The reactions of thymine and thymidine with ozone

Article abstract of DOI:10.1039/b204067k

The ozonolysis of thymine and thymidine has been investigated by a product study complemented by kinetic studies using spectrophotometry, conductometry and stopped-flow with optical and conductometric detection. Material balance has been obtained. Ozonolysis of thymine (k = 3.4 × 10⁴ dm³ mol^-1 s^-1) leads to the formation of the acidic (pK_a = 4) hydroperoxide 1-hydroperoxymethylene-3-(2-oxopropanoyl)urea 5 (～34%), neutral hydroperoxides (possibly mainly 1-hydroperoxyhydroxymethyl-3-(2-oxopropanoyl)urea 6, total ～41%) and H₂O₂ (25%, with corresponding formation of 1-formyl-5-hydroxy-5-methylhydantoin 11). The organic hydroperoxides decay (～1.1 × 10^-3 s^-1 at 20°C, 1.3 × 10^-4 s^-1 at 3°C) releasing formic acid (formation of 5-hydroperoxy-5-methylhydantoin 18) and also to some extent H₂O₂ (and 11). After 100 min, the formic acid yield is 75%. Upon treatment at high pH, it increases to 100%. Reduction of the organic hydroperoxides with bis(2-hydroxyethyl) sulfide (k = 50 dm³ mol^-1 s^-1) leads to 11 whose subsequent treatment with base yields 5-hydroxy-5-methylhydantoin 13 in 100% yield. It is suggested that the Criegee ozonide formed upon reaction with ozone at the C(5)-C(6) double bond opens heterolytically in two directions with subsequent opening of the C(5)-C(6) bond. In the preferred route (75%), the positive charge resides at C(6). Deprotonation at N(1) gives rise to 5, while its reaction with water yields 6. Loss of formic acid yields 5-hydroperoxy-5-methylhydantoin 18. Reduction of 5 and 6 with the sulfide yields 11. In the minor route (25%), the positive charge remains at C(5) followed by a reaction with water. The resulting α-hydroxy hydroperoxide rapidly loses H₂O₂ (formation of 11). In basic solution, singlet dioxygen is formed (8%). The concomitant product, 5,6-dihydroxy-5,6-dihydrothymine has been detected. In the ozonolysis of thymidine, the rapid formation of conductance (k = 0.55 s^-1) is due to the release of acetic acid (18%). In this reaction a short-lived hydroperoxide is destroyed. As a consequence of this, 25 s after ozonolysis the total hydroperoxide yield is only ～78% (including 8% H₂O₂). The products corresponding to acetic acid are suggested to be CO₂ and N-(2-deoxy-β-D-erythropentofuranosyl)formylurea 22. A number of organic hydroperoxides have been detected by HPLC by post-column derivatisation with iodide. An acidic hydroperoxide such as 5 in the case of thymine is not among the products. Upon sulfide reduction, the organic hydroperoxides yield mainly (43-50%) N₁-(2-deoxy-β-D-erythropentofuranosyl)-5-hydroxy-5- methylhydantoin 23. The reasons for some striking differences in the ozonolyses of thymine and thymidine are discussed.

Full text of DOI:10.1039/b204067k

View Article Online / Journal Homepage / Table of Contents for this issue
The reactions of thymine and thymidine with ozone
Roman Flyunt,†a,b Jacob A. Theruvathu,‡a Achim Leitzkea,b and Clemens von Sonntag*a,b
a Max-Planck-Institut für Strahlenchemie, Stiftstr. 34-36, PO Box 101365,
D-45413 Mülheim an der Ruhr, Germany. E-mail: Clemens@vonSonntag.de;
Fax: ϩ49-208-306-3951
2
b Institut für Oberflächenmodifizierung (IOM), Permoserstr. 15, D-04303 Leipzig, Germany
Received (in Cambridge, UK) 26th April 2002, Accepted 12th July 2002
First published as an Advance Article on the web 6th August 2002
The ozonolysis of thymine and thymidine has been investigated by a product study complemented by kinetic studies
using spectrophotometry, conductometry and stopped-flow with optical and conductometric detection. Material
balance has been obtained. Ozonolysis of thymine (k = 3.4 × 104 dm3 molϪ1 sϪ1) leads to the formation of the acidic
(pKa = 4) hydroperoxide 1-hydroperoxymethylene-3-(2-oxopropanoyl)urea 5 (∼34%), neutral hydroperoxides (possibly
mainly 1-hydroperoxyhydroxymethyl-3-(2-oxopropanoyl)urea 6, total ∼41%) and H2O2 (25%, with corresponding
formation of 1-formyl-5-hydroxy-5-methylhydantoin 11). The organic hydroperoxides decay (∼1.1 × 10Ϫ3 sϪ1 at 20 ЊC,
1.3 × 10Ϫ4 sϪ1 at 3 ЊC) releasing formic acid (formation of 5-hydroperoxy-5-methylhydantoin 18) and also to some
extent H2O2 (and 11). After 100 min, the formic acid yield is 75%. Upon treatment at high pH, it increases to 100%.
Reduction of the organic hydroperoxides with bis(2-hydroxyethyl) sulfide (k = 50 dm3 molϪ1 sϪ1) leads to 11 whose
subsequent treatment with base yields 5-hydroxy-5-methylhydantoin 13 in 100% yield. It is suggested that the Criegee
ozonide formed upon reaction with ozone at the C(5)–C(6) double bond opens heterolytically in two directions
with subsequent opening of the C(5)–C(6) bond. In the preferred route (75%), the positive charge resides at C(6).
Deprotonation at N(1) gives rise to 5, while its reaction with water yields 6. Loss of formic acid yields 5-hydroperoxy-
5-methylhydantoin 18. Reduction of 5 and 6 with the sulfide yields 11. In the minor route (25%), the positive charge
remains at C(5) followed by a reaction with water. The resulting α-hydroxy hydroperoxide rapidly loses H2O2
(formation of 11). In basic solution, singlet dioxygen is formed (8%). The concomitant product, 5,6-dihydroxy-5,6-
dihydrothymine has been detected. In the ozonolysis of thymidine, the rapid formation of conductance (k = 0.55 sϪ1
)
is due to the release of acetic acid (18%). In this reaction a short-lived hydroperoxide is destroyed. As a consequence
of this, 25 s after ozonolysis the total hydroperoxide yield is only ∼78% (including 8% H2O2). The products
corresponding to acetic acid are suggested to be CO2 and N-(2-deoxy-β--erythropentofuranosyl)formylurea 22. A
number of organic hydroperoxides have been detected by HPLC by post-column derivatisation with iodide. An acidic
hydroperoxide such as 5 in the case of thymine is not among the products. Upon sulfide reduction, the organic
hydroperoxides yield mainly (43–50%) N1-(2-deoxy-β--erythropentofuranosyl)-5-hydroxy-5-methylhydantoin 23.
The reasons for some striking differences in the ozonolyses of thymine and thymidine are discussed.
Ozone has also been known for a long time to cause damage
Introduction
to DNA of eukariotic cells, e.g., in yeast,8 in plants9 and also in
In drinking-water processing, ozone is gaining in importance,
human cells.10 Although ozone toxicity is of considerable con-
because it is not only a good oxidant but also a powerful
cern in industrial countries resulting in daily reports on ozone
disinfectant which readily copes with bacteria and viruses.1–4
levels in air, very little information as to the mechanism of its
Mechanistic details of their inactivation are as yet poorly
DNA damaging action is as yet available.
understood.
The reaction of ozone with nucleic acids and their constitu-
In the inactivation of viruses by the much more reactive OH
ents has already attracted some attention.11–21 Recently, we have
radical (for compilations of ozone and OH radical rate
supplemented the available data on ozone rate constants with
constants see refs. 5,6), two targets have been recognised, the
the nucleobases (the sugar moiety does not react with ozone at
protein coat and the DNA(RNA).7 Apparently, a good fraction
a comparable rate),22 and from this study it follows that the
of the OH radicals can penetrate through the protein coat and
guanine and thymine moieties are considerably more reactive
react with the nucleic acids located inside. In contrast, in the
than the cytosine and adenine ones.
case of bacteria, OH radicals created in the bulk water are
The reaction of ozone with thymidine has already been
effectively scavenged by the material that makes up the bacterial
studied,20 but a number of questions, especially as to mech-
membrane, and the resulting damage is insufficient to kill the
anistic details, remained open. For this reason, we have
bacterium.7 In the case of the inactivation of viruses by ozone,
reinvestigated the thymidine system based on what we have
DNA damage may also be a major cause, since in proteins only
learned from a detailed study on the products, part of them
a few of their constituents, cysteine, cystine, tryptophan and
short-lived intermediates, of its base moiety, thymine.
tyrosine, react fast with ozone.5 Less can be predicted for the
inactivation of bacteria by ozone.
Experimental
Thymine (Fluka) and thymidine (Acros) were used as received.
The reagents bis(2-hydroxyethyl) sulfide (Aldrich) and catalase
(Boehringer Mannheim) and the reference material 5-hydroxy-
5-methylhydantoin (Toronto Research Chemicals) were also
† On leave from Institute of Physico-Chemistry, National Academy of
Science of the Ukraine, Naukova Street 3a, UA-79053 L’viv, Ukraine.
‡ On leave from School of Chemical Sciences, Mahatma Gandhi
University, Kottayam-686560, Kerala, India.
1572
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b204067k

View Article Online
commercially available. Ozone was generated with the help
of a dioxygen-fed ozonator (Philaqua Philoz 04, Gladbeck).
The ozone content of the aqueous ozone stock solutions was
at this position [reactions (2) and (4)]. This has been rational-
ised by a stabilisation of the carbocation formed in reaction
(2).
determined spectrophotometrically using ε(260 nm)23,24
=
In the reaction with thymine and thymidine, ozone will add
to the C(5)–C(6) double bond. Due to the electron-donating
properties of N(1), the Criegee intermediate formed in reaction
(1) will preferentially open according to reaction (2). It will be
shown that the substituents at N(1) (H in thymine, 2-deoxy-
ribosyl in thymidine) have a pronounced effect on the sub-
sequent reactions of the carbocation formed in reaction (2). In
thymidine, this short-lived intermediate can only react accord-
ing to reaction (4), but in thymine it can also deprotonate at
N(1).
3300 dm3 molϪ1 cmϪ1. Solutions were made up in Milli-Q-
filtered (Millipore) water. For ozonation, stock substrate-
containing solutions were mixed with an aliquot of the ozone
stock solution.
Formic and acetic acids were determined by ion chromato-
graphy (Dionex DX-100, AS-9 HC, eluent: 1 × 10Ϫ2 mol dmϪ3
NaHCO3 at 1 ml minϪ1).
Liquid chromatography–mass spectrometry (LCMS) meas-
urements (Hewlett-Packard, HP1100/HP5989B) were carried
out by electrospray ionisation (ESI) in the positive mode using
acetonitrile–water (1 : 1) as eluent.
Thymine
Hydrogen peroxide and organic hydroperoxides were deter-
mined with molybdate-activated iodide.25 This reagent consists
of two components. One contains iodide (0.4 mol dmϪ3),
molybdate (1.6 × 10Ϫ4 mol dmϪ3) and KOH (3.6 × 10Ϫ2 mol
Thymine 1 reacts with ozone with a rate constant of 4.2 × 104
dm3 molϪ1 Ϫ1, and upon deprotonation (pKa = 9.9) the rate
s
constant increases to 3 × 106 dm3 molϪ1 sϪ1 22 At pH 6.5, where
.
the experiments to be reported next have been carried out,
the contribution of the anion in product formation can be
neglected. Under typical experimental conditions ([thymine] =
1 × 10Ϫ3 mol dmϪ3, [O3] = 1 × 10Ϫ4 mol dmϪ3) the ozone half-life
is 0.016 s, i.e. the reaction proceeds at the time scale of the
mixing of the two solutions.
dmϪ3), the other potassium hydrogen phthalate (0.1 mol dmϪ3
)
as buffer (pH ∼4).25 Equal volumes of the probe and the
reagents are mixed, and the formation of I3Ϫ is measured at 350
nm (ε = 25500 dm3 molϪ1 cmϪ1). Some experiments have also
been carried out without molybdate as catalyst to differentiate
between highly reactive hydroperoxides and hydrogen peroxide.
The latter reacts very slowly under these conditions.
Conductometry and ion chromatography. When the ozonoly-
sis is carried out in a conductometric cell, the conductance rises
with biphasic kinetics (Fig. 1). The first step is too fast to be
For gas chromatography–mass spectrometry (GCMS), a
Hewlett-Packard (5890 Series II) setup was used. For better
volatility, the samples were trimethylsilylated (TMS) with
bis(trimethylsilyl)trifluoroacetamide (BSTFA, Macherey und
Nagel). 5-Hydroxy-5-methylhydantoin yields the 3-TMS deriv-
ative, MW = 346 Da, m/z (%): 331 (100), 216 (7), 147 (46),
73 (63). The cis/trans isomers of 5,6-dihydro-5,6-dihydroxy-
thymine (thymine glycol) were identified as their 4-TMS deriv-
atives, MW = 448 Da, m/z (%): 433 (10), 405 (8), 331 (75),
147 (25), 73 (100).
Stopped-flow experiments were carried out using a Biologic
SFM3 and a laboratory-made set-up. This allowed us to follow
the kinetics of the reaction either spectrophotometrically with a
diode array detector (Tidas-16, J&M, Aalen) or conducto-
metrically. For this purpose, a laboratory-made setup (similar
to the one described before)26,27 has been used.
Conductometric measurements on the longer time-scale were
carried out with a conductometer (CDM3, Radiometer).
Results and discussion
Fig. 1 Ozonolysis of thymine in aqueous solution at 18 ЊC. Formation
of acid (expressed as mol acid per mol ozone consumed) as a function
of time as followed by conductance measurements. Inset: the slow part
of conductance increase at 3 ЊC (᭺) and 18 ЊC (•) plotted as if it were of
first-order kinetics.
The general reaction mechanism of ozone with olefins in aqueous
solution
According to a detailed rate and product study that also
involved the characterisation of short-lived hydroperoxides,
there is a considerable influence of the electron-donating/
withdrawing power of substituents (D vs. A) on the final prod-
ucts [cf. reactions (1)–(5)].28
resolved kinetically by means of conventional conductometry
(see Fig. 1). The second shows a half-life of 10.5 min at room
temperature (k ≈ 1.1 × 10Ϫ3 sϪ1). At 3 ЊC, the conductance build-
In fact, an electron-donating group D such as a methyl sub-
stituent favours the formation of the α-hydroxy hydroperoxide
up is eight times slower (k ≈ 1.3 × 10Ϫ4 Ϫ1; cf. inset in Fig. 1).
s
For a compilation of rate constants see Table 1.
The calibration of the conductometric set-up has been done
with sulfuric acid. The acids that are formed in this system are
weak acids (the acidic hydroperoxide 5 has a pKa of 4.0, see
below, and formic acid one of 3.75) and are hence partially
protonated ([ozone] = 2 × 10Ϫ4 mol dmϪ3 in these experiments).
The “prompt” acid yield corrected for partial protonation is
then calculated at ∼34% (with respect to consumed ozone).
Further acid is released in the slow process whereby its yield
increases to ∼75%.
Formic acid yields have been measured by ion chromato-
graphy. The formic acid yield measured about 5 min after
ozonolysis is noticeably lower (∼43%) than after ∼100 min
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
1573

View Article Online
Table 1 Compilation of rate constants relevant to this study
Reaction
Rate constant
Reference
Thymine ϩ O3
products
4.2 × 104 dm3 molϪ1 sϪ1
3.4 × 104 dm3 molϪ1 sϪ1
3 × 106 dm3 molϪ1 sϪ1
> 70 sϪ1
Ref. 22
This work
Ref. 22
Thymine anion ϩ O3
Fast acid release
products
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
Slow formic acid release, 18 ЊC
Slow formic acid release, 3 ЊC
Decay of 5, 18 ЊC
1.1 × 10Ϫ3 sϪ1
1.3 × 10Ϫ4 sϪ1
1.0 × 10Ϫ3 sϪ1
Slow H2O2 release, 18 ЊC
∼1 × 10Ϫ3 sϪ1
5 (and 6) ϩ R2S
products
products
50 dm3 molϪ1 sϪ1
43 dm3 molϪ1 sϪ1
7.5 dm3 molϪ1 sϪ1
0.4 dm3 molϪ1 sϪ1
1 dm3 molϪ1 sϪ1
5 (and 6) ϩ IϪ
18 ϩ IϪ
products
4Ϫ
5 (and 6) ϩ Fe(CN)6
products
4Ϫ
18 ϩ Fe(CN)6
products
Table 2 Compilation of yields (with respect to ozone consumed) in the ozonolysis of thymine
Process
Yield (%)
Reference
“Prompt” acid formation (conductometry)
Formic acid at ∼100 min (conductometry)
Formic acid at ∼100 min (IC)
∼34
∼75
75
100
Absent
100
93
25
25
40
75
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
Formic acid at high pH (IC)
Acetic acid
Total hydroperoxide (immediate)
Total hydroperoxide (2 h)
Hydrogen peroxide (immediate, catalase assay)
Hydrogen peroxide (immediate, R2S assay)
Hydrogen peroxide (after 1 h)
Organic hydroperoxides (immediate)
Organic hydroperoxides (after 1 h)
1-Hydroperoxymethylene-3-(2-oxopropanoyl)urea 5
5-Hydroperoxy-5-methylhydantoin 18 (after 1 h)
5-Hydroxy-5-methylhydantoin 13 (after R2S treatment)
5-Hydroxy-5-methylhydantoin 13 (after R2S and OHϪ
treatment)
53
34
53
67
100
Singlet dioxygen (at high pH)
8
Ref. 29
(∼75%). In neutral solutions, this value does not increase
further with time. A treatment with base (KOH, pH 10.8,
90 min, 18 ЊC) brings the formic acid yield to 100%. For a
compilation of yields see Table 2.
In order to test whether the prompt acid release can also be
resolved kinetically, stopped-flow experiments with conducto-
metric detection were carried out. As can be seen from Fig. 2,
the build-up of conductance follows first order-kinetics,
and the rate of reaction depends on the thymine concen-
tration.
The rate constant derived from the data shown in the inset of
Fig. 2 (k = 3.4 × 104 dm3 molϪ1
s
Ϫ1) agrees with the value
22
obtained by competition kinetics (4.2 × 104 dm3 molϪ1 sϪ1
)
considering the errors involved in these measurements, especi-
ally in the value obtained by competition kinetics. It is thus
concluded that in this fast conductance increase the rate-
determining step is the reaction of ozone with thymine. It will
be shown below that an acidic hydroperoxide, assigned to 5, is
formed. Compared to the reaction of ozone with thymine
which gives rise to the thymine-derived Criegee ozonide, the
decay of this Criegee ozonide (leading to 5) and the subsequent
deprotonation of 5 are fast.
Formation and decay of hydroperoxides. The total hydro-
peroxide yield is 100% as assayed with molybdate-activated
iodide. Upon treatment of the ozonated solution with catalase,
the hydroperoxide yield is reduced to 75%. Thus, H2O2 may be
present right after ozonolysis in 25% yield. However, we have
observed that formic peracid, a conceivable product in the pres-
ent system, is also rapidly degraded by catalase,30 and further
experiments have been carried out to ascertain that the
hydroperoxide that is eliminated by catalase is indeed H2O2. To
this point, bis(2-hydroxyethyl) sulfide (1 × 10Ϫ3 mol dmϪ3) has
been added to the ozonated solution. This sulfide reacts fast
with formic peracid (k = 220 × dm3 molϪ1 sϪ1 30 yielding formic
)
acid (i.e. leading to an increase in conductance, pKa(HC(O)-
OOH) = 7.1, pKa(HC(O)OH) = 3.75), but no additional
increase of conductance is observed. On the contrary, a
decrease in the conductance occurs (k = 50 dm3 molϪ1 sϪ1), and
the “slow” increase in conductance shown in Fig. 1 is also
suppressed. This is taken as evidence that the observed “fast”
formation of conductance is due to an acidic hydroperoxide (5,
see below). The “slow” increase in conductance must also have
Fig. 2 Ozonolysis of thymine in aqueous solution at 18 ЊC. Formation
of an acid (assigned to 5) in the reaction of ozone with thymine as
followed by stopped-flow with conductometric detection. The solid line
through the data points is a first-order fit. Inset: kobs as a function of the
thymine concentration.
1574
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582

View Article Online
(formic acid releasing) hydroperoxides as precursors (5 and 6,
suggested, see below), and formic peracid is not a product of
the ozonolysis of thymine.
separated kinetically. The change in kobs with time shows a
half-value of ∼13 min (data not shown).
The reactivity of the two sets of organic hydroperoxides
4Ϫ
The reaction of bis(2-hydroxyethyl) sulfide cannot be used
to determine the yield of reactive hydroperoxides, since its
sulfoxide is difficult to detect by HPLC due to its low absorp-
tion coefficient in the accessible wavelength region. However,
methionine also undergoes the corresponding reaction with
reactive hydroperoxides, and its sulfoxide is more readily
detected.30 Using this assay, the yield of sulfide-reactive
hydroperoxides is 75%. The complement to 100%, a value of
25%, has been determined by measuring the hydroperoxide
yield remaining after the destruction of sulfide-reactive
hydroperoxides with bis(2-hydroxyethyl) sulfide (Fig. 3). It has
(immediate, 5 and 6, and late, 18) toward Fe(CN)6 is very
similar. Right after ozonolysis and destruction of H2O2 with
catalase, the rate constant of the then prevailing organic
hydroperoxides 5 and 6 was k = 0.4 dm3 molϪ1 sϪ1 and after
keeping the samples for one hour to let 5 and 6 decay into 18,
the rate constant was k = 1 dm3 molϪ1 Ϫ1. When based on the
s
iodide assay, 5 and 6 yield two mol Fe(CN)63Ϫ, while 18 yields
2.9 mol. Fe()-based determinations of hydroperoxides often
lead to higher than stoichiometric (2 mol Fe() per mol
hydroperoxide) relationships (one electron reduction of organic
hydroperoxides leads to the formation of alkoxyl radicals which
may undergo fragmentation yielding alkyl radicals; subsequent
O2-addition and reduction gives rise to new hydroperoxides,
cf. ref. 30). Interestingly, the reactivities of these two sets of
hydroperoxides towards the two reductants studied here, iodide
and Fe(CN)64Ϫ, are reversed.
HPLC and UV-spectroscopy. Right after ozonolysis, a
strongly UV-absorbing (λmax = 256 nm) product (Fig. 4, assigned
Fig. 3 Ozonolysis of thymine. Normalised yields of total hydro-
peroxides (᭺) and hydrogen peroxide (•) as a function of time after
ozonolysis. Hydrogen peroxide yields were determined after the organic
hydroperoxides had been destroyed with sulfide.
been shown that H2O2 does not react with bis(2-hydroxyethyl)
sulfide at an appreciable rate.30 Based on the catalase and the
sulfide assays, it is hence concluded that H2O2 is present in 25%
yield right after ozonation.
Thus, at least three kinds of hydroperoxides are formed
immediately upon ozonation, an acidic hydroperoxide (assigned
to 5, see below), a neutral hydroperoxide (6, suggested, see
below) and H2O2. The organic hydroperoxide remaining one
hour after ozonation is a further neutral hydroperoxide
(assigned to 18, see below).
As is seen from Fig. 3, the total hydroperoxide yield is not
much (∼7%) decreased over time. A fraction of the sulfide-
reactive ones decays yielding partly H2O2. This reaction pro-
ceeds with a half-life of ∼12 min according to a computer
fit through the data (• in Fig. 3). Within error limits, it
follows the same kinetics as the “slow” acid release reported
above (t1/2 = 10.5 min), i.e. a new organic hydroperoxide
(assigned to 18, see below) appears during the decay of the
primary ones. The yield of H2O2 during this decay of the pri-
marily formed hydroperoxides (5 and 6, see below) is much
lower (∼15%) than that of formic acid (∼75%). The loss of total
hydroperoxide yield (∼7%) also occurs at the same time scale
(cf. Fig. 3).
Fig. 4 UV absorption spectra (and retention times with water as
eluent) of the products formed in the ozonolysis of thymine. Frame A:
•, the anion of 1-hydroperoxymethylene-3-(2-oxopropanoyl)urea 5a
(7.6 min); ᭺, 5-hydroperoxy-5-methylhydantoin 18 (6 min). Frame B:
ꢀ, 1-formyl-5-hydroxy-5-methylhydantoin 11 (8.3 min); ᭝, 5-hydroxy-
5-methylhydantoin 13 (5.4 min, reference material available).
to 5a, see below) is observed by HPLC on a reversed-phase
column with a retention time of 7.6 min using water as eluent.
There is also a second product (assigned to 11, see below) elut-
ing at 8.3 min. This has only an end-absorption tailing towards
220 nm. Both products must be quite polar, since they elute
much earlier than thymine (15.6 min). As will be discussed
below, we are forced to account for a further organic hydro-
peroxide (6, suggested). Compared to 5, hydroperoxide 6 is
expected to have a lower UV-absorption and may be masked if
its retention time is very close to that of 5.
When the pH of the eluent is decreased (with sulfuric acid), 5
elutes somewhat later with a concomitant shift of its absorption
maximum towards 237 nm (cf. inset in Fig. 5). From the data
shown in the main graph of Fig. 5, the pKa value of 5 is
calculated at 4.0. For taking the spectra in Figs. 4 and 5, differ-
ent set-ups (including columns) were used. It is noted that the
spectrum of 5a given in Fig. 5 lacks the short-wavelength
absorption shown in Fig. 4. Possibly, on the column used for
running the spectra shown in Fig. 5 the separation was better,
and in Fig. 4 the spectrum of 5a is contaminated by contribu-
tions of 6. For this most likely weakly-absorbing hydroperoxide
no UV-spectrum is available. The suggestion that 5 and 6
coelute is supported by post-column derivatisation with molyb-
date-activated iodide, where right after ozonolysis only two
Hydrogen peroxide does not react with iodide at an appre-
ciable rate unless activated by molybdate, but the more reactive
hydroperoxides do. The hydroperoxides present right after
ozonolysis (5 and 6, see below) react with iodide (without
molybdate activation) more rapidly (k = 43 dm3 molϪ1 sϪ1
,
determined by stopped flow) than the one remaining after one
hour (18, for its identification see below; k = 7.5 dm3 molϪ1 sϪ1).
Accordingly, the apparent kobs of the reaction changes with
time, since the rate constants of these hydroperoxides differ
only by a factor of ∼5.7, i.e. these two reactions are not well
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
1575

View Article Online
product. As expected from the above, 11 does not lose formic
acid at natural pH, i.e. it does not decay into 13 under these
conditions.
The molar absorption coefficient of 5a at 256 nm must be
higher than that of thymine at the same wavelength. When
equal volumes of thymine (4 × 10Ϫ4 mol dmϪ3) and ozone
(2.25 × 10Ϫ4 mol dmϪ3) solutions were mixed, the absorption at
256 nm rose by a factor of 1.18 (taking the dilution by the
ozone solution into account) and subsequently decayed again
(k = 1.0 × 10Ϫ3 sϪ1, see above) to a level which is compatible with
the expected remaining thymine concentration and no signifi-
cant absorption of the remaining products at 256 nm. Taking
the yield of the 256 nm species as 34% (based on the imme-
diate conductance formation, cf. Fig. 1), its molar absorption
coefficient is calculated to be 2.5 × 104 dm3 molϪ1 cmϪ1
,
i.e. more than three times that of thymine at its maximum at
265 nm (ε = 7.9 × 103 dm3 cmϪ1 sϪ1).
Fig.
5 Percentage of undissociated 1-hydroperoxymethylene-3-(2-
LCMS. The LCMS-ESI technique in the positive mode was
used for the determination of the molecular weights of some of
the key products. This technique has the disadvantage that
some compounds do not give rise to a pronounced spectrum
even if present at a reasonably high concentration. Thus, they
escape detection. Yet, some of the hydroperoxides were detect-
able. Some of these, e.g. 5, rapidly decay at room temperature.
At 3 ЊC, the decay of 5 is much slower (1.3 × 10Ϫ4 sϪ1, cf. inset in
Fig. 1). This eight-fold prolonged lifetime enabled us to carry
out LCMS measurements with a sample that had been ozo-
nated at 3 ЊC. A peak at m/z = 175, (M ϩ H)ϩ, was observed
indicating that the molecular weight of 5 is 174 Da, i.e. it
contains three more oxygens compared to thymine. The mass
spectrum of 5 was weak, since even at 3 ЊC a considerable frac-
tion of it had decayed. No mass spectral evidence was obtained
for 6, the other inferred primary organic hydroperoxide. Prod-
uct 11, the complement of H2O2 formed immediately upon
ozonation (see below), does not give rise to a mass spectrum
under these experimental conditions.
oxopropanoyl)urea 5 as a function of pH as obtained by HPLC. The
pH was varied by varying the pH of the eluent. UV spectra of 5 (solid
line: λmax = 237 nm, eluent pH 2.6) and its anion (5a, dashed line:
λmax = 256 nm, eluent pH 7).
hydroperoxides, H2O2 and a rather broad peak due to organic
hydroperoxide(s) (retention times under these conditions 7.6
min, 22% yield, and 9.7 min, 78% yield) were detected. One
hour after ozonation the H2O2 peak has increased (39% of total
hydroperoxide), and the peak of the organic hydroperoxide
(61%, attributed to 18) is now sharp and shifted to 9.5 min.
The hydroperoxide 5 decays by the same kinetics (Fig. 6,
When 5 had decayed, the mass spectrum of 18 could be taken
and showed a pronounced m/z = 147, (M ϩ H)ϩ, together with
∼10% m/z = 293 (2M ϩ H)ϩ i.e. its molecular weight is 146 Da.
Similarly to product 11, the final product 5-hydroxy-5-methyl-
hydantoin 13 (authentic reference material was available) does
not give rise to a mass spectrum under these conditions.
As mentioned above, 5 and 18 are hydroperoxides and are
eliminated by the addition of bis(2-hydroxyethyl) sulfide, and as
a consequence of this, compounds giving rise to an LCMS
response are no longer present except for the resulting sulfoxide
(MW = 138 Da) which shows a pronounced signal at m/z = 139,
(M ϩ H)ϩ together with some m/z = 277 (2M ϩ H)ϩ. The parent
sulfide does not give an LCMS signal.
Fig. 6 Ozonolysis of thymine. Decay of the anion of 1-hydroperoxy-
methylene-3-(2-oxopropanoyl)urea 5a and build-up of 5-hydroperoxy-
5-methylhydantoin 18 as a function of time. Inset: first-order kinetic
plots of the data.
Assignment of the products. Product 5 has a molecular
weight of 174 Da. It is a strongly oxidising hydroperoxide,
i.e. the hydroperoxide function must be activated by electron-
withdrawing groups. It is a fairly strong acid (pKa = 4.0). The
spectral shift from 237 to 256 nm upon deprotonation requires
a considerable conjugation of the π system. Its structure is
assigned to 1-hydroperoxymethylene-3-(2-oxopropanoyl)urea.
Product 11 is not a peroxide. Upon hydrolysis, 11 is converted
into 5-hydroxy-5-methylhydantoin 13 and formic acid. Hence,
11 is assigned to 1-formyl-5-hydroxy-5-methylhydantoin.
Product 18 is a hydroperoxide with the molecular weight of
146 Da. Its precursors are 5 and 6 (see below). Upon reduction
of 18 with sulfide, 5-hydroxy-5-methylhydantoin 13 is formed.
Hence, 18 is assigned to 5-hydroperoxy-5-methylhydantoin.
inset: k = 1.0 × 10Ϫ3 sϪ1) as found for the release of formic acid
(k = 1.1 × 10Ϫ3 Ϫ1, see Table 1). It gives rise to the hydro-
s
peroxide 18 which elutes at 6.0 min. It absorbs at shorter wave-
length (λmax = 220 nm) than its precursor 5 (cf. Fig. 4).
The hydroperoxides 5 and 18 are both wiped out upon the
addition of bis(2-hydroxyethyl) sulfide which shows that they
must be strongly oxidising hydroperoxides (cf. ref. 30). The
hydroperoxide 5 (together with 6, see below) is reduced to 11,
while 18 is reduced to 13. When the sulfide was added one hour
after ozonation (to let the primary hydroperoxides (5 and 6)
decay into 18), the yield of 5-hydroxy-5-methylhydantoin 13
was 67% (reference material for the quantification of 13 was
available). When this solution was subsequently treated with
NaOH at pH 10.5 overnight, product 11 also disappeared,
and the yield of 5-hydroxy-5-methylhydantoin 13 increased to
∼100%. When the sulfide is added immediately after ozonolysis,
1-formyl-5-hydroxy-5-methylhydantoin 11 is the only observed
Ozonolysis at high pH. Thymine deprotonates at high pH
(pKa = 9.9), and the rate of reaction with ozone becomes con-
siderably faster (k = 3 × 106 dm3 molϪ1 Ϫ1).22 Concomitantly,
s
the reaction mechanism must be partially altered, since now the
1576
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582

View Article Online
formation of singlet dioxygen, O2(1∆g) is observed in 8% yield.29
At pH 11.6, the formation of the 256 nm species 5 is no longer
observed. This must be due to a change in reaction mechanism.
Its formation would not have escaped our attention, since the
kinetics of the decay of 5 at this pH is less than a factor of 2
faster than at pH 4 (shown by adjusting a thymine solution
ozonated at natural pH to pH 11.6).
reaction of thymine 1 with ozone (Fig. 2) is due to the form-
ation of 5a and a proton. In the sequence of reactions that lead
to 5a and Hϩ, the rate-limiting step is the reaction of ozone
with 1 [reaction (6)]. From the conductivity data shown in
Fig. 1, the yield of 5 must be close to 34% of ozone consumed.
Subsequent hydrolytic processes lead to the release of formic
acid (t1/2 ≈ 10.5 min at room temperature) monitored by the
“slow” increase in conductivity (Fig. 1). These conductivity
data suggest that the combined yield of the formic acid-
releasing hydroperoxides (attributed to 5 and 6) is close to 75%,
i.e. the yield of 6 is ∼41%. The kinetics of this “slow” rise in
conductivity is mirrored by a very similar kinetics of the dis-
appearance of the optical absorption of 5/5a. The fact that 5
has a pKa value of 4.0 and formic acid one of 3.75 warrants the
presence of additional precursors of the formic acid that is
released in the “slow” process. It is suggested that both 5 and 6
contribute here. The evaluation of build-up kinetics is fraught
with considerable errors, especially when more than one species
is involved.32 When the individual rate constants are not
drastically different, they cannot be disentangled, and the plots
shown in the inset of Fig. 1 are mainly given to show the
marked reduction in rate of hydrolysis upon reducing the tem-
perature to 3 ЊC. This enabled us to run the mass spectra of the
hydroperoxides by LCMS. The molecular weight of 5 is 174 Da,
agreeing with mass spectral data. There is no mass spectral
evidence for the formation of 6, but in view of its low concen-
tration at the time of measurement and the absence of any mass
spectral response in the case of 11 and 13, this lack of confirm-
ation does not carry much weight and thus does not rule out the
formation of 6. After hydrolysis, a hydroperoxide with the
molecular weight of 146 Da remains. It is assigned to 18. A
tentative suggestion as to the mechanism of the hydrolysis of 5
and 6 will be given below.
Thymine glycol is observed by GCMS after trimethylsilyl-
ation.
Mechanistic aspects. Ozone is a strongly electrophilic agent31
and will hence preferentially add to the C(5)-position of thy-
mine 1 [reaction (6)], as other electrophilic agents such as the
ؒ
ؒ
OH radical and the H atom do (cf. ref. 7). The zwitterion 2
will close the ring thereby forming the Criegee intermediate 3
[reaction (7)]. The Criegee intermediate 3 can now open the
ring in two directions [reactions (8) and (12)]. The zwitterion 4
formed in the predominating reaction (8) can eliminate the pro-
ton at the neighbouring nitrogen [reaction (9)]. Hydroperoxides
have generally high pKa values [cf. pKa(H2O2) = 11.8; pKa-
(HC(O)OOH) = 7.1], and the hydroperoxide function of 5 will
thus remain protonated at a pH < 7. The acidity observed for 5
[pKa(5) = 4.0] is thus due to a deprotonation at N(3) [reaction
(11)]. In competition with reaction (9), the zwitterion 4 may
react with water yielding the α-hydroxyalkyl hydroperoxide 6
[reaction (10)]. One also may consider that the zwitterion 4 may
form an N(3)–C(6) bond upon losing the acidic proton at N(3).
However, the strain exerted by the four-membered ring may not
be much in favour of this reaction, and hence this possibility is
not shown in the reaction scheme.
The hydroperoxide 5 has a conjugated π system and shows a
strong absorption at 237 nm (cf. Fig. 5). The proton at N(3) is
acidic [equilibrium 11, pKa(5) = 4.0], and the absorption maxi-
mum is shifted to 256 nm upon deprotonation. The fast build-
up of conductivity that occurs on the same time scale as the
In competition with reaction (8), the Criegee intermediate 3
can decay according to reaction (12). The zwitterion 7 may give
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
1577

View Article Online
rise to the hydroperoxides 8 and 9 [reactions (13)–(15)]. The
hydroperoxide 9 is an α-hydroxyalkyl hydroperoxide. Such
hydroperoxides often eliminate H2O2 so rapidly that it is not
possible to determine their lifetime. A case in point is hydroxy-
benzyl hydroperoxide.33 On the other hand, there are also
α-hydroxyalkyl hydroperoxides such as hydroxymethyl hydro-
peroxide which have very long lifetimes in neutral solution
and release H2O2 rapidly only at high pH.34 The structural
parameters that determine the stability of α-hydroxyalkyl
hydroperoxides are as yet not known. We suggest that hydro-
peroxide 9 belongs to the unstable ones and that the H2O2
observed right after ozonolysis is due to reactions (16) and/or
(18). The resulting product 10 will convert to 1-formyl-5-
hydroxy-5-methylhydantoin 11. This product is rather stable
and releases formic acid only upon treatment at high pH. The
yields of immediate H2O2 and formic acid released at high pH
are both ∼25%. It is thus suggested that reaction (14) dominates
over reaction (13) and that reaction (15) is slow compared to
reactions (16) and/or (18). However, the ozonolysis of thym-
idine which will be discussed below takes a very different route,
and acetic acid is released from an intermediate analogous
to 9. No acetic acid is formed in the thymine system that
we are concerned with here. This requires that in the thymine
system there must be a much faster process competing with a
potentially also possible acetic acid-releasing process. These
two systems differ in the substitution at N(1). The rapid release
of H2O2 in the thymine system may thus involve the N(1)
proton. We thus suggest that H2O2 is released in the con-
certed pathway (18) which may well proceed with a relay of
water molecules to accommodate a good transition state for the
reaction to take place. In thymidine, the rate of acetic acid
release is 0.55 sϪ1 (see below). To compete with such a process
effectively, the rate of reaction (18) must be considerably
faster.
The reduction of hydroperoxide 5 by sulfide is depicted
in reaction (19) [followed by reactions (20) and (17)] (hydro-
peroxide 6 will undergo analogous reactions), and since 1-
formyl-5-hydroxy-5-methylhydantoin 11 eliminates formic acid
only at high pH [reaction (21)], not only the conductance signal
increase observed right after the addition of ozone (dissociation
of 5) but also the subsequent slow increase due to the release of
formic acid disappears upon the addition of the sulfide. After
hydrolysis at high pH, the only remaining product is 5-hydroxy-
5-methylhydantoin 13 (100% after the addition of sulfide,
reference material was available).
The “slow” release of formic acid with concomitant decay
of hydroperoxide 5 and the other related hydroperoxide 6 may
be rationalised by assuming that these hydroperoxides form
α-hydroxyendoperoxides such as 14 [equilibrium (22)]. Hydro-
peroxides readily undergo such a reaction with carbonyl com-
pounds. Often, the equilibrium constants are quite high, and
the instability of α-hydroxyalkyl hydroperoxides [cf. reaction
(16)] is then due to the inefficiency of the back-reaction at the
low concentrations of the reactants (∼10Ϫ4 mol dmϪ3) that we
are concerned with here. This situation does not prevail in the
intramolecular addition reaction (22). Addition of water to 14
leads to 15 [reaction (23)]. It is noted that upon endoperoxide
formation, 6 immediately gives rise to 15. It is important that in
the next step no N-formyl compound is formed. This would not
hydrolyse as readily (t1/2 ≈ 13 min) as observed. Instead, we
suggest reaction (24) which leads to the amide 16. This may
then undergo ring closure forming the hydantoin derivative 17
[reaction (25)] which subsequently hydrolyses [reaction (26)].
Mass spectral evidence for the formation of the hydroperoxidic
hydantoin 18 has been given above. Its reduction by sulfide to
5-hydroxy-5-methylhydantoin 13 [reaction (27)] has been ascer-
tained as well. It has been mentioned above that the decay
kinetics of 5 and the build-up of formic acid are practically
identical and do not deviate too much from a first-order rate
law (cf. inset in Fig. 1). This could be accounted for if the rate
determining step in this sequence of reactions is the formation
of the endoperoxide 14 (or 15 from 6).
In the absence of sulfide, the hydroperoxides 5 and 6 mainly
decay into the hydroperoxide 18 and formic acid. A small
fraction also loses H2O2 in competition (cf. Fig. 3). Since the
suggested mechanism involves an α-hydroxyalkyl peroxide as
intermediate (15 may convert into 9), this is not unexpected.
During this decay, the total hydroperoxide yield is also reduced
to a small extent (7%). For this minor pathway that must lead to
the formation of higher oxidised products, we do not present a
mechanistic suggestion.
At high pH, thymine deprotonates [equilibrium (28), pKa =
9.9]. Since the acidity of N(1) and N(3) is very similar, anions
1a and 1b are both present at comparable concentrations.35 It is
tempting to predict that 1a has a higher electron density in the
C(5)–C(6) double bond which would favour reaction (31).
However, thymidine which can only deprotonate at N(3) shows
a similar increase in rate upon deprotonation as thymine.
Hence, no prediction as to the preference of reaction (31) vs.
reaction (29) can be made as yet.
The hydrotrioxide anion 2a formed in reaction (31) has no
positive charge at C(6) for a rapid ring closure to the Criegee
intermediate. This will enhance the lifetime of this hydro-
trioxide, i.e. it stands a chance of decomposing into 19 and
singlet dioxygen, O2(1∆g) [reaction (32)].
Under these conditions the singlet dioxygen yield is 8%.29
This value is low when compared with the yield of O2(1∆g)
formation in the case of the 5-chlorouracil anion (∼44%).29 The
reasons for this surprisingly low yield are not yet fully under-
stood. Spin conversion and elimination of ground-state triplet
dioxygen may compete. This process is energetically favoured
by 105 kJ molϪ1. This effect is most prominent in the case of
the reaction of IϪ and BrϪ with ozone.30 In these systems,
spin-orbit coupling via the heavy-atom effect favours singlet
triplet conversion, and singlet dioxygen formation is dramatic-
ally reduced. Here, where we compare thymine with 5-chloro-
uracil, this explanation does not hold, since it would only
explain a reduced O2(1∆g) yield in the case of 5-chlorouracil.
When a hydrotrioxide anion has a long lifetime and the ener-
getics of singlet dioxygen elimination disfavour the reaction
to proceed effectively, loss of ground-state triplet dioxygen has
1578
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582

View Article Online
Fig. 7 Build-up of conductance as a function of time in the reaction
of thymidine (8 × 10Ϫ4 mol dmϪ3) with ozone (6 × 10Ϫ5 mol dmϪ3).
Inset: the fast build-up resolved by the stopped-flow technique.
formic acid, acetic acid is formed here as well (the latter is
absent in the case of thymine). The yield of this fast acid form-
ation is ∼18%, in agreement with the yield of acetic acid (see
Table 3) when correction for the only partial dissociation of
acetic acid (pKa = 4.8) is made. Addition of sulfide 30 s after
ozonation suppressed significantly the yield of formic acid but
did not influence the acetic acid yield. It is thus concluded that
acetic acid is released during this fast conductance rise.
The slower part of the conductance rise is due to the form-
ation of formic acid, and its kinetics can be described by a first-
order process (k ≈ 9 × 10Ϫ3 sϪ1). When the ozonated sample was
kept for two hours at pH 11.6, the formic acid yield increased
from ∼76% to ∼100%. The eluent used for IC (1 × 10Ϫ2 mol
dmϪ3 bicarbonate, pH 8.6) is slightly basic and thus labile
formyl compounds may release formic acid upon chromato-
graphy beyond the amounts detected during the 8 min used to
follow the conductance increase shown in Fig. 7 (main graph).
For a compilation of yields see Table 3.
been considered to proceed with quite some efficiency.36 Fur-
thermore, protonation of the hydrotrioxide anion by water may
compete with its decay and other reactions of the hydrotrioxide
may take place. Hydrotrioxides have often been assumed to be
intermediates in the ozonation of aliphatic compounds,37–39 but
have recently been quite well characterised,40 although their
decay pathways are still not yet fully elucidated. It is an import-
ant observation that the decay of the hydrotrioxide derived
from propan-2-ol is strongly catalysed by water.40 On the singlet
level, the water-catalysed decomposition of H2O3 is fairly well
understood theoretically (and thus also that of organic hydro-
trioxides).41 Yet, singlet dioxygen was not detected in the ozon-
olysis of aqueous propan-2-ol solution at room temperature.29
After a few seconds, no iodide or Fe() oxidising species was
detectable (our own observations). Thus, it seems that water
catalysis may have resulted in the release of ground-state (trip-
let) dioxygen. The present system is also poorly understood,
and a more detailed product study than is presented here is
highly desirable.
Product 19 formed in reaction (32) is an isopyrimidine
species. Isopyrimidines are well-documented intermediates
formed in the free-radical chemistry of uracil and its derivatives
(cf. refs. 42,43). In the present system, addition of water to the
N(1)–C(6) double bond will result in the formation of thymine
glycol 20 [reaction (33)]. Its formation has been ascertained (see
above), but no attempts have been made as yet to quantify this
and other conceivable products.
Formation and decay of hydroperoxides. When the
molybdate-activated iodide reagent is added immediately, i.e.
∼1–2 s after the addition of ozone, the total hydroperoxide yield
thus determined is close to 100%. When the reagent is added
after 25 s, only ∼78% hydroperoxides are detected. Hence, there
must be a very short-lived organic hydroperoxide with a yield of
∼22%. An addition of catalase reduces the total hydroperoxide
yield by 8%, i.e. only little H2O2 is formed. As determined by
stopped-flow, the reactive organic hydroperoxide(s) present
after 1 min react with iodide (without molybdate activation)
with a rate constant of ca. 100 dm3 molϪ1 Ϫ1, and after 8 min
s
when most of the formic acid is released (cf. Fig. 7) a lower rate
constant of around 45 dm3 molϪ1 sϪ1 is measured. After the
rapid loss, the total hydroperoxide yield remains practically
stable over time, and in contrast to thymine no H2O2 is released.
Upon the addition of sulfide, the organic hydroperoxides are
eliminated. The remaining hydroperoxide detectable with
molybdate-activated iodide is H2O2 (confirmed by its kinetics).
Thymidine
Thymidine reacts with ozone with a rate constant of 3 × 104
dm3 molϪ1 sϪ1 and its anion with a rate constant of 1.2 × 106
dm3 molϪ1 sϪ1 22
. In contrast to thymine, no singlet dioxygen is
formed at high pH.29 The products that have been reported in
an earlier study20 are also so markedly different from those
reported above for the thymine reaction, that it seemed worth-
while to have a closer look at mechanistic details.
HPLC and LCMS. Upon HPLC four major UV-absorbing
peaks (retention times with water as eluent: 11.4, 12.8, 14.9 and
25.0 min; thymidine elutes at 45 min; H2O2 is not detectable by
this method) disappeared upon sulfide addition and are thus
attributed to reactive hydroperoxides. In contrast to thymine, an
acidic hydroperoxide is not formed. Post-column derivatisation
with molybdate-activated iodide monitors also weakly oxidis-
ing hydroperoxides such as H2O2 and allowed us to quantify
their yields. Since a lower flow rate had to be used, the retention
times are longer than those mentioned above (7.6 min, H2O2
(11.5%), 9.5 min (13.5%), 15.3 min (37%), 25.4 min (4%), 26.4
Conductometry and ion chromatography. Similarly to thy-
mine, there is a fast and a slow build-up of conductance upon
the addition of an ozone solution (Fig. 7).
In contrast to thymine, the fast process is two orders of mag-
nitude slower (0.55 sϪ1) than the reaction of thymidine with
ozone under these conditions (60 sϪ1). This indicates that this
fast increase in conductance cannot be due to the formation of
an acidic hydroperoxide (as 5 in the case of thymine). Besides
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
1579

View Article Online
Table 3 Compilation of yields (with respect to ozone consumed) in the ozonolysis of thymidine
Process
Yield (%)
Reference
Acetic acid
18
∼18
∼40–45
76
100
∼100
78
8
8
8
70
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
This work
Ref. 20
Fast acid release (as acetic acid, conductometry)
Acid release (8 min, as acetic plus formic acids, conductometry)
Formic acid (IC)
Formic acid after 2 h at high pH (IC)
Total hydroperoxide (immediate)
Total hydroperoxide (25 s–1.5 h)
Hydrogen peroxide (25 s, catalase assay)
Hydrogen peroxide (25 s, R2S assay)
Hydrogen peroxide (after 1 h)
Organic hydroperoxides (after 25 s)
Organic hydroperoxides (after 1 h)
N1-(2-deoxy-β--erythropentofuranosyl)-5-hydroxy-5-methylhydantoin, two
isomers 23
N1-(2-deoxy-β--erythropentofuranosyl)-5-hydroxy-5-methylhydantoin, two
isomers 23 (after R2S treatment)
N-(2-deoxy-β--erythropentofuranosyl)formamide 21
N-(2-deoxy-β--erythropentofuranosyl)formylurea 22
Singlet dioxygen (at high pH)
70
19.5
43–50
This work
19
18
Absent
Ref. 20
Ref. 20
Ref. 29
min (4%), 29.9 min (3%), 32.5 min (5%)). The larger number of
products as compared to thymine is largely due to the fact that
now meso/( ) isomers are formed. In contrast to thymine, after
five hours only small changes (retention times, number of
peaks, yields) are observed in this HPLC chromatogram.
Upon LCMS of a sample ozonated at 3 ЊC, three species were
detected with characteristic peaks at m/z = 162, m/z = 177
(together with 117 and 353) and m/z = 263 (together with 117).
The latter peak was broad and is possibly due to two isomers.
Upon reduction with bis(2-hydroxyethyl) sulfide the product
characterised by m/z = 263 disappeared, and a new product with
m/z = 247 (together with 117) is observed. Thus, the m/z = 263
species must be a hydroperoxide. The m/z = 117 peak in these
mass spectra is typical of the 2-deoxyribosyl moiety, i.e. all
these products retain the sugar moiety.
syl)-5-hydroxy-5-methylhydantoin 23 (MW = 246 Da). The
formation of 23 has been reported before.20
Further species with weak signals are observed in this heavily
ozonated sample, but it would be too speculative to try an
assignment.
A few mg were isolated by preparative HPLC as reference
material. After rotary evaporation, this material remained oily,
and an HPLC chromatogram still showed some impurities.
Using this material for calibration, the combined yield of the
two hydantoins 23 was determined at 43%. Considering that the
reference material contained impurities we also calculated their
yield based on the assumption that they had the same molar
absorption coefficient as their parent, 5-hydroxy-5-methyl-
hydantoin 13 for which reliable material was available. We then
obtain ∼50%. In any case, the yield of 23 is much higher than
the reported20 value of only 19.5%. In the latter study,20 the
hydroperoxides were subjected without prior reduction to a
laborious work-up, and possibly only a fraction underwent
H2O2 elimination, while the rest was degraded.
In the earlier study, the formation of N-(2-deoxy-β--erythro-
pentofuranosyl)formamide 21 (MW = 161) has been reported
and its yield determined at 19%.20 This product seems to be
present under our conditions as well (characterised by a
prominent peak at m/z = 162, the M ϩ 1 ion).
UV spectroscopy. Equal volumes of thymidine (2 × 10Ϫ4 mol
dmϪ3) and ozone (1.52 × 10Ϫ4 mol dmϪ3) solutions were mixed
and the ensuing UV spectra run at 40 s and 10 min. From the
measured UV spectra, the contribution of the remaining thym-
idine (now present at 4.8 × 10Ϫ5 mol dmϪ3) was subtracted.
The difference spectra are shown in Fig. 8. While the species
The m/z = 177 species is assigned to N-(2-deoxy-β--erythro-
pentofuranosyl)urea 22. Its molecular weight is 176, and the
observed ion would be (M ϩ H)ϩ. The also observed m/z = 353
is accounted for by the cluster ion (2M ϩ H)ϩ. The previous
study20 does not report the formation of this product.
The m/z = 263 is due to a hydroperoxide. The primary
hydroperoxide would have a molecular weight of 308 Da and
should show an m/z = 309 (M ϩ H)ϩ. Such a species is not
observed, either because of its poor response or due to its rapid
decay (considerable time after ozonolysis had elapsed before the
mass spectra could be taken; the LCMS set-up was in a distant
building). The molecular weight of the secondary hydro-
peroxide (loss of formic acid) N1-(2-deoxy-β--erythropento-
furanosyl)-5-hydroperoxy-5-methylhydantoin 24 is 262 Da, and
we assign the observed 263 species to its (M ϩ H)ϩ ion. For this
product, we would expect two isomers, and the broadness of the
peak would account for this.
Fig. 8 Thymidine (7.6 × 10Ϫ5 mol dmϪ3) and product spectra after
reacting equal volumes of thymidine (•, 2 × 10Ϫ4 mol dmϪ3) and ozone
(1.5 × 10Ϫ4 mol dmϪ3) solutions and subtracting the remaining
thymidine. Spectra taken at ∼40 s (∆) and 10 min (ꢀ). Running the
spectra took 12 s.
The m/z = 247 species is only observed after reduction with
sulfide. Its precursor is the hydroperoxide 24, and we attribute it
to the (M ϩ H)ϩ ion of N1-(2-deoxy-β--erythropentofurano-
1580
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582

View Article Online
with λmax = 238 nm decays, one with λmax = 226 nm builds up.
The kinetics at these two wavelengths are the same (k ≈ 9 × 10Ϫ3
s
the Criegee ozonide [cf. reactions (6) and (7)]. Subsequent ring
opening according to the major pathway [cf. reaction (8)] leads
to a zwitterion which only can deprotonate and yield a C–N
double bond, when the nitrogen carries a hydrogen as substitu-
ent (as in thymine). Since in thymidine N(1) carries the
2-deoxyribosyl group instead, deprotonation and concomitant
formation of an acidic hydroperoxide such as 5 is no longer
possible. Here, the reaction analogous to reaction (10) that
would lead to 2-deoxyribosyl-substituted 6 is the most likely
one to occur. For this species we do not have mass spectral
evidence, but it is tempting to attribute the 238 nm species to
this intermediate. The subsequent formation of formic acid,
Ϫ1). This value agrees with the release of formic acid (slow step
in Fig. 7, main graph).
The same experiment as shown in Fig. 8 has been repeated by
stopped-flow. There, the first observable product is also the
species characterised by λmax = 238 nm. The kinetics of the
spectra only reveal the development of two products. Right
after ozonolysis the 238 nm species is present and decays follow-
ing first-order kinetics (k ≈ 9 × 10Ϫ3 sϪ1) into the 226 nm species.
Since the yields of the 238 nm and 226 nm species are
not known, we are not able to give their molar absorption co-
efficients, but they must be > 104 dm3 molϪ1 cmϪ1 at their
maxima.
cf. Fig. 7, will lead to 24 via an endoperoxide (cf. 6
15
16 17 18) as depicted for the aglycon in reactions
The kinetics that are observed by UV spectroscopy are paral-
leled by the slow conductance rise (cf. Fig. 7) which is due to
formic acid release. The process observed by fast conductom-
etry (k = 0.55 sϪ1) is not detected by UV spectroscopy.
(25)–(27). The release of formic acid which occurs at the same
rate as the decay of the 238 nm species could account for the
formation of N1-(2-deoxy-β--erythropentofuranosyl)-5-hydro-
peroxy-5-methylhydantoin 24 and the latter may thus have an
absorption maximum at 226 nm. This is in agreement with the
UV spectra of the hydroperoxides detected by HPLC. For 24
there is also mass spectral evidence (see above).
For the minor pathway of the decay of the Criegee ozonide
shown for thymine as reaction (12) the fast hydantoin form-
ation and concomitant release of H2O2 does not take place due
to the lack of the hydrogen at N(1) (see above). Instead, acetic
acid is released. The elimination of acetic acid with a concomi-
tant drop in the total hydroperoxide yield is reminiscent of the
rapid33 decay of 2-hydroperoxy-2-hydroxyacetic acid [reaction
(34)].
Mechanistic aspects. Although there are considerable similar-
ities between thymine and thymidine in their reactions with
ozone, there are also marked differences. In the case of thymine,
the total hydroperoxide yield is 100% and decreases very little
(by 7%) over more than two hours. With thymidine, the
hydroperoxide yield is close to 100% only right after ozonolysis,
but already after 25 s it has dropped to ∼78%. Formation of
H2O2 is important (25%) with thymine, but minor (8%) with
thymidine. In thymidine, there is a rapid (0.55 sϪ1) release of
acetic acid (18%), a product that is not formed in the case
of thymine. An acidic hydroperoxide such as 5 (34% in the case
of thymine) is not formed with thymidine. Common to both
systems is the full yield (∼100%) of formic acid after reduc-
tion of hydroperoxides with sulfide and treatment at high pH.
Yet, while 5-hydroxy-5-methylhydantoin 13 then reaches 100%
in the case of thymine, the corresponding thymidine product
23 is only formed in a 43–50% yield. Much of this deficit can
be accounted for by fragment products such as N-(2-deoxy-
β--erythropentofuranosyl)formamide 21 and N-(2-deoxy-β-
Here, we suggest that the primary hydroperoxide 25 can
deprotonate at N(3) [reaction (35)], the N(3)H is acidified by
two carbonyls in the α-position and further electron-
withdrawing groups in the β-position, and that the ensuing
anion undergoes the fragmentation reaction (36). The rate of
acetic acid release is 0.55 sϪ1 under the conditions of Fig. 7. The
product of reaction (36), 27, will be unstable and hydrolyse
thereby eliminating CO2 [reaction (37)]. The product that is
formed in this reaction, N-(2-deoxy-β--erythropentofurano-
syl)-N-formylurea 28, as well as its further degradation product
N-(2-deoxy-β--erythropentofuranosyl)formamide 21 [cf. reac-
tion (39)], has been reported to be formed in high yield (cf.
Table 3).20 Above, it has been suggested that N-(2-deoxy-β--
erythropentofuranosyl)urea 22 [cf. reaction (38)] may also be
among the products.
-erythropentofuranosyl)urea 22.
A
major problem in
presenting a somewhat detailed mechanism for thymidine
ozonolysis, is the lack of a material balance. It has already been
pointed out above that we also cannot rely on the yields pre-
sented in the earlier study,20 since considerable degradation
of hydroperoxidic material must have occurred upon work
up. Mechanistically, they20 account for the formation of the
N1-(2-deoxy-β--erythropentofuranosyl)-5-hydroxy-5-methyl-
hydantoins 23 which are their (and also our) major products by
an elimination of H2O2. However, H2O2 is only formed in 8%
yield (Table 3), and 24 is stable for hours (unless subjected to
severe chromatographic conditions). In the earlier study,20 it
has also not been realised that acetic acid is a major product
(18%). Especially the formation of the latter is a kind of key to
mechanistic differences between thymine and its nucleoside.
It is reasonable to assume that in both systems ozone attack
will be preferentially at C(5) with the subsequent formation of
The sum of the reported yields20 of 21 and 28 is too high
(37%) to balance the formation of acetic acid (18%, this work).
Thus, they must be formed from other precursors (e.g. 24)
during work-up as well. If N-formylurea is a good model for 28,
the latter must hydrolyse rather slowly (at natural pH, 0.1 M
N-formylurea releases only 0.04% formic acid within two hours.
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582
1581

View Article Online
9 R. A. Floyd, M. S. West, W. E. Hogsett and D. T. Tingey,
Plant Physiol., 1989, 91, 644.
10 R. H. Fetner, Nature, 1962, 194, 793.
11 N. Shinriki, K. Ishizaki, A. Ikehata, T. Yoshizaki, A. Nomura,
K. Miura and Y. Mizuno, Biochim. Biophys. Acta, 1981, 655, 323.
12 K. Ishizaki, N. Shinriki, A. Ikehata and T. Ueda, Chem. Pharm.
Bull., 1981, 29, 868.
As the pH is lowered, hydrolysis becomes progressively faster).
The relatively high yield of 21 (19%)20 may nevertheless have
been formed during the laborious work-up, but degradation
processes of labile (peroxidic) intermediates that have as yet not
been revealed may contribute as well.
13 N. Shinriki, K. Ishizaki, K. Miruba, T. Ueda and F. Harada,
Chem. Pharm. Bull., 1983, 31, 3601.
14 K. Ishizaki, N. Shinriki and T. Ueda, Chem. Pharm. Bull., 1984, 32,
3601.
15 N. Shinriki, K. Ishizaki, S. Sato, K. Miura, K. Sawadaishi and
T. Ueda, Chem. Pharm. Bull., 1984, 32, 3636.
16 K. Miura, T. Ueda, N. Shinriki, K. Ishizaki and F. Harada,
Chem. Pharm. Bull., 1984, 32, 651.
17 K. Sawadaishi, K. Miura, E. Ohtsuka, T. Ueda, K. Ishizaki and
N. Shinriki, Nucleic Acids Res., 1985, 13, 7183.
18 G. Unnikrishnan, K. Gopakumar and D. Krishnan, Radiat. Phys.
Chem., 1986, 28, 281.
19 J. Van der Zee, T. M. A. R. Dubbelman and J. Van Steveninck,
Free Radical Res. Commun., 1987, 2, 279.
20 I. Girault, D. Molko and J. Cadet, Free Radical Res., 1994, 20, 315.
21 I. Girault, S. Fort, D. Molko and J. Cadet, Free Radical Res., 1997,
26, 257.
22 J. A. Theruvathu, R. Flyunt, C. T. Aravindakumar and C. von
Sonntag, J. Chem. Soc., Perkin Trans. 2, 2001, 269.
23 L. Forni, D. Bahnemann and E. J. Hart, J. Phys. Chem., 1982, 86,
255.
24 E. J. Hart, K. Sehested and J. Holcman, Anal. Chem., 1983, 55, 46.
25 A. O. Allen, C. J. Hochanadel, J. A. Ghormley and T. W. Davis,
J. Phys. Chem., 1952, 56, 575.
26 P. Dowideit, Doctoral thesis, Ruhr-Universität Bochum, 1996.
27 L. Prager, P. Dowideit, H. Langguth, H.-P. Schuchmann and
C. von Sonntag, J. Chem. Soc., Perkin Trans. 2, 2001, 1641.
28 P. Dowideit and C. von Sonntag, Environ. Sci. Technol., 1998, 32,
1112.
29 F. Muñoz, E. Mvula, S. E. Braslavsky and C. von Sonntag, J. Chem.
Soc., Perkin Trans. 2, 2001, 1109.
30 R. Flyunt, A. Leitzke, G. Mark, E. Mvula, E. Reisz, J. A.
Theruvathu and C. von Sonntag, unpublished results.
31 J. Hoigné and H. Bader, Water Res., 1983, 17, 173.
32 H. Strehlow, Rapid Reactions in Solution, VCH, Weinheim, 1992.
33 A. Leitzke, E. Reisz, R. Flyunt and C. von Sonntag, J. Chem. Soc.,
Perkin Trans. 2, 2001, 793.
Conclusion
Although the ozonolyses of thymine and thymidine have many
mechanistic aspects in common, there is a noticeable influence
of the substituent at N(1) on the pathways taken beyond the
formation of the Criegee ozonide. In thymine, the Criegee ozo-
nide opens into an intermediate that is capable of deproton-
ation at the neighbouring N(1)H. This new kind of reaction in
ozone chemistry is not available to thymidine due to the lack of
a hydrogen at this position. There are also major differences in
the minor pathway. While acetic acid is released in the case of
thymidine (18%), no such process occurs with thymine. Con-
cerning ozone-induced DNA damage, it may be of importance
that highly reactive hydroperoxides are formed which may lead
to further DNA lesions, e.g., upon reaction of these hydro-
peroxides with transition metal ions (cf. Fenton-type reac-
tions).44 On the other hand, these hydroperoxides might be
readily destroyed by a sulfur compound such as gluthathione
which is quite abundant (near millimolar)45 in cells. In eukario-
tic cells where it is difficult to rationalise how ozone may reach
the nucleus, DNA lesions such as 8-hydroxyguanine9 may even
be caused by hydroperoxidic intermediates generated in the
reaction of ozone with cellular components other than DNA.
Certainly, there must be a long chain of events from the first
reaction of ozone with cellular components to the dramatic
morphologic alterations observed,46 e.g., in cells of the lung
exposed to ozone.
Acknowledgements
J.A.T. would like to thank the World Laboratory for a stipend
enabling him to carry out part of his doctoral thesis at the
Max-Planck-Institut für Strahlenchemie. We also thank Mr
H.-W. Klein (Max-Planck-Institut für Kohlenforschung) for
running the LCMS and assisting us with the interpretation of
the spectra. The continuing support by Professor R. Mehnert,
enabling us to finalise the work at the IOM, is highly
appreciated.
34 E. Bothe and D. Schulte-Frohlinde, Z. Naturforsch., Teil B, 1978, 33,
786.
35 H. Morita, J. S. Kwiatkowski and A. Tempczyk, Bull. Chem. Soc.
Jpn., 1981, 54, 1797.
36 P. S. Nangia and S. W. Benson, J. Am. Chem. Soc., 1980, 102, 3105.
37 P. S. Bailey, Ozonation in Organic Chemistry. Vol. 2. Nonolefinic
Compounds, Academic Press, New York, 1982.
38 D. H. Giamalva, D. F. Church and W. A. Pryor, J. Am. Chem. Soc.,
1986, 108, 7678.
39 M. N. Schuchmann and C. von Sonntag, J. Water Supply Res.
Technol.—AQUA, 1989, 38, 311.
40 B. Plesnicar, J. Cerkovnik, T. Tekavec and J. Koller, Chem. Eur. J.,
2000, 6, 809.
41 J. Koller and B. Plesnicar, J. Am. Chem. Soc., 1996, 118, 2470.
42 M. I. Al-Sheikhly, A. Hissung, H.-P. Schuchmann, M. N.
Schuchmann, C. von Sonntag, A. Garner and G. Scholes, J. Chem.
Soc., Perkin Trans. 2, 1984, 601.
References
1 G. Bünning and D. C. Hempel, GWF, Wasser/Abwasser, 1999, 140,
173.
2 R. T. Toledo, F. E. Escher and J. C. Ayres, Appl. Microbiol., 1998, 26,
592.
3 K. V. Ellis, Crit. Rev. Environ. Control, 1991, 20, 341.
4 E. Dahi, Water Res., 1976, 10, 677.
5 P. Neta, R. E. Huie and A. B. Ross, J. Phys. Chem. Ref. Data, 1988,
17, 1027.
6 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross,
J. Phys. Chem. Ref. Data, 1988, 17, 513.
43 M. N. Schuchmann, M. Al-Sheikhly, C. von Sonntag, A. Garner
and G. Scholes, J. Chem. Soc., Perkin Trans. 2, 1984, 1777.
44 J. A. Theruvathu, C. T. Aravindakumar, R. Flyunt and C. von
Sonntag, J. Am. Chem. Soc., 2001, 123, 9007.
7 C. von Sonntag, The Chemical Basis of Radiation Biology, Taylor
and Francis, London, 1987.
8 H. Dubeau and Y. S. Chung, Mutation Res., 1982, 102, 249.
45 P. Wardman and C. von Sonntag, Methods Enzymol., 1995, 251, 31.
46 G. L. Huber, R. J. Mason, M. LaForce and N. J. Spencer,
Arch. Intern. Med., 1971, 128, 81.
1582
J. Chem. Soc., Perkin Trans. 2, 2002, 1572–1582