Modeling the hotochemistry of the reference phototoxic drug lomefloxacin by steady-state and time-resolved experiments, and DFT and post-HF calculations

Article abstract of DOI:10.1002/chem.200701099

The irradiation in water of 1ethyl-6,8-difluoro-7(3-methylpiperazino)3- quinolone-2-carboxylic acid (lomefloxacin), a bactericidal agent whose use is limited by its serious phototoxicity (and photomutagenicity in the mouse), leads to formation of the aryl cation in position eight that inserts into the 1-ethyl chain. Trapping of the cation was examined and it was found that chloride and bromide straightforwardly add in position eight, but with iodide and with pyrrole the l-(2-iodoethyl) and the l-[2-(2-pyrrolyl)ethyl] derivatives are formed. Flash photolysis reveals the triplet of lomefloxacin. a short-lived species (λ_max = 370 nm. τ = 40 ns) that generates the triplet cation (λ_max = 480 nm, τ ≈ 120 ns). The last intermediate is quenched both by halides and by pyrrole. DFT and post-HF methods have shown that the triplet is the lowest state of the cation (ΔG _ST = 13.3 kcal mol^-1) and intersystem crossing (ISC) to the singlet has no role because a less endothermic process occurs, that is, intramolecular hydrogen abstraction from the N-ethyl chain (9.2 kcal mol ¹) that finally leads to cyclization. The halides form weak complexes with the triplet cation (k_q from 4.9×10⁸ for CF to 7.0× 10⁹M^-1 S^-1 for I ). With Cl ^- and Br^- ISC occurs in the complex along with C ₈-X bond formation. However, this latter process is slow with bulky iodide and with neutral pyrrole, and in these cases moderately endothermic electron transfer (ca. 7 kcal mor^-1) yielding the 8-quinolinyl radical occurs. Hydrogen exchange leads to a new radical on the 1-ethyl chain and to the observed products. These findings suggest that the mutagenic activity of the DNA-intercalated drug involves attack of the photogenerated cation to the heterocyclic bases.

Full text of DOI:10.1002/chem.200701099

FULL PAPER
DOI: 10.1002/chem.200701099
Modeling the Photochemistry of the Reference Phototoxic Drug
Lomefloxacin by Steady-State and Time-Resolved Experiments, and DFT
and Post-HF Calculations
Mauro Freccero,*[a] Elisa Fasani,[a] Mariella Mella,[a] Ilse Manet,[b] Sandra Monti,*[b] and
Angelo Albini*[a]
Abstract: The irradiation in water of 1-
ethyl-6,8-difluoro-7(3-methylpiperazi-
no)3-quinolone-2-carboxylic acid (lo-
mefloxacin), a bactericidal agent whose
use is limited by its serious phototoxici-
ty (and photomutagenicity in the
mouse), leads to formation of the aryl
cation in position eight that inserts into
the 1-ethyl chain. Trapping of the
cation was examined and it was found
that chloride and bromide straightfor-
wardly add in position eight, but with
iodide and with pyrrole the 1-(2-io-
doethyl) and the 1-[2-(2-pyrrolyl)ethyl]
derivatives are formed. Flash photoly-
sis reveals the triplet of lomefloxacin, a
short-lived species (lmax =370 nm, t=
40 ns) that generates the triplet cation
(lmax =480 nm, tꢀ120 ns). The last in-
termediate is quenched both by halides
and by pyrrole. DFT and post-HF
methods have shown that the triplet is
cyclization. The halides form weak
complexes with the triplet cation (kq
from 4.9”108 for Clꢁ to 7.0”109 mꢁ1 sꢁ1
for Iꢁ). With Clꢁ and Brꢁ ISC occurs in
ꢁ
the complex along with C8 X bond for-
the lowest state of the cation (DGST
=
mation. However, this latter process is
slow with bulky iodide and with neutral
pyrrole, and in these cases moderately
endothermic electron transfer (ca.
7 kcalmolꢁ1) yielding the 8-quinolinyl
radical occurs. Hydrogen exchange
leads to a new radical on the 1-ethyl
chain and to the observed products.
These findings suggest that the muta-
genic activity of the DNA-intercalated
drug involves attack of the photogener-
ated cation to the heterocyclic bases.
13.3 kcalmolꢁ1) and intersystem cross-
ing (ISC) to the singlet has no role be-
cause
a
less endothermic process
occurs, that is, intramolecular hydrogen
abstraction from the N-ethyl chain
(9.2 kcalmolꢁ1) that finally leads to
Keywords: carbocations · cations ·
density functional calculations
·
fluoroquinolones · photochemistry
Introduction
ular for highly active and usually cytotoxic compounds, for
which restricting the action to specific targets is of primary
importance, as in the case of DNA-targeting antitumor
drugs.[1a,b] Among the possible activation methods, photo-
chemistry is distinguished by its ability to generate highly re-
active intermediates under peculiarly mild conditions, pro-
vided that transmission of light to the desired target can be
achieved and a suitable photoactivable moiety is present.
Until now, photoinduced DNA alkylation has been obtained
for a limited number of compounds, in most cases pertaining
either to the group of psoralens[1c] or to that of quinone me-
thides.[1d,e] Thus, it would be useful to devise further classes
of molecules that can be photoactivated and damage selec-
tively a cell component, in particular DNA. Inspiration for
this search may be obtained by reviewing literature cases of
the phototumorigenic effect of xenobiotics, in particular of
drugs, because in this case the mechanism of action, and
thus, the location in the cell, is known. Understanding how
the noxious effect is caused may suggest a way for using
In situ activation of drugs is an emerging research area and
may open the way to new therapeutic applications, in partic-
[a]Prof. M. Freccero, Prof. E. Fasani, Prof. M. Mella, Prof. A. Albini
Department of Organic Chemistry, University of Pavia
Via Taramelli 10, 27100 Pavia (Italy)
Fax : (+39)0382-98323
E-mail: mauro.freccero@unipv.it
angelo.albini@unipv.it
[b]Dr. I. Manet, Dr. S. Monti
Istituto per la Sintesi Organica
e la Fotoreattività-ISOF, CNR Area della Ricerca
Via P. Gobetti 101, 40129 Bologna (Italy)
E-mail: monti@isof.cnr.it
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author: Cartesian coordi-
nates, energies, and thermal corrections to energy, enthalpy, and free
energy of the stationary points.
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
653

that compound or some derivative for the selective attack
on a tumor.
On the basis of the foregoing, it appears that searching
for a mechanism of the photomutagenic and genotoxic
effect by lomefloxacin has a more general interest than as-
sessing a limitation in the therapeutic use of this molecule.
An unusual mechanism may be involved, based on a photo-
reaction of the drug with a biomolecule [Eq. (3)], which
may be the model for a new class of photoactivated drugs.
However, it is difficult to guess from literature reports
which bimolecular reaction may be involved, because irradi-
ation of 1 in water[10a] (as well as in plasma)[10b] causes an ef-
ficient (F=0.55) unimolecular reaction to give pyrroloqui-
A case in point is that of the fluoroquinolone lomefloxa-
cin (1), an inhibitor of bacterial gyrase[2,3] effective against a
variety of infections, but presently less often used because
of negative side effects, including significant phototoxicity.
In fact, this drug is known to cause single-strand DNA
cleavage and tumorigenesis in animals.[4,5] Indeed, the activi-
ty of 1 is so remarkable that this molecule has been pro-
posed as a standard of photomutagenic action.[4] This distin-
guishes 1, a 6,8-difluoro derivative, from most fluoroquino-
lones (bearing a single fluorine in position six) that do not
exhibit such activity, although being somewhat phototoxic.
The reason of this enhanced activity is unclear at present, al-
though it has been ascertained that this is not due to a
higher skin concentration after administration of the drug.[5a]
Interestingly, the mechanism underlying the photomuta-
genic effect of 1 seems to be peculiar. In fact, reactive
oxygen species (ROS) have most often been suggested to be
the active species in drug-photoinduced DNA damage
[Eq. (1)]:
nolone 3. Evidence that this process involves heterolysis of
+
ꢁ
the C8 F bond to form carbocation 2 that inserts into the
b position of the N-ethyl group (see Scheme 1) was presen-
ted.[10a] Furthermore, a transient attributed to this cation has
been detected by flash photolysis (see below).[11] Thus, a
highly reactive intermediate, such as aryl cation 2+,[12,13] is
generated by photolysis, but is consumed intramolecularly,
with no interference even by solvent water (no 8-hydroxy-
quinolone 3’ is formed).[14]
1
3
1
Cꢁ
ð1Þ
drug þ hn ! drug ! drug þ O2 ! O2 or O2
However, in vitro experiments showed that oxygen activa-
1
Cꢁ
tion yielding either O2 or O2 is modest in the case of 1,
but more efficient with other fluoroquinolones.[6a,b] Further-
more, photobiological studies showed that with this drug not
only formation of 8-oxoguanine, the typical mark of oxida-
tive damage,[6c,d] but also cyclobutane pyrimidine dimeriza-
tion occurred[5b] and the latter phenomenon was correlated
with UVA-induced tumorigenesis. The occurrence of pyrimi-
dine dimerization suggests a different mechanism, that is,
the initiating step is energy transfer from the drug in the
triplet state[7] to the pyrimidine chromophore [Eq. (2)]and
that this causes the damage:
Scheme 1. Photoreaction of 1 in water.
1
3
drug þ hn ! drug ! drug þ DNA base !
3DNA base ! dimers
Results
ð2Þ
Photochemical studies: In view of the above, an in-depth
study of the photochemistry of lomefloxacin appeared
worthwhile. As shown in the following, intriguing results
were obtained and elaborated computational work was re-
quired for the rationalization. In previous work, we had
identified an intermolecular reaction via cation 2+, that is,
Cl/F exchange at position eight leading to compound 4a by
irradiation of 1 in the presence of NaCl (Scheme 2).[15] The
efficiency of chloride trapping was determined by measuring
the ratio of the cyclization yields in the presence vs in the
absence of Clꢁ. According to Scheme 2, the ratio of the cy-
clized product 3 in the presence vs in the absence of the
anion was expected to depend linearly on the Clꢁ concentra-
tion, [3]/[3]0 =1 + (kCl/ki) [Clꢁ]. This was indeed observed
with Clꢁ (under constant ionic strength, obtained by adding
a suitable amount of NaClO4, see Experimental Section)
and the rate-constants ratio for chloride addition versus in-
tramolecular insertion could be calculated, kCl/ki =60mꢁ1.
Evidence has been reported that this mechanism is in-
volved in the phototoxic effect of monofluoroquinolones,[8]
in agreement with the fact that such molecules exhibit a
long-lived triplet (tT ꢀ1 ms), as indicated by flash photolysis
studies in water.[9] However, once again lomefloxacin be-
haves differently, as only a weak absorption is observed in
the place of the intense transient triplet signal of the mono-
3
fluoro derivatives,[9] suggesting that triplet 1 is short lived
and the photophysical pathway of [Eq. (2)]is less likely. A
final possibility is that irradiation causes a chemical reaction
of the excited drug itself (or of a photoproduct of it) with
DNA,[6e] possibly establishing a covalent bond [Eq. (3]:
drug þ hn ! 1,3drug þ DNA ! products
or 1,3drug ! products
ð3Þ
products þ DNA ! products
654
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 653 – 663

Photochemistry of Lomefloxacin
FULL PAPER
Scheme 3. Photoreaction of 1 with pyrrole.
different reaction paths were involved, with attack either at
C8 or at the side chain b position. We further checked that
quenching of the fluorescence of 1 was modest (KSV =3mꢁ1
for pyrrole).
Scheme 2. Photoreaction of 1 with halides.
We now further tested the irradiation in the presence of
halides, and found that with bromide the analogous bromo-
quinolone 4b was formed (70% vs 3% 3 with 0.05m NaBr
at 35% conversion of 1; further irradiation caused a signifi-
cant reductive debromination of 4b, a secondary photoreac-
tion as proven by separate irradiation of pure 4b). Trapping
was more efficient than with chloride (kBr/ki =520mꢁ1). As
for fluoride, an effect was obtained only when a large
amount was added and involved a slowing of the photocon-
version of 1 to 3, reasonably because fluoride intercepted
the cation to give back the starting compound. By treating
the data as above, the rate ratio was measured as kF/ki =
5mꢁ1.
When iodide was considered, a faster reaction was ob-
served (kI/ki =915mꢁ1), but the striking fact was that the
product structure was different. The main product did result
from overall F/I exchange, but position eight was reduced
and the iodine atom was located on the N-ethyl chain in the
b position (structure 5), as indicated by spectroscopic analy-
sis (see Experimental Section). The expected 8-iodo deriva-
tive (4c) was only a minor
Detection of transients: Laser flash photolysis experiments
at lexc =355 nm have been carried out both by ourselves[9]
and, more recently, by Cuquerella et al.[11] We reported a
weak shortlived absorption (lmax ca. 500 nm), that the latter
authors evidenced to be quenched by chloride and bromide
(kCl =4.1”108, kBr =3.6”109 mꢁ1 sꢁ1). However, an accurate
reexamination of the evolution of the spectra now revealed
that there were actually two short-lived transients. The dif-
ference spectrum taken at 25 ns from pulse end was charac-
terized by two main bands peaking at 370 and 480 nm (see
Figure 1). Time evolution occurred by biexponential kinet-
ics, the absorbance profiles clearly exhibiting a growing
phase followed by a decay at 480 nm and a monotonic de-
crease toward a constant value at 370 nm. Neither of the
two bands was appreciably quenched by O2 or N2O.
A kinetic model with three species involved in two con-
secutive reactions with rate constants k1 and k2 proved to
suit perfectly the whole spectral range. The separated spec-
tra of these species (difference spectra with respect to the
ground state of 1) were reconstructed by means of the
product (62% 5, 7% 4c, traces
of 3 with 0.05m NaI at 25%
conversion).
Recalling that phenyl cations
are known to react with p nu-
cleophiles,[13] fluoroquinolone 1
was irradiated also in the pres-
ence of pyrrole. Indeed, a new
product was formed that incor-
porated the heterocycle, char-
acterized by reductive defluori-
nation in position eight and in-
sertion of pyrrole into the
ethyl side chain (product 6,
Scheme 3).
Thus, a diagnostic trapping
of cation 2+ occurred in the
presence of both halide anions
and of a neutral p nucleophile
such as pyrrole (not of a neu-
tral nucleophile such as
water, though). However, two
Figure 1. Transient absorption spectra upon excitation at 355 nm of an N2O saturated 1.4”10ꢁ3 m solution of 1
in 1”10ꢁ3 m NaHCO3 buffer of pH 6.9 at 258C at various delays from pulse end. Insets: A) reconstructed abso-
lute spectra and B) kinetics at 370 nm (black) and 490 nm (grey).
n
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
655

A. Albini, M. Freccero, S. Monti et al.
SPECFIT/32 program, using
the time constants k1 =(2.5Æ
0.3)”107 sꢁ1 and k2 =(8.3Æ
0.2)”106 sꢁ1 and are shown in
inset A of Figure 1.
The
above-mentioned
quenching by chloride and
bromide was confirmed, but
this involved only the 480 nm
band. The same applied to
iodide, as can be appreciated
in Figure 2. Particularly in-
structive were experiments at
a
low concentration of the
quencher. With iodide at 5”
10ꢁ4 m it could be clearly dis-
tinguished that the 370–380 nm
band was not quenched,
whereas the 480–490 nm was
(see insets A and B). Further-
more, the 490 nm band in turn
evolved toward a much longer-
lived absorption evidenced at
345 nm (compare the forma-
tion profile shown in inset C
with the decay in inset A).
Figure 2. Transient absorption spectra upon excitation at 355 nm of an N2O saturated 1.4”10ꢁ3 m solution of 1
in the presence of 0.5”10ꢁ3 m NaI in 1”10ꢁ3 m NaHCO3 buffer of pH 6.9 at 258C at various delays from pulse
end. Insets: kinetics at A) 380 nm, B) 490 nm, and C) 345 nm in the absence (black) and presence of NaI (grey
[NaI]=1”10ꢁ3 m for 490 nm).
Figure 3 shows the results
obtained in the presence of
pyrrole. With 5”10ꢁ2
m
of
quencher, the formation of an
absorption in the region 450–
520 nm was barely observable,
whereas the 365 nm absorption
disappeared with practically
unchanged kinetics (see inset).
At 5”10ꢁ3 m, pyrrole had little
effect on the observed transi-
ents.
The above experiment show
that the first-formed transient,
absorbing more strongly in the
near UV region (lmax =365 nm,
tffi40 ns), was not quenched by
either halides or pyrrole and
was the precursor of the fur-
ther transient in the visible
region (lmax =480 nm and life-
Figure 3. Transient absorption spectra upon excitation at 355 nm of an N2O saturated 1.4”10ꢁ3 m solution of 1
in the presence of 0.05m pyrrole in 1”10ꢁ3 m NaHCO3 buffer of pH 6.9 at 258C at various delays from pulse
end. Insets: kinetics at 370 nm and 490 nm in the absence (black) and presence (grey) of 0.05m pyrrole.
time ca. 120 ns), the one that was quenched. Besides con-
firming the quenching constants already reported for Clꢁ
and Brꢁ, we further measured the constant for iodide (kI =
7.0”109 mꢁ1 sꢁ1) and estimated a lower limit for that for pyr-
role (kpy >2”109 mꢁ1 sꢁ1). [The rate constant for the trapping
of the second intermediate by pyrrole was estimated as kpy >
2”109 mꢁ1 sꢁ1, considering that this intermediate did not ap-
preciably accumulate at [Py]of 5”10 ꢁ2 m and thus, the
decay rate in the presence of a quencher (k’2) must be mark-
edly larger than its rate of formation (k1 =2.5”107 sꢁ1).
Thus, for a k’2/k1 ratio > ꢀ4, it resulted that kpy [py]> ꢀ1”
108 sꢁ1.]
Computational Studies
Geometry and energy of the cation spin states: The forma-
tion of products 5 and 6 and the change in product distribu-
tion in going from bromide to iodide were quite puzzling.
This fostered computational work, aimed to model the prop-
656
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 653 – 663

Photochemistry of Lomefloxacin
FULL PAPER
3
1
erties and the reactivity of the suggested intermediate cation
2+ and to test whether the experimentally observed two
modes of trapping were reproduced. This was done in gas
and condensed phases by using DFT and post-HF methods.
for triplet 7+ and a puckered aromatic ring for singlet 7+
(Figure 4). Both methods predicted the triplet to be more
stable than the singlet, but the MP2 method seemed to over-
estimate the stability of the open-shell species. At the
U(R)B3LYP/6-31+G(d,p) level of theory DETꢁS =13.6 kcal
N
molꢁ1 was predicted in the gas phase, with a minimal solva-
tion effect (13.3 kcalmolꢁ1 in water, see Table 1).
Monomolecular reactivity of singlet and triplet aminoquino-
linyl cations: Next, the potential energy surface (PES) of
the unimolecular reactions involving the singlet and triplet
aminoaryl cations 17+ and 37+ was explored to locate a
viable path for the observed intramolecular attack on the N-
ethyl chain. The stationary points (minima and transitions
states, TSs) were searched for both in the gas phase and in
aqueous solution (by using the PCM solvation model).
A
path involving intramolecular insertion into the
carbon–hydrogen bond was characterized for the singlet
cation (see Scheme 4 and Table 1). This was an almost bar-
rierless process in water, with an early transition state
TS-17+ (DG°H O =0.6–1.1 kcalmolꢁ1, depending on solvent-
2
Scheme 4. Calculated intramolecular paths from quinolinyl ion 7+.
cavity definition, see Figure 5).[16] From this TS the system
evolved along the reaction coordinate toward a much more
Table 1. Electronic energies (E, in Hartree), relative electronic energies (DEgas, in kcalmolꢁ1), thermal contri-
bution to free energy (dG), free energy in the gas phase relative to 17+ (DGgas), free energy in aqueous solu-
tion (Gsol), free energy in aqueous solution relative to 17+ (DGH O), reaction free energy (DG°H O) for quinolin-
A somewhat simplified model
structure that incorporated the
essential features of 2+ was
used, namely, 7-amino-1-ethyl-
4-oxo-1,4-dihydro-quinolin-8-yl
cation (7+, see Scheme 4).
2
2
yl cations 7+ and some of their derivatives.
Stationary points ꢁE DEgas
17+
0.0
dG
DGgas
ꢁGsol
DGH O
DG°
H2O
2
610.2013657[a]
100.4
0.0
(0.0)
610.282163[a,c]
610.290864[a,d]
(606.476167)[b,c]
(606.484550)[b,d]
610.303710[a,c]
610.312659[a,d]
(606.550972)[b,c]
0.0
0.0
–
(606.386342015)[b]
N
U
DFT and MP2 calculations
were
U(R)B3LYP/6-31+G
MP2/6-31+G(d,p) levels of
performed
at
37+
610.2233267[a]
100.2
ꢁ13.6
ꢁ13.3[a,c]
ꢁ13.5[a,d]
ꢁ46.9[b,c]
–
AHCTREUNG
(606.450723523)[b]
ACHTREUNG
theory. Solvation effects on ge-
ometries and energies were
computed by means of the po-
larizable continuum model
(PCM) using both UA0 and
UAHF radii (see Supporting
Information for details). Of
the two spin states of the
cation, both DFT and post-HF
methods predicted in good
agreement a planar geometry
(606.559549)[b,d] ꢁ47.1[b,d]
[f]
TS-17+
TS-37+
–
99.8[g]
98.3
610.280461[a,c,e]
610.289827[a,d,e]
610.290895[a,c,e]
610.294337[a,d,e]
610.417001[a,c]
610.323023[a,c]
610.330062[a,d]
1.1
0.6
1.1
0.6
9.2
610.2063026[a]
–5.2
ꢁ8.8[a,c]
ꢁ4.3[a,d]
ꢁ78.9
ꢁ26.4
ꢁ25.4
8+
9+
610.3227159[a]
610.2404845[a]
106.1
99.6
ꢁ70.4
ꢁ25.5
ꢁ78.9
ꢁ13.1
ꢁ11.9
[a]U(R)B3LYP/6-31G +
to the hydrogen that is going to be transferred. [f] TS-17+ not located at the RB3LYP/6-31G+
theory in the gas phase. [g]In water solution at the RB3LYP/6-31G +(d,p) level.
A
ACHTREUNG
AHCTREUNG
AHCTREUNG
stable
(DG°H O =ꢁ78.9 kcal
2
molꢁ1) intermediate, a benzeni-
um (Wheland) cation (8+), and
hence to pyrroloquinoline 11.
On the other hand, hydro-
gen abstraction was the viable
path from the most stable state
of the cation, triplet 37+, and
led to a triplet diradical cation,
9+, by passing through TS-37+.
This required an activation
Figure 4. Geometries of singlet and triplet aminoaryl cations 1,37+ in gas phase at the B3LYP/6-31+G
(d,p)
G
level.
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
657

A. Albini, M. Freccero, S. Monti et al.
weight of the halide (see
Table 2). This was an impor-
tant point because triplet
cation 37+ would have to un-
dergo intersystem crossing
(ISC) to 17+ before coupling
with the halide to give prod-
ucts 10, and the previous for-
mation of a complex might
provide this opportunity, fa-
vored by the heavy-atom effect
in the decreasing series Iꢁ!
Brꢁ!Clꢁ (see Discussion).
Figure 5. Calculated transition states TS-17+ and TS-37+. Reaction activation free energies (in kcalmolꢁ1) and
distances (in Š) are shown for both the gas phase and aqueous solution (in parentheses), at the B3LYP/6-
31+G(d,p) level.
A
free energy of 9.2 kcalmolꢁ1 in water. In turn, cation 9+
could undergo intersystem crossing and cyclize to the above
isomer 8+. As seen above, cyclization to 11 (Scheme 4) was
indeed the experimentally observed process in the irradia-
tion of lomefloxacin in water and calculations revealed that
both the singlet and the triplet ions could be the intermedi-
ates in it, through a non-activated and activated path, re-
spectively (see below).
Table 2. Reaction free energy (DGcomplex) for the complexation of triplet
cation 7+ by Clꢁ, Brꢁ, and Iꢁ in the gas phase (DGcomplex-gas) and in aque-
ous solution (DGcomplex-aq).
Complex 37+-Xꢁ
37+-Clꢁ
37+-Brꢁ
DGcomplex-gas
DGcomplex-aq
DGsol Xꢁ
Bimolecular reactivity of singlet and triplet aminoquinolinyl
ꢁ112.1[a]
ꢁ119.3[b]
ꢁ106.9[c]
ꢁ102.7[c]
ꢁ0.2[a]
ꢁ11.1[b]
+0.6[c]
ꢁ2.3[c]
ꢁ74.6
ꢁ68.6
1
cations: As intuitively expected, singlet cation 7+ collapsed
to the corresponding eight-halo derivative (see structure 10,
Scheme 4) when in the vicinity of a halide anion. In con-
trast, computation evidenced that approaching of a chloride,
37+-Iꢁ
ꢁ60.7
[a]At the B3LYP/6-31 +G
(d,p) with LAN2DZ basis set for Brꢁ. [c]At B3LYP/6-31 +G
CEP-31G basis set for Brꢁ and Iꢁ.
ACHTREUNG
3
bromide, or iodide anion to triplet cation 7+ did not lead to
A
ACHTREUNG
bond formation, but to electrostatic pairing with the anion
close to the 7-amino group, which bore most of the positive
charge (see Figure 6).
The complexes were marginally stabilized in water, the
net stabilization resulting from the opposite contributions of
ion pairing and of loss of the solvation of the halide anion,
respectively decreasing and increasing with the atomic
As mentioned above, the experiments showed that prod-
ucts of type 10 were formed with Clꢁ and Brꢁ, but of type
12 with iodide (see Scheme 5). This changed path must be
rationalized. Because it involved the side chain, just as cycli-
zation to 11, a possibility was
that 12 arose from the same in-
termediate
as
cyclization,
namely, from cation 8+. It may
be thought that this reacted in
two ways. Thus, in the conver-
sion of 8+ to 11 water acted as
a base. Should it act as a nucle-
ophile, ring opening to give a
product of structure 12 would
indeed result. The viability of
these mechanisms was tested
computationally, by locating
the transition state from 8+ to
11 in water [TS-8+-11
(H2O)]
A
and toward 12 with either
ꢁ
water [TS-8+-12
A
the nucleophile [TS-8+-12(Iꢁ)].
At the B3LYP level of theory,
using both 6-31G(d) and 6-
Figure 6. Side and upper views of the complexes 37–X at the B3LYP/6-31+G
(d,p) level with CEP-31G basis
A
set for Brꢁ and Iꢁ, in the gas phase and in aqueous solution (in parenthesis). LAN2DZ basis set for Brꢁ gives
very similar geometries.
31+(d,p) basis sets, the ring-
A
658
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 653 – 663

Photochemistry of Lomefloxacin
FULL PAPER
104.1 kcalmolꢁ1]were very similar, and the same held for
pyrrole (experimental Eox 1.31 V).[18]
The results are gathered in Table 4 and suggest that the
monoelectronic reduction of both cations 1,37+ by iodide
anion was slightly endoergonic (5 to 7 kcalmolꢁ1, see foot-
notes of Table 4) and virtually identical data were obtained
for pyrrole.
Table 4. Computed (vertical) reduction potentials of triplet (37+) and sin-
glet (17+) quinolonyl cations, of cation 9+, iodine atom, and pyrrole radi-
cal cation in aqueous solution, compared to experimental values (for
iodine and pyrrole).
Scheme 5. Possible reactions of cation 8+ with bases/nucleophiles.
Redox reaction
E [V](vs NHE)
ꢁ
I !I
+1.33,[a] 1.30[b]
+1.12[c]
C
1
3
+
C
7 !7 (v)
opening processes induced by water and iodide anion were
largely endothermic in aqueous solution, with an activation
free energy greater than 37 kcalmolꢁ1. In contrast, the de-
protonation of the Wheland cation 8+ to give 11 was a bar-
rierless process (Table 3).
+
7 !7 (v)
+1.02[c]
C
pyrrole +(v)+e!pyrrole
+1.31,[a] 1.35[c]
+1.12[c]
C
+
C
9 !13
[a]Experimental value. [b]Computed at B3LYP/LAN2DZ with solva-
tion effects evaluated by the PCM solvation model using UA0 radii, and
taking into account Ggas,H+ =ꢁ6.28 kcalmolꢁ1
,
Gsolv,H+ =ꢁ262.23 kcal
molꢁ1, DHf(H2)=104.1 kcalmolꢁ1. [c]Calculated from the computed acti-
vation free energy in water [at PCM-U(R)B3LYP/6-31G+(d,p) using
Table 3. Activation free energy in the gas phase (DG° in kcalmolꢁ1
)
gas
LAN2DZ for Iꢁ]for the reactions: 17+ +I !7 (v)+I +4.9 kcalmol
;
ꢁ
ꢁ1
and in aqueous solution (DG°H O), relative to the complexes 8+–H2O and
C
C
2
ꢁ
ꢁ1
+
ꢁ
37+ +I !7 (v)+I +7.2 kcalmol
;
pyrrole (v)+I !pyrrole+I ꢁ0.4
8+–Iꢁ, for the transition states involved in the deprotonation and ring-
opening reactions of benzenium cation 8+.
C
C
C
C
kcalmolꢁ1; 9+ +I !13 +I +4.9 kcalmolꢁ1; taking into consideration the
ꢁ
C
C
ꢁ
C
experimental redox potential for the couple I /I (1.33 V, vs NHE) by
Stationary points
DG°
DG°
gas
H2O
[c]
1
+
ꢁ
C
C
DG=ꢁnF[DE
( 7 /7 )ꢁDE
N
TS-8+-11
TS-8+-12
N
32.8[a] (32.2)[b]
61.9[a] (61.8)[b]
34.0[e]
–
G
48.2[d]
37.5[f]
TS-8+-12(Iꢁ)
Electron-transfer path: It appeared unlikely that electron
[a]B3LYP/6-31G(d) [b]B3LYP/6-31 +G
water, deprotonation occurred with no barrier in the presence of an an-
A
1
transfer was important for singlet 7+, for which a barrier-
less intramolecular process was available to give 8+ (al-
though coupling with a paired anion to give 10 may com-
pete). However, in the case of the triplet cation, reduction
cillary water molecule. [d]PCM-B3LYP/6-31 +G(d,p). [e]B3LYP/6-
31G(d) and CEP basis set on Iꢁ. [f]PCM-B3LYP/6-31G(d) and CEP
basis set on iodide anion.
C
to radical 7 by iodide (or pyrrole) confronted a lower barri-
er than hydrogen abstraction (7.2 vs 9.2 kcalmolꢁ1). The en-
suing chemistry of the radical was explored and this was
found to convert, after thermal relaxation, to the alkyl radi-
Thus, this mechanism did not explain the reaction with
iodide (and the analogous process with pyrrole), which must
involve interaction of the nucleophile directly with ion 7+,
and yet differ from simple coupling that would give product
11. An alternative we considered was that an electron-trans-
fer process was occurring, rather than an electrophile–nucle-
ophile combination. The thermochemistry of the monoelec-
tronic reduction of both spin states of the cation (1,37+) to
C
cal 13 (Scheme 6) by intramolecular hydrogen abstraction
C
from the ethyl side chain. The first-order saddle point TS-7
ꢁ1
C
C
connecting 7 to 13 lay only 3.8 kcalmol above the reac-
tant in water, therefore, it did not represent the rate-deter-
mining step of this reaction (Figure 7).
C
quinolinyl radical 7 by iodide anion and by pyrrole was
evaluated in aqueous solution. The free-energy change for
the vertical oxidation of pyrrole to its radical cation and the
vertical reduction of ions 7+ were computed at B3LYP/6-
31G+(d,p) level of theory, and that for the reduction of
A
iodide radical to iodide anion at B3LYP/CEP-31G and
B3LYP/Lan2DZ by using the PCM solvation model. It was
gratifying to find that the computational results were close
to the experimental data where available. Thus, the calculat-
ꢁ
C
ed free-energy change for the I /I process and that obtained
from the measured redox potential through a thermochemi-
ꢁ
C
cal cycle [E
A
trode (NHE),[17a] calculated by taking Ggas,H+ =ꢁ6.28 kcal
molꢁ1, Gsolv,H+ =ꢁ262.23 kcalmolꢁ1,[17b] and DHf(H2)=
3
Scheme 6. Electron-transfer path from cation 7+.
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
659

A. Albini, M. Freccero, S. Monti et al.
fact, comparing the lifetime of this state (31, 40 ns) with that
of the monofluoroquinolones (ca. 1 ms), suggests that frag-
mentation occurs with an efficiency of around 95%, well in
accord with the high quantum yield of the photoreaction;
the short lifetime explains the lack of O2 quenching. The re-
3
constructed spectra show that 1 has no absorption in the
visible region and is the precursor of a further transient with
lmax =480 nm (Figure 1). The latter is again unaffected by
O2, but is quenched by halides (rate increases to diffusion-
controlled with iodide) and by pyrrole. If this were the sin-
glet cation, one would expect that coupling with any anion,
with no spin barrier, would occur at the same rate. The ob-
served difference supports that the latter transient is triplet
C
Figure 7. Structure of transition state TS-7 . Reaction activation free
energy (in kcalmolꢁ1) and distances (in Š) are shown for both the gas
phase and aqueous solution (in parentheses), at the B3LYP/6-31+G(d,p)
N
3
cation 2+ (see below), which is the expected state from the
level.
3
cleavage of triplet 1.
The computational study on the simplified model 7+
showed that the two spin states of the cation are quite dif-
ferent in structure and chemistry. As far as the structure is
concerned, the difference mirrored that found with 4-amino-
phenyl cation, with the planar triplet lowest in energy and
the singlet (ca. 13 kcalmolꢁ1 above it) characterized by a
puckering out of plane of C8. Again, as with phenyl cations,
the singlet is a localized cation with an empty s orbital at
C8, whereas the triplet has one electron in the s orbital and
one unpaired electron in the
In turn, this alkyl radical could react with Iꢁ or co-formed
iodine atom (or, respectively, with pyrrole or its radical
cation) to generate the final products 12 (Scheme 6). Fur-
ther calculations showed that distonic diradical cation 9+
C
was also reduced to radical 13 at an accessible potential and
electron transfer from iodide and pyrrole was likewise
slightly endothermic (4.3 and 6.6 kcalmolꢁ1, respectively),
thereby offering a further path to 12 (Table 5).
Table 5. Electronic energies (E, in Hartree), relative electronic energies (DEgas, in kcalmolꢁ1), thermal contri-
aromatic
p
system. This
C
bution to free energy (dG), relative (to 7 ) free energy in the gas phase (DGgas), free energy in aqueous solu-
tion (GH O), relative (to 7 ), free energy in aqueous solution (DGH O) for radicalic intermediates and transition
endows the aromatic moiety
with radical-cation character,
with strong donation from the
amino group (see the simpli-
fied structures in Scheme 4).
Examination of the potential
energy surface (PES) for the
reactions of the two states of
the cation evidenced a well-
differentiated chemistry. In the
ꢁ
C
2
2
states.
Stationary points
ꢁE
DEgas
dG
DGgas
ꢁGH O
DGH O
2
2
610.4958393[a]
610.4763145[a]
610.51024[a]
0.0
12.3
ꢁ9.0
101.2
99.0
99.9
0.0
10.1
ꢁ10.3
610.508769[a,b]
610.499151[a,b]
610.527240[a,b]
0.0
3.8
ꢁ12.9
C
7
C
TS-7
C
13
[a]U(R)B3LYP/6-31G +(d,p). [b]UA0 radii.
Discussion
1
+
case of singlet 7 , intramolecular insertion in the b C H
bond is facile, with an early (and barely observable) transi-
tion state. The detailed mapping of this process highlighted
a close resemblance with insertion reactions via singlet car-
benes,[19] with simultaneous rotation of the ethyl group and
bond formation. The triplet quinolinyl cation, on the other
hand, has no non-activated path available. The most favora-
ble reaction channel, hydrogen abstraction to give distonic
diradical cation 9+, confronts a 9.2 kcalmolꢁ1 barrier. With
Previous evidence indicated that the photochemistry of lo-
mefloxacin (1) in water involved release of fluoride selec-
tively from position eight and the process was unimolecular
with a reaction quantum yield Fr of 0.55, much higher than
that exhibited by 6-monofluoroquinolones, not undergoing
unimolecular cleavage (e.g., 0.06 for norfloxacin). Because
the fluorescence quantum yields and lifetimes of 6,8-difluor-
oquinolones such as 1 and 6-monofluoroquinolone deriva-
tives are similar (e.g., FF =0.08 for 1 and 0.11 for norfloxa-
cin, with tF =1 and 1.5 ns, respectively), photofragmentation
should not occur from the singlet excited state.
two unpaired electrons in two orthogonal orbitals, 7+ re-
3
sembles a triplet carbene or a radical at C8. Indeed, compar-
3
ison of the structures of cation 7+ and radical 7 , as well as
C
On the other hand, the large increase in reaction quantum
yield was accompanied by a dramatic change in the T–T ab-
sorption. Monofluoroquinolones exhibit an intense and rela-
tively long-lived triplet absorption (t=1 ms) with lmax in the
range 500–620 nm.[8] In the case of 1, careful examination of
the time-resolved spectra evidences a short-lived (ca. 40 ns)
transient absorbing at around 370 nm (Figure 1), unaffected
by the additives tested, that fits well with the expectation of
a triplet undergoing a fast monomolecular fragmentation. In
the almost identical transition states for hydrogen abstrac-
tion, TS-37+ and TS-7 (see Figures 5 and 7) demonstrates
C
that the triplet cation has a well-defined sp2 radical charac-
ter at C8.
3
To summarize, intermediate 7+ is formed first, but has
only activated paths available, leading either to diradicalic
1
9+ or to 7+ through ISC, both paths finally reaching cy-
clized 11. The latter path, however, confronts a markedly
higher (by more than 5 kcalmolꢁ1) barrier and thus, it ap-
660
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 653 – 663

Photochemistry of Lomefloxacin
FULL PAPER
pears unlikely that the singlet is formed even as a secondary
intermediate and, if it were formed, the steeply descending
path leading to 8+ and entropic factors would make trap-
ping difficult.
which is intramolecular insertion to give 3. However, fluoro-
quinolones are known to insert into the distorted DNA seg-
ment bound to the gyrase active site, either by intercalating
between the bases on both sides of the double helix or by
replacing cytosine and forming hydrogen bonds with a gua-
nine on the opposite strand.[20] Thus, when photoexcitation
involves a molecule of 1 intercalated with DNA, the likeli-
hood of an attack by 2+ on a neighboring biomolecule is
high. The model reaction with pyrrole in solution suggests
that electrophilic attack on a DNA base could occur in that
case ([Eq. (4)], compare [Eq. (3)]). Another possibility, in
view of the accessible reduction potential of the quinolinyl
cation, is that electron, rather than energy, transfer occurs
[Eq. (5)]and causes pyrimidine dimerization. [6,21]
As for intermolecular processes, these do not involve the
excited states, as the quenching of both the fluorescence and
the T–T absorption is unimportant under the conditions
used. Indeed, experiments showed that both types of prod-
ucts, the 8-substituted quinolones 10 and products 12 bear-
ing the substituent at the b-ethyl position, arise from a
single intermediate, the transient detected at 480 nm in flash
photolysis and identified as the triplet cation, as supported
by the correspondence of the rate-constants ratio from flash
photolysis (Clꢁ/Brꢁ/Iꢁ 1:11.2:14.0) with those from steady-
state experiments (Fꢁ/Clꢁ/Brꢁ/Iꢁ =0.08:1:8.7:15.3). No fur-
ther common intermediate is involved, as a thorough search
for a nucleophile-driven ring opening in cation 8+ gave no
result, proving that this path leads only to cyclized 11.
The different quenching rates of the 480 nm transient sup-
ports, as mentioned, the triplet multiplicity. Furthermore,
the observed fast increase as the atomic weight of the halide
quencher increases confirms that heavy-atom-induced ISC
has a role. Thus, it seems reasonable to conclude that ISC
3
1 þ hn ! 2þ
ð4aÞ
ð4bÞ
ð5aÞ
ð5bÞ
ð5cÞ
32þ þ DNA base ! products
3
1 þ hn ! 2þ
32þ þ DNA base ! 2C þ DNA baseCþ
DNA baseCþ ! dimers
3
1
from 7+ to 7+ is insignificant in neat water because of the
high endothermicity. In the weakly stabilized 7+–Xꢁ com-
3
Conclusion
plex (see Figure 6) the barrier is much lower, as ISC in the
cation occurs simultaneously with the highly exothermic
coupling with the anion to give the 8-halogeno derivatives
10 (Scheme 6). However, when passing from Clꢁ and Brꢁ to
bulky Iꢁ bond formation at position eight, flanked by
branched substituents at both sides, is expected to slow
down. The other path individuated by computation is elec-
tron transfer that is favored with a good oxidant, such as
iodide, which is in fact active at very low concentrations
(Figure 2). Likewise, a neutral nucleophile such as pyrrole is
not sufficiently reactive for attacking the hindered electro-
philic site in cation 7+, but being a good donor it reacts
through electron transfer to give a product of type 12, just
like iodide. Electron transfer is an activated process, but
with a smaller barrier than hydrogen abstraction, and occurs
Fluoroquinolones, and in particular 8-fluoro derivatives such
as lomefloxacin, are distinguished for their high photo-
A
ꢁ
teristic photoreaction: C F bond heterolysis to give an aryl
cation. In this work, photoproduct studies and time-resolved
techniques were exploited for characterizing bimolecular re-
actions, and DFT and post-HF calculations were used for
the first time for the detailed rationalization of the chemis-
try of the cation, which has proven to be particularly in-
volved. The triplet quinolinyl cation 32+ formed from 31
(both intermediates identified by flash photolysis) is a viable
precursor of all the observed processes. Theoretical model-
ing reproduced intramolecular cyclization of the cation, as
well as its trapping by nucleophiles (when charged or good
donors), giving two different products types, either with the
entering group at C8 or, when electron transfer intervenes,
at the b position in the ethyl chain.
3
+
C
C
according to the sequence 7 !7 !13 (with activation
energy of the first, and rate-determining, step, of 6–7 kcal
molꢁ1), or, less likely, after hydrogen abstraction (37+!9+!
ꢁ1
C
13 , activation energy of the first step, 9.2 kcalmol ). Final-
The chemical study suggests that the genotoxic effect of 1
involves either addition of the corresponding triplet aryl
cation onto a DNA base or electron-transfer-induced dime-
rization of the bases. More generally, the above discussion
suggests that chemical studies of phototoxic drugs may
concur to the understanding of their action on DNA, and
possibly help to devise new models for light-activated anti-
tumor drugs.
ly, the change in mechanism is characterized by the appear-
3
ance of a new transient that arises from 7+ (see inset C in
C
Figure 2); this may correspond to radical 7 .
Thus, the two different intermolecular reactions arise
3
from the triplet cation 7+, either through ISC in the com-
plex and ionic coupling or through electron transfer, respec-
tively, the choice between the two paths being determined
by the bulk and the oxidation potential of the quencher.
The rationalization of the photochemistry of fluoroquino-
lone 1 presented above suggests a mechanism for the ob-
served mutagenic effect that fits with the experimental re-
sults. Photolysis of lomefloxacin in solution generates the
Experimental Section
Preparative photochemical reactions:
A solution of lomefloxacin 1
3
highly reactive, but short-lived, cation 2+, the only fate of
(140 mg, 5”10ꢁ4 m) in bidistilled water (800 mL) containing the appropri-
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
661

A. Albini, M. Freccero, S. Monti et al.
ate amount of additive was placed in an immersion-well apparatus,
stirred, and flushed with nitrogen for 30 min and then irradiated by
means of a 125-W high-pressure mercury lamp at 208C while maintaining
the nitrogen flux. Monitoring by HPLC (Hypersil ODS2, 250”4.6 mm,
5 mm, pH-3 phosphate buffer/MeCN 8:2 as eluant, flux 0.6 mLminꢁ1, l=
275 nm) showed that the starting material was consumed after 60 min.
The solution was stirred for 2 h with 2”400 mL 1% ethyl chloroformate
in chloroform. The organic layers were reunited, dried, concentrated and
when required treated with ethereal diazomethane. The solution was
evaporated and the residue subjected to chromatography on silica gel
eluting with chloroform/methanol mixtures of 98:2!95:5. The products
from the photolysis of lomefloxacin were thereby isolated as the respec-
tive urethanes (in some case also methyl esters by treatment with diazo-
methane). The structure assignment was based on elemental analyses and
spectroscopic characterization, in particular by 1H, 13C NMR (in CDCl3,
300 MHz with TMS as internal standard), and appropriate bidimensional
techniques. Key spectroscopic data of new compounds are reported
below.
(brs, 1H), 6.5 (d, JHꢁF =7 Hz, 1H), 6.7 (brs, 1H), 7.75 (d, JHꢁF =13 Hz,
1H), 8.35 (s, 1H), 9.2 ppm (brs, 1H); 13C NMR (75 MHz): d =15.1
(CH3), 15.85 (CH3), 28.6 (CH2), 39.1 (CH2), 47.5 (CH), 49.8 (CH2), 51.9
(CH3), 54.9 (CH2), 55.8 (CH2), 61.9 (CH2), 105.1 (CH), 106.7 (CH), 108.8
(CH), 109.5 (C), 112.9 (d, JCꢁF =22 Hz, CH), 118.6 (CH), 121.0 (C), 123.2
(d, JCꢁF =6 Hz, CH), 127.8 (C), 137.2 (C), 145,5 (d, JCꢁF =10 Hz), 149.1
(CH), 153.0 (d, JCꢁF =250 Hz, C), 155.8 (C), 165.5 (C), 173.4 ppm (C); el-
emental analysis calcd (%) for C25H29FN4O5 (484.52): C 61.97, H 6.03, N
11.56; found: C 62.1, H 6.0, N 11.9.
Small-scale irradiations: Small-scale photochemical irradiation for deter-
mining the relative rate of reactions with additives were carried out on
10-mL portions of aqueous solution of 1 in serum-capped quartz tubes.
These were irradiated in a merry-go-round apparatus by means of two
15-W phosphor-coated lamps (center of emission 313 nm). Deoxygena-
tion of the solution was obtained by flushing for 30 min with argon. The
experiments in the presence of halides were conducted with a total salt
concentration of 0.2m by adding the appropriate amount of NaClO4. The
substrate conversion and product yields were determined by HPLC (see
above) on the basis of appropriate calibration curves.
Computational methods: All calculations were carried out by using the
C.02 version of the Gaussian 03[22] program suite.
The geometric structures of the reactants and transition states were fully
optimized both in the gas phase and in aqueous solution by using the
hybrid density functional (U)B3LYP[23] with the 6-31+G
ACHTREUNG
AHCTREUNG
fuse functions is mandatory for a reliable evaluation of energies in anion-
ic systems such as TS-8-Nu.
8-Bromo-1-ethyl-6-fluoro-7-[1-(3-methyl-4-ethylaminocarbonylpiperazi-
no)]-4-quinolone-3-carboxylic acid methyl ester (14a): 1H NMR
(300 MHz, CDCl3): d=1.3 (t, J=7 Hz, 3H), 1.4 (d, J=7 Hz, 3H), 1.5 (t,
J=7 Hz, 3H), 2.95, 3.45 (2”m, 2H), 3.15, 3.65 (2”m, 2H), 3.2, 4.0 (2”
m, 2H), 3.9 (s, 1H), 4.15 (q, J=7 Hz, 2H), 4.4 (m, 1H), 4.65 (m, 2H), 8.2
(d, J=12 Hz, 1H), 8.75 ppm (s, 1H); 13C NMR (75 MHz, CDCl3): d=
14.6 (CH3), 15.8 (CH3), 15.9 (CH3), 38.9 (CH2), 47.4 (CH), 51.6 (CH2),
The basis-set functions specifically used for Br and I atoms have been
the effective core potentials (ECP) CEP/31G[24] and LANL2DZ[25] with
comparable results. With these basis sets only valence electrons for Br
and I atoms are treated explicitly.
Thermal contributions (dG) to free energy (DG) were computed from
B3LYP/6-31+G(d,p) structures and harmonic frequencies by using the
G
51.9 (CH2), 52.1 (CH2), 55,4 (CH2), 61.5 (CH2), 109.1 (C), 113.3 (d, JCꢁ
=
F
harmonic oscillator approximation and the standard expressions for an
ideal gas in the canonical ensemble at 298.15 K and 1 atm.
24 Hz, CH), 126.2 (C), 129.5 (C), 137.4 (C), 143.7 (C, d, J=15 Hz), 152.5
(CH), 155.5 (C), 157.0 (d, JCꢁF =260 Hz, C), 166.1 (C), 172.2 ppm (C); el-
emental analysis calcd (%) for C21H25BrFN3O5 (498.34): C 50.61, H 4.97,
N 9.86; found: C 50.4, H 5.0, N 9.5.
The optimization of the stationary points in the solvent bulk were calcu-
lated by means of the self-consistent reaction field (SCRF) method using
the PCM (polarizable continuum model)[17b] as implemented in the C.02
version of Gaussian 2003. The cavity is composed by interlocking spheres
centered on non-hydrogen atoms with radii obtained by the HF paramet-
rization by Barone known as the united atom topological model
(UAHF).[27] Such a model includes the non-electrostatic terms (cavita-
tion, dispersion and repulsion energy) in addition to the classical electro-
static contribution. We remind readers that the energies resulting from
PCM computations have the status of free energies, as they take implicit-
ly into account thermal and entropic contributions of the solvent. Howev-
er, because they do not include the thermal contributions (dG) of solute
molecular motions to the free energy (DG), gas-phase thermal contribu-
tion of solute molecular motions were added to evaluate the correspond-
ing free energy in water (DGsol). UAO parametrization of the cavity has
been also used and has yielded almost identical results in geometries and
very similar data on energies.
8-Iodo-1-ethyl-6-fluoro-7-[1-(3-methyl-4-ethylaminocarbonylpiperazino)]-
4-quinolone-3-carboxylic acid methyl ester (14b): 1H NMR (300 MHz,
CDCl3): d (at 458C)=1.3 (t, J=7 Hz, 3H), 1.4 (t, J=7 Hz, 3H), 1.5 (d,
J=7 Hz, 3H), 2.9, 3.25 (2”m, 2H), 3.1, 4.05 (2”m, 2H), 3.2, 3.8 (2”m,
2H), 3.9 (s, 3H), 4.2 (q, J=7 Hz, 2H), 4.4 (m, 1H), 4.7 (m, 2H), 8.2 (d,
J=10 Hz, 1H), 8.8 ppm (s, 1H); 13C NMR (75 MHz): d=14.5 (CH3),
15.5 (CH3), 16.8 (CH3), 39.1 (CH2), 47.3 (CH), 51.8 (2CH2), 52.1 (CH3),
55.1 (CH2), 61.3 (CH2), 91.8 (d, JCꢁF =18 Hz, C), 115.2 (d, JCꢁF =24 Hz,
CH), 125.9 (C), 140.8 (C), 146.0 (d, JCꢁF =16 Hz, CH), 152.2 (CH), 155.4
(C), 156.4 (d, JCꢁF =250 Hz, C), 165.6 (C), 172.3 ppm (C); elemental anal-
ysis calcd (%) for C21H25FIN3O5 (545.34): C 46.25, H 4.62, N 7.71; found:
C 46.5, H 4.8, N 7.7.
1-(2-Iodoethyl)-6-fluoro-7-[1-(3-methyl-4-ethylaminocarbonylpiperazi-
no)]-4-quinolone-3-carboxylic acid methyl ester (14c): 1H NMR
(300 MHz, CDCl3): d=1.3 (t, J=7 Hz, 3H), 1.4 (d, J=7 Hz, 3H), 3.0, 3.4
(2”m, 2H), 3.15, 3.6 (2”m, 2H), 3.5, 4.15 (2”m, 2H), 3.55 (t, J=7 Hz,
2H), 4.2 (m, 2H), 4.5 (m, 1H), 4.65 (t, J=7 Hz, 2H), 6.7 (d, JHꢁF =4 Hz,
1H), 8.0 (d, JHꢁF =10 Hz, 1H), 8.7 ppm (s, 1H); 13C NMR (75 MHz): d=
2.5 (CH2), 14.5 (CH3), 15.5 (CH3), 38.4 (CH2), 46.9 (CH), 49.45 (CH2),
54.4 (CH2), 55.9 (CH2), 61.6 (CH2), 103.1 (CH), 108.3 (C), 113.2 (d, J=
22 Hz, CH), 120.4 (d, JCꢁF =15 Hz, C), 136.6 (C), 146.6 (d, JCꢁF =12 Hz,
C), 147.6 (CH), 153.3 (d, JCꢁF =250 Hz, C), 155.1 (C), 166.6 (C),
167.0 ppm (C); elemental analysis calcd (%) for C20H23FIN3O5 (531.32):
C 45.21, H 4.36, N 7.91; found: C 45.5, H 4.0, N 8.1.
Nanosecond laser flash photolysis: The laser beam (a JK-lasers Nd/YAG
operated at 355 nm, pulse FWHM 20 ns) was focused on a 3 mm high
and 10 mm wide rectangular area of the cell and the first 2 mm in depth
were analyzed at a right angle geometry. The incident pulse energies
used were <17 mJcmꢁ2 (5 mJ per pulse). The minimum response time of
the detection system was of ca. 2 ns. The bandwidth used in the spectro-
kinetic measurements was typically 2 nm. The sample absorbance at
355 nm was typically 0.2 over 1 cm. Oxygen was removed by vigorously
bubbling the solutions with a constant flux of N2O. The solution, in a
flow cell of 1 cm path, was renewed after few laser shots. The tempera-
ture was 295Æ2 K. The detector system was perturbed from 390 nm to
480 nm by the intense emission of 1, generated by the laser excitation.
These problems were minimized by using neutral density filters at the en-
trance slit of the monochromator and pulsing the 150 W high pressure
Xe lamp at high currents (ꢀ200 mA for 1 ms) to increase the intensity of
1-[2-(2-Furanyl)-ethyl]-6-fluoro-7-[1-(3-methyl-4-ethylaminocarbonylpi-
perazino)]-4-quinolone-3-carboxylic acid methyl ester (14d): 1H NMR
(300 MHz, CDCl3): d=1.3 (t, J=7 Hz, 3H), 1.4 (t, J=7 Hz, 3H), 2.9,
3.35 (2”m, 2H), 3.1, 3.55 (2”m, 2H), 3.35 (m, 2H), 3.4, 4.15 (2”m, 2H),
3.8 (s, 3H), 4.2 (m, 2H), 4.4 (m, 1H), 4.45 (m, 2H), 6.0 (brs, 1H), 6.15
662
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 653 – 663

Photochemistry of Lomefloxacin
FULL PAPER
the analyzing light. In spite of this, transient spectra were not significant
before 10–30 ns from pulse end. Acquisition of absorption signals were
performed by a home made program using Asyst 3.1 (Software Technolo-
gies, Inc.).
ref. [14a]) and with o-propylchlorobenzene (ref. [14b]); a) T. Cohen,
J. Lipowits, J. Am. Chem. Soc. 1964, 86, 2515; b) K. Hori, T. Sonoda,
M. Harada, S. Yamazaki-Nishida, Tetrahedron 2000, 56, 1429.
[15]E. Fasani, M. A. Rampi, A. Albini, J. Chem. Soc. Perkin Trans. 2
1999, 1901.
[16]In fact, we located an early transition structure at the B3LYP/6-
31+G(d,p) level only in water, not in gas. At MP2/6-31G(d) a TS in
G
gas had an activation energy of 3.1 kcalmolꢁ1
.
Acknowledgements
ꢁ
C
[17]a) Experimental redox potential
E
A
H. A. Schwarz, B. H. J. Bielski, J. Phys. Chem. 1986, 90, 1445; b) V.
Barone, M. Cossi, J. Phys. Chem. A 1998, 102, 1995.
We thank Dr. Francesco Manoli for technical support and MIUR for par-
tial support of this work.
[18]H. D. Tabba, K. M. Smith, J. Org. Chem. 1984, 49, 1870.
[19]R. D. Bach, M. D. Su, E. Aldabbagh, J. L. Andres, H. B. Schlegel, J.
Am. Chem. Soc. 1993, 115, 10237.
[1]a) S. S. Wolkenberg, D. L. Boger, Chem. Rev. 2002, 102, 2477;
b) S. R. Rajski, R. M. Williams, Chem. Rev. 1998, 98, 2723; c) F. P.
Gasparro, B. Liao, P. J. Foley, X. M. Wang, J. M. McNiff, Environ.
Mol. Mutagen. 1998, 31, 105; d) M. Chatterjee, S. E. Rokita, J. Am.
Chem. Soc. 1994, 116, 1690; e) S. N. Richter, S. Maggi, S. Colloredo-
Mels, M. Palumbo, M. Freccero, J. Am. Chem. Soc. 2004, 126, 13973.
[2]L. J. Piddock, M. C. Hall, R. Wise, Antimicrob. Agents Chemother.
1990, 34, 1088.
[3]a) N. J. Traynor, M. D. Barratt, W. W. Lovell, J. Ferguson, N. K.
Gibbs, Toxicol. in Vitro 2000, 14, 275; b) N. J. Neumann, A. Blotz,
G. Wasinska-Kempka, M. Rosenbruch, P. Lehmann, H. J. Ahr,
H. W. Vohr, J. Photochem. Photobiol. B 2005, 4, 25.
[4]a) L. Marrot, J. P. Belaidi, C. Jones, P. Perez, L. Riou, A. Sarasin,
J. R. Meunier, J. Invest. Dermatol. 2003, 121, 596; b) L. Marrot, C.
Agapakis-Causse, Mutat. Res. 2000, 468, 1.
[5]a) G. Klecak, F. Urbach, H. Urwyler, J. Photochem. Photobiol. B
1997, 37, 174; b) T. Itoh, H. Miyauchi-Hashimoto A. Sugihara, K.
Tanaka, T. Horio, J. Invest. Dermatol. 2005, 125, 554.
[6]a) L. J. Martinez, R. H. Sik, C. Chignell, Photochem. Photobiol.
1998, 67, 399; b) N. Wagai, K. Tawara, Arch. Toxicol. 1992, 66, 392;
c) L. K. Verna, D. Chen, G. Schluter, G. M. Williams, Cell Biol. Toxi-
col. 1998, 14, 237; d) J. E. Rosen, A. K. Prahalad, G. Schluter, D.
Chen, G. M. Williams, Photochem. Photobiol. 1997, 65, 990; e) C. E.
Perrone, K. C. Takahashi, G. M. Williams, Toxicol. Sci. 2002, 69, 16.
[7]a) N. J. Traynor, N. K. Gibbs, Photochem. Photobiol. 1999, 70, 957;
b) S. Sauvaigo, T. Douki, F. Odin, S. Caillat, J. L. Ravanat, J. Cadet,
Photochem. Photobiol. 2001, 73, 230.
[8]F. Boscà, V. Lhlaubet-Vallet, M. C. Cuquerella, J. V. Castell, M. A.
Miranda, J. Am. Chem. Soc. 2006, 128, 6318.
[9]S. Monti, S. Sortino, E. Fasani, A. Albini, Chem. Eur. J. 2001, 7,
2185.
[20]a) J. Heddle, A. Maxwell, Antimicrob. Agents Chemother. 2002,
1805; b) L. A. Mitscher, Chem. Rev. 2005, 105, 559; c) L. L. Shen, J.
Baranowski, G. Pernet, Biochemistry 1989, 28, 3879.
[21]Guanosine is more easily oxidized than pyrrole, E=1.29 V vs NHE,
and thymine less easily; at any rate, the process observed will
depend on the closest-lying bases in the preferred intercalation;
a) S. Steenken, S. V. Jovanovic, J. Am. Chem. Soc. 1997, 119, 617;
b) A. Joy, A. K. Ghosh, G. B. Schuster, J. Am. Chem. Soc. 2006, 128,
5346.
[22]Gaussian 03, Revision C.02, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomer-
y, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyen-
gar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N.
Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y.
Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg,
V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O.
Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Fores-
man, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski,
B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L.
Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Na-
nayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen,
M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Pittsburgh
PA, 2003.
[23]a) A. D. Becke, J. Chem. Phys. 1993, 98, 1372; b) H. L. Schmider,
A. D. Becke, J. Chem. Phys. 1998, 108, 9624.
[24]a) W. Stevens, H. Basch, J. Krauss, J. Chem. Phys. 1984, 81, 6026;
b) T. R. Cundary, W. Stevens, J. Chem. Phys. 1993, 98, 5555.
[25]a) W. R. Wadt, P. J. Hay, J. Chem. Phys. 1985, 82, 270; b) P. J. Hay,
W. R. Wadt, J. Chem. Phys. 1985, 82, 284; c) W. R. Wadt, P. J. Hay, J.
Chem. Phys. 1985, 82, 299.
[26]For a more comprehensive treatment of solvation models, see: C. J.
Cramer, D. G. Truhlar, Chem. Rev. 1999, 99, 2161.
[27]V. Barone, M. Cossi, J. Tomasi, J. Chem. Phys. 1997, 107, 3210.
[10]a) E. Fasani, F. F. Barberis Negra, M. Mella, S. Monti, A. Albini, J.
Org. Chem. 1999, 64, 5388; b) H. de Vries, G. Beijersbergen van He-
negouwen, J. Photochem. Photobiol. B 2000, 58, 6.
[11]M. C. Cuquerella, M. A. Miranda, F. Boscà, J. Phys. Chem. B 2006,
110, 6441.
[12]P. J. Stang in Divalent Carbocations (Eds.: Z. Rappoport, P. J. Stang)
Wiley, Chichester, 1997, p. 451.
[13]M. Fagnoni, A. Albini, Acc. Chem. Res. 2005, 38, 713.
ꢁ
[14]Examples of intramolecular insertion into a C H bond have been
Received: July 17, 2007
Published online: October 17, 2007
reported previously with a o-diazoniumdibenzylbenzamide salt (see
Chem. Eur. J. 2008, 14, 653 – 663
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
663