820
Chem. Res. Toxicol. 1997, 10, 820-827
Photophysical and Photochemical Characterization of a Photosensitizing Drug: A Combined Steady State Photolysis and Laser Flash Photolysis Study on Carprofen† Francisco Bosca,‡,§ Susana Encinas,‡ Paul F. Heelis,*,§ and Miguel A. Miranda*,‡ Departamento de Quı´mica/Instituto de Tecnologı´a Quı´mica UPV-CSIC, Universidad Polite´ cnica de Valencia, Avenida de los Naranjos s/n, 46022 Valencia, Spain, and Faculty of Science, Health and Medical Studies, North East Wales Institute, Mold Road, Whexham, Clwyd LL11 2AW, United Kingdom Received March 7, 1997X
Carprofen (1a) is a photosensitizing nonsteroidal anti-inflammatory drug. It undergoes photodehalogenation from its triplet excited state. The resulting aryl radical (II) is able to abstract hydrogen atoms from model lipids, mediating their peroxidation by a type I mechanism. This aryl radical intermediate appears to be responsible for the observed photobiological effects of carprofen. The active involvement of the triplet state has been confirmed by direct detection of this species in laser flash photolysis and by quenching experiments with cyclohexadiene and naphthalene. Carprofen also photosensitizes singlet oxygen production with a quantum yield of 0.32. A minor reaction pathway is photodecarboxylation, which occurs from the excited singlet state and leads to an acetyl derivative (1b). In the case of the dehalogenated photoproduct (2a), photodecarboxylation to the ethyl (2d) and acetyl (2b) derivatives, together with singlet oxygen production (quantum yield ) 0.18), is also possible. However, the biological activity of 2a in the linoleic acid photoperoxidation and photohemolysis tests is markedly lower than that of 1a, which constitutes further evidence in favor of the important role of photodehalogenation in the adverse photobiological effects of carprofen.
Introduction Several nonsteroidal anti-inflammatory 2-arylpropionic acids have shown an unusual potential to induce photosensitivity disorders (1-4). In this context, a number of case reports have appeared on photosensitization by carprofen (1a) (5-8), which have been reproduced by means of photopatch tests studies (9). Thus, a multicenter trial performed with more than 1000 patients showed that (out of 32) carprofen was the third most potent test chemical according to the incidence of phototoxic as well as photoallergic reactions (9). Experimental results obtained from in vivo studies using the mouse tail technique also reflected the significant phototoxic properties of this drug (10). In vitro assays have further confirmed the photobiological properties of 1a; for instance positive responses were observed in the photohemolysis assay and the Candida albicans phototoxicity test (11). Likewise, carprofen has been shown to undergo photobinding to cell constituents, which is the primary event involved in the development of photoallergy (12). A detailed knowledge of the photophysical and photochemical properties of a photosensitizing drug is obviously essential to the understanding of its mechanism of action. In this connection, photodegradation of carprofen has been the subject of a previous study, in which the dechlorinated 2-(2-carbazolyl)propionic acid (2a) was detected as the major photoproduct (13). However, the † Dedicated to Professor Waldemar Adam on the occasion of his 60th birthday. ‡ Universidad Polite ´ cnica de Valencia. § North East Wales Institute. X Abstract published in Advance ACS Abstracts, June 1, 1997.
S0893-228x(97)00037-4 CCC: $14.00
Chart 1. Chemical Structures of Carprofen (1a) and Its Stable Photoproducts (1b,c and 2a-d)
nature of the excited state involved, as well as key photophysical data such as intersystem crossing quantum yields or singlet oxygen formation, has not been determined. Free radicals arising from the dehalogenation and decarboxylation processes, together with singlet oxygen and superoxide radical anion, were speculated to be responsible for the carprofen-induced photohemolysis. However, the role played by these different pathways and their relative contributions to the overall photochemical damage remain to be established. Against this background, we have investigated the photophysics and photochemistry of carprofen in order to gain further insight into the molecular basis of the observed photobiological effects. Our studies have included fluorescence and phosphorescence spectroscopy, laser flash photolysis, singlet oxygen detection, photoproduct analysis, and quenching experiments. As a result, we have unambiguously established that the key process in photosensitization by carprofen is © 1997 American Chemical Society
Phototoxicity of Carprofen
cleavage of the carbon-halogen bond, which occurs in the excited triplet state and leads to aryl radicals and chlorine atoms. These highly reactive species appear to be responsible for the extensive free radical-mediated biological damage.
Chem. Res. Toxicol., Vol. 10, No. 7, 1997 821 (φf ) 0.42 in ethanol) (16). Carprofen phosphorescence emission spectra were measured in ethanolic solution at 77 K (λexcitation ) 330 nm). Triplet State Measurements. Extinction coefficients of the triplet states in ethanol were estimated by monitoring the energy transfer reaction between these compounds and ground state NP
Experimental Procedures 3
Chemicals. Carprofen, naphthalene (NP),1 and sodium dodecyl sulfate (SDS) were provided by Sigma (St. Louis, MO). Carbazole, duroquinone (DQ), perinaphthenone, and 1,3-cyclohexadiene (CHD) were purchased from Aldrich (Steinheim, Germany). Reduced glutathione (GSH) and sulfuryl chloride were from Merck (Darmstadt, Germany). Acetone, acetonitrile, absolute ethanol, and methanol (HPLC grade) were from SDS (Peypin, France). All other chemicals were of reagent grade. Tiaprofenic acid (TP) (Surgamic) was from Rusell Iberica (Madrid). Phosphate-buffered saline (PBS) consisted of 0.050 M phosphate buffer and 0.136 M NaCl solution (pH 7.2). 3-Chlorocarbazole (14) was synthesized from carbazole by treatment with sulfuryl chloride in dichloromethane at room temperature. Purification of the product was achieved by recrystallization in cyclohexane. The final purity was higher than 99.9% as assessed by gas chromatography. Instrumentation. Electron impact mass spectra were obtained on a Varian-Saturn II instrument. High-performance liquid chromatography analysis was performed on a Hitachi apparatus equipped with a L-6250 intelligent pump and a L-400 fixed wavelength ultraviolet detector. Separations of the photoproducts were achieved on a Tracer high-performance liquid chromatography Spherisorb column (ODS-2, 10 µm packing) using methanol-water-acetone-acetic acid (56:30:12:2) as mobile phase. Ultraviolet spectra were registered by a Shimadzu, ultraviolet-visible scanning spectrophotometer (2101PC) with a slit width of 5 nm. Fluorescence and phosphorescence measurements were made on a Perkin Elmer MPF-43A spectrofluorimeter. The laser flash photolysis system was as previously described (15) and was based on a J. K. Lasers System 2000 Neodymium/ YAG laser emitting pulses at 355 nm with energies of up to 100 mJ and a pulse duration of 20 ns. The analysis system was a pulsed Xe lamp producing a flat output for a duration of 600 µs. The luminescence (1270 nm) from singlet oxygen was detected by a Judson J16-85P-RO5M germanium photodiode (5 mm2) closely coupled to the laser photolysis cell in a right-angle geometry. A 5 mm thick (5 cm diameter) piece of Ar-coated silicon metal was placed between the diode and cell to act as a narrow band filter for the 1270 nm luminescence. The photodiode output current was amplified with a Judson Model 100 preamplifier. The output from the amplifier was fed into a Philips digital storage oscilloscope (PM 3311) via a colinear 150 MHz, 20 dB amplifier. Fluorescence and Phosphorescence Measurements. Photolysis of samples was reduced by using slit widths corresponding to wavelengths of less than 2.5 nm for excitation and emission. Corrected fluorescence emission spectra were taken using dilute ethanol and acetonitrile solutions of carprofen (1a), its photoproduct (2a), and the model compounds 3-chlorocarbazole (3) and carbazole (4) (absorbance ) 0.4) at room temperature. The samples were excited at 300 nm, and emission measurements were performed in the region 305-500 nm. Fluorescence quantum yields (φf) were determined using integrated peak areas by comparison with carbazole as standard 1 Abbreviations: carprofen triplet state, 31a*; 1,3-cyclohexadiene, CHD; duroquinone, DQ; duroquinone triplet state, 3DQ*; extinction coefficient, ; fluorescence quantum yield, φf; intersystem crossing quantum yield, φisc; monodeuterated methanol, CH3OD; naphthalene, NP; naphthalene triplet state, 3NP*; optical density, OD; phosphatebuffered saline, PBS; red blood cells, RBC; reduced glutathione, GSH; singlet lifetime, τs; singlet oxygen quantum yield, φ∆; sodium dodecyl sulfate, SDS; tiaprofenic acid, TP; triplet lifetime, τΤ.
1a* + NP f 1a + 3NP*
In these experiments, deaerated ethanol solutions of 1a (8 × 10-4 M) containing NP (2 × 10-4-2 × 10-3 M) were excited using a 355 nm laser pulse. Under these conditions, more than 99% of the absorbed light is absorbed by the carprofen. The extinction coefficients () of the triplet states were calculated using the equation (17):
k2/(k2 - k1) × OD(3NP*(415 nm)) × (31a*(490 nm)) ) OD(31a*(490 nm)) × (3NP*(415 nm)) where the OD values refer to the optical density of the carprofen triplet state (490 nm) at the beginning of the reaction and the NP triplet state (415 nm) at the end of the reaction, k1 is the 31a* decay rate constant without NP, and k is the different 2 31a* decay rate constant obtained with varying concentrations of NP. The extinction coefficient of 3NP in ethanol was taken to be 40 000 dm-3 mol-1 cm-1 (18). In this way a value for of 14 200 ( 260 dm-3 mol-1 cm-1 was obtained for carprofen triplet state. By the same procedure the extinction coefficients of 2a and 3 using 3 × 10-3 M ethanol solutions were obtained. These parameters were used to calculate the intersystem crossing quantum yields (φisc) for 1a, 2a, and 3 in deaerated ethanol solutions. Excitation of duroquinone (DQ) and carbazole derivatives was carried out separately. The four solutions had identical optical density (OD) at the excitation wavelength (0.275 at 355 nm). Under these conditions φisc was obtained by application of the following formula (17):
φisc(1a) ) φisc(DQ) × OD(31a*(490 nm)) × (3DQ*(490 nm))/OD(3DQ*(490 nm)) × (31a*(490 nm)) where the OD values refer to the optical density of the carprofen triplet state and duroquinone triplet state at 490 nm. The duroquinone extinction coefficient (18) and triplet state quantum yield (19) in ethanol were taken to be 5580 dm-3 mol-1 cm-1 and 0.94, respectively. Singlet Oxygen Measurement. The singlet oxygen quantum yield (φ∆) of carprofen (1a), 2a, 3, and 4 was determined in acetonitrile solutions using the same absorbance value (0.50) at 266 nm for each compound. Laser excitation at various lowpulse energies yielded for each molecule the slope of the laser pulse energy versus the emission intensity of the singlet oxygen produced. A singlet oxygen quantum yield (φ∆) of 0.95 for perinaphthenone in acetonitrile was used (20). Photolysis of 1a. Solutions of 1a in different solvents (4 × 10-3 M, 10 mL/tube) were irradiated with Pyrex-filtered light from an OSRAM-HLQ 125 W medium pressure Hg lamp located inside an immersion well photoreactor (Applied Photophysics Model 3230) for 15 min. The photon flux incident on the tubes (measured by potassium ferrioxalate actinometry) (21) was 5 × 1016 photons s-1. Hence samples received, on average, 2.3 J cm-2. The irradiation of 1a was performed in methanol, methanol/ KOH (10-2 M), and PBS under aerobic and anaerobic atmospheres, as well as in aerated buffered solutions of red blood cells (RBC; 3.3 × 106 cells mL-1). Parallel experiments were done with the photoproduct 2a. These photoreactions were monitored by reverse phase high-performance liquid chromatography using the above-mentioned conditions. A fixed wavelength of 330 nm was used for detection. Irradiation of 4 × 10-3 M methanol and monodeuterated methanol (CH3OD) solutions of 2a and 3-chlorocarbazole (3) under argon atmosphere were monitored by gas chromatography/mass spectrometry.
822 Chem. Res. Toxicol., Vol. 10, No. 7, 1997
Bosca et al.
Isolation and Identification of the Photoproducts. Isolation of the photoproducts was carried out by the same reverse phase high-performance liquid chromatography using the above-mentioned conditions. Compounds 1b and 2a,b (Chart 1) have been previously described and were characterized by their spectral properties (13, 22, 23). In some experiments the free carboxylic acids (1a and 2a) were derivatized to the corresponding methyl esters (1c and 2c) by treatment with methanol/sulfuric acid (5%) under reflux for 2 h. The structures were assigned by mass spectrometry. Compound 1c showed diagnostically important peaks at m/z 289 (31%) [M+ + 2], 287 (94%) [M+], 230 (11%) [M+ + 2 - COOCH3], 228 (33%) [M+ COOCH3], 193 (100%) [M+ - COOCH3 - Cl], 165 (29%) [M+ CH3CHCOOCH3 - Cl], 59 (84%) [COOCH3+], and 2c showed them at m/z 253 (87%) [M+], 194 (46%) [M+ - COOCH3], 167 (29%) [M+ - CH2dCHCOOCH3], 59 (90%) [COOCH3+]. Quenching of Excited Carprofen: (a) By Cyclohexadiene. Methanolic drug solutions (2 × 10-3 M) were introduced in Pyrex tubes containing different amounts of cyclohexadiene (CHD/drug molar ratio between 1 and 30) and irradiated for 15 min with a 125 W medium pressure Hg lamp inside the immersion well photoreactor described above. Experiments were performed under both aerobic and anaerobic conditions. The controls were solutions of cyclohexadiene without drugs, as well as solutions of the latter without CHD, at the same concentrations. Controls were irradiated under identical conditions, and a set of duplicate samples was not irradiated. The resulting cyclohexadiene dimers were analyzed directly by GC. In order to monitor the drug photoproducts by gas chromatography, the free carboxylic acids were derivatized to the corresponding methyl esters. This was performed by treatment with methanol/sulfuric acid (5%) under reflux for 2 h. (b) By Naphthalene. Methanolic drug solution (4 × 10-3 M) containing naphthalene (10-2 M) was irradiated for 15 min with a 125 W medium pressure Hg lamp inside an immersion well photoreactor, under aerobic conditions. The photoreaction was monitored by reverse phase high-performance liquid chromatography using the above-mentioned conditions. A fixed wavelength of 330 nm was used for detection. Photoperoxidation of Linoleic Acid by the Presence of Carprofen and Its Major Photoproduct in Micellar Aqueous Solutions. Solutions of linoleic acid (10-3 M) and sodium dodecyl sulfate (SDS; 10-2 M) in PBS, containing either 1a or 2a (10-5 M), were irradiated through Pyrex with a 400 W mercury lamp for 5, 10, and 20 min, keeping the temperature at 37 °C by means of a thermostated bath. Tiaprofenic acid (TP) was used as a reference. The reaction was monitored by ultraviolet-visible spectrophotometry, following the appearance and subsequent increase of a new absorption maximum at λ ) 233 nm, due to the conjugated dienic hydroperoxides derived from linoleic acid (4, 24). The controls were solutions of linoleic acid without carprofen, tiaprofenic acid, or 2a, as well as solutions of the latter (compounds 1a, 2a, and TP) without linoleic acid, at the same concentrations. Controls were irradiated under identical conditions, and a set of duplicate samples was not irradiated. Photohemolysis Tests. Buffered drug solutions (10-4 M) were transferred to plastic cuvettes and RBC added, so that the resultant suspension had an optical density of approximately 0.5 at 650 nm (3.3 × 106 cells mL-1). Duplicate samples of the resulting suspensions, as well as samples containing the drug solutions only, were exposed to ultraviolet-A radiation. For this purpose a 400 W medium pressure mercury lamp was used as the light source. The distance between the lamp and the samples was 5 cm. The photon flux incident on the cuvettes (measured as before) was 2 × 1016 photons s-1. Hence samples received, on average, 12.9 J cm-2 in 1 h. A control experiment was performed in the dark. The samples were measured at 650 nm every 10 min during irradiation with an ultraviolet-visible spectrophotometer. Further details of the procedure can be found elsewhere (25).
Figure 1. Absorption (lightface line) and uncorrected fluorescence (boldface line) spectra of carprofen (1a) in ethanol. Table 1. Fluorescence Quantum Yield of Carbazole Compounds φf acetonitrile
ethanol
compd
N2
O2
N2
O2
1a 2a 3 4
0.068 0.435 0.060 0.422
0.064 0.253 0.056 0.259
0.067 0.431 0.058 0.420
0.062 0.265 0.055 0.264
Results Fluorescence and Phosphorescence. Corrected fluorescence spectra of carprofen (1a) in acetonitrile and ethanol solutions showed two bands around 367 and 352 nm (Figure 1). Similar spectra were obtained for 3-chlorocarbazole (3). The major photoproduct 2a and carbazole (4) displayed two bands around 362 and 347 nm. Singlet energies obtained from the fluorescence band maxima of 1a and 3 were ca. 81 kcal mol-1. For 2a and 3 the corresponding values were ca. 82 kcal mol-1. Potassium salts of 1a and 2a showed the essentially identical spectra to their acid forms. The fluorescence quantum yields (φf) were determined using carbazole as standard (φf ) 0.42 in ethanol), in acetonitrile and ethanol in the presence or absence of oxygen, and are shown in Table 1. A sharp decrease of φf was associated with chlorocarbazole molecules (1a and 3) and with the presence of oxygen, especially in compounds 2a and 4. Oxygen quenching of carbazole φf has been previously reported (26). The φf values of the potassium salts of compounds 1a and 2a yielded identical results. Fluorescence quenching by cyclohexadiene was also observed for 2a and 4 but not for 1a or 3 (data not shown). The uncorrected phosphorescence spectrum of carprofen (1a) in ethanol at 77 K showed two bands around 445 and 428 nm (not shown). Thus, the triplet energy level was ca. 69 kcal mol-1. Under the same conditions, 2a showed two bands at 448 and 430 nm, and hence the triplet energy level was ca. 68 kcal mol-1. Nanosecond Laser Flash Photolysis. Laser flash photolysis of deaerated 4 × 10-4 M methanolic solutions of carprofen (1a, and its potassium salt), its photoproduct (2a, and its potassium salt), and the model compound 3-chlorocarbazole (3) was carried out in order to detect the formation of triplet states and other intermediate
Phototoxicity of Carprofen
Chem. Res. Toxicol., Vol. 10, No. 7, 1997 823
Figure 2. Transient absortion spectra of a nitrogen-saturated methanol solution of carprofen (1a; 4 × 10-4 M) measured 300 ns (solid line) and 4 µs (dotted line) after the laser pulse (355 nm).
Figure 3. Transient absortion spectra of a nitrogen-saturated PBS solution of 1a (2 × 10-4 M) measured after 150 ns (A) and 2000 ns (B). Plot C is the difference (A - B) to observe better 1a triplet and e-(aq). Table 3. Irradiation of Carprofen in Methanolic Mediuma products (%)b
Table 2. Triplet Photophysical Data of Carprofen and Carbazole Derivatives
a
compd
τΤ (µs)a
φisca
φ∆b
1a 2a 3 4
3.3 28 4.1 43 (170c)
0.37 0.35 0.36 0.36c
0.32 0.18 0.32 0.17
In ethanol. b In acetonitrile. c In benzene (29).
species. Compounds 1a, 2a (and their potassium salts), and 3 yielded the same two transient species. The first transient had an absorption band at λmax ca. 430 nm and was assigned to the triplet state (27). This species was quenched by oxygen, with rate constants of ca. 5 × 109 dm-3 mol-1 s-1 for 1a, 2a, and 3. The second transient absorbing at λmax ca. 640 nm was assigned to the corresponding carbazolyl radical (R2N•), which is in excellent agreement with previous work on related compounds (27, 28). A decrease in its optical density was produced by the presence of naphthalene (2 × 10-3 M) or oxygen (2.2 × 10-3 M) in the solutions. Figure 2 shows the triplet and the carbazolyl radical obtained from 1a. Identical transient absorption spectra were obtained from carprofen (1a) in deaerated ethanolic solutions (data not shown), but in this solvent the triplet state lifetimes (τΤ) were somewhat higher. Thus, while in deaerated ethanol the τΤ of 1a was 3.3 µs, in methanol it was 1.3 µs. Table 2 shows the triplet lifetimes of carprofen and carbazole derivatives in ethanolic solutions. These data were known for the parent carbazole in benzene (29). As a general rule, the introduction of chlorine in the carbazole ring (compounds 1a and 3) produced a marked decrease of the triplet lifetimes. Flash photolysis of deaerated 2 × 10-4 M PBS solutions of carprofen showed the presence of the hydrated electron e-(aq), in addition to the 1a triplet (τΤ ca. 0.2 µs) and the carbazolyl radical (Figure 3). As indicated above in the materials and methods section, deaerated ethanol solutions of 1a (8 × 10-4 M) and 2a and 3 (3 × 10-3 M) containing NP (2 × 10-4-2 × 10-3 M) were excited by a 355 nm laser pulse. The triplet extinction coefficients () at 490 nm were found to be 14 200 ( 260 dm-3 mol-1 cm-1 for carprofen, 22300 ( 1050 dm-3 mol-1 cm-1 for 2a, and 6090 ( 350 dm-3 mol-1 cm-1 for 3. Using these parameters the intersystem crossing quantum yields (φisc) of 1a, 2a, and 3 were
solvent
atmosphere
2a
MeOH MeOH MeOH/KOH MeOH/KOH
anaerobic aerobic anaerobic aerobic
87 40 99 64
2b
2c
1a
1b
1c
13
11 3
18
4
2
13
2
8
Carprofen at 4 × M concentration was irradiated with Pyrex-filtered light of a medium pressure mercury lamp for 15 b min. Determined by reverse phase high-performance liquid chromatography. Photoproducts are listed in order of elution. a
10-3
obtained in deaerated ethanol solutions (Table 2). The values were very similar in all cases. Time-resolved luminescence from singlet oxygen was measured at 1270 nm using an appropriate diode as a detector to obtain the quantum yield of singlet oxygen formation (φ∆) for carprofen (1a), 2a, 3, and 4 using acetonitrile solutions (Table 2). Chlorocarbazole derivatives (compounds 1a and 3) had a φ∆ significantly higher than their dehalogenated analogues (2a, 4). Photochemistry of 1a. When irradiation of 1a was performed in deaerated methanolic solutions, 2a was obtained as the major product together with lower amounts of 2c. The initial drug was completely consumed. However, under aerobic conditions the conversion was less than complete. Compounds 2a,c were again obtained, but in this case minor amounts of 1b,c and 2b were produced. In the same experiment with methanol/KOH, photodegradation was much cleaner, leading to 2a as the only photoproduct under an inert atmosphere. Under aerated conditions small amounts of 1b and 2b were also produced. As can be seen in Table 3, the photoproduct distribution depended upon the irradiation conditions. Compound 2a was always predominant. The number of photoproducts increased in the presence of oxygen. Compounds 1c and 2c were not detected during irradiation in methanol/KOH. When irradiation of 1a was carried out in aerated buffered solutions of RBC, a photoproduct distribution similar to that characteristic of the methanol/KOH aerobic photolysis was obtained (data not shown). Nevertheless, in PBS solutions without RBC, complex photomixtures were obtained in either aerobic or anaerobic conditions. One of these components was compound 2a, but as a general rule, polymerization was the major process (data not shown).
824 Chem. Res. Toxicol., Vol. 10, No. 7, 1997
Bosca et al.
Table 4. Irradiation of the Major Photoproduct 2a in Different Mediaa products (%)b solvent
atmosphere
2a
MeOH MeOH MeOH/KOH MeOH/KOH PBS PBS
anaerobic aerobic anaerobic aerobic anaerobic aerobic
87 80 89 85 92 90
2b
2c
15
