Prooxidant Activity of Ferrioxamine in Isolated Rat Hepatocytes and

desferrioxamine:iron ratio, was added to isolated rat hepatocytes and to linoleic ... free desferrioxamine, ferrioxamine can participate in redox cycl...
0 downloads 0 Views 102KB Size
Chem. Res. Toxicol. 1999, 12, 365-370

365

Prooxidant Activity of Ferrioxamine in Isolated Rat Hepatocytes and Linoleic Acid Micelles Stefania Bergamini, Cristina Rota, Mariagrazia Staffieri, Aldo Tomasi, and Anna Iannone* Department of Biomedical Sciences, University of Modena, 41100 Modena, Italy Received June 19, 1998

The complex iron-desferrioxamine (ferrioxamine) is considered chemically unreactive, and not able to participate in redox cycle reactions. Desferrioxamine-dependent toxicity is, however, described in both human and animal studies. The aim of this work was to test the possibility that chelated iron, under certain circumstances, could enter redox reactions, giving an explanation of desferrioxamine side effects. Carefully prepared ferrioxamine, to obtain a 1:1 desferrioxamine:iron ratio, was added to isolated rat hepatocytes and to linoleic acid micelles. A strong prooxidant and cytotoxic effect was observed in the cells, also potentiating tert-butyl hydroperoxide-induced lipid peroxidation. In micelles, the prooxidant effect was observed only in the presence of ascorbate, which is oxidized during the process, giving rise to ascorbyl radical. Ferrioxamine, under the experimental conditions used, did not release iron, indicating that the prooxidant effect was due to iron redox cycling. The addition of desferrioxamine prevented both ferrioxamine- and tert-butyl hydroperoxide-induced lipid peroxidation and cytotoxicity. Concurrently, a nitroxide radical was detected, an indication of the radical scavenger activity of the hydroxamic moiety. No radical species was observed when ferrioxamine was added to the same system. The prooxidant effect of ferrioxamine gives a possible explanation of the reported human and animal desferrioxamine toxicity. When, in compartmentalized regions, the ratio of desferrioxamine:metal reaches 1:1, ferrioxamine is formed. In the absence of metalfree desferrioxamine, ferrioxamine can participate in redox cycling reactions, initiating lipid peroxidation and cytotoxicity.

Introduction Iron bound to desferrioxamine (i.e., ferrioxamine) is commonly considered to be unreactive in vivo. Desferrioxamine (DFO1), a bacterial siderophore obtained from the Streptomyces pilosus (1, 2), is highly specific for Fe3+ (KS of 1031). It also reacts with ferrous ion, although the complexation involves an initial oxidation to the ferric state. The complex formed when ferrous ion reacts with DFO appears to be the same as when ferric ions react (3). Both reactions are pH-dependent, since the formation of ferrioxamine (FOA) from the reaction of DFO with Fe2+ is faster at pH 7, and the reaction of DFO with Fe3+ is faster at acidic pH. DFO binds iron in a 1:1 molar ratio (hexadentate chelator), and it is generally accepted that it chelates iron with sufficient avidity to render it chemically unreactive (i.e., unable to participate in redox cycling). In iron chelates with citrate, ADP, ATP, and GTP, at least one of the six ligands of iron is left free to maintain the catalyzing activity. It was shown that as fewer ligands are involved in chelation, the catalyzing activity becomes higher (4). Even though DFO is widely used for the prevention and treatment of iron overload in humans (5, 6), in * To whom correspondence should be addressed: Department of Biomedical Sciences, University of Modena, via Campi 287, 41100 Modena, Italy. Phone: ++39-59-428633. Fax: ++39-59-428623. Email: [email protected]. 1 Abbreviations: DFO, desferrioxamine; FOA, ferrioxamine; EDTA, ethylenediaminetetraacetic acid; EPR, electron paramagnetic resonance; LA, linoleic acid; MDA, malondialdehyde; TBA, thiobarbituric acid; t-buOOH, tert-butyl hydroperoxide; TCA, trichloroacetic acid.

laboratory animals it has been shown to aggravate the acute inflammatory response (7, 8), and enhances the toxicity of alloxan (9) and paraquat (10). In humans, DFO has been shown to induce neurotoxic effects (11) and retinopathy (12). An explanation for these side effects could be the presence of a prooxidant effect of DFO, which has been suggested by several authors, with the production of hydroxyl radicals derived from the reaction of this chelator with Fe2+ (13-15). On the other hand, in experimental model systems, DFO is often used to inhibit the lipid peroxidation process and to confirm the role of iron in lipid peroxidation systems (16), since it inhibits iron catalysis of the HaberWeiss reaction. Recently, it has been demonstrated that DFO can act as a chain-breaking antioxidant in biological membranes, independent of its iron chelating properties (16-18). Two mechanisms contribute to the antioxidant effect of DFO: (i) the inhibition of free radical formation by the chelation of redox active iron and (ii) a radical scavenger activity by which DFO neutralizes free radicals donating hydrogen atoms; the resulting nitroxide radical is poorly reactive since it is stabilized by resonance. The products of DFO reaction with iron and with free radicals are represented in Scheme 1. The effects of DFO therapy could be considered as a balance between antioxidant and prooxidant effects. The DFO molecule can work as a radical scavenger or as a chelator subtracting potentially reactive iron. However, this latter reaction, involving ferrous ion, can produce reactive species (i.e., hydroxyl radical).

10.1021/tx980149c CCC: $18.00 © 1999 American Chemical Society Published on Web 03/13/1999

366 Chem. Res. Toxicol., Vol. 12, No. 4, 1999

Bergamini et al. Scheme 1

An additional aspect has to be considered: the possibility that once the iron is chelated by DFO, the product (FOA), under certain circumstances, could promote a redox cycling of the metal. This could explain the report by Osheroff (10) who concluded that DFO increases the lethality of paraquat, since the paraquat radical cation could reduce FOA to an Fe2+ complex. In addition, a prooxidant effect of FOA, with the production of hydroxyl radicals, has been hypothesized by Borg and Schaich (19). On the other hand, the fact that Streptomyces pilosus removes iron from the FOA, which possibly acts as an iron donor in the incorporation of iron in the porphyrins (20), should serve as a caution regarding the notion that iron bound to DFO can no longer enter into redox or other biochemical reactions. A lack of information exists about FOA effects in biological systems. However, it is known that in animals it is more toxic than DFO itself and ferric chelates of EDTA or diethylenetriaminepentaacetic acid (DTPA) (21). In this work, we investigated the possibility that iron from the FOA stimulates the iron-dependent lipid peroxidation in biological systems, comparing it to the effects of DFO itself. The effect of this compound has been tested using isolated liver hepatocytes and linoleic acid (LA) micelles as experimental model systems.

Experimental Procedures Materials. KH2PO4 and Na2HPO4, used for the preparation of phosphate buffer, ethanol, ferric chloride, and ferrous sulfate were supplied by BDH Italia S.r.l. (Milan, Italy). Chelex-100 resin was from Bio-Rad Laboratories (Hercules, CA). Collagenase, tris(hydroxymethyl)methylamine, tert-butyl hydroperoxide (t-buOOH), a 0.4% trypan blue solution, and linoleic acid were purchased from Sigma-Aldrich S.r.l. (Milan, Italy). Desferrioxamine (Desferal) was purchased from Ciba-Geigy S.p.A. (Varese, Italy). Thiobarbituric acid (TBA), L-ascorbic acid, and malondialdheyde-bis(dimethylacetal) were purchased from Merck (Darmstadt, Germany). All buffer solutions were made up in deionized water, and pH adjustments were made with either HCl or NaOH. The Fe2+

and Fe3+ solutions were prepared as described by Yegorov (22) in 0.2 mM HCl and used within 5 h of preparation. DFO was dissolved in water, or Tris or phosphate buffer, depending on the sample composition. FOA was made up by mixing an equimolar concentration of a water solution of DFO and an Fe3+ solution prepared in 0.2 mM HCl. The t-buOOH solution was prepared in medium C (for the composition, see Isolation of Hepatocytes); the L-ascorbate was prepared in medium C or Tris buffer. Isolation of Hepatocytes. Adult male Wistar rats (200250 g) were fed a standard laboratory diet and water ad libitum. Hepatocytes were isolated using a method previously described (23), where the final medium (medium C) for suspending the isolated hepatocytes was modified and Chelex-treated. Medium C was composed of 85.5 mM NaCl, 20 mM KCl, 50 mM Hepes/ NaOH, 1 mM CaCl2, 2 mM MgSO4, 0.8 mM Na2HPO4, and 10 mM glucose. The pH of this solution was adjusted to 7.4 and then filtered through a Chelex-100 column. The first 50 mL was discarded, since it is known that the prewashing procedure is necessary to avoid artifacts due to leaching material from the resin (24). The pH was checked at the end of this procedure, since a slight modification was always observed. Cells were suspended in medium C to obtain a final concentration of 16 × 106 cells/mL. Viability was measured by the Trypan blue exclusion test (25), and only cell preparations with a viability of >85% were used for the experiments. Micelle Preparation. Micelles were made by using a Thermobarrel Extruder (Lipex Biomembranes Inc., Vancouver, BC). Linoleic acid (2 mg/mL) was suspended by simple vortexing in medium C. This suspension was then processed through the extruder to obtain micelles. Biochemical Assay. The extent of lipid peroxidation was measured by the thiobarbituric acid (TBA) test. Rat hepatocytes were incubated at 37 °C in a shaking water bath in the presence of t-buOOH alone or with DFO or FOA, at the concentrations indicated in the tables and figures. Aliquots of the hepatocyte suspension (0.7 mL) were removed at regular intervals, and an equivalent amount of 20% trichloroacetic acid (TCA) was added. Samples were centrifuged at 2000g for 10 min to remove the precipitate; 1 mL of clear supernatant was added to 1 mL of 0.67% TBA. The solution was heated for 10 min in a boiling water bath and then cooled in ice-cold water. The malondialdehyde (MDA) content in samples was measured at 535 nm, by using a standard calibration curve of malondialdehyde-bis(dimethylacetal).

Prooxidant Activity of Ferrioxamine

Figure 1. Calibration plot of the DFO-Fe3+ complex. The curve was obtained by adding known amount of ferric chloride to 1 mM DFO in water and following the formation of the ferrioxamine (values plotted along the x-axis as micrograms of ferric ions) at 430 nm. In the experiments where the linoleic acid (LA) micelles were used, the extent of MDA formation was measured with the spectrophotometer after processing the samples through an equivalent amount of exane to extract the lipids. The pink aqueous phase, containing the MDA, was then separated from the exane by centrifugation at 2000g for 15 min. EPR Measurements. Experiments performed with LA micelles were carried out in medium C, and the sample was composed of LA (1.6 mg/mL), 4 mM L-ascorbate, and 0.8 mM FOA or 0.8 mM DFO. Samples for EPR measurements in experiments with cells were composed of 5 × 106 hepatocytes, 1 mM t-buOOH, and 1 mM DFO or FOA. The samples were incubated for 1 min and drawn into a gaspermeable Teflon tube (Zeus Industrial Products, Raritan, NJ) and the tubes placed in the EPR cavity with a flow of air at 37 °C. The EPR spectra were recorded on a Bruker ESP 300E spectrometer with the following settings: incident microwave power of 10 mW, modulation amplitude of 1 G, a time constant of 41 ms, a sweep time of 84 s, and a scan width of 50 G. A standard Bruker flow dewar was used for temperature regulation in the EPR cavity. Optical Techniques. Special care was used in preparing the FOA solution, since many factors interfere with the reaction between DFO and iron in forming FOA. The formation of the colored complex Fe3+-DFO at 430 nm was detected on a HP8452A spectrophotometer. FOA was made up using either Fe3+ or Fe2+ (0.2 mM) and DFO at an equimolar concentration. Reactions were carried out in water, Tris, or phosphate buffer, at different pHs. The mixture of all the reagents but DFO was used as a blank. To determine the absolute Fe3+ concentration in the samples, a calibration curve was developed in deionized water (Figure 1). Detection of Iron Release from FOA. The effect of ascorbate on iron release from FOA was tested following absorbance changes at 430 nm. Reactions were performed in the presence of LA micelles (prepared in medium C) or 50 mM Tris buffer (pH 7.4). In micelles, the FOA and ascorbate concentrations were the same as for MDA measurements; in Tris buffer, the ascorbate concentration was 2 mM and that of FOA 0.4 mM. The FOA absorbance spectrum was monitored up to 2 h in experiments performed with micelles and 24 h in Tris buffer. Statistical Analysis. The data are expressed as mean ( SE. Significance was estimated using the Student’s two-tailed t test; the criterion for significance was p < 0.01.

Results The effect of FOA was tested in two biological model systems (cells and micelles), where the antioxidant effect of DFO was also demonstrated by using an organic

Chem. Res. Toxicol., Vol. 12, No. 4, 1999 367

Figure 2. Effect of pH and phosphate on ferrioxamine formation. The extent of formation over time of the colored complex DFO-Fe3+ at 430 nm was calculated using the calibration curve shown in Figure 1 and expressed as micrograms of ferric iron. Reaction mixtures were made in the presence of 0.1 M Tris or phosphate buffer, whose pHs were changed as indicated in the figure, and compared to water.

hydroperoxide (t-buOOH) to induce the lipid peroxidation process. Preliminary experiments were performed to obtain authentic FOA to be used in our experiments, having checked the conditions of complete saturation of DFO with Fe3+. This is an important issue since (i) unsaturated DFO in the mixture acts as an antioxidant and (ii) free iron could, on the other hand, lead to misleading results. Formation of FOA. The formation of FOA was followed at 430 nm for up to 20 min under different experimental conditions. Reaction mixtures were made up at room temperature, in water, or 0.1 M Tris or phosphate buffer (pH as indicated in Figure 2). In deionized water, and in Tris buffer at pH 5.5, the reaction of DFO with Fe3+ was very fast and quickly reached a steady state (Figure 2). Tris buffer at pH 7.4 significantly inhibits this reaction, indicating a pH-dependent mechanism. The reaction was also inhibited by phosphate, since phosphate buffer at pH 7.4 or 5.5 significantly inhibits FOA formation (Figure 2). In the presence of ferrous ion, FOA formation was not inhibited by phosphate and the highest degree of complexation was readily achieved at pH 7.4 (data not shown). The FOA used in the experimental model systems described in this work was prepared by mixing an equimolar concentration of DFO dissolved in water and FeCl3 dissolved in 0.2 mM HCl. Experiments in Cells. Isolated rat hepatocytes were incubated for 1 h at 37 °C with FeCl3 (0.2 mM). As expected, this treatment induced lipid peroxidation, as measured by MDA detection (Table 1). FOA addition caused a significant increase in the extent of lipid peroxidation, higher than that observed in the presence of iron alone. The further addition of 0.2 mM DFO completely inhibited the FOA-induced peroxidation. DFO was also highly effective in protecting hepatocytes in a time course experiment, where the well-known organic hydroperoxide t-buOOH was employed to induce lipid peroxidation (Figure 3). The addition of FOA to t-buOOH-treated hepatocytes caused, instead, a further increase in MDA levels, compared to those of the cells treated with the hydroperoxide alone (Figure 3). DFO protection was further confirmed in a 3 h time course experiment where cell viability was measured by the trypan blue exclusion test (Figure 4). FOA, as above,

368 Chem. Res. Toxicol., Vol. 12, No. 4, 1999

Bergamini et al.

Table 1. Antioxidant Effect of Desferrioxamine and Prooxidant Effect of Ferrioxamine in Isolated Rat Hepatocytesa incubation conditions

nmol of MDA/1E6 cells

control FeCl3 ferrioxamine desferrioxamine ferrioxamine and desferrioxamine

0.107 ( 0.035 (13) 0.733 ( 0.192b,c (3) 2.347 ( 0.142b (4) 0.050 ( 0.017 (15) 0.041 ( 0.030c (3)

a Freshly isolated hepatocytes were incubated at 35 °C for 1 h. Desferrioxamine, ferrioxamine, and FeCl3 were present at final concentrations of 0.2 mM. The values are means ( SE (number of experiments). b P < 0.001 with respect to that of control. c P < 0.001 with respect to that of ferrioxamine.

Figure 4. Effect of desferrioxamine and ferrioxamine on hydroperoxide cytotoxicity. Hepatocyte viability was measured by the trypan blue exclusion test. The composition of the samples was identical to that described in the legend of Figure 3. Values are the mean of four to six separate experiments.

Figure 3. Effect of desferrioxamine and ferrioxamine on hydroperoxide-induced lipid peroxidation. Hepatocytes (5 × 106 cells/mL) were incubated at 37 °C in the presence of 1 mM t-buOOH with or without 0.2 mM DFO or FOA. Values represent the mean of four or five separate experiments.

not only did not protect but also actually caused an increase in the number of trypan blue positive cells with respect to the control (Figure 4). The effect on cell viability was not different when t-buOOH-treated hepatocytes were compared with t-buOOH- and FOA-treated hepatocytes. The mechanism of antioxidant action was then studied by EPR. When hepatocytes were incubated for 1 min at room temperature in the presence of 1 mM t-buOOH and 1 mM DFO, the nitroxide radical described in Scheme 1 was detected (Figure 5A). Unequivocally, the EPR signal was a triplet of triplets, with hyperfine coupling constants consistent with that described for the DFO nitroxide radical (26-28). The DFO radical was not detected when hepatocytes were incubated in the presence of FOA (Figure 5B), giving a clear indication that the antioxidant action of the molecule depends on the availability of the hydroxamic groups, which are unavailable when iron is chelated. Experiments in Micelles. LA micelles were used to better define the prooxidant action of FOA. The addition of FOA to LA micelles did not initiate the peroxidation

Figure 5. EPR detection of the desferrioxamine nitroxide radical. Spectrum A was observed in hepatocytes (5 × 106 cells/ mL) treated with 1 mM desferrioxamine and 1 mM t-buOOH. When desferrioxamine was substituted with 1 mM ferrioxamine, no EPR spectrum was detected.

process as measured by the extent of MDA formation. The simultaneous addition of 0.8 mM FOA and 4 mM ascorbate caused lipid peroxidation (Table 2). To avoid artifacts due to trace iron, possibly not bound to DFO, FOA was also prepared by using an excess of DFO (1:0.9 DFO:Fe). Even in this case, the prooxidant effect of FOA was demonstrated (Table 2). The further addition of DFO to the FOA-ascorbate system led to a significant antioxidant effect, enhanced by increasing DFO concentrations. The prooxidant effect of ascorbate-FOA was less efficient than that induced by FeCl3-ascorbate (Table 2). Control experiments were performed by incubating micelles in the presence of DFO, FeCl3, or ascorbate (alone or along with DFO); the MDA production was not different from control (data not shown). FOA-Ascorbate Interaction. Ascorbate addition to micelles was necessary for FOA-dependent lipid peroxidation. To rule out the possibility of iron release promoted by ascorbate, FOA absorption at 430 nm was measured in either micelles or Tris buffer in the presence of

Prooxidant Activity of Ferrioxamine

Chem. Res. Toxicol., Vol. 12, No. 4, 1999 369

Table 2. Antioxidant Effect of Desferrioxamine and Prooxidant Effect of Ferrioxamine in Micellesa incubation conditions

nmol of MDA/mL

control ferrioxamine ascorbate and ferrioxamine (1:0.9 desferrioxamine:Fe) ascorbate and ferrioxamine ascorbate, ferrioxamine, and 0.4 mM desferrioxamine ascorbate, ferrioxamine, and 0.8 mM desferrioxamine FeCl3/ascorbate

0.296 ( 0.092 (7) 0.302 ( 0.248 (8) 3.501 ( 0.223b (4) 3.629 ( 0.347b (11) 1.644 ( 0.318b,c (7) 1.446 ( 0.317b,c (7) 11.538 ( 1.079b (7)

a Micelles were incubated at 37 °C for 2 h. FOA (1:1 DFO:Fe), DFO, and FeCl3 were present at final concentrations of 0.8 mM, and the final ascorbate concentration was 4 mM. MDA production in micelles incubated in the presence of DFO, FeCl3, or ascorbate (alone or with DFO) was not different from control (data not shown). The values are means ( SE (number of experiments).b P < 0.001 with respect to that of control. c P < 0.001 with respect to that of ascorbate and FOA.

Figure 6. EPR detection of the ascorbyl radical in LA micelles. Spectrum A was observed in LA micelles with ascorbate. Spectrum B shows the ascorbyl radical detected when ascorbate was incubated with 0.8 M FOA and LA micelles. Spectrum C shows the signal detected when DFO was added instead of FOA.

ascorbate. These experiments showed that the total amount of FOA remained unchanged up to 2 h in micelles and 24 h in Tris buffer (data not shown), indicating that lipid peroxidation in our experimental systems was not induced by an ascorbate-dependent iron release from FOA. Further experiments were carried out using EPR spectroscopy to study the interaction between ascorbate and FOA. When micelles were incubated at 37 °C under aerobic conditions in the presence of L-ascorbate and 0.8 mM FOA, a doublet with a spectral feature typical of the ascorbyl radical (g value of 2.005 and hyperfine splitting constant of 1.75 G) was detected (Figure 6B). In the absence of FOA, a small concentration of ascorbyl radical was also detected (Figure 6A). Reactions performed in the presence of DFO and ascorbate did not cause any significant change in the ascorbyl radical signal (Figure 6C).

Discussion On the basis of results from the experiments performed in different buffers and at different pHs, great care was taken to obtain a fully saturated DFO. Also, a small concentration of unreacted DFO, as that observed in the

case of the formation of FOA in phosphate buffer or at pH 7.4 (Figure 2), could affect the outcome of the experiments designed to evaluate FOA-dependent activity. DFO hydroxamic groups, not bound to iron, confer a radical scavenging activity to the molecule. In our experiments, when the chelating agent was added to an oxidizing model system (hepatocytes and t-buOOH), cell peroxidation was inhibited, viability improved, and a nitroxide radical, evidence of an electron transfer reaction, detected. It is known that the nitroxide radical formation results in a lipid chain-breaking antioxidant activity (18, 29). On the other hand, DFO nitroxide could also act as a damaging species, rapidly reacting with ascorbate, tocopherol, and sulfhydryl groups, resulting in protein damage (27, 30). EPR experiments clarified, also, that the observed radical, a carbonyl-conjugated nitroxide [CH2-N(O•)CO], was not compatible with the simultaneous Fe3+ coordination. In fact, no such radical was observed when, under the same experimental conditions, carefully prepared FOA was added to isolated cells instead of DFO (see Scheme 1 and Figure 5). Various chelating agents may promote lipid peroxidation, depending on the ratio of chelator:iron ion (16). The addition of ferrous ion to membrane phospholipids can either promote or inhibit the peroxidation in the presence of chelators, such as EDTA, diethylenetriaminepentaacetic acid, and bathophenanthrolinesulfonate. However, it is common to assume that a hexadentate chelator such as DFO binds iron with sufficient avidity to render it chemically unreactive within the range of physiological redox potentials (16, 31, 32). Nonetheless, cyclic voltammetry of FOA demonstrates a reversible oneelectron reduction at ca. -0.45 V (33), which is within the range of a number of reducing enzymes (34) and may well give the molecular basis of a potential FOA prooxidant activity. This possibility was tested by the addition of carefully iron-saturated DFO to cells and micelles, which resulted in an intense prooxidant activity, as measured by the extent of MDA production. This activity was mediated by a reducing agent, since ascorbate was necessary to induce lipid peroxidation in micelles. Two possible prooxidant mechanisms could be involved: ferrous ion release or redox cycling of the FOA-linked metal. The latter is more likely since no evidence of iron release was collected in experiments where ascorbate was added to FOA; the absorption spectrum of this latter did not exhibit decreases in the presence of ascorbate. Moreover, the increase of the ascorbyl radical signal observed in the reaction of LA micelles with FOA and ascorbate clearly indicates that ascorbate one-electron oxidation is taking place; this reaction is likely coupled to the reduction of FOA-linked Fe3+ to Fe2+. At physiological pH, Fe2+ is easily oxidized back to Fe3+ (3), and iron redox cycling can act as an oxygen radical-generating system (13). The possibility of redox cycling is clearly dependent on the ratio of metal:chelator, as reported by Buettner (35) and Halliwell (36). In the case of DFO:iron, a 1:1 ratio is required to observe the oxidizing activity; any higher ratio confers the usual antioxidant activity, due both to the radical scavenging activity and to iron chelation. In this study, we have demonstrated that hepatocytes render FOA (1:1 DFO:iron) redox active, conferring a substantial prooxidant activity to the molecule and

370 Chem. Res. Toxicol., Vol. 12, No. 4, 1999

increasing the extent of hydroperoxide-induced lipid peroxidation, which is at its turn iron-dependent (37, 38). The prooxidant effect of FOA certainly suggests that DFO may increase the level of oxidative damage in vivo, through the likely formation of FOA, which undergoes redox cycling, causing lipid peroxidation, cytotoxicity, and toxicity to the animal. Our results give a possible explanation of the reported animal and human DFO toxicity in vivo (7-11), where acute maculopathy and phlebitis (12, 39, 40) are reported. Typical doses of DFO in human therapy should give a high drug:metal ratio; however, within compartmentalized regions, the DFO: metal ratio may reach 1:1, giving toxic FOA-dependent effects.

Acknowledgment. This work was supported by CNR Grant 96.05002.ST74.

References (1) Bickel, H., Gaumann, E., Keller-Schierlein, W., Prelog, V., Vischer, E., Wettstein, A., and Zahner, H. (1960) Uber eisenhaltige wachstumsfaktoren, die sideramine, und ihre antagonisten, die einsehaltigen antibiotika sideromycine. Experientia 16, 128-168. (2) Bickel, H., Hall, G. E., Keller-Schierlein, W., Prelog, V., Vischer, E., and Wettstein, A. (1960) Uber die kontituton von ferrioxamin B. Helv. Chim. Acta 43, 2129-2138. (3) Goodwin, J. F., and Whitten, C. F. (1965) Chelation of ferrous sulphate solutions by desferrioxamine B. Nature 205, 281-283. (4) Porter, J. B. (1996) Evaluation of new iron chelators for clinical use. Acta Haematol. 95, 13-25. (5) Modell, B., Letsky, E. A., Flinn, D. M., Peto, R., and Weatherall, D. J. (1982) Survival and desferrioxamine in thalassaemia major. Br. Med. J. 284, 1081-1084. (6) McLaren, G. D., Muir, W. A., and Kellermeyer, R. W. (1983) Iron overload disorders: natural history, pathogenesis, diagnosis and therapy. CRC Crit. Rev. Clin. Lab. Sci. 19, 205-265. (7) Yoshino, S., Blake, D. R., and Bacon, P. A. (1984) The effect of desferrioxamine on antigen-induced inflammation in the rat air pouch. J. Pharm. Pharmacol. 36, 543-545. (8) Blake, D. R., Hall, N. D., Bacon, P. A., Dieppe, P. A., Halliwell, B., and Gutteridge, J. M. P. (1983) Effect of a specific iron chelating agent on animal models of inflammation. Ann. Rheum. Dis. 42, 89-93. (9) Grankvist, K., and Marklund, S. L. (1983) Opposite effects of two metal chelators on alloxan-induced diabetes in mice. Life Sci. 33, 2535-2540. (10) Osheroff, M. R., Schaich, K. M., Drew, R. T., and Borg, D. C. (1985) Failure of desferrioxamine to modify the toxicity of paraquat in rats. J. Free Radical Biol. Med. 1, 71-82. (11) Blake, D. R., Winyard, P., Lunec, J., Williams, A., Good, P. A., Graves, S. J., Gutteridge, J. M. C., Rowley, D. A., Halliwell, B., Cornish, A., and Hidler, R. C. (1985) Cerebral and ocular toxicity induced by desferrioxamine. Q. J. Med. 219, 345-355. (12) Rahi, A. H., Hungerford, J. L., and Ahmed, A. I. (1986) Ocular toxicity of desferrioxamine: light microscopic histochemical and ultrastructural findings. Br. J. Ophthalmol. 70, 373-381. (13) Klebanoff, S. J., Waltersdorph, A. M., Michel, B. R., and Rosen, H. (1989) Oxygen-based free radical generation by ferrous ions and deferoxamine. J. Biol. Chem. 264, 19765-19771. (14) Miller, D. M., Spear, N. H., and Aust, S. D. (1992) Effects of deferrioxamine on iron-catalized lipid peroxidation. Arch. Biochem. Biophys. 295, 240-246. (15) Braughler, J. M., Chase, R. L., and Pregenzer, J. F. (1986) Stimulation and inhibition of iron-dependent lipid peroxidation by desferrioxamine. Biochem. Biophys. Res. Commun. 153, 933938. (16) Gutteridge, J. M. C., Richmond, R., and Halliwell, B. (1979) Inhibition of the iron-catalyzed formation of hydroxyl radicals from superoxide and of lipid peroxidation by desferrioxamine. Biochem. J. 184, 469-472. (17) Paya’, M., Halliwell, B., and Hoult, J. R. S. (1992) Peroxyl radical scavenging by a series of coumarins. Free Radical Res. Commun. 17, 293-298. (18) Hartley, A., Davies, M., and Rice-Evans, C. (1990) Desferrioxamine as a lipid chain-breaking antioxidant in sickle erythrocyte membranes. FEBS Lett. 264, 145-148.

Bergamini et al. (19) Borg, D. C., and Schaich, K. M. (1986) Prooxidant action of desferrioxamine: Fenton-like production of hydroxyl radicals by reduced ferrioxamine. J. Free Radicals Biol. Med. 2, 237-243. (20) Keberle, H. (1964) The biochemistry of desferrioxamine and its relation to iron metabolism. Ann. N.Y. Acad. Sci. 119, 758-768. (21) Catsch, A., and Harmuth-Hoene, A. E. (1978) Pharmacology and therapeutic applications of agents used in heavy metal poisoning. In The chelation of heavy metals (Levine, W. G., Ed.) pp 107239, Pergamon Press, Oxford, U.K. (22) Yegorov, D. Y., Kozlov, A. V., Azizova, O. A., and Vladimirov, Y. A. (1993) Simultaneous determination of Fe(III) and Fe(II) in water solutions and tissue homogenates using desferal and 1,10phenanthroline. Free Radical Biol. Med. 15, 565-574. (23) Iannone, A., Hu, H., Tomasi, A., Vannini, V., and Swartz, H. M. (1989) Metabolism of aqueous soluble nitroxides in hepatocytes: effects of cell integrity, oxygen, and structure of nitroxides. Biochim. Biophys. Acta 991, 90-96. (24) Van Reyk, D. M., Brown, A. J., Jessup, W., and Dean, R. T. (1995) Batch-to-batch variation of chelex-100 confounds metal-catalysed oxidation, leaching of inhibitory compounds from a batch of chelex-100 and their removal by a pre-washing procedure. Free Radical Res. 23, 533-535. (25) Jeejeebhoy, K. N., Greenberg, G. R., Phillips, M. J., BruceRobertson, A., and Sodtke, U. (1975) Albumin, fibrinogen and transferrin synthesis in isolated rat hepatocyte suspensions. A model for the study of plasma protein synthesis. Biochem. J. 146, 141-155. (26) Morehouse, K. M., Flitter, W. D., and Mason, R. P. (1987) The enzymatic oxidation of desferal to a nitroxide free radical. FEBS Lett. 222, 246-250. (27) Davies, M. J., Donkor, R., Dunster, C. A., Gee, C. A., Jonas, S., and Willson, R. L. (1987) Desferrioxamine (Desferal) and superoxide free radicals. Formation of an enzyme-damaging nitroxide. Biochem. J. 246, 725-729. (28) Soriani, M., Mazzuca, S., Quaresima, V., and Minetti, M. (1993) Oxidation of desferrioxamine to nitroxide free radical by activated human neutrophils. J. Free Radical Biol. Med. 14, 589-599. (29) Denicola, A., Souza, J. M., Gatti, R. M., Augusto, O., and Radi, R. (1995) Desferrioxamine inhibition of the hydroxyl radical-like reactivity of peroxynitrite: role of the hydroxamic groups. Free Radical Biol. Med. 19, 11-19. (30) Mordente, A., Meucci, E., Miggiano, G. A. D., and Martorana, G. E. (1990) Prooxidant action of desferrioxamine: enhancement of alkaline phosphatase inactivation by interaction with ascorbate system. Arch. Biochem. Biophys. 277, 234-240. (31) Cederbaum, A. I., and Dicker, E. (1983) Inhibition of microsomal oxidation of alcohols and of hydroxyl-radical-scavenging agents by the iron chelating agent desferrioxamine. Biochem. J. 210, 107-113. (32) Starke, P. E., and Farber, J. L. (1985) Ferric iron and superoxide ions are required for the killing of cultured hepatocytes by hydrogen peroxide. J. Biol. Chem. 260, 10099-10104. (33) Cooper, S. R., McArdle, J. V., and Raymond, K. N. (1978) Siderophore electrochemistry: relation to intracellular iron release mechanism. Proc. Natl. Acad. Sci. U.S.A. 75, 3551-3554. (34) Borg, D. C., Schaich, K. M., and Forman, A. (1984) Autoxidative cytotoxicity: is there metal-dependent formation of hydroxyl radicals? Are there “crypto-hydroxyl” radicals? In Oxygen Radicals in Chemistry and Biology (Bors, M. S. W., and Tait, D., Eds.) pp 123-130, Walter de Gruyter, Berlin. (35) Buettner, G. R. (1986) Ascorbate autoxidation in the presence of iron and copper chelates. Free Radical Res. Commun. 1, 349353. (36) Halliwell, B. (1989) Protection aginst tissue damage in vivo by desferrioxamine: what is its mechanism of action? Free Radical Biol. Med. 7, 645-651. (37) Hogberg, J., Orrenius, S., and O’Brien, P. (1975) Further studies on lipid-peroxide formation in isolated hepatocytes. Eur. J. Biochem. 59, 449-455. (38) Thornalley, P. J., Trotta, R. J., and Stern, A. (1983) Free radical involvement in the oxidative phenomena induced by tert-butyl hydroperoxide in erythrocytes. Biochim. Biophys. Acta 759, 1622. (39) Good, P. A., Claxson, A., Morris, C. J., and Blake, D. R. (1990) A model for desferrioxamine-induced retinopathy using the albino rat. Ophthalmologica 201, 32-36. (40) Dennerlein, J. A., Lang, G. E., Stahnke, K., Kleihauer, E., and Lang, G. K. (1995) Ocular findings in Desferal therapy. Ophthalmologe 92, 38-42.

TX980149C