Photoinduced C-F bond cleavage in some fluorinated 7-amino-4-quinolone- 3-carboxylic acids

Article abstract of DOI:10.1021/jo982456t

The photochemistry of some fluorinated 7-amino-4-quinolone-3-carboxylic acids used in therapy as antibacterials and known to be phototoxic has been investigated in water. All of them undergo heterolytic defluorination, and this appears to be a path for the generation of aryl cations in solution. 6- Fluoro derivatives such as norfloxacin (Φ(dec) = 0.06) and enoxacin (Φ(dec) = 0.13) give the corresponding phenols. Insertion of an electron-donating substituent makes defluorination inefficient; thus, ofloxacin, an 8-alkoxy derivative, is found to be rather photostable (Φ(dec) = 0.001) and reacts in part via a process different from defluorination (degradation of the N-alkyl side chain). With a 6,8-difluoro derivative, lomefloxacin, the reaction is more efficient (? = 0.55) and selective for position 8. Contrary to the previous cases, the aryl cation undergoes insertion in the neighboring N- ethyl group rather than solvent addition (a carbene-like chemistry). With all of the above fluoroquinolones an intensive triplet-triplet absorption is detected and is quenched by sulfite (k(q) = (1-5) x 10⁸ M^-1 s^-1). Under this condition, reductive defluorination via the radical anion takes place. The relation of the above chemistry to the phototoxicity of these drugs is commented upon briefly.

Full text of DOI:10.1021/jo982456t

5388
J . Org. Chem. 1999, 64, 5388-5395
P h otoin d u ced C-F Bon d Clea va ge in Som e F lu or in a ted
7-Am in o-4-qu in olon e-3-ca r boxylic Acid s
E. Fasani,† F. F. Barberis Negra,† M. Mella,† S. Monti,‡ and A. Albini*,†
Department Organic Chemistry, University of Pavia, v. le Taramelli 10, 27100 Pavia, Italy, and CNR
Inst. Photochemistry, v. Gobetti 101, 40129 Bologna, Italy
Received December 16, 1998
The photochemistry of some fluorinated 7-amino-4-quinolone-3-carboxylic acids used in therapy
as antibacterials and known to be phototoxic has been investigated in water. All of them undergo
heterolytic defluorination, and this appears to be a path for the generation of aryl cations in solution.
6-Fluoro derivatives such as norfloxacin (Φdec ) 0.06) and enoxacin (Φdec ) 0.13) give the
corresponding phenols. Insertion of an electron-donating substituent makes defluorination inef-
ficient; thus, ofloxacin, an 8-alkoxy derivative, is found to be rather photostable (Φdec ) 0.001) and
reacts in part via a process different from defluorination (degradation of the N-alkyl side chain).
With a 6,8-difluoro derivative, lomefloxacin, the reaction is more efficient (Φ ) 0.55) and selective
for position 8. Contrary to the previous cases, the aryl cation undergoes insertion in the neighboring
N-ethyl group rather than solvent addition (a carbene-like chemistry). With all of the above
fluoroquinolones an intensive triplet-triplet absorption is detected and is quenched by sulfite (kq
) (1-5) × 108 M-1 s-1). Under this condition, reductive defluorination via the radical anion takes
place. The relation of the above chemistry to the phototoxicity of these drugs is commented upon
briefly.
The great chemical inertness of the carbon-fluorine
by Chignell et al.6,7 has shown that both 1O2 and O2
•-
bond allows the use of fluorinated derivatives, e.g.,
fluoroaromatic or heteroaromatic compounds, as drugs.
Although reaction of the C-F bond has been obtained in
these derivatives only under strong reductive conditions,
the situation may be different in a photochemical reac-
tion. We were attracted to this field by the numerous
recent reports of light-related adverse effects of fluoro-
quinolones (FQs), a family of drugs that have excellent
antibacterial properties and are frequently prescribed.
Unfortunately, these have a phototoxic1 and photoaller-
genic effect2 as well as, in some cases, a photomutagenic
and phototumorigenic effect.3 This is a major limitation
to their use in therapy.
are indeed produced, but rather inefficiently, and at any
rate that there is no correlation between the relative
efficiency in activating oxygen and the phototoxic poten-
tial of the same series of FQs.
Another possibility is that FQs react photochemically
with some cell component. Evaluation of this path
requires that the photochemistry of these drugs is
ascertained. At present there are sparse reports about
some FQ photoreactions, in some case involving the
isolation of the photoproducts, in other cases only partial
evidence about the type of reaction occurring. Reported
reactions include dimerization for ciprofloxacin8 and
degradation of the alkylamino side chain for levofloxacin.9
Recently, indication that defluorination occurs for FQs
containing a fluorine atom in position 8, such as lom-
efloxacin,10,11 fleroxacin,11 and orbifloxacin,5,12 has been
reported.
The chemical mechanism underlying such effects has
not been clarified up to now. Evidence has been reported
for the role of reactive oxygen species such as singlet
oxygen, the superoxide anion, or the hydroxyl radical4,5
to the FQs induced phototoxicity. However, a recent study
Thus, it appeared worthwhile to establish the photo-
reactivity of a series of FQs in aqueous solution under
uniform conditions. The present work has been carried
out on four drugs of this family13 and shows that
heterolytic defluorination is consistently the main pro-
* To whom correspondence should be addressed. Fax +39 382
507323. E-mail: albini@chifis.unipv.it.
(1) Ferguson, J . Photochem. Photobiol. 1995, 62, 954. Ball, P.;
Tillotson, G. Drug Saf. 1995, 13, 343. Lietman, P. S. Drugs 1995, 49,
794. Norrby, S. R.; Lietman, P. S. Drugs 1993, 45 (Suppl. 3), 59.
(2) (a) Horio, T.; Miyauchi, H.; Aoki, Y.; Harada, M. J . Dermatol.
Sci. 1994, 7, 130. (b) Tokura, Y.; Nishijma, T.; Yagi, H.; Furukawa,
F.; Takigawa, M.. Photochem. Photobiol. 1996, 64, 838.
(6) Bliski, P.; Martinez, L. J .; Koker, E. B.; Chignell, C. F. Photo-
chem. Photobiol. 1996, 64, 496.
(3) (a) Klecak, G.; Urbach, F.; Urwyler, H. J . Photochem. Photobiol.
B 1997, 37, 174. (b) Maekinen, M.; Forbes, P. D.; Stenbaek, F. J .
Photochem. Photobiol. B 1997, 37, 182. (c) Urbach, F. J . Photochem.
Photobiol. B 1997, 37, 169. (d) Reavy, H. J .; Traynor, N. J .; Gibbs, N.
K. 1997 Photochem. Photobiol. 1997, 66, 368.
(7) Martinez, L. J .; Sik, R. H.; Chignell, C. F. Photochem. Photobiol.
1998, 67, 399.
(8) Tiefenbacher, E. V.; Haen, E.; Przybilla, B.; Kurz, H. J . Pharm.
Sci. 1994, 83, 463.
(9) Yoshida, Y.; Sato, E.; Moroi, R. Drug. Res. 1993, 43, 601.
(10) Fasani, E.; Mella, M.; Caccia, D.; Tassi, S.; Fagnoni, M.; Albini,
A. Chem. Commun. 1997, 1329. (b) Fasani, E.; Profumo, A., Albini, A.
Photochem. Photobiol. 1998, 68, 666.
(11) Martinez, L.; Li, G.; Chignell, C. F. Photochem. Photobiol. 1997,
65, 599.
(12) Morimura, T.; Nobuhara, Y.; Matsukura, H. Chem. Pharm.
Bull. 1997, 45, 373.
(13) For a preliminary report on some photoreactions of lomefloxa-
cin, see ref 10a.
(4) Robertson, D. G.; Epling, G. A.; Klely, J . S.; Bailey, D. L.; Song,
B. Toxicol. Appl. Pharmacol. 1992, 111, 221. (b) Wagai, N.; Tawara,
K. Free Rad. Res. Commun. 1992, 17, 387. (c) Wagai, N.; Tawara, K.
Arch. Toxicol. 1992, 66, 392-397. (d) Umezawa, N.; Arakane, K.; Ryu,
A.; Mashiko, S.; Hirobe, M.; Nagano, T. Arch. Biochem. Biophys. 1997,
342, 275. (e) Shimoda, K.; Nomura, M.; Kato, M. Fundam. Appl.
Toxicol. 1996, 31, 133-140.
(5) Morimura, T.; Kohno, K.; Nobuhara, Y.; Matsukura, H. Chem.
Pharm. Bull. 1997, 45, 5.
10.1021/jo982456t CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/07/1999

Photoinduced C-F Bond Cleavage
J . Org. Chem., Vol. 64, No. 15, 1999 5389
Sch em e 1
Sch em e 4
Sch em e 2
Sch em e 5
Sch em e 3
The study was carried out in dilute ((1-2) × 10-4 M)
neutral aqueous solution. The fluoroquinolones are amino
acids and are present as zwitterions in the ca. pH 6-8
range.15 Product isolation and characterization involved
previous functionalization of the photolysis mixture (in
most cases by acylation of the amino group, extraction,
methylation of the carboxylic group, and chromatographic
separation, see Experimental Section). Smaller scale, but
otherwise identical, experiments were also carried out
and monitored by HPLC. These served for measuring the
quantum yield of reaction (Φdec), as well as for the
cess. In one case, the resulting aryl cation undergoes an
unexpected “carbene-like” chemistry.
Resu lts
The quinolones used in therapy are 1,4-dihydro-1-alkyl-
4-oxoquinoline-3-carboxylic acids. More precisely, the
drugs most largely used at present (third-generation
quinolones) all bear a fluoro in position 6 and a â-ami-
noethylamino group in position 7 as additional substit-
uents.14 Therefore, we chose for the present study the
1-ethyl-6-fluoro-7-piperazinyl derivative (norfloxacin, 1a )
as the reference term as well as some of its simple
derivatives. These were selected as examples of the
structural variations adopted in pharmacological re-
search with the aim of modulating the activity. Apart
from a modification of the alkyl chains, presumed not to
affect the photochemical behavior, such elaboration
mostly involves position 8 in the quinolone ring. There-
fore, we chose an 8-aza derivative (enoxacin, 1b), an
8-fluoro derivative, (lomefloxacin, 1c), and an 8-alkoxy
derivative (ofloxacin, 1d ) (see Schemes 1-5). This choice
allowed the exploration of the structure effect on the
photoreactivity as well as comparing four out of five of
the FQs commonly used in present clinical practice.
-
potentiometric measure of the fluoride released (ΦF ).
Irradiation of an argon-flushed solution of 1a led to
liberation of fluoride and formation essentially of a single
compound. This was derivatized as indicated above and
isolated. The product thus obtained was identified as a
derivative of phenol 2a from its analytical and spectro-
scopic properties (see Scheme 1; in this case, as well as
in the following ones, the nonderivatized compounds are
(15) Ross, D. L.; Riley, C. M. Int. J . Pharm. 1992, 83, 267. Ross, D.
L.; Riley, C. M. Int. J . Pharm. 1990, 63, 237.
(16) Unpublished work by one of the present authors (S.M.). (b)
Sortino, S.; De Guidi, G.; Giuffrida, S.; Monti, S.; Velardita, A.
Photochem. Photobiol. 1998, 67, 167.
(17) Taken from Figure 1 in: Bilski, P.; Martinez, L. J .; Koker, E.
B.; Chignell, C. F. Photochem. Photobiol. 1998, 68, 20.
(18) Cornelisse, J . In CRC Handbook of Organic Photochemistry and
Photobiology; Horspool, W. H., Song, P. S., Eds.; CRC Pub.: Boca
Raton, 1995; p 250. Cornelisse, J .; Havinga, E. Chem. Rev. 1975, 75,
353. Mutai, K. Nakagaki, R.; Tukada, H. Bull. Chem. Soc. J pn. 1985,
58, 2066. Van Riel, H. C. H. A.; Lodder, G.; Havinga, E. J . Am. Chem.
Soc. 1981, 103, 7257. Van Eijk, A. M. J .; Huizer, A. H.; Varma, C. A.
G. O.; Marquet, J . J . Am. Chem. Soc. 1989, 111, 88. Van Eijk, A. M.
J .; Huizer, A. H.; Varma, C. A. G. O. J . Photochem. 1985, 29, 415.
(14) Neu, H. C. In The 4-Quinolones; Crumplin, G. C., Ed.; Springer-
Verlag: London, 1990; p 1.

5390 J . Org. Chem., Vol. 64, No. 15, 1999
Fasani et al.
Ta ble 1. P a r a m eter s for th e P h otoch em ica l Rea ction s of
F lu or oqu in olon es 1a -c
reaction quantum yielda (%)
2-
b
sub-
strate
Na2SO3/
kq(SO3 )
H2O
0.06
0.13
0.55
H2O/air NaClO4 NaHSO3 τF (ns) (M-1 s-1
)
1a
1b
1c
1d
0.01
0.08
0.5
0.09
0.19
0.5
0.004
0.027
0.28
<1
<1
<1
4
1.2 × 108
4.8
>1
1.0
0.0012c 0.0016d 0.0013c e0.0001e
a
Quantum yield for the substrate consumption. Unless other-
wise noted, an equimolecular amount of fluoride ((5%) is liberated.
b
Quenching rate by sulfite of the transient absorption at 600 nm.
c 80% fluoride liberated. 85% fluoride liberated. e Amount of
d
fluoride too low for an accurate measurement.
indicated in the schemes; for the actually characterized
compounds, see the Experimental Section). The quantum
yield for the consumption of 1a was 0.06 (see Table 1),
and the molar amount of fluoride set free was equal
within 5% to consumed 1a . The solutions used were
buffered at pH 7.2 by adding 5 × 10-4 M NaHCO3 (the
natural pH of 1a and of the other FQs is 6.1-6.5). In
this way, the pH did not change appreciably during the
experiments. With an unbuffered solution hydrogen
fluoride release caused a pH drop down to ca. 4.5 at
complete conversion, but the reaction course did not
change significantly.
Irradiation of an air-equilibrated solution led to the
same process, but with a diminished quantum yield
(0.01). The effect of salts was also tested (with Ar-flushed
solutions). In the presence of 0.1 M NaClO4, the quantum
yield for defluorination to give 2a was enhanced to 0.09.
In 0.02 M NaHSO3-NaSO3 buffer (pH 7), on the other
hand, a different defluorination process occurred. This
led to the 6-unsubstituted quinolone 3a with a low
quantum yield (0.004).
The naphthyridine derivative 1b likewise underwent
clean defluorination (100% of the substrate consumption)
to give phenol 2b in neat water (or NaHCO3 buffer) with
Φ ) 0.13, decreasing to 0.08 in air-equilibrated solution
and increasing to 0.19 in the presence of 0.1 M NaClO4.
In sulfite buffer reductive defluorination to give 3b was
observed instead (Φ ) 0.027, see Scheme 1 and Table 1).
The 6,8-difluoro derivative 1c turned out to react more
efficiently than the previous derivatives (Φ ) 0.55). The
process involved mono-defluorination selectively from
position 8; no phenol was isolated, contrary to the case
of 1a ,b, and the main product was the tricyclic derivative
4 (see Scheme 2). In this case, the effect of both oxygen
and NaClO4 on the reaction quantum yield was minimal
(10%). In sulfite buffer, the efficiency of the photoreaction
was diminished, though not as strongly as in the previous
cases (Φ ) 0.28), and gave two mono-defluorinated
derivatives, the above tricyclic compound 4 and quinolone
3c. Increasing the sulfite buffer concentration up to 0.1
M caused neither a further drop of the quantum yield
nor a change in the product distribution.
F igu r e 1. (a) Absorption changes observed in a 7.3 × 10-5
M
solution of enoxacin in Ar-saturated sulfite buffer 0.1 M, pH
7.06, upon excitation with a 2 mJ , 20 ns laser pulse at 355
nm: (b) 20 ns, end-of-pulse; (O) 35 ns and (*) 250 ns after
time “zero”(see Experimental Section). (b) Dependence of the
rate constants extracted by biexponential treatment of the time
profiles of ∆A at 700 nm, k1 and k2, on the sulfite anion
concentration. Inset: time profile of ∆A at 700 nm at 0.01 M
sulfite and biexponential fit corresponding to 1/τ1 ) k1 ) 0.71
× 107 s-1 and 1/τ2 ) k2 ) 0.22 × 107 s-1
.
tion was a fluorine-containing derivative, diamide 5
(Scheme 3). In the presence of sulfite, the quantum yield
dropped by 1 order of magnitude, and we were unable to
identify any photoproduct.
The above quinolones showed a blue fluorescence with
a maximum around 380 nm, with the difference that
lifetime was <1 ns for 1a -c, while 1d showed τ ) 4 ns.
Addition of 0.02 M sulfite caused only a minor effect on
the fluorescence.
Laser flash photolysis of the four quinolones 1a -d in
aqueous solutions led in every case to transient absorp-
tion in the 500-650 nm region, which was attributed to
the population of the triplet state.16a The transients, with
lifetimes in the microsecond or submicrosecond range,
were quenched in the presence of sulfite, and rate
constants in the order of 108 M-1 s-1 were measured (see
Table 1).
With 1b a more detailed investigation of the transient
behavior in aqueous solution and in the presence of
sulfite at pH 7 was performed. The transient spectra
observed with 0.1 M sulfite at several delays after the
laser are shown in Figure 1a. At the end of the 20 ns
laser pulse the absorption spectrum showed two bands.
The first one, with a maximum at 500 nm, was attributed
to the triplet state (compare ref 16b), while the latter
band (maximum at 670 nm) was due to a second
transient resulting from sulfite quenching of the triplet
The 8-alkoxyquinolone 1d was the most photostable
of the series. The decomposition quantum yield in Ar-
flushed water was 0.0012 and changed only marginally
in air-equilibrated solution or in the presence of 0.1 M
NaClO4 (Table 1). Contrary to the above cases, defluori-
nation was not the only process, though it remained
predominant, since the fluoride liberated corresponded
to 80% of the substrate decomposed. However, the only
product isolated and recognized from preparative irradia-

Photoinduced C-F Bond Cleavage
J . Org. Chem., Vol. 64, No. 15, 1999 5391
state during the laser pulse. During the following 15 ns,
in fact, the 500 nm band disappeared, while the 670 nm
band further increased. The kinetics of the absorbance
changes over the whole wavelength range was well
described by a biexponential function. The time constants
τ1 and τ2 were ca. 20 and 75 ns, respectively. After 230
ns, the transient absorption was characterized by a large
band with maximum at ca. 650 nm, which decayed in
the scale of tens of microseconds. In the absence of sulfite
the 500 nm absorption decay was monoexponential with
a time constant of 850 ns; during the triplet decay no
increase of the absorption in the region 600-800 nm or
formation of the longer-lived 650 nm species was ob-
served.
reductive defluorination, observed in sulfite buffer with
all quinolones except 1d ; and (d) oxidative degradation
of the alkylamino side chain, observed with 1d in a slow
process.
Defluorination according to paths a or b (or c in the
presence of sulfite) was the only reaction observed for
1a -c. In every case, the amount of HF released was
identical ((5%) to the extent of the photodecomposition.
The quantum yield for this process was diminished under
air with 1a and 1b, but only slightly in the case of 1c.
On the contrary, a significant fraction of the (slow)
photoreaction of 1d led to a product conserving the
fluorine (path d), and in this case the quantum yield was
higher under air.
The study of the effect of the sulfite concentration on
the kinetics of the fast processes showed that, besides
the triplet, the 670 nm transient was also quenched by
sulfite. In Figure 1b, the linear dependence of the two
rate constants, extracted by biexponential analysis of the
absorbance changes, on the sulfite concentration is
reported. The values of kq ) 4.8 × 108 M-1 s-1 for the
triplet and kq ) 1.2 × 108 M-1 s-1 for the 670 nm
transient were derived. In the inset of Figure 1b the time
profile at 700 nm and the corresponding best fitting for
0.01 M sulfite are reported.
The photochemical behavior of 1b in phosphate buffer
0.01 M had been previously investigated.16b The triplet
decay rate had been reported to be 1.1 × 107 s-1. Since
in aqueous solution, in the absence of phosphate, the
triplet lifetime of 1b is 850 ns, it is concluded that
quenching of the triplet by the phosphate ions is also
operative. This effect, which is determinant for the
medium-dependent photodecomposition of fluoroquino-
lones,10b is presently under investigation.
Excited Sta tes In volved . Light excitation of the
quinolones under study populated the triplet by ISC from
the singlet state, as indicated both by the observation of
phophorescence in a glassy matrix and by the transient
observed by nanosecond flash photolysis. The case of 1b
has been examined in some detail. The strong absorption
with a maximum at 500 nm corresponds to the T-T
absorption and has a lifetime of 850 ns in water at pH
7.2. As expected, it was quenched by oxygen in the same
way as the photochemical reaction. The same transient
had been previously reported in phosphate buffer, where
the lifetime is shorter, and evidence for its attribution
to the triplet and it being the photoreactive state had
been reported.16b A value of ΦISC ) 0.85 was evaluated,
though we must caution that this may be overestimated.16b
We observed that the intensity of the end-of-pulse
transient absorption was similar with all the quinolones,
except for 1c where it is weaker. Furthermore, there are
some differences in the reaction of 1c on one hand and
1a ,b on the other one.
The above fluoroquinolones are soluble only in polar
media. The photochemical behavior in organic solutions
was not systematically investigated. However, explor-
ative irradiations of 1b and 1c in trifluoroethanol and
in methanol showed that the photoreactivity was much
lower than in water.
First, the photodecomposition of 1a and 1b was only
moderately efficient, and both the transient and the
photoreaction were quenched by oxygen, while 1c de-
composed quite efficiently and oxygen quenching was
limited. Second, sulfite quenched both the transient and
process a (for 1a ,b) or b (for 1c), but to a different degree.
With 0.02 M sulfite buffer, the reaction quantum yield
for 1c dropped by 24% and a new product, compound 3c,
was formed in addition to that formed in water (com-
pound 4). An increase of the sulfite concetration up to
0.1 M caused no further change. With 1a and 1b, on the
other hand, the drop in quantum yield was much larger
(by a factor of 15 and 5, respectively) and the reduced
quinolones 3a ,b substituted the phenols (2a ,b) formed in
water. Third, with none of the above substrates the
absorption spectrum as well as the fluorescence spectrum
and efficiency were significantly affected by the addition
of 0.02 M sulfite, while as seen above the reaction
quantum yield dropped, and dramatically so, with 1a and
1b.
Discu ssion
Rea ction Obser ved . The present data show that
fluoroquinolones are photosensitive, though with largely
different efficiency. The present investigation was carried
out with lamps centered at 320 nm or with a Pyrex-
filtered mercury arc (λ > 300 nm), but the absorption of
these drugs extended to ca. 360 nm, and thus, the UV-A
component of both natural and artificial indoor light is
active. Indeed, dilute solutions of the most sensitive of
these drugs, viz. 1c, decomposed in hours under the
ordinary laboratory illuminations, and it is not surprising
that a phototoxic effect is so easily observed. The photo-
chemistry of these compounds is quite varied as far as
both the product distribution and the reaction quantum
yield (the latter changing over 4 orders of magnitude
depending on structure and medium) are concerned. The
observed photoreactions were as follows: (a) nucleophilic
substitution of fluoride, observed with both 1a and 1b
and yielding the corresponding phenols (defluorination
occurred also with 1d , as apparent from the fluoride set
free, though with much lower quantum yield than with
the other FQs (in this case, however, no defluorinated
product was isolated)); (b) substitution of fluoride through
an intramolecular alkylation, as observed with 1c; (c)
Therefore, the substitution of a hydroxyl for a fluoro
group in quinolones 1a ,b (reaction a) undoubtedly oc-
curred via the triplet state; the more efficient defluori-
nation of 1c (reaction b) arose again from the triplet
(shorter lived and thus less easily quenched) and in part
from the singlet state. As for process c, the reductive
defluorination observed with all three of the above
quinolones in the presence of sulfite, this involved in
every case the triplet as demonstrated by the effect of
this ion on the decay of the T-T absorption (see below
for the path involved).

5392 J . Org. Chem., Vol. 64, No. 15, 1999
Fasani et al.
Deflu or in a tion via th e Excited Sta te: Ar yl Ca tion
Rea ctivity. As seen above, C-F bond cleavage was the
only primary photochemical event with 1a -c. This
cannot be a homolytic process, since the BDE of the
aromatic C-F bond (ca. 120 kcal M-1) largely exceeds
that of the singlet (ca. 82 kcal M-1, from fluorescence)
and the triplet state (ca. 69 kcal M-1, from phosphores-
cence) of these molecules. Two paths are viable, viz. the
addition-elimination mechanism via a σ complex and
heterolytic fragmentation (see Scheme 4 for the case of
1a ,b). The former mechanism has been demonstrated for
several photoinduced aromatic substitutions, usually
occurring via the triplet state.18 The latter one has been
recently suggested for a few electron-donating substituted
fluoroaromatics19 and is favored with the present quino-
lones by the internal charge-transfer character toward
the ring in the ππ* excited state (see formula 1*). This
is due to the presence of both electron-donating (amino)
and electron-withdrawing (fluorine) substituents and is
evidenced by the considerable Stokes shift of the fluo-
rescence (e.g., for 1b 5000 cm-1 in dioxane, growing to
8000 in D2O)17 and the low fluorescence lifetime (<1 ns
for 1a -c) typical of such push-pull states.
and that the final product arises through cyclization onto
the N-ethyl group, not by solvolysis. The regioselectivity
can be qualitatively understood by considering the struc-
ture of the intermediate cation. An aryl cation in position
6 (6, see Scheme 4) receives no stabilization from the
neighboring amino group, since this is only possible in
mesomeric formulas such 6′, which are unimportant
because they are no more aromatic. However, the situ-
ation is different for a cation in position 8, such as 7 from
1c (Scheme 5), since this gains a significant stabilization
through the contribution of structure 7′, where the charge
is on the nitrogen and aromaticity is conserved in the
pyridone moiety.20 The participation of this mesomer
explains why C-F cleavage is so much faster from
position 8 than from position 6. As a consequence,
defluorination is regioselective and more effective
with 1c.
Furthermore, the structure of the cation allows us to
understand the different chemistry observed. The obvious
reaction from σ cation 6 (X ) CH, N) is water addition,
and thus, if this is the intermediate, it is reasonable that
phenols 2a and 2b are the exclusive products (Scheme
4). In the case of cation 7, however, the charge is
distributed between the carbon and the nitrogen atoms
through the contribution of mesomeric formula 7′. This
is a π cation where the carbon atom in position 8 is a
carbene. Indeed, in this case, the exclusive process is a
typical carbene reaction, intramolecular insertion into a
C-H bond of the CH3 in the neighboring N-ethyl group
(Scheme 5). This may be envisaged as a two-step radical
or a concerted reaction; in any case, the attack occurs at
the more accessible position (i.e., at the unactivated
methyl group, not at the methylene R to the amino
group), as it is general in carbene reactions.21
The medium and structure effects on the reaction were
in accord with both mechanisms. That a ionic mechanism
was involved was indicated by the occurrence of the
photocleavage in water but only to a much lesser degree
in a solvent with a lower dielectric constant such as
methanol or trifluoroethanol, and by the reactivity
increase upon increasing the ionic strength (see the
experiments with NaClO4). Furthermore, the reaction
quantum yield (in argon flushed water) decreased in the
order Φdec(1c) ) 4 × Φdec(1b) ) 9 × Φdec(1a ) ) 460 ×
Aryl cations are known to be high-energy, and thus
quite unselective, intermediates22 and have been spec-
troscopically characterized in a rigid matrix.23 In solu-
tion,24 these are intermediates in the displacement
reactions of arenediazonium cations (in the absence of
strong bases, reducing agents, or light)25 and in a few
solvolyses.26 There is both computational and experimen-
tal evidence for mesomeric25a,27 and hyperconjugative26
stabilization of such species, but no indication of a change
in the reaction course. The present data offer a new
insight, by demonstrating a clear-cut change of the
chemical behavior in a 2-aminoaryl cation where π
Φ
dec(1d ), showing a correlation with the electronegativity
of the group in position 8, respectively dC(-F)-, dN-,
dC(-H)-, and dC(-OR)-. This may be rationalized as
due either to electron-withdrawing groups making nu-
cleophile attack on the excited-state easier or to these
groups increasing charge-transfer toward the ring in the
excited state and, thus, the incipient heterolytic weaken-
ing of the C-F bond. On the other hand, a mesomeric
donating group such as the alkoxy group in 1d both slows
down the addition of nucleophiles and balances the
charge transfer.
(20) The stabilisation of the pyridone structure via neutral and
zwitterionic mesomers is well-known, see, e.g.: Katritzky, A. R.;
Lagowski, J . M. Adv. Heteroatom. Chem. 1963, 1, 339. Elguero, J .;
Marzin, C.; Katritzky, A. R.; Linda, P. The Tautomerism of Hetero-
cycles; Academic Press: New York, 1976; p 71. Hall, G. G.; Hardisson,
A.; J ackman, L. M. Tetrahedron 1963, 19 (Suppl. 2), 101.
(21) See, e.g.: Majerski, Z.; Hamersak, Z.; Sarac-Arneri, R. J . Org.
Chem. 1988, 53, 5053. Paquette, L. A.; Wilson, J . E.; Menzal, R. P.;
Allen, G. R. J . Am. Chem. Soc. 1972, 94, 7761. Sasaki, T.; Eguchi, S.;
Kiriyama, T. J . Am. Chem. Soc. 1963, 91, 212.
Since the fluoroquinolones are amphoteric, it is impos-
sible to explore the dependence of the photosubstitution
quantum yield on [OH-], as has been done for known
bimolecular photosubstitution. However, the fact that
defluorination is most efficient in the case of 1c, where
no hydroxyl group is incorporated, supports that unimo-
lecular cleavage is the mechanism.
The photochemistry of quinolone 1c differs from that
of analogues 1a ,b in that defluorination is not only more
effective but also selective for position 8, not position 6,
(22) Zollinger, H. Angew. Chem., Int. Ed. Engl. 1978, 17, 141.
Lorand, J . P. Tetrahedron Lett. 1989, 30, 7337.
(23) Gasper, S. M.; Devadoss, C.; Schuster, G. B. J . Am. Chem. Soc.
1995, 117, 5206. Ambroz, H. B.; Kemp, T. J . J . Chem. Soc., Perkin
Trans. 2 1976, 1420. Ambroz, H. B.; Kemp, T. J . J . Chem. Soc., Perkin
Trans. 2 1980, 768. Ambroz, H. B.; Kemp, T. J . Chem. Soc. Rev. 1976,
8, 353. Scaiano, J . C.; Nguyen, K. T. J . Photochem. Photobiol. A: Chem.
1983, 23, 269. Chaudhuri, A.; Loughlin, J . A.; Romsted, L. S.; Yao, J .
J . Am. Chem. Soc. 1993, 115, 8351.
(19) Zhang, G.; Wan, P. Chem. Commun. 1994, 19. Yang, N. C.;
Huang, A.; Yang, D. H. J . Am. Chem. Soc. 1989, 111, 8069. Durand,
A. P.; Brown, R. G.; Worrall, D.; Wilkinson, F. J . Chem. Soc., Perkin
Trans. 2 1998, 365-370.
(24) Gas-phase studies have been also reported; see, e.g.: Angelini,
G.; Sparapani, C.; Speranza, M. Tetrahedron 1984, 40, 4865. Fornarini,
S.; Speranza, M. J . Am. Chem. Soc. 1985, 107, 5358. Angelini, G.;
Fornarini, S.; Speranza, M. J . Am. Chem. Soc. 1982, 104, 4773.

Photoinduced C-F Bond Cleavage
J . Org. Chem., Vol. 64, No. 15, 1999 5393
Sch em e 6
than to chemical reaction. The decomposition quantum
yield in sulfite buffer 0.02 M is lower than in neat water
(where defluorination proceeds directly from the excited
state), and the decrease is the stronger the less electron-
withdrawing is the substituent. With the less quenchable
1c the decrease is moderate (by a factor of 2). Among
the other fluoroquinolones, Φ decreases by a factor of 5
for 1b, of 15 for 1a , and of .10 for 1d . Thus, when the
radical anion is more stabilized (and presumably longer-
lived), defluorination by this mechanism is more competi-
tive. In the difluoro derivative 1c, the cleavage is again
selective for position 8, since the same factors that
stabilize the cation in position 8 similarly operate on the
radical.
The thus-formed σ aryl radicals 8 and 9, respectively,
are then reduced by sulfite to the corresponding anion,
and the ensuing protonation completes the observed
overall process, viz. reductive defluorination to give
quinolones 3a -c (see Scheme 6).
Contrary to the photochemistry in neat water, further
transients are observed in the presence of sulfite and can
be assigned with reference to Scheme 6. The 670 nm
transient (empty circles in Figure 1a) formed by quench-
ing of the triplet state of 1b by sulfite (k1 ) 4.8 × 108
M-1 s-1, see Figure 1b) and growing within 15 ns reason-
ably is the radical anion 1b•-. This species is relatively
long-lived, and the strong C-F bond cleaves inefficiently.
The fragmentation of this species presumably occurs as
a unimolecular process leading to radical 8b. The long-
living (tenths of microseconds) species absorbing at 650
nm that remains after decay of the two first transients
(see the lower spectrum in Figure 1a) is thus reasonably
radical 8b. In the final step, the radical is then further
reduced as mentioned above. It should be noticed that
the second transient is also quenched by sulfite (k1 ) 1.2
× 108 M-1 s-1, see Figure 1b), though the significance of
this phenomenon is uncertain at the moment.
P ip er a zin yl Sid e Ch a in Oxid a tion . The photochem-
istry of 1d clearly differs from that of the other quino-
lones. This is due to the fact that the alkoxy group in
position 8 strongly limits the charge transfer to the ring
in the excited state (see above). In fact, in contrast to
the other three quinolones considered, the much less
reactive 1d has τfluo ) 4 ns, and thus, the singletsand
presumably also the tripletsstate come nearer to “nor-
mal” ππ* states, with low ICT character. This hinders
defluorination, and the quantum yield of decomposition
drops in the order of 10-3. Potentiometric analysis reveals
that fluoride is released also in this case, but the amount
of F- observed is less than 100% of decomposed 1d . We
did not isolate fluorine-free photoproducts in this case,
nor did Yoshida et al. in a previous study in air-
equilibrated water.9 The product isolated arises from
oxidative degradation of the piperazinyl side chain. Such
a degradation is a common occurrence in the photochem-
istry of dialkyarylamines and often involves triplet-
sensitized hydrogen abstraction followed by oxygen ad-
dition or further oxidation of the R-amino radical; quantum
yields generally are in the same order as observed here.33
Since long irradiation times are required (notice also that
this is the only quinolone for which the degradation in
the presence of air is somewhat faster than under
deoxygenated condition), it appears likely that in pre-
delocalization is possible: in this case, a carbene-like
chemistry is observed instead of the “normal” nucleophile
addition.
Deflu or in a tion via th e Ra d ica l An ion . The sulfite
anion quenches the triplet state and causes reductive
defluorination of the quinolones. These heterocycles are
expected to be easily reduced in the excited state. As an
example, for 1b, Ered) -1.30 V vs SCE in water,28 and
thus Ered(1b3*) ) Ered + Eexc ) 2.70 V. Thus, it is
reasonable that the efficient triplet quenching (rate
constant >108 M-1 s-1) is an electron-transfer process
leading to the radical anion (1a -c•-). This species frag-
ments to the corresponding σ aryl radical (8 or 9, see
Scheme 6) with structure-dependent efficiency. The fluo-
robenzene radical anion has been characterized,29 and
there are few known defluorinations via the radical
anion,30 although in polysubstituted aromatics other
substituents are as a rule lost in preference to fluo-
rine.31,32 Thus, defluorination from the radical anion is
expected to occur inefficiently, even though it is facili-
tated with the present substrates by the substituent
stabilization of the resulting aryl radicals. Indeed, electron-
transfer quenching of the excited states by sulfite mainly
leads to physical decay via back-electron transfer rather
(25) Swain, C. G.; Sheats, J . E.; Harbinson, K. G. J . Am. Chem. Soc.
1975, 97, 783. Bergstrom, R. G.; Landells, R. G. M.; Wahl, G. W.;
Zollinger, H. J . Am. Chem. Soc. 1976, 98, 3301. Speranza, M.; Keheyan,
Y.; Angelini, G. J . Am. Chem. Soc. 1983, 105, 6377.
(26) Apeloig, Y.; Arad, D. J . Am. Chem. Soc. 1985, 107, 5285.
Himeshima, Y.; Kabayashi, H.; Sonoda, T. J . Am. Chem. Soc. 1985,
107, 5286.
(27) Dill, J . D.; Schleyer, P. v. R.; Pople, J . A. J . Am. Chem. Soc.
1977, 99, 1. Ambroz, H. B.; Kemp, T. J .; Przybytniak, G. K. J .
Photochem. Photobiol. A: Chem. 1997, 108, 149. Fanghaenel, E.;
Kriwanek, J .; Moerke, W. J . Prakt. Chem. 1985, 327, 117. Ambroz, H.
B.; Kemp, T. J .; Przybytniak, G. K. J . Photochem. Photobiol. A: Chem.
1992, 68, 85.
(28) Unpublished results by A. Profumo from this laboratory.
(29) Andrieux, C. P.; Blocman, C.; Saveant, J . M. J . Electroanal.
Chem. Interfacial Electrochem. 1979, 105, 413.
(30) Kariv-Miller, E.; Vajstnar, Z. J . Org. Chem. 1985, 50, 1394.
Rossi, R. A.; Bunnett, J . F. J . Am. Chem. Soc. 1972, 94, 683.
(31) Buick, A. R.; Kemp, T. J .; Neal, G. T.; Stone, T. J . J . Chem.
Soc. A 1969, 666. Hauser, K. J .; Bartack, D. E.; Hawley, H. D. J . Am.
Chem. Soc. 1973, 95, 6033. Volke, J .; Manousek, O.; Troyepolskaya,
T. V. J . Electroanal. Chem. Interfacial Electrochem. 1977, 85, 163.
(32) Bunnet, J . F. Acc. Chem. Res. 1978, 11, 413; 1992, 25, 2. Rossi,
R. A.; de Rossi, R. H. Aromatic Substitution by the SRN1 Mechanism;
American Chemical Society: Washington, D.C., 1983
(33) See, e.g.: Albini, A.; Fasani, E.; Pietra, S.; Sulpizio, A. J . Chem.
Soc., Perkin Trans. 2 1984, 1689.

5394 J . Org. Chem., Vol. 64, No. 15, 1999
Fasani et al.
Sch em e 7
parative experiments on dilute solutions traces of air
penetrate into the vessel and oxygen is involved in the
oxidation leading to the observed product 5. Therefore,
we have no firm evidence for a detailed mechanism in
this case, but it is apparent that the photochemistry of
1d can be classed along with that of other aromatic
amines, while it is very different from the peculiar
reactivity of 1a -c having an ICT state.
Con clu sion . This work reveals that irradiation in
water of amino-substituted (hetero)aryl fluorides leads,
in some case quite efficiently, to heterolytic fragmenta-
tion of the C-F bond, a process for which there is little
precedent.
From the chemical point of view, the main interest is
that a new access to aryl cations in solution is thus
opened as an alternative to the decomposition of diazo-
nium salts. From the present study, it appears that the
chemistry of such cations is not uniform, varying between
water addition and carbene-like intramolecular insertion
depending on the structure. It may be that a similar
study of simpler models will afford new evidence about
the structure and reactivity of aryl cations. Furthermore,
a different pathway for defluorination, in this case via
the radical anion, has been evidenced in the presence of
sulfite.
From the pharmaceutical point of view, the present
data suggest that incorporating a fluorine in a drug, a
practice often advantageously used in optimizing the
pharmacological profile and technical characteristics,
may on the other hand introduce an undesired photore-
activity, at least in aromatics. The correlation of the
photolability with the charge transfer to the ring in the
excited state suggests a way for predicting the signifi-
cance of such a side effect on the basis of the substituents.
The observed reactions may be relevant for the re-
ported phototoxic effects. Since, as seen above, oxygen
activation is not the mechanism, phototoxicity depends
on a photoreaction of the drug with a component of the
cell. We observed an efficient carbene-like chemistry with
the defluorinated cation from 1c (actually the most
phototoxic of the quinolones we studied).2a,3a,4c,34,35 In
water, this occurs intramolecularly and involves the
N-ethyl chain, but in the cell it is likely that such an
intermediate can find other C-H bonds for which inser-
tion is convenient, thus initiating the degradation of some
cell component.
P r ep a r a tive Ir r a d ia tion s. Solutions (1 to 2 × 10-4 M) of
the drugs in water or the appropriate buffer (1.4 L) were
irradiated in an immersion well apparatus by means of a
Pyrex-filtered 500 W medium-pressure mercury arc (Helios
Italquartz) at 17 °C. The course of the reaction was monitored
by HPLC (see below). When a convenient (50-90%) fraction
of the starting material had been consumed, the aqueous
solution was brought to neutrality and treated as follows.
Meth od A. The solution was extracted by 3 × 0.45 L of
chloroform with 1% ethyl chloroformate, and the organic phase
was washed with NaHCO3 to neutrality and then with water,
and then dried. Freshly prepared ethereal diazomethane was
added; after 30 min, the excess reagent was decomposed with
AcOH and the solution evaporated. Chromatography of the
residue on a silica gel column (120 g, 35-70 µm, chloroform-
methanol mixture from 99:1 to 90:10 as eluant) gave the
products.
Meth od B. Same as in A, omitting the treatment with
diazomethane.
Meth od C. The aqueous solution was evaporated under
reduced pressure.
The method used and the products characterized are given
in the following text. See Scheme 7 for the formulas of the
derivatized photoproducts.
1a in Wa ter (Meth od B). 1-Eth yl-1,4-d ih yd r o-6-h y-
dr oxy-4-oxo-7-(4-eth oxycar bon yl-1-piper azin yl)qu in olin e-
3-ca r boxylic a cid (2a ′): colorless crystals; mp > 260 °C
(nitroethane); 1H NMR [(CD3)2SO] δ 1.2 (t, 3H, J ) 7 Hz), 1.5
(t, 3H, J ) 7 Hz), 3.25 (m, 4H), 3.6 (m, 4H), 4.08 (q, 2H, J )
7 Hz), 4.55 (q, 2H, J ) 7 Hz), 7.08 (s, 1H), 7.68 (s, 1H), 8.82 (s,
1H), 10.2 (broad s, exch., 1H), 15.6 (broad s, exch, 1H); 13C
NMR [(CD3)2SO] δ 18.5, 18.55, 47.0, 53.0, 53.3, 54.8, 109.3,
110.4, 112.6, 124.4, 137.9, 150.3, 152.5, 153.0, 158.5, 170.7,
175.8. Anal. Calcd for C19H23N3O6: C, 58.60; H, 5.95; N, 10.79.
Found: C, 58.7; H, 5.9; N, 10.6.
Exp er im en ta l Section
Ma t er ia ls a n d In st r u m en t s. Nofloxacin (1a ), enoxacin
(1b), and ofloxacin (1d ) were purchased from Sigma Chemical
Co. (Milan) and used without further purification. Lomefloxa-
cin hydrochloride, from the same supplier, was dissolved in
water (0.02 M solution), and 1 M NaOH was added to
neutrality. The free base (1c, mp 231-4 °C) was extracted with
chloroform. All other chemicals and solvents were reagent
grade or better. The pH 7 sulfite buffer was prepared from
1.26 g of Na2SO3 and 0.95 g of Na2S2O5 per liter for 0.02 M
solution.
1a in Su lfite Bu ffer (Meth od B). 1-Eth yl-1,4-d ih yd r o-
4-oxo-7-(4-et h oxyca r b on yl-1-p ip er a zin lyl)q u in olin e-3-
ca r boxylic a cid (3a ′): colorless solid from chromatography,
containing a small amount of the ethylcarbamate of the
1
starting substrate 1a ; mass 373 m/e M+; H NMR (CDCl3) δ
1.3 (t, 3H, J ) 7 Hz), 1.6 (t, 3H, J ) 7 Hz), 3.35 (m, 4H), 3.75
(m, 4H), 4.2 (q, 2H, J ) 7 Hz), 4.45 (q, 2H, J ) 7 Hz), 6.6 (d,
1H, J ) 1 Hz), 7.06 (dd, 1H, J ) 1, 9 Hz), 8.2 (d, 1, J ) 9 Hz),
8.55 (s, 1H).
Fluorescence lifetimes were measured (in phosphate buffer)
by Dr. R. Argazzi (University of Ferrara) by means of a single-
photon-counting apparatus. The other data were obtained from
standard instruments.
1b in Wa ter (Meth od C). 1-Eth yl-6-h yd r oxy-1,4-d ih y-
d r o-4-oxo-7-(1-p ip er a zin yl)-1,8-n a p h th yr id in e-3-ca r box-
ylic a cid (2b): colorless crystals; mp > 260 °C; 1H NMR
[(CD3)2SO] δ 1.45 (t, 3H, J ) 7 Hz), 3.0 (m, 4H), 3.9 (m, 4H),
4.5 (q, 2H, J ) 4 Hz), 7.65 (s, 1H), 8.7 (s, 1H); 13C NMR [(CD3)2-
SO] δ 18.9, 48.6, 50.6, 50.8, 111.3, 117.9, 118.6, 145.9, 147.1,
149.3, 156.3, 170.4, 179.8. Anal. Calcd for C15H18N4O4: C,
(34) Chatelat, A. A.; Albertini, S.; Gocke, E. Mutagenesis 1996, 11,
497.
(35) Domagala, J . M. J . Antimicrob. Chemother. 1994, 33, 685.

Photoinduced C-F Bond Cleavage
J . Org. Chem., Vol. 64, No. 15, 1999 5395
56.59; H, 5.70; N, 17.60. Found: C, 56.7; H, 5.8; N, 17.3. When
treated with ethyl chloroformiate in chloroform, this compound
gave a material identical to that obtained by method B (see
below).
68.9, 105.8/105.2 (d, J C-F ) 23 Hz), 110.8/110.7, 120.0/119.9
(d, J C-F ) 18 Hz), 123.6, 128.8/128.7 (d, J C-F ) 7 Hz), 142.4
(d, J C-F ) 67 Hz), 145.6, 155.1/155.6 (d, J C-F ) 250 Hz), 162.6/
162.5, 163.3, 165.6/165.8, 172.0/172.1. Anal. Calcd for C19H22
-
Meth od B. 1-Eth yl-6-eth oxyca r bon yloxy-1,4-d ih yd r o-
4-oxo-7-(4-eth oxyca r bon yl-1-p ip er a zin yl)-1,8-n a p h th yr i-
d in e-3-ca r boxylic a cid (2b′): colorless crystals; mp 159 °C
FN3O4: C, 60.79; H, 5.91; N, 11.19. Found: C, 60.6: H, 6.0:
N, 11.0.
Sm a ll Sca le Exp er im en ts. Photochemical reactions were
carried out on 10 mL portions of aqueous solution of the drugs
(1 × 10-4 M) 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). When
required, deoxygenation of the solution was obtained by
flushing for 1 h with argon passed through a Supelco carrier
gas purifier.
The reactions were also carried out in spectrophotometric
quartz couvettes on an optical bench, in that case using 3 mL
samples of 0.5 or 1 × 10-4 M drug solution and irradiating by
a focused beam from an Osram 200 W high-pressure mercury
arc and an interference filter (transmission peak at 317 nm,
bandwidth 8 nm). These experiments as well as those above
were carried out at low conversion (<30, in most cases <20%),
and the reported Φ values are the average of at least three
independent measurements ((10%).
The substrate decomposition was determined by HPLC
(Waters model 501 apparatus) with optical detection (Waters
490E, λ 276 nm). An inverse-phase Merck Purosphere RP-18
e (5 µm) column (3 × 125 mm) was used. The eluant was an
85:15 mixture of pH 3 buffer (prepared by adding phosphoric
acid to a 0.75% solution of triethylamine until pH 3 was
reached) and acetonitrile. A few microliters of a solution of an
appropriate standard (1c for 1a , 1d for 1b and 1c, 1b for 1d )
were added to the photolyzate before analysis.
The fluoride concentration was measured by means of an
Orion SA520 potentiometer using a selective electrode (Orion
F-94-09) after addition to the 10 mL photolysate of 2 mL of
Orion TISAB III buffer (ammonium acetate, ammonium
chloride, 1,2-cyclohexylidene-R,δ-dinitrilotetraacetic acid) and
dilution to 22 mL with distilled water.
1
(methanol); H NMR (CDCl3) δ 1.3 (t, 3H, J ) 7 Hz), 1.42 (t,
3H, J ) 7 Hz), 1.5 (t, 3H, J ) 7 Hz), 3.68 (m, 4H), 3.78 (m,
4H), 41.18 (q, 2H, J ) 7 Hz), 4.38 (q, 2H, J ) 7 Hz), 4.42 (q,
2H, J ) 7 Hz), 8.7 (s, 1H), 8.8 (s, 1H), 14.9 (s, exch, 1H); 13C
NMR (CDCl3) δ 14.1, 14.5, 14.7, 43.2, 47.3, 47.6, 61.7, 65.9,
109.5, 114.5, 129.4, 135.9, 145.5, 146.2, 146.9, 152.2, 154.2,
166. 6, 177.2. Anal. Calcd for C18H22N4O6: C, 55.38; H, 5.68;
N, 14.35. Found: C, 55.3; H, 5.8; N, 14.2.
1b in Su lfite Bu ffer (Meth od B). 1-Eth yl-1,4-d ih yd r o-
4-oxo-7(4-eth oxyca r bon yl-1-p ip er a zin yl)-1,8-n a p h th yr i-
d in e-3-ca r boxylic a cid (3b′): light yellow solid from chro-
matography; mass 374 m/e M+; 1H NMR (CDCl3) δ 1.3 (t, 3H,
J ) 7 Hz), 1.6 (t, 3H, J ) 7 Hz), 3.65 (m, 4H), 3.75 (m, 4H),
4.2 (q, 2H, J ) 7 Hz), 4.45 (q, 2H, J ) 7 Hz), 6.82 (d, 1H, J )
9 Hz), 8.45 (d, 1H, J ) 9 Hz), 8.7 (s, 1H). Anal. Calcd for
C
18H22N4O5: C, 57.74; H, 5.92; N, 14.93. Found: C, 57.8; H,
5.9; N, 14.8.
1c in Wa ter (Meth od A). 8-F lu or o-1,2-d ih yd r o-9-(4-
eth oxycar bon yl-3-m eth yl-1-piper azin yl)-6-oxo-6H-pyr r olo-
[3,2,1-ij]qu in olin e-5-ca r boxylic a cid m eth yl ester (4′):
light yellow crystals; mp 233-234 °C (ethanol); 1H NMR
(CDCl3) δ 1.3 (t, 3H, J ) 7 Hz), 1.4 (d, 3H, J ) 7 Hz), 3.1(m,
2H), 3.4 (m, 2H), 3.3 and 4.0 (AB part of an ABX system, 2H),
3.6 (t, 2H, J ) 8 Hz), 3.9 (s, 3H), 4.15 (q, 2H, J ) 7 Hz), 4.4 (X
part, 1H), 4.55 (t, 2H, J ) 8 Hz), 7.6 (d, 1H, J ) 12 Hz), 8.45
(s, 1H); 13C NMR (CDCl3) δ 14.6, 15.2, 22.3, 39.2, 47.1, 50.3,
51.7, 51.8, 55.0, 61.4, 110.2 (d, J C-F ) 25 Hz), 110.7, 120.7 2
(d, J C-F ) 7 Hz), 123.9 2 (d, J C-F ) 7 Hz), 140.6, 140.65 2 (d,
J C-F ) 15 Hz), 142.9, 155.3, 156.3 (d, J C-F ) 250 Hz), 166.4,
173.0. Anal. Calcd for C21H24FN3O5: C, 60.42; H, 5.80; N,
10.07. Found: C, 60.3; H, 5.8; N, 10.0.
1c in Su lfite Bu ffer (Meth od A) (Besid es 4′). 1-Eth yl-
6-flu or o-1,4-d ih yd r o-4-oxo-7-(4-eth oxyca r bon yl-3-m eth -
yl-1-p ip er a zin yl)qu in olin e-3-ca r boxylic a cid m eth yl es-
The light flux was measured by ferrioxalate actinometry.36
La ser F la sh P h otolysis. Nanosecond laser flash photolysis
experiments were performed by means of a Nd:YAG J K lasers
(pulse 20 ns full width at half-maximum (fwhm), 355 nm). The
setup for the nanosecond absorption measurements has been
described previously.16b The laser beam was focused on a 3
mm high and 10 mm wide rectangular area of the cell, and
the first 2 mm was analyzed at a right angle geometry. The
energy used was ca. 2 mJ /pulse. Time “zero” was taken at the
onset of the laser pulse. Spectral resolution was 2 nm. The
sample absorbance was ca. 0.6 at 355 over 1 cm. Oxygen was
removed by vigorously bubbling the solutions with a constant
flux of argon, previously passed through a water trap to
prevent evaporation of the sample. Care was taken to renew
the solution at each laser shot. Temperature was 295 ( 2 K.
1
ter (3c′): colorless crystals; mp 220 °C (methanol); H NMR
(CDCl3) δ 1.38 (t, 3H, J ) 7 Hz), 1.4 (d, 3H, J ) 7 hz), 1.55 (t,
3H, J ) 7 Hz), 2.85 and 3.4 (two m, 2H), 3.0 and 3.5 (two m,
2H), 3.45 and 4.05 (two m, 2H), 3.9 (s, 3H), 4.2 (m, 4H), 4.4
(m, 1H), 6.7 (d, J ) 7 Hz), 8.05 (d, 1H, J ) 14 Hz), 8.4 (s, 1H);
13C NMR (CDCl3) δ 14.2, 14.5, 15.3, 38.6, 46.9, 48.8, 49.6 (d,
J C-F ) 2 Hz), 51.9, 54.8 (d, J C-F ) 2 Hz), 61.4, 103.8 (d, J C-F
) 2 Hz), 110.2, 113.7 (d, J C-F ) 22 Hz), 123.9 (d, J C-F ) 6
Hz), 135.9, 145.1 (d, J C-F ) 11 Hz), 151.5, 153.2 (d, J C-F
)
246 Hz), 155.2, 166.5, 172.9. Anal. Calcd for C21H26FN3O5: C,
60.13; H, 6.25; N, 10.02. Found: C, 59.9; H, 6.3; N, 9.9.
1d in Wa ter (Extr a ction w ith Ch lor ofor m a n d Tr ea t-
m en t w ith CH2N2). 9-F lu or o-2,3-d ih yd r o-3-m eth yl-10-(N′′-
m et h yl-N′,N′′-d ifor m yl-2-a m in oet h yla m in o)-7H-p yr id o-
[1,2,3-d e]-1,4-ben zoxa zin e-6-ca r boxylic a cid m eth yl ester
(5′): colorless crystals; mp >260 °C (nitroethane); 1H NMR
(CDCl3) δ (the NCHO signals are split in two due to hindered
rotation of the amide groups) 1.6 (d, 3H, J ) 7 Hz), 2.7 (s,
3H), 3.0 (s, 3H), 3.4-3.7 (m, 2H), 3.8-4.1 (m, 2H), 3.9 (s, 3H),
4.35-4.55 (m, 3H), 7.8 (d, 1H, J ) 10 Hz), 7.9/7.92 (s, 1H),
8.18/8.2 (d, 1H, J ) 2 Hz), 8.45 (s, 1H); 13C NMR (CDCl3) δ
(almost all signals split in two) 17.7/17.9, 29.0/34.6, 41.8/40.9,
43.0/47.3 (d, J C-F ) 3 Hz), 52.25/52.2, 54.3/57.5, 54.3/57.5, 69.2/
Ack n ow led gm en t. Generous support of this work
by Istituto Superiore della Sanita`, Rome, in the frame-
work of the program “Physicochemical Properties of
Drugs and their safe Use” is gratefully acknowledged.
We thank Drs. Monti and Sortino for communicating
some of their results prior to publication
J O982456T
(36) Hatchard, L. C.; Parker, C. A. Proc. R. Soc. 1956, A253, 318.