Investigation of flavin-containing DNA-repair model compounds

Article abstract of DOI:10.1021/ja964097u

Irradiation of DNA with UV-B light causes the formation of mutagenic DNA lesions such as cis-syn and trans-syn cyclobutane pyrimidine dimers. DNA photolyases are flavin-dependent repair enzymes which directly revert the mutagenic cis-syn pyrimidine dimers

Full text of DOI:10.1021/ja964097u

7440
J. Am. Chem. Soc. 1997, 119, 7440-7451
Investigation of Flavin-Containing DNA-Repair Model
Compounds
Robert Epple, Ernst-Udo Wallenborn, and Thomas Carell*
Contribution from the Department of Organic Chemistry, Swiss Federal Institute of Technology,
ETH-Zentrum, UniVersita¨tstrasse 16, 8092 Zu¨rich, Switzerland
ReceiVed NoVember 27, 1996X
Abstract: Irradiation of DNA with UV-B light causes the formation of mutagenic DNA lesions such as cis-syn and
trans-syn cyclobutane pyrimidine dimers. DNA photolyases are flavin-dependent repair enzymes which directly
revert the mutagenic cis-syn pyrimidine dimers into the corresponding monomers by a light-facilitated repair reaction.
To gain deeper insight into the repair process, we recently prepared flavin-containing model compounds which are
able to mimic the repair reaction (Carell, T.; Epple, R.; Gramlich, V. Angew. Chem., Int. Ed. Engl. 1996, 35, 620-
623). This publication now contains a detailed description of the synthesis of a series of related model compounds
and a comprehensive investigation of their cleavage properties. The results obtained help to unravel the requirements
necessary for an efficient, flavin-mediated cleavage of pyrimidine dimers and provide insight into the factors on
which the enzymatic repair process depends. The investigation of the cleavage reaction with cis-syn, trans-syn,
and trans-anti cyclobutane pyrimidine dimer model compounds reveal an enhanced vulnerability for the cis-syn
isomer. The trans-syn dimer is 10 times more stable. These results are comparable to those observed in a recent
study on the E. coli enzyme. The excellent solubility of some of the model compounds has allowed a medium-
dependent investigation of the flavin-initiated cleavage reaction. Increased cleavage efficiencies are observed in
polar solvents such as water (φ ) 0.06) and acetonitrile (φ ) 0.05). The quantum yields decrease by a factor of 4
in solvents with very low polarity such as dioxane (φ ) 0.01). These results are not in agreement with earlier
solvent-dependent evaluations performed with non-flavin-containing model compounds (Hartzfeld, D. G.; Rose, S.
D. J. Am. Chem. Soc. 1993, 115, 850-854). The results, however, suggest that the unusually polar flavin-binding
pocket, observed in the X-ray crystal structure of the E. coli. photolyase, might be required to increase the catalytic
repair efficiency. Investigations of the cleavage reaction in the presence of acid and base in organic solvents emphasize
the strict requirement for a deprotonated reduced riboflavin chromophore. The determined pH values for half-
maximal (pH ) 6.5) and maximal (7 e pH e 9) cleavage efficiencies are in agreement with the pKa value (pKa )
6.3) of the reduced riboflavin and reveal that physiological conditions are required to reach maximum catalytic
cleavage efficiency.
Introduction
by nature to remove the UV-induced photolesions from the
genome.2 DNA photolyases are flavin-containing repair en-
zymes which catalyze the efficient repair of the cis-syn
cyclobutane pyrimidine dimers.9 Elegant studies of various
photolyases2,9,10 have provided the currently accepted mecha-
nism of the repair process. The photolyase-catalyzed mono-
merization of pyrimidine dimers requires sunlight which initiates
an electron-transfer process from a reduced flavin cofactor in
the enzyme to the DNA damage.9,11,12 Spontaneous monomer-
ization of the resulting radical anion and transfer of the surplus
electron back to the riboflavin cofactor concludes the catalytic
cycle.
The irradiation of cells with UV-B light (280-320 nm) causes
the formation of cis-syn cyclobutane pyrimidine dimers as the
major photoproducts present in irradiated DNA.1,2 The structur-
ally related, highly mutagenic trans-syn lesions are formed to
a much smaller extent, predominantly upon irradiation of single-
stranded DNA.2,3 Both DNA lesions are involved in the
development of various skin cancers such as basal cell and
squamous cell carcinomas and melanomas.4-6 The observed
depletion of the ozone layer causes an increasing UV-B radiation
level and therefore a higher risk of skin cancer.7,8 This scenario
fuels research interests to elucidate the mechanisms developed
Although the X-ray structure of the E. coli photolyase13 gives
a detailed picture of the cofactor arrangement and its protein
environment, several important elements of the enzymatic
process remain unsolved: (1) It is not currently known how
the enzyme mediates the excited-state electron-transfer processes
* Corresponding author.
X Abstract published in AdVance ACS Abstracts, July 15, 1997.
(1) Blackburn, G. M.; Davies, R. J. H. J. Chem Soc. C 1966, 2239-
2244. W. Setlow, R. B. Photochem. Photobiol. 1968, 7, 643-649.
(2) Friedberg, E. C.; Walker, G. C.; Siede, W. DNA Repair and
Mutagenesis; ASM Press: Washington, DC, 1995.
(3) Taylor, J.-S. Acc. Chem. Res. 1994, 27, 76-82. Taylor, J.-S. J. Chem.
Educ. 1990, 67, 835-841.
(9) For recent reviews, see: Sancar, A. Biochemistry 1994, 33, 1-9.
Kim, S.-T.; Sancar, A. Photochem. Photobiol. 1993, 57, 895-904. Heelis,
P. F.; Kim, S.-T.; Okamura, T.; Sancar, A. J. Photochem. Photobiol., B
1993, 17, 219-228. Heelis, P. F.; Hartman, R. F.; Rose, S. D. J. Chem.
Soc. ReV. 1995, 289-297. Carell, T. Angew. Chem., Int. Ed. Engl. 1995,
34, 2491-2495.
(10) Sancar, G. B. Mutat. Res. 1990, 236, 147-160.
(11) Kim, S.-T.; Heelis, P. F.; Sancar, A. Biochemistry 1992, 31, 11244-
11248.
(12) Kim, S.-T.; Heelis, P. F.; Okamura, T.; Hirata, Y.; Mataga, N.;
Sancar, A. Biochemistry 1991, 30, 11262-11270.
(13) Park, H.-W.; Kim, S.-T.; Sancar, A.; Deisenhofer, J. Science 1995,
268, 1866-1872.
(4) Thielmann, H. W. In Recent Results in Cancer Research; Hecker,
E., Jung, E. G., Marks, F., Tilgen, W., Eds.; Springer-Verlag: Heidelberg,
1993; Vol. 128 (Skin Carcinogenesis in Man and in Experimental Models),
pp 275-297.
(5) Ames, B. N.; Gold, L. S.; Willett, W. C. Proc. Natl. Acad. Sci. U.S.A.
1995, 92, 5258-5265.
(6) Cleaver, J. E. Arch. Dermatol. 1993, 129, 348-350.
(7) Kerr, J. B.; McElroy, C. T. Science 262, 1032-1034.
(8) Daya-Grosjean, L.; Dumaz, N.; Sarasin, A. J. Photochem. Photobiol.
1995, 28, 115-124.
S0002-7863(96)04097-8 CCC: $14.00 © 1997 American Chemical Society

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7441
Figure 1. Representation of the electron-transfer-based cycloreversion of the model compounds 1 and 2. Reaction products are the monomers 3
and 4. The electron donor is a reduced flavin. The parent compounds 1 and 2 mimic the light-driven DNA-repair process catalyzed by the enzyme
DNA photolyase.
in order to achieve almost quantitative repair efficiency (φ )
0.6), and (2) it remains mysterious why photolyases are unable
to recognize and efficiently repair the minor but highly
mutagenic trans-syn photolesion.14 This trans-syn isomer
poses a serious mutagenic threat, especially in organisms which
possess a single-stranded DNA genome during their life
cycle.15,16
So far, the investigation of the dependencies of the repair
reaction with flavin model systems were limited to the analysis
of mixed solutions containing flavin derivatives and cyclobutane
dimer models.17-20 These studies were hampered by low
quantum yields of the photochemical cycloreversion process (φ
) 10-3-10-4) and by the limited solubility of the reaction
partners in solvents other than water. Recent solution studies,
performed by S. D. Rose and co-workers, in which flavins
initiate a cleavage chain reaction, have shown that deprotonation
of the reduced flavin is a strict requirement.20 This investigation
provided estimates of the pH values required for half-maximal
(pH ≈ 7.5) and maximal (pH ≈ 8.5) repair efficiency.20 These
high values, however, are not in agreement with the pKa of the
reduced riboflavin (pKa ) 6.3),21 and they raise the question of
how photolyases achieve maximal repair efficiency under
physiological conditions (pH ) 7).
9,10-dicyanoanthracene sensitizers.23,24 In contrast, less efficient
cleavage of the trans-syn isomers was reported with phenan-
threne and p-dicyanobenzene sensitizers25 and in studies using
γ-radiolysis.26
We have recently communicated the preparation of covalently
linked flavin-containing model compounds.27 Initial cleavage
experiments indicated a higher vulnerability of the cis-syn
isomer in the flavin-mediated monomerization process in
comparison to the trans-syn and the trans-anti isomer. We
now report the synthesis of a series of related model compounds,
derived from the parent compounds 127 and 2, which cleave
photoinduced into the photosplit products 3 and 4 as depicted
in Figure 1. An HPLC-based assay was developed, which
allows quantification of the cleavage rates and determination
of the quantum yields. The most active compounds feature
quantum yields of up to φ ) 10-1-10-2, which is 2-3 orders
of magnitude higher than previously observed with flavin-
containing solutions.17-19 The presented investigation of the
monomerization reaction shows that the flavin-induced repair
reaction is weakly solvent dependent and most efficient in
solvents possessing a high dielectric constant. This is in contrast
to earlier observations made with non-flavin-containing model
compounds.28-31 Our results, however, suggest an explanation
for the unusually polar flavin-binding pocket observed in the
X-ray structure of the E. coli enzyme.13 pH-dependent mea-
surements, performed with more stable amide-linked model
compounds show maximal cleavage efficiency at 7 pH e 9, in
agreement with the known pKa values of the reduced riboflavin
and of the cyclobutane uridine dimer lesion. This result helps
The cleavage rates of cis-syn, trans-syn, and trans-anti
cyclobutane dimers were measuredsusing nonflavin-containing
model systemssto investigate potential cleavage rate differ-
ences. The results, however, were inconclusive. Higher
cleavage rates for trans-syn isomers were determined in
experiments performed with anthraquinone,22 p-chloranil, and
(14) Kim, S. T.; Malhotra, K.; Smith, C. A.; Taylor, J.-S; A. Sancar
Biochemistry, 1993, 32, 7065-7068. See also: Ben-Hur, E.; Ben-Ishai, R.
Biochim. Biophys. Acta 1968 166, 9-15.
(23) Wenska, G.; Pasczyc, S. J. Photochem. Photobiol., B 1990, 8, 27-
37.
(24) Herbert, M. A.; LeBlanc, J. C.; Weinblum, D.; Johns, H. E.
Photochem. Photobiol. 1969, 9, 33-44.
(25) Pac, C.; Kubo, J.; Majima, T.; Sakurai, H. Photochem. Photobiol.
1982, 36, 273-282. For a Huckel calculation, see: Hartman, R. F.;
VanCamp, J. R., Rose, S. D. J. Org. Chem. 1987, 52, 684-689.
(26) Podmore, I. D.; Heelis, P. F.; Symons, M. C. R.; Pezesk, A. J. Chem.
Soc., Chem. Commun. 1994, 1005-1006.
(27) Carell, T.; Epple, R.; Gramlich, V. Angew. Chem., Int. Ed. Engl.
1996, 35, 620-623.
(28) VanCamp, J. R.; Young, T.; Hartman, R. F.; Rose, S. D. Photochem.
Photobiol. 1987, 45, 365-370. Kim, S.-T.; Rose, S. D. Photochem.
Photobiol. 1990, 52, 789-794. Kim, S.-T.; Rose, S. D. J. Photochem.
Photobiol. B. Biol. 1992, 12, 179-191. Kim, S.-T.; Rose, S. D. J. Phys.
Org. Chem. 1990, 3, 581-586. For other model studies, see: Austin, R.;
McMordie, S.; Begley, T. P. J. Am. Chem. Soc. 1992, 114, 1886-1887.
(29) Hartzfeld, D. G.; Rose, S. D. J. Am. Chem. Soc. 1993, 115, 850-
854.
(15) LeClerq, J. E.; Borden, A.; Lawrence, C. W. Proc. Natl. Acad. Sci.
U.S.A. 1991, 88, 9685-9689. Gibbs, P. E. M.; Kilbey, B. J.; Banerjee, S.
K.; Lawrence, C. W. J. Bacteriol. 1993, 175, 2607-2612. Gibbs, P. E. M.;
Lawrence, C. W. Nucleic Acids Res. 1993, 21, 4059-4065. Lawrence, C.
W.; Gibbs, P. E. M.; Borden, A.; Hersfall, M. J.; Kilbey, B. J. Mutat. Res.
1993, 299, 157-163. Hersfall, M. J.; Lawrence, C. W. J. Mol. Biol. 1994,
235, 465-471.
(16) Smith, C. A.; Wang, M.; Jiang, N.; Che, L.; Zhao, X.; Taylor, J.-S.
Biochemistry 1996, 35, 4146-4154.
(17) Rokita, S. E.; Walsh, C. T. J. Am. Chem. Soc. 1984, 106, 4589-
4595.
(18) Miyake, K.; Masaki, Y.; Miyamoto, I.; Yanagita, S.; Ohno, T.;
Yoshimura, A.; Pac, C. Photochem. Photobiol. 1993, 58, 631-636.
(19) Jorns, M. S. J. Am. Chem. Soc. 1992, 114, 3133-3136.
(20) Hartman, R. F.; Rose, S. D. J. Am. Chem. Soc. 1992, 114, 3559-
3560.
(21) Hemmerich, P; Veeger, C.; Wood, H. C. S. Angew. Chem., Int. Ed.
Engl. 1965, 4, 671-688.
(22) Ben-Hur, E.; Rosenthal, I. Photochem. Photobiol. 1970, 11, 163-
168.
(30) Hartman, R. F.; Rose, S. D. J. Org. Chem. 1992, 57, 2302-2306.
(31) For an AM1 calculation of activation energies and solvent depend-
encies, see: Heelis, P. F. J. Mol. Model. 1995, 1, 18-21.

7442 J. Am. Chem. Soc., Vol. 119, No. 32, 1997
Epple et al.
Scheme 1. Model Compounds for the Flavin-Initiated Cycloreversion of Pyrimidine Dimers by a Light-Driven
Electron-Transfer Processa
a All compounds were prepared from the cis-syn, trans-syn, and trans-anti cyclobutane uracil dimers 5-717 and the flavin alcohol 8 and the
flavin amine 9.
to clarify how photolyases achieve nearly quantitative repair
efficiency (φ > 0.6) under physiological conditions.
Finally, more detailed measurements of the cleavage proper-
ties of various cyclobutane uracil dimer isomers, incorporated
into the model compounds, allowed quantification of the
increased cleavage rate for the cis-syn pyrimidine dimer.
group in 10 with trifluoroacetic acid anhydride.34 Reaction of
11 with 1-bromo-2-methoxyethane (12)35 afforded 13, which
was hydrogenated to 14. The flavin synthesis developed by R.
Kuhn and co-workers36 afforded the methoxyethyl flavin 15.
The final compound 8 was obtained after ethylation to 16 and
cleavage of the methoxy protection group. Treatment of 8 with
the pyrimidine dimer dicarboxylic acids 5-7 after in situ BOP37
activation of the carboxylic acids and subsequent chromatog-
raphy yielded the three bisflavin model compounds 1, 17, and
18.27 To eliminate potential intramolecular flavin-flavin
π-stacking interactions38 which might obscure the quantum yield
measurements, preparation of the monoflavin-substituted model
compounds 19-22 was required. These compounds were
synthesized by reaction of the corresponding activated pyrimi-
dine dimer dicarboxylic acids 5-7 with 1 equiv of the
hydroxyethyl flavin 8 and the subsequent addition of a large
excess of benzyl alcohol (or pentyl alcohol) to the reaction
mixture.
Results and Discussion
Design and Preparation of the Model Compounds. The
model compounds presented in Scheme 1 were designed as
previously reported.27 As the lesion component, a cis-syn, a
trans-syn, and a trans-anti N1,N1′-dicarboxymethyl cyclobu-
tane uracil dimer was covalently attached to one or two flavin
chromophores. The three pyrimidine dimer dicarboxylic acids
5-7 were prepared as described. The structures of the three
isomers were confirmed by X-ray crystal structure analysis for
the cis-syn and the trans-syn isomers 5 and 6.27,32,33
The synthesis of the hydroxyethyl and aminoethyl 6,7-
dimethylflavin chromophores 8 and 9, required for the synthesis
of the model compounds, are depicted in Scheme 2. The
preparation of the hydroxyethyl flavin was recently com-
municated.27 The synthesis includes protection of the amino
The synthesis of the aminoethyl-substituted flavin 9 (Scheme
2) was required for the preparation of amide-linked model
(34) P. Kirsch, Dissertation, Heidelberg, Germany, 1993. See also: Staab,
H. A.; Zipplies, M. F.; Mu¨ller, T.; Storch, M.; Krieger, C. Chem. Ber. 1994,
127, 1667-1680.
(35) Roblin, R. O.; Lampen, J. O.; English, J. P.; Cole, Q. P.; Vaughan,
J. R. J. Am. Chem. Soc. 1945, 67, 290-294.
(36) Kuhn, R; Weygand, F. Ber. Dtsch. Chem. Ges. 1934, 67, 1409-
1413. Kuhn, R; Weygand, F. Ber. Dtsch. Chem. Ges. 1935, 68, 1282-
1288.
(32) Carell, T.; Epple, R.; Gramlich, V. HelV. Chim. Acta, manuscript
in preparation.
(33) Cochran, A. G.; Sugasawara, R.; Schultz, P. G. J. Am. Chem. Soc.
1988, 110, 7888-7890. Jacobsen, J. R.; Cochran, A. G.; Stephans, J. C.;
King, D. S.; Schultz, P. G. J. Am. Chem. Soc. 1995, 117, 5453-5461.
(37) Castro, B.; Evin, G.; Selve, C.; Seyer, R. Synthesis 1977, 413-469
(38) Slifkin, M. A. In Charge-Transfer Interactions of Biomolecules;
Academic Press: New York, 1971; Chapter 7.

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7443
Scheme 2a
a(a) (CF3CO)2O, room temperature (rt), 90%; (b) 1-bromo-2-methoxyethane35 (12), K2CO3, DMF, 100 °C, 89%; (c) H2, Pd/C, HOAc, rt; (d)
HOAc, boric acid, alloxane monohydrate, (e) EtI, K2CO3, DMF, rt, 83%; (f) BBr3, CH2Cl2, 0 °C, 81%; (g) 1.7 equiv of 1-(bromoethyl)phthalimide
(23), 175 °C, 23.5%; (h) H2, Pd/C, HOAc, 50 °C; (i) alloxane monohydrate, B(OH)3, HOAc, rt, 47%; (j) EtI, K2CO3, DMF, rt, 86%; (k) concentrated
HCl, reflux, 92%.
compounds required for measurements at high pH values.
Reaction of 10 with (1-bromoethyl)phthalimide (23) afforded
the phthalimido anilide 24. Hydrogenation of 24 in acetic acid
and treatment of the resulting aniline derivative 25 with alloxane
and boric acid in acetic acid yielded the flavin derivative 26,
which was ethylated at N(3) with ethyl iodide to give 27. Due
to the instability of the isoalloxazine of phthalimide 27 in the
presence of hydrazine,39 we cleaved the phthalimide moiety by
refluxing compound 27 in concentrated HCl and isolated the
target compound 9 as its stable hydrochloride salt. The flavin
ethylamine 9 could be condensed with cis-syn cyclobutane
uracil dimer dicarboxylic acid 5 to afford 2 by using in situ
BOP37 activation of the carboxylic acids and very slow addition
of triethylamine to the reaction mixture. To perform measure-
ments in nonpolar solvents such as dioxane, the monoflavin
monopentylamide model compound 28 was prepared in a similar
fashion through reaction of the cis-syn cyclobutane dicarboxylic
acid 5 with 1 equiv of the flavin ethylamine 9 and subsequent
quenching of the reaction mixture with pentylamine.
Figure 2. Fluorescence spectra of compound 28 in H2O (10-5 mol/L)
(s, 28 oxidized; - - -, 28 reduced) after addition of either sodium
dithionite solution or after reduction performed with hydrogen and Pd/
barium sulfate catalyst (excitation at 450 nm).
ated. Due to the amphoteric character of the reduced flavin41,42
and the observation that efficient cleavage of pyrimidine dimers
requires a deprotonated flavin chromophore,19,20 the aqueous
reduction solutions were buffered at pH 8.0 with 0.5 M Tris‚HCl
to ensure the predominant presence of the deprotonated, reduced
flavin species (pKa ) 6.3). In apolar media, reduction of the
model compounds was achieved catalytically with hydrogen and
Pd on barium sulfate. To ensure that the catalyst causes no
interference with the measurements, varying amounts of catalyst
were addedswithin the generally used rangesto the reaction
mixture. Similar cleavage rates were determined within the
experimental error. This catalytic procedure yields clear and
optically transparent solutions below 350 nm, which enables
exclusive excitation of the reduced flavin chromophores at all
required wavelengths. In these measurements, deprotonation
of the reduced flavins was accomplished with triethylamine.
As depicted in Figure 2, the fluorescence of the sample is
strongly diminished after reduction. This allowed the complete
The ethyl group at the flavin-N(3) center is the only difference
between the substitution pattern of our model compounds and
of the reaction partners (FAD chromophore and the dimer in
damaged DNA) involved in the natural enzymatic repair process.
The ethyl group increases the solubility of the model compounds
without affecting the redox potential of the riboflavin chro-
mophore.40 In addition, the X-ray crystal structure of the E.
coli photolyase reveals no hydrogen bond between the enzyme
and the flavin-N(3)-H position, which could alter the redox
potential of the isoalloxazine unit in the photolyase flavin-
binding pocket.13
Investigation of the General Cleavage Properties. For the
photolysis experiments in polar media, we prepared 10-4
M
solutions of the model compounds in water, ethylene glycol, or
DMF. In these solvents, the flavin chromophore is easily
reduced by addition of aqueous sodium dithionite solution. This
method affords clear solutions which were subsequently irradi-
(41) Ghisla, S.; Massay, V.; Lhoste, J.-M.; Mayhew, G. Biochemistry
1974, 13, 589-597.
(42) Two excellent flavin reviews: Bruice, T. C. Acc. Chem. Res. 1980,
13, 256-262. Walsh, C. Acc. Chem. Res. 1980, 13, 148-155.
(39) Ing, H. R.; Manske R. H. F. J. Chem. Soc. 1926, 2348-2351.
(40) Breilinger, E.; Niemz, A.; Rotello, V. M. J. Am. Chem. Soc. 1995,
117, 5379-5380.

7444 J. Am. Chem. Soc., Vol. 119, No. 32, 1997
Epple et al.
Figure 4. Concentration dependence of the overall splitting quantum
yield (φ) determined with the model compound 28. The irradiation
experiment was performed in ethylene glycol at 366 nm; reduction was
achieved with aqueous dithionite solution (pH 8.0).
the amount of 28 (elution time ≈ 17 min) and the increase of
the photoproduct 4 (elution time ≈ 8 min) are clearly evident.
In an effort to quantify the cleavage efficiency, peak areas were
integrated and standardized by division through the total peak
area. These normalized values were plotted against time as
shown in Figure 3B for the measurement presented in Figure
3A. The data points for the amount of starting material 28 were
fitted with an exponential decay function. An exponential rise
to maximum function was required to fit the data points obtained
for the amount of photocleaved product 4. This proves that
the cleavage reaction obeys first-order kinetics, as expected for
a purely intramolecular process.
Additional control experiments confirmed the intramolecu-
larity of the cleavage process: (1) Irradiation of mixtures
containing reduced methoxyethyl flavin 16 and the cis-syn
cyclobutane uracil dimer dibenzyl esters gave no carboxy-
methyl uracil benzyl ester, even after prolonged irradiation, and
(2) a concentration-dependent investigation of the quantum yield
performed with the cis-syn model compound 28 indicated no
change of quantum yields between 10-3 and 10-5 M concentra-
tions within experimental error (Figure 4). At concentrations
of approximately 10-4 M, higher quantum yields with a smaller
error were obtained because the integration of HPLC peaks was
the most precise in this concentration range. Therefore, 10-4
M solutions were used for all further measurements.
Two additional control experiments were performed to
validate the strict requirement of a reduced flavin sensitizer:
(1) Irradiation of the cyclobutane uracil dibenzyl ester alone or
in the presence of reducing agents yielded no photoproduct, and
(2) irradiation of the model compounds without prior reduction
of the flavin chromophore gave no photoconversion even after
prolonged irradiation.
To calculate the quantum yields φ (φ ) number of photo-
cleaved model compounds/number of absorbed photons), the
initial rates for the cleavage reaction were extracted from the
fitted curve. The intensity of the light beam was measured by
ferrioxalate actinometry.44,45 The number of photons absorbed
by the model compound was calculated using the absorbance
of the irradiated sample at 366 nm, determined prior to the
measurement. Quantum yields were calculated on the basis of
Figure 3. Irradiation experiment performed with 28 in MeOH (3 mL
of a 10-4 mol/L solution). Reduction was performed by catalytic
hydrogenation with Pd/barium sulfate catalyst. Addition of 50 µL of
NEt3 (0.1 M final concentration) ensured complete deprotonation of
the reduced flavin chromophores (irradiation at 366 nm). (A) HPLC
chromatograms obtained after 0, 2, 4, 6, 9, 12, 15, and 18 min of
irradiation (Lichrosphere 100/5 column from Macherey-Nagel). Solvent
systems: 55% A and 45% B over 9 min, then switching to 80% A and
20% B for an additional 19 min (A, methanol; B, water) (detection at
450 nm). Retention times: 7.3 min (cleavage product 4), 17.4 min
(model compounds 28). (B) Plot of relative peak areas vs time and
curve fits obtained with an exponential rise to maximum and an
exponential decay function.
reduction of the sample to be monitored during a photolysis
experiment.41-43 As a result of the small, but significant UV
spectroscopic differences43 (vide infra) of the various species
(oxidized, reduced, and reduced and deprotonated), the proto-
nation state of the isoalloxazine moiety could be determined
precisely before each measurement in all solvents used.42
The reduced and deprotonated samples are stable in the dark,
in an inert atmosphere for several hours as determined by TLC
and by reverse-phase HPLC chromatography of a reoxidized
sample. Upon exposure to daylight or to monochromatic light
of, for example, 366 or 400 nm, however, the cis-syn and the
trans-syn model compounds react cleanly to give only one new
product. Analysis of the reaction mixture by reverse-phase
HPLC and co-injection of the synthesized expected reaction
products 3 or 4 confirmed that the light-dependent cleavage
reaction yields only 3 and 4, even after prolonged irradiation.
The splitting rates for the photoinduced cleavage reactions
were determined by reverse-phase HPLC analysis of aliquots
removed from the irradiated reaction mixture and immediately
reoxidized by exposure to oxygen. The series of chromatograms
presented in Figure 3A were obtained upon irradiation of model
compound 28 in methanol. The time-dependent decrease of
(44) Hatchard, C. G.; Parker, C. A. Proc. R. Soc. A 1956, 235, 518-
536.
(45) Braun, A. M.; Maurette, M. T.; Oliveros, E. In Photochemical
Technology; John Wiley and Sons: England, 1991; pp 77-81. Murov, S.
L. In Handbook of Photochemistry; Marcel Dekker: New York, 1973; pp
119-123.
(43) Dudley, K. H.; Ehrenberg, A.; Hemmerich, P.; Mu¨ller, F. HelV.
Chim. Acta 1964, 150, 1354-1383.

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7445
syn dimer-containing model compounds 19 and 22. Due to the
absorption of the dithionite at the irradiation wavelength of 366
nm, we checked the results with all model compounds by
irradiation at 400 nm, where the dithionite absorption is
negligible. Irradiation at 400 nm and also with white light,
however, yielded the same relative cleavage rate differences for
the cis-syn, the trans-syn, and the trans-anti cyclobutane
uracil isomers.
On the basis of our measurements, we conclude that the
different flavin-initiated cleavage rates of the three investigated
isomers are determined solely by the different cyclobutane
constitutions and configurations. The observed increased stabil-
ity of the trans-syn isomer most likely also effects the
repairability of the natural cis-syn and trans-syn DNA lesions
by the enzyme DNA photolyase.2,14 For the enzymatic process,
A. Sancar and co-workers determined a very low and probably
unspecific binding of the trans-syn isomer by the E. coli
photolyases, with a drastically reduced repair quantum yield of
φ g 0.003 (compare φ > 0.6 for the cis-syn isomer).14 Our
result suggests that the inherent stability of the trans-syn dimer
might be another factor influencing the low enzymatic repair
efficiency.
Figure 5. Dependencies of quantum yield on the configuration and
constitution of the cyclobutane ring. The trans-anti (18 and 21), trans-
syn (17 and 20), and cis-syn (1, 19, and 22) model compounds were
measured in ethylene glycol with irradiation at 366 nm. Reduction was
performed with aqueous dithionite solution (pH 8.0).
To gain deeper insight into the possible reason for the
increased vulnerability of the cis-syn dimer toward the flavin-
mediated cleavage reaction, we analyzed the crystal structures
of the cis-syn and the trans-syn isomers for significant
differences. C. Pac et al. explained different cleavage rates in
oxidative processes as being due to different through-bond
interactions of the relevant orbitals in the three isomers.25 The
crucial orbital-orbital overlaps depend critically on the dihedral
angle between the interacting orbitals and therefore on the ring
pucker. C. Pac suggested that the sterically encumbered cis-
syn dimer causes an increased orbital overlap, which then results
in a higher cleavage rate.25,47
In the generally accepted view of the reductive splitting
process, the donated electron initially occupies the π* orbital
of the CdO bond.9,48,49 Delocalization of electron density into
the antibonding C(5)-C(5′) σ* orbital weakens this bond and
causes the initial bond breakage. For our system, the interaction
between the C(4′)dO(4′) and C(4)dO(4) π* orbitals and the
C(5)-C(5′) σ* orbital is therefore of crucial importance for the
cleavage efficiency. Upon electron uptake, the C(5)-C(5′) bond
strength could be reduced by different extents in both isomers.
This might then cause different cleavage rates. The analysis
of the X-ray crystal structures27 of the cis-syn and the trans-
syn cyclobutane uracil dimer dibenzyl esters reveals, as expected,
only very small differences of the cyclobutane bond lengths
and torsion angles. The orientation of the C(4)dO(4) double
bond relative to the C(5)-C(5′) bond is, however, significantly
different. In both isomers, the descriptive torsion angles O(4)-
C(4)-C(5)-C(5′) differ on one side of the dimer by ap-
these results. All photolysis measurement was performed three
times, and the averaged data gave an individual error for each
measurement of (8%. Measurements of the parent compounds
1 and 2 in water at pH 8.0 revealed a remarkably high quantum
yield of φ ) 0.064. This is 2-3 orders of magnitude higher
than the quantum yields obtained previously in experiments
performed with mixed solutions of riboflavin derivatives and
cyclobutane pyrimidine dimers.17-19 The present molecules are
therefore the first coValently linked flaVin-pyrimidine dimer
model compounds which are able to mimic the enzymatic repair
reaction.
Cleavage of the Cyclobutane Pyrimidine Dimer Depending
on the Configuration and Constitution. The photoinduced
splitting properties of the three bisflavin model compounds 1,
17, and 18, which contain a cis-syn, a trans-syn, or a trans-
anti cyclobutane uracil dimer, were analyzed.27 All compounds
were dissolved in ethylene glycol (sodium dithionite reduction,
buffered at pH 8.0) and irradiated at 366 nm (Figure 5).
Remarkably different cleavage efficiencies were observed for
the three isomers. The fastest cleavage, with a quantum yield
of φ ) 0.034, was observed for 1, which contains the cis-syn
cyclobutane uracil dimer. Under identical conditions, a quantum
yield of only φ ) 0.003 was measured for the corresponding
trans-syn compound 17, which is thus approximately 10 times
more resistant toward the flavin-induced cycloreversion. In
agreement with earlier observations,25 the trans-anti cyclobu-
tane uracil dimer 18 is unable to undergo the cyclobutane
cleavage, even after prolonged irradiation. All three model
compounds 1, 17, and 18 were also measured in acetonitrile
and in ethanol (data not shown) to exclude the possibility of
solvent effects. In these solvent systems, similar quantum yield
differences were observed for the three cyclobutane uracil
dimers. Again, the cis-syn isomer was found to be ap-
proximately 10 times more sensitive toward electron-transfer-
induced cycloreversion.
(47) Pac, C.; Ohtsuki, T.; Shiota, Y.; Yanagida, S.; Sakurai, H. Bull.
Chem. Soc. Jpn. 1986, 59, 1133-1139. Semiempirical calculations predict
unfortunately an almost planar cyclobutane ring and are therefore not able
to simulate pucker determined effects. We performed a transition state search
using GAMESS(US)50 on the UHF/PM3 level. For both isomers, the
transition states corresponding to the C(5)-C(5′) bond cleavage were found.
Frequency analysis verified that the stationary points corresponded to first-
order transition states. The activation energies were obtained as the
difference between the transition state energies and the energies of the
corresponding UHF/PM3-minimized anions. The calculations yielded as
expected comparable activation energies of 3.9 kcal/mol for the cis-syn
and 3.6 kcal/mol for the trans-syn isomer,49 which underlines the inability
of semiempirical calculations to reproduce the observed cleavage rate
differences.
Since flavins are known to build π-stacked aggregates,38
which might obscure the electron-transfer step,46 we also
investigated the cleavage properties of the four monoflavin
model compounds 19-22, in which an intramolecular stacking
of two flavin units is impossible. As depicted in Figure 5
increased cleavage rates were again determined for the cis-
(48) Splitting rate in a dimethylaniline model system: Yeh, S. R.; Falvey,
D. E. J. Am. Chem. Soc. 1991, 113, 8557-8558.
(49) Voityuk, A. A.; Michel-Beyerle, M. E.; Ro¨sch, N. J. Am. Chem.
Soc. 1996, 118, 9750-9758.
(46) Marcus, R. A.; Sutin, N. Biochem. Biophys. Acta 1985, 811, 265-
322.

7446 J. Am. Chem. Soc., Vol. 119, No. 32, 1997
Epple et al.
Table 1. Calculated Mayer Bond Orders51 for the Neutral Dimer
and Its Radical Anion, Based on the X-ray Crystal Structure of the
Dibenzyl Ester17,32 with RHF/4-31G-Optimized Hydrogen Atoms as
Input Geometry (Program: GAMESS(US))50
neutral dimer
anionic dimer
C(5)-C(5′) C(6)-C(6′) C(5)-C(5′) C(6)-C(6′)
isomer
cis-syn
bond
bond
bond
bond
UHF/STO-3G
UHF/6-31G*
UHF/DZV
0.958
0.947
0.850
0.945
0.938
0.852
0.884
0.778
0.692
0.940
0.936
0.857
trans-syn
UHF/STO-3G
UHF/6-31G*
UHF/DZV
0.963
0.964
0.890
0.931
0.914
0.870
0.906
0.826
0.748
0.924
0.924
0.884
proximately 20°! To investigate how this difference might
influence the cleavage properties, ab initio calculations were
performed using the program GAMESS(US).50 The X-ray
crystal structures were used as geometries for modeling both
the neutral molecules and their radical anions. The calculated
Mayer bond orders are listed in Table 1.51 Electron transfer to
both isomers results in the expected reduction of the C(5)-
C(5′) bond orders. The decreases in bond order were calculated
to be approximately 19% for the cis-syn isomer and, on
average, only 15% for the trans-syn dimer. The calculations
therefore show the expected trend. The bond weakening is more
pronounced in the cis-syn dimer radical anion, and this isomer
is cleaved 10 times faster than the trans-syn dimer. More
detailed calculations are now required to fully understand and
to quantify the increased lability of the cis-syn isomer.
Nevertheless, the rudimentary bond order calculations suggest
an attractive preliminar explanation for the observed rate
differences.25
Figure 6. Measurement of the dependency of the quantum yield on
the polarity of the medium with the model compounds 2 and 28.
Irradiation was performed at 366 nm; reduction was achieved by the
addition of aqueous dithionite solution buffered at pH 8.0. The
measurements were performed in water (ꢀ ) 78.3), in water/ethylene
glycol (Etgl) mixtures, in pure ethylene glycol (ꢀ ) 37.7), and in DMF
(ꢀ ) 36.7).52
Table 2. Solvent Dependence of Quantum Yields for Compound
28
solvent
ꢀra
Φb
1,4-dioxane
benzene
ethyl acetate
tert-butyl alcohol
ethanol
methanol
acetonitrile
2.21
2.27
6.02
12.47
24.55
32.66
35.94
0.016
c
0.024
0.037
0.044
0.040
0.046
Medium-Dependent Measurements of the Cleavage Reac-
tion.52 To investigate how the polar flavin-binding pocket of
the E. coli photolyase might influence the cleavage reaction,
medium-dependent measurements were performed with the
monoflavin-monopentylamide model compound 28. To vali-
date the results, further measurements were carried out with
the bisflavin model compound 2.
a
Relative permittivity (dielectric constant) for the pure liquid at 25
°C.52 b Quantum yields of model compound 28. c No quantum yield
detectable even after prolonged irradiation.
In addition, the cleavage efficiency was measured in various
organic solvents. The model compound 28 was catalytically
reduced, and triethylamine (50 µL, final concentration 0.1 M)
was added to ensure complete deprotonation of the flavin
chromophore (determined by UV/vis spectroscopy). The quan-
tum yields calculated for the cleavage reaction in various organic
solvents are given in Table 2. In these measurements, com-
pound 28 features the highest quantum yields in the most polar
solvents. A gradual decrease of the cleavage efficiency was
observed in organic solvents as a function of decreasing polarity.
The irradiation experiments performed in organic media
support the results obtained in water/ethylene glycol mixtures
and suggest that the appearance of highly charged intermediates
during the cleavage reaction is unlikely.53 We observe only a
4-fold decrease of the quantum yield over the entire polarity
range investigated. The best cleavage efficiencies were ob-
served in polar media. The result of the polarity-dependent
measurement is in perfect agreement with the X-ray crystal
structure of the E. coli photolyase, which shows a riboflavin
Initially, the cleavage efficiency was investigated in pure
water and the polarity of the medium was reduced by addition
of ethylene glycol or ethanol (data not shown): these solvents
possess similar characteristics but have different dielectric
constants. We also measured the cleavage efficiency in pure
ethylene glycol and in DMF (Figure 6). The required aqueous
solutions were buffered at pH 8.0. UV/vis measurements
performed prior to the irradiation experiments ensured complete
reduction and deprotonation of the flavin units. The highest
quantum yield (φ ) 0.062) in the cleavage reaction of 28 was
obtained in pure water. Upon addition of ethylene glycol, a
gradual decrease of the quantum yield was observed until φ )
0.033 was reached in pure ethylene glycol. The systematic
reduction of the medium polarity results in a significant decrease
of the quantum yield for the cleavage reaction. Decreasing the
solvent polarity extends the half-life of the model compound
28 from 10 min in water to 20 min in ethylene glycol. A second
series of measurements performed with the bisflavin amide
model compound 2 (Figure 6) confirmed the small solvent
dependence and increased cleavage rates in the most polar
solvent mixture H2O/ethylene glycol (2:1).
(52) Reichardt, C. In SolVent and SolVent Effects in Organic Chemistry;
VCH: Weinheim, Germany, 1988.
(53) Another explanation for the observed solvent effects is that the
conformation of the cyclobutane uridine dimer is altered in different media.
Although this cannot be ruled out, the descriptive torsion O(4)-C(4)-
C(5)-C(5′), which is responsible for the orbital overlap, must change by
several degrees in order to influence the cleavage rate by a factor of 4. We
believe currently that such a torsion angle change, within a rigid cyclobutane
pyrimidine dimer framework, is unlikely.
(50) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.;
Gordon, M. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.;
Su, S. J.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem.,
1993, 14 (11), 1347-1363.
(51) Mayer, I. Int. J. Quantum Chem. 1986, 29, 477-483.

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7447
Figure 7. UV spectra of the model compound 28 in H2O (10-5 mol/
L): reduction was performed by catalytic hydrogenation over Pd/barium
sulfate catalyst [s, oxidized 28 (pH 3-10); - ‚ -, reduced 28 (pH 3-5);
‚‚‚, reduced 28 (pH 6); - - -, reduced 28 (pH 7-10)]. Citrate buffers
(0.5 M) were used for pHs 3, 4, 5, and 6. Tris.HCl buffers (0.5 M)
were employed for pHs 7, 8, 9, and 10.
cofactor tightly bound in an unusually polar binding pocket by
an array of hydrogen bonds.13 A salt bridge is oriented parallel
to the isoalloxazine chromophore within van der Waals contact.
Several water molecules are in close proximity to the isoallox-
azin ring system, and both heterocycles of the FAD cofactor
are partly solvent-accessible through an opening in the surface
of the enzyme. The fact that the amino acids involved in the
flavin-binding pocket are highly conserved in all eight known
microbial photolyases suggests that the polar environment of
the cofactor is of general importance for the enzymatic repair
reaction.13 On the basis of the solvent-dependent studies, we
presume that this polar environment might be in part responsible
for the high catalytic efficiency observed for these enzymes.
We believe that the solvent effect measured with our model
compounds is related to the ability of the system to deprotonate
the reduced riboflavin. The reduced riboflavin has a low pKa
value (pKa ) 6.3),21 allowing it to be readily deprotonated under
physiological conditions, thus generating a negatively charged
electron donor (FlH-). pH-dependent measurements were
carried out to further investigate the dependence of the cleavage
reaction in our model compounds on the degree of protonation
of the reduced flavin.
pH-Dependent Measurements of the Cleavage Reaction.
The cleavage reaction was investigated in water at various pH
values and in different organic solvents such as acetonitrile,
ethanol, or dioxane in the presence of acetic acid or triethy-
lamine. The degree of protonation of the reduced flavin was
closely monitored by UV spectroscopy. Figure 7 depicts the
UV spectra obtained from 10-5 M solutions of oxidized 28 and
the hydrogenated model compound 28 below pH 5 (neutral
species), at pH 6, and above pH 7 (deprotonated species). The
measured spectra are in agreement with the pKa value of 6.321
determined for the reduced riboflavin and with literature spectra
obtained for similar flavin species.43 In organic media, we
observed the reduced and protonated flavin species directly after
catalytic reduction and also in the presence of acetic acid.
Addition of triethylamine yielded the deprotonated form, as
determined by UV spectroscopy. These solutions (10-4 M) were
then irradiated at 366 nm. The calculated quantum yields are
depicted in Figure 8A,B. In organic solvents, efficient splitting
of the cyclobutane ring is strictly limited to solutions containing
Figure 8. (A) pH dependence of quantum yields measured with 28 in
H2O at different pH values (0.5 M citrate buffers for pHs 3, 4, 5, and
6; 0.5 M Tris.HCl buffers for pHs 7, 8, and 9; 0.05 M phosphate buffers
for pHs 10 and 11). The value at pH 11 possesses a greater error due
to beginning of decomposition of the model compound. (B) Measure-
ment of the quantum yield of 28 in acetonitrile, ethanol, and dioxane.
Right side: in the presence of triethylamine. Left side: in the absence
of triethylamine.
triethylamine. In solutions without base, the measured photo-
cleavage is very low. Addition of acetic acid during an
irradiation experiment caused the immediate termination of the
cleavage reaction. Thus the reduced, neutral riboflavin species
is unable to perform photocleavage. This observation is in
agreement with earlier measurements performed by S. D. Rose
and co-worker20 and by M. S. Jorn19 with flavin and cyclobutane
pyrimidine dimer solutions. Supporting results were obtained
by measurements performed with buffered aqueous solution.
Here, we observed negligible splitting below pH 5 and maximal
splitting efficiencies above pH 7. The half-maximal quantum
yield was observed around pH 6. In contrast to earlier studies,20
the model compounds are repaired with close to maximum
efficiency at pH 7. These results show that maximal cleavage
efficiency is reached under physiological conditions. At pH 9,
the cleavage efficiency decreases again, possibly due to the
deprotonation of the cyclobutane uracil dimer unit (pKa )
10.7).19 Above pH 11, decomposition of the model compounds
was observed, thus preventing further measurements.
Conclusions of the Solvent and pH-Dependent Measure-
ments. Earlier medium-dependent measurements with model
compounds which possess an arylamine- or indol-type electron
donor, instead of a reduced and deprotonated flavin cofactor,
showed increased cleavage efficiencies in nonpolar media such
as hexane/dioxane (95:5).28,29 These previously investigated
[Donor-Dimer] systems yield a zwitterionic intermediate

7448 J. Am. Chem. Soc., Vol. 119, No. 32, 1997
Epple et al.
Scheme 3. Schematic Representation of the Cleavage
Process in a Donor Cyclobutane Pyrimidine Dimer System
(Donor-D) and in a Flavin Cyclobutane Pyrimidine Dimer
System (FlH2-D) (D, Dimer; M, Monomer)
X-ray structure of the E. coli enzyme13 may be required to
maximize the repair efficiency. The dependence of the cleavage
reaction on the protonation state of the reduced flavin is in
agreement with earlier observations19,20 and underlines the strict
necessity to deprotonate the reduced flavin chromophore. The
model compounds reveal maximal repair efficiency under
physiological conditions (pH > 7), which is in agreement with
the pKa value of 6.321 for the reduced riboflavin. Above pH 9,
the quantum yield decreases, due to deprotonation of the dimer
unit.
The spectral region between 190 and 360 nm is currently
not accessible in measurements performed with the enzyme
because the enzyme requires sulfite containing buffers for the
reduction of the enzyme-bound flavin. Reduction of our model
compounds by catalytic hydrogenation enables the preparation
of active model compound solutions with an optical transparancy
from 190 to 360 nm. This possibility allows, for the first time,
the investigation of transient intermediates which possess
absorption bands in this spectral region and the measurement
of their lifetimes by picosecond laser photolysis spectroscopy.
The results of these measurements will be reported in due
course.
[Donor--Dimer+] after photoinduced electron transfer to the
dimer unit. It is currently believed that this intermediate
possesses a high driving force for an unproductive reversal of
the electron-transfer process, which then regenerates the [Donor-
Dimer] state.28,29 On the basis of the observation that the
cleavage in these systems proceeds most efficiently in nonpolar
media, it was postulated that the competing back electron
transfer might be slowed in the Marcus inverted region.29
In contrast, a photoinduced electron transfer from a depro-
tonated flavin to the pyrimidine dimer [FlH--Dimer] results
in a simple shift of the negative charge from the flavin unit to
the dimer [FlH•-Dimer-] as depicted in Scheme 3. The
occurrence of a [FlH•-Dimer-] intermediate, which is not
significantly more or less polar than the initial state, is strongly
supported by the minimal medium dependence observed for the
cleavage reaction (factor four). Such an intermediate was
previously suggested by T. Okamura et al. based on picosecond
laser photolysis studies.54 The lifetime of the [FlH•-Dimer-]
state is of crucial importance for the efficient photoinduced
cycloreversion of cyclobutane pyrimidine dimers by reduced
and deprotonated flavins. The [FlH•-Dimer-] intermediate can
either split to yield [FlH--M + M] or can collapse by a back
electron transfer to give [FlH--Dimer]. Due to the nonzwit-
terionic character of this intermediate, a much smaller driving
force is expected for the unproductive charge-shift reaction.
These considerations and the medium-dependent measurements
seem to exclude the “Marcus inverted region” argument as an
explanation for the high repair efficiency in flavin/pyrimidine
dimer systems. It is possible, however, that the polar environ-
ment and the known stability of the FADH• are the crucial
factors influencing the stability of the [FADH•-Dimer-]
intermediate. In the enzymatic case, the FADH• is known to
be exceptionally well stabilized if bound in the enzymatic
binding pocket,9 and most purified photolyases contain the
riboflavin cofactor in its neutral blue radical form. This
indicates that the FADH• stability might also be important for
the enzymatic repair efficiency.
Experimental Section
General Methods. All materials were obtained from commercial
suppliers and were used without further purification. Solvents of
technical quality were distilled prior to use. The aqueous buffers were
prepared using deionized distilled water. For reactions under an inert
gas atmosphere, nitrogen of standard quality was used. For analytical
thin layer chromatography, precoated silica gel plates (Merck 60-F254)
were used. Staining of amino compounds was performed with
ninhydrin. Flash chromatography was performed using Silica gel
(Merck, 0,040-0.063 mm) and Silica gel-H (Fluka, 0.005-0.040 mm).
Melting points are uncorrected and were determined on a Bu¨chi Smp
20. IR spectra were recorded on a Perkin-Elmer 1600 FT-IR using
KBr pellets or CHCl3 solutions. UV spectra were recorded on a Varian
Cary 5 UV/vis spectrophotometer in 1 cm quartz cuvettes. Fluorescence
spectra were measured on a Spex 1680 0.22 m double spectrometer
with a bandwidth of 3.4 nm in 1 cm quartz cuvettes. NMR spectra
were recorded on a Varian Gemini 200 (200 MHz (1H), 50 MHz (13C)),
Varian Gemini 300 (300 MHz (1H), 75 MHz (13C)), and Bruker AM-
500 (500 MHz (1H), 125 MHz (13C)). The chemical shift (δ) is reported
in parts per million downfield from tetramethylsilane (TMS, δ ) 0
ppm). Altermatively, the resonance of residual solvent protons were
used as the reference. EI-mass spectra and FAB-mass spectra were
measured by the staff of the mass spectrometry facilities of the ETH
Zurich on a Hitachi-Perkin-Elmer VG TRIBRID with 70 eV ionization
energy (EI) and on a ZAB-2 SEQ with 3-nitrobenzyl alcohol as the
matrix (FAB). Elemental analysis were performed by the microanalysis
laboratory of the ETH Zurich. HPLC chromatograms were obtained
with a Knauer HPLC instrument (HPLC pumps 64, Knauer variable-
wavelength UV detector, Knauer degasser) using HPLC grade solvents.
Summary and Outlook. The model compounds presented
herein mimic the photoinduced, reduced flavin-dependent cleav-
age of cyclobutane pyrimidine dimers (φ ) 0.06). This process
is exploited by nature to remove the harmful cyclobutane
pyrimidine dimer lesion present in UV-B-irradiated DNA. The
investigation of the cleavage reaction with differently configured
cyclobutane uracil dimers revealed an enhanced cleavage
efficiency for the cis-syn dimer, which is the major substrate
for DNA photolyases. The trans-syn cyclobutane uracil dimer
demonstrates an increased resistance toward flavin-induced
reductive cycloreversion. This resistance might be one factor
influencing the low trans-syn repair activity observed for the
E. coli DNA photolyase.14
N-(2′-Methoxyethyl)-4,5-dimethyl-2-nitroaniline (13). 1-Bromo-
2-methoxyethane37 (12) and 4,5-dimethyl-2-nitro-N-(trifluoroacetyl)-
aniline36 (11) were prepared according to the literature. A solution
containing 11 (8.00 g, 30.5 mmol) and K2CO3 (16.0 g, 116 mmol) 100
mL in DMF (100 mL) was heated to 100 °C. At this temperature,
1-bromo-2-methoxyethane (12) (12.0 g, 86.3 mmol) was carefully
added. The solution was kept at reflux temperature for additional 90
min. Then the solvent was removed in vacuo. Water was added to
the residual material, and the slurry was extracted with 250 mL of
CHCl3 three times. The combined organic layers were dried with
MgSO4, filtered, and concentrated in vacuo. Isolation of the product
by silica gel flash column chromatography (6 × 12 cm, toluene f
toluene/EtOAc (10:1)) afforded 13 (6.12 g, 89%) as an orange solid:
mp 33-35 °C; IR (CHCl3) 3378 (w), 3011 (w), 2922 (w), 1632 (s),
1574 (s), 1507 (s),1409 (m), 1336 (m), 1241 (s), 1156 (m), 1121 (m),
Investigation of the solvent-dependent cleavage process
revealed that reaction rates increase in polar media. These
studies show that the polar flavin environment observed in the
(54) Okamura, T.; Sancar, A.; Heelis, P. F.; Begley, T. P.; Hirata, Y.;
Mataga, N. J. Am. Chem. Soc. 1991, 113, 3143-3145. Essenmacher, C;
Kim, S.-T; Atamian, M,; Babcock, G. T.; Sancar, A. J. Am. Chem. Soc.
1993, 115, 1602-1603. Rustandi, R. R.; Jorns, M. S. Biochemistry 1995,
34, 2284-2288.
1
1022 (w), 1006 (w), 850 (w) cm-1; H NMR (200 MHz, CDCl3) δ
2.18 (s, 3 H), 2.27 (s, 3 H), 3.43 (s, 3 H), 3.49 (t, J ) 5.4 Hz, 2 H),

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7449
3.67 (t, J ) 5.4 Hz, 2 H), 6.64 (s, 1 H), 7.93 (s, 1 H), 8.10 (m, 1 H);
13C NMR (50 MHz, CDCl3) δ 18.73, 20.91, 43.01, 59.31, 70.84, 114.38,
114.43, 124.86, 126.88, 144.40, 147.53; EI-MS (70 eV) m/z 224 (20,
M - H+), 179 (100, [M - NO2]+), 106 (15), 28 (6).
bromoethyl)phthalimide (23, 8.00 g, 31.5 mmol) were mixed and
pulverized. This solid material was heated in a culture tube at 175 °C
for 5 h. The melted solids were allowed to cool to room temperature.
The obtained solid material was treated with 200 mL of boiling EtOH.
The obtained slurry was heated to reflux for 10 min and subsequently
sonicated at room temperature for an additional 30 min to dissolve all
product. The slurry was allowed to stand at 4 °C overnight. The brown
precipitate was filtered off and recrystallized from EtOH twice. 24
(1.44 g, 23.5%) was obtained as brownish needles: mp 213-215 °C;
IR (CHCl3) 3378 (w), 1772 (w), 1717 (s), 1633 (w), 1572 (m), 1506
10-(2′-Methoxyethyl)-7,8-dimethyl-10H-benzo[g]pteridine-2,4-di-
one (15). A suspension of Pd/C catalyst (50 mg) in AcOH (2 mL)
was slowly added to a solution of 13 (3.00 g, 13.3 mmol) in 50 mL of
AcOH. The mixture was sparged with hydrogen and stirred for 15 h
at room temperature. The reaction mixture, containing the diamine
14, was filtered through a Celite pad. Then alloxane monohydrate (7.00
g, 43.7 mmol) and boric acid (15.0 g, 243 mmol) were added. The
yellow slurry was stirred for 6 h at room temperature. Next, 500 mL
of water were added and the reaction mixture was extracted three times
with 250 mL of CHCl3 each. The organic layers were combined,
washed with saturated aqueous Na2CO3 solution, dried with MgSO4,
and concentrated in vacuo. Recrystallization of the residual material
from AcOH/H2O gave 15 (1.79 g, 46%) as a gold brown powder: mp
279-281 °C (dec); IR (CHCl3) 3378 (w), 3010 (w), 1717 (m), 1677-
1
(m), 1394 (m), 1339 (w), 1122 (m), 1106 (m) cm-1; H NMR (300
MHz, CDCl3) δ ) 2.16 (s, 3 H), 2.28 (s, 3 H), 3.63 (q, J ) 6.3 Hz, 2
H), 4.01 (t, J ) 6.3 Hz, 2 H), 6.80 (s, 1 H), 7.74 (m, 2 H), 7.87 (m, 2
H), 7.92 (s, 1 H), 8.10 (q, J ) 6.3 Hz, 1 H); 13C NMR (50 MHz,
CDCl3) δ ) 18.72, 20.91, 36.82, 41.55, 114.36, 123.83 (2C), 125.37,
126.94, 130.75 (2C), 132.26, 134.55 (2C), 143.86, 147.73, 168.65 (2C);
EI-MS (70 eV) m/z 339 (13, M+), 304 (8), 179 (100), 160 (28), 147
(11), 133 (10), 104 (16), 77 (21). Anal. Calcd for C18H17N3O4‚1H2O
(357.36): C, 60.50; H, 5.36; N, 11.76. Found: C, 60.47; H, 4.91; N,
11.75.
(m), 1600 (w), 1581 (m), 1547 (s), 1344 (w), 1261 (w), 1117 (w) cm-1
;
1H NMR (200 MHz, CDCl3) δ 2.45 (s, 3 H), 2.55 (s, 3 H), 3.30 (s, 3
H), 3.92 (t, J ) 5.2 Hz, 2 H), 4.91 (t, J ) 5.2 Hz, 2 H), 7.69 (s, 1 H),
8.04 (s, 1 H), 8.62 (br.s, 1 H); 13C NMR (50 MHz, (CD3)2SO) δ 18.37,
20.28, 43.50, 58.06, 67.83, 116.29, 130.38, 130.98, 133.25, 135.35,
136.70, 145.86, 149.83, 155.12, 159.48; EI-MS (70 eV) m/z 300 (3,
M+), 242 (100, [M - C3H6O]+), 171 (80), 156 (28), 44 (34). Anal.
Calcd for C15H16N4O3‚1/2H2O (309.32): C, 58.25; H, 5.50; N, 18.12.
Found: C, 58.77; H, 5.42; N, 18.03.
7,8-Dimethyl-10-(2′-phthalimidoethyl)-10H-benzo[g]pteridine-
2,4-dione (26). 24 (200 mg, 0.589 mmol) was dissolved in 20 mL of
acetic acid (20 mL) at 50 °C. At this temperature, a suspension of
Pd/C catalyst (15 mg) in AcOH (1 mL) was added to the reaction
solution. The mixture was kept under hydrogen at 50 °C for 30 h.
The colorless solution was subsequently allowed to cool to room
temperature and filtered through a Celite pad. Alloxane monohydrate
(400 mg, 2.50 mmol) and boric acid (800 mg, 12.96 mmol) were added,
and the slurry was stirred at room temperature for an additional 16 h.
The suspension was diluted with 500 mL of water and extracted with
200 mL of CHCl3. The combined organic layers were dried with
MgSO4 and concentrated in vacuo. The resulting yellow oil was
solidified through addition of 200 mL of diethyl ether. The solid
product was filtered off to yield 26 (115 mg, 47%) as a yellow
powder: mp 298-300 °C; IR (CHCl3) 3378 (w), 1772 (w), 1717 (s),
1633 (w), 1583 (m), 1550 (s), 1394 (m), 1244 (m), 1167 (w), 989 (m),
3-Ethyl-10-(2′-methoxyethyl)-7,8-dimethyl-10H-benzo[g]pteridine-
2,4-dione (16). A solution of iodoethane (2.50 g, 16.0 mmol) was
added dropwise to a mixture of 15 (1.09 g, 3.63 mmol) and K2CO3
(3.00 g, 21.7 mmol) 50 mL in DMF. The reaction slurry was stirred
overnight at room temperature, then diluted with 500 mL of water and
extracted with 200 mL of CHCl3 each. The combined organic extracts
were dried with MgSO4, filtrated, and concentrated in vacuo. The
residual material was subjected to flash chromatography on silica gel
(3 × 15 cm, acetone/CH2Cl2 (1:1)) to give 16 as an orange oil.
Recrystallization from MeOH/H2O afforded 16 (0.990 g, 83%) as
orange platelets: mp 204-205 °C; IR (CHCl3) 3000 (w), 1706 (w),
1651 (m), 1583 (m), 1548 (s), 1439 (w), 1339 (w), 1117 (w) cm-1; 1H
NMR (200 MHz, CDCl3) δ ) 1.31 (t, J ) 7.1 Hz, 3 H), 2.43 (s, 3 H),
2.53 (s, 3 H), 3.29 (s, 3 H), 3.91 (t, J ) 5.3 Hz, 2 H), 4.17 (q, J ) 7.1
Hz, 2 H), 4.88 (t, J ) 5.3 Hz, 2 H), 7.65 (s, 1 H), 8.02 (s, 1 H); 13C
NMR (50 MHz, CDCl3) δ ) 12.78, 19.18, 21.25, 36.87, 44.94, 58.93,
69.22, 116.28, 131.85, 131.92, 134.63, 135.39, 136.20, 147.03, 148.40,
155.17, 159.41; EI-MS (70 eV) m/z 328 (4, M+), 270 (100), 242 (23),
199 (32), 171 (22), 44 (16). Anal. Calcd for C17H20N4O3 (328.37): C,
62.18; H, 6.14; N, 17.06. Found: C, 62.23; H, 6.13; N, 17.05.
3-Ethyl-10-(2′-hydroxyethyl)-7,8-dimethyl-10H-benzo[g]pteridine-
2,4-dione (8). A solution of 16 (0.980 g, 2.71 mmol) in CH2Cl2 (50
mL) was cooled to 0 °C. Then BBr3 (3.5 mL, 36.3 mmol) dissolved
in 10 mL of CH2Cl2 was slowly added with a syringe under nitrogen.
The solution was stirred at 0 °C for an additional 4.5 h. A mixture of
H2O (2.5 mL) and acetone (5 mL) was added dropwise and very
carefully to hydrolyze unreacted BBr3. The mixture was stirred for
another 15 min, then 300 mL of acetone and 200 mL of H2O were
added. The organic solvents were removed in vacuo. The residual
aqueous product solution was allowed to stay at 0 °C for 24 h. The
precipitate 8 was removed by filtration and recrystallized from acetone/
H2O. 8 (0.690 g, 81%) was obtained in form of yellow needles: mp:
205-208 °C (dec); IR (CHCl3) 3689 (w), 3011 (s), 1706 (w), 1650
(m), 1583 (m), 1548 (s), 1461 (w), 1439 (w), 1339 (w), 1133 (m), 928
(w), 622 (m) cm-1; 1H NMR (200 MHz, CDCl3) δ 1.27 (t, J ) 7.1 Hz,
3 H), 2.44 (s, 3 H), 2.54 (s, 3 H), 3.00 (t, J ) 5.5 Hz, 1 H), 4.09 (q,
J ) 7.1 Hz, 2 H), 4.21 (t, J ) 5.5 Hz, 2 H), 4.92 (t, J ) 5.5 Hz, 2 H),
7.64 (s, 1 H), 8.02 (s, 1 H); 13C NMR (50 MHz, CDCl3) δ 13.20, 19.67,
21.76, 37.37, 47.39, 60.18, 116.28, 132.05, 132.78, 135.43, 137.08,
137.30, 148.27, 149.45, 155.72, 159.84; EI-MS (70 eV) m/z 314 (1,
M+), 298 (13), 270 (43), 199 (40), 171 (25), 77 (27), 44 (100). Anal.
Calcd for C16H18N4O3‚1H2O (332.34): C, 57.77; H, 6.02; N, 16.85.
Found: C, 57.80; H, 5.62; N, 16.63.
1
894 (m) cm-1; H NMR (300 MHz, (CD3)2SO) δ 2.34 (s, 3 H), 2.40
(s, 3 H), 4.03 (t, J ) 6.0 Hz, 2 H); 4.91 (t, J ) 6.0 Hz, 2 H), 7.79 (m,
4 H), 7.82 (s, 1 H), 7.88 (s, 1 H), 11.26 (s, 1 H); MS (FAB+) 416
(100, [M + H]+), 391 (4), 371 (7), 242 (4), 174 (5), 107 (12). Anal.
Calcd for C22H17N5O3‚2H2O (451.44): C, 58.53; H, 4.69; N, 15.51.
Found: C, 58.70; H, 4.39; N, 15.59.
3-Ethyl-7,8-dimethyl-10-(2′-phthalimidoethyl)-10H-benzo[g]pte-
ridine-2,4-dione (27). Iodoethane (0.75 mL, 9.28 mmol) was added
to a suspension of 26 (240 mg, 0.578 mmol) and dry K2CO3 (800 mg,
5.79 mmol) in 20 mL of DMF. The reaction slurry was stirred for 24
h at room temperature. The mixture was diluted with 100 mL of water
and extracted with 200 mL of CHCl3 three times. The combined
organic layers were dried with MgSO4, filtrated, and concentrated in
vacuo. The crude residue was subjected to flash chromatography with
silica-H adsorbent (3 × 20 cm, CHCl3/MeOH (12:1)). 27 (220 mg,
86%) was obtained as a yellow powder: mp 254-256 °C; IR (CHCl3)
1772 (w), 1716 (s), 1656 (m), 1589 (m), 1549 (s), 1439 (w), 1394 (m),
1333 (w) cm-1; 1H NMR (300 MHz, (CD3)2SO) δ 0.95 (t, J ) 6.9 Hz,
3 H), 2.33 (s, 3 H), 2.39 (s, 3 H), 3.75 (q, J ) 6.9 Hz, 2 H), 4.00 (t,
J ) 5.6 Hz, 2 H), 4.93 (t, J ) 5.6 Hz, 2 H), 7.75 (m, 4 H), 7.84 (s, 1
H), 7.91 (s, 1 H); 13C NMR (75 MHz, (CD3)2SO) δ ) 12.70, 18.60,
20.54, 34.66, 35.71, 41.87, 115.86, 123.13 (2C), 130.78, 131.38, 131.47
(2C), 134.24, 134.49 (2C), 136.08 (2C), 147.03, 149.28, 154.11, 158.96,
167.98 (2C); MS (FAB+) 444 (100, [M + H]+). Anal. Calcd for
C24H21N5O4‚1H2O (461.48): C, 62.47; H, 5.02; N, 15.18. Found: C,
62.59; H, 4.69; N, 15.12.
3-Ethyl-7,8-dimethyl-10-(2′-aminoethyl)-10H-benzo[g]pteridine-
2,4-dione (9). A suspension of 27 (455 mg, 1.03 mmol) in 30 mL of
concentrated HCl was stirred at reflux temperature for 4 h. The reaction
mixture was concentrated in vacuo. The residue material was triturated
with 20 mL of MeOH and filtrated. The solid material was washed
with acetone twice to yield 9 (330 mg, 92%) as a yellow powder: mp
286-288 °C; IR (CHCl3) 3444 (m), 1713 (s), 1615 (s), 1580 (s), 1542
(s), 1458 (m), 1343 (m), 1250 (s), 1193 (m), 1022 (w), 929 (w), 878
(w), 806 (w), 773 (w), 443 (w) cm-1; 1H NMR (300 MHz, (CD3)2SO)
δ 1.16 (t, J ) 6.9 Hz, 3 H), 2.42 (s, 3 H), 2.50 (s, 3 H), 3.20 (m, 2 H),
4,5-Dimethyl-2-nitro-N-(2′-phthalimidoethyl)aniline (24). 4,5-
Dimethyl-2-nitroaniline (10, 3.00 g, 18.1 mmol) and an excess of N-(2-

7450 J. Am. Chem. Soc., Vol. 119, No. 32, 1997
Epple et al.
3.94 (q, J ) 6.9 Hz, 2 H), 4.91 (t, J ) 6.0 Hz, 2 H), 7.98 (s, 1 H), 8.05
(s, 1 H), 8.17 (m, 3 H); 13C NMR (75 MHz, (CD3)2SO) δ 12.82, 18.63,
20.41, 35.93, 35.99, 41.11, 116.05, 130.68, 131.28, 134.30, 136.20,
136.46, 147.31, 149.63, 154.68, 159.25; MS (FAB+) 314 (100,
[M-Cl]+), 271 (15), 232 (7).
(w), 3389 (w), 3022 (w), 1706 (s), 1656 (m), 1600 (s), 1550 (s), 1467
(m), 1406 (w), 1356 (w), 1272 (m), 1056 (w), 972 (w), 922 (w), 850
(w) cm-1; 1H NMR (400 MHz, (CD3)2SO) δ 1.14 (t, J ) 7.1 Hz, 3 H),
2.38 (s, 3 H), 2.50 (s, 3 H), 3.38 (m, 1 H),3.39 (m, 1 H), 3.89 (d, J )
17.5 Hz, 1 H), 3.90 (q, J ) 7.1 Hz, 2 H), 3.99 (d, J ) 17.5 Hz, 1 H),
4.09 (d, J ) 17.8 Hz, 1 H), 4.13 (d, J ) 17.8 Hz, 1 H), 4.29 (m, 1 H),
4.35 (m, 1 H), 4.44 (m, 1 H), 4.54 (m, 1 H), 4.91 (m, 2 H), 5.03 (d, J
) 12.4 Hz, 1 H), 5.08 (d, J ) 12.5 Hz, 1 H), 7.29 (m, 5 H), 7.86 (s,
1 H), 7.39 (s, 1 H), 10.65 (s, 1 H), 10.71 (s, 1 H); 13C NMR (125
MHz, (CD3)2SO) δ 12.78, 18.65, 20.52, 35.84, 38.23, 38.29, 38.91-
40.00 (1C), 42.55, 48.10, 48.25, 59.08, 61.05, 65.96, 115.97,127.58,
127.73, 128.00, 128.26, 128.34, 130.86, 131.01, 133.92, 135.41, 135.86,
136.20, 146.69, 149.09, 151.44, 151.54, 154.51, 158.98, 169.13, 169.16,
169.56, 169.60; HRMS (FAB+) m/z calcd for C35H35N8O10 ([M + H]+)
727.2476, found 727.3066.
1-(Carboxymethyl)uracil (trans-anti)-Cyclobutane Dimer Mono-
benzyl Ester Monoflavin Ester (21): mp 173-175 °C (dec); IR
(CHCl3) 3378 (w), 3289 (w), 1744 (m), 1709 (s), 1672 (m), 1656 (m),
1583 (m), 1549 (s), 1500 (w), 1467 (m), 1406 (w), 1372 (m), 1267
(w) cm-1; 1H NMR (300 MHz, (CD3)2SO) δ 1.14 (t, J ) 6.9 Hz, 3 H),
2.39 (s, 3 H), 2.50 (s, 3 H), 3.47 (m, 1 H), 3.76 (m, 1 H), 3.83 (d, J )
17.5 Hz, 1 H), 3.90 (q, J ) 6.9 Hz, 2 H), 4.01 (d, J ) 17.5 Hz, 1 H),
4.02 (d, J ) 17.8 Hz, 1 H), 4.14 (m, 1 H), 4.22 (d, J ) 17.8, 1 H),
4.35 (m, 1 H), 4.53 (m, 2 H), 4.94 (m, 2 H), 5.18 (m, 2 H), 7.38 (m,
5 H), 7.90 (s, 1 H), 7.91 (s, 1 H), 10.56 (s, 1 H), 10.69 (s, 1 H); MS
(FAB+) 727 (100, [M + H]+).
General Procedure for the Preparation of the 1-(Carboxy-
methyl)uracil Cyclobutane Dimer Bisflavin Esters (1, 17, and 18).
The synthesis of the 1-(carboxymethyl)uracil cyclobutane dimer starting
materials 5-7 was achieved as decribed by us recently.17 BOP (150
mg, 0.323 mmol) was added to a solution of the corresponding cis-
syn, trans-syn, or trans-anti 1-(carboxymethyl)uracil cyclobutane
dimers (5-7, 50 mg, 0.147 mmol) in 1.5 mL of DMF. This solution
was stirred at room temperature for 10 min. Then a solution of 8 (100
mg, 0.318 mmol) in 3 mL of DMF was added to the activated dimer
solution. Ten drops of triethylamine were added to start the reaction,
which was stirred at room temperature for 24 h. The reaction mixture
was diluted with 100 mL of water and extracted with 200 mL of CHCl3
three times. The combined organic layers were dried with MgSO4,
filtrated, and concentrated in vacuo. The remaining orange oil was
solidified through trituration with diethyl ether. The residual material
was subjected to flash chromatography with silica-H as adsorbent (3
× 15 cm, CHCl3/MeOH (10:1)). The yields for the three model
compounds 1, 17, and 18 were 67 mg (49%) for 1, 118 mg (84%) for
17, and 36 mg (27%) for 18. All compounds were obtained as a yellow
powder.
1-(Carboxymethyl)uracil (cis-syn)-Cyclobutane Dimer Bisflavin
Ester (1): HRMS (FAB+) m/z calcd for C44H45N12O12 ([M + H]+)
933.3280, found 933.3320. For all other analytical data see ref 27.
1-(Carboxymethyl)uracil (trans-syn)-Cyclobutane Dimer Bis-
flavin Ester (17): HRMS (FAB+) m/z calcd for C44H45N12O12 ([M +
H]+) 933.3280, found 933.3441. For all other analytical data see ref
27.
1-(Carboxymethyl)uracil (trans-anti)-Cyclobutane Dimer Bis-
flavin Ester (18): HRMS (FAB+) m/z calcd for C44H45N12O12 ([M +
H]+) 933.3280, found 933.3412. For all other analytical data see ref
27.
1-(Carboxymethyl)uracil (cis-syn)-Cyclobutane Dimer Mono-
pentyl Ester Monoflavin Ester (22) was synthesized following the
general procedure described for the preparation of 19-21. Instead of
benzyl alcohol, 1-pentanol was used. 22 was obtained in 12% (23 mg)
yield after chromatography with silica-H: mp 146-148 °C (dec); IR
(CHCl3) 3689 (w), 3389 (w), 1733 (m), 1711 (s), 1650 (m), 1583 (m),
1550 (s), 1467 (m), 1406 (w), 1339 (w), 1267 (m), 1017 (w), 922 (w)
1
cm-1; H NMR (500 MHz, (CD3)2SO) δ 0.85 (t, J ) 7.0 Hz, 3 H),
1.14 (t, J ) 7.0 Hz, 3 H), 1.27 (m, 4 H), 1.56 (quint., J ) 7.0 Hz, 2
H), 2.40 (s, 3 H), 2.50 (s, 3 H), 3.63 (m, 1 H), 3.71 (m, 1 H), 3.73 (d,
J ) 17.4 Hz, 1 H), 3.74 (d, J ) 17.4 Hz, 1 H), 3.91 (q, J ) 7.1 Hz,
2 H), 4.03 (m, 2 H), 4.06 (d, J ) 17.4 Hz, 1 H), 4.10 (d, J ) 17.4 Hz,
1 H), 4.14 (m, 2 H), 4.52 (m, 2 H), 4.81 (m, 1 H), 5.03 (m, 1 H), 7.87
(s, 1 H), 7.93 (s, 1 H), 10.50 (s, 1 H), 10.53 (s, 1 H); 13C NMR (125
MHz, (CD3)2SO) δ 12.84, 13.79, 18.73, 20.60, 21.69, 27.42, 27.68,
35.94, 37.95, 38.38, 42.82, 46.92, 47.14, 54.80, 54.91, 61.41, 64.66,
116.23, 130.94, 131.09, 134.04, 135.98, 136.33, 146.77, 149.22, 152.54,
152.60, 154.75, 159.12, 166.85, 167.06, 168.60, 168.61; HRMS (FAB+)
m/z calcd for C33H39N8O10 ([M + H]+) 707.2789, found 707.2824.
1-(Carboxymethyl)uracil (cis-syn)-Cyclobutane Dimer Bisflavin
Amide (2). The synthesis of 2 was achieved using the general
procedure described for the preparation of 1, 17, and 18. Instead of
the flavin alcohol 8, the flavin amine 9 (100 mg, 0.286 mmol) was
used. The reaction was stirred for only 5 h. The reaction mixture
was worked up, and the residual, solidified material was not subjected
to flash chromatography but recrystallized three times from methanol.
2 was obtained (60 mg, 44%) as a yellow powder: mp: 270-272 °C;
IR (KBr) 3416 (m), 3322 (m), 1704 (s), 1650 (s), 1584 (s), 1547 (s),
1461 (s), 1337 (m), 1267 (m), 1230 (s), 1194 (m), 1172 (m), 1150
General Procedure for the Preparation of 1-(Carboxymethyl)-
uracil Cyclobutane Dimer Monobenzyl Ester Monoflavin Ester
(19-21). A solution of the corresponding (cis-syn, trans-syn, or
trans-anti)-1-(carboxymethyl)uracil cyclobutane dimers (5-7, 100 mg,
0.294 mmol) and an excess of BOP (300 mg, 0.646 mmol) were
dissolved in 4 mL of DMF and stirred at room temperature for 10 min.
After the addition of a solution of 8 (46 mg, 0.147 mmol) in 2 mL
DMF (2 mL) and of 10 drops of triethylamine, the reaction was stirred
for 5 h at room temperature. Then benzyl alcohol (40 mg, 0.38 mmol)
was added to the reaction mixture, which was stirred for an additional
20 h at room temperature. The reaction mixture was diluted with 100
mL of water and extracted with 200 mL of CHCl3 three times. The
combined organic layers were dried with MgSO4, filtrated, and
concentrated in vacuo. The remaining orange oil was solidified through
trituration with diethyl ether. The solid residual material was subjected
to flash chromatography with silica-H as adsorbent (3 × 15 cm, CHCl3/
MeOH (10:1)). 19 was obtained with 21% yield (46 mg), 20 with
21% yield (46 mg) and 21 with 4% yield (9 mg) as a yellow powder.
1-(Carboxymethyl)uracil (cis-syn)-Cyclobutane Dimer Monoben-
zyl Ester Monoflavin Ester (19): mp 153-157 °C (dec); IR (CHCl3)
3389 (w), 2989 (w), 1711 (s), 1650 (m), 1583 (m), 1550 (s), 1467 (m),
1
(m), 1018 (w), 928 (w), 845 (w), 807 (w), 770 (w) cm-1; H NMR
(400 MHz, (CD3)2SO) δ 1.14 (t, J ) 6.6 Hz, 6 H), 2.39 (s, 6 H), 2.49
(s, 6 H), 3.39 (d, J ) 16.7 Hz, 2 H), 3.47 (m, 4 H), 3.62 (m, 2 H), 3.92
(q, J ) 6.6 Hz, 4 H), 4.09 (m, 2 H), 4.11 (d, J ) 17.7 Hz, 2 H), 4.63
(m, 4 H), 7.90 (s, 2 H), 7.94 (s, 2 H), 8.30 (m, 2 H), 10.43 (s, 2 H);
13C NMR (125 MHz, (CD3)2SO) δ 12.88, 18.72, 20.82, 35.66, 35.95,
38-42 (1C), 42.92, 48.08, 54.88, 115.99, 130.90, 131.07, 134.11,
135.93, 136.24, 146.75, 149.01, 152.37, 154.68, 159.13, 167.30, 168.36;
HRMS (FAB+) m/z calcd for C44H47N14O10 ([M + H]+) 931.3599, found
931.3593.
1-(Carboxymethyl)uracil (cis-syn)-Cyclobutane Dimer Mono-
pentyl Amide Monoflavin Amide (28). 28 was synthesized following
the procedure for the preparation of 22. The flavin amine 9 was used
instead of the flavin alcohol 8. The obtained crude orange oil was
dissolved in MeOH and filtered to remove the bisflavin amide side
product 2. The obtained filtrate was subjected to flash chromatography
with silica-H as the absorbent (3 × 20 cm, CHCl3/MeOH (7:1 f 5:1)).
1
1406 (w), 1339 (w), 1267 (m, 1056 (w), 1017 (w), 922 (w) cm-1; H
NMR (500 MHz, (CD3)2SO) δ 1.11 (t, J ) 7.1 Hz, 3 H), 2.38 (s, 3 H),
2.49 (s, 3 H), 3.63 (m, 1 H), 3.71 (m, 1 H), 3.72 (d, J ) 17.4 Hz, 1 H),
3.83 (d, J ) 17.4 Hz, 1 H), 3.89 (q, J ) 7.1 Hz, 2 H), 4.09 (d, J )
17.4 Hz, 1 H), 4.13 (d, J ) 17.4 Hz, 1 H), 4.14 (m, 2 H), 4.52 (m, 2
H), 4.79 (m, 1 H), 5.03 (m, 1 H), 5.13 (s, 2 H), 7.35 (m, 5 H), 7.86 (s,
1 H), 7.91 (s, 1 H), 10.49 (s, 1 H), 10.57 (s, 1 H); 13C NMR (125
MHz, (CD3)2SO) δ 12.85, 18.73, 20.62, 35.95, 37.90, 38.45, 42.84,
46.96, 47.13, 54.80, 54.98, 61.44, 66.14, 116.23, 127.88 (2C), 128.12,
128.43 (2C), 130.96, 131.10, 134.06, 135.65, 136.00, 136.32, 146.79,
149.21, 152.57, 152.64, 154.76, 159.14, 166.86, 167.09, 168.55, 168.61;
HRMS (FAB+) m/z calcd for C35H35N8O10 ([M + H]+) 727.2476, found
727.2527.
1-(Carboxymethyl)uracil (trans-syn)-Cyclobutane Dimer Mono-
benzyl Ester Monoflavin Ester (20): mp 201 °C; IR (CHCl3) 3500

FlaVin-Containing DNA-Repair Model Compounds
J. Am. Chem. Soc., Vol. 119, No. 32, 1997 7451
28 (51 mg, 27%) was obtained as a yellow powder: mp 197-199 °C
Type No.111-OS, 10 mm light path) equipped with a magnetic stirrer
and a rubber septum. The solution was purged for 20 min with nitrogen.
Then 25 µL (50 µL for diflavin compounds) of a sodium dithionite
solution (50 mg in 5 mL aqueous buffer) was added. For reduction
by catalytic hydrogenation, small amounts of reduction catalyst (Pd
on Barium sulfate, Fluka) were added into the cuvette and the reaction
mixture was stirred and purged first with nitrogen for 5 min, then with
hydrogen for additional 20 min. Afterward the reaction mixture was
allowed to stand under hydrogen at room temperature for another 5
min. The disappearance of the fluorescence of the solution (excitation
at 366 nm, detection at 520 nm) was checked before and after the
reduction to ensure complete reduction. For the pH-dependent
measurements 0.5 M buffers were prepared. Citrate buffers were used
for pHs 3, 4, 5, and 6, Tris.HCl buffers were employed for pHs 7, 8,
and 9, and a phosphate buffer was used for pHs 10 and 11. Other
substances added to the reaction mixtures such as acetic acid or
triethylamine were added prior to the reduction. The irradiation of
the cuvettes containing 3 mL samples was performed at 366 or 400
nm in a fluorescence spectrometer Spex 1680 equipped with a 0.22 m
double-grating monochromator, a 450 W xenon lamp with a band-
pass of 3.4 nm, and a magnetic stirrer in the sample compartment. All
samples were stirred during the irradiation. During a series of
measurements, the intensity of the light beam was repetitively
determined by following the recommended procedures for potassium
ferrioxalate actinometry.44,45 The error in the number of light quants
emitted from the lamp was determined to be (10%. Separation of the
photoproduct and the starting material was achieved on an analytical
Lichrosphere (5/100, Merck) HPLC column (4 mm × 250 mm) by
using 55% A and 45% B over 9 min to elute the photoproduct. Then,
switching to 80% A and 20% B for additional 19 min caused the elution
of the starting material (A, methanol; B, water). The peak areas were
visualized and integrated by Knaur Eurochrom 2000 software. To
eliminate inaccuracies of peak area stemming from fluctuations of the
detection lamp, the determined peak areas of the photoproduct and the
starting material were devided by the total peak area determined in the
chromatogram (peak area photoproduct plus peak area starting material).
The increase of photoproduct and the decrease of starting material were
plotted against time and the obtained data fitted using a monoexpo-
nential decline or rise to maximum function with the program
DeltaGraph Pro 3.5. To determine the experimental error of our
measurements, the cleavage reaction of the model compound 28 in
ethylene glycol with dithionite reduction was measured several times
at different days. The obtained quantum yields were identical within
an experimental error of (8%.
(dec); IR (CHCl3) 3333 (m), 3222 (m), 1706 (s), 1650 (m), 1583 (m),
1
1550 (s), 1467 (m), 1339 (w), 1272 (m), 1017 (w), 928 (w) cm-1; H
NMR (500 MHz, (CD3)2SO) δ 0.85 (t, J ) 7.0 Hz, 3 H), 1.14 (t, J )
7.0 Hz, 3 H), 1.24 (m, 4 H), 1.37 (quint., J ) 7.0 Hz, 2 H), 2.41 (s, 3
H), 2.50 (s, 3 H), 3.02 (q, J ) 7.0 Hz, 2 H), 3.34 (d, J ) 17.4 Hz, 1
H), 3.44 (d, J ) 17.4 Hz, 1 H), 3.49 (m, 2 H), 3.63 (m, 2 H), 3.92 (q,
J ) 7.1 Hz, 2 H), 4.01 (m, 1 H), 4.11 (m, 1 H), 4.12 (d, J ) 16.5 Hz,
1 H), 4.13 (d, J ) 16.5 Hz, 1 H), 4.58 (m, 1 H), 4.74 (m, 1 H), 7.90
(s, 1 H), 7.92 (t, J ) 7.0 Hz, 1 H), 7.96 (s, 1 H), 8.25 (t, J ) 7.0 Hz,
1 H), 10.39 (s, 1 H), 10.42 (s, 1 H); 13C NMR (125 MHz, (CD3)2SO)
δ 12.89, 13.87, 18.74, 20.82, 21.79, 28.52, 28.75, 35.74, 35.96, 38.19,
38.51 (2C), 42.99, 48.08, 48.14, 54.49, 55.35, 116.04, 130.95, 131.09,
134.13, 135.95, 136.28, 146.75, 149.11, 152.39, 152.42, 154.73, 159.17,
166.95, 167.14, 167.55, 168.40; HRMS (FAB+) m/z calcd for
C33H41N10O8 ([M + H]+) 705.3109, found 705.3126.
1-(Carboxymethyl)uracil Flavin Ester (3). 1-(Carboxymethyl)-
uracil was synthesized according to the literature.33 A solution of
1-(carboxymethyl)uracil (100 mg, 0.588 mmol) and (300 mg, 0.646
mmol) of BOP in 3 mL of DMF was stirred at room temperature for
10 min. After the addition of a solution of 8 (185 mg, 0.589 mmol) in
DMF (3 mL), 10 drops of triethylamine were added and the reaction
mixture was stirred for an additional 18 h at room temperature. The
reaction mixture was diluted with 100 mL of water and extracted three
times with 100 mL of CHCl3. The combined organic layers were dried
with MgSO4, filtrated, and concentrated. The remaining orange oil
was solidified through addition of diethyl ether. The residual material
was subjected to flash chromatography with silica-H as adsorbent (3
× 15 cm, CHCl3/MeOH (10:1)) to yield 3 (165 mg, 60%) as an orange
powder: mp 190-192 °C; IR (CHCl3) 1756 (w), 1694 (s), 1650 (m),
1
1583 (m), 1550 (s), 1461 (w), 1339 (w), 1189 (w) cm-1; H NMR
(400 MHz, (CD3)2SO) δ 1.16 (t, J ) 7.0, 3 H), 2.41 (s, 3 H), 2.50 (s,
3 H), 3.93 (q, J ) 7.0, 2 H), 4.39 (s, 2 H), 4.58 (t, J ) 5.3 Hz, 2 H),
4.94 (t, J ) 5.3 Hz, 2 H), 5.00 (d, J ) 7.9 Hz, 1 H), 7.51 (d, J ) 7.9
Hz, 1 H), 7.87 (s, 1 H), 7.94 (s, 1 H), 11.29 (s, 1 H); 13C NMR (100
MHz, (CD3)2SO) δ 12.85, 18.73, 20.62, 35.94, 42.70, 48.27, 61.64,
101.03, 116.08, 130.95, 131.07, 134.03, 136.03, 136.20, 145.43, 146.87,
149.15, 150.77, 154.70, 159.08, 163.48, 168.11; MS (FAB+) m/z 466
(100, [M + H]+), 391 (13), 371 (28), 329 (11).
1-(Carboxymethyl)uracil Flavin Amide (4) was synthesized in
analogy to 3 with the flavin amine 9 instead of the flavin alcohol 8. 4
was obtained in 35% yield as an orange powder: mp >265 °C (dec);
IR (CHCl3) 1694 (s), 1650 (m), 1583 (m), 1550 (s), 1456 (w), 1339
(w), 1194 (m), 1118 (w), 1156 (w) cm-1; 1H NMR (400 MHz, (CD3)2-
SO) δ 1.15 (t, J ) 6.9, 3 H), 2.41 (s, 3 H), 2.49 (s, 3 H), 3.51 (q, J )
5.6 Hz,2 H), 3.93 (q, J ) 6.9, 2 H), 4.20 (s, 2 H), 4.68 (t, J ) 5.6 Hz,
2 H), 5.55 (d, J ) 7.9 Hz, 1 H), 7.44 (d, J ) 7.9 Hz, 1 H), 7.85 (s, 1
H), 7.95 (s, 1 H), 8.37 (t, J ) 5.6 Hz, 1H), 11.25 (s, 1 H); 13C NMR
(100 MHz, (CD3)2SO) δ 12.89, 18.75, 20.80, 35.97, 40.30, 42.91, 49.27,
100.70, 115.92, 130.93, 131.07, 134.14, 135.94, 136.25, 146.22, 146.79,
149.14, 150.91, 154.71, 159.17, 163.81, 167.78; MS (FAB+) m/z 467
(37, [M + H]+),447 (100), 391 (16), 371 (24).
Acknowledgment. We thank Prof. F. Diederich for the
generous support of this research and Prof. U. Wild for enabling
us to perform the GAMESS(US) calculations (E.-U. W.). This
work was supported by the Schweizer Nationalfond and by the
German Fonds der Chemischen Industrie with a Liebig-
Fellowship to T.C. We thank J. Wytko-Weiss and R. Tykwinsky
for a carefull inspection of the manuscript.
Irradiation Experiments, Determination of the Quantum Yield,
and Data Evaluation. Solutions (10-4 M) of the model compound in
the corresponding solvents were prepared in quartz cuvettes (Hellma
JA964097U