ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS
Vol. 342, No. 2, June 15, pp. 275–281, 1997
Article No. BB970124
Participation of Reactive Oxygen Species in Phototoxicity
Induced by Quinolone Antibacterial Agents
Naoki Umezawa,* Kumi Arakane,† Akemi Ryu,† Shinro Mashiko,‡
Masaaki Hirobe,* and Tetsuo Nagano*,1
*Faculty of Pharmaceutical Sciences, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113, Japan;
†Research Laboratory, KOSÉ Corporation, 1-18-4 Azusawa, Itabashi-ku, Tokyo 174, Japan; and
‡Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe 651-24, Japan
Received December 31, 1996, and in revised form March 24, 1997
To elucidate the mechanism of phototoxicity in-
duced as a side effect by some of the new quinolone
antibiotics, we studied sparfloxacin (SPFX), lomeflox-
acin, enoxacin, ofloxacin, and ciprofloxacin. We first
examined the photosensitized formation of reactive
oxygen species such as singlet oxygen (1O2) and super-
oxide anion (O0 2 ) mediated by the new quinolones. Al-
though a large number of studies have been reported,
there is no direct evidence that these drugs generate
reactive oxygen species. We employed a near-infrared
emission spectrometer to detect 1O2-specific emission
(1268 nm), and the nitroblue tetrazolium reduction
method to detect O0 2 . All the quinolones investigated
in this study were found to produce 1O2 . Four drugs,
but not SPFX, produced O0
2 . We also examined photo-
dynamic DNA strand-breaking activity as a possible
mechanism to explain the participation of reactive ox-
ygen species in the phototoxicity of the drugs. All the
drugs exhibited photodynamic DNA strand-breaking
activity. The inhibitory effect of scavengers of reactive
oxygen species indicated that the main active species
was 1O2 . The DNA strand-breaking activity was corre-
lated not with the 1O2-forming ability, but with the af-
finity of the drugs for DNA. This result may be due to
the short lifetime of 1O2 . These data suggested that
the phototoxicity of the new quinolones was related to
DNA damage caused by reactive oxygen species, espe-
cially 1O2 . q 1997 Academic Press
Key Words: new quinolones; phototoxicity; singlet ox-
ygen; superoxide anion; photodynamic DNA strand-
breaking activity.
New quinolones with potent anti-bacterial activity
based on the inhibition of bacterial DNA gyrase (1,
1
To whom correspondence should be addressed. Fax: /81-3-5684-
2395.
0003-9861/97 $25.00
Copyright q 1997 by Academic Press
All rights of reproduction in any form reserved.
AID ABB 0124 / 6b38$$$161 05-09-97 23:02:51
2) are used clinically for general bacterial infectious
diseases. However, some of them have been reported
to induce phototoxicity as a side effect (3), followed by
erythema, pigmentation, and hardening or peeling of
skin. Some studies of structure –activity relationships
have been made on quinolone-induced phototoxicity (4,
5), and it appears that the phototoxicity is determined
by the nature of the 8-position substituent. Halogen
caused the greatest photoreactive activity, while hy-
drogen and methoxy substituents had little effect.
However, the mechanism of the phototoxicity has not
been established. Since reactive oxygen species may
be involved, we investigated the generation of such
species by five commercially available quinolone anti-
biotics (Fig. 1), i.e., sparfloxacin (SPFX),2 lomefloxacin
(LFLX), enoxacin (ENX), ofloxacin (OFLX), and ci-
profloxacin (CPFX).
Many drugs cause phototoxicity, and some of them
are known to produce 1O2 photodynamically (6, 7). Na-
lidixic acid, a so-called old quinolone (Fig. 1) whose
structure is similar to those of the new quinolones but
lacks fluorine, is a photosensitizer (8, 9). Some of new
quinolones are also reported as photosensitizers (10).
In addition, quinolone-induced membrane lipid peroxi-
dation (11) and cytotoxicity in cultured cells (12) have
been reported, and these effects are inhibited by vari-
ous scavengers of reactive oxygen species. It has also
been reported that pretreatment of Balb/c mice with
scavengers of reactive oxygen species provided signifi-
cant protection against the ear-swelling reaction in-
duced by oral administration of quinolones followed by
2
Abbreviations used: SPFX, sparfloxacin; LFLX, lomefloxacin;
ENX, enoxacin; OFLX, ofloxacin; CPFX, ciprofloxacin; Hp, hemato-
porphyrin; NBT, nitroblue tetrazolium; SOD, superoxide dismutase;
DABCO, 1,4-diazabicyclo-[2,2,2]-octane; DMSO, dimethyl sulfoxide;
OC, open circular; CC, closed circular.
275
arca
276 UMEZAWA ET AL.
FIG. 1. Chemical structures of quinolones.
UVA irradiation (13). Therefore, it is considered that
reactive oxygen species participate in the phototoxicity
of the drugs. Scheme 1 shows possible reactions of trip-
let excited drugs with oxygen. However, there is no
report of the direct and quantitative detection of the
generation of reactive oxygen species mediated by the
new quinolones. Moreover, since there is some doubt
about the specificity of the scavengers of reactive oxy-
gen species, it is still unclear what kind of oxidants
are generated.
In this work, we examined this point by measuring
the emission at 1268 nm accompanying the O2 (1Dg) r
O2 (3Sg0) transition. Figure 2 shows our experimental
setup (see Experimental Procedures). This emission is
highly specific to 1O2 . Using this method, it was possi-
SCHEME 1. Possible reactions of triplet excited drugs with oxy-
gen. Sens, photosensitizer; D, electron donor.
AID ABB 0124 / 6b38$$$162 05-09-97 23:02:51
FIG. 2. Near-infrared emission spectrometer.
ble to detect 1O2 quantitatively and to compare the 1O2-
forming ability (14). We also used the nitroblue tetrazo-
lium (NBT) reduction method to detect O20 ; this method
is widely used and can measure O20 quantitatively (15).
We also investigated photodynamic DNA strand-
breaking activity as a model of phototoxic injury, since
many phototoxic compounds have this activity (16–19).
It is known that the new quinolones interact with DNA
(1, 20, 21), and some of them, i.e., tosufloxacin and
ENX, showed strand-breaking activity (22). In that
study, a fluorescent lamp (Toshiba, mellow white FL
15N) with a broad emission spectrum from 350 to 750
nm was used, but quinolones have weak absorbance in
this range. Therefore, we used a UVA lamp to provide
sufficient spectral overlap for excitation.
It is worth noting that the extreme photosensitive
reaction seen in patients with xeroderma pigmentosum
is associated with a congenital defect of DNA repair
mechanisms (23).
EXPERIMENTAL PROCEDURES
Chemicals. Ofloxacin was provided by Daiichi Pharmaceutical
Co., Ltd. (Tokyo), lomefloxacin by Shionogi & Co., Ltd. (Osaka, Ja-
pan), ciprofloxacin by Bayer Yakuhin Co., Ltd. (Shiga, Japan), and
sparfloxacin and enoxacin by Dainippon Pharmaceutical Co., Ltd.
(Osaka, Japan). b-Carotene, superoxide dismutase (SOD; Cu/Zn
form), catalase, NaN3 , 1,4-diazabicyclo-[2,2,2]-octane (DABCO), and
mannitol were purchased from Wako Pure Chemical Industries, Ltd.
Plasmid pBR 322 DNA was purchased from Wako Pure Chemical
Industries, Ltd., and calf thymus DNA (sodium salt, type I) was from
Sigma Chemical Co.
Measurement of 1O2 emission. Our experimental setup (Fig. 2) con-
sisted of an Ar laser (Innova 70-4; Coherent Inc., U.S.A.) with a liquid
nitrogen-cooled near-infrared Ge-detector (model 403HS; Applied De-
tector Co., U.S.A.) connected to the exit slit of a monochromator (Model
CT10; JASCO, Japan) having a blaze wavelength at 1250 nm to mini-
mize the grating loss. An IR-80 cutoff filter with 0% transmittance at
arca
 REACTIVE OXYGEN IN QUINOLONE-INDUCED PHOTOTOXICITY 277

less than 750 nm and 35% transmittance at 800 nm was placed at

the entrance slit of the monochromator. A collecting lens focused the
monochromator output onto the detector crystal. The Ar laser output
was chopped at 800 Hz by an acousto-optic modulator (A-160; Hoya,
Japan) driven by a driver (110-DS; Hoya). The laser light contained
three wavelengths in the UVA range (334.0, 351.1, and 363.8 nm).
The signal output from the Ge-detector was fed to a Model 124A
lock-in amplifier via a Model 116 preamplifier (both from E. G. & G.
Princeton Applied Research, U.S.A.) and synchronized with the inter-
nal reference signal of the lock-in amplifier. The signal output from
the lock-in amplifier was fed to an XY recorder, and the emission
spectrum was recorded by scanning the grating with a motor. In order
to minimize photobleaching of the drugs, the solution was circulated
by a peristaltic pump through a quartz flow cell (3 1 3 mm).
NBT reduction method. The assay medium contained 0.24 mM
NBT, 0.1 mM EDTA, and new quinolones in 50 mM sodium phosphate
buffer (pH 6.5). The medium (10 ml) was exposed to sunlight for 1.5
h. These experiments were done at the same time and at the same
place (in Tokyo, during fine weather in February). Diformazan gener-
ated by reduction of NBT was quantitated in terms of the absorbance
at 560 nm.
SOD was added to the assay medium to examine whether the
formation of diformazan was due to O0 2 .

Photodynamic DNA strand-breaking activity. Plasmid DNA stock

solution was 0.3 mg/6 ml pBR322 plasmid DNA in 10 mM Tris–HCl
buffer (pH 7.9) containing 1 mM EDTA.
Five microliters of a new quinolone solution (final concentration
10–1000 mM) in dimethylsulfoxide (DMSO) and 9 ml of the buffer
were added to 6 ml of DNA stock solution in 1.5-ml Eppendorf tubes.
The reaction mixtures were irradiated with a UVA lamp (Toshiba;
FL 20S BLB) for 1 h at room temperature, at a distance of 10 cm
from the samples. After irradiation, 5 ml of 0.1% bromophenol blue
in 30% glycerol in TBE buffer (89 mM Tris, 89 mM boric acid, 2
mM EDTA, pH 8.0) was added to the mixture. Electrophoresis was
performed with 1.0% agarose gel containing 0.5 mg/ml ethidium bro-
mide. Fifteen-microliter aliquots were loaded onto agarose gel and
electrophoresed in 0.51 TBE buffer at 70 V for 2 h. After migration,
the gels were photographed under irradiation with UV light (UVP
dual-intensity transilluminator) through a red filter.
FIG. 4. Near-infrared emission spectra of LFLX under various con-
ditions. These spectra were measured under the following conditions:
(a) 0.5 mM LFLX in CH3OH, (b) 0.5 mM LFLX in CH3OH after N2
bubbling for 1 h, (c) 0.5 mM LFLX in CH3OD, (d) 0.5 mM LFLX
in CHCl3 :CH3OH (1:1), (e) 0.5 mM LFLX and 5 mM b-carotene in
CHCl3 :CH3OH (1:1). Laser power, 80 mW (a–c), 35 mW (d and e);
light source, Ar laser; wavelength, 334.0, 351.1, and 363.8 nm (UVA
range).
In the experiment with scavengers of reactive oxygen species, 5 ml
of a new quinolone solution (final concentration 1 mM) in DMSO and
9 ml of the buffer containing a scavenger were added to 6 ml of DNA
stock solution in 1.5-ml Eppendorf tubes. The final concentrations
of scavengers were as follows: SOD, 100 mM (O0 2 scavenger); catalase,
10 mM (H2O2 scavenger); NaN3 , 100 mM (1O2 scavenger); DABCO,
100 mM (1O2 scavenger); and mannitol, 100 mM (•OH scavenger).
The resulting photographs were scanned into a Macintosh com-
puter and expressed numerically using the NIH Image program (de-
veloped at the U.S. National Institutes of Health and available from
the Internet by anonymous FTP from zippy.nimh.nih.gov or on floppy
disk from the National Technical Information Service, Springfield,
VA, Part No. PB95-500195GEI), a public domain image processing
and analysis program. Relative proportions of open circular (OC)
form DNA and closed circular (CC) form DNA were calculated from
the peak areas. Induction rate and inhibition rate were calculated
by means of the following equations (16),

d0b
induction rate (%) Å 1 100,
a
e0c
inhibition rate (%) Å 1 100,
a0c
FIG. 3. Near-infrared emission spectra of the quinolones. These
spectra were measured under the following conditions: (a) 0.5 mM
SPFX in CHCl3 , laser power 40 mW; (b) 0.5 mM LFLX in CH3OH, where a is the CC form (%) after photoirradiation in the absence of
laser power 90 mW; (c) 0.05 mM ENX in CHCl3 , laser power 40 mW; photosensitizer, b is the OC form (%) after photoirradiation in the
(d) 0.25 mM OFLX in CHCl3 , laser power 40 mW; (e) 0.5 mM CPFX absence of photosensitizer, c is the CC form (%) after photoirradiation
in CH3OH, laser power 75 mW. Light source, Ar laser; wavelength, in the presence of photosensitizer, d is the OC form (%) after photoir-
334.0, 351.1, and 363.8 nm (UVA range). radiation in the presence of photosensitizer, and e is the CC form

AID ABB 0124 / 6b38$$$162 05-09-97 23:02:51 arca

278 UMEZAWA ET AL.

in the medium. When the solvent was changed from

FIG. 5. Concentration–emission intensity (1268 nm) curves of the
new quinolones (closed circle, LFLX; open square, CPFX). These mea-
surements were carried out under the following conditions: solvent,
CH3OH; light source, Ar laser; wavelength, 334.0, 351.1, and 363.8
nm (UVA range); laser power, 85 mW.
CH3OH to CH3OD (a solvent known to slow down the
nonradiative inactivation of 1O2 , thus prolonging its
lifetime), the emission increased (Fig. 4c). This method
(deuterium isotope effect) is used frequently to confirm
the presence of 1O2 . The emission at 1268 nm was also
inhibited by the addition of a 1O2 scavenger, b-carotene.
Figure 4d shows the emission of LFLX in CH3OH:CHCl3
(1:1) excited with UVA. As b-carotene is very sparingly
soluble in CH3OH, CH3OH:CHCl3 (1:1) was used. When
5 mM b-carotene was added, the emission decreased
(Fig. 4e). The decrease of the emission intensity was
dependent on the concentration of b-carotene (data not
shown). These results confirmed that the emission at
1268 nm did arise from 1O2 . This means that these new
quinolones are photosensitizers.
Figure 5 shows the dependency of the emission inten-
sity at 1268 nm on the concentration of the quinolones.

(%) after photoirradiation in the presence of photosensitizer and

scavenger.
DNA-binding affinity. Calf thymus DNA was purified by ultra-
centrifugation in 10 mM Tris–HCl buffer (pH 7.9) containing 1 mM
EDTA (40,000g for 30 min). The concentration of this solution was
determined by measuring the absorbance at 260 nm (in this case,
13.76 mM). We used these values: OD260 Å 50 mg/ml, molecular weight
per nucleotide 330.
The percentage of drugs bound to DNA was determined by a mem-
brane filtration method using Centrifree micropartition devices
(Millipore; UFC3LCC) that have thin membranes (cellulose) with an
Mr cutoff of 5000. Binding mixtures containing 0.1 mM new quino-
lones and calf thymus DNA (0.344–6.88 mM) were denatured at
907C for 15 min and then gradually cooled to 47C. After 1 night, the
mixtures were transferred to the Centrifree devices and centrifuged
at 5000g at 47C. As the molecular weight of calf thymus DNA was
more than 5000, drugs bound to DNA could not pass through the
membrane. The absorbance of free drugs was measured, and the
concentration of free drugs and the percentage of bound drugs to
DNA were calculated. The results were corrected for the concentra-
tion of drugs bound to the membrane itself.

RESULTS AND DISCUSSION

1
O2 Formation
We measured near-infrared emission from laser-ex-
cited solutions of the new quinolones in CH3OH (LFLX,
CPFX) or CHCl3 (SPFX, ENX, OFLX). The 1O2 emission
at 1268 nm was detected from all the new quinolones
used in this study (Fig. 3), though the baseline was
slanting at various angles, possibly owing to the fluo-
rescence of the quinolones. FIG. 6. (a) O0 2 -forming ability of new quinolones (open circle, SPFX;
In order to confirm that the emission arose from 1O2 , closed circle, LFLX; open triangle, ENX; closed triangle, OFLX; open
we conducted further experiments. Figure 4a shows the square, CPFX). (b) Effect of SOD on NBT reduction mediated by the
near-infrared emission spectrum of LFLX in CH3OH. new quinolones (open circle, SPFX; closed circle, LFLX; open trian-
gle, ENX; closed triangle, OFLX; open square, CPFX). Assay media
The emission disappeared when nitrogen gas was bub- contained 200 mM new quinolone, 0.24 mM NBT, and 0.1 mM EDTA
bled into the solution to exclude O2 (Fig. 4b). This result in 50 mM sodium phosphate buffer (pH 6.5) and were exposed to
indicated that the emission depended on O2 dissolved sunlight for 1.5 h.

AID ABB 0124 / 6b38$$$162 05-09-97 23:02:51 arca

 REACTIVE OXYGEN IN QUINOLONE-INDUCED PHOTOTOXICITY 279

FIG. 7. Photodynamic DNA strand-breaking activities of 1 mM new

quinolones and hematoporphyrin.

The emission mediated by LFLX and CPFX increased

FIG. 8. Dose-dependent increase of OC form (%) generated by pho-
in a dose-dependent manner. The emission of SPFX, todynamic action of the new quinolones and hematoporphyrin (open
ENX, and OFLX could not be detected in CH3OH. Be- circle, SPFX; closed circle, LFLX; open triangle, ENX; closed triangle,
cause the lifetime of 1O2 in CHCl3 was longer than in OFLX; open square, CPFX; closed square, Hp).
CH3OH, the emission of SPFX, ENX, and OFLX was
detected. The 1O2-forming ability of the new quinolones
used in this study was as follows; LFLX, CPFX ú dynamic DNA strand-breaking activity (Fig. 7). From
SPFX, ENX, OFLX. their mobilities on a gel, the forms were identified as
supercoiled CC, and nicked OC. There was no activity
O20 Formation in the absence of drugs or irradiation. Hematoporph-
We used the NBT reduction method to detect O20 with yrin (Hp), which generates 1O2 in a high yield, had low
EDTA as an electron donor. In vivo, the electron donors activity. Figure 8 shows the dose-dependent increase of
may be various amines or electron transfer systems. the photodynamic DNA strand-breaking activity. The
The sample solution was exposed to sunlight. The ab- order of photodynamic DNA strand-breaking activity
sorbance of diformazan at 560 nm increased in a dose- of the new quinolones was as follows: SPFX ú ENX ú
dependent manner, while no change was observed CPFX ú LFLX ú OFLX.
without irradiation. We added various scavengers of reactive oxygen spe-
Figure 6-a shows the production of O0
2 mediated by the cies to the reaction mixtures, in order to identify the
five quinolones. All of them, except SPFX, produced O20 main active species. We used SOD (O20 scavenger), cata-
in a dose-dependent manner. The order of O20-forming lase (H2O2 scavenger), NaN3 , DABCO (1O2 scavengers),
ability was as follows: OFLX ú ENX ú CPFX ú LFLX and mannitol (•OH scavenger). Each scavenger was
ú SPFX. It is interesting that SPFX did not produce added in excess. Table I summarizes the inhibition by
O20 at all, but the reason why is unknown. the scavengers. NaN3 and DABCO strongly inhibited
In order to examine whether the NBT was reduced the DNA strand-breaking activity. These results indi-
by O20 , SOD was added to the sample solution. Figure cated that the main active species for DNA strand-
6b shows that SOD inhibited the reduction almost com-
pletely. This result suggested that NBT was mainly
reduced by O20 . TABLE I
The known order of phototoxicity is SPFX ú LFLX, Inhibition of Photodynamic DNA Strand-Breaking Activity
ENX ú OFLX ú CPFX (5, 12, 24, 25). The 1O2-forming by Various Scavengers of Reactive Oxygen Species
or the O20-forming abilities of the new quinolones did
not correlate directly with the order of phototoxicity. Inhibition (%)
Thus the question arises, do reactive oxygen species
participate in the phototoxicity? SOD Catalase NaN3 DABCO Mannitol
(100 mM) (10 mM) (100 mM) (100 mM) (100 mM)
Photodynamic DNA Strand-Breaking Activity SPFX —a 5.1 84.0 86.1 0.5
We next examined the photodynamic DNA strand- LFLX —a 18.3 69.9 68.9 —a
ENX 0.3 1.3 72.5 92.5 1.1
breaking activity to estimate the role of reactive oxygen
OFLX —a 60.9 105.7 98.4 —a
species in the toxicity. Plasmid DNA was irradiated CPFX —a 12.7 69.6 85.6 8.8
with a UVA lamp in the presence of 1 mM (final concen-
tration) quinolones. All the new quinolones had photo- a
No inhibition was observed.

AID ABB 0124 / 6b38$$$163 05-09-97 23:02:51 arca

280 UMEZAWA ET AL.
FIG. 9. Binding affinity of the new quinolones to calf thymus DNA
(open circle, SPFX; closed circle, LFLX; open triangle, ENX; closed
triangle, OFLX; open square, CPFX).
breaking was 1O2 . Since it has already been shown that
all the new quinolones used in this study produce 1O2
photodynamically, this conclusion is not unexpected.
But the DNA strand-breaking activity was not corre-
lated with the 1O2-forming ability of each drug, and
hematoporphyrin had an especially low activity. We
hypothesized that since 1O2 has quite a short lifetime
and a short distance of diffusion, the binding affinity
of the drugs to DNA could be an important factor for
the phototoxicity.
Binding Affinity of the New Quinolones to DNA
Figure 9 shows the relationship between the percent-
age of bound drug to DNA and the concentration of
DNA, evaluated by a membrane filtration method. We
fixed the concentration of the quinolones and changed
the concentration of DNA. The percentage of bound
drug to DNA was dependent on the concentration of
DNA and was considered to reflect the binding affinity
to DNA. The order of binding affinity to DNA was as
follows: SPFX ú ENX, CPFX ú LFLX, OFLX. This
order is consistent with the above hypothesis.
The mechanism of phototoxicity is very complex.
Skin is composed of many cell types, and many factors
can contribute to toxicity, for example, distribution of
the drugs to skin. It was reported that SPFX remained
in rat skin for a relatively long time compared with
other new quinolones (26), and the concentration of
CPFX in rat skin was much lower compared with the
others (27). The accumulation in skin may be relevant
to the high phototoxicity of SPFX, while the low concen-
tration of CPFX in skin could correlate with the low
phototoxicity of CPFX. In addition to the distribution
of the drugs to skin, other properties such as stability
to light, lifetime of the triplet excited state, and triplet
AID ABB 0124 / 6b38$$$163 05-09-97 23:02:51
quantum yield may affect the phototoxicity of the
drugs. These new quinolones are quite unstable in sun-
light (data not shown). And our attempts to measure
the lifetime of the triplet excited state were unsuccess-
ful. We could not detect the absorption of the triplet
excited states, possibly because of the instability of the
new quinolones under UV irradiation (YAG laser, 355
nm). Moreover, intracellular distribution, antioxida-
tive activity of cells, effective wavelength range of irra-
diation, and irradiation period may also affect the pho-
totoxicity.
In summary, we have demonstrated that the new
quinolones produce 1O2 and O20 upon exposure to light.
Our results indicated that reactive oxygen species, es-
pecially 1O2 , generated by photoreaction of the new
quinolones can damage DNA. Although this was an in
vitro reaction, such a reaction may well occur in the
human body. The order of reactive oxygen-forming abil-
ity was not correlated with the order of phototoxicity.
But some of the factors shown in the previous para-
graph may contribute to this lack of correlation. The
results of DNA strand-breaking activity suggested it.
We speculate that 1O2 could play an important role in
phototoxicity caused as a side effect of these antibiotics.
REFERENCES
1. Shen, L. L., and Pernet, A. G. (1985) Proc. Natl. Acad. Sci. USA
82, 307–311.
2. Shen, L. L., Mitscher, L. A., Sharma, P. N., O’Donnell, T. J., Chu,
D. W. T., Cooper, C. S., Rosen, T., and Pernet, A. G. (1989) Bio-
chemistry 28, 3886–3894.
3. Takayama, S., Hirohashi, M., Kato, M., and Shimada, H. (1995)
J. Toxicol. Environ. Health 45, 1–45.
4. Marutani, K., Matsumoto, M., Otabe, Y., Nagamuta, M., Tanaka,
K., Miyoshi, A., Hasegawa, T., Nagano, H., Matsubara, S., Kam-
ide, R., Yokota, T., Matsumoto, F., and Ueda, Y. (1993) Antimi-
crob. Agents Chemother. 37, 2217–2223.
5. Domagala, J. M. (1994) J. Antimicrob. Chemother. 33, 685–706.
6. Hasan, T., and Khan, A. U. (1986) Proc. Natl. Acad. Sci. USA
83, 4604–4606.
7. Arai, T., Nishimura, Y., Sasaki, M., Fujita, H., Matsuo, I., Sakur-
agi, H., and Tokumaru, K. (1991) Bull. Chem. Soc. Jpn. 64,
2169–2173.
8. Moore, D. E., Hemmens, V. J., and Yip, H. (1984) Photochem.
Photobiol. 39, 57–61.
9. Dayhaw-Barker, P., and Truscott, T. G. (1988) Photochem. Pho-
tobiol. 47, 765–767.
10. Robertson, D. G., Epling, G. A., Kiely, J. S., Bailey, D. L., and
Song, B. (1991) Toxicol. Appl. Pharmacol. 111, 221–232.
11. Wada, K., Saniabadi, A. R., Umemura, K., Takiguchi, Y., and
Nakashima, M. (1994) Pharmacol. Toxicol. 74, 240–243.
12. Wagai, N., and Tawara, K. (1992) Arch. Toxicol. 66, 392–397.
13. Wagai, N., and Tawara, K. (1991) Toxicol. Lett. 58, 215–223.
14. Nagano, T., Arakane, K., Ryu, A., Masunaga, T., Shinmoto, K.,
Mashiko, S., and Hirobe, M. (1994) Chem. Pharm. Bull. 42,
2291–2294.
15. Auclair, C., and Voisin, E. (1985) in CRC Handbook of Methods
arca
 REACTIVE OXYGEN IN QUINOLONE-INDUCED PHOTOTOXICITY
for Oxygen Radical Research (Greenwald, R. A., Eds.), pp. 123–
132, CRC Press, Boca Raton, FL.
16. Iwamoto, Y., Itoyama, T., Yasuda, K., Uzuhashi, T., Tanizawa,
H., Takino, Y., Oku, T., Hashizume, H., and Yanagihara, Y.
(1992) Chem. Pharm. Bull. 40, 1868–1870.
17. Iwamoto, Y., Itoyama, T., Yasuda, K., Morita, T., Shimizu, T.,
Masuzawa, T., and Yanagihara, Y. (1993) Biol. Pharm. Bull. 16,
1244–1247.
18. Fujita, H., and Matsuo, I. (1988) Chem.–Biol. Interactions 66,
27–36.
19. Artuso, T., Bernadou, J., Meunier, B., Piette, J., and Paillous,
N. (1991) Photochem. Photobiol. 54, 205–213.
20. Shen, L. L., Kohlbrenner, W. E., Weigl, D., and Baranowski, J.
(1989) J. Biol. Chem. 264, 2973–2978.
AID ABB 0124 / 6b38$$$164 05-09-97 23:02:51
281
21. Palù, C., Valisena, S., Ciarrocchi, G., Gatto, B., and Palumbo,
M. (1992) Proc. Natl. Acad. Sci. USA 89, 9671–9675.
22. Iwamoto, Y., Kurita, A., Shimizu, T., Masuzawa, T., Uno, K.,
Yagi, M., Kitagawa, T., Oku, T., and Yanagihara, Y. (1994) Biol.
Pharm. Bull. 17, 654–657.
23. Friedberg, E. C., Ehmann, U. K., and Williams, J. I. (1979) Adv.
Radiat. Biol. 8, 85–174.
24. Wagai, N., Yamaguchi, F., Sekiguchi, M., and Tawara, K. (1990)
Toxicol. Lett. 54, 299–308.
25. Horio, T., Miyauchi, H., Asada, Y., Aoki, Y., and Harada, M.
(1994) J. Dermatol. Sci. 7, 130–135.
26. Akiyama, H., Torigoe, R., Yamada, T., Abe, Y., Kanamoto, A.,
Shimoe, K., Kanzaki, H., and Arata, J. (1991) Chemotherapy
(Jpn.) 39(s-4), 145–148.
27. Yamamoto, Y., Ikeda, M., and Arata, J. (1985) Chemotherapy
(Jpn.) 33(s-7), 966–969.
arca