Mechanism and Kinetics of Photoisomerization of a Cyclic Disulfide, trans-4,5-Dihydroxy-1,2-dithiacyclohexane

Article abstract of DOI:10.1021/jp037598s

The photolysis of trans-4,5-dihydroxy-1,2-ditniacyclohexane in aqueous and CH₂Cl₂ solution yields two isomers of 2,3-dihydroxy-l- mercaptotetrahydrothiophene, characterized by ¹H and ¹³C NMR and negative ion electrospray mass spectrometry. Product formation in water is independent of the presence of oxygen and the concentration of trans-4,5-dihydroxy-1,2-dithia-cyclohexane and involves intramolecular 1,5-H-transfer, followed by cyclization through thiophilic or nucleophilic addition, or 1,2-H- transfer, followed by cyclization through the recombination of a sulfur- and a carbon-centered radical. In contrast, the quantum yields for reaction in CH₂Cl₂ solutions are dependent on both oxygen concentration and on the concentration of trans-4,5-dihydroxy-1,2- dithiacyclohexane. The latter results are consistent with an intermolecular H-transfer between an initial dithiyl diradical and trans-4,5-dihydroxy-1,2- dithia-cyclohexane. Time-resolved laser flash photolysis studies indicate rapid product formation on the submicrosecond time scale and support the intermediacy of α-mercaptoalkyl radicals.

Full text of DOI:10.1021/jp037598s

J. Phys. Chem. A 2004, 108, 2247-2255
2247
Mechanism and Kinetics of Photoisomerization of a Cyclic Disulfide,
trans-4,5-Dihydroxy-1,2-dithiacyclohexane
Lorena B. Barro´n,† Kenneth C. Waterman,‡ Piotr Filipiak,§ Gordon L. Hug,§
Thomas Nauser,† and Christian Scho1neich*,†
Department of Pharmaceutical Chemistry, 2095 Constant AVenue, UniVersity of Kansas,
Lawrence, Kansas 66047, Pfizer Global Research and DeVelopment, Eastern Point Road, Groton,
Connecticut 06340, and Radiation Laboratory, UniVersity of Notre Dame, Notre Dame, Indiana 46556
ReceiVed: NoVember 25, 2003; In Final Form: January 30, 2004
The photolysis of trans-4,5-dihydroxy-1,2-dithiacyclohexane in aqueous and CH2Cl2 solution yields two isomers
of 2,3-dihydroxy-1-mercaptotetrahydrothiophene, characterized by 1H and 13C NMR and negative ion
electrospray mass spectrometry. Product formation in water is independent of the presence of oxygen and the
concentration of trans-4,5-dihydroxy-1,2-dithia-cyclohexane and involves intramolecular 1,5-H-transfer,
followed by cyclization through thiophilic or nucleophilic addition, or 1,2-H-transfer, followed by cyclization
through the recombination of a sulfur- and a carbon-centered radical. In contrast, the quantum yields for
reaction in CH2Cl2 solutions are dependent on both oxygen concentration and on the concentration of trans-
4,5-dihydroxy-1,2-dithiacyclohexane. The latter results are consistent with an intermolecular H-transfer between
an initial dithiyl diradical and trans-4,5-dihydroxy-1,2-dithia-cyclohexane. Time-resolved laser flash photolysis
studies indicate rapid product formation on the submicrosecond time scale and support the intermediacy of
R-mercaptoalkyl radicals.
Introduction
such as disproportionation, intra- and intermolecular hydrogen
transfer, and radical-radical recombination.
The disulfide bond, R-S-S-R, represents an important
structural element in proteins.1 Moreover, disulfide entities are
part of polysulfide systems, which are found in an ever-
increasing list of natural products with promising therapeutic
properties.2-5 Aliphatic disulfides generally show absorption
spectra with λmax < 300 nm, depending on the dihedral angle
of the C-S-S-C system.6 Depending on the nature of R, the
exposure of aliphatic disulfides to UV light results in homolytic
cleavage of either the S-S bond (reaction 1), generating a pair
of thiyl radicals (RS•), or the C-S bond (reaction 2), generating
a carbon-centered radical and a perthiyl radical (RSS•).7,8
RSSR f 2 RS•
(1)
(2)
A pair of thiyl radicals generated from a high molecular
weight protein disulfide should remain longer in a solvent cage
than an analogous pair of thiyl radicals from compounds of
lower molecular weight. In proteins, the separation of the radical
pair will depend on the stability of the secondary and tertiary
structure, the kinetics of unfolding, the conformational dynamics
of the specific protein domain surrounding the radical pair, and
the diffusion coefficient(s) of the polypeptide chain(s). Hence,
under such conditions the relative efficiencies of the different
reaction channels may be significantly different than with low
molecular weight disulfides, though no systematic studies of
these reactions in proteins have been undertaken.
Two objectives motivated the present study. First, we believe
that the photolysis of a cyclic disulfide would mimic the
situation of a thiyl radical pair within a protein, with restricted
mobility. Second, we wanted to establish the photochemical
reactions of a cyclic disulfide, which could potentially be used
as an initiator for the controlled photochemical initiation of
radical reactions in the solid state (solid-state hydrogen-transfer
reactions are of fundamental importance10-13 and may be
initiated through the controlled generation of thiyl radicals).
RSSR f RSS• + R•
These geminate pairs of product radicals are generated within
a solvent cage. They enter various reaction channels, depending
on the electronic configuration of the radical pair in the solvent
cage (viz., singlet vs triplet), the rate of escape from the solvent
cage, and the presence of oxygen and additional electron or
hydrogen donors. For example, an important reaction of thiyl
radicals is the reversible addition of oxygen, yielding thiylper-
oxyl radicals (RSOO•) (reaction 3) which can subsequently
rearrange irreversibly to sulfonyl radicals (reaction 4) (for
2-mercaptoethanol in aqueous solution, k3 ) 2.2 × 109 M-1
s-1, k-3 ) 6.2 × 105 s-1, and k4 ) 2 × 103 s-1).9 These reactions
are in competition with various other potential reaction pathways
* Corresponding author. Phone: (785) 864-4880. Fax: (785) 864-5736.
E-mail: schoneic@ukans.edu.
† University of Kansas.
‡ Pfizer Global Research and Development.
§ University of Notre Dame.
10.1021/jp037598s CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/03/2004

2248 J. Phys. Chem. A, Vol. 108, No. 12, 2004
Barro´n et al.
SCHEME 1
Therefore, we selected to study the photolysis and mechanisms
of product formation for the cyclic organic disulfide trans-4,5-
dihydroxy-1,2-dithiacyclohexane (DTTox) (structure 1; Scheme
1). The photochemically generated geminate radical pair (here,
a diradical; structure 2) will remain in close contact as both
radical sites are connected by an aliphatic linker. We will show
that photolysis results predominantly in the isomerization to the
novel structure 5, involving an initial pair of thiyl radicals which
react via a 1,2- and/or 1,5-H-shift, followed by bond formation
between a thiyl and a carbon-centered radical and/or nucleophilic
cyclization, respectively. The presence of oxygen does not affect
the product yields in the protic solvent water, whereas it does
so in the nonprotic solvent dichloromethane (CH2Cl2). The latter
result suggests a solvent dependence of the H-shift for thiyl
radicals, comparable to analogous features of alkoxy radicals
(RO•).14
of an arc lamp power supply (Model 68806; Oriel Instruments,
Stratford, CT), a convective lamp housing (Oriel Instruments)
equipped with a 100 W (ozone free) xenon lamp, a monochro-
mator (Model 77250; Oriel Instruments), and an integrating
sphere (Labsphere, Inc., North Sutton, NH). The light beam was
guided into the monochromator with a fused silica plano convex
lens (Oriel Instruments) and from the monochromator into the
integrating sphere via a biconvex fused silica lens (Oriel
instruments). From the integrating sphere, the reflected light
was collected and quantified by a labsphere integrating sphere
system control (SC-5500; Labsphere, Inc.). Data acquisition was
performed over an RS232 interface using software written in
Delphi (Borland, Scotts Valley, CA). The photolysis sample
was placed in a specifically adapted quartz cuvette into the
integrating sphere, and the monochromator wavelength was set
to 297 ( 2 nm (slit width: 280 µm). All samples were prepared
in air- or N2-saturated deionized water or organic solvents of
highest commercially available purity. Usually, photolysis times
in the Rayonet photoreactor did not exceed 15 min, whereas
comparable photolytic yields at a single wavelength in the
integrating sphere system required photolysis times of up to 24
h.
Experimental Section
Materials.
trans-4,5-Dihydroxy-1,2-dithiacyclohexane
(DTTox; 1) was obtained from Sigma-Aldrich Chemical Co.
(St. Louis, MO) and used as received. Methylene chloride
(CH2Cl2) and HPLC-grade acetonitrile (CH3CN) were purchased
from Fisher Scientific (Pittsburgh, PA) and used as received.
Water was distilled by a Labconco purification system. D2O
was received from Cambridge Isotope Laboratories, Inc.
(Andover, MA), and 3H-naphtho[2,1-b]pyran-2-carboxylic acid,
10-(2,5-dihydro-2,5-dioxo-1H-pyrrol-1-yl)-9-methoxy-3-oxo-,
methyl ester (ThioGlo-1) was supplied by Covalent Associates
(Woburn, MA).
Photochemical Reactions. Photochemical reactions were
carried out either in a Rayonet photoreactor (Southern New
England Ultraviolet Company, Branford, CT) or in a custom-
built photolysis system. The Rayonet photoreactor was equipped
with up to eight RPR-3500 lamps, which emit light between
305 and 410 nm, overlapping with the low-energy part of the
UV absorbance of DTTox (absorbance ca. 250-325 nm; λmax
) 280 nm). The custom-built photolysis system was composed
Quantum Yields. Quantum yields were determined with the
integrating sphere system. To calibrate the integrating sphere
system, chemical actinometry was carried out using the ferric
oxalate actinometer.15 A 900 µL aliquot of the actinometry
solution was placed in a quartz cuvette that had been modified
to fit the integrating sphere. The actinometry solution was
irradiated at 297 nm at various times. The reflected light was
collected as current by the SC-5500 integrating sphere system
control. After irradiation, the solution was vortexed and an 800
µL aliquot of the irradiated solution was transferred to a 10
mL flask. A 4 mL aliquot of 0.2% 1,10-phenanthroline solution
and 400 µL of 0.6 M sodium acetate in 1% H2SO4 were added.
The solution was mixed and allowed to equilibrate for 30 min
in the dark at room temperature. After incubation, the absorbance
at 510 nm was measured. By knowing the amount of photons

Photoisomerization of a Cyclic Disulfide
J. Phys. Chem. A, Vol. 108, No. 12, 2004 2249
used, a correlation was achieved between the actual current and
the photons provided by the light source.
HPLC Analysis. Product separation was carried out on a
Hewlett-Packard 1050 instrument equipped with a UV detector.
A 100 µL aliquot of diluted sample was injected using an
Agilent Technologies 1100 autosampler onto an HPLC column
(Phenomenex ODS, 250 × 4.5 mm) and eluted using a linear
acetonitrile/trifluoroacetic acid gradient with UV detection at
214 nm. The mobile phase started with 5% mobile phase B
[0.1% (v/v) trifluoroacetic acid/90:10% (v/v) acetonitrile/water]
at 0 min and increased linearly to 30% mobile phase B within
30 min at a flow rate of 0.8 mL/min. Products were isolated
and lyophilized for characterization.
Quantification of H2O2. Hydrogen peroxide was measured
spectrophotometrically by detecting its titanium sulfate complex
at 410 nm, using ꢀ410 ) 700 L mol-1 cm-1 16
. Solutions of 1.6
× 10-2 M DTTox were prepared in water and methylene chloride
and irradiated for 10 min using the Rayonet photoreactor
equipped with eight RPR-3500 lamps. Aqueous solutions were
immediately incubated with titanium sulfate and analyzed by
UV spectrophotometry. For the determination of H2O2 in CH2-
Cl2, the organic solvent was extracted with water before
incubation with titanium sulfate and analysis by UV spectro-
photometry.
Figure 1. HPLC analyses of control (upper panel) and photolyzed
(lower panel) DTTox. For experimental details, see the Experimental
Section.
current YAG-based system and that previously reported was
the use of a Digikrom 240 monochromator. It, along with the
other peripheral instrumentation, is under computer control (PC
with a Pentium II processor). The monitor light source was a 1
kW pulsed xenon lamp with an optical path length of 0.5 cm
through a rectangular quartz (optically flat) cell. The optical
path length of the laser through the sample was 1 cm, and the
laser beam was perpendicular to the monitoring beam. The
optical cell was part of a gravity-driven flow system.
ESI-Q-TOF MS/MS. Isolated products were lyophilized and
reconstituted in 80% methanol containing 5 mM ammonium
hydroxide for negative ion electrospray ionization (ESI) MS/
MS analysis. The samples were infused directly into the ESI
source. ESI-MS/MS spectra were acquired on a Q-TOF-2
(Micromass Ltd. Manchester U.K.) hybrid mass spectrometer
operated in the negative MS mode with data acquisition by the
time-of-flight (TOF) analyzer. The instrument was operated at
isotopic resolution.
Results
Product Characterization. Figure 1 shows representative
chromatograms, recorded at 214 nm, before and after a 10 min
photolysis of 1.6 × 10-2 M DTTox in air-saturated water. Peaks
I and II identify two major reaction products which elute with
tR ) 9.1 and 10.2 min, respectively. When both products were
monitored with a diode array detector, the corresponding UV
spectra show an absorbance maximum at λ max ≈ 245 nm with
no absorbance at λ > 270 nm (data not shown).
1
NMR Spectroscopy. H, COSY (correlation spectroscopy),
13C, and HMQC (heteronuclear multiple-quantum coherence)
measurements were performed on a 500 MHz Bruker Avance
spectrometer. Isolated products and reagent controls were
analyzed in deuterium oxide (D2O).
FT-IR Spectroscopy. Isolated products were lyophilized and
reconstituted with 100 µL of ethanol. Aliquots (10 µL) were
placed on disposable polyethylene IR cards, and the solvent was
evaporated. The spectra were recorded on a Nicolet Magna-IR
560 instrument using the OMNIC software.
Fluorescence Spectroscopy. Isolated products at a concen-
tration of 3.8 µM were incubated for 30 min with 150 µM
ThioGlo-1 at room temperature. In order to quantitate the
amount of free thiols labeled with ThioGlo-1, a standard curve
of 0-8 µM N-Ac-Cys was prepared. The excitation wavelength
was 360 nm, and the emission wavelength was 530 nm.
Fluorescence yields were measured with a BIO-TEK FL600
microplate fluorescence reader.
Laser Flash Photolysis. The 266 nm excitation source used
in the nanosecond flash photolysis experiments was a Quanta-
Ray Pro 230 pulsed Nd:YAG laser, operated at 10 Hz. An
optical attenuator (variable, Newport Model 935-10) was placed
in the laser path in order to reduce the dose delivered to the
sample cell to 10-12 mJ/10 ns pulse. The detection system
used was similar to one described previously17 with the
following modifications: The present work used a LeCroy
LC574A storage oscilloscope (1 GHz). It has a long word length
(500 000 points when run as a single channel) that allowed
multiple time scales to be saved for each kinetic trace as
described elsewhere.17 The shortest time scale with the LC574A
was 0.25 ns/point. One other material difference between the
Products I and II were subjected to negative ion ESI-MS/
MS analysis, yielding identical molecular ions [M-H]- with
m/z 151 (molecular weight of 152). This molecular weight is
identical to that of DTTox, suggesting that I and II are
isomerization products. Further evidence that the products are
isomers of DTTox was provided by MS/MS experiments, which
yield two dominant fragment ions with m/z 75 and m/z 117,
corresponding to the loss of CS2 and the loss of H2S,
respectively. More structural information was gained by H/D
exchange experiments. Products I and II were isolated by
reversed-phase HPLC, lyophilized, and dissolved into a mixture
of D2O and CD3OD (8:2, v/v) before negative ion ESI-MS/MS
analysis. The resulting molecular ions for both I and II showed
m/z ) 153, indicating three exchangeable protons on each
product. The two mass unit increase occurs because loss of a
deuteron is expected in the negative ion electron spray process,
while the other two exchanged deuterons remain. Consistent
with this, the same H/D exchange method shows that for the
reactant DTTox, its two exchangeable protons result in a mass
increase of one (m/z ) 152).
The structures of products I and II were elucidated using 1H,
13C, COSY, and HMQC NMR spectroscopy. As a control, both
DTTox and its reduced, open chain product dithiothreitol (DTT;
1
structure 6 in Scheme 1) were analyzed as well. For DTT, H

2250 J. Phys. Chem. A, Vol. 108, No. 12, 2004
Barro´n et al.
NMR analysis indicates six nonexchangeable protons. The
methylene protons show a chemical shift of 2.6 ppm, whereas
the methine protons display a chemical shift of 3.6 ppm. For
DTTox, the methylene protons show a chemical shift of 2.8 and
3.1 ppm, whereas the methine protons are detected at 3.48 ppm.
Both control samples, DTTox and DTT, show no protons with
chemical shifts higher than 3.7 ppm.
In product I, five nonexchangeable protons are detected, with
three of them showing chemical shifts above 3.7 ppm (Figures
S1-S3, Supporting Information). Two signals detected at 2.6
ppm (dd, Jvic ) 3.9 Hz, Jgem ) 10.2 Hz, 1H) and 3.25 ppm (dd,
Jvic ) 5 Hz, Jgem ) 11.8 Hz, 1H) likely correspond to a set of
two methylene protons. The other three protons are detected
with chemical shifts of 3.95 ppm (m, 1H), 4.35 ppm (m, 1H),
and 4.6 ppm (d, J ) 4.2 Hz, 1H).
Product II shows a similar picture with five nonexchangeable
protons (Figures S4-S6, Supporting Information). Again, a set
of two methylene protons is detected at 2.85 ppm (dd, Jvic
)
7.2 Hz, Jgem ) 11.1 Hz, 1H) and 3.0 ppm (dd, Jvic ) 6.2 Hz,
Jgem ) 11.1 Hz, 1H), and the three additional protons have
chemical shifts of 3.85 ppm (t, J ) 7.2 Hz, 1H), 4.08 ppm (m,
1H), and 4.14 ppm (d, J ) 6.7 Hz, 1H).
Figure 2. Representative COSY NMR of product I (expanded region).
Important connectivities are indicated by the arrows. A summary of
the connectivities is given in Table 1. The resonance at ca. 3.15 ppm
is caused by a degradation product of I during the time required for
recording of the COSY NMR.
The 13C NMR analysis shows four distinct carbon atoms for
both product I (detected at 36, 47, 76, and 79 ppm) and product
II (33, 46, 76, and 85 ppm).
TABLE 1: Connectivities of Products I and II as
Determined by COSY NMR
An HMQC experiment was performed to establish the
correlations between the proton and 13C signals. In both products
I (Figure S7, Supporting Information) and II (Figure S8,
Supporting Information), the protons detected below 3.3 ppm
are attached to a carbon in the 30-40 ppm region. Comparison
with authentic DTTox and DTT confirm the presence of one
methylene group. In product I, the protons at 3.95 and 4.35
ppm correlate with the carbons at 79 and 76 ppm, respectively.
This confirms the presence of two methine protons, both in close
proximity to a hydroxyl group. For product II, a similar
correlation was observed between the protons at 3.85 and 4.05
ppm and the carbons at 85 and 76 ppm. In both products I and
II, the remaining proton correlates with a carbon, which is
attached to two sulfur atoms. In order to confirm the con-
nectivities in both products, a COSY experiment was performed,
which is displayed in Figure 2. The respective connectivities
are summarized in Table 1.
Based on the data provided by negative ion ESI-MS/MS
and NMR analysis, we propose that products I and II represent
two epimers of 2,3-dihydroxy-1-mercaptotetrahydrothiophene
(structure 5, Scheme 1). The exact stereochemical configurations
of the epimers cannot be assigned based on the NMR data.
Further evidence for structure 5 was provided by specific
labeling of the free mercapto group with 3H-naphtho[2,1-b]-
pyran-2-carboxylic acid, 10-(2,5-dihydro-2,5-dioxo-1H-pyrrol-
1-yl)-9-methoxy-3-oxo-, methyl ester (ThioGlo-1). ThioGlo-1
is a maleimide-based label, which fluoresces (λexc ) 379 nm;
λem ) 510 nm) after Michael addition of a thiol to the maleimide
moiety.18 By comparison to known concentrations of a standard
thiol, N-acetylcysteine, ThioGlo-1 labeling of products I and II
indicates a stoichiometry of one free mercapto group per
molecule of I and II (for determinations of the concentrations
of I and II, see below: Product Quantification).
product I
connectivity
product II
connectivity
H (ppm)
2.6
H (ppm)
2.85
3.25
4.35
3
4.05
3.25
3.95
4.35
2.6
4.35
3
2.65
4.05
4.35
4.6
3.85
4.05
4.05
4.15
2.6
3.25
3.95
2.6
3
3.85
4.6
3.95
4.15
3.85
scopic analyses did not show any signal in the 1020-1240 cm-1
region, where the >CdS group would absorb.19,20 When
products I and II were monitored with a diode array detector
coupled on-line to the HPLC system, we did not detect any
significant absorbance around 530 nm, which would be expected
for a thioaldehyde.19
Importantly, HPLC (as well as MS/MS and one- and two-
dimensional NMR analysis) did not indicate significant yields
of any other reaction product besides products I and II
immediately after photolysis. However, upon prolonged storage
these products I and II proved unstable, with the appearance of
degradation products. These degradation products have not been
characterized further. We specifically note that the photolysis
of DTTox did not lead to measurable yields of DTT (6).
Product Quantification. The response factors for products
I and II during HPLC analysis with 214 nm UV detection were
1
determined by H NMR experiments, comparing the absolute
yields of protons in products I and II to the quantity of lost
protons through consumption of DTTox. An aqueous (D2O)
solution containing 1.6 × 10-2 M DTTox and 0.01% (v/v) tert-
butyl alcohol as an internal standard was prepared as a control
sample, and the relative NMR intensities of all protons
measured. An air-saturated aqueous (D2O) solution of 1.6 ×
10-2 M DTTox was photolyzed for 10 min prior to the addition
of a final content of 0.01% (v/v) tert-butyl alcohol. NMR
We note that the data from MS/MS and ThioGlo-1 labeling
also support the thioaldehyde shown in structure 4 (Scheme 1).
However, 13C NMR, FTIR and UV-vis spectroscopic analysis
show no evidence for the presence of a thioaldehyde: we
observed no 13C resonances around 250 ppm, where the
thiocarbonyl carbon atom would be expected;19 FTIR spectro-

Photoisomerization of a Cyclic Disulfide
J. Phys. Chem. A, Vol. 108, No. 12, 2004 2251
measurements were made of all protons from DTTox, product
I, and product II relative to those of the internal standard tert-
butyl alcohol. The total integrated proton signal lost as a result
of DTTox consumption corresponded quantitatively to that
recovered in products I and II. Moreover, the relative distribution
of NMR peak intensity between the remaining DTTox and the
products I and II correlated with the relative peak intensities of
these products during HPLC analysis. Hence, the response
factors of DTTox and products I and II during HPLC analysis
with 214 nm UV detection are identical within experimental
error. On the basis of this analysis, the photolysis of 1 generates
products I and II at a ratio of 2:1.
Quantum Yields and the Effects of Solvent Nature and
Oxygen. By use of an integrating sphere, quantum yields for
DTTox photolysis in water were determined through equations
a and b, where Pabs is the number of photons absorbed, Pc is
the number of photons reflected from a control sample lacking
DTTox, and Ps is the amount of photons reflected from a sample
containing DTTox.
Φ ) moles of product/moles of absorbed photons (a)
Pabs ) Pc - Ps
(b)
The moles of product produced were calculated from the peak
areas of products I and II recorded during HPLC analysis (vide
supra). For air-saturated aqueous solutions, we obtained Φ )
1.04 ( 0.15 independent of the concentration of DTTox between
0.33 × 10-3 and 1.6 × 10-2 M. These experiments were
performed with the aqueous solution open to air to ensure air
saturation during the photolysis. For CH2Cl2, such experiments
were not feasible because of the evaporation of the solvent
during a 24 h photolysis. Therefore, the following comparison
between aqueous and CH2Cl2 solutions was made based on short
periods of irradiation in the Rayonet photoreactor, taking into
account a slightly higher absorptivity of DTTox in CH2Cl2
compared to that with H2O (vide infra):
The photolytic yields of product I and II were independent
of oxygen concentration (air- vs N2-saturated solutions) in water
but showed oxygen dependence in CH2Cl2. A 10 min light
exposure in the Rayonet photoreactor of 1.6 × 10-2 M DTTox
in air-saturated solvent yielded product I at (7.2 ( 0.3) × 10-3
M in water and (12 ( 0.6) × 10-3 M in CH2Cl2. The relative
ratio, [12 × 10-3 M]/[7.2 × 10-3 M] ) 1.7, of these product
yields in CH2Cl2 versus H2O is rationalized by the 1.7-fold
higher absorptivity of DTTox in CH2Cl2 compared to that in
H2O in the spectral region overlapping with the emission
spectrum of the RPR-3500 lamps. Based on the combined yields
of products I and II, we calculate that for 1.6 × 10-2 M DTTox
in CH2Cl2, Φ ) 1.0. Similarly, a 10 min Rayonet photoreactor
Figure 3. LFP of Ar-saturated aqueous solutions of 1.6 × 10-3
M
DTTox in water. (A) Absorbance spectra recorded within specific periods
after the laser flash. (B) Absorption vs time trace recorded at 280 nm.
The solid line represents a first-order fit of the experimental data. For
experimental details, see the Experimental Section.
this solvent. No H2O2 was detected during photolysis of air-
saturated solutions of DTTox in CH2Cl2 (<10 µM, based on
the LOD from calibration curves).
Laser Flash Photolysis. Laser flash photolysis (LFP) experi-
ments were performed in order to characterize the primary
species generated during DTTox photolysis. The optical spectra
recorded at various time points after 266 nm LFP of 1.6 mM
DTTox in Ar-saturated water and CH2Cl2 are displayed in
Figures 3A and 4A, respectively. A comparison of both figures
reveals some differences between the two solvents, as expected
based on the product studies in both solvents presented above.
In water (Figure 3A), the spectrum recorded between 400 and
600 ns after the flash is characterized by a strong absorbance
at λ < 290 nm, some photobleaching between 290 and 320 nm,
and a weak structureless absorbance between 340 and 460 nm.
The absorbance characteristics between 280 and 320 nm likely
reflect the superposition of two processes, photobleaching of
exposure of a N2-saturated aqueous solution of 1.6 × 10-2
M
DTTox, yielded (7.5 ( 0.4) × 10-3 M of product I indicating
no oxygen dependence on the reaction yield in water. In contrast,
in N2-saturated CH2Cl2 (1.6 × 10-2 M DTTox), photolysis yields
dropped by about 33% to (8.0 ( 0.5) × 10-3 M. Based on the
combined yields of products I and II in N2-saturated CH2Cl2,
we calculate Φ ) 0.66.
At lower concentrations of DTTox (0.33 × 10-2 M), Φ
remained equal to 1.0 in water for both air- (vide supra) and
N2-saturated solutions. In CH2Cl2, the quantum yield dropped
to Φ ) 0.47 in air-saturated solution and to Φ ) 0.07 in N2-
saturated solution. Thus, the photolytic formation of products I
and II is independent of both DTTox and O2 concentration in
water but highly dependent on both terms in CH2Cl2.
Motivated by the oxygen dependence of product formation
in CH2Cl2, we monitored the photolytic formation of H2O2 in
the ground-state absorbance of DTTox (λ
formation of a transient Y with λ max,Y < 320 nm and ꢀ280,Y
) 280 nm) and
max
>
ꢀ280,DTTox. Optical spectra recorded at longer times after LFP
demonstrate that, ultimately, transient Y decomposes into
product(s) with lower extinction coefficients at 280 nm com-
pared to those of ground-state DTTox. This result is consistent
with the steady-state photolysis experiments, which show that
products I and II have no UV absorbance at λ > 270 nm (vide
supra). Figure 3B shows the respective absorption-versus-time

2252 J. Phys. Chem. A, Vol. 108, No. 12, 2004
Barro´n et al.
profile at 280 nm after LFP of DTTox in Ar-saturated H2O.
Transient Y disappears in a first-order process (k ) 9.1 × 104
s-1), which ultimately results in photobleaching at this wave-
length.
Usually, thiyl radicals show weak absorption bands with λmax
) 330 nm (ꢀ330 < 600 M-1 cm-1, except for penicillamine thiyl
radicals) and no significant absorbance between 270 and 300
nm.21 Hence, it is unlikely that transient Y is a thiyl radical.
The spectrum also displays at most minor yields of perthiyl
radicals, which are characterized by a distinct strong absorbance
with λmax ≈ 370 nm (ꢀ
≈ 1700 M-1 cm-1).7,22-24 We can
370
also exclude the formation of significant photolytic yields of
disulfide radical anions from DTTox (RSSR•-) which display a
strong absorbance band with λmax(RSSR•-) ) 390 nm (ꢀ390
)
6600 M-1 cm-1).25 At the same time, no radical cations of
DTTox can be observed, which by analogy to several aliphatic
disulfides, would absorb with λmax(RSSR•+) ≈ 410-440 nm (ꢀ
≈ 2000 M-1 cm-1).26 The spectral data in water do not permit
the identification of Y. However, by analogy with the spectral
characteristics of R-alkylthioalkyl radicals (R-C•H-S-R′),
which absorb with λmax ) 280-290 nm (ꢀ ≈ 3000 M-1 cm-1)27,
the data would be consistent with the intermediacy of an
R-mercaptoalkyl radical (R-C•H-SH) (vide infra). We do not
believe that the observed transient represents the triplet state of
the disulfide for the following reason: transient Y decomposes
with k ) 9.1 × 104 s-1. Our steady-state photolysis experiments
demonstrate no effect of triplet oxygen on product yields in
water. Usually, the reaction of excited triplet states with oxygen
proceeds with k ≈ 109 M-1 s-1 28,29
Based on [O2] ) 2.5 ×
.
10-4 M in air-saturated water, we calculate a pseudo-first-order
rate constant for the potential reaction of any triplet excited state
with oxygen in water of k ≈ 2.5 × 105 s-1. Therefore, if the
290 nm transient were the triplet excited state, we would have
expected significantly lower yields of products I and II in air-
versus N2-saturated solution.
The optical spectrum recorded between 100 and 200 ns after
LFP of DTTox in CH2Cl2 (Figure 4A) shows an intense
absorbance at λ < 325 nm, displaying a shoulder at 290 nm.
Within 3.5 µs after LFP, this shoulder disappears, leaving an
optical spectrum which is characterized by a structureless
absorption band increasing toward the UV, and remains rather
stable for 160 µs (an absorption-vs-time profile, recorded at 275
nm is shown in Figure 4B). Importantly, also in CH2Cl2 no
significant yields of perthiyl radicals, disulfide radical anions,
and disulfide radical cations are present at any time after the
laser flash. Figure 4C shows a difference spectrum obtained by
subtraction of the spectrum obtained between 2.5 and 3.5 µs
after the pulse from the spectrum obtained between 100 and
200 ns after the pulse. This computation yields a better picture
of the 280 nm absorbance, which is immediately present after
the laser flash. Again, this 280 nm species could indicate the
presence of an R-mercaptoalkyl radical and/or a derivative
therefrom (vide infra).
Figure 4. LFP of Ar-saturated solutions of 1.6 × 10-3 M DTTox in
CH2Cl2. (A) Absorbance spectra recorded within specific periods after
the laser flash. (B) Absorption vs time trace recorded at 275 nm. (C)
Difference spectrum computed by subtraction of the spectrum recorded
between 2.5 and 3.5 µs from the spectrum recorded between 100 and
200 ns after the laser flash. For experimental details, see the
Experimental Section.
Discussion
A mechanistic discussion of our results needs to take into
account the following key observations: (i) DTTox undergoes
photoisomerization into two epimeric 2,3-dihydroxy-1-mercap-
totetrahydrothiophenes. (ii) Based on LFP experiments, we can
discard the intermediacy of perthiyl radicals, disulfide radical
cations, and disulfide radical anions. (iii) LFP experiments are
consistent with the formation of R-mercaptoalkyl radicals,
especially in H2O. (iv) LFP and steady-state photolysis experi-
ments in CH2Cl2 further argue against photoionization of DTTox
as the resulting solvated electrons would react instantaneously
with CH2Cl2. In this case, only disulfide radical cations would
remain as potential precursors for products I and II. (v) There
is a 50% higher yield of products I and II upon photolysis of
DTTox in air-saturated compared to N2-saturated CH2Cl2. (vi)
There is no difference in product yields between air- and N2-
saturated solutions of DTTox in H2O. (vii) No other reaction
products are observed in either solvent, with or without oxygen.

Photoisomerization of a Cyclic Disulfide
J. Phys. Chem. A, Vol. 108, No. 12, 2004 2253
SCHEME 2
A mechanism accounting for all these observations in H2O
is displayed in Scheme 1. The disulfide bridge of DTTox in water
a nearly exclusive nucleophilic attack of the mercapto group
on the thiocarbonyl carbon (reaction 9). A possible rationale
for that could be a kinetic preference of pentacyclization
(reaction 9) versus hexacyclization (reaction 8). On the other
hand, high quantum yields of 5 would be well explained by a
solvent-assisted 1,2-H-shift (reaction 10), followed by carbon-
sulfur bond formation (reaction 11). This mechanism would
require that k10 . k7, which is contrary to the theoretical
prediction of activation barriers for intramolecular H-shifts in
ground-state carbon-centered radicals (vide supra).33 However,
such differences in activation barriers may be less pronounced
if diradical formation and H-shift proceed in a concerted manner,
originating from the excited triplet state (see dashed arrow in
Scheme 1). Such a mechanism could also rationalize the rapid
formation of a species absorbing with λmax ) 290 nm (poten-
tially the R-mercaptoalkyl radical; vide supra) during LFP,
which is observable at times of <1 µs after the laser flash in
both solvents, H2O and CH2Cl2. Such rapid formation would
not be expected for ground-state thiyl radicals based on pulse
radiolysis studies of â-mercaptoethanol: here, a first-order rate
constant for the 1,2-H-shift of â-hydroxyalkylthiyl radicals,
shows a UV absorbance between 250 and 330 nm with λmax
)
280 nm. This red shift compared to linear aliphatic disulfides
(λmax ≈ 250 nm) is due to significant splitting of the sulfur lone
pair orbitals into a higher energy n- and a lower energy n+
orbital.30,31 This splitting is the result of a considerable deviation
of the C-S-S-C dihedral angle in the hexacyclic dithiane from
90°, a value generally adopted by sterically unhindered linear
aliphatic disulfides. The longer wavelength absorbance of DTTox
is then ascribed to the n- f σ* transition. Our LFP results and
product studies suggest that the electron is excited almost
exclusively into the σ* orbital of the S-S bond, resulting in
the homolytic cleavage of the S-S bond (reaction 5) (excitation
into the σ* orbital of the C-S bond would lead to homolytic
C-S bond cleavage, but perthiyl radicals were not detected by
LFP). This is consistent with earlier electron spin resonance
(ESR) studies on the photolysis of several cyclic dithianes,
including 1,2-dithiacyclohexane, which showed the formation
of dithiyl radicals.32 The resulting diradical could undergo ring
closure to regenerate DTTox (reaction 6); however, in the
electronic triplet state such ring closure will be comparatively
slow. Two alternative processes will be discussed, both of which
will lead to product 5, a 1,5-H-shift (reaction 7) followed by
reaction 9, and a 1,2-H-shift (reaction 10) followed by reaction
11 (it is important to note that, by analogy to alkoxyl radicals,14
the 1,2-H-shift is likely solvent assisted). Theoretical calculations
on H-transfer processes in alkyl radicals33 show that the 1,5-
H-shift is associated with a lower activation barrier compared
to the 1,2-H-shift, due to a lower ring strain in the cyclic
transition state. In fact, 1,5-H-shifts are synthetically utilized
organic reactions of free radicals.34-36 However, the 1,5-H-shift
would yield thiocarbonyl 4. By analogy to previous examples,37
the latter is expected to undergo thiophilic addition to regenerate
DTTox (reaction 8). Our results show that product 5 is formed
with quantum yields up to 1.0. Hence, if 4 were an intermediate
in the formation of 5, the ring closure of 4 would need to involve
HO-CH2-CH2-S•, in H2O was estimated to k ≈ 2 × 103 s-1 9
.
For CH2Cl2, the mechanism needs to be modified to account
for the effects of oxygen and DTTox concentration. We propose
the following reactions, which are displayed in Scheme 2.
There is clear evidence that 1,2-H-shifts of alkoxyl radicals
are solvent assisted,14 catalyzed by solvents which contain
hydroxyl groups and have nucleophilic properties. If similar
parameters would control 1,2-H-shifts in thiyl radicals, we would
expect a considerably less efficient 1,2-H-shift in CH2Cl2. As a
result, a fraction of diradical 2 may abstract a hydrogen atom
from excess substrate in a bimolecular reaction (reaction 12),
yielding the monothiyl radical 7 and the carbon-centered radical
8. The following reactions are hypothetical but are all formulated
by analogy to known processes of thiyl radicals and R-mercap-
toalkyl radicals.9 The monothiyl radical 7 could reversibly add
oxygen (reaction 13) and reversibly shift the hydrogen (reaction

2254 J. Phys. Chem. A, Vol. 108, No. 12, 2004
Barro´n et al.
14). An irreversible exit from these two equilibria would be
oxygen addition to the carbon-centered radical (reaction 15),
reactions in the solid state. In fact, our preliminary data show
that thiyl radicals of DTTox oxidize added mercaptanes in the
solid state.
•
followed by elimination of HO2 (reaction 16). However, no
•
dismutation of HO2 to H2O2 and O2 appears to be occurring
Acknowledgment. We are grateful to Pfizer, Inc. and NIH
CA072987 for financial support. The work described herein was
also supported by the Office of Basic Energy Sciences of the
U. S. Department of Energy. This paper is Document No.
NDRL-4485 from the Notre Dame Radiation Laboratory.
since significant yields of H2O2 in CH2Cl2 are not detected (vide
supra). The carbon-centered radical 8 could potentially undergo
ring opening (reaction 17), 1,2-H-shift (reaction 18), ring closure
•
(reaction 19), and ultimate reduction of radical 9 by HO2
(reaction 20). The mechanism in Scheme 2 satisfies the observed
dependence of photolytic product yields on (i) the presence of
oxygen and (ii) the concentration of DTTox (controlling the rate
Supporting Information Available: 1H NMR spectra
(Figures S1-S6) and HMQC spectra (Figures S7 and S8) of
products I and II. This material is available free of charge via
the Internet at http://pubs.acs.org.
•
of reaction 12). Reduction by HO2 (reaction 20) is postulated
based on the absence of any of its dismutation product, H2O2,
after photolysis. We note that, theoretically, radical 2 could
abstract hydrogen atoms also from the methine carbons of
38
References and Notes
DTTox which are activated by the R-hydroxy substituents.
However, the resulting R-hydroxyalkyl radicals would likely
undergo C-S homolysis38 ultimately generating radicals, which
would not be precursors for products I and II. As a consequence,
such a mechanism should have generated additional photoprod-
ucts, which were experimentally not observed.
(1) Creighton, T. E. Proteins; W. H. Freeman and Company: New
York, 1993.
(2) Nicolaou, K. C.; Dai, W.-M. Angew. Chem., Int. Ed. Engl. 1991,
30, 1387-1416.
(3) Nicolaou, K. C.; Smith, A. L. Acc. Chem. Res. 1992, 25, 497-
503.
(4) Greer, A. J. Am. Chem. Soc. 2001, 123, 10379-10386.
(5) Brzostowska, E. M.; Greer, A. J. Am. Chem. Soc. 2003, 125, 396-
404.
(6) Boyd, D. B. J. Am. Chem. Soc. 1972, 94, 8799-8804.
(7) Morine, G.; Kuntz, R. R. Photochem. Photobiol. 1981, 33, 1-5.
(8) Grant, D. W.; Stewart, J. H. Photochem. Photobiol. 1984, 40, 285-
290.
(9) Zhang, X.; Zhang, N.; Schuchmann, H.-P.; von Sonntag, C. J. Phys.
Chem. 1994, 98, 6541-6547.
(10) Scheffer, J. R. In Organic Solid State Chemistry; Desiraju, G. R.,
Ed.; Elsevier: Amsterdam, 1987; pp 1-45.
(11) McBride, J. M. Acc. Chem. Res. 1983, 16, 304-312.
(12) Byrn, S. R.; Pfeiffer, R. R.; Stowell, J. G. Solid-State Chemistry of
Drugs; SSCI, Inc.: West Lafayette, IN, 1999.
(13) Feng, X. W.; McBride, J. M. J. Am. Chem. Soc. 1990, 112, 6151-
6152.
Based on the oxygen-dependent increase of the quantum
yields of 5 in CH2Cl2, we conclude that the 290 nm species
formed during LFP in CH2Cl2 is most likely not the triplet state.
In CH2Cl2, the 290 nm species decays into another transient by
first-order kinetics with t1/2 ) 0.5 µs, corresponding to k ) 1.4
× 106 s-1. Taking a circa 8-times higher solubility of O2 in
organic solvents compared to H2O, we would expect a pseudo-
first-order reaction of the triplet state with O2 with k ≈ 2 × 106
s-1. Hence, the presence of O2 should cause a measurable
decrease of product formation in contrast to our experimental
data, which demonstrate that O2 is a prerequisite for product
formation in CH2Cl2.
The direct photolytic cleavage of disulfides may play an
important role for protein degradation in vivo during the
exposure to light, for example, for proteins located in the eye
or the skin. In addition, several authors have obtained evidence
for both C-S and S-S cleavage of disulfides through interaction
with triplet excited states.39-42 On the basis of these observa-
tions, the homolytic cleavage of disulfides does not necessarily
require direct exposure to light. For example, triplet excited
ketones form during the thermal decomposition of dioxetanes,
which are prominent biological oxidation products.43 These
triplet excited states have the potential for direct44 (via energy
transfer) and indirect45 (via peroxyl radicals) DNA damage and
should, in principle, react similarly with proteins.
(14) Konya, K. G.; Paul, T.; Lin, S.; Lusztyk, J.; Ingold, K. U. J. Am.
Chem. Soc. 2000, 122, 7518-7527.
(15) Murov, S. L.; Carmichael, I.; Hug, G. L. Handbook of Photochem-
istry, 2nd ed.; Marcel Dekker: New York, 1993.
(16) Eisenberg, G. M. Ind. Eng. Chem. Anal. Ed. 1943, 15, 327-328.
(17) Thomas, M. D.; Hug, G. L. Comput. Chem. (Oxford) 1998, 22,
491-498
(18) Wright, S. K.; Viola, R. E. Anal. Biochem. 1998, 265, 8-14.
(19) Ando, W.; Ohtaki, T.; Suzuki, T.; Kabe, Y. J. Am. Chem. Soc.
1991, 113, 7782-7784.
(20) Silverstein, R. M.; Webster, F. X. Spectrometric Identification of
Organic Compounds, 6th ed.; John Wiley & Sons: New York, 1998.
(21) Hoffman, M. Z.; Hayon, E. J. Phys. Chem. 1973, 77, 990-996.
(22) Elliot, A. J.; McEachern, R. J.; Armstrong, D. A. J. Phys. Chem.
1981, 85, 68-75.
(23) Burkey, T. J.; Hawari, J. A.; Lossing, F. P.; Lusztyk, J.; Sutcliffe,
R.; Griller, D. J. Org. Chem. 1985, 50, 4966-4967.
(24) Everett, S. A.; Scho¨neich, Ch.; Stewart, J. H.; Asmus, K.-D. J.
Phys. Chem. 1992, 96, 306-314.
Conclusion
(25) Von Sonntag, C. In Sulfur-Centered ReactiVe Intermediates in
Chemistry and Biology; Chatgilialoglu, C., Asmus, K.-D., Eds.; NATO ASI
Series, Series A: Life Sciences; Plenum Press: New York, 1990; Vol. 197,
pp 359-366.
(26) Bonifacic, M.; Scha¨fer, K.; Mo¨ckel, H.; Asmus, K.-D. J. Phys.
Chem. 1975, 79, 1496-1502.
(27) Hiller, K.-O.; Asmus, K.-D. Int. J. Radiat. Biol. 1981, 40, 597-
604.
(28) Gijzeman, O. L. J.; Kaufman, F.; Porter, G. J. Chem. Soc., Faraday
Trans. 2 1972, 69, 708-720.
(29) Turro, N. J. Modern Molecular Photochemistry; The Benjamin/
Cummings Publishing Co., Inc.: Menlo Park, CA, 1978; pp 589-590.
(30) Kosower, N. S.; Kosower, E. M. In Free Radicals in Biology; Pryor,
W. A., Ed.; Academic Press: New York, 1976; Vol. II, pp 55-84.
(31) Creed, D. A. Photochem. Photobiol. 1984, 39, 577-583.
(32) Smissman, E. E.; Sorenson, J. R. J. J. Org. Chem. 1965, 30, 4008-
4010.
The photoisomerization of trans-4,5-dihydroxy-1,2-dithiacy-
clohexane (DTTox) into 2,3-dihydroxy-1-mercaptotetrahydro-
thiophene 5 is initiated by homolytic cleavage of the disulfide
bond. In water, an oxygen-independent mechanism yields 5 via
either 1,5- or 1,2-H-shift followed by nucleophilic cyclization
or recombination of a sulfur- with a carbon-centered radical,
respectively. In the aprotic solvent CH2Cl2, a potential mech-
anism dependent on the concentration of oxygen and DTTox
involves intermolecular H-abstraction and the intermediary
formation of oxygen-centered radicals. Products analogous to
5 may represent intermediates to stable covalent aggregates in
proteins, especially if the free mercapto group would oxidize
further to sulfonic acid. On the basis of the reactions displayed
in Schemes 1 and 2 and the solid nature of DTTox, this disulfide
could represent a suitable initiator for controlled free radical
(33) Viskolcz, B.; Lendvay, G.; Ko¨rtve´lysi, T.; Seres, L. J. Am. Chem.
Soc. 1996, 118, 3006-3009.

Photoisomerization of a Cyclic Disulfide
J. Phys. Chem. A, Vol. 108, No. 12, 2004 2255
(34) Barton, D. H. R.; Beaton, J. M.; Geller, L. E.; Pechet, M. M. J.
Am. Chem. Soc. 1960, 82, 2640-2641.
(35) Lopez, J. C.; Alonso, R.; Fraser-Reid, B. J. Am. Chem. Soc. 1989,
111, 6471-6473.
(40) Byers, G. W.; Gruen, H.; Giles, H. G.; Schott, H. N.; Kampmeier,
J. A. J. Am. Chem. Soc. 1972, 94, 1016-1018.
(41) Rosenfeld, S. M.; Lawler, R. G.; Ward, H. R. J. Am. Chem. Soc.
1972, 94, 9255-9256.
(36) Dorigo, A. E.; McCarrick, M. A.; Loncharich, R. J.; Houk, K. N.
J. Am. Chem. Soc. 1990, 112, 7508-7514.
(37) Viola, H.; Hartenhauer, H.; Mayer, R. Z. Chem. 1988, 28, 269-
277.
(38) Akhlaq, M. S.; von Sonntag, C. J. Am. Chem. Soc. 1986, 108,
3542-3544.
(39) Walling, C.; Rabinowitz, R. J. Am. Chem. Soc. 1959, 81, 1137-
1143.
(42) Gruen, H.; Schott, H. N.; Byers, G. W.; Giles, H. G.; Kampmeier,
J. A. Tetrahedron Lett. 1972, 37, 3925-3928.
(43) Cilento, G.; Adam, W. Free Radical Biol. Med. 1995, 19, 103-
114.
(44) Epe, B.; Muller, E.; Adam, W.; Saha-Mo¨ller, C. R. Chem.-Biol.
Interact. 1992, 85, 265-281.
(45) Adam, W.; Arnold, M. A.; Saha-Mo¨ller, C. R. J. Org. Chem. 2001,
66, 597-604.