%
TFA. Neutralization of the solution containing VIIIa is
followed by conversion of this compound to the Z form of
Va, the product isolated immediately after this treatment;
examination of Table 2 predicts that this isomer is consider-
ably more stable than the alternate E form. Upon standing, Va
is converted to more stable closed-ring hydrates. These two
hydrates have comparable calculated solvated energies; thus
the final product is predicted to consist of a mixture of closed-
ring hydrates Ia and IIa. The DE for the reaction of Va to form
Ia, the most stable hydrate, can be estimated using data in
Table 2, from which a value of DE = )13.88 kJ is obtained.
Finally, in a final irreversible thermal reaction, the two
hydrates gradually decompose to produce m5C.
An extensive examination of the photohydration products
of cytidine (Cyd) was published in 1978 (31); in this study,
exchangeable hydrogens were replaced with deuterium prior to
irradiation. Irradiation was carried out in D2O and the
reaction followed by proton NMR. In this work, it was shown
that deuterium oxide added across the 5,6-double bond to
form a hydrate with OD attached to the 6-position of the
nucleosidic pyrimidine base; evidence was presented indicating
that two diastereomeric products were present in the NMR
sample. Experimental evidence in the literature is consistent
with the supposition that water also adds across the 5,6-double
bond in dCyd to give the corresponding diastereomeric
hydrates (for reviews see [2,5]). These two products can be
partially resolved via HPLC on an analytical reverse phase
column (see below). However, isolated purified dCyd hydrates
have not been studied in detail using NMR and other tools for
structural characterization. Their instability towards reversion
to parent compound and deamination to give 2¢-deoxyuridine
hydrates appears to have precluded structural studies; these
matters are discussed by Douki et al. (32).
In the course of follow-up experiments, building on studies
of the photoreactivity of dCyd with aliphatic amines in
aqueous solution (33), it was noted that irradiation of dCyd
in the absence of amines led to formation of a compound that
was strongly retained on a reverse phase column. This
compound was found to have a spectrum very similar in
profile to that of adducts of dCyd with ethylamine, but with a
different kmax. In the following, we present the results of some
exploratory studies, including UV spectroscopic and HPLC
experiments, which suggest that this compound corresponds to
an opened-ring form of dCyd hydrate.
Facile conversion of Va(E) to Va(Z) could be mediated by
the aldehyde species IX. Table 2 indicates that the computed
solvated energy of this species lies intermediate between the
corresponding energies of these two compounds.
As mentioned above, the half-lives of Va and Vb at 40ꢁC in
distilled water were about 5 and 10 min respectively. It was also
noted that at the same temperature, over a 20 min period when
little Va was present, about 22% of combined putative Ia and
IIa disappeared; in a similar situation, about 24% of combined
putative Ib and IIb disappeared over a 30 min period. These
kinetic results suggest that free energy barriers of significant size
exist at each step along the reaction pathways Va fi (putative
Ia, IIa) fi m5C and Vb fi (putative Ib, IIb) fi DMC.
Study of the rates of individual steps in these reactions as a
function of temperature would be useful, so as to quantify the
enthalpies and entropies of activation for each of these steps.
It is possible that other plausible mechanisms can be
proposed to explain the experimental results described in this
paper. However, the mechanism put forth in Scheme 3 appears
to be a good starting point, as it is consistent with experimental
data. In addition, the various proposed protonated intermedi-
ates (i.e. VIIa, VIIIa), appearing in the mechanism are predicted
by quantum chemical calculation to be the energetically most
stable in aqueous solution, as compared to various alternative
isomeric species that could be proposed in their place.
When the progress of photoreaction of an aqueous 0.2 mM
solution of dCyd (Xa) is followed using UV spectroscopy,
three phenomena are observed. The first is the loss of
absorbance at 270 nm, the kmax of dCyd and the second is
the increase in an absorbance band with kmax = 241 nm
(corresponding to production of photohydration products of
dCyd). However, in addition to these two absorption changes,
a weak absorption band appears that tails to the red; this band
extends to about 330 nm.
When HPLC of a 250 lL sample of a 0.2 mM solution that
had been irradiated for 90 min at 254 nm was run immediately
after photolysis, the chromatogram displayed in Fig. 1 was
obtained. This chromatographic separation was made on
Column B, using the following gradient with 100 mM NaCl
and MeOH as eluents: 0 min: 0% MeOH; 2 min: 0% MeOH;
6 min: 12% MeOH; 7 min: 12% MeOH; 7 min 15 s: 0%
MeOH; 10 min: 0% MeOH. The flow rate was 2 mL min)1. In
Panel (a), the HPLC trace using a detection wavelength of
241 nm (the absorption maximum of the putative dCyd
hydrates), dCyd elutes at 4.92 min and two diastereomeric
hydrates elute at 1.01 and 1.24 min. (In the following, we will
term these two products hydrates, even though definitive
characterization by NMR remains to be done. This is based on
the fact that each of these compounds, upon heating, is
converted to dCyd [along with a small amount of 2¢-deoxy-
uridine].) Two compounds with absorption spectra similar to
those of the hydrates elute at 1.75 and 3.40 min; in some runs,
the latter peak appears to be considerably broader than shown
in Fig. 1. These latter two compounds may be products
analogous to the O6,5¢-cyclo-5,6-dihydro-2¢-deoxyuridine
Evidence for an opened-ring form of the 2¢-deoxycytidine
hydrate
In the following section, we present preliminary results that
suggest that the two 2¢-deoxycytidine hydrates each undergo a
thermal reaction to produce the same opened-ring hydrate
isomer. Such a reaction is illustrated in Scheme 5, where Xb,
one of the hydrate diastereomers (in this case with OH
attached in the R configuration to the pyrimidine ring) reacts
to form a putative ring-opened product (perhaps with a
structure given by Xc).
NH2
NH2
NH2
H
N
N
N
k2
hν
H
H
k-1
k-2
OH
O
OH
O
O
N
dRib
NHdRib
Xc
N
dRib
Xb
Xa
Scheme 5.

828 Martin D. Shetlar and Janet Chung
Figure 1. The HPLC chromatogram resulting from injection of 250 lL of 0.2 mM aqueous dCyd that had undergone photoreaction. The injection
was made immediately after irradiation of the parent solution at 254 nm for 90 min; a Shiseido Capcell UG120 4.6 · 150 mm column was used for
separation. (a) HPLC trace when detection was done at k = 241 nm, while (b) shows the corresponding trace with detection at 311 nm.
(The identities of the various peaks, as well as the HPLC gradient and flow rate conditions used, are discussed in the text.) (c) Spectra of the peaks
shown as eluting at 3.7 and 7.4 min in (b). The spectra in (c) are normalized, thus appearing on the same scale.
products reported by Cadet et al. (34); however, because of the
small amounts in which they are produced and their instability
towards the conditions used in our work-up procedures, we
were unable to isolate them for further characterization. Of
particular interest here, however, are the two main peaks in the
chromatogram shown in Panel (b), where the detection
wavelength is 311 nm. The absorption spectra corresponding
to these two peaks (shown in Panel [c]), obtained by utilizing
the ‘‘on the fly’’ spectral capabilities of the diode array
detector, have maxima at wavelengths where they could
contribute to the redshifted weak absorption band that was
observed in the UV spectrum of the reaction mixture described
above. Indeed, the two spectra display a remarkable resem-
blance in profile to the corresponding spectra observed for
compounds Va and Vb.
and D2 (telution = 7.4 min), we irradiated 1 L of 0.2 mM dCyd
for 90 min at 254 nm in the cold in 165 mL batches as
described in Materials and Methods. This solution was taken
to dryness in stages using rotatory evaporation at 40ꢁC; the
residue was dissolved in 2 mL of distilled water. The resulting
solution was partitioned on Column A using the following
gradient with water serving as the aqueous eluent: 0% MeOH;
4 min: 0% MeOH; 6 min: 12% MeOH; 18 min: 12% MeOH;
21 min 15 s: 0% MeOH; 25 min: 0% MeOH. The flow rate
was 4 mL min)1. The fractions (dCH1 and dCH2), corre-
sponding to the two hydrates and eluting in peaks centered at
4.2 and 5.1 min, were collected and the HPLC of each freshly
purified fraction was run on Column B, using the following
gradient at 2 mL min)1: 0% MeOH; 4 min: 0% MeOH;
5 min: 18% MeOH; 9 min: 18% MeOH; 9 min 18 s: 0%
MeOH; 14 min: 0% MeOH. In each case, in addition to the
parent hydrate peak, a small amount of the second hydrate
To determine if the dCyd hydrates could be precursors to
these two products, hereafter termed D1 (telution = 3.7 min)

Photochemistry and Photobiology, 2011, 87 829
and a small peak corresponding to D2, eluting at 7.2 min,
came off as well. The UV spectrum of the material eluting
in this peak was identical to that of the peak eluting at
7.4 min in Panel (b) in figure (kmax = 299 nm). The peak D1,
shown in Panel (b) of Fig. 1 as eluting at 3.75 min, did not
appear in these chromatograms. We did some exploratory
experiments that showed that D1 is the same compound
described by Urata and Akagi (35); these workers found that
irradiation of dCyd in 0.2 mM solution at pH 4 resulted in
production of 2-deoxyribolactone. In addition, UV spectral
monitoring of the progress of the reaction indicated that an
unidentified product with a kmax at 311 nm increased in
amount with irradiation time. We repeated this work; how-
ever, in addition to following the reaction by UV spectroscopy,
we analyzed the reaction mixture using HPLC. Analysis of the
reaction mixture resulting from 32 min irradiation at 254 nm
(3 mL of 0.2 mM dCyd in a quartz spectrophotometer cuvette,
pH 4, placed next to a BLE-IT155 lamp) on Column B was
carried out (elution gradient: 0 min: 0% MeOH; 2 min: 0%
MeOH; 6 min: 12% MeOH; 7 min: 12% MeOH; 7 min 15 s:
0% MeOH; 10 min: 0% MeOH; flow rate: 2 mL min)1). The
main peak observed, when detection was done at either
311 nm or 299 nm, was that corresponding to D1
(telution = 3.5 min), as shown by identity of UV spectra and
via coinjection with a similarly irradiated solution made up in
distilled water. Only a trace of material corresponding to D2
was present in the solution irradiated at pH 4. We did not
pursue the identification of D1 further.
To obtain additional support for the hypothesis that the
material corresponding to peak D2 was chemically related to
the known hydrates, we isolated this substance and followed
its thermal reaction via spectroscopy and chromatography.
A solution of dCyd in distilled water (2 mM, 50 mL) was
irradiated at 254 nm in a cold room (about 4ꢁC) for 16 h in a
Southern New England Ultraviolet Company photochemical
reactor. Forty-five milliliter of the irradiated solution was
concentrated to 0.5 mL, using rotatory evaporation at 32ꢁC.
Thirty microliters of this material was injected on Column B
and a separation was made using distilled water flowing at
2.25 mL min)1. The material corresponding to D2, eluting at
10.2 min, was collected in a 0.6 mL quartz spectrophotometer
cell and this cell was then immediately situated in a cell holder
whose temperature was controlled by a circulating water bath.
After temperature equilibration, the HP diode array spec-
trometer was used in the kinetic mode to collect UV spectra at
A discussion of some of the challenges that will need to be met
to obtain a definitive structure of the material corresponding
to D2 is given below.
Could dCyd hydrates similarly exist within the context of
DNA as ring-closed forms (Xb) in pseudoequilibrium with an
opened-ring form, as shown in the second reaction in Scheme 5
(drawn with the unproven assumption that the material
corresponding to D2 has the structure Xc)? If so, then achieving
a complete understanding of the photochemistry and photobi-
ology of cytosine residues within DNA (or RNA) could require
dealing with an unanticipated degree of complexity in terms of
the photoproducts present. For example, repair of hydrates
that exist within DNA in both closed-ring and opened-ring
structures could occur at different rates or, perhaps, could even
require differing modes of enzymatic detection and removal
within damaged DNA; similarly, interactions of other DNA-
binding proteins (e.g. those involved in transcription, replica-
tion or establishment of DNA structure within the chromo-
some) with damaged DNA could be different for the two forms
of hydrate. One question that can be asked is whether an
opened-ring hydrate could continue to exist for a reasonable
length of time after a closed-ring hydrate was removed by
repair or decomposition. Answering the question for photo-
damaged DNA in vitro or in vivo will require sophisticated
experimentation. However, the type of data displayed in
Figure S1 does allow an answer to be given to this question
at the level of the nucleoside. In particular, we have determined
the enthalpy of activation and entropy of activation for
reversion of nucleosidic D2 to closed-ring hydrates in the
doubly distilled water eluting from Column B (measured
pH = 8.8) and in sodium phosphate buffer at near physiolog-
ical pH. First, we repeated the experiment run at 25.3ꢁC (see
above) at three additional temperatures in the distilled water
eluting from the HPLC column. Using measured absorbance
values at kmax = 299 nm as data points, we found that the
reversion reaction at each temperature obeyed first order
kinetics (i.e. graphs of the natural log of the measured
absorbance at 299 nm versus time were linear). The first order
rate constants determined at the four temperatures were as
follows: 12.8ꢁC, 0.0151 min)1; 18.7ꢁC, 0.0254 min)1; 25.3ꢁC,
0.0472 min)1; 29.9ꢁC, 0.0637 min)1. Another set of experi-
ments was conducted in 4 mM sodium phosphate buffer (pH
7.54). In these experiments, 960 lL of freshly collected material
was diluted with 40 lL of 100 mM phosphate and put in a
1.2 mL cuvette. After temperature equilibration, four sets of
overlaid spectra similar to those shown in Figure S1 were
obtained. From these runs, the following set of first order rate
appropriate intervals.
A representative set of raw data
obtained at 25.3ꢁC, with a spectrum run every 3 min, is shown
in Figure S1. This figure shows that, over a period of 60 min,
the absorbance at 299 nm decreases with time, while the
absorbance at 242 nm (kmax for dCyd hydrate) concurrently
increases. Thus, D2 is converted to compounds having UV
spectral traces very similar to those of the hydrates H1 and H2.
The UV spectrum of the material corresponding to D2 is
very similar in profile to the corresponding spectra seen for the
amine adducts (IVa, IVb and IVc) and hydrates (Va and Vb) of
m5C and DMC. This indicates that extensive double bond
conjugation must be present in the structure of this product
and strongly suggests that the structure has an opened-ring
structure. Although we have not established a definitive
structure for D2, it seems likely that the structure may be or
have similarities to the structure displayed as Xc (Scheme 5).
constants was calculated: 15.1ꢁC, 0.0287 min)1
; 18.2ꢁC,
0.0415 min)1; 21.2ꢁC, 0.0552 min)1; 26.3ꢁC, 0.0858 min)1. In
each of the runs described above, the correlation coefficient was
0.989 or greater.
It can be noted the rate constant for reversion of Cyd
hydrate to Cyd in unbuffered solution at 25ꢁC and pH 8 has
been determined by deBoer et al. (36) to be 0.0036 min)1; at
pH 9, the corresponding value was 0.0045 min)1. If it is
assumed that similar numbers obtain for the dCyd hydrate,
then conversion of D2 to hydrate appears to be on the order of
10 times faster than the conversion of hydrate to parent
nucleoside in unbuffered water. Likewise, the rate constant for
reversion of dCyd hydrate to parent at 26ꢁC has been
evaluated to be 0.0066 min)1 (37) in 4 mM phosphate buffer

830 Martin D. Shetlar and Janet Chung
at pH 7. This compares to the value of 0.0858 min)1 given
above for conversion of D2 to hydrate in 4 mM phosphate
buffer at pH 7.5 at 26.3ꢁC. Thus, it can be estimated that the
dCyd hydrate decomposes to dCyd in the neighborhood of 13
times slower under these conditions than the reversion of D2
to hydrate. Rate constants for conversion of dCyd hydrate to
D2 at corresponding temperatures and pH values remain to be
determined. Such rate constants are required for rigorous
kinetic analysis of the dCyd hydrate-D2-dCyd system. How-
ever, detection of D2 in the dCyd hydrate system, along with
the implications of the kinetic parameters available at the
nucleoside level, suggest that it should not be ruled out that
dCyd hydrates, contained within DNA, could have a suffi-
ciently long lifetime to establish a pseudoequilibrium with D2
before their eventual reversion to dCyd.
The Eyring transition state theory of chemical kinetics
indicates that the temperature dependence of the rate constant
for a reaction in solution can be mathematically described as
shown in the equation given immediately below (see, for
example, Chang [38], pp. 476–480). In this equation DH‡ and
DS‡ are, respectively, the enthalpy of activation and entropy of
activation, j is the transmission coefficient, R is the gas
constant, T is the temperature in K, N is Avogadro’s number
and h is Planck’s constant.
(37ꢁC). In 4 mM phosphate, the corresponding values are
4.39 · 10)3 min, t1 ⁄ 2 = 158 min (0ꢁC) and k = 0.282 min)1
,
t1 ⁄ 2 = 2.5 min (37ꢁC). If the value of t1 ⁄ 2 in phosphate buffer
at 37ꢁC can be extrapolated to ring-opened dCyd hydrates
contained in DNA in the environment of a cellular nucleus,
then this value suggests that the dCyd opened-ring hydrate,
once formed, could have a sufficiently long lifetime for it to be
regarded as a separate lesion in a biological context.
We now return to the challenges of obtaining a definitive
structure for the compound corresponding to D2. Ideally such
a structure should be based on NMR spectroscopic and mass
spectrometric characterization of D2. The rate constant data
discussed above suggests that elegant experimentation will be
required to obtain such information. For example, the path to
obtaining a pure sample of D2 suitable for proton or C-13
NMR has difficulties. Based on the kinetic data discussed
above, if purified D2 is to be maintained in a pure state, then
the mode of sample preparation should avoid elevated
temperatures; otherwise conversion to closed-ring hydrate will
occur. However, even such a gentle technique as lyophilization
of aqueous solvent from the purified frozen sample requires at
least several hours of time. Assuming a 3-h lyophilization
period in which an aqueous sample is maintained in the frozen
state at 0ꢁC, a little bit more than 50% of the original
compound will have reverted to closed-ring hydrate at the end
of the lyophilization process. Procedures for putting the freeze-
dried sample into a suitable NMR solvent (e.g. D2O) are likely
to lead to even more loss, while extended NMR examination,
particularly those involving two-dimensional techniques, will
likewise lead to an even greater degree of decomposition. Use
of a technique such as stopped flow HPLC-NMR could be
useful in obtaining a proton NMR spectrum of putative Xc.
Beyond preparation of a sample of D2 free of parent hydrate
for NMR purposes, an additional problem exists in obtaining
a definitive molecular mass for D2, particularly if it has the
structure Xc. The closed-ring hydrate and D2 would then have
the same molecular mass; thus, parent D2 cannot be distin-
guished from its closed-ring hydrate decomposition product(s)
via simple mass spectrometric measurements. Utilization of
HPLC ⁄ mass spectrometry or other advanced methods may
allow the problem of molecular mass determination to be
solved. The observation that D2 can be eluted from a Capcell
reversed phase column with distilled water could be helpful in
such studies. Summarizing, definitive structural characteriza-
tion of D2 via physical techniques could provide a worthy
challenge for future investigators.
z
z
k ¼ jðRT=NhÞeDS =ReꢀDH =RT
If j is set equal to unity, this equation can be rewritten in the
form of ln (k ⁄ T) = ln (R ⁄ Nh) + DS‡ ⁄ R ) DH‡ ⁄ RT.
According to this equation, the value of DH‡ can be evaluated
from the slope of a graph of ln (k ⁄ T) versus 1 ⁄ T, while the y-
intercept of such a graph yields a quantity from which DS‡ can
be calculated. We analyzed the kinetic data for reversion of D2
to hydrate using this equation; the resulting plot yielded values
of DH‡ = 62.6 kJ and DS‡ = )94.6 J K)1 for the reaction
converting D2 to hydrate in distilled water. The correlation
coefficient describing the linear fit of the rate constant data for
the Eyring plot was 0.996. A similar analysis of the data ob-
tained in phosphate buffer indicates that DH‡ = 78.6 kJ and
DS‡ = )36.2 J K)1 with correlation coefficient 0.994. Note
that these two sets of activation parameters cannot necessarily
be directly compared, as they were obtained at different pH
values. Furthermore, the data for the phosphate buffer may
reflect inclusion of catalytic pathways for conversion of D2 to
hydrate that do not occur in the experiments in distilled water.
As deBoer et al. (36) have shown, the rate of decomposition of
dCyd hydrates is increased in the presence of various buffers,
as compared to the corresponding rates when buffers were
absent. (Phosphate was not among the buffers studied, al-
though it was shown that the rate of loss of the photohydrate
of cytidylic acid is greatly enhanced, as compared to the cor-
responding rate for Cyd hydrates [36].) The effects of the
various buffers on the kinetics of these reactions were analyzed
in terms of acid-base catalysis. Conceivably, the presence of
buffer components could similarly introduce new pathways for
conversion of D2 to hydrate.
Preliminary results suggest that the cytosine photohydrate and
1-methylcytosine photohydrate also undergo ring-opening to
yield analogs of D2
To examine the possibility that the cytosine hydrate can also
exist in an alternative ring-opened form, we examined the
HPLC of freshly irradiated Cyt. It is interesting to note that, as
was the case with the hydrates of dCyd, definitive structural
studies of the photohydrates of cytosine and 1-methylcytosine
(e.g. involving NMR and mass spectrometric studies of purified
compound) have evidently not been published, probably
because their instability makes such studies difficult. Therefore,
based on their thermal reversibility (as described in the relevant
literature [reviewed in 2,5]), we have assumed that the dominant
From the values obtained for DH‡ and DS‡ the rate constant
and half-life (t1 ⁄ 2) for reversion of D2 to hydrate can be
evaluated at other temperatures of interest (e.g. 0ꢁC and 37ꢁC).
In distilled water, the desired values are k = 4.43 · 10)3 min)1
,
t1 ⁄ 2 = 156 min (0ꢁC) and k = 0.12 min)1, t1 ⁄ 2 = 5.8 min

Photochemistry and Photobiology, 2011, 87 831
product arising from UV irradiation of each of these com-
pounds at low concentration is indeed a hydrate analogous to
that shown in Xb. Irradiation of 100 mL of 0.2 mM Cyt at
254 nm was carried out for 60 min as described above for dCyd.
After concentration to 1.4 mL via rotatory evaporation at
40ꢁC, 500 lL of the irradiated solution was chromatographed
on Column B using the following gradient with 100 mM NaCl
and MeOH as eluents: 0 min: 0% MeOH; 1 min: 0% MeOH;
5 min: 15% MeOH; 5 min 30 s: 15% MeOH; 5 min 45 s: 0%
adducts of m5C, Cyt, dCyd and related compounds can readily
be isolated in opened-ring forms, this report gives the first
indication that water adducts of such compounds can similarly
exist as such isomeric forms in amounts detectable by HPLC.
Acknowledgements—We gratefully acknowledge NSF Grant CHE-
0131203 and services provided by the Bio-organic, Biomedical Mass
Spectrometry Resource (A. L. Burlingame, Director), supported by the
NIH Division of Research Resources Grant RR 01614 and especially
thank David Maltby who took special pains to obtain needed mass
spectra in a timely manner.
MeOH; 9 min: 0% MeOH. The flow rate was 2 mL min)1
.
A peak eluting between 0.6 and 1.1 min corresponded to Cyt
hydrate. A variety of additional peaks also eluted, including at
least one ureidoacrylonitrile product (27). Of particular interest
was a relatively small peak (compared with the size of the
hydrate and Cyt peaks) eluting at 3.7 min; this peak contained
material with a UV absorption spectrum that displayed a
kmax = 297 nm; the profile of the absorption band was very
similar to that observed for D2 in the case of dCyd. Reinjection
of 250 lL of the collected Cyt hydrate, utilizing the same
column and gradient conditions, again led to an HPLC
chromatogram that contained a small peak with a retention
time of 3.5 min and containing material with kmax = 297 nm.
The same experimental protocol was applied to the
1-methylcytosine system. In this case, the hydrate eluted
between 0.8 and 1.6 min, while material eluting in a small peak
at 6.6 min had the UV spectroscopic character of a compound
similar in nature to D2 (kmax = 297 nm). Once again, a
number of other products also eluted from the column,
including previously characterized ureidoacrylonitriles (27).
The HPLC chromatogram resulting from reinjection of 250 lL
of collected hydrate, using the same column and elution
conditions as previously, showed a small peak eluting at
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
Figure S1. A set of UV spectral traces showing the decay of
the intensity of the spectrum corresponding to D2, the putative
opened-ring form of the hydrate, and the corresponding
increase in the intensity of the spectrum corresponding to the
closed-ring dCyd hydrate(s).
Appendix S1. Experimental details concerning the isolation
of IVb, IVc and Vb, as well as a description of the methods for
determining the yields of IVa and IVb and the preparation of
VI.
Please note: Wiley-Blackwell is not responsible for the
content or functionality of any supporting information sup-
plied by the authors. Any queries (other than missing material)
should be directed to the corresponding author for the article.
REFERENCES
1. Friedberg, E. C., G. C. Walker, W. Seide and R. D. Wood (2005)
DNA Repair and Mutagenesis, 2nd edn. ASM Press, Washington,
DC.
6.6 min with the same spectral profile and value of kmax
.
The small amount of putative D2 analog present in either
the isolated Cyt hydrate or the corresponding 1-methylcyto-
sine hydrate samples, as revealed by HPLC, precluded
isolating sufficient amounts of these putative opened-ring
hydrates to conduct studies to observe the reversal of these
compounds to closed-ring hydrates. Thus, the results of the
studies described for these two compounds must be regarded
only as suggestive.
2. Fisher, G. J. and H. E. Johns (1976) Pyrimidine photohydrates. In
Photochemistry and Photobiology of Nucleic Acids, Vol. 1 (Edited
by S. Y. Wang), pp. 169–224. Academic Press, New York.
3. Fisher, G. J. and H. E. Johns (1976) Pyrimidine photodimers. In
Photochemistry and Photobiology of Nucleic Acids, Vol. 1 (Edited
by S. Y. Wang), pp. 225–294. Academic Press, New York.
4. Wang, S. Y. (1976) Pyrimidine bimolecular photoproducts. In
Photochemistry and Photobiology of Nucleic Acids, Vol. 1 (Edited
by S. Y. Wang) pp. 295–356. Academic Press, New York.
5. Cadet, J. and P. Vigny (1990) The photochemistry of nucleic acids.
In Bioorganic Photochemistry, Vol. 1 (Edited by H. Morrison),
pp. 1–272. John Wiley and Sons, New York.
CONCLUDING REMARKS
There are reports in the literature indicating that each of the
three major pyrimidine nucleobases (Ura, Thy and Cyt) forms
a photohydrate when irradiated in water. However, a hydrate
with a similar structure has not been reported for m5C. We
have prepared an opened-ring structural isomer of this
compound; this synthesis relied on treatment of an m5C-
ammonia adduct with dilute acid. It has been shown that this
opened-ring structure (N-carbamoyl-3-hydroxy-2-methylacry-
lamidine (Va)) decomposes thermally to form products that
are putatively cis and trans 5,6-dihydro-6-hydroxy-5-methyl-
cytosine (Ia and IIa), one of which can be detected in
irradiated aqueous m5C. Analogous results are described for
Vb, the opened-ring water adduct corresponding to Ib and IIb.
Other work showed that the photohydrates of dCyd in
aqueous solution coexist with a compound that exhibits the
UV spectroscopic properties expected of the corresponding
ring-opened water adduct. While it is well known that amine
6. Davies, R. J. H. (1995) Ultraviolet radiation damage in DNA.
Biochem. Soc. Trans. 23, 407–418.
7. Ruzsicska, B. P. and D. G. E. Lemaire (1995) DNA photochem-
istry. In CRC Handbook of Organic Photochemistry and Photobi-
ology (Edited by W. M. Horspool and P.-S. Song), pp. 1289–1315.
CRC Press, Boca Raton, FL.
8. Ravanat, J.-L., T. Douki and J. Cadet (2001) Direct and indirect
effects of UV radiation on DNA and its components. J. Photo-
chem. Photobiol. B 63, 88–102.
9. Cadet, J., S. Courdavault, J.-L. Ravanat and T. Douki (2005)
UVB and UVA radiation-mediated damage to isolated and cel-
lular DNA. Pure Appl. Chem. 77, 947–961.
10. Cadet, J., E. Sage and T. Douki (2005) Ultraviolet radiation-
mediated damage to cellular DNA. Mutation Res. 571, 3–17.
11. Shetlar, M. D. (1980) Cross-linking of proteins to nucleic acids by
ultraviolet light. Photochem. Photobiol. Rev. 5, 105–197.
12. Saito, I. and H. Sugiyama (1990) Photoreactions of nucleic acids
and their constituents with amino acids and related compounds.
In Bioorganic Photochemistry, Vol. 1 (Edited by H. Morrison),
pp. 317–340. John Wiley and Sons, New York.

832 Martin D. Shetlar and Janet Chung
13. Meisenheimer, K. M. and T. H. Koch (1997) Photocross-linking
of nucleic acids to associated proteins. Crit. Rev. Biochem. Mol.
Biol. 32, 101–140.
26. Shugar, D. and J. J. Fox (1952) Spectrophotometric studies of
nucleic acid derivatives and related compounds as a function of
pH. Biochim. Biophys. Acta 9, 199–218.
14. Moore, A. M. and C. H. Thomson (1957) Ultraviolet irradiation
of pyrimidine derivatives: I. 1,3-dimethyluracil. Can. J. Chem. 35,
163–169.
15. Robertson, K. D. (2003) DNA methylation and chromatin-
unraveling the tangled web. Oncogene 21, 5361–5369.
16. Privat, E. and L. Sowers (1996) Photochemical deamination
and demethylation of 5-methylcytosine. Chem. Res. Toxicol. 9,
745–750.
17. Vairapandi, M. and N. J. Duker (1994) Excision of ultraviolet-
induced photoproducts of 5-methylcytosine from DNA. Mutation
Res. 315, 85–94.
18. Zuo, S., R. J. Boorstein, R. P. Cunningham and G. W. Teebor
(1995) Comparison of the effects of UV irradiation on
5-methyl-substituted and unsubstituted pyrimidines in alter-
nating pyrimidine-purine sequences in DNA. Biochemistry 34,
11582–11590.
27. Shaw, A. A. and M. D. Shetlar (1990) 3-Ureidoacrylonitriles:
Novel products from the photoisomerization of cytosine,
5-methylcytosine and related compounds. J. Amer. Chem. Soc.
112, 7737–7742.
28. Sulkowska, A. (1997) Effect of temperature on the stability of
association of pyrimidine bases with serum albumin: Proton NMR
study. Appl. Spectroscopy 51, 428–432.
29. Knittel, J. and R. Zavod (2002) Drug design and relationship of
functional groups to pharmacologic activity. In Foye’s Principles
of Medicinal Chemistry (Edited by D. A. Williams and T. L.
Lemke), pp. 37–67. Lippincott Williams and Wilkins, Philadel-
phia, PA.
30. Shetlar, M. D., J. Chung and K. Hom (2010) Photoinduced ring
opening reactions and subsequent thermal ring closure reactions in
the thymine-ammonia system and related systems. Photochem.
Photobiol Sci. 9, 1128–1243.
19. Wierzchowski, K. L. and D. Shugar (1960) Further studies
on the photochemistry of pyrimidines, with special reference
to 5- and 6-substituted derivatives in relation to photore-
activation in the T-even bacteriophages. Acta Biochim. Pol. 7,
63–84.
20. Celewicz, L. and M. D. Shetlar (1992) The photochemistry of
5-methylcytosine and 5-methyl-2¢-deoxycytidine in aqueous solu-
tion. Photochem. Photobiol. 55, 823–830.
31. Liu, F.-T. and N. C. Yang (1978) Photochemistry of cytosine
derivatives. 2. Photohydration of cytosine derivatives. Proton
magnetic resonance study on the chemical structure and property
of photohydrates. Biochemistry 17, 4877–4885.
32. Douki, T., G. Vadesne-Bauer and J. Cadet (2002) Formation of
2¢-deoxyuridine hydrates upon exposure of nucleosides to gamma
radiation and UVC-irradiation of isolated and cellular DNA.
Photochem. Photobiol. Sci. 1, 565–569.
21. Dorwin, E. L., A. A. Shaw, K. Hom, P. Bethel and M. D. Shetlar
(1988) Photoexchange products of cytosine and 5-methylcytosine
with N-a-acetyl-L-lysine and L-lysine. J. Photochem. Photobiol. B
2, 265–278.
22. Saito, I., H. Sugiyama, S. Ito, N. Furukawa and T. Matsuura
(1981) A novel photoreaction of thymidine with lysine. Photoin-
duced migration of thymine from DNA to lysine. J. Amer. Chem.
Soc. 103, 1598–1600.
23. Shetlar, M. D., K. Hom, S. Distefano, K. Ekpenyong and J. Yang
(1988) Photochemical reactions of cytosine and 5-methylcytosine
with methylamine and n-butylamine. Photochem. Photobiol. 47,
779–786.
24. Shaw, A. A. and M. D. Shetlar (1989) Ring-opening photoreac-
tions of cytosine and 5-methylcytosine with aliphatic alcohols.
Photochem. Photobiol. 49, 267–271.
33. Hom, K. (1991) Photochemical reactions of 2¢-deoxycytidine,
2¢-deoxyuridine and related pyrimidine nucleosides and nucleo-
bases with ethylamine. Ph.D. dissertation, University of Califor-
nia, San Francisco.
34. Cadet, J., L.-S. Kan and S. Y. Wang (1978) O6,5¢-Cyclo-5,
6-dihydro-2¢-deoxyuridine. Novel deoxyuridine photoproducts.
J. Amer. Chem. Soc. 100, 6715–6720.
35. Urata, H. and M. Akagi (1991) Photo-induced formation of the
2-deoxyribonolactone-containing nucleotide for d(ApCpA);
effects of neighboring bases and modification of deoxycytidine.
Nucleic Acids Res., 19, 1773–1778.
36. DeBoer, G., O. Klinghoffer and H. E. Johns (1970) Reversal
mechanisms for the photohyrates of cytosine and its derivatives.
Biochim. Biophys. Acta 213, 253–268.
37. Schott, H., U. Hommel and G. Gauglitz (1988) Examination of
the photokinetics of 1-b-D-arabinofuranosylcytosine and deoxy-
ribocytidine. Photochem. Photobiol. 47, 517–522.
38. Chang, R. (2000) Physical Chemistry for the Chemical and
Biological Sciences. University Science Books, Sausalito, CA.
25. Shaw, A. A. and M. D. Shetlar (1989) The photochemistry of
2-alkoxycytosines in phosphate buffer and its link to cytosine
photochemistry in alcoholic solution. Photochem. Photobiol. 49,
273–277.