Advanced glycation endproducts are well described as photosensitizers being capable of generating ROS when they are irradiated with UVA light. This fact has been proposed as one of the causes of cataract development (6–22). Nonetheless, their actual potential for damaging eye lens proteins is poorly understood. Specific damage to some reactive amino acids within lens proteins as a consequence of singlet oxygen generation has been measured (18), regardless of the low oxygen concentration found in the lens, which is unable to sustain a significant singlet oxygen formation (14).
In a previous work, we have demonstrated that UVA-visible irradiation of lysine-derived AGEs at low oxygen concentration can induce the photodecomposition of free Trp and Trp-residues in human serum albumin (21). In the present paper, we extend the study to eye lens proteins, using simple models to elucidate the sensitizing mechanism that leads to photo-crosslinking and oxidation of the proteins.
Advanced glycation endproducts were prepared considering the important role of lysine and arginine in the generation of known glycation products. For this purpose, equimolar amounts of N-Ac-Lys and N-Ac-Arg were incubated with a high concentration of Glc. The absorption and emission properties of the sensitizer used in this work are shown in Fig. 1. A major fluorophore was detected at Ex330/Em390 nm along with a minor fluorophore at Ex380/Em440 nm. These data are consistent with the generation of known Lys–Arg and Lys–Lys crosslinks (26,27). This mixture of glycation products was used directly as a sensitizer to study the photochemical behavior of AGEs.
AGE-sensitized tryptophan photodecomposition quantum yields
To investigate the photosensitizing mechanism of AGEs, Trp was used as a model target and the photodecomposition quantum yields at 367 nm of this amino acid were determined at different oxygen concentrations. FeCy, a known electron-scavenger (28), was used with the aim to interfere with electron-transfer mechanism, whereas deuterium oxide buffer was used so as to enhance singlet oxygen-mediated photodamage.
The results shown in Table 1 support a combined Type-I–Type-II mechanism for all oxygen concentrations, being Type I and Type II favored at low and high oxygen concentrations, respectively. The fraction of Type-I (ftype I) and Type-II (ftype II) mechanisms that contributes to Trp photodecomposition depends on both of the following: Trp and oxygen concentration, and the rate constants of the reactions between triplet AGEs and Trp or O2, as expressed by the equations below:
Table 1. Quantum yields for AGE-sensitized Trp decomposition at 367 nm under different oxygen pressures.
|Trp photodecomposition quantum yields (φ367 × 103)|
|O2 (%)||Control||Added FeCy*||D2O buffer†|
where ket, kic, kP and kq correspond to the rate constants of the electron-transfer between triplet AGEs and Trp, internal conversion, phosphorescence emission and quenching of triplet AGEs by oxygen, respectively.
Given that Trp concentration was the same in all the experiments, the contribution of Type-I and Type-II photoprocesses depends solely on the oxygen concentration and the values for the rate constants of the processes.
At 5% O2, Type-I photosensitizing mechanism is predominant because the presence of FeCy considerably drops the quantum yield to 51% with respect to the control. A similar result was reported by our group for the predominantly Type-I sensitizer, riboflavin (29). Under the same conditions, D2O increases the damage 2.6 times, which is low if we consider that deuterium oxide prolongs the lifetime of singlet oxygen about 10–17 times compared to water (30). This result indicates a small contribution of singlet oxygen at low oxygen concentration.
At 20% O2, the quantum yield drops to 64% with respect to that at 5% O2. This result is consistent with AGEs acting preferentially as Type-I sensitizers, and therefore, the four-fold increase in the oxygen concentration disfavors the interaction between the triplet sensitizer and tryptophan, and this is reflected in a lower damage to the target. The presence of FeCy does not inhibit the process to a great extent (down to 80% of the control), which agrees with the fact that at 20% oxygen, Type-I mechanism is disfavored. A high oxygen concentration favors the quenching of the triplet state of the sensitizer by ground state oxygen in a Type-II mechanism that generates singlet oxygen, and this is observed as a five-fold increase in the quantum yield when the reaction is performed in deuterium oxide buffer.
At 100% O2, the quantum yield rises again to 89% of the value observed at 5% O2. This result was surprising because for a Type-I sensitizer, an even greater decrease in the quantum yield was expected compared to the value at 20% O2. A plausible explanation for this behavior is that at 100% O2, the increased amount of singlet oxygen generated in this condition compensates the loss in Type-I mechanism, balancing the total efficiency of the process. The influence of FeCy and D2O in the quantum yield shows the same behavior as that observed at 20% O2, being consistent with the occurrence of Type-II mechanism at high oxygen concentrations.
Notwithstanding the fact that an enhancement of the photodamage in the presence of deuterium oxide is observed for all oxygen concentrations, supporting the participation of Type-II mechanism, the increase is small compared to the expected enhancement of the lifetime of singlet oxygen. The ratio between the quantum yields in D2O and in H2O is 2.62 at 5% O2, which is low compared to that for known Type-II sensitizers such as methylene blue, where a ratio of 8.29 in oxygen-saturated solutions is observed (31). In the same work, a ratio of 4.99 is obtained for riboflavin in aerated solutions, which agrees well with the values of 4.97 and 4.94 observed for 20% and 100% oxygen concentrations found for AGEs in our conditions. These data indicate that Type-I photoprocesses are the principal mechanism by which AGEs induce photodamage under low oxygen conditions.
Photocrosslinking of bovine lens proteins sensitized by AGEs
As the results described above show the presence of Type-I photosensitizing mechanism, and this would lead to the formation of radical intermediates, we investigated the possibility of a radical-mediated photo-crosslinking of lens proteins. Figure 2a shows the time-dependent formation of crosslinking when the total water-soluble bovine eye lens proteins are irradiated with UVA-visible light in the presence of AGEs at 5% O2 concentration.
Figure 2. SDS-PAGE analysis of (a) bovine lens proteins, (b) α-, (c) βH- and (d) βL-crystallins, irradiated during 10, 20, 30, 40, 50 and 60 min with UVA-visible light in the presence of AGEs (absorbance of 0.2 at 365 nm) in 100 mm phosphate buffer pH 7.4 under a 5% oxygen atmosphere. The main crosslinking band represents protein dimers with an average molecular weight of 41 kDa.
Download figure to PowerPoint
At initial time, eye lens proteins contain a negligible amount of crosslinking generated in vivo. When these proteins are incubated in the dark in the presence of AGEs (dark control) or irradiated during 1 h in the absence of the chromophores (sensitizer control), the electrophoretic pattern of these samples does not show any modifications (data not shown). SDS-PAGE analyses were performed in the presence of a reducing agent so as to exclude disulfide bonds and leave only the irreversible covalent crosslinking. The main band centered at 41 kDa represents the formation of dimers with molecular weights in the range 36–47 kDa, arising from the crosslinking of the crystallin subunits with an average molecular weight of about 20 kDa (32). The photochemical aggregation of lens crystallins due to UV irradiation has been described before, although in these studies the nature of the crosslinking was probably the direct photoexcitation of Trp residues or its oxidation products (33–35). Lee et al. described the crosslinking of lens proteins as a result of the sensitizing activity of the chromophores extracted from brunescent cataractous lenses in the presence of air (36); however, no further investigations were performed in this area.
Crystallins represent 95% of the total lens proteins, of which the major components are α- and β-crystallins (40% each). To observe what happened with each one of the fractions, a separation through size exclusion chromatography was conducted. Isolated α-, βH- and βL-crystallins were irradiated under the same conditions as those for the whole lens proteins homogenate, as described in Materials and Methods. Crosslinking of all the fractions was observed. Alpha- and beta-crystallins showed different grades of crosslinking, decreasing in the order α- < βH- < βL-crystallins (Fig. 2b–d, respectively). These results are consistent with an early study from our group using riboflavin as sensitizer for lens crystallins, where a good correlation between the decomposition of Trp residues and protein aggregation was found (37). These data point to Trp as a key target in the photoprocesses that lead to the observed crosslinking, as the number of Trp residues within the fractions are 2.6- and 2.7-fold higher for the beta fractions (βH- and βL-crystallins, respectively) than the alpha fraction (37), which is consistent with the experimental observations. Nevertheless, the participation of other amino acids cannot be discarded.
Photocrosslinking mechanism of bovine lens proteins
To further investigate the mechanism of AGE-sensitized protein crosslinking, a series of experiments were performed by irradiating 3 mg mL−1 lens proteins solutions in the presence of AGEs adjusted to an absorbance of 0.2 at 365 nm under increasing oxygen concentrations (5%, 20% and 100%). Figure 3 shows the SDS-PAGE analysis of the samples irradiated during 60 min under different oxygen pressures. A rise in the crosslinking degree with the increase in oxygen concentration can be observed, indicating that Type-II mechanism and/or secondary reactions mediated by oxygen begin to be important in the aggregation process.
Figure 3. SDS-PAGE analysis of bovine lens proteins irradiated during 10, 20, 30, 40, 50 and 60 min with UVA-visible light in the presence of AGEs (absorbance of 0.2 at 365 nm) in 100 mm phosphate buffer pH 7.4 under (a) 5%, (b) 20% and (c) 100% oxygen concentrations.
Download figure to PowerPoint
Subsequently, the influence of FeCy and D2O on lens proteins photo-crosslinking was investigated for every oxygen concentration. Figure 4a shows the effect of the additives at 5% O2. A great decrease in the crosslinking was observed in the presence of FeCy, indicating the occurrence of Type-I interaction between AGE-sensitizers and the proteins. In the presence of D2O buffer, only a slight increase in the crosslinking was observed, giving evidence for a small participation of singlet oxygen in the crosslinking at low oxygen concentration, as expected due to the predominance of Type-I mechanism.
Figure 4. Quantification of the photoinduced crosslinking of bovine lens proteins when a 3 mg mL−1 solution is irradiated with UVA-visible light in the presence of AGEs (absorbance of 0.2 at 365 nm) under different oxygen pressures. Panels (a–c) show the effect of FeCy and D2O at 5%, 20% and 100% O2, respectively. Initial crosslinking was normalized to zero in all cases. Symbols: (▪) 5% O2, () 20% O2, () 100% O2, (∇) Added FeCy and (◊) D2O buffer. Data are the mean ± SD of three experiments. The percentage of crosslinking corresponds to the integrated area of the bands above 37 kDa referred to the area of the total protein.
Download figure to PowerPoint
Carbon-centered radicals produced on the protein are prone to react with oxygen, generating reactive species that can further enhance the production of neutral radicals (38). The reactions are the following:
It is important to note that the oxygen concentration present in the reaction medium can modulate both, the occurrence of this oxygen-mediated radical mechanisms and the prevalence of Type-I or Type-II mechanisms.
Notwithstanding the fact that the reaction of neutral radicals with oxygen is an efficient process, protein radical recombination could also make a contribution to protein crosslinking, especially at low O2 concentration (Eq. 8).
Although this reaction does not require oxygen to proceed and hence could be favored in its absence, a reduced protein crosslinking was observed when the solutions were irradiated under a nitrogen atmosphere (data not shown). This result is consistent with earlier studies from our group that indicate that the efficiency of AGE-sensitized photoprocesses decreases dramatically under oxygen-free conditions due to the impossibility of recovering the sensitizer in its active form, limiting the extension of the reaction (22).
The improvement in the efficiency of this recombination process in the case of lens proteins can be explained by the conformational properties of the lens crystallins that possess a quaternary structure configuring multimeric complexes (39–41) in which the subunits are in close vicinity, which could facilitate radical recombination. When a model protein such as glucose 6-phosphate dehydrogenase is irradiated in the presence of AGEs in the same conditions described in this paper, crosslinking is not observed despite the fact that an important photosensitized inactivation is clearly appreciated (data not shown). A similar behavior with respect to this Type-I–mediated crosslinking has been observed for riboflavin, whose photochemical efficiency on eye lens proteins has been compared with those of AGEs (9). The interaction between triplet riboflavin and crystallins together with the induction of protein crosslinking at low oxygen concentration has been demonstrated (37). The amino acids susceptible to Type-I interaction are mainly Trp, Tyr and His (42,43). Formation of Trp–Trp (44) and Tyr–Tyr (45,46) crosslinks through radical mechanisms has been observed. These processes are characterized by very low efficiencies in anaerobic conditions because oxygen is required to prevent photobleaching of the sensitizer and promote its regeneration (22). In spite of this, riboflavin-sensitized crosslinking of proteins other than crystallins had not been observed until recently (47), using an extremely high concentration of the target protein, suggesting that the radical recombination is highly dependent on the proximity of the participating groups and emphasizing that this condition should be especially favorable in crystallin multimeric complexes.
At 20% O2 (Fig. 4b) and 100% O2 (Fig. 4c), the influence of FeCy is smaller than that at 5% O2 and this is consistent with a higher proportion of Type-II mechanism over Type I at higher oxygen concentrations. The effect of D2O on the enhancement of the crosslinking is more marked than that at 5% O2, giving evidence for the contribution of singlet oxygen to the observed crosslinking at high oxygen concentrations.
The experiments performed at high oxygen concentration, in which the interactions between triplet AGEs and oxygen is favored, showed an efficient covalent protein aggregation. The involvement of 1O2 in the photodynamic crosslinking of proteins has been reported previously (48–51). It has been proposed that photodynamically generated singlet oxygen interacts with photo-oxidizable amino acids residues such as His, Cys, Trp and Tyr in one protein molecule to generate reactive species. These in turn interact nonphotochemically with residues of these types or with free amino groups in another protein molecule to form a crosslink such as His–His or His–Lys (52–55). In the case of His–His singlet oxygen-mediated crosslinking, the first step is the 1,4-cycloaddition of singlet oxygen to the His-imidazole ring to give a stable endoperoxide. This then undergoes changes followed by nucleophilic addition of a second imidazole ring and the elimination of one molecule of water to give His–His crosslinks (52).
Photo-oxidation of lens proteins and peroxide formation sensitized by AGEs
Carbonyl groups can be introduced into proteins as a consequence of the sensitized photo-oxidation of amino acids such as Trp, His and Tyr in both Type-I and Type-II mechanisms (56,57). Therefore, the presence of carbonylated proteins due to AGEs photosensitizing activity at 5%, 20% and 100% O2 was detected with the Oxyblot assay.
Lens proteins are known to bind nonspecifically to other proteins. To minimize nonspecific binding, the antibodies were incubated in the presence of native lens proteins to observe only DNP-specific signals. Clear blots were obtained for all oxygen concentrations, as depicted in Fig. 5. Negative controls, which included both irradiated and nonirradiated samples that were not derivatized with DNPH, showed no signal. On the other hand, derivatized samples, both irradiated and nonirradiated, which were only incubated with the secondary antibody, also failed to produce any signal. Figure 5 shows an increase in the carbonyl content throughout the irradiation period at all oxygen concentrations. The results show that the amount of carbonyl groups formed increases with the rise in oxygen concentration, which suggests an important role of oxygen-mediated reactions in the photo-oxidation of lens proteins. Both the secondary reactions of radical species formed in Type-I mechanism and the decomposition of singlet oxygen intermediates appear to be important. Additional experiments showed that the photo-oxidation of lens proteins is partially inhibited by FeCy and increased in the presence of D2O, in the same manner as the effect observed for protein crosslinking (data not shown).
Figure 5. Oxyblot analysis of bovine lens proteins (3 mg mL−1) irradiated with UVA-visible light during 60 min, in the presence of AGEs (absorbance of 0.2 at 365 nm) under different oxygen pressures. Panels (a), (b) and (c) show the increase in carbonyl content at 5% O2, 20% O2 and 100% O2, respectively. These Oxyblots correspond to the SDS-PAGE shown in Fig. 3.
Download figure to PowerPoint
For a better accuracy in the interpretation of the results, carbonyl groups were quantified by the DNPH spectrophotometric method. At initial time, bovine lens proteins contain a small amount of carbonyl groups (1.87 nmol mg−1 of protein) and this quantity does not change significantly during a 1 h incubation period in the presence of AGEs (<1 nmol mg−1 protein). Carbonyl group formation was observed exclusively when the samples were irradiated in the presence of AGEs, with maximal values obtained at 60 min of irradiation and corresponding to 8.70, 10.49 and 12.35 nmol mg−1 of protein for 5%, 20% and 100% O2, respectively. These results are in agreement with the corresponding Oxyblots shown in Fig. 5. It is noteworthy that carbonyl group introduction due to photo-oxidation is particularly low and may not represent a major end-result of AGE-sensitized photoprocesses.
Recent studies point out that the main products of Type-I and Type-II photosensitizing mechanisms could be the generation of reactive peroxides (24,58,59). Singlet oxygen, generated in Type-II mechanism, reacts mainly with Trp-, Tyr- and His-residues, with the consequent formation of hydroperoxides and/or endoperoxides (60,61). Type-I interaction can also generate hydroperoxides through the reactions of radical intermediates with oxygen (Eq. 7) and later hydrogen abstraction (Eq. 8) (38). The peroxides formed by either Type-I or Type-II mechanisms can further decompose thermally or undergo a catalytic breakdown in the presence of metal ions in a pseudo-Fenton reaction (62), promoting the formation of radicals that could eventually lead to the crosslinking of lens proteins.
To investigate the formation of peroxides in the process of photo-crosslinking of lens proteins, these species were determined by a modified FOX-2 method. The results shown in Fig. 6 show the generation of peroxides during the irradiation of lens proteins in the presence of AGEs at 5%, 20% and 100% oxygen concentrations. The amount of peroxides generated during the irradiations was higher for 5% O2 concentration, followed by 100% O2 and 20% O2. The addition of catalase (65 U mL−1) after the irradiations almost completely abolishes the peroxides detected, indicating the presence of H2O2 as the main peroxide generated in the reaction medium. Under our experimental conditions, hydrogen peroxide accounted for more than 98% of the total peroxides generated in the photoprocesses at 5% and 20% oxygen concentrations, while protein hydroperoxide formation was detected exclusively at 100% O2, with values lesser than 7% of the total generated peroxides. Taking into account the experimental conditions of this study, a low yield of protein peroxides is expected as these species are particularly prone to decompose in the presence of metals and UV light (63–65), which were both present in our system. However, the decomposition of these intermediates gives rise to oxidation products and radical species that are consistent with the oxidation and crosslinking observed for lens proteins. Formation of H2O2 was not increased in the presence of 500 U mL−1superoxide dismutase (data not shown) indicating that the main source of hydrogen peroxide is associated to the reaction between molecular oxygen and the reduced sensitizer (66,67), as proposed in the reactions below:
Figure 6. Generation of peroxides during the irradiation of bovine lens proteins (3 mg mL−1) in the presence of AGEs (absorbance of 0.2 at 365 nm) under different oxygen pressures. The addition of CAT 65 U mL−1 after the irradiations eliminated most of the peroxides detected (dashed lines). Symbols: (□) 5% O2, () 20% O2 and () 100% O2. Data are the mean ± SD of two experiments.
Download figure to PowerPoint
The lack of increase in H2O2 production in the presence of superoxide dismutase suggests that superoxide radical anion generation by the reaction between AGEs radical anion and oxygen could not be significant in our system or that this specie is rapidly consumed in additional reactions with carbon-centered radicals.
Given the presence of hydrogen peroxide in the reaction medium together with the ubiquitous presence of metal traces, the generation of hydroxyl radicals is very probable. These species constitute a new source of radicals that lead to crosslinking and/or oxygen-mediated reactions, where the oxygen concentration modulates the prevalence of one or another process.