Conclusions.

While it has long been known that the lethal effects of photosensitizing dyes such as the thiazine, xanthenes, and acridines involve oxygen, little has been known about the mechanism by which these dyes exert their lethality. This study presents evidence that toxicity due to the photodynamic effect is not primarily an effect of singlet oxygen, but that, instead, OH. is an important mediator of photodynamic lethality. The OH. is not produced directly by the dye upon illumination, but depends upon the presence of oxidizable substrates and appropriately chelated iron; little reaction occurs in the absence of oxygen, light, dye, iron, chelator, or oxidizable substrate. Evidently, the dye is first excited by the absorption of light of the appropriate wavelength; in this excited state, it is then capable of reduction by the substrate. After reduction, the dye reacts with triplet state dioxygen, producing O2- plus the ground state form of the dye. The O2- dismutes to form hydrogen peroxide, spontaneously or with the aid of SOD. Either O2- or the excited dye reduces the chelated iron(III) to iron(II). Hydrogen peroxide then reacts with the iron(II) to form OH.. This OH. then attacks any cellular components in the vicinity of the reaction, including, particularly, DNA, which forms an intercalated complex with many of the dyes.

Results which support this scenario include the following. Superoxide was found to be produced by all dyes tested, intercalated or not. NADH was reduced by the dyes, again whether intercalated or not. Hydroxyl radical was produced by the dyes, whether intercalated or not, and this production was inhibited by the presence of catalase. Moreover, dye sensitized production of OH. was augmented by addition of H2O2 to reaction mixtures. The strand scission of DNA by the dyes was prevented by catalase and, to some extent, superoxide dismutase, as well as by every hydroxyl radical scavenger tested. Lethality in E. coli was reduced by enhanced levels of catalase and of a DNA repair enzyme.

The dyes that have been found to act in this fashion include the thiazines, methylene blue, thionin, azure c, and toluidine blue o; a phenazine, neutral red; the acridines, proflavin, acridine orange and acridine yellow; and the xanthenes, rose bengal and fluorescein. While similar concentrations of these dyes have different degrees of effectiveness in these reactions, they all appear to act in much the same manner. Other dyes that have been investigated briefly and found to apparently act in a similar fashion include lucifer yellow and hematoporphyrin derivative. These dyes were all studied in order to determine whether their reaction mechanisms are similar. In general, they seem to be extremely similar. The greatest difference between the dye classes studied appears to be in their reactivity toward different oxidizable substrates, and their access to the intracellular compartment of E. coli. NADH and glutathione were both quite effective with all of the dyes studied except for quinacrine, while GTP and GMP were found to be particularly effective in the reaction with rose bengal, but less so in reactions with the thiazines, and even less so in reactions with neutral red. Tryptophan and tyrosine were much more effective with azure c than with rose bengal, proflavin, or neutral red. Other reaction participants, however, appear to be identical in their reactions with all active dye classes, as would be expected since all of the reactions are identical after O2- is produced and chelated iron reduced. One exception lies in the varying ability of different classes of excited dyes to react with iron(III), substituting for O2-, as would be expected since in this reaction the specific structure of the dye in question may affect the course of the reaction. Under the conditions of the DNA strand scission experiments, rose bengal, proflavin, acridine orange, lucifer yellow, neutral red, fluorescein, thionin, hematoporphyrin derivative, azure c, and toluidine blue all showed at least partial protection with SOD, while methylene blue did not; however, under the harsher conditions of the hydroxyl radical experiments-higher light intensity and iron and chelator concentrations-SOD showed less protection for methylene blue, neutral red, proflavin, and fluorescein, indicating that the semi-reduced dye is reducing the iron. More clearcut is the result showing that OH. is produced in the presence of DTPA-chelated iron by thionin, methylene blue, fluorescein, proflavin, acrdine orange, and neutral red; the only exception was lucifer yellow. As superoxide is not capable of reducing DTPA-chelated iron as it does EDTA-chelated iron, it appears that all of these dyes, when excited and subsequently reduced, are capable of this.

Work by Martin et al.200,183 established that lethality by dyes is a radical phenomenon; therefore, it is clear that intracellular metals and chelators do exist to carry out the Haber-Weiss reaction in vivo. However, until now there has been no suggestion as to what substances of biological relevance may act as chelators, or what substances besides NADH may act as oxidizable substrates for the dyes. In vitro studies presented in this thesis provide a number of relevant suggestions as to which substrates and chelators may be important, how much iron is necessary, and how important DNA may be as a target for the radicals. Substrates that can be oxidized by the photoexcited dyes include cysteine, tryptophan, tyrosine, glutathione, and GMP. Chelators that allow iron to catalyse the production of OH. include DTPA, EDTA, ATP, ADP, citrate, GTP, 2,3-DHB, picolinic acid, 2,4 dipyridyl, oxalic acid, pyrophosphate, and DNA itself. Many of the above results were shown by multiple detection methods in order to avoid uncertainties in the interpretation of results.

It was also shown that some of our predictions about the in vivo behavior of dyes can be verified. In vitro results showed that reductants were used up by reactions with the dyes; in vivo, levels of reduced glutathione were depleted. In vivo studies presented in this thesis also add to the information we have on the mechanisms of dye-mediated lethality. While previous studies showed that preinduction of the defensive enzymes SOD and catalase protected against killing by photosensitizing dyes, this study adds specificity to that information by showing that increased intracellular levels of catalase alone or endonuclease IV alone also lend protection. Catalase plasmid protection demonstrates that peroxide is an important mediator in dye lethality, while endonuclease IV protection shows that DNA is an extremely important target in dye-mediated lethality since endonuclease IV has no other enzymatic function than the repair of damaged DNA. Protection against lethality by hydroxyl radical scavengers and reduction in strand scission in vivo in the presence of scavengers both show that OH. is an important mediator of lethality in vivo. All of the factors that influence OH. production & removal similarly affect the ability of dyes to damage plasmid DNA. Moreover, the in vitro results on SOD protection suggest a reason for the fact that SOD plasmids are not very effective at conferring protection in vivo. Thus, the in vitro and in vivo results are gratifyingly consistent.

Another finding concerns E. coli B's greater sensitivity to phototoxicity and free radical toxicity in comparison with E. coli K. The increased sensitivity may be due to a deficiency in repair enzymes. This idea had originally been suggested by the fact that a K12 strain mutant in exonuclease III was similar to E. coli B in sensitivity.

The fact that E. coli B is more highly sensitive than E. coli K12 to kill by photodynamic action and to kill by the O2- generating redox active herbicide paraquat29 further suggests that a free radical generating mechanism underlies both effects and that targets of damage are similar. DNA damage by paraquat and paraquat's induction of the SOS response have been demonstrated by Brawn and Fridovich.131 Potential future experiments include placing the nfo plasmid in E. coli strain xthA to determine whether endonuclease IV confers resistance to dyes in an exonuclease III deficient strain, as both function as AP endonucleases and it is thought that it is this function that is important in repairing damage by oxygen radicals.126

Oxygen free radical generation results from a variety of different sources, including the dye-mediated photodynamic effect as well as others. The natural defenses of the cell against these active oxygen species include the front-line defenses of the enzymes superoxide dismutase and the hydroperoxidases; in case of failure of these enzymes to prevent oxygen free radicals from attacking cellular components, photodynamically-damaged DNA can be repaired by nucleases which remove damaged single-stranded sections in order to allow replication of undamaged sequences. Different sources of damage-radiation, hydrogen peroxide, near-ultraviolet light, and dye-sensitized visible light-are defended against by the same systems, suggesting that they all involve similar mechanisms & similar targets. The protection of catalase against photodynamic DNA damage and killing found in this study parallels the fact that near ultraviolet damage is protected against by catalase.115 The sensitivity of the xthA strain, mutant in exonuclease III, to photodynamic kill compares to its sensitivity to kill by hydrogen peroxide

and by near uv.

Thus, all light- and oxygen-dependent lethal effects in bacteria may have a common mechanistic basis involving free radical chemistry.

A number of agents are noted for their ability to generate oxygen free radicals, among them the antitumor antibiotic adriamycin, bleomycin, streptonigrin, the diabetogenic agent alloxan, and the herbicides paraquat and diquat. It appears from this study, comparing oxygen radical yields of these substances with those of the photosensitizing dyes in the classes including the xanthenes, acridines, thiazines, and phenazines, that these dyes may be even more potent producers of these highly toxic radicals. All of the requirements outlined in this thesis are met by the contents of normal cells, and are therefore of in vivo significance; all, then, that should be required in order for these reactions to take place in vivo is a supply of a dye that functions as the dyes considered in this investigation do. If the dyes used in cosmetics and foods are capable of entering cells, damage to DNA may occur, possibly leading to mutation or cancer if cellular defenses are inadequate. Whether this damage will occur in people for exposure to a given dye depends only on the accessibility of dyes to the interior of cells and the presence of the various components required within a single cellular compartment. The fact that toxicity has been seen for dyes including neutral red177 and rose bengal166 in humans indicates that these requirements are probably met, and suggests caution in the current use of related dyes in foods, drugs and cosmetics.

The major finding of this investigation, that the hydroxyl radical rather than singlet oxygen is a major mediator of photodynamic toxicity, suggests that a large body of photodynamic-related research, performed in the years before the elucidation of biologically relevant reactions that produce OH., may require some reinterpretation.