This review focuses on damage photosensitized by the fluoroquinolone (FQ) antibacterial drugs. Different models are employed to study biosubstrate photodamage mediated by FQs (organisms, cells, isolated biomolecules and super molecules). Being that the effect of environment (polarity of the medium, ions, pH, binding with bio-molecules, etc.) is crucial in FQ photochemistry, photobiological reactions can be consequently dramatically influenced. Thus, the photosensitization processes induced by FQs are here discussed taking into account that such extensive and cross-targeted pathological implications request an excursus covering photosensitization in systems of increasing molecular complexity. In vivo and in vitro evidences for photoallergy, phototoxicity, photomutagenesis and photocarcinogenesis mediated by FQs are discussed.
In the last three decades literature reports have shown how a great variety of drugs are reactive under exposure to environmental light, where UVA radiation represents about 90% of the solar UV radiation reaching earth’s surface. This process represents one of the factors responsible for the generation of adverse side effects (1–3). UVA can generate photosensitization reactions both by penetration into dermis and by reaching dermal blood flux: in each case, drugs can trigger such reactions.
Among the phototoxic effects induced by chemicals and pharmaceutical compounds, those involving fluoroquinolone (FQ) family, despite its efficiency in treatment of a broad spectrum of infections, are photosensitization reactions of a variable degree of severity, due to exposure to UVA radiation (4,5).
Fluoroquinolones are well-established antibacterial drugs, whose pharmacological activity is associated with inhibition of the bacterial topoisomerase. They contain a fluorine atom attached at position C-6 of the bicyclic skeleton; in some cases, a second halogen substituent is present at position C-8. Chart 1 shows the main structures of FQs relevant for the purpose of this review.
Literature reports many examples of preclinical data for FQs mediated skin phototoxicity. These reactions include skin lesions with different degrees of severity, from cutaneous reactions to skin cancer (carcinoma and melanoma) (4,5). Lomefloxacin (LOM), Ciprofloxacin (CIP), Ofloxacin (OFL) (4,5) and Fleroxacin (FLE) (4–6) are very powerful photochemical carcinogens. The types of tumors observed in FQs treated animals include benign neoplasms such as papillomas, keratoacanthomas, solar keratoses and malign neoplasms like squamous cells carcinomas. These latter can be originated from a malignant change or ulcerations in surface epithelium and represent definite indicators of cutaneous carcinogenicity. In addition, epidermal hyperplasia and dermal fibrosis were observed. For example, animals exposed to LOM and UV exhibited a number of tumors with a specific type of neoplastic progression (4).
Several reports deal with phototoxicity evidences in humans for different clinical consequences such as erythema with the use of Grepafloxacin (GRE) (7,8), CIP (8), LOM (7,9), OFL, BAYy3118 (BAY) (7), FLE (10), Sitafloxacin (SIT), Enoxacin (ENO), Norfloxacin (NOR), Sparfloxacin (SPA) (11) and Trovafloxacin (TRO) (12). A review reports clinical data about the incidence of rashes and other cutaneous reactions in FQ therapy (13); the rates of rash vary from <1%–3% or higher if longer courses of therapy are given. An analysis of spontaneous reports and FQ consumption data from three Italian regions showed that phototoxic reactions were more frequent with LOM (14).
All these harmful consequences can be induced mainly through generation of reactive oxygen species (ROS) (15–17) and/or energy transfer processes (18). This is the reason of the very large number of photophysical, photochemical and photobiological studies on this class of drugs, including some significant reviews covering single parts of this large field of investigation (15,19–22). The aim of this review is thus to collect all significant literature in this topic of the photochemical and photobiological sciences, i.e. the photosensitization reactions of FQs involving various biotargets and their biological consequences including allergy, up to mutagenicity and genotoxicity. These extensive and cross-targeted pathological implications request thus an excursus covering photoinduced damages in systems of increasing molecular complexity preceded by an essential introduction on FQ photochemistry.
We mainly selected publications from the past decade, although we did not exclude commonly referenced and highly regarded older publications. We selected, where existing, also high quality systematic reviews, to provide readers with more details and references than this review can accommodate.
FQ Photochemistry and Photophysics, an Overview
The photoreactivity of FQs is markedly modulated by the nature and position of the substituents attached to the quinolone structure (see Chart 1). This is based on photophysical and photochemical studies on a large number of FQs, such as LOM (23–27), CIP (24,28), NOR (18,23–30), ENO (18,23,25–29) OFL (18,23,25,27,31), Pefloxacin (PEF) (18) and Rufloxacin (RUF) (18,32,33). In addition, the pH of the reaction medium plays an important role (29,32,34,35). FQs undergo a variety of photochemical processes such as dehalogenation (24–26,29,34,36), oxidation of an amino substituent at C-7 (29), decarboxylation (32), production of superoxide anion O2−• and generation of singlet oxygen 1O2 (37,38). The production of 1O2 does not correlate with the order reported for their phototoxicity. The more phototoxic drugs tend instead to produce O2−• at a faster rate, although there is not a one-to-one correspondence between the relative rates of O2−• production and the phototoxicity ranking of the FQs (38). Besides, FQs with a triplet energy value above a threshold value (in the range 265–269 kJ/mol) always induce cyclobutane formation of thymine dimers (T<>T) (18,27,39). These findings can explain why phototoxicity does not necessarily correlate with the photodegradation quantum yield (40,41). The internal charge-transfer character of the excited state determines the efficiency of the most common reaction, which is the heterolytic C-F bond fragmentation. This is in agreement with the electronegativity of the substituent in position 8 and by the stabilization of the resulting aryl cation (larger for the 8-cation than for the 6-cation) (23).
The existence of the highly reactive species, identified as a singlet aryl cation by photodehalogenation of LOM and BAY can justify the photogenotoxic properties associated with these antibacterial drugs, likely due to direct reaction of these cations with DNA (42). Modeling studies confirm the hypothesis that the mutagenic activity of the DNA-intercalated drug involves attack of the photogenerated cation to the heterocyclic bases (43). In the cation formed by photoinduced C-F bond cleavage in FLE, intramolecular reaction with the N-Et chain is prevented by the electron-withdrawing effect of fluorine and intermolecular attack by nucleophiles is facilitated (44). Photodehalogenation seems to be the main process involved in the phototoxic effect of FQs; however, this photodegradation process is not significant in CIP, NOR (38), Moxifloxacin (MOX) (45,46), OFL (38) and RUF (32,47). For these two latter FQs, irradiation under aerobic conditions gives rise to N-demethylation of the piperazinyl ring via photoionization from singlet state and formation of hydrated electrons, as well as 1O2 generation from drug triplet excited state. In the absence of oxygen, RUF undergoes decarboxylation and opening of the piperazinyl ring (32). The veterinary antibacterial Marbofloxacin (MAR) undergoes an unusual homolytic photocleavage of the N-N bond, yielding a phenoxy radical that undergoes dimerization, reduction, or disproportionation (48).
The effect of environment (biomolecules and cells, polarity of the medium, ions, pH, etc.) is crucial in photobiological reactions, and, consequently photochemistry could be dramatically influenced.
Photochemistry in biomimicking systems
Micelles and liposomes. A drastic increase of photostability of the highly photoreactive and photocarcinogenic LOM can be achieved through a selective and efficient sequestering of its cationic form by a suitable micellar system at neutral pH (49). Besides, the effect of trapping of NOR in sodium dodecyl sulfate micelles results in a net increase of photostability and in the suppression of the radical photodecomposition mechanism occurring in phosphate buffered solution. This favors an ionic pathway (50). In the liposomal encapsulation of CIP, OFL and LOM, the presence and type of lipids influence the ways of photodegradation process. For example, the presence of unsaturated fatty acid chains in the liposomal bilayer alters the LOM photodegradation pathways, making the CO2 loss more common and increasing the frequency of dehydrogenation followed defluorination (51). The photosensitizing activity of ENO in aerated conditions provokes lipid bilayer damage. TBARS formation indicates the degree of lipid peroxidation and efflux of trapped markers the extent of membrane damage. The inhibition of liposome damage by both EDTA and deferoxamine suggests that OH• radicals can be formed via a Fenton reaction involving traces of iron ions and H2O2 (52).
Amino acids and proteins. Moxifloxacin cation radical is able to photooxidize amino acids, as evidenced by pulse radiolysis experiments in the presence of tryptophan (Trp) and tyrosine (Tyr) (46). Sarafloxacin (SAR) triplet was also found to oxidize Trp to its radical with concomitant formation of the drug anion radical (53). Commonly, radical photosensitization caused by the formation of ROS, such as OH•, O2−• and H2O2 is defined Type I, whereas a Type-II photodamage occurs via1O2. The radical pathway can be also evaluated on the basis of Trp derivatives photogeneration. Photoproducts such as tryptamine are Type-I diagnostic. Other compounds, such as 3-OH-kynurenine, are Type-II diagnostic products. 4-OH-chinoline is instead obtained by a cooperative reaction pathway. Some of these photoproducts are also Trp metabolites (54 and literature there cited). Singlet oxygen reacts with the amino acids Trp, Tyr, histidine (His), methionine and cysteine. The substitution with the electron-donating methyl group in indole, phenol and imidazole rings increases the quenching rate constant (55). The oxidative destruction of the aromatic ring correlates with the substitution effects. The photooxidation takes place by formation of an exciplex between 1O2 and the quencher. This formation is favored by the electron donor ability of the amino acid residue (55). In this context CIP and, in a lesser extent, OFL, affect Tyr integrity, whereas LOM and NOR are not photoactive; 1O2 does not play a significant role (56). Contrarily to the reported inefficacy of NOR towards Tyr, laser flash photolysis and pulse radiolysis studies showed that triplet state of this FQ, together with those of ENO and CIP, can oxidize Tyr as well as Trp (28). The FQ radical anions and oxidized radicals of Trp and Tyr were directly observed. Under aerobic conditions, the photooxidation of Trp and Tyr involves both Type-I and -II mechanisms. LOM, CIP and OFL and not NOR, are weak photosensitizers towards His (53). On increasing molecular complexity, the presence of bovine serum albumin (BSA) does not alter the NOR transient species formation, but they decay faster, especially the triplet state (56). Photoexcited MOX produces cation radicals that react with BSA, leading to the possible formation of haptens in photoallergic reactions (46).
Bases and nucleic acids. Guanine is the most susceptible DNA target with regard to oxidation induced by ROS (57,58). The cation radical of MOX reacts with thymine and guanosine (46). The triplet aryl cation photochemically generated from LOM and FLE, bearing a fluorine atom at position 8, attacks the most nucleophilic nucleotide, and forms covalent adducts (59). Differently, guanosine monophosphate quenches the triplet state of NOR (that lacks the fluorine at position 8), similarly to what occurs in the presence of the inorganic phosphate anion (29). Commonly, different dGuo products come from FQs photoexcitation. These derivatives are diagnostic for the photosensitization mechanism, that is an attack of dGuo via a Type-I (oxazolone derivative) (60) or via a Type-II (dSp, Spiroiminodihydantoin) pathway (61). A Type-II photosensitizing mechanism was assigned in the oxidation of free dGuo in the presence of NOR (62), OFL (62,63) and RUF (63,64). In RUF photomediated process, Cu2+ inhibits formation of Type-II products (65). The reduction of Type-II products, dSp, is higher than the copper(II) concentration used, pointing to a catalytic pathway. The significant decrease of dSp counteracts the almost negligible increase of Type-I derivatives, caused by an electron extraction from dGuo from the radical cation of RUF (65). A Type-I photosensitizing mechanism is operative in ENO and LOM sensitized free base (62). Using oligonucleotides instead free bases, the detection of alkali labile sites (ALS) at 5′-G of a GG step, or selectively at consecutive G sites, is due to Type-I pathway, whereas uniform cleavage at all the guanines is caused by Type-II mechanism. Indeed, as shown by PAGE electrophoresis sequencing technique, DNA photosensitization by LOM, FLE and CIP occurs on 32P-labeled oligonucleotides from human DNA c-Ha-ras-1 proto-oncogene and p53 tumor suppressor gene (66,67). The distribution obtained for LOM, FLE and CIP does not show any preference for neighboring guanines. Based on the obtained fragment pattern and on the effect of NaN3 addition or the use of deuterium oxide as solvent, it has been proposed that 1O2 is the primary reactive species in DNA-photosensitized oxidation by these three FQs (67). In addition, some FQ-DNA complexes reveal a remarkably different photoreactivity. Complexation of Levofloxacin (LEV), levorotary optical isomer of OFL, and MOX with native DNA leads to formation of new transient species. These were tentatively assigned to the triplet state of the drugs complexed with the biomolecule (45). The LOM-calf thymus DNA complex is characterized by a photodefluorination yield similar to that observed for the free drug, whereas the ENO-DNA complex is not photoreactive (68).
FQ photodegradation and localization in cells. In this picture, for a correct evaluation of photosensitization processes, it also seems important to monitor the FQ intracellular photodegradation. Consistent with their photoinstability in solutions, NOR, OFL, LOM, CIP and BAY are readily photobleached by UVA in HS68 fibroblasts (69). The photodegradation of intracellular RUF and its photoproduct formation occurs in human fibroblasts (70) and in yeasts (71), approximately with the same quantum yield calculated in the absence of cells. Fluorescence image analysis shows that lysosomes of live human HaCaT keratinocytes are a preferential site of LOM, Clinafloxacin (CLI), MOX and BAY localization and phototransformations. As the lysosomal environment is relatively acidic, low pH may affect the drug reactivity and fluorescence (72).
pH, polarity of the medium, ions, metals and environment protection
pH and polarity. Indeed, for BAY, CLI, LOM and MOX, fluorescence provides a convenient way to monitor their presence within cells and to follow their photoreactivity during irradiation (72). Authors propose that the temporal profiles of fluorescence red shifts and its intensity decrease are consistent with dehalogenation of CLI, LOM and BAY. Their temporal fluorescence in vitro further clarifies the difference in phototoxicity noted for these drugs. NOR and the naphthyridine analog ENO give the corresponding 6-hydroxy derivatives by irradiation in water at pH 7.2 and, with lower efficiency, at pH 4.5 and 10 (35). The photodegradation quantum yield of ENO depends on pH, being maximum in neutral conditions (34). The 6,8-difluoro derivative LOM is selectively defluorinated from position 8 over the entire pH range considered (1–10) (35). This drug is photodegraded in the presence of aliphatic amines via reductive elimination of the fluorine in the 8-position and alteration of the piperazine side-chain. The C8 site has a carbene rather than a localized cation character (73). Again, influence of solvent polarity and proticity on the photochemical properties of NOR suggests that the phototoxicity of this drug and other FQs may strongly depend on its localization in hydrophilic or hydrophobic cell/tissue regions (74).
Ions. In the presence of sulfite or phosphate buffer a defluorination mechanism, induced by electron transfer from the inorganic anions to NOR, ENO and LOM triplet states (and not OFL), was proposed (25,26,29). For ENO and NOR, a transient band previously attributed to the defluorinated cation (34), was reassigned to the radical anion (26,29). However, similarly to the case of SAR, it has been shown that the process consists of an energy transfer reaction yielding a second triplet formed from the first triplet by reacting with strong bases (53). Moreover, due to a decrease of the triplet energy when sulfur (RUF) instead of oxygen (OFL) is attached to position 8 of the FQ ring system, phosphate anions are able to quench OFL triplet but not RUF triplet (63). Defluorination is also observed in the photoreaction of NOR in aqueous solution. This leads to production of the corresponding 6-hydroxy derivatives via a direct attack by hydroxide anions to the excited triplet state (75). In the presence of chloride ion, Orbifloxacin (ORB) converts into a photoproduct substituted by chlorine at the 8-position (8-Cl ORB) (76).
Metals. The neutral and anionic forms of NOR complex calcium and magnesium (the antibacterial exists mainly as a complex in the blood plasma); 1O2 generation is decreased by complexation. At pH ≥ 7.4, complexation to Ca2+ and Mg2+ increases the rate of NOR photodegradation (77). Cu2+ in the micromolar range modulates RUF photodegradation rate. Indeed, in the presence of metal ions, the drug photobleaching is reduced under aerobic conditions, and no change in the nature of the photoproducts occurs. In anaerobic media, RUF photodegradation rate increases with increasing Cu2+ concentration and the photoproducts distribution changes. A static quenching with formation of a complex between the RUF in the ground state and copper(II) is proposed on the basis of fluorescence reduction upon complexation (65). pH influences both the metal complexation and the formation of intramolecular hydrogen bond between the cheto and the carboxy moieties of the FQs. This competition between these processes influences fluorescence yield, singlet oxygen generation and photodegradation rate (65,77).
Environment protection. Recently, the ubiquity of FQs in surface waters including hospital effluents has also stimulated studies on their fate in the aqueous euphotic zone and methods of detoxification (78,79). In these systems, CIP, Danofloxacin (DAN), LEV, SAR, Difloxacin (DIF), Enrofloxacin (ENR), Gatifloxacin (GAT) and Balofloxacin (BAL) undergo both direct photolysis and self-sensitized photo-oxidation via hydroxyl radicals (OH•) and 1O2.
FQ Impact on the Main Biomolecules and Their Biological Consequences
As reported above, the photosensitizing properties of some FQs are accredited to the photodehalogenation reactions (24–26,29,34,36,38). For other members of this class photosensitization was attributed to the formation of ROS, such as OH•, O2−• and H2O2 (Type I), 1O2 (Type II), as well as to triplet–triplet energy transfer (18,27,38–40,62). These processes are able to modify cell components, including lipids, proteins and nucleic acids. Consequently, the photosensitization reactions have been studied in several systems with different molecular complexity, from single biomolecules to cells, both in vitro and in vivo. In Scheme 1 the various colored lines constitute the interaction network between each FQ and the various biological models and compartments target of photosensitization according to the references reported in this review. Tables 1 and 2 report a summary of relevant cited photoinduced damages and biological consequences. Concerning the mechanism of FQs action, it is mainly reported the ability to induce photoperoxidation of membrane lipids and to photodamage DNA at various degrees of severity. Recently, a cooperative cell damage mechanism was proposed, based on triggering of DNA toxicity by lipid peroxidation products (80). Some authors also proposed that FQ phototoxicity is the result of a multistep mechanism: first, local photo-oxidative stress takes place and, second, some of the photoproducts exert genotoxic actions (81). Thus, subcellular localization is one of the important parameters that may determine these effects (45,69,72,82,83).
Table 1. Summary of relevant cited photoinduced damages and biological consequences on membrane and cytoplasmatic compartments.
Lipids, membrane damage, cytotoxicity, irritation and inflammation
Lipids. Most of FQs induce lipid peroxidation. CIP induces photoperoxidation of linoleic acid used as simple lipid model system. The linoleic acid photoperoxidation is higher under sunlight followed by UVB and UVA irradiation, and it is dose and concentration dependent under all radiation doses (84). Experiments of linoleic acid peroxidation using various additives as antioxidant/quencher, confirm the involvement of ROS in CIP induced lipid peroxidation via a mixed Type-I and -II mechanism (84). The photosensitization potential of CIP was also evidenced in the peroxidation of squalene by measurement of thiobarbituric acid-reactive substances (TBARS) after UVA exposure. CIP phototoxic potential was compared with that of LOM and FLE. It resulted that LOM is the strongest sensitizer, followed by FLE and CIP, whereas ENO has moderate activity; OFL, LEV and NOR are weak sensitizers (85).
Red blood cells. The phototargeting on lipids was also considered in human blood. LOM, NOR, ENO (86), LEV (45) and OFL (45,86) induce lipid peroxidation on human red blood cells (RBC) isolated membranes (ghost). The photoperoxidation induced by OFL and LEV increases sharply with the irradiation time, whereas MOX is almost inactive (45). O2−• seems to be involved in this process (45). OFL and LEV also cause lysis of albino mouse RBC. The absence of membrane damage in the presence of preirradiated solutions of these two drugs indicates that the photoproducts are not involved in the photohemolytic effect (45). ENO also leads to photohemolysis of human RBC (52). The mechanism of photosensitization responsible for the membrane damage depends on the oxygen concentration and follows a different path with respect to that operative for plasmid-DNA cleavage. Among oxygenated radicals, the OH• seems the species mainly responsible for membrane damage (52). NOR, CIP, LOM and ENO show dose-related increases in photohemolysis, whereas SPA is not active (87). Photohemolysis can be also induced by LEV in an oxygen-independent way (45).
Membrane damage in cells. Membrane photooxidation is responsible for CIP phototoxicity in fibroblast cell lines L-929 and NIH-3T3 of mice: photogenerated ROS might trigger phototoxicity on cell lines as well as peroxidation of linoleic acid. ROS initiate cellular lipid peroxidation, which finally leads to cell death (84). LEV, MOX and OFL also induce in vitro cytotoxicity of Balb/c 3T3 cells (45). The involvement of O2−• and/or radicals is in agreement with the lipid peroxidation. CIP, NOR, LOM and OFL photoinduce membrane damage, determined by release of cytosolic lactate dehydrogenase in lens epithelial cells (HLE B-3). Apoptosis and necrosis, evaluated by flow cytometry and loss of cell viability, follow the damage photosensitized by these drugs (82). OFL induces photosensitization in two other human cell lines, HL60 and K562 (88). This drug is inserted in the cell membranes at concentrations directly related with its concentration used for the experiment and incubation time. This produces an increase in the generalized polarization of a fluorescent membrane probe. Polarization values are similar to those produced by membrane lipid oxidation. The results suggest the cell membrane as a target of the OFL adverse action, with a possible mechanism involving the formation of ROS, which trigger, in turn, the lipid peroxidation chain reaction.
Cytotoxicity. A large number of FQs mediated photocytotoxic effects can be revealed by a combination of cell viability tests, such as the NR (Neutral Red) assay. LOM induced photocytotoxicity was assessed by viability evaluation of the skin model (KeraSkin) in combination with in vitro ocular NR test (89). The KeraSkin™, a commercially available engineered human epidermal equivalent, is used as an in vitro model of the epidermis. This model (diameter = 8 mm) consists of normal human-derived keratinocytes, multiple viable cell layers, functional stratum corneum. Other in vitro tests confirmed LOM photocytotoxicity, as measured by cell viability and cytokines release (90). CIP photocytotoxicity was confirmed by a panel of tests including NR and MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assays (84). In in vivo experiments using ear analysis and NR cell test of rodents, LOM and SPA were also found to be photocytotoxic (91). LOM, GAT, SPA, PEF, CIP, OFL photocytotoxicity was evaluated by the Photo Hen Egg Test-C Chorioallantoic Membrane model (92). Apoptotic cells, where an extensive DNA damage activates programmed cell death genes, can be observed at least few hours after irradiation. RUF does not cause apoptosis in yeast (70,71); BAY, NOR, CIP, FLE and LOM (93) are inactive in V79 fibroblast. Apoptosis was observed in keratinocytes by caspase-3 activation and FAS-L gene induction (94). In addition, ENO and SPA lead to apoptosis in HaCaT keratinocytes. Apoptosis and consequent necrosis were evaluated by flow cytometric enumeration of annexin V+7−AAD− and annexin V+7−AAD+ cells, respectively. The expression of apoptosis-related molecules, caspase-3 and poly (ADP-ribose) polymerase (PARP), was tested by flow cytometric and Western blotting analyses (95).
Irritation and inflammation. Photosensitization testing models, such as the Integrated Model for the Differentiation of chemical-induced allergic and irritant Skin reactions (IMDS) (96), ensure an affordable screening of FQ phototoxicity. These tests allow determining how a compound may react after application to the skin and irradiated with solar simulated light. More precisely, IMDS allows discriminating between a contact sensitizer and a skin irritant with a combination of primary ear swelling analysis and cell counting of the ear-draining lymph nodes. In addition, inclusion of UV irradiation in the local lymph node assay enables discrimination of photoallergic from photoirritant reactions after dermal application. These assays are relevant for the endpoint studied on mice, such as the prediction of chemical-induced allergic side effects. These cannot be easily detected by other existing assays. Indeed, the in vitro tests are of proven value only in predicting irritant or toxic effects (96). ENO, SPA, BAY, LOM, OFL were tested by IMDS on guinea pig (96). Skin photoirritation was evaluated and scored by the appearance of erythema and swelling of the mice ears at various times after irradiation (91). The phototoxic potentials of 6,8-dihalogenated FQs LOM, SPA, FLE, CLI and of 6-FQs CIP, NOR, LEV and Tosufloxacin (TOS) are influenced not only by the substituent at position 8 but also by that at position 1. Differently, the 8-methoxy quinolones GAT and MOX did not induce any phototoxicity in ear of rodents (97). Incubation with ENO, LEV, OFL (98), SPA (98,99) and LOM (99) under UVA irradiation stimulates BALB/c 3T3 mouse fibroblast cells to release prostaglandin E2 (PGE2) and 6-keto-PGF1α, but not leukotriene B4. The PGE2 release is inhibited by cyclooxygenase inhibitors, antioxidants, protein kinase C (PKC) inhibitors and a tyrosine kinase (TK) inhibitor, but not by antibodies against tumor necrosis factor α (TNFα) and interleukin-1 (IL-1). This leads to the hypothesis that: (1) ROS generated from FQs under UVA irradiation trigger PG release from dermal fibroblasts via PKC and TK activation, resulting in skin inflammation; and (2) 5-lipoxygenase products, histamine, TNFα or IL-1 are ruled out from the mechanism (98,99).
Protein and photoallergy
Due to the multifaceted role of proteins in the various cellular compartments, such as membrane integrity, immunological involvement in the interaction with photohaptens, up to that of DNA replication, it seems important to dedicate particular attention to this photosensitization target. The binding interactions between OFL, CIP, NOR and LOM and BSA or RNAse, were investigated by use of steady state absorption and emission techniques. The binding constant value is around 105m−1 and the number of sites per drug molecule involved in the interaction is close to the unit (56). Photosensitized protein oxidation by drugs, with the consequent modification of their structure, is thought to be responsible for the occurrence of phototoxic phenomena such as photoallergy and loss of key biological functions. Thus, considering that proteins represent about the 60% of the dry weight of a cell, these macromolecules may represent an important target in the phototoxic action of FQs (56).
BSA and RNAse. FQs photo-oxidize proteins and induce protein–protein cross-links in isolated model proteins. Protein photo-oxidation by CIP, LOM and NOR was monitored following the decrease of Trp content in BSA as changes in its characteristic fluorescence. Differently from what is observed for CIP and NOR, an increase of fluorescence in LOM experiments suggests the formation of a photoadduct (56); OFL was found ineffective. Similar behavior is observed in Tyr from RNAse, suggesting a common mechanism for the FQs induced photodamage (56). The formation of haptens in MOX induced photoallergic reactions can be ascribed to cation radicals that react with BSA (46). To confirm that FQs can function immunologically as photohaptens, it was proposed that lysine (Lys) affords OFL photocoupling of peptides and OFL-photomodified peptides on class II molecules stimulate pathogenetic T cells in FQ photoallergy (100). This was supposed because Lys is preferentially degraded in model BSA that is UVA-conjugated with OFL. This finding seems in contrast with the observed inefficacy of OFL in protein photooxidation (56).
Other proteins. In presence of LEV, MOX (45), CIP, NOR, LOM (56) and OFL (45,56), light induced cross-linking of spectrin, a protein associated with the cytoplasmatic side of the RBC membrane, occurs. In this way, the formation of intermolecular cross-link induces an increase of the rigidity of the plasmatic cellular membranes with a consequent alteration of the cellular functions. OFL is the most active cross-linking photoagent; this is in good agreement with the phototoxicity order observed in the 3T3 cells (45). The good correlation between lipid peroxidation, protein photodamage and cellular phototoxicity, indicates that these compounds exert their toxic (photoallergic) effects mainly in the cellular membrane (45). To determine the site of peptides/proteins photobound to OFL and to assess the T-cell antigenicity of FQ-photocoupled peptides, murine Langerhans cells were used as photoadduct-presenting cells (100). In clinical photoallergy to FQs, authors assume that the drug photobinds to self-peptides on MHC class II molecules of Langerhans cells, thereby sensitizing and restimulating T cells. The development of FQ with lower photoaffinity to peptide/protein may avoid the photosensitive adverse effect of these widely used drugs. CIP, NOR, LOM, OFL were found to photopolimerize the lens protein α-crystallin (82). The extent of photopolymerization was attributed to the ability of these FQs to produce 1O2 in aqueous solutions at pH 7. Thus, these drugs, if taken systematically or injected intravitreally are potentially phototoxic to the eye and could contribute to early cataractogenesis (82).
Native DNA. The ability of FQs to produce ROS during irradiation seems to be indirectly linked with their phototoxic effect on DNA, although it does not necessarily correlate with the photodegradation quantum yield (40,41). FQs photosensitization on naked DNA (usually linear double strand DNA fragments) gives frequently 8-OH-dGuo as the main oxidation product. This product is a biomarker for mutagenic and carcinogenic events, with probable implication in aging processes (101). RUF (63,64), OFL (62,63), CIP, BAY, MOX (40), LOM (40,62), ENO and NOR (62) are able to photosensitize formation of this biomarker on naked DNA. A cooperative Type-I and -II mechanism was reported for RUF (64). Besides, H2O2, an unstable product of FQs photolysis (47), can react with traces of metal ions in order to promote the formation of the OH• radical via Haber–Weiss reaction, resulting in DNA base oxidation and strand cleavage (57,58). In parallel, the O2−• formed by drug photoionization could be able to extract a hydrogen atom from the sugar moiety that also results in cleavage of the duplex (58). 1O2 by itself is also able to induce DNA modifications with particular efficiency, acting selectively on guanine moiety according to its oxidation rate of 5 × 106m−1 s−1 (102–104). This value is much higher than that relative to oxidation of the other nucleobases. The 6,8-dihalogenated FQs LOM and BAY show the highest oxidative potential; by contrast, the 6-methoxy derivative MOX induces very low levels of 8-OH-dGuo (41). Other modified bases like oxidized pyrimidines (thymidine glycols, 5-formyl-2′-deoxyuridine and 5-(hydroxymethyl)-2′-deoxyuridine) were also detected using LOM and ENO. These two FQs and NOR (but not OFL) also form T<>T dimers (62), according to their triplet energy value (18,27). Indeed, this latter represents the major factor determining the photosensitized formation of T<>T dimers via energy transfer from excited FQ triplets to thymine triplet. The 3FQ energies levels must be above to a threshold level of 265–269 kJ mol−1 to make the drug a potential photosensitizer via T<>T formation (19). Some authors reported formation of T<>T dimers in cellular and isolated DNA induced by LOM (62,105). LOM exhibits relatively good triplet yields (25). This condition is required to allow triplet–triplet energy transfer but is not sufficient, as seen for OFL, which is inefficient in inducing T<>T formation. Moreover, LOM photodegradation is not affected by the presence of DNA (68), which would be against the energy transfer process.
Plasmids. FQs induce base damage in form of bulk oxidations in purines and pyrimidines of plasmids (circular supercoiled double strand DNA) as well as DNA strand cleavage. This latter is generated from alterations such as single strand breaks (SSB) and double strand breaks (DSB). For photocleavage experiments plasmid containing FQs are subjected to sensitization and the induced structural DNA changes are detected by agarose gel electrophoresis. This technique is very sensitive, because only one SSB is enough to convert the supercoiled circular form (Form I) into the nicked relaxed form (Form II) by means of SSB, whereas DSB lead to the formation of linear DNA (Form III). These three forms exhibit different electrophoretic mobilities.
Base oxidation: The plasmid model also reveals DNA base damages when agarose gel electrophoresis is combined with the use of specific DNA repair enzymes. Formamidopyrimidine DNA N-glycosylase (Fpg) reveals modified purine bases including 8-OH-dGuo, 2,4-diamino-6-hydroxy-5-formamidopyrimidine and 4,6-diamino-5-formamidopyrimidine. Endonuclease III (Endo III) is almost specific for oxidized pyrimidine bases such as thymine and cytosine glycols or 5-hydroxy-5-methylhydantoin. T<>T are evidenced by T4 endonuclease V or Micrococcus luteus endonuclease. Indeed, LEV (45) and OFL (39,45), as well as 6,8 dihalogenated FQs LOM and BAY (81,106) show a higher potential in plasmid DNA base oxidation of both purine and pyrimidine, whereas methoxy substitution (MOX) tends to reduce this effect (45). T<>T formation has been evidenced for ENO and PEF (18), as well as for NOR (18,27). OFL (18,39) and RUF (18) are unable to photoinduce this lesion, according to the low triplet energy value of these two drugs, as reported above (18,27). RUF induces formation of 8-OH-dGuo on plasmid DNA (70) with a yield similar to that achieved in sonicated calf thymus DNA (63). Its occurrence in these two biomodels can be attributed to one-electron oxidation reactions (107).
Cellular DNA. As often reported in this review, environmental conditions can influence the switches between the photosensitization mechanisms (e.g. Type-I/II reactions) (1–3,115). Moreover, the participation of FQs microbiologically active photodegradation products (116) and the role played by competitive intracellular endogenous photosensitizers can interfere with these mechanisms. For example, the complexity of endogenous skin photosensitizers, with regard to molecular structure, pathways of formation, mechanisms of action, and the diversity of relevant skin targets contributed to an underestimation of the importance of endogenous sensitizers in skin photodamage (22,117). Such variables interfere in cell photosensitization pathways, thus making difficult to rationalize FQs mechanisms of cell damage.
Base oxidation: Cellular DNA irradiation in the presence of RUF (70,71), BAY (118), LOM (41,62,119,120), NOR, OFL, ENO (62) and CIP (120) yields 8-OH-dGuo. The values are normally lower than those obtained in native DNA, as found in OFL, ENO, LOM and NOR (62), as well as in RUF (63,70,71). MOX induces very low levels of 8-OH-dGuo (118). Both purine and pyrimidine damage has been detected for ENO, NOR, OFL and LOM (62). Moreover, photosensitized T<>T formation has been reported in THP-1 cells from tumoral monocytes for NOR, LOM and ENO, but not for OFL (62). RUF potential to photogenerate 8-OH-dGuo in cell DNA was tested on a model organism, yeast Saccharomyces cerevisiae (71). As expected, cell biomarker levels are always lower with respect to those obtained in the other DNA models (free dGuo, bovine calf thymus DNA, plasmid, fibroblasts) (70).
Cleavage: Comet assay (i.e. single cell gel electrophoresis [SCGE]) is a rapid, simple, and sensitive technique for measuring DNA damage in individual eukaryotic cells. It has been used for in vivo and in vitro studies in the evaluation of the (photo)toxic potential of FQs, as well as the (photo)genotoxic potential of radiations. It can be used to quantify the presence of a wide variety of DNA lesions including DSB, SSB, ALS and apurinic-sites (AP-sites). In addition, this assay can be combined with the use of specific antibodies or with DNA repair enzymes to evaluate the level of modified purine and pyrimidine bases. Comet assay allows verifying cellular DNA repair ability by measuring the decrease of the DNA damage when cells are incubated after FQs photosensitization (70,93). Photoinduced Comet formation has been revealed with ENO (62), RUF (70,71), FLE (93), NOR (62,93), BAY (93,106), SPA (112,121), LOM (62,93,94,105,106,122,123), CLI (122), CIP (93,122,124) and OFL (62,112). This latter was shown inefficient by other authors (124). Repair enzymes in Comet assay have revealed purine bases oxidations through formation of Fpg-sensitive sites on tumoral cells photosensitized by ENO, LOM, OFL and NOR (62). Differently from what obtained in MS/MS analysis of extracted cellular DNA, only LOM, which acts principally via a Type-I mechanism, induced a significant number of endo III-sensitive sites, diagnostic for oxidized pyrimidines (62). Moreover, T<>T has been evidenced for LOM using enzymatic treatment in human keratinocyte line (HaCaT) (105) or immunochemical detection using monoclonal antibodies in human fibroblasts and keratinocytes (94). The data of Comet assay of FQs photosensitized cells are complementary to those obtained by plasmid photosensitization. Indeed, a structural correlation to genomic DNA seems to be a reasonable way in order to give further details about photodamaged cell DNA revealed by unwinding in electrophoresis. Comet assay allows verifying yeasts cellular DNA repair ability in RUF photodamage (125). Photogenotoxicity has been assessed with Comet assay in UVA photosensitized rodents with SPA (121), CLI, CIP (122) and LOM (122,123).
Photogenotoxicity and photocarcinogenicity
Although mammalian cells possess effective repair mechanisms for oxidative damage, photoproducts, and dimers, these repair mechanisms can be overloaded. Eventually, unrepaired damage leads to gene mutations or chromosomal damage in exposed cells and to tumors in the skin. Therefore, in vitro photomutagenicity testing in mammalian cells may be an early and easy-to-measure predictor of the photocarcinogenic potential of a pharmaceutical (21).
Some mutations lead to a gene (or more) injury. This kind of cell damage was observed after UVA photosensitization in vivo and in vitro. Several authors defined photogenotoxic in vitro evidences. These can include:
1 Comets formation both on human and mouse cells photoinduced by LOM (93,106,123) and BAY (93,106), CIP, FLE, NOR (93), and on wild type yeasts in RUF photosensitization (71).
2 Positivity to Ames test in bacterial systems. This test consists in induction of the change of the genetic material of an organism that increases the frequency of mutations above the natural background level. LOM, FLE and CIP resulted phototoxic by means of standard Salmonella typhimurium (TA 100 and TA 104) reverse mutation test batteries (126).
3 Photosensitized point mutation in a gene locus and consequent formation of forward mutations. This photomutagenic potency was explored with RUF using a panel of yeast (S. cerevisiae) mutants affected in different DNA repair pathways (125). Similar experiments are reported with LOM and BAY analyzing the frequency of revertant and recombinant colony in the mutant yeast S. cerevisiae (D7-rad3) (106). In addition, point mutations in mammalian cells were used as routine genotoxicity tests: BAY and LOM, but not MOX, resulted mutagenic in the H(G)PRT (Hypoxanthine Guanine PhosphoRibosyl Transferase) test in V79 Chinese hamster cells (127).
4 Lomefloxacin photoinduced triggering of various stress response genes (heme-oxygenase-1, p53, MDM2, GADD45) in some human skin cells (fibroblasts, keratinocytes and Caucasian melanocytes) (94), as well as p53 activation or melanogenesis stimulation by LOM and BAY (106).
5 Positivity to photo-Chromosomal Aberration (photo-CA) assay, based on detection of photoclastogens (genotoxic chromosome damage). This damage was reported for CIP (126,128), FLE, LOM (126,128,129), CLI (128,129) SPA, ENO, LEV, GRE, SIT and Gemifloxacin (GEM) (129).
6 Micronucleus formation on Chinese hamster V79 cells: NOR, ENO, CIP, OFL, TRO, SPA, LOM, CLI (130), LOM, GRE and BAY (131) resulted positive for this test.
These photogenotoxic evidences can result in several clinical consequences. For example, ENR is able to photoinduce retinal degeneration and blindness in felines by means of photogenotoxicity in the feline ABCG2 gene (132). Genotoxic damage by means of induction of tumors, ear swelling response, immuno-histochemical and histological examination as well as sunburn cell formation counting was found in skin cells of LOM and OFL photosensitized rodents (133). The correlation of magnitude in the skin micronucleus test for several FQs is similar to that found in previous in vitro photochemical clastogenicity studies, i.e. photo-CA assay, and with their reported in vivo phototoxic potentials (129).
As many mutations can cause cancer, photomutagens can be photocarcinogenic (4–6,93). There is an indication that the determination of photoclastogenicity might be a good measure for the assessment of a photocarcinogenic potential. However, mechanistical studies in the case of a photoclastogenic substance seem to be necessary to support the risk assessment. Indeed, between the FQs resulted positive to the photo-CA assay, only CIP (4,5), FLE (4–6) and LOM (4,5,133), resulted photocarcinogenic. In addition, in the same experiment for photogenotoxicity described above, rodents deficient in T<>T-DNA repair systems resulted positive for LOM photoinduced skin tumor, but not for OFL (133). In in vivo experiments, CIP, LOM, OFL and FLE has shown an increased number of benign neoplasms, but only LOM developed malignant tumors, like squamous cell carcinomas (4,5).
Targeting of Cellular Organelles
Several studies involving cytoplasm regard lysosomes. The accumulation of FQs in lysosomes probably results from their Lewis acid-base properties with pKa close to neutrality, which confers them lysosomotropic properties. By definition, lysosomotropism implies partition of ionic species in the cytosol and in the acidic lysosomes with respect to their charge (134). Different ionic species of FQs, therefore, undergo partition in various compartments and organelles of living cells according to their ionic equilibria. Thus, lysosomes are a preferential site of localization of FQs in keratinocytes (72), lens epithelial cells (82) and fibroblasts (69,135).
In a recent study on live human epidermal HaCaT keratinocytes, laser scanning confocal microscopy was used to detect and to follow the fluorescence changes of MOX and of three dihalogenated FQs (CLI, BAY, LOM) during in situ irradiation by confocal laser in real time (72). Preferential localization of CIP, NOR, LOM and OFL in lysosomes of human lens HLE B-3 epithelial cells was determined by the fluorescence overlay of the FQs and lysosomal probe Lyso Traker (82). This is in agreement with a previous paper, where lysosomes were found to be the primary site of phototoxicity of these FQs for HS 68 skin fibroblasts (69). The endocytotic pathway is profoundly altered by the UVA-induced photosensitization of HS 68 fibroblasts by the FQs antibiotics LOM, BAY, NOR and CIP (135). The endocytosis of low-density lipoproteins (LDL) loaded with two carbocyanine dyes compatible for effective Forster-type Resonance Energy Transfer (FRET), has been used as a model system. Perturbation of the membrane by the FQs is revealed by the change in the rate of exchange of the donor dye from the LDL to the cell membrane as compared to untreated cells. FQs partly inhibit the lysosomal degradation of LDL, as demonstrated by the disappearance of FRET between the donor and the acceptor dyes. In addition, upon irradiation, the four FQs readily destroy the actin filament network, involved in the fusion of mature endosomes with lysosomes. All these data suggest that several components of the endocytotic pathway are impaired by photosensitization with these FQs (135).
The microspectrofluorometry technique shows that although OFL, LOM, NOR, BAY and CIP are readily incorporated into lysosomes of HS68 human skin fibroblasts, also a weak staining of the whole cytoplasm occurs, especially with NOR (69). Thus, besides lysosomes, these five FQs can be found in the whole cytoplasm (83). Therefore, it may be that cytoplasm components other than lysosomes can also be target of FQs sensitization. As a result, photosensitized damage to other cytoplasmatic sites than lysosomes can also be expected (83). In particular, FQs photosensitization on the cell respiration machinery, which implies the integrity of mitochondria in HS 68 fibroblasts, was studied (83). Thus, using microfluorometry and rhodamine 123 as a specific fluorescent probe which is released from mitochondria by light absorption, it was demonstrated that under UVA irradiation NOR and CIP readily damage mitochondrial membranes. By contrast, no photoinduced damage can be observed with OFL, LOM and BAY, the latter being the most phototoxic derivative towards cell viability of HS 68 fibroblasts (69). Photocleavage of mitochondrial DNA was also reported with BAY in embryonic turkey liver (136). Alteration of the mitochondrial functions photoinduced by OFL, LEV and MOX was studied in Balb/c 3T3 cells following mitochondrial potential alteration by the fluorescence of a lipophilic cation commonly used for the assessment of the mitochondrial potential. This potential declines in the case of OFL and LEV, whereas it remains unaltered in presence of MOX (45).
This review illustrates as the photosensitization reactions of FQs, involving various biotargets, lead to several photomodifications in vitro. A comparison between the FQs discussed in this review and present on the market shows that LOM has significantly more photosensitizing potential than the other drugs, whereas MOX shows very mild photosensitization effects. At this regard, we have considered the role of chemical structures within FQ family on the phototoxic potential. Indeed, the photoreactivity of FQs is markedly modulated by the nature and position of the substituents attached to the quinolone structure. Photodehalogenation seems to be the main process involved in the phototoxic effect of FQs and the 6,8-dehalogenated FQs show the more significant evidences, such as the carcinogenic potential of LOM and FLE, and SPA withdrawing from the market. The efficiency of the C-F bond fragmentation is influenced by the substituent in position 8. Again, methoxy substitution (MOX) tends to reduce the photosensitizing activity. Nevertheless, some observations cannot be rationalized. For example, the production of 1O2 does not correlate with the order reported for FQ phototoxicity that, in addition, does not necessarily correlate with the photodegradation quantum yield. Moreover, the photodehalogenation process is not significant in some FQ, such as CIP, that photoinduces benign neoplasms, and RUF, that resulted photomutagenic in vitro.
In addition, we found and cited throughout the review some of the significant discrepancies reported in the literature, such as the attribution of some transient bands, reactions of FQs excited states with amino acids residues and FQs relative efficiency both of protein photo-oxidation and nuclear photoinduced damage, as well as rationalization of T<>T dimers formation in DNA.
The overall results described in the review, which represent the fruit of extensive work performed by many research teams in the word, take part in an overall picture of the possible biological and/or pathological implications of a variable degree of severity, including photoallergy, photomutagenesis up to photocarcinogenesis. This could represent a warning for researchers to take always into account the extensive and cross-targeted implications, covering photoinduced damages in systems of increasing molecular complexity.
[ Guido de Guidi ]
Guido de Guidi was born in 1956 in Piacenza, Italy. He moved to Catania in 1972 where he was graduated and got a PhD in Biochemistry and Physiology. In 1993, he was visiting Scientist at the California Institute of Technology, Pasadena, California (Prof. J. Barton Group). Since 1991 he was Assistant Professor in General and Inorganic Chemistry and since 1998 he is Professor of General and Inorganic Chemistry at the University of Catania. His main international collaborations are with the Universidad Politecnica de Valencia, Espana (Miguel A. Miranda), the Institute Curie, CNRS, Paris Orsay (Evelyne Sage) and the University of Ottawa, Canada (J. C. Scaiano). He acts as reviewer for several journals in the field of photo(biological)chemistry such as Photochemistry and Photobiology, Photochemical and Photobiological Sciences, Journal of Photochemistry and Photobiology A and B, Chemical Research in Toxicology. He has published more than one hundred papers dealing mainly with (1) thermal and photochemical reactivity of coordination compounds; (2) photosensitization mechanisms in biological systems induced by xenobiotic agents; and (3) environmental chemistry.
[ Giuseppina Bracchitta ]
Giuseppina Bracchitta was born in 1983 in Ragusa, Italy. She received her education at the University of Catania. In July 2008, she was graduated in Biological Science defending a thesis on effects of Methylene Blue and Naproxen in Tryptophan photosensitization, following the photoinduced formation of different amino acid photoproducts. At the beginning of 2009, she began a PhD course in Biochemical and Biomolecular Sciences under the supervision of Professor De Guidi. The research project of this PhD course deals with molecular mechanisms of drug photosensitization in biological systems of increasing complexity. For this project, she attended 4 months of training at the Institute Curie, CNRS, Paris Orsay, under the tutorship of Evelyne Sage and Pierre-Marie Girard.
[ Alfio Catalfo ]
Alfio Catalfo was born in Catania, Italy, but was largely raised in Dormagen, Germany. Following his undergraduate studies at the University of Catania, he earned his PhD degree in Biochemical and Biomolecular Sciences at the same institution in 2004 under the direction of Professor Guido De Guidi. From 2004 to 2010, he was a postdoctoral fellow at the same university with Professor Guido De Guidi. He currently works on toxicology resulting from environmental pollution.