The (5-4) and (6-4) adducts of 1-methylthymine and their Dewar valence isomers

Article abstract of DOI:10.1111/j.1751-1097.2011.00932.x

Previous studies of the photochemical reaction of 1-methylthymine (MeT) in frozen aqueous solution have indicated that four cyclobutane type dimers are formed. We have restudied this system and have found that, in addition to cyclobutane dimers, both a (5-4) adduct and a (6-4) adduct of MeT are formed in significant amounts. Upon standing in aqueous solution, the (5-4) adduct is susceptible to reaction to form an isomeric form of the parent adduct, possibly via ring-opening and closure reactions at C-6 of the saturated pyrimidine ring component of the adduct. Irradiation of each of these three adducts with UVB light produces a pair of Dewar-type adducts. The nine products were individually characterized by mass spectrometry, proton NMR spectroscopy and UV spectroscopy. A less comprehensive study showed that irradiation of thymidine in frozen aqueous solution produces a diastereomeric pair of (5-4) adducts, along with the previously known diastereomeric pair of (6-4) adducts. Irradiation of 1-methylthymine (center) at 254 nm in frozen aqueous solution produces, in addition to cyclobutane dimers, both a (6-4) adduct (left) and a (5-4) adduct (right). The initially formed (5-4) adduct is unstable toward thermal conversion to another isomeric (5-4) adduct. Irradiation of each of these adducts in fluid solution with UVB results in the formation of two Dewar valence isomers. Similarly, irradiation of thymidine, frozen in ice at -78.5°C, produces a diastereomeric pair of (5-4) adducts, along with a corresponding pair of previously known (6-4) adducts.

Full text of DOI:10.1111/j.1751-1097.2011.00932.x

Photochemistry and Photobiology, 2011, 87: 802–817
The (5-4) and (6-4) Adducts of 1-Methylthymine and Their Dewar
Valence Isomers
Martin D. Shetlar* and Janet Chung§
Department of Pharmaceutical Chemistry, School of Pharmacy, University of California, San Francisco,
CA, USA
Received 13 December 2010, accepted 30 March 2011, DOI: 10.1111/j.1751-1097.2011.00932.x
of (6-4) adduct formation Thy, upon irradiation with UVC
ABSTRACT
light in frozen aqueous solution, reacts to form 5-hydroxy-
6-4¢-(5-methylpyrimidin-2¢-one)-5,6-dihydrothymine (IIIa) (4).
Mixed CBDs and (6-4) adducts have also been reported (e.g.
CBD or (6-4) adducts containing both Thy and Cyt); such
products play particularly important roles in understanding
the photochemistry occurring in dinucleotides, as well as in
DNA itself.
Previous studies of the photochemical reaction of 1-methylthy-
mine (MeT) in frozen aqueous solution have indicated that four
cyclobutane type dimers are formed. We have restudied this
system and have found that, in addition to cyclobutane dimers,
both a (5-4) adduct and a (6-4) adduct of MeT are formed in
significant amounts. Upon standing in aqueous solution, the (5-4)
adduct is susceptible to reaction to form an isomeric form of the
parent adduct, possibly via ring-opening and closure reactions at
C-6 of the saturated pyrimidine ring component of the adduct.
Irradiation of each of these three adducts with UVB light
produces a pair of Dewar-type adducts. The nine products were
individually characterized by mass spectrometry, proton NMR
spectroscopy and UV spectroscopy. A less comprehensive study
showed that irradiation of thymidine in frozen aqueous solution
produces a diastereomeric pair of (5-4) adducts, along with the
previously known diastereomeric pair of (6-4) adducts.
The (6-4) adducts, particularly in the context of dinucleo-
tides or DNA, have been found to undergo a UVB light-
induced secondary photoreaction to form the so-called Dewar
photoproducts. A generalized representation of the Dewar
valence isomer containing two Thy moieties is given by IV; the
first complete characterization of this type of adduct was given
by Taylor et al. (5,6), who identified it as the product arising
from UVB irradiation of the (6-4) adduct of thymidylyl-(3¢-5¢)-
thymidine (TpT). At about the same time, Nishio et al. (7)
prepared Dewar-type compounds via irradiation of 1,4,6-
trisubstituted pyrimidin-2-ones; for example, l,4,6-dimethyl-
pyrimidin-2-one, which can be regarded as being analogous to
the 2¢-pyrimidone moiety in IIIa (but with a methyl group
replacing hydrogen at N1 and a methyl group at C6 rather
than C5), yielded the corresponding Dewar-type product 3,4,6-
dimethyl-2-oxo-1,3-diazabicyclo[2.2.0]hex-5-ene. In a number
of cases, similar reactions occur with derivatized forms of the
nucleobase; these have often been used as models for reactivity
of the nucleobase itself (e.g. 1-methylthymine [MeT] used as a
model for Thy).
The literature for these various types of photoreactions up
to about 1975 is reviewed in (8) (CBDs), (9) (hydrates) and (10)
((6-4) adducts). For general reviews of the photochemistry of
pyrimidine bases and related compounds, including later work
on each of the types of reaction mentioned above, see (11–13);
more recent reviews, with a stronger focus on the occurrence of
these products in a nucleic acid context, are given in (14,15).
In addition to the types of products mentioned above, other
products have been reported whose occurrence, thus far, has
been limited to particular nucleobases and closely related
compounds. When thawed solutions of Cyt, cytidine and
2¢-deoxycytidine are examined, after irradiation with 254 nm
light in the frozen state at dry ice temperature ()78.5ꢀC), it is
found that compounds with the general type of structure given
by V are formed (16,17). The Ura analog of this type of
compound has been synthesized via a protocol involving
photoreaction between 4-thiouracil and Ura (18) and recently
INTRODUCTION
The photochemistry of thymine (Thy) and related compounds
has received considerable study over the past 50 years. Indeed,
the discovery of the so-called ‘‘ice dimer’’ of Thy (I) (see
Scheme 1) in 1960 (1,2), formed in frozen aqueous solutions of
Thy irradiated with 254 nm light, can be viewed as a seminal
event in the history of the photochemistry and photobiology of
nucleic acids. Since then, it has been found that pyrimidine
nucleobases occurring in DNA and RNA, as well as their
nucleosides and nucleotides, undergo a number of types of
photochemical reaction; many of the reactions also occur in
the context of dinucleotides, polynucleotides, DNA and RNA.
In addition to cyclobutane dimer (CBD) formation, light-
induced hydration and formation of so-called (6-4) adducts are
among the most commonly encountered reactions. For exam-
ple, uracil (Ura), cytosine (Cyt) and Thy each undergo a
photohydration reaction in which water adds across the
5,6-double bond; the structure of the Ura photohydrate,
6-hydroxy-5,6-dihydrouracil, is shown as II (3). As an example
*Corresponding author email: shetlar@cgl.ucsf.edu (Martin D. Shetlar)
§Current address: Janet Chung, City of Hope, Molecular and Cell Biology,
Beckman Research Institute, Duarte, CA, USA
ꢁ 2011 The Authors
Photochemistry and Photobiology ꢁ 2011 The American Society of Photobiology 0031-8655/11
802

Photochemistry and Photobiology, 2011, 87 803
O
is evidence that such products involving Thy can be formed,
although not with both reaction partners being major DNA or
RNA nucleobases, nucleosides or closely related compounds.
After irradiation of an aqueous mixture of Thy and 4-
thiouridine at 335 nm, a diastereomeric pair of stable (5-4)
adducts was isolated and characterized by Blazek et al. (25).
These products were indicated to be secondary thermal
reaction products, arising via replacement of a 6SH group in
the originally formed (5-4) adducts by a 6OH. The two
products were interconvertible upon heating, suggesting that
the two forms differed in having the OH and 2¢-pyrimidone
moieties cis and trans with respect to one another. More
recently, Clivio et al. (26) found that irradiation of a peptide
nucleic acid (PNA) species, containing a Thy appropriately
separated from a 4-thiothymine moiety, resulted in the
formation of a Thy(5-4)Thy adduct within the context of the
PNA. Again this product appears to be a secondary thermal
reaction product, arising from replacement of an SH moiety by
an OH group. Interestingly, when 5-chloro-4-thiouridine and
5-fluoro-4-thiouridine react photochemically with Thd in
water, a more complex tricyclic ring system was formed,
rather than (5-4) adducts (27).
O
O
O
MeMe
OH
Me
Me
H
N
H
N
H
H
N
N
H
N
H
O
O
OH
O
O
N
H
N
H
N
H
N
H
H
N
II
H
IIIa
I
O
O
H
H
N
OH
Me
H
H
R
N
H
H
NH2
R
N
N
NH2
Me
O
O
N
N
R
N
N
H
H
O
N
NH2
O
H
N
O
N
R
N
R
V
IV
VI
R
O
H
N
O
O
OH
Me
O
H
O
N
H
O
CH2
Me
H
N
H
Me Me
N
OH
H
H
N
N
N
N
H
O
O
H
O
O
N
H
N
H
VIII
VII
Scheme 1.
In this paper, we present evidence that MeT forms both a
(6-4) adduct (IIIb) and a (5-4) adduct when irradiated in frozen
aqueous solution (see Scheme 2). We also found that this
initially observable (5-4) photoproduct isomerizes upon stand-
ing to form a second (5-4) adduct. Evidence to be presented
later suggests that the initially isolatable adduct can be
identified as Xb, in which the 2¢-pyrimidone ring and the
hydroxyl group are trans with respect to one another.
Corresponding evidence will be provided that suggests that
the thermal isomerization product is IXb, in which the
hydroxyl group and the 2¢-pyrimidone ring are cis with respect
to one another. We also demonstrate that each of these three
adducts, when irradiated by UVB light, is transformed into a
mixture of two diastereomeric Dewar valence isomers. Finally,
we present evidence that irradiation of thymidine under the
same conditions produces two (5-4) type adducts, as well as the
previously characterized (6-4) adducts (28–30). (The number-
ing system is displayed in the structure for IIIb; asterisks in the
has been shown to be present in reaction mixtures resulting
from irradiation of Ura at 254 nm in frozen aqueous solution
on dry ice (19). It appears likely that these compounds result
from an unstable precursor via loss of the elements of
ammonia or water. In the case of products with structure V,
one possible precursor (a (5-4) adduct) would have the general
structure given by VI, although alternative structures for this
intermediate have also been proposed ([10], p. 336).
The so-called ‘‘spore photoproduct’’ (5,6-dihydro-5-(a-
thyminyl)thymine, VII) is another type of product formed
under a variety of reaction conditions. For example, irradia-
tion of Thy in thin films produces this compound, while
irradiation of thymidine (Thd) in thin films and in frozen
aqueous solution forms the thymidine analog of VII. Irradi-
ation of highly dehydrated DNA (e.g. that contained in the
spores of microorganisms) likewise produces this molecule;
this compound is a predominant source of UV damage in such
systems, and repair systems to remove this damage have been
discovered. (For a review of the early literature, see [11]; more
recent photochemical work is discussed in [20], while a
comprehensive recent review is given in [21].)
O
R
N
R
N
O
O
OH
Me
H
4
H
N3
*
6
5
O
O
N
Me
N
2
1
*
N
R
Another interesting type of product found when solutions
of Thy are irradiated in aqueous solution at )78.5ꢀC is the Thy
trimer hydrate (VIII), first reported by Wang (22); the three-
dimensional structure of VIII is described in Karle (23), based
on results from X-ray diffraction studies. This compound is
formed as a secondary photoreaction product, possibly via
reaction of the putative oxetane precursor of III with Thy (24).
Recently, mass spectral and chemical evidence has been
presented indicating that formation of the Ura analog of VIII
occurs when Ura is similarly irradiated (19).
As the above discussion indicates, a considerable amount is
known about the photochemistry of Thy, both as the
nucleobase and when incorporated into nucleosides, oligonu-
cleotides and nucleic acids. However, there are some questions
that are yet unanswered. One is ‘‘Can Thy or closely related
compounds be photochemically induced to form stable (5-4)
adducts, analogous in structure to those shown in VI?’’ There
Me
O
Me
4' 5'
H
H
*
N
N
*
H
Me
Me
H
OH
6'
N 3'
OH
1'
2'
N
R
*
N
R
N
R
*
H
O
O
O
Xa: R = H
Xb: R = Me
Xc: R = dRib
R
IIIa: R = H
IXa: R = H
IIIb: R = Me
IXb: R = Me
IIIc: R = dRib
IXc: R = dRib
O
R
O
O
N
N
OH
H
*
N
*
Me
*
O
*
N
O
*
N
Me
H
Me
*
Me
H
N
N
R
*
H
O
*
N
Me
Me
H
OH
*
H
*
N
OH
N
N
R
*
H
O
N
R
*
R
O
O
IVa: R = H
XIIa: R = H
XIIb: R = Me
XIIc: R = dRib
XIa: R = H
XIb: R = Me
XIc: R = dRib
IVb: R = Me
IVc: R = dRib
Scheme 2.

804 Martin D. Shetlar and Janet Chung
freezing process. Four lamps, contained in two Spectroline X-Series
lamp housings, were placed across the top of the pan containing the
solutions; irradiation was done with the pan resting on top of the dry
ice slabs. The fluence rate at the surface of the frozen aqueous layer in
the pan averaged ca 25 J m)2 s)1, as measured using a Spectronics
DM-254N Ultraviolet Meter (Westbury, NY).
various structures in Scheme 2, including those at bridgehead
nitrogens, denote chiral centers.)
It should be noted that IIIb, IXb and Xb each exists in two
enantiomeric forms. The structures shown in Scheme 2 are for
the 5S,6R (IIIb), 5S,6R (IXb) and 5S,6S (Xb) forms. (The other
(6-4) adduct, in which the 2¢-pyrimidone ring and the OH
group are trans with respect to one another, has not been
observed for Thy and closely related compounds. However,
the analogous adduct has been observed when Ura is irradi-
ated in the frozen state [19].) Structure IVb corresponds to one
of two possible Dewar valence isomers with the same 5S,6R
configuration as IIIb; the one displayed as IVb has a 5S,6R,6¢R
configuration, while the second would have a 5S,6R,6¢S
stereoconfiguration. There is a corresponding pair of Dewar
adduct enantiomers for the the 5R,6S enantiomer of IIIb,
namely 5R,6S,6¢R and 5R,6S,6¢S. As the 5S,6R,6¢R configu-
ration is the mirror image of the 5R,6S,6¢S configuration and,
similarly, the 5S,6R,6¢R configuration is the mirror image of
the 5R,6S,6¢S configuration (corresponding to the generalized
structure of IVb, in which stereochemistry is not considered),
there are two diastereomers of IVb, each having two enantio-
meric forms. Performance of similar analyses for the general-
ized structures corresponding to XIb and XIIb, the (5-4) Dewar
valence isomers, indicates that the situation is the same for
these compounds.
Acetone-photosensitized reactions of MeT in aqueous solution
containing 20% vol ⁄ vol acetone were done with 2 mM MeT. About
170 mL of the solution was placed in a 180 mL quartz tube with a
24 ⁄ 40 outer taper seal joint (Southern New England Ultraviolet Co.,
Hampton, CT). This joint was stoppered with a Pyrex adaptor that
converted a 24 ⁄ 40 inner taper seal joint to 19 ⁄ 38 outer taper seal; an
appropriately sized rubber septum was fitted over the 19 ⁄ 38 joint and
an 8 inch long needle was pushed through the septum. To the end of
this needle was attached a length of Teflon tubing that extended to
near the bottom of the irradiation vessel; an additional bleed needle
was also inserted through the septum. The tube was positioned in the
space between two Pyrex filtered 15 Watt Spectronics UVB (BLE-
1T158) lamps housed in a Spectroline XX-15A lamp holder. The lamp
holder was placed in an upright position, setting at ca 15ꢀ angle to a
supporting wall. The solution was bubbled with nitrogen (99.997%
purity; Puritan Medical Supply, Overland Park, KS) for 20 min and
then irradiated for 120 min under bubbling N2, passed through a glass
frit into an intermediate vessel partially filled with 80% ⁄ 20% vol ⁄ vol
water ⁄ acetone.
Exploratory studies of the photochemistry of MeT. Prior to
preparative studies, preliminary studies were conducted to determine
appropriate concentrations of MeT for use in larger scale studies. The
results of these studies, described in Appendix S1, indicate that
solutions 0.2 mM in MeT gave the best results of the concentrations
examined and this concentration was used in the preparation of parent
solutions for irradiation.
HPLC analysis of an irradiated MeT system: an example. Figure 1
displays the HPLC chromatogram obtained via injection of 100 lL of
250-fold concentrate of a 0.2 mM solution that had been irradiated for
32 min on dry ice as described in Materials and Methods. It was run
on Column A using the following water ⁄ methanol gradient (Gradient
A): 0 min, 10% methanol (MeOH); 3.5 min, 10%; 9 min, 20%;
13 min, 20%; 15 min, 10%; 20 min, 10%, flow rate: 5 mL min)1. The
percent conversion of parent MeT to products was estimated by HPLC
to be 62.3%. From spectra obtained using the ‘‘on the fly’’ spectral
capture capability of the diode array detector, it could be concluded
that Peaks 2, 7, 8 and 10 probably correspond to the four CBDs (31).
Similarly, Peak 3, the very small Peak 5 and Peak 6 displayed spectra
similar in profile to the spectrum of the Thy (6-4) adduct (4). The peak
labeled 9 has a spectrum that is ‘‘double humped’’ in nature with kmax
at 275 and 309 nm. The two peaks labeled 11a and 11b both had kmax
values at 275 nm. By comparison of the integrated peak areas for
Peaks 3 and 6 at 315 nm (assuming equal e315 values), it can be
estimated that ca 2.45 times as much product elutes in Peak 3 as
compared to Peak 6. A very rough estimate of the fraction of total
product, existing as the compounds eluting in Peaks 3 and 6, can be
made by dividing the sum of the peak areas corresponding to these
substances, measured at 230 nm, by the total peak area of all products.
This fraction has a value of 0.13.
‘‘On the fly spectra’’ of Peak 1a–f indicate that many of the
corresponding compounds have maxima between 270 and 280 nm. It
should be emphasized that the absorbance scale for the HPLC trace in
Panel B in Fig. 1 (detection at 271 nm) has a much smaller range than
that given in Panel A (detection at 225 nm). Thus, the actual sizes of
Peaks 1a–f and 11a,b as manifested by peak area or height, are much
smaller than those of the other numbered product peaks shown in
Panel A.
Preparative scale irradiation and isolation of products in the MeT
system for structural characterization. For preparative runs, we
generally irradiated 1000 or 2000 mL of solution in 500 mL batches.
For these runs, the parent solution was irradiated for 64 min,
thawed, refrozen and reirradiated for an additional 64 min. When
this solution was analyzed by HPLC, the percent conversion of
parent MeT to products was 66.9%. This irradiation protocol
increased the yield of the material eluting in Peak 3 by 3.5-fold and
the yield of material in Peak 6 by 1.8-fold (based on area peaks
measured at 315 nm) as compared to the systems irradiated for
32 min.
MATERIALS AND METHODS
General aspects. MeT, Thd and Thy were obtained from Sigma (St.
Louis, MO). HPLC solvents were from Fisher (Fair Lawn, NJ), while
NMR solvents were provided by Aldrich (Milwaukee, WI). Prepara-
tive separations were accomplished on a Shiseido Capcell UG120
10 · 250 mm column (5 mm particle size; Yokohama, Japan) (Column
A), while analytical HPLC separations were usually carried out on a
Capcell UG120 4.6 · 150 mm column (5 mm particle size) (Column
B). Other columns used were Column C, a Microsorb Phenyl column
(4.6 · 250 mm, 5 mm particle size; Varian, Walnut Creek, CA), a
Microsorb Amino column (4.6 · 250 mm, 5lm particle size) (Column
D) and a Capcell Phenyl column (4.6 · 150 mm, 5 lm particle size)
(Column E). The HPLC system used was a Rainin binary gradient
pumping system (Emeryville, CA) coupled to a Hewlett-Packard
1040A diode array detector (Palo Alto, CA). Before injection, all
HPLC samples were subjected to spin filtration on Costar Spin-X
microcentrifuge tubes containing a 0.2 lm nylon filter (Corning
Incorporated, Corning, NY). Rotatory evaporations were accom-
plished on a Buchi R-200 Rotovapor (New Castle, DE) coupled to a
¨
mechanical pump; two dry ice traps were placed in series between
sample and pump.
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).
Unfiltered Spectronics BLE IT155 lamps, with output mainly at
254 nm and housed in Spectroline XX-15A lamp housings (Westbury,
NY), were used for irradiations of solutions in the ice phase;
Spectronics BLE IT158 lamps, with output centered at 312 nm and
contained in the same housings, were used to photosensitize reactions
in solutions containing acetone or to convert (6-4) and (5-4) adducts to
Dewar valence isomers.
Irradiation methods. Irradiations of MeT in the frozen state at
254 nm were done on either 250 or 500 mL batches of aqueous 0.2 mM
solution, while those for Thd were similarly done with 2 mM Thd. Each
batch was placed in a 13 inch · 9 inch nonstick baking pan (Bradshaw
International, Rancho Cucamonga, CA) and frozen on two
25 cm · 25 cm dry ice slabs ()78.5ꢀC) placed side by side. During
freezing the pan was covered with a 15 inch · 10 inch Pyrex baking
dish; this minimizes frosting of the surface of the solution during the

Photochemistry and Photobiology, 2011, 87 805
Figure 1. The HPLC chromatogram resulting from injection of 100 lL of a 250-fold concentrate, obtained from a solution of 0.2 mM MeT that
had been irradiated for 32 min on dry ice as described in Materials and Methods. Panel A shows the chromatogram seen with detection at 230 nm,
panel B provides a similar trace obtained with detection at 271 nm and panel C displays the trace obtained using 315 nm as the detection
wavelength. The chromatogram was obtained as described in Materials and Methods. Detailed identification of the various peaks is discussed in
Materials and Methods; however, the peaks labeled 3 and 6 in Panel C correspond to compounds with UV absorption properties similar to those
expected for (6-4) or (5-4) adducts, while the small broad peak labeled 5 shows similar UV absorbance at longer wavelengths overlapped by
absorbance from the tail of the MeT peak. MeT, 1-methylthymine.
For product isolation, a slightly different gradient (Gradient C) was
developed that gave almost the same kind of chromatographic profile
(with different retention times and peak heights), but yet provided
enhanced resolution of Peaks 9 and 9a (see Fig. 1). This gradient was the
following: 0 min, 8% MeOH; 12 min, 8%; 18 min, 21%; 21 min, 26%;
those amounts contained in other fractions, we did not pursue
identification of the various compounds in these complex reaction
mixtures.
We found that the composition of MeT photoreaction mixtures
depended on the time between irradiation and workup. Analysis of two
irradiated samples that had stood, one for 7 days at 4ꢀC and the other
for 7 days at room temperature, indicated that the peak area (315 nm)
for P3, while decreasing ca 2% over a 7 day period at 4ꢀC, diminished
by 48% over the same time period at room temperature, while the
corresponding (very small) area for P5 increased noticeably at 4ꢀC and
much more at room temperature. These results suggest that P5 is a
secondary product, possibly resulting from thermal decomposition of
P3. The other compounds in the reaction mixture, including the
material in Peak 6, appeared to be quite stable.
24 min, 26%; 24.5 min, 8%; 32 min, 8%; flow rate: 4 mL min)1
.
During the course of a representative preparative HPLC purification, a
total of 3.3 mL of concentrate was injected in four batches; this
concentrate was obtained by rotatory evaporation of 1000 mL of
irradiated solution. The following fractions were collected: F1 (3–
14.5 min), F2 (14.5–17.4 min), F3 (17.4–18.4 min), F4 (18.4–20.2 min),
F5 (20.2–21.2 min), F6 (21.2–21.9 min), F7 (21.9–22.9 min), F8
(22.9–23.9 min), F9 (23.9–24.9 min), F10 (24.9–25.4 min) and F11
(25.4–29.5 min).
Irradiation of thymidine in the frozen state on dry ice yields products
that have UV absorption characteristic of (5-4) and (6-4) adducts. As
will be discussed in detail in Results and Discussion, early studies by
Varghese (28,32) of the photochemistry of Thd in ice and in thin films
indicated that two compounds with UV spectra similar to those
expected for (6-4) and ⁄ or (5-4) adducts. Hydrolysis of one of these
products in 0.5 M HCl produced XIII, the dehydration product of IIIa
(Scheme 3). This indicted that at least one (6-4) adduct had been
produced. The other product was not definitively identified. In
The various fractions were examined by analytical HPLC on
Column B. Using 88 ⁄ 12 H2O ⁄ MeOH flowing at 2 mL min)1, it was
determined that the main products found in F2 (P(roduct) 2, elution
time [te]: 1.6 min), F7 (P7) (te: 2.7 min), F8 (P8) (te: 3.2 min) and F10
(P10) (te: 4.1 min) were putatively CBDs, based on their absorption
spectra; F4 contained parent MeT. The chromatogram of F3 showed
that it contained a dominant product (P3) (te: 1.9 min) that had a (6-4)
adduct-like absorption spectrum; a similar product (P6) was found in
F6 (te: 2.4 min). It was found that F5 contained small amounts of a
distinct similar type of product P5 (te: 2.3 min), along with a small
amount of an impurity (te: 2.6 min) whose spectrum did not display an
absorption maximum above 220 nm. Fraction F9 contained the
product with a ‘‘double-humped’’ absorption spectrum (P9). (For
convenience’s sake, a summary of the correspondence between Peak
number in Fig. 1, fraction number, compound number [as defined
above] and assigned structures for the compounds eluting in the
various HPLC peaks is given for the MeT system in Table S1. This
table also contains analogous information for the thymidine system,
discussed below.) Both F1 and F11 were quite complex mixtures. The
HPLC chromatogram obtained from analysis of F11, using the
conditions described above, showed that it contained a large number
of products (at least eight), with five of them having kmax in the
neighborhood of 275 nm; similarly, the chromatogram from HPLC
O
CH3
H
N
O
N
H
CH3
H
N
XIII
O
N
H
analysis of F1, using 96 ⁄ 4 H2O ⁄ MeOH flowing at 2 mL min)1
,
showed at least 12 peaks. Since, as pointed out previously, the amounts
of the compounds contained in F1 and F11 are small, as compared to
Scheme 3.

806 Martin D. Shetlar and Janet Chung
unpublished work by Shaw and Cadet (summarized in [11], p. 80), it
was indicated that two diastereomers of the Thd (6-4) adduct had been
structurally characterized and that two related decomposition products
were isolated as well. Douki et al. (30) also described the production of
the two Thd (6-4) adducts; using these adducts as photosubstrates,
they indicated that they had also prepared the corresponding Dewar
valence isomers of these two (6-4) adducts. Interestingly, it has been
found that irradiation of thymidylyl-(3¢-5¢)-2¢-deoxyadenosine in the
frozen state results in the formation of an intermolecular (6-4) adduct
linkage involving two Thds (33).
(2 mM) at 254 nm for 80 min in the frozen state (in 250 mL batches),
we concentrated the resulting solution to 2 mL via rotatory evapo-
ration at 40ꢀC. After spin filtration (slow, because of the high viscosity
of the concentrate), ca 1.65 mL of the concentrate was recovered. This
material was separated on Column A, using the following gradient:
0 min, 6% MeOH; 5 min, 6%; 11 min, 20%; 15 min, 20%; 15 min
10 s, 6%; 20 min, 6%; flow rate: 5 mL min)1. The concentrate was
injected in seven batches and the desired product peaks were detected
at 320 nm; only those peaks displaying strong absorbance at this
wavelength were collected. The following five fractions were obtained:
F3¢, 9.2–9.8 min; F5¢, 10.2–11.1 min; F7¢, 14.3–15.2 min; F10¢, 16.4–
17.1 min and F11¢ (17.3–18.1 min). After further HPLC work, F3¢,
F5¢, F7¢ and F10¢ each yielded distinct pure compounds (P3¢, P5¢, P7¢
and P10¢) with absorption spectra similar to those expected for (6-4)
or (5-4) adducts, while F11¢ gave a compound with a ‘‘double-
humped’’ spectrum. The purification protocols used are described in
Appendix S1. We present evidence in Results and Discussion that P3¢
and P5¢ are Thd (6-4) adducts and that P7¢ and P10¢ are (5-4) adducts
of Thd.
Analytical HPLC of Thd photoreaction mixtures. To determine if
compounds that could potentially be identified as (5-4) adducts are
produced in UV-irradiated thymidine, in analogy with the MeT system,
we irradiated Thd (2 mM) in four 250 mL batches, each for 80 min in the
frozen state, with 254 nm light as described above for MeT. The
percentage conversion of Thd to products was determined by analytical
HPLC to be ca 39%. The irradiated solution was analyzed on Column
B, using the following gradient: 0 min, 6% MeOH; 6 min, 6%; 10 min,
16%; 13.5 min, 16%; 14 min, 6%; 18 min, 6%; flow rate: 3 mL min)1
.
The HPLC resulting from injection of 100 lL of freshly irradiated
solution is shown in Fig. 2. The materials in Peaks 1 and 2 both have UV
spectra with a maximum at around 265 nm, as measured via the spectral
capture feature of the diode array HPLC detector. The compounds
eluting in Peaks 3, 5, 7, 9 and 10 displayed spectra that had absorbance
at wavelengths between 300 and 335 nm, as would be expected of (6-4)
or (5-4) adducts. However, the material eluting in Peak 10 showed two
maxima, one at 271 nm and the other at 319 nm, suggesting that this
peak corresponds to a mixture of one of the spore products and (6-4) (or
(5-4)) adduct. The minor product eluting in Peak 9 may be a (6-4) or
(5-4) adduct decomposition product. The products eluting in Peaks 4, 6
and 8 had spectra characteristic of CBD, while the material eluting in
Peak 11 consisted mainly of a product with a spectral maximum at
271 nm (probably being one of the spore adducts [32]); however, there
was also a substance eluting in this peak that displayed a small amount
of absorbance between 300 and 335 nm. For the purposes of the work
reported here, we concentrated on the materials eluting in Peaks 3, 5, 7
and 10, although in the course of our work a number of other products
were identified as compounds already reported in the literature. The
relative percentage yields for the compounds eluting in these four peaks
can be estimated from the corresponding integrated peak areas,
measured at 320 nm. The values obtained are as follows: Peak 3:
17.2%; Peak 5: 31.2%; Peak 7; 21.6%; Peak 10; 30%.
RESULTS AND DISCUSSION
In the initial part of the presentation that follows, we discuss
the identification of the major products found after the
irradiation of MeT in frozen aqueous solution at –78.5ꢀC.
These photoproducts include CBDs, as well as a (6-4) adduct
and a (5-4) adduct, both formed as initially observable
photoproducts. We also discuss a thermal isomerization
reaction that converts the initially observed (5-4) adduct to a
secondary (5-4) product. We will then discuss the UVB light-
induced conversion of the MeT (6-4) adduct and the two MeT
(5-4) adducts to corresponding Dewar valence isomers. Finally,
we will discuss the identification of (5-4) adducts of Thd.
Isolation and characterization of the initially observable (6-4)
and (5-4) photoadducts and the secondary thermal (5-4)
adduct of MeT
Preparative HPLC of Thd photoreaction mixtures. We used the
following conditions for a first pass HPLC separation; these condi-
tions allowed us to focus on those peaks containing absorbance that
might correspond to (5-4) adducts. After irradiation of 1000 mL of dT
In the following, we will focus on the protocols used
for purification and characterization of the products
Figure 2. HPLC chromatograms, as measured at three different wavelengths, resulting from injection of 100 lL of freshly irradiated Thd.
Irradiation and HPLC conditions, along with a commentary on peak identification, is given in Materials and Methods.

Photochemistry and Photobiology, 2011, 87 807
corresponding to the peaks labeled 3 and 6 in Fig. 1. As in
Materials and Methods, the material eluting in these two
peaks is termed P(roduct) 3 and P6 respectively in the
following development.
Appendix S1. The UV spectrum of purified P5 showed
kmax = 314 nm and kmin at 257 nm. The ESI mass spectrum
displayed peaks with molecular masses of 281.2 [M + H]+,
303.1 [M + Na]+ and 319.1 [M + K]+, consistent with this
compound being either IXb or Xb. The proton NMR spectral
data, given in Table 1, are also consistent with this assign-
ment. Each of the expected resonances is present at appro-
priate chemical shifts and intensities. Of particular interest is
the observation that when the spectrum of P5 is run in
DMSO-d6, C6H and the C6OH are coupled to one another.
Addition of D2O results in disappearance of the OH
resonance and collapse of C6H to a singlet. The data
gathered for P5 are thus consistent with this compound
having a structure given as either IXb or Xb in Scheme 2. It
can be noted that the chemical shift values for the resonances
of the C6H in P3 (4.61) and P5 (5.30) differ by ca 0.7 ppm;
the J values describing the interaction of C6H with C6OH
also differ for these two compounds (P3: J = 4.7) and P5
(J = 3.0).
Purification and characterization of P3 and P6. In general,
fractions F3 and F6, isolated from preparative separations of
concentrated irradiated MeT solutions as described in Mate-
rials and Methods, contained essentially pure P3 and P6
respectively and did not require further purification.
The UV spectrum of P3 had a maximum at 316 nm and a
minimum at 253 nm, while the ESI mass spectrum displayed
peaks at 281.2 [M + H]+, 303.1 [M + Na]+ and 319.1
[M + K]+; these values correspond to those expected for
compounds with a molecular formula of C12H16N4O4 (molec-
ular mass: 280). The proton NMR data (Table 1) are
consistent with P3 being a (5-4) adduct with either the
structure IXb or Xb (both shown in Scheme 2). In general,
all expected resonances are present with the expected chemical
shifts. The key observation, however, is the presence of a C6
proton (d = 4.61) coupled to the OH proton when the
spectrum is run in DMSO-d6. This can only occur if the OH
is located at the 6-position as well. Addition of D2O results in
the C6 resonance collapsing to a singlet, as would be expected
after hydrogen–deuterium exchange by a C6OH proton. This
rules out the alternative possibility that P3 is a (6-4) adduct.
Compound P6 displays a UV absorbance maximum at
318 nm and a minimum at 261 nm. The ESI mass spectrum
shows peaks at 281.2 [M + H]+, 303.1 [M + Na]+ and 319.1
[M + K]+, with the parent peak dominating the spectrum;
these values are consistent with identification of P6 as IIIb. The
proton NMR data, given in Table 1, also support this
identification. The expected resonances for a compound with
structure IIIb are present, except for one methyl group (which
is likely buried under the water peak). In contrast to P3, the
OH group, observed in DMSO-d6, is not coupled to another
proton.
Tentative structural assignments for the stereochemistry of P3
and P5
We now consider the problem of whether P3 should be
assigned to have the structure IXb and P5 the structure Xb (or
vice versa). This question cannot be answered with certainty.
While a (6-4) adduct with the trans configuration has been seen
as a minor photoproduct when Ura is irradiated in the frozen
state (19) (the cis form being favored in that case), there are
not any precedents for stable (5-4) adducts of nucleobases and
related compounds. It is generally believed that (6-4) adducts
(e.g. IIIb) are formed via reaction of a compound containing
an initially formed oxetane ring. Thermal ring opening of the
oxetane, perhaps mediated by water molecules within the ice
matrix, presumably leads to the observed adduct(s). In the case
of Thy, the oxetane ring system appears to be unstable at
temperatures above )80ꢀC (24); at temperatures above this
value, the oxetane intermediate evidently reacts in the ice
matrix, possibly mediated via relatively mobile water molecules
within the matrix, to form (6-4) adducts. (For a discussion of
these matters, see [11], pp. 80–86 and the presentation in [19],
pp. 94–95.) A reaction scheme showing a proposed mechanistic
pathway leading from oxetane adduct formation to final
Characterization of P5, the (5-4) adduct resulting from thermal
isomerization of P3. The thermal decomposition of P3 upon
standing, to form a second adduct (P5) with the same UV
spectral profile, was described in Materials and Methods. We
describe its preparation in larger amounts and its isolation in
Table 1. Proton NMR data for the (6-4) and (5-4) adducts of MeT and the Dewar valence isomers of these adducts.
Adduct
C5CH3
C5¢CH3
N1CH3
N1¢CH3
C6H
C6¢H
OH
N3H
P3 (Xb)
1.49
1.36
1.35
1.37
1.29
1.29
1.52
1.39
1.40
1.93
1.75
1.73
2.00
1.69
1.69
2.08
1.82
1.78
2.92
2.89
2.86
2.83
2.81
2.86
2.72
2.84
2.85
3.39
2.91
2.91
3.36
2.85
2.88
*
4.61(J = 4.7)
4.58 (J = 4.9)
4.54 (J = 4.8)
5.30 (J = 3.0)
4.83 (J = 5.4)
4.74 (J = 4.5)
4.57
7.89
5.03
5.03
8.00
5.06
5.09
7.93
5.02
5.06
6.74 (J = 4.8)
6.66 (J = 4.9)
6.80 (J = 4.8)
6.81 (J = 3.0)
6.86 (J = 5.4)
6.84 (J = 5.0)
5.81
10.67
10.41
10.42
10.38
10.63
10.57
10.23
10.20
10.36
P3D1 (XIIb-1)
P3D2 (XIIb-2)
P5 (IXb)
P5D1 (XIb-1)
P5D2 (XIb-2)
P6 (IIIb)
P6D1 (IVb-1)
P6D2 (IVb-2)
2.84
2.88
4.16
4.10
5.73
5.64
*This resonance was unobservable, as the corresponding peak was obscured by the water peak.
Proton NMR spectra were run at 600 MHz on a Varian INOVA NMR spectrometer in DMSO-d6; TMS was used as an internal standard. Each of
the peaks gave an appropriate integration area for the number of protons expected. Coupling constants in Hz are given in parentheses after the
chemical shifts of the corresponding proton and are in ppm.

808 Martin D. Shetlar and Janet Chung
products for the reaction leading to (6-4) adducts in the Ura
system is presented in fig. 5 in (19).
relationship to one another. A similar attack on carbon B
(labeled 6 in Fig. 4), generating Int-2b, would yield an end
product in which the hydroxyl group is trans to the 2¢-
pyrimidone ring; the resultant compound would correspond to
Xb. Thus, pathways can be envisioned leading from the
oxetane displayed in Fig. 4 to both IXb and Xb via an SN2
mechanism. Inspection of an analogous structure for the
diastereomer 5R,6R,4¢R leads to a similar conclusion.
Which of these two pathways (nucleophilic attack at C4¢ or
C6) would be expected to dominate if this mechanism is
operative? Figure S1 shows colored representations of two
other highly informative views of the 5R,6R,4¢S oxetane
isomer with all hydrogens present. Examination of this figure,
looking at the relative degree of stereochemical blockage of the
two carbons, indicates that C6 is significantly more accessible
to backside attack by an incoming water than C4¢. Since attack
at C6 leads to the formation of Xb, rather than IXb, this
suggests that, if this mechanism is operative, then P3 (the final
photoproduct) corresponds to Xb and P5 (the secondary
thermal reaction product) to IXb. Analogous visualizations of
the 5R,6R,4¢R diastereomer, also displayed in Fig. S1, allow
the same conclusion to be drawn. These conclusions are in
agreement with the known properties of SN2 reactions, for
which attack of a nucleophile at a tertiary carbon (e.g. C4¢) is
generally accepted to be much slower than attack at secondary
carbons (e.g. C6).
The other possible mechanism to be considered is one in
which the oxetane reacts via carbocation intermediates as
shown in Fig. 5. In this case, rather than backside attack of
water (or other nucleophile) at C4¢ and C6, ring opening
occurs via generation of carbocations at these sites; in the
mechanism shown, this process involves putatively relatively
mobile water molecules in the ice matrix. In the top panel of
Fig. 5 Int-1, the intermediate oxetane undergoes a ring-
opening process that generates a tertiary cationic site at C4¢
(Int-2c) that can lose a proton to generate the 2¢-pyrimidone
ring. The resulting IXb has the 6-hydroxy group and 2¢-
pyrimidone ring cis with respect to one another. In the bottom
panel, the ring-opening process generates a secondary carbo-
cation at C6 (Int-2d). This species would be expected to react
with mobile water both above and below the ring containing
the carbocation site to eventually form, after loss of an
Presumably (5-4) adducts are also formed via a reactive
process involving an analogous oxetane ring. (Figure 3 dis-
plays a reaction scheme indicating one possible set of reactions
leading from photoinduction of oxetane formation to produc-
tion of final (5-4) adducts.) However, based on the results
given above, it cannot be ascertained whether IXb or Xb is the
product initially resulting from the corresponding ring-open-
ing reaction of the parent oxetane. A definitive answer must
await X-ray diffraction studies of crystals of one or both of
these compounds. However, suggestive evidence points toward
Xb, the adduct with the 2¢-pyrimidone and hydroxyl groups
trans to one another, being P3; if this is true, then P5 would
correspond to IXb. This conclusion has been reached, based on
the results of quantum chemical calculations using Spartan 08
for Macintosh (Wavefunction, Inc., Irvine, CA).
Assuming that only suprafacial cycloaddition occurs in the
reaction of the C4 carbonyl group on one MeT with the C5C6
double bond of a second MeT, a putative oxetane-type
intermediate involved in the formation of (5-4) adducts can
exist in two diastereomeric forms (each form comprised of an
enantiomeric pair). A view of the structure of one of these
forms of the putative oxetane, generated by determining the
equilibrium geometry of this compound via quantum mechan-
ical calculation, is shown in Fig. 4; this particular form has a
stereochemical designation of 5R,6R,4¢S. (See the caption of
Fig. 4 for a more detailed description of how this structure was
obtained.) For the sake of visual clarity, all of the hydrogens
(except that at C6) are removed. There are at least two types of
mechanism by which intermediate oxetane can react to form
(5-4) adducts. One involves reactions that are SN2 in nature,
while the other can be regarded as an SN1 mechanism in which
oxetane ring opening occurs via carbocation intermediate(s).
In the first mechanism, backside attack (relative to the oxetane
oxygen [labeled Ox in Fig. 4] by a nucleophile [e.g. water] at
carbon A in Fig. 3 [labeled 4¢ in Fig. 4]) would break the
oxetane ring bond to the 2¢-pyrimidone ring and generate the
intermediate Int-2a (Fig. 3); loss of a proton from the added
water and protonation of the oxygen formerly associated with
the oxetane would lead to IXb (Fig. 3); in this compound, the
hydroxyl group and the 2¢-pyrimidone ring are cis in their
O
O
H
Me
O
N
O
N
NMe
NMe
H
Me
H
N
H
N
H2O
H
-
OH
Me
N
H2O
Me
OH
Me
O
O
H
OH
O
N
O
N
H
H
Me
O
Me
Me
H
Me
H
N
N
H
h
A
Me
Int-2a
IXb
H
2
N
O
O
O
N
B
O
H
N
O
N
Me
O
O
N
NMe
H
H2O
Me
NMe
Me
Me
H
H
N
H
N
H2O
-
H
Int-1
OH
Me
Me
H
H
O
N
O
N
OH
OH
Me
Me
Xb
Figure 3. Reaction pathways leading to the formation of IXb and Xb from the common oxetane intermediate Int-1 via a backside displacement
Int-2b
mechanism, as discussed in the text.

Photochemistry and Photobiology, 2011, 87 809
however, predict that different products will be produced. The
backside attack mechanism indicates that Xb would be the sole
product, while the mechanism involving a carbocation at C4¢
predicts that IXb should be the sole product.
Me
6’
Based on the experimental information currently available, a
decision cannot be made as to whether the product P3 is IXb or
Xb. However, additional information derived from quantum
mechanical calculation provides support for the interpretation
that P3 is Xb and P5 is IXb. One would expect that the observed
thermal isomerization of initially observed P3 to P5 would
occur, so as to produce the thermodynamically more stable
compound, eventually reaching an equilibrium in which the
most stable compound predominated. While DG0, the Gibbs
free energy change in a reaction, is the rigorous measure of the
direction and degree of reaction under standard state condi-
tions, calculated values of DE should provide a rough guide as
to whether IXb or Xb is the more stable thermodynamically.
Using Spartan 08 for Macintosh, we calculated the energies of
the equilibrium geometries of the most stable conformers of
IXb and Xb at 0 K in the gas phase, using the Hartree-Fock
model with the 6-31G* basis set. The result of these calculations
indicated that IXb (Ecalc = )2575714.7 kJ mol)1) is consider-
ably more stable than Xb (Ecalc = )2575705.9 kJ mol)1) in the
gas phase. Using these gas phase values, Spartan also calculated
the corresponding energies of IXb and Xb in aqueous solution
via a semiempirical calculation method, based on the SM5.4
model of Chambers et al. (34). The calculated values of E for
the two hydrated species were Ecalc = )2575763.1 kJ mol)1
(IXb) and Ecalc = ) 2575742.8 kJ mol)1 (Xb). Using the above
two sets of results, the value of DE for the reaction Xb fi IXb
in the gas phase can be evaluated as )8.8 kJ, while in aqueous
solution the corresponding energy is DE = )20.3 kJ at 0 K. At
this level of sophistication, the results indicate that IXb, the cis
isomer, is the more energetically stable of the two, both in the
gas phase and in aqueous solution. These results are consistent
with the assignment of P3 as Xb and the assignment of its more
stable thermal isomerization product P5 as IXb. It can thus be
suggested that thermal isomerization of Xb to IXb in aqueous
solution is a thermodynamically favorable process and that a
solution containing these two compounds will reach a compo-
sition in which putative IXb is the predominant component.
As pointed out above, some observations of the decompo-
sition of P3 and formation of P5 upon standing of freshly
irradiated solutions of MeT at 4ꢀC and at room temperature
suggested that P3 reacts to form P5. To gain a more
quantitative feeling for the composition of a solution contain-
ing P3 and P5 after it reaches an equilibrium composition, we
took portions of a solution of P3 that, after purification, had
stood for ca 3.5 years in the frozen aqueous state at )20ꢀC.
Application of HPLC to this solution indicated that 89% of
the total material in this solution was P3 and 11% P5 (based on
the assumption that the emax values are the same); this
corresponds to a peak area ratio of R = [P5] ⁄ [P3] of 0.12.
We heated this solution at 100ꢀC for varying lengths of time,
quenched the reaction in ice and redetermined R. It was found
that R increased with time of heating; however, there was no
change in R between 30 and 60 min of heating. We evaluated
the final value of R ( = Keq) to be ca 5.1. This corresponds to a
value of DG0 = )5.05 kJ for the reaction P3 fi P5 at 100ꢀC.
A similar set of experiments was carried out on a freshly
irradiated reaction mixture, resulting from irradiation of
1’
Me
Ox
5’
3’
Me
4
4’
5
6
Me
1
Figure 4. A view of a ball-and-stick model of the 5R,6R,4¢S isomer of
the putative oxetane precursor of the (5-4) adduct of MeT. All
hydrogens, except that located at C6, are hidden. Oxygen atoms are
denoted by black balls, nitrogens by light gray balls and carbon atoms
are coded by medium gray. The oxygen of primary interest is labeled
Ox. Backside attack of water at C4¢ in line with the C4¢-Ox bond would
lead to IXa and analogous attack at C6 would produce Xb. This model
was constructed using Spartan 08 for Macintosh to calculate the
equilibrium ground state conformation. The sequence of calculations
used was as follows. A set of conformers was sought via molecular
mechanics calculations using the MMFF force field. Only one low
energy conformer was found. The equilibrium geometry of this
conformer was determined using semiempirical calculations utilizing
the PM3 model. The equilibrium conformation determined in this
round of calculation was then reoptimized using the Hartree-Fock
model with the 3-21G basis set. The equilibrium conformation was
then given a final optimization using the Hartree-Fock model and the
6-31G* basis set. An informative alternative view of this oxetane
isomer, as well as the 5R,6R,4¢R diastereomer, is shown in color in
Fig. S1.
equivalent of H3O+, comparable, but not necessarily equal,
amounts of IXb and Xb. As shown in Fig. 1, the area of Peak 5
(measured at 315 nm) is much smaller than that of P3, even
after 300-fold concentration at 40ꢀC. Indeed, we found, at
most, only trace amounts of material corresponding to Peak 5
upon HPLC of 0.2 mM MeT, freshly irradiated in frozen
solution. We conclude that contributions from the bottom
pathway in Fig. 5 are not significant in this system. The top
pathway, however, cannot be ruled out based on product
studies, as it predicts the formation of a single product, namely
IXb.
Summarizing the above discussion, it can be concluded that
both a mechanism involving backside attack at C6 and one
involving attack of water on a carbocation at C4¢ are possible
mechanisms for the production of P3. These two mechanisms,

810 Martin D. Shetlar and Janet Chung
Me
N
H2O
Me
O
O
O
H
H
O
O
O
N
N
Me
Me
N
H
4'
N
NMe
H
Me
Me
H
H
H
N
+
N
N
H
Me
OH
O
Me
O
O
OH
H-OH
O
N
N
N
H
H
Me
Me
Me
H
Int-1
IXb
H2O
Int-2c
Me
N
O
O
H
O
H
N
O
O
O
N
Me
HN
Me
Me
Me
NMe
H
NMe
H
H
H
H
N
N
N
H2O
O
H
6
+
O
Me
Me
O
O
H-OH
H
O
H
N
N
N
OH
Me H
Me
Me
Int-1
Figure 5. Reaction pathways leading to the formation of IXb and Xb from the common oxetane intermediate Int-1 via carbocationic species, as
Xb
Int-2d
discussed in the text.
500 mL of 0.2 mM MeT in the frozen aqueous state as
R
N
R
N
R
N
O
described previously. This solution was concentrated 100-fold
and analyzed on Column B using the following gradient
flowing at 2 mL min)1: 0 min, 6% (MeOH; 5 min, 6%;
5.5 min, 14%; 8 min, 14%; 8.25 min, 6%; 11 min, 6%). Using
this gradient, P3 eluted at 3.7 min and P5 at 4.8 min. In these
experiments, P5 was barely detectable after rotatory evapora-
tion of the parent solution at 40ꢀC. The ratio [P5] ⁄ [P3]
achieved a final value of 5.2 after 60 min at 100ꢀC. These
results lend experimental support to the idea that P5 is
considerably more stable than P3.
O
O
O
O
N
N
O
N
Me
Me
H
N
H
Me
N
H
Me
H
OH
N
Me
Me
O
OH
H
N
R
H
N
R
O
H
O
:
N
O
R
Xb [R = Me]
IXb
Ald
+
+
+
+
-H
+H
-H
+H
R
N
R
R
N
O
O
O
N
In view of multiple indicators suggesting that P3 is Xb and
P5 is XIb, we will make this as an assumption during the
remainder of this report. These results also provide support for
favoring a mechanism of formation of P3 that involves
backside displacement (Fig. 3) over a mechanism involving
carbocation intermediates, as shown in Fig. 5. It should be
kept in mind, however, that while supported by quantum
mechanical calculations, further experimental work (e.g. X-ray
diffraction studies, as in [35]) remains to be done to verify these
assignments. In the meanwhile, caution should be exercised in
assuming that (5-4) adducts (and (6-4) adducts, see [19]), found
as initially observable photoproducts in pyrimidine nucleo-
base, nucleoside and related systems, will necessarily have a
configuration in which the hydroxyl and 2¢-pyrimidone moi-
eties are cis with respect to one another.
O
O
O
N
N
N
-H O
2
+H O
2
Me
Me
Me
H
H
H
N
N
N
Me
Me
-H O
2
Me
H
OH
+H O
2
+
OH
2
+
+
N
R
H
N
R
H
N
R
Int-5
O
O
O
2
Int-4
Int-3
Figure 6. Possible reaction pathways leading to the isomerization of
Xb to IXb, as described in the text.
the dihydropyrimidine ring; this mechanism is detailed in
fig. 1.18 in (11). An alternative mechanism is that shown in the
lower part of Fig. 6. In this mechanism, protonation of the 6-
hydroxy group of Xb occurs to yield Int-3, followed by the loss
of water to form a carbocation with a chiral center at C5
(shown in a resonance form in Int-4) followed by the addition
of water to give Int-5; deprotonation of Int-5 leads to IXb.
Attack of water can occur from either side of the bottom ring in
Int-4; as the two sides are stereochemically inequivalent, the
corresponding rates of addition will not necessarily be equal.
Thus, after equilibrium has been reached, the relative amounts
of IXb and Xb may be quite different (in agreement with results
discussed above, in which it was found that at 100ꢀC, the
equilibrium ratio of [P5] ⁄ [P3] takes a value of 5.1). Similarly,
the same set of steps run in reverse would lead to Xb if IXb were
to be the starting material.
The mechanism of interconversion of Xb and IXb. By what
mechanism could Xb be converted into IXb? One possible
pathway is shown in the upper portion of Fig. 6, in which
water-mediated conversion of Xb to an intermediate aldehyde
(Ald in Fig. 6) is followed by ring closure to form IXb.
Eventually an equilibrium mixture of IXb and Xa would be
achieved, with the thermodynamically favored isomer domi-
nating. Such a mechanism has been proposed to account for the
interconversion of four forms of the ‘‘photohydrate’’ of Thd,
which similarly contain a hydroxyl group at the 6-position of

Photochemistry and Photobiology, 2011, 87 811
six Dewar valence isomers, corresponding to two each for IIIb,
IXb and Xb. (The possibility of production of two chromato-
graphically resolvable Dewar valence isomers for each parent
(6-4) adduct does not appear to have been previously noted; it
arises from the chirality of C¢6 in these Dewar valence isomers,
as discussed at the end of the Introduction.)
Stereochemistry of P6, the (6-4) adduct of MeT
Discussions similar to those given above can be given for
reaction of parent oxetane to form the MeT (6-4) adduct (P6).
These arguments conclude that P6 most likely has structure
IIIb, in which C5-hydroxyl and 2¢-pyrimidone at C6 are cis in
their configuration with respect to one another. These matters
are pursued in Appendix S1. Also given there are the results of
quantum mechanical calculations that indicate the MeT (6-4)
adduct IIIb is more stable than the corresponding adduct with
the C5-hydroxyl group trans to the 2¢-pyrimidone group at C6.
Prefatory comments. In general, the method of choice of
preparing the Dewar valence isomers is irradiation with UVB
light. With increasing exposure, the characteristic absorption
spectrum of the parent (6-4) adduct gradually disappears in the
case of thymidylyl-(5¢-3¢)-thymidine and closely related dinu-
cleoside monophosphates. The resultant Dewar adduct has
little absorbance above 250 nm. An illustration comparing the
spectra of the (6-4) adduct of thymidylyl-(5¢-3¢)-2-deoxyuridine
and the corresponding Dewar valence isomer is provided by
fig. 3 in Lemaire and Ruzsicska (40). Irradiation of Dewar
valence isomers can result in conversion of such compounds
back to their parent (6-4) adduct (see, for example, [6] where
this reaction of the TpT Dewar valence isomer is described).
In Appendix S1, we describe the methodology and results of
exploratory experiments that show that two (5-4) adducts (P3
and P5) and the (6-4) adduct (P6) of MeT each form two
Dewar valence isomers upon UVB irradiation. Also described
there are the preparation, isolation and purification of the
various Dewar valence isomers on a scale suitable for
spectroscopic and chemical studies. Below we describe the
spectroscopic and chemical characterization of these Dewar
adducts.
Yields of IIIb and Xb
We carried out a study designed to estimate the isolated yields of
IIIb and Xb. We irradiated 1000 mL of 0.2 mM MeT on dry ice
for 2 · 64 min time periods. Comparative analytical HPLC of
the unirradiated and freshly irradiated solution showed that ca
64.5% of the parent had reacted, corresponding to 18.2 mg of
MeT being converted to product. The irradiated solution was
concentrated to 3.3 mL and separated on Column A using
Gradient C as described previously. The purity of collected F3
and F6 was checked via HPLC on Column B using 94 ⁄ 6
H2O ⁄ MeOH flowing at 2 mL min)1. Both fractions were quite
pure. Assay of collected F3 by UV spectroscopy (using the
assumption that emax = 4940, that of IIIa) showed that we had
isolated 6.6 lmol (1.84 mg) of Xb, while a similar assay for F6
indicated that 1.56 lmol (437 lg) of IIIb was isolated. This
corresponds to yields of 10.1% (Xb) and 2.4% (IIIb). These
values suggest that the (5-4) adduct is preferentially produced,
by a factor of ca 4.2, under the reaction conditions used for this
experiment. This ratio is in reasonable agreement with the ratio
of the peak areas corresponding to Xb and IIIb found (3.6, when
measured at 225 nm) when 100 lL of freshly irradiated solution
was injected on Column B, using the following HPLC gradient:
0 min, 8% MeOH; 3.5 min, 8%; 5 min, 16%; 8.75 min, 16%;
Characterization of the Dewar valence isomers of P3 (Xb). The
two Dewar products P3D1 and P3D2, obtained as described in
Appendix S1, were subjected to ESI mass spectrometry, which
indicated that the molecular mass of each was consistent with
that expected for a dimeric form of MeT. Both mass spectra
displayed strong peaks with values of 303.1 [M + Na]+ and
319.1 [M + K]+, and a very weak peak at 281.1, correspond-
ing to [M + H]+. The proton NMR spectra of both P3D1
and P3D2, run in DMSO-d6, displayed all of the expected
resonances (Table 1). As noted for the parent compound, the
peaks assigned to C6H and C6OH are coupled with the
magnitude of the coupling constant being about the same as
that observed for the corresponding peaks in P3. To the
sample corresponding to P3D1, we examined the effect of
adding a small amount of D2O to the NMR sample; as
expected, this spin–spin splitting disappeared. It can be noted
that the d values for the various proton resonances for P3D1
and P3D2 are quite similar, supporting the supposition that
they are similar in structure. As mentioned above, neither of
the compounds displays a UV absorbance maximum at
k > 210 nm. These spectral data are consistent with the two
compounds being identified as two diastereomeric Dewar-type
adducts with the general structure shown as XIIb (based on the
identification of P3 as Xb; see above).
9 min, 8%; 12 min, 10%, flow rate: 2 mL min)1
.
Identification of the MeT CBD
The CBDs of MeT has been previously studied in detail by
Stepien et al. (31). In Appendix S1, we describe the results of
studies that showed that the compounds contained in F2 (P2),
F7 (P7), F8 (P8) and F10 (P10) were indeed CBDs and that the
stereochemistry of these products can be assigned as follows:
P2: c,s; P7: t,s; P8: c,a; P10: t,a. These studies employed
methylation technology (36), along with authentic samples of
the CBDs of Thy prepared as described in (37).
Preparation and characterization of the Dewar adducts of IIIb,
IXb and Xb
As mentioned in the Introduction, irradiation of the (6-4)
adducts of a number of dinucleoside monophosphates with
UVB light converts these compounds into products in which
the 2¢-pyrimidone portion of the molecule has been converted
to a Dewar valence isomer (see, for example [6, 38–42]). The
production and biological importance of such Dewar adducts
in the context of DNA, as well as their parent (6-4) adducts,
has been considered in detail (for reviews, see [14, 15]). In the
following, we describe the preparation and characterization of
Characterization of the Dewar valence isomers of P6 (IIIb). The
molecular masses of P6D1 and P6D2, prepared as described
in Appendix S1, were determined using ESI mass spectrom-
etry. Both compounds have the expected mass of 280,
displaying peaks at 281.1 [M + H]+, 303.1 [M + Na]+
and 319.1 [M + K]+. Proton NMR spectra, run in

812 Martin D. Shetlar and Janet Chung
DMSO-d6, produced the data given in Table 1. All of the
expected resonances for a Dewar adduct are present for both
compounds; however, as compared to the NMR spectra for
the two diastereomers corresponding to XIIb, neither the
C6H nor the OH protons are split. This supports the
hypothesis that the OH moiety is located at C5 of the
dihydropyrimidine ring, rather than at C6. Taken as a whole,
the above data are consistent with P6D1 and P6D2 being two
diastereomeric Dewar valence isomers of IIIb, one of which is
shown in Scheme 2 as IVb.
It should be noted that the d value for the C6¢H in each of
the two Dewar adducts P6D1 and P6D2 is near 5, while the
value of d for the C6¢H proton in IIIb is 7.93. (A similar
statement can be made about the corresponding compounds
discussed immediately above and below.) For the TpT (6-4)
adduct and the TpT Dewar valence isomer, the corresponding
d values (in D2O) are 5.33 and 8.00 (6). These observations,
while not made in the same solvent, tend to support the idea
that the compounds being described are indeed Dewar
valence isomers.
methodology of the calculations are given in the caption to
Fig. S3.) The data in Table S2 indicate that the difference in
the energies of XIIb-1 and XIIb-2 is considerably larger in
magnitude, both in the gas phase and in the aqueous phase,
than the corresponding energy differences for the other two
sets of Dewar valence isomers. Direct correlations of product
ratios with differences in a thermodynamic quantity are not
rigorous; however, it is interesting to note that the observed
product ratio of the two isomers formed in the photoinduced
reaction of P3 (putatively Xb) to form two Dewar adducts
(putatively XIIb-1 and XIIb-2) was 8 ⁄ 1 (see Appendix S1),
while the corresponding observed product ratios for the
photoreactions of P5 (putatively IXb) and P6 (IIIb) were ca
1.4 and 1.25 respectively (Appendix S1).
Effects of irradiating at k = 254 nm on the various Dewar
adducts. To provide additional evidence that our assignments
as Dewar valence isomers, as discussed above, were correct, we
irradiated an appropriately diluted solution of each of the six
Dewar adducts at 254 nm in a quartz spectrophotometer cell
and followed the spectrum as a function of time irradiated. In
all six cases, the spectrum of the corresponding (5-4) or (6-4)
adduct became increasingly strong with increasing time of
irradiation. This behavior is similar to that previously
observed for the Dewar valence isomer of the TpT (6-4)
adduct (6).
Characterization of the Dewar valence isomers of P5. For
completeness sake, we also prepared and characterized the two
corresponding photoproducts of P5, namely P5D1 and P5D2
(see Appendix S1).
Using ESI mass spectrometry, the molecular masses of these
two products were determined. The spectrum of P5D1 showed
a very strong peak at 303.2 ([M + Na]+) and a much weaker
peak at 319.2 ([M + K]+), while the corresponding peaks for
P5D2 showed peaks at 281.2 [M + H]+, 303.2 [M + Na]+
and 319.2 [M + K]+, with the latter two peaks being the
strongest. The NMR spectra corresponding to these two
compounds, each run in DMSO-d6, displayed all of the
expected resonances (see Table 1). Of particular interest are
the resonances corresponding to C6H and C6OH. As was the
case for P3D1 and P3D2, these two protons are coupled to one
another in P5D1 and P5D2. This is consistent with the
assignment of these two compounds to be the two diastereo-
meric Dewar adducts having the generalized structure XIb.
Effect of heating on P6 (IIIb) in the presence and absence of acid
A 150 lL portion of purified P6 was placed in each of two
screw-top 1.5 mL Sarstedt plastic tubes. To one was added
enough 1 M HCl to make the solution 0.1 M in acid. Both were
placed in a boiling water bath for 10 min. The acid solution
was then evaporated to dryness by rotatory evaporation and
redissolved in 150 lL of water. The HPLC of each of the
resulting reaction mixtures was run on Column E (92 ⁄ 8
H2O ⁄ MeOH flowing at 2 mL min)1). There was no evidence
for decomposition of P6 in either solution. In a separate
experiment, a sample was evaporated to dryness and taken up
in 11.7 M HCl. This solution was heated for 30 min at 100ꢀC.
After removal of acid by rotatory evaporation and redissolu-
tion in water, HPLC was run on Column B, using both 90
Quantum mechanical calculations of the energies of the
equilibrium geometries of the diastereomeric forms of the
Dewar valence isomers of the MeT (6-4) adduct and the MeT
(5-4) adducts. As shown above, irradiation of P3, P5 and P6
produces, in each case, two Dewar valence isomers with
HPLC properties such that they can be readily separated on a
phenyl reverse phase column. One question that can be asked
is whether, in fact, the diastereomeric Dewar valence adducts
of the MeT (6-4) and (5-4) adducts would be predicted via
theoretical methods to have significantly different equilibrium
geometries and, thus, potentially different energies. To
address this question, we determined the equilibrium geom-
etries and the corresponding energies for the members of each
pair of such Dewar valence isomers using Spartan 08 for
Macintosh. In each case, the geometric relationship of the
Dewar moiety to the dihydropyrimidine ring was significantly
different for the two diastereomers. As an example of the
result of these calculations, Fig. S3 provides a visual depiction
of the results of such calculations for the two diastereomers
corresponding to IVb, while Table S2 provides the values of
the energies calculated in each of the six cases. (Details of the
H2O ⁄ 10 MeOH (2 mL min)1
) and 75 H2O ⁄ 25 MeOH
(1.6 mL min)1) as eluents. No evidence for decomposition
was observed. In particular, material with the properties of P9
(the material with a double-humped spectrum, see Materials
and Methods) was not present, suggesting that this material is
not a (6-4) adduct dehydration product (4). (The actual nature
of P9 has been determined and will be discussed in a later
publication.) The stability of P6 (IIIb) to acid-induced dehy-
dration contrasts to the lability of IIIa toward the same
reaction. As the number of examples is limited, it will require
further study to determine if only (6-4) adducts with a
hydrogen at N1 on the dihydropyrimidine portion of the
adduct can undergo acid-mediated dehydration similar to that
undergone by IIIa (For thymidine (6-4) adducts, loss of sugar
under mild acid hydrolysis conditions, leading to IIIa, fol-
lowed by loss of water under stronger acid hydrolysis
conditions [to form XIII] has been shown to occur [28].
Analogs of XIII with 2¢-deoxyribose moieties at N1 [and ⁄ or
N1¢] do not appear to have been reported.)

Photochemistry and Photobiology, 2011, 87 813
diated at 254 nm, the absorbance at 317 nm increased with
increasing time of exposure. However, other photoreactions
appeared to be occurring as well, as extra absorbance at
wavelengths between 260 and 275 nm increased in parallel
with the appearance of the absorbance at kmax of the (6-4)
adduct. Although we did not pursue the nature of the reactions
responsible for production of this extra absorbance, these
results suggest that caution should be exercised in assuming
that conversion of (6-4) adducts to Dewar adducts by UVB,
and then back to parent (6-4) adducts with 254 nm radiation,
is a completely reversible process.
The acid hydrosylates of P7¢ and P10¢ were similarly worked
up and analyzed on Column B, using 82 ⁄ 18 H2O ⁄ MeOH.
While the parent material disappeared and several products
were formed, none of these was XIII. (We did not attempt to
identify the various acid hydrolysis products.) It can be
concluded that P7¢ and P10¢ do not display the acid hydrolysis
behavior expected of (6-4) adducts; thus, these data do not
contradict the supposition that these two compounds are (5-4)
adducts.
The (5-4) adducts of thymidine
As mentioned in Materials and Methods, we carried out
exploratory experiments to determine if compounds with the
expected properties of (5-4) adducts were produced when Thd
is irradiated in the frozen state and indicated that these
experiments gave an affirmative answer. In the following
section, we give the results of chemical studies designed to
determine which of the four compounds of interest (P3¢, P5¢,
P7¢¢ and P10¢; see Materials and Methods for the meaning of
these designations) are likely to be (6-4) adducts and which are
candidates to be (5-4) adducts. We then describe the results of
our studies of the (5-4) candidates using UV spectroscopy,
mass spectrometry and proton NMR spectroscopy, as well as
chemical and photochemical methods. We will first consider
P3¢ and P5¢, showing that these two products correspond to (6-
4) adducts. We will then provide strong evidence that P7¢ and
P10¢ constitute a pair of diastereomeric Thd (5-4) adducts.
Demonstration that P3¢ and P5¢ are (6-4) adducts and that P7¢
and P10¢ are possibly (5-4) adducts. A definitive distinction
between (6-4) adducts and (5-4) adducts can be made by
heating these compounds at 100ꢀC in reasonably concentrated
HCl. As noted above, Thd (6-4) adducts (i.e. IIIc) undergo loss
of sugars at N1 and N1¢ upon heating in mild acid to form IIIa
and then, upon heating in strong acid, lose water to form XIII
(28). Compounds such as IXc or Xc would not be expected to
undergo acid-induced reaction to form the compound XIII.
We utilized this expected difference in behavior to assign the
appropriate adduct type for P3¢, P5¢, P7¢ and P10¢.
Structural characterization of P7¢ and P10¢. The following
spectroscopic data are also consistent with the assignment of
P7¢ and P10¢ to be two diastereomeric forms of the thymidine
(5-4) adduct; one possible diasteromeric form is shown as IXc
in Scheme 2. Compounds P7¢ and P10¢ have nearly identical
UV absorption spectra (kmax = 317 and kmin = 251 nm,
respectively). Both compounds display the expected ESI mass
spectral behavior for thymidine (5-4) adducts, with peaks
appearing at 485.2 [M + H]+, 507.2 [M + Na]+ and 523.2
[M + K]+. The proton NMR data obtained from freshly
purified samples of the two compounds, run in D2O with TSP
as a reference compound, are given in Table 2. All of the
expected peaks are present. In particular, each compound
shows resonances corresponding to two methyl (Me) groups,
one with chemical shift near d = 1.65 and the other near
d = 2.1; this is consistent with each of these two compounds
containing a Me group within a dihydropyrimidine ring and a
Me attached to a vinyl group. Similarly, each compound
displays a resonance that can be assigned to a vinyl proton
near d = 8.3, as well as a resonance (d near 5.2) that
corresponds to a proton attached to C6 in a dihydropyrim-
idine ring. For each compound, the chemical shifts and
coupling constants (estimated via simulation) for the two sets
of sugar protons given in Table 2 are also in line with the
expectation for protons located within 2¢-deoxyribofuranose
ring systems.
We placed 50 lL of purified P3¢, P5¢, P7¢ and P10¢ into
1.5 mL plastic screw-top vials (Sarstedt) and added to each
vial 50 lL of 37% (11.7 M) HCl. The four vials were heated for
40 min at 100ꢀC. The contents of each vial were concentrated
to dryness by rotatory evaporation and the resulting residue
was taken up in 500 lL of distilled water. Using Column B
with 94 ⁄ 6 H2O ⁄ MeOH, flowing at 3 mL min)1, 25 lL injec-
tions of the acid-treated P3¢ and P5¢ showed that almost
complete conversion of the parent compound to XIII
(Scheme 3), eluting at 2.63 min, had occurred, based on
identity of the product retention times and UV spectrum with
that of authentic XIII (prepared as described in [4]). When
corresponding samples were heated for 20 min in 10% HCl,
conversion was not complete; a considerable amount of IIIa
was present as well (as observed by Varghese [28] under mild
acid hydrolysis conditions). These observations provide evi-
dence that P3¢ and P5¢ are diastereomeric (6-4) adducts of Thd.
The results of a study of the effects of UVB light on P3¢ and
P5¢ are consistent with the assignment of these two compounds
as (6-4) adducts. Douki et al. (30) have previously indicated
that the (6-4) adducts of Thd are readily converted to the
corresponding Dewar valence isomers. In our confirmatory
study, spectroscopic monitoring of the results of irradiation of
P3¢ and P5¢ with UVB radiation showed, in agreement with the
results of Douki et al., these two compounds displayed the
spectroscopic behavior expected if conversion of (6-4) adducts
to corresponding Dewar adducts was occurring. With increas-
ing time of irradiation with UVB, the absorption at kmax
disappeared; eventually the observed spectrum displayed little
absorbance at wavelengths above 260 nm. When the quartz
cuvette containing either photoreacted P3¢ or P5¢ was reirra-
In summary, spectroscopic data given above, along with the
results from acid hydrolysis experiments on P3¢, P5¢, P7¢ and
P10¢, are consistent with P7¢ and P10¢ being diastereomeric
forms of a (5-4) adduct; however, as discussed below, it remains
an open question for each of these adducts as to whether it has
the type of structure shown in Xc or, alternatively, IXc.
Exploratory studies of the stability of P7¢ and P10¢
To gain some idea of the kinds of chemistry that occurs when
the putative (5-4) adducts of thymidine are heated or,
alternatively, irradiated with UVB light, we carried out some
preliminary studies of the reactive properties of these two
compounds.

814 Martin D. Shetlar and Janet Chung
Table 2. Proton NMR data for the (5-4) adducts of thymidine.
Adduct
P7¢
P10¢
Ring proton(s)
d (Dihydropyrimidine)
d (2¢-Pyrimidone)
d (Dihydropyrimidine)
d (2¢-Pyrimidone)
CH3
H6
1.67
5.20
2.12
8.30
1.65
5.26
2.12
8.33
Sugar protons
d (A)
d (B)
d (A)
d (B)
H1¢
H2¢
H2¢¢
H3¢
H4¢
H5¢
H5¢¢
6.22
2.38
2.30
4.41
3.94
3.69
3.74
6.26
2.38
2.65
4.45
4.18
3.91
3.80
6.24
2.37
2.64
4.46
4.17
3.93
3.82
6.27
2.39
2.23
4.41
3.96
3.78
3.73
Coupled protons
J
J
J
J
H1¢–H2¢
H1¢–H2¢¢
H2¢–H2¢¢
H2¢–H3¢
H2¢¢–H3¢
H3¢–H4¢
H4¢–H5¢
H4¢–H5¢¢
H5¢–H5¢¢
7.2
6.7
14.0
6.8
4.5
3.9
4.7
3.8
6.1
6.3
5.5
14.4
5.6
6.7
4.3
3.4
4.7
8.9
6.0
13.8
6.3
3.0
3.1
3.6
4.8
6.3
14.3
6.0
4.8
4.5
3.3
5.2
12.5
12.2
12.6
12.3
Spectra were run in D2O, using TSP as a reference compound; P7¢ is the earlier eluting (5-4) adduct (Peak 7) and P10¢ is the later eluting product
(Peak 10) in Fig. 2. For P7¢, Total Correlation Spectroscopy (TOCSY) was used to partition the sugar protons into two spin systems, designated
‘‘A’’ and ‘‘B.’’ The ring component that each spin system is associated with has not been determined. For P10¢, the assignment of sugar protons to a
particular ring system was based on the assignments by Shaw (29). The chemical shifts (d) and magnitudes of the coupling constants (J) reported for
the sugar protons were estimated with SpinWorks (Version 3.1, 2009, Kirt Marat, University of Minnesota), using the simulation module, NUM-
MRIT, with default parameters (NUMMRIT algorithm reference: J. S. Martin and A. R. Quirt, J. Magn. Reson. 5, 318 [1971]). For each adduct,
the simulated proton NMR spectrum closely matched the experimental spectrum.
Photostability of P7¢ and P10¢. We did a limited set of
experiments that suggest that P7¢ and P10¢ undergo isomeri-
zation to corresponding Dewar valence isomers when irradi-
ated with UVB light. Purified P7¢ and P10¢ (absorbance: 0.6)
were each placed in a quartz cuvette and placed next to an
unfiltered UVB lamp (see Materials and Methods). In the case
of P7¢, ca 92% of the absorbance at 318 nm disappeared over a
period of 8 min of irradiation. A spectrum without an
absorption maximum above 230 nm replaced the original
spectrum. A similar behavior was seen for P10¢. When these
UVB-irradiated solutions were placed in front of a 254 nm
lamp and reirradiated, the absorbance at 318 nm increased
with time of exposure over 8 min. As was the case with the Thd
(6-4) adducts, ‘‘extra’’ absorbance appeared in the wavelength
range between 260 and 275 nm.
When the UVB-irradiated P7¢ was analyzed on Column B
using 93 ⁄ 7 H2O ⁄ MeOH flowing at 2 mL min)1 as eluent and
detection at 225 nm, two peaks appeared with UV spectra
corresponding to those anticipated for Dewar valence iso-
mers. A large peak (te = 8.1 min) eluted just prior to the
parent peak (te = 8.6 min), while a much smaller peak eluted
at 9.4 min. The peak area of the larger peak, measured at
225 nm, constituted ca 92% of the total of the two peak areas
corresponding to Dewar adducts. Similarly, the irradiated
solution of P10¢ was analyzed using 84 ⁄ 16 H2O ⁄ MeOH
flowing at 2 mL min)1. Two peaks, eluting at 1.5 and
1.8 min, came off the column with UV spectra consistent
with these compounds being Dewar valence isomers of P10¢;
the parent compound eluted at 3.3 min. The peak area of the
peak eluting at 1.8 min corresponded to ca 94% of the total
peak area for these two peaks. (As indicated above, in the
analogous results for irradiation of the MeT (5-4) adduct P3
[Xb] [in which the 6-hydroxyl and 2¢-pyrimidone ring are
putatively trans to one another], the predominant Dewar
adduct made up ca 89% of the total product, based on peak
area measurements.) In a separate experiment, P10¢ was
irradiated as above for 12 min and HPLC was run using
90 ⁄ 10 H2O ⁄ MeOH flowing at 2 mL min)1. About 90% of
the parent compound (te = 6.5 min) was converted to Dewar
products, the predominant one eluting at 3.3 min. The
solution was reirradiated at 254 for 12 min and HPLC run
again. About 50% of the Dewar was consumed, while the
amount of P10¢ had increased approximately five-fold.
Several other small product peaks, some having absorbance
between 250 and 280 nm, appeared as well. These two sets of
results suggest that the UVB light-mediated reaction of P7¢
and P10¢ does indeed result in Dewar valence isomerization;
furthermore, this reaction shows a reasonable degree of
reversibility when the putative Dewar adducts are irradiated
at 254 nm. The definitive identification of these products as
Dewar valence isomers must await NMR and mass spectral
studies.

Photochemistry and Photobiology, 2011, 87 815
Taken as a whole, the results of the above studies with UVB
light are consistent with the expectation for compounds that
contain a 2¢-pyrimidone ring system and provide supporting
evidence for the assignment of the structures of P7¢ and P10¢ as
(5-4) adducts.
that obtained by Shaw, while the chemical shift values given by
Shaw have values very close in magnitude to those given in
Table 2 for P10¢. Indeed, we concluded that dThd(6-4)Pyo II
and P10¢ are the same compound and used the data given by
Shaw to guide our assignment of the sugar protons to
appropriate rings of P10¢, as given in Table 2.
Thermal stability of P7¢ and P10¢. As noted in the discussion of
the product P3 of MeT, it decomposes upon standing to form
the isomeric compound P5. While we have not studied the
thermal reactions of P7¢ and P10¢ in detail, we noted that these
compounds are unstable when allowed to stand at room
temperature. (For this reason, solutions containing P7¢ and
P10¢ were stored long term in the frozen state; furthermore,
these two compounds were repurified immediately before
preparing NMR samples.) When P7¢ is incubated for 16 h at
room temperature in D2O and then analyzed on Column B,
using 82 ⁄ 18 H2O ⁄ MeOH flowing at 2 mL min)1, a second
peak (te = 1.60 min) appears, eluting shortly after the parent
peak (te = 1.47 min); the parent area constituted approxi-
mately 2 ⁄ 3 of the total peak area for the two peaks. In
addition, two considerably smaller peaks eluted at 2.1 and
2.25 min. After a similar incubation of P10¢, a second peak
(te = 2.4 min) again eluted after the parent peak (te = 2.25
min); the fraction of the total peak area corresponding to
parent was once more ca 2 ⁄ 3. Again, two much smaller peaks
elute at 1.6 and 1.75 min. For both P7¢ and P10¢, each of the
peaks corresponding to decomposition products displayed UV
absorption spectra with nearly the same profile as the parent
peak. These results suggest that P7¢ and P10¢ undergo thermal
isomerization reactions to form compounds containing the 2¢-
pyrimidone moiety. It remains to be determined, however,
whether these reactions result in the formation of (5-4) adducts
in which the hydroxyl group and the 2¢-pyrimidone ring have
changed their cis ⁄ trans relationship with respect to one
another or, instead, reflect the results of sugar isomerization
reactions (or both).
Following Varghese’s lead, we subjected samples of P7¢ and
P10¢ to heating in 0.1 M NH4OH at 100ꢀC for 30 min and then
ran the HPLC of the resultant samples (after rotatory evapo-
ration to dryness and redissolution to the same volume with
water). We used Column B with 90 ⁄ 10 H2O ⁄ MeOH flowing at
2 mL min)1 as eluent. One of the main reaction products could
be identified as Thd by identity of both te (2.8 min) and the UV
spectrum with that of an authentic sample; there were also a
number of products with spectra similar to that expected for
MPD. These may be sugar isomerization products of either
MPD or of the parent adduct; however, we did not pursue the
identity of these compounds. When P3¢ and P5¢ were similarly
subjected to treatment with 0.1 M NH4OH, it was determined
that these two compounds were much more stable to reaction.
When HPLC analysis was conducted as above, there were only
traces of decomposition products with (6-4) adduct-like spectra
and a barely detectable amount of Thd.
The observations, described in the paragraphs immediately
above, strongly suggest that the Td4 product of Varghese has the
same nature as the Thd (5-4) adducts. Indeed, it is possible that
Td4 was a mixture of P¢7 and P¢10, as the resolution of the
chromatographic methods used by Varghese to isolate this
compound was intrinsically lower than that of the methods used
in this work. (Indeed, the same can be said about Varghese’s
photoproduct Td1, which was possibly a mixture of P3¢ and P5¢.)
An unanswered question: do P7¢ and P10¢ have structures of the
type given by IXc or Xc.? This question cannot be answered
definitively, based on the experimental information presented
in this paper. While Dewar product ratios for P7¢ and P10¢,
discussed above, are similar to those seen for P5, the putative
MeT (5-4) trans adduct, these observations can be regarded
only as suggestive that P7¢ and P10¢ have a trans configuration.
Thus, the question of whether the 6-hydroxyl and 2¢-pyrim-
idone ring in P7¢ and P10¢ are cis or trans with respect to one
another remains to be answered, possibly by X-ray diffraction
studies if high-quality crystals can be obtained.
Stability of P7¢ and P10¢ in 0.1 M aqueous ammonia. It is very
likely that (5-4) adducts of Thd have been observed at least
twice in previously reported work, but not recognized as such.
Varghese (28), in his early work on the photochemistry of Thd
in frozen solution, isolated two products with UV absorption
characteristic of a 2¢-pyrimidone ring. As alluded to previ-
ously, he assigned one of these (denoted as Td1 in his paper) as
a (6-4) adduct, based on the acid hydrolysis behavior of this
compound. Varghese did not make a final structural assign-
ment for Td4, the second adduct, but concluded that it was an
unstable isomer of IIIc. However, he did a couple of chemical
studies with results suggesting that, in fact, Td4 could be
different in nature from a (6-4) adduct. In particular, he found
that acid hydrolysis of Td4 led to Thy and assorted other
compounds, rather than the anticipated IIIa. Furthermore, he
found that when he treated Td4 with 0.1 M NH4OH for 30 min
at 80ꢀC, he recovered Thd and 5-methyl-2-pyrimidone deoxy-
ribonucleoside (MPD) (43) as products.
Concluding remarks
We have found that MeT forms both a (5-4) and a (6-4) adduct
when irradiated in frozen aqueous solution at )78.5ꢀC; the (5-
4) adduct, in which the C6¢ hydroxyl group and the 2¢-
pyrimidone ring at C5¢ appear to be trans with respect to one
another, undergoes a thermal isomerization, forming a second
(5-4) adduct in which the 2¢-pyrimidone and hydroxyl groups
are putatively cis to one another. We have also shown that
thymidine forms a pair of diastereomeric (5-4) adducts, as well
as two (6-4) adducts, previously indicated to have been isolated
and characterized. Stable (5-4) adducts, resulting from photo-
reaction of DNA pyrimidine nucleosides or closely related
compounds, such as MeT, have not been previously reported.
In addition, we have demonstrated that irradiation of each
of the MeT adducts (both (5-4) and (6-4)) with UVB light
results in formation of a pair of the diastereomeric Dewar
More recently, Shaw (in his thesis [29]) reported a com-
pound with a proton NMR spectrum very much like that
obtained for one of the (5-4) adducts described here (P10¢);
however this compound was assigned as a (6-4) adduct. (In his
nomenclature, the compound was termed dThd(6-4)Pyo II.)
The profile of the NMR spectrum of P10¢ was very similar to

816 Martin D. Shetlar and Janet Chung
2. Wang, S. Y. (1960) Reversible behavior of the ultra-violet irradi-
ated deoxyribonucleic acid and its apurinic acid. Nature 188, 844–
846.
3. Moore, A. M. (1958) Ultraviolet irradiation of pyrimidine deriv-
atives: II. Note on the synthesis of the product of reversible
photolysis of uracil. Can. J. Chem. 36, 281–283.
4. Varghese, A. J. and S. Y. Wang (1968) Thymine–thymine adduct
as a photoproduct of thymine. Science 160, 186–187.
5. Taylor, J.-S. and M. P. Cohrs (1987) DNA, light, and Dewar
pyrimidinones: The structure and biological significance of TpT3.
J. Am. Chem. Soc. 109, 2834–2835.
6. Taylor, J.-S., D. S. Garrett and M. P. Cohrs (1988) Solution-state
structure of the Dewar pyrimidinone photoproduct of thymidylyl-
(3¢-5¢)-thymidine. Biochemistry 27, 7206–7215.
valence isomers separable by HPLC; preliminary evidence
suggests that a similar reaction occurs for the two Thd (5-4)
adducts. These diastereomeric pairs evidently arise from the
production of a chiral center at C6¢ in the valence isomerization
reaction. Quantum chemical calculations indicate that the two
members of each diastereomeric pair of Dewar adducts,
differing in chirality at C6¢, have significantly different equi-
librium geometries along with different calculated energies.
It remains to be determined whether (5-4) adducts involving
Thy residues are formed when denatured DNA or DNA in
structural forms alternative to the B form are irradiated.
Intrastrand photoreactions have been shown to lead to high
yields of anti isomers of the Thy dimer when DNA in the A
form is irradiated with UVC (44), while irradiation of human
telomeric G-quadruplex DNA similarly leads to the formation
of Thy CBD with anti conformations (45). It would be
interesting to determine if (5-4) adducts are formed in these or
other non-B DNA systems, arising from photoreaction of Thy
contained in different strands or loops within such structures.
7. Nishio, T., A. Kato, C. Kashima and Y. Omme (1980) Photo-
chemical electrocyclization of 1,4,6-trisubstituted pyrimidin-2-
ones to 2-oxo-I,3-diazabicyclo[2.2.0] hex-5-enes. J. Chem. Soc.
Perkin Trans. 1, 607–610.
8. 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.
9. 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.
10. 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.
11. 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.
12. Davies, R. J. H. (1995) Ultraviolet radiation damage in DNA.
Biochem. Soc. Trans. 23, 407–418.
13. 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.
Acknowledgements—We gratefully acknowledge NSF Grant CH-
0131203 and the Bio-organic, Biomedical Mass Spectrometry Resource
(A.L. Burlingame, Director), supported by the NIH Division of
Research Resources Grant RR 01614; in particular, we thank David
Maltby for obtaining the required mass spectra in a timely manner. We
also acknowledge helpful discussions with Professor Norman Oppen-
heimer. Finally, we thank the referees for their critical reading of the
manuscript and their helpful suggestions for improving it.
SUPPORTING INFORMATION
14. Cadet, J., E. Sage and T. Douki (2005) Ultraviolet radiation-
mediated damage to cellular DNA. Mutat. Res. 571, 3–17.
15. 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.
16. Rhoades, D. F. and S. Y. Wang (1971) A new photoproduct of
cytosine. Structure and mechanism studies. J. Am. Chem. Soc. 93,
3779–3781.
17. Rhoades, D. F. and S. Y. Wang (1971) Photochemistry of poly-
cytidylic acid, deoxycytidine and cytidine. Biochemistry 10, 4603–
4611.
18. Bergstrom, D. E. and N. J. Leonard (1972) Photoreaction of 4-
thiouracil with cytosine. Relation to photoreactions in Eschericia
coli transfer ribonucleic acids. Biochemistry 11, 1–9.
19. Shetlar, M. D. and V. J. Basus (2011) The photochemistry of
uracil: A reinvestigation. Photochem. Photobiol. 87, 82–102.
20. Douki, T. and J. Cadet (2003) Formation of the spore photo-
product and other dimeric lesions between adjacent pyrimidines in
UVC-irradiated dry DNA. Photochem. Photobiol. Sci. 2, 433–436.
21. Desnous, C., D. Guillaume and P. Clivio (2010) Spore photo-
product: A key to bacterial eternal life. Chem. Rev. 110, 1213–1232.
22. Wang, S. Y. (1971) Thymine phototrimer. J. Am. Chem. Soc. 93,
2768–2771.
23. Karle, I. L. (1976) Crystal and molecular structure of photo-
products from nucleic acids. In Photochemistry and Photobiology
of Nucleic Acids Vol. 1 (Edited by S. Y. Wang), pp. 483–519.
Academic Press, New York.
Additional Supporting Information may be found in the online
version of this article:
Appendix S1. Experimental details and results from calcu-
lations.
Figure S1. Perspectives of two diastereomeric forms of the
putative oxetane precursor of the (5-4) adduct of MeT, shown
as both ball and stick and space-filling models.
Figure S2. Perspectives of two diastereomeric forms of the
putative oxetane precursor of the (6-4) adduct of MeT, shown
as both ball and stick and space-filling models.
Figure S3. Tube and ball and stick representations of the
5R,6S,6¢R and 5R,6S,6¢S Dewar valence isomers produced by
photoreaction of the 5R,6S diastereomer of the (6-4) adduct of
MeT.
Table S1. Correlation between HPLC peak number, frac-
tion number, product designation and assigned structure for
the irradiated MeT and thymidine systems.
Table S2. Calculated energies for the equilibrium geometries
of the two diastereomeric Dewar valence isomers of IVb, XIb
and XIIb.
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.
24. Rahn, R. O. and L. Hosszu (1969) Photochemical studies of
thymine in ice. Photochem. Photobiol. 10, 131–137.
25. Blazek, E. R., J. L. Alderfer, W. A. Tabaczynski, V. C. Stamoudis,
M. E. Churchill, J. G. Peak and M. Peak (1993) A 5-4 pyrimidine-
pyrimidone photoproduct produced from mixtures of thymine and
4-thiouridine irradiated with 334 nm light. Photochem. Photobiol.
57, 255–265.
26. Clivio, P., D. Guillaume, M. T. Adeline, J. Hamon, C. Riche and
J.-L. Fourrey (1998) Synthesis and photochemical behavior of
peptide nucleic acid dimers and analogues containing 4-thiothy-
REFERENCES
1. Beukers, R. and W. Berends (1960) Isolation and identification of
the irradiation product of thymine. Biochim. Biophys. Acta 41,
550–551.

Photochemistry and Photobiology, 2011, 87 817
mine: Unprecedented (5-4) photoadduct reversion. J. Am. Chem.
Soc. 120, 1157–1166.
27. Skalski, B., K. Taras-Goslinska, A. Dembska, Z. Gdaniec and
S. Franzen (2010) Photoinduced fluorescent cross-linking of
5-chloro- and 5-fluoro-4-thiouridines with thymidine. J. Org.
Chem. 75, 621–626.
28. Varghese, A. J. (1970) Photochemistry of thymidine in ice. Bio-
chemistry 9, 4781–4787.
29. Shaw, A. A. (1987) A study of the direct effects of ionising and far
ultraviolet radiation on nucleic acids. Ph.D. thesis, University of
Surrey.
37. Shetlar, M. D., V. J. Basus, A. M. Falick and A. Mujeeb (2004)
The cyclobutane dimers of 5-methylcytosine and their deamina-
tion products. Photochem. Photobiol. Sci. 10, 968–979.
38. Taylor, J.-S., H.-F. Lu and J. J. Kotyk (1990) Quantitative con-
version of the (6-4) photoproduct of TpdC to its Dewar valence
isomer upon exposure to simulated sunlight. Photochem. Photo-
biol. 51, 161–167.
39. Douki, T., L. Voituriez and J. Cadet (1993) Characterization of
the (6-4) photoproduct of 2¢-deoxycytidylyl(3¢->5¢)-thymidine
and of its Dewar valence isomer. Photochem. Photobiol. 53, 293–
297.
30. Douki, T., M. Court and J. Cadet (2000) Electrospray–mass
spectrometry characterization and measurement of far-UV-
induced thymine photoproducts. J. Photochem. Photobiol. B, Biol.
54, 145–154.
31. Stepien, E., R. Lisewski and K. L. Wierzchowski (1973) Cyclob-
utane dimers of 1-methylthymine: Isolation, identification and
properties. Acta Biochim. Pol. 20, 297–311.
40. Lemaire, D. E. and B. P. Ruzsicska (1993) Quantum yields and
secondary photoreactions of the photoproducts of dTpdT, dTpdC
and dTpdU. Photochem. Photobiol. 57, 755–757.
41. Douki, T. and J. Cadet (1994) Formation of cyclobutane dimers
and (6-4) photoproducts upon far-UV photolysis of 5-methylcy-
tosine-containing dinucleoside monophosphates. Biochemistry 33,
11942–11950.
32. Varghese, A. J. (1971) Photochemistry of thymidine as a thin solid
film. Photochem. Photobiol. 13, 357–364.
42. Celewicz, L., M. Mayer and M. D. Shetlar (2005) The photo-
chemistry of thymidylyl-(3¢-5¢)-5-methyl-2¢-deoxycytidine in
aqueous solution. Photochem. Photobiol. 81, 414–418.
43. Laland, S. G. and G. Serck-Hanssen (1964) Synthesis of pyrimi-
din-2-one deoxyribosides and their ability to support the growth
of the deoxyriboside-requiring organism Lactobacillus acidophilus
R26. Biochem. J. 90, 76–81.
44. Douki, T., G. Laporte and J. Cadet (2003) Inter-strand photo-
products are produced in high yield with A-DNA exposed to UVC
radiation. Nucleic Acids Res. 31, 3134–3142.
33. Bose, S., S. Kumar, R. J. H. Davies, S. K. Sethi and J. A.
McCloskey (1984) The photochemistry of d(T-A) in aqueous
solution and in ice. Nucleic Acids Res. 12, 7929–7947.
34. Chambers, C. C., G. D. Hawkins, C. J. Cramer and D. G. Truhlar
(1996) Model for aqueous solvation based on Class IV atomic
charges and first solvation shell effects. J. Phys. Chem. 100, 16385–
16398.
35. Karle, I. L. (1969) Crystal structure of a thymine–thymine adduct
from irradiated thymine. Acta Crystallogr. Sect. B 25, 2119–2126.
36. Blackburn, G. M. and R. J. H. Davies (1966) The photochemistry
of 1-methylthymine and 3-methylthymine. J. Chem. Soc. C, 1342–
1345.
45. Su, D. G., H. Fang, M. Gross and J.-S. A. Taylor (2009) Photo-
crosslinking of human telomeric G-quadruplex loops by anti
cyclobutane thymine dimer formation. Proc. Natl Acad. Sci. USA
106, 12861–12866.