Characterization of the Cycle of Iron-mediated Electron Transfer from Adriamycin to Molecular Oxygen *

The anticancer drug adriamycin binds iron and these complexes cycle to reduce molecular oxygen (Zweier, J. L. (1984) J. Biol. Chem. 259, 6056-6058). Optical absorption, EPR, and Mossbauer spectroscopic data are correlated with polarographic 0 2 consumption and chemical Fez+ extraction measurements in order to characterize each step in this cycle. Fe3+ binds to adriamycin at physiologic pH forming a complex with an optical absorbance maximum at 600 nm. EPR signals at g = 4.2 and g = 2.01, and a doublet Mossbauer spectrum with isomer shift 6 = 0.57 mm/s and quadrupole splitting AEQ = 0.74 mm/s are observed indicating that the Fe3+ bound to adriamycin is high spin S = 512. Under anaerobic conditions the absorbance maximum at 600 nm decreases with an exponential decay constant = 0.77 h-’, and the EPR and Mossbauer spectra of Fe3+-adriamycin similarly decrease as the Fe3+ is reduced to EPR silent Fez+. The Fez+-adriamycin complex which is formed exhibits a Mossbauer spectrum with 6 = 1.18 mm/s and AEQ = 1.82 mm/s indicative of high spin Fez+. As the EPR spectra of Fe3+adriamycin decrease on reduction of the Fe3+ to Fez+ a signal of the oxidized adriamycin free radical appears at g = 2.004 with line width of 8 G. On exposure to 0 2 the absorption maximum at 600 nm, the Fe3+ EPR, and the Fe3+ Mossbauer spectra all return. Polarographic measurements demonstrate that O2 is consumed and that HzOz is formed. Addition of high affinity Fez+ chelators block Oz consumption indicating that Fe2+ formation is essential for O2 reduction. This cycle of iron-mediated O2 reduction can explain the formation of the reactive reduced oxygen and adriamycin radicals which are thought to mediate the biological activity of adriamycin.

at g = 4.2 and g = 2.01, and a doublet Mossbauer spectrum with isomer shift 6 = 0.57 mm/s and quadrupole splitting AEQ = 0.74 mm/s are observed indicating that the Fe3+ bound to adriamycin is high spin S = 512. Under anaerobic conditions the absorbance maximum at 600 nm decreases with an exponential decay constant = 0.77 h-', and the EPR and Mossbauer spectra of Fe3+-adriamycin similarly decrease as the Fe3+ is reduced to EPR silent Fez+. The Fez+-adriamycin complex which is formed exhibits a Mossbauer spectrum with 6 = 1.18 mm/s and AEQ = 1.82 mm/s indicative of high spin Fez+. As the EPR spectra of Fe3+adriamycin decrease on reduction of the Fe3+ to Fez+ a signal of the oxidized adriamycin free radical appears at g = 2.004 with line width of 8 G. On exposure to 0 2 the absorption maximum at 600 nm, the Fe3+ EPR, and the Fe3+ Mossbauer spectra all return. Polarographic measurements demonstrate that O2 is consumed and that HzOz is formed. Addition of high affinity Fez+ chelators block Oz consumption indicating that Fe2+ formation is essential for O2 reduction. This cycle of iron-mediated O2 reduction can explain the formation of the reactive reduced oxygen and adriamycin radicals which are thought to mediate the biological activity of adriamycin.
The anthracycline antibiotic, adriamycin, is one of the most potent anticancer drugs in clinical use. It is effective in the treatment of such common tumors as carcinoma of the lung, breast, and ovary (1). Unfortunately, adriamycin is also toxic to heart muscle, and it is a very active mutagen and carcinogen (2). The drug is thought to mediate its therapeutic and toxic effects via the formation of reactive adriamycin and reduced oxygen radicals; however, the mechanism of radical formation is not known (2-4). Recently a number of laboratories have determined that the binding of iron to adriamycin may be a critical step in the generation of these reactive radicals (5)(6)(7)(8)(9).
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Adriamycin is known to chelate iron.
Step association constants of lo", lo'', and 104. 4 have been estimated for the association of the first, second, and third adriamycin, respectively, with Fe3+ yielding an overall association constant of 1033.4 (10).
In the past we have shown that adriamycin, through chelation of Fe3+, can function as a catalyst for 0, reduction by the physiologic reducing agents glutathione and cysteine (6,7). Activation of molecular 0, to radical species is a feature common to several iron-chelates, but the reaction catalyzed by Fe3+-adriamycin appears unusual because the complex binds tightly to erythrocyte-ghost membranes and catalyzes their destruction in the presence of glutathione (6). More recently, we have shown that Fe3+-adriamycin is also able to bind to and cleave double-stranded DNA (7). In both sets of experiments, we observed that the chelate formed by adriamycin and Fe3+ catalyzes destruction of the erythrocyte ghosts or DNA in the absence of reducing agents. In the case of DNA, this activity is completely blocked by addition of catalase, suggesting that Hz02 had been generated and is a critical intermediate in DNA cleavage. The generation of H20, under these circumstances suggests that iron-adriamycin is able to reduce molecular oxygen in the absence of added electron donors such as the thiols. In support of this hypothesis, it has been reported that a complex of Fe3+-ADP and adriamycin can reduce ferricytochrome c in the absence of O2 (8). Finally, EPR studies of the iron-adriamycin complexes suggest that adriamycin reduces its bound Fe3+ to Fe2+ with subsequent electron transfer to molecular oxygen (5).
In the present study we definitively demonstrate that the iron-adriamycin complexes cycle to reduce 0,. By correlating optical absorbance, electron paramagnetic resonance, and Mossbauer spectroscopic data with polarographic O2 consumption and chemical Fez+ extraction measurements each step in the cycle of 0, reduction is characterized.

Materials
Adriamycin hydrochloride (>98% pure by HPLC') was supplied by the Drug Synthesis and Chemistry Branch of the National Cancer Institute or purchased from Aldrich. Bathophenanthroline disulfonic acid disodium salt hydrate, ultrapure NHIHPPOI, acetohydroxamic acid, and ferrous ammonium sulfate hexahydrate (99.999%) were purchased from Aldrich. Ferric-acetohydroxamic acid was prepared daily by dissolving 3 molar eq of the chelating agent in doubly distilled water and adding 1 molar eq of solid FeC13.6H20 (ACS grade, Allied Chemicals). For EPR experiments a ferric chloride standard solution was prepared by the methods of Aisen et al. (11). For the Mossbauer The abbreviations used are: HPLC, high-performance liquid chromatography; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid BPS, bathophenanthroline; AHA, acetohydroxamic acid.

6820
This is an Open Access article under the CC BY license. experiments a similar 57FeC13 solution was prepared from 97% isotopically pure 57Fe metal purchased from New England Nuclear. All other reagents were of the highest quality commercially available. Doubly distilled water was used in all experiments.

Methods
Optical Studies-Optical absorption spectra were obtained with a Hewlett-Packard UV/Vis spectrophotometer model 8450.4, which is equipped with microprocessor and storage memory that allows the recording and storage of spectra from 200 to 800 nm each second. The anaerobic time course of the reaction of adriamycin with Fe3+ was performed in a quartz anaerobic cuvette equipped with a side arm stoppered with a silicon rubber gas chromatography septum. Four ml of buffered adriamycin were degassed according to previously published techniques (6). Stock ferric acetohydroxamate was added through a gas-tight 10-pl syringe to reach the final concentration desired.
EPR Experiments-Fe3+-adriamycin was prepared by dissolving the desired amount of the drug in doubly distilled water, mixing the desired amount of FeCla, and then titrating the pH from the initial value of approximately 2.5 to the desired pH with 0.05 N NaOH. Titration of the pH by addition of 50 mM Hepes, 0.1 M KC1 buffer at pH 7.4 yielded spectral results identical to those obtained with NaOH. Anaerobic Fe3+-or Fez+-adriamycin complexes were prepared either in vacuum or in 100% nitrogen atmosphere as described previously (5). EPR spectra were recorded with a Varian E-9 spectrometer operating at X band using 100 kHz modulation frequency and a TE 102 cavity. EPR spectra were obtained over the temperature range 10-100 K using an Air Products variable temperature apparatus with EPR dewar insert. Spectra were also obtained at 77 K using a liquid nitrogen EPR dewar. The microwave frequency and magnetic field were calibrated using techniques similar to those described previously (12).
Mossbauer-Anaerobic complexes were prepared by purging with nitrogen gas in a specially constructed glass apparatus fused to a transfer tube which connects to a lucite Mossbauer cell. After the desired time the anaerobic sample was poured through the transfer tube into the Mossbauer cell which was then immediately frozen in liquid nitrogen. After freezing the seal between the glass transfer tube and the Mossbauer cell was opened allowing insertion of the cell into the Mossbauer dewar.
Transmission Mossbauer spectra were obtained using a 512-chan-ne1 spectrometer operated in a constant acceleration mode. A 75-mCi source of 57C0 diffused into a Rh matrix in conjunction with a Jannis flow cryostat were employed to carry out the measurements at the various temperatures. The theoretical fit of the data was performed assuming a lorentzian shape for the absorption lines, and the isomer shifts (6) are given relative to metallic iron at room temperature.
Liquid Chromatographic Experiments-Solutions of Fe3+ and adriamycin were prepared under anaerobic conditions as described for the optical spectroscopy studies. The reaction was monitored spectroscopically for loss of absorbance at 600 nm. After 1.5-h incubation, iron as Fez+ was then extracted according to a modified method (13). This technique depends upon the fact that the Fez+-phenanthroline complex is stable in the presence of oxygen. Four hundred p1 of 0.01 M 4,7-diphenyl-l,lO-phenanthroline (Sigma) in 95% ethanol was added by a gas-tight syringe to 4 ml of the degassed reaction mixture through the side arm of the anaerobic cuvette. The cuvette was opened, and 3.1 ml of a degassed solution containing 10% NH4HzP04, glacial acetic acid, and 1 M HCl (2:0.80.3, v/v/v/) was added quickly with rapid mixing. The solution had a final pH of 2.5 and was extracted by mixing with 4.4 ml of chloroform/methanol (4:1, v/v). At this pH, only the aglycon forms of adriamycin and the Fe(I1)phenanthroline chromogen can be extracted. The organic phase was collected and dried under a stream of N,, resuspended in 2 ml of methanol, and 10-pl aliquots were analyzed by HPLC. The liquid chromatographic analysis of the samples was performed by automatic injection into an HPLC system fitted with a Shoeffel UV/Vis detector model GM 770 and a Shoeffel fluorimetric detector model 970. Simultaneous analysis by detection of absorption at 535 nm and by fluorescence (excitation 229 nm; emission 580 nm) was performed. Samples were run isocratically at 2 ml/min using a Waters RCM-100 silica column and a mobile phase containing 85% CH3CN and 15% of a 0.2% solution of ammonium hydroxide adjusted to pH 4 with formic acid. The 4,7-phenyl-l,lO-phenanthroline was eluted as a single peak at 8 min under these conditions. A standard curve for quantitation was constructed from a stock of Fe(NH4)2(S04)Z ex-tracted as above and serially diluted in methanol in the range of concentrations of 1 to 100 p~. The correlation between the area under the peak eluted at 8 min, and concentration was linear with r > 0.99.
The same samples described above were analyzed by HPLC according to a previously described method for adriamycin and its metabolites. The analysis showed that less than 5% of the total drug present was extracted as the aglycon plus a poorly resolved peak with the same retention time as adriamycin.

Iron-mediated Electron Transfer from Adriamycin to O2
Polarographic Experiments-0, consumption studies were performed with a Clark-type electrode (Yellow Spring Instruments) on a Gilson model 5/6 Oxygraph equipped with a water-jacketed cell for temperature control. Superoxide dismutase (Sigma, type I from bovine blood) was used without further purification. Catalase (Sigma, from bovine liver, 2 X crystallized) was pretreated by mixing with a slurry of Chelex 100 resin in water at 4 "C for 30 min. This preparative step resulted in higher activity of the enzyme and more reproducible results. Control experiments were performed with enzymes that had been boiled for 10 min. The different buffer solutions were equilibrated with air at 37 "C and assumed to contain the same concentration of dissolved 0 2 as water (207 pM at 37 "C). The oxygraph was calibrated daily with a solution of water equilibrated with air for 2 h at 37 "C. All experiments were performed in the dark.
Polarographic studies were performed with ferric-acetohydroxamate as the source of Fe3+ for adriamycin. Control experiments were performed with FeC13 to rule out possible artifacts due to hydroxamic acid. No significant differences were observed, and ferric-acetohydroxamate was routinely used because it results in a lesser degree of variability at neutral pH.
Calculations and Computer Fitting-Equations were fitted to the data using the MLAB fitting routine program. The rate constant for the 600-nm absorption decay of the anaerobic Fe3+-adriamycin complex was calculated by least squares fit to the equation, where a is the measured absorbance at time t; the extrapolated absorbance value at time 0; and K the calculated rate constant. The pH titration curves of 0, consumption were similarly fitted to the Henderson-Hasselbalch equation.

RESULTS AND DISCUSSION
The anticancer drug adriamycin forms complexes with Fe3+ over the physiologic pH range, pH 6.5-8.5, which give rise to distinctive optical absorption, EPR, and Mossbauer spectra. On addition of Fe3+ to adriamycin a new absorption band centered at 600 nm appears along with hypochromicity of the main peak at 480 nm (Figs. lA and 2). These complexes give rise to EPR spectra at 77 K with signals at g = 4.2 and g = 2.01 (Fig. 3). Over the temperature range from 10-100 K the EPR spectra are similar to those observed at 77 K. No new signals are observed on lowering the temperature to 10 K. ( Fig. 4A). Both the temperature behavior of the EPR spectrum and the isomer shift values of the Mossbauer spectra indicate that the Fe3+ added to adriamycin is bound as high spin Fe3+, S = 5/2 (14).
It has previously been determined from optical absorbance titrations of Fe3+ uersus adriamycin that the Fe3+-adriamycin complex has a stiochiometry of 1:3 (6). In the presence of an excess of adriamycin the spectra of aerobic Fe3+-adriamycin complexes remain unchanged. The optical, EPR, and Mossbauer spectra of complexes with an Fe3+:adriamycin ratio of 1:lO are unchanged after 6 h under aerobic conditions at room temperature. Under anaerobic conditions, however, the spectra of the Fe3+-adriamycin complexes dramatically change as a function of time. As shown in Figs. lA and 2 the addition of Fe3+ to an anaerobic solution of adriamycin at pH 7.0 results in the appearance of an absorbance at 600 nm that reaches a maximum in 30 s. After the first 3 min, however, the intensity of the absorbance at 600 nm progressively declines over the next 1.5 h, following a monoexponential decay with an apparent rate constant of 0.77 h" (Figs. 1B and 2). After 2 h the optical absorption spectra appears identical to that of anaerobic preparations of Fez+-adriamycin (Fig. 5). Introduction of Oz rapidly restores the 600-nm absorbance to its initial value (Figs. 1C and 2). The EPR spectra of anaerobic complexes of Fe3+-adriamycin also decrease as a function of time (Fig. 3). As reported previously Fez+-adriamycin is EPR silent, so the decrease in the Fe3+-adriamycin spectra suggest the reduction of the bound Fe3+ to Fez+ (5). A sharp signal appears at g = 2.004 and increases as the Fe3+ signals decrease (Fig. 3). The signal at g = 2.004 has a line width of only 8 G, and it is superimposed on the much broader 225-G line width Fe3+ signal at g = 2.01. The line width and g value of this signal suggest that it is due to a free radical. This signal appears to correspond to the oxidized adriamycin free radical which is formed as adriamycin reduces its bound Fe3+ to Fez+.
On re-exposure to Oz the Fe3+ EPR spectra reappear, and the free radical signal disappears with both of these spectral changes completed within 5 min. Under anaerobic conditions the Mossbauer spectrum of Fe3+-adriamycin also decreases as a function of time. After incubating the sample under anaerobic conditions for 3 h one can notice the emergence of an absorption peak in the high velocity range accompanied by a decrease in the intensity of the higher velocity peak of the original doublet (Fig. 4B). Analysis and simulation of this spectrum indicates that the observed change is due to a decrease in the intensity of the original doublet with the formation of a new component which contributes a doublet with an isomer shift 6 = 1.18 f 0.02 mm/s and quadrupole splitting A& = 1.82 f 0.02 mm/s. These parameters are characteristic of Fez+, S = 2 (14), which indicates that the Fe3+ with S = 5/2 is reduced to high spin Fez+, S = 2. On exposure to Oz the Fez+ doublet decreases with a corresponding reappearance of the original Fe3+ doublet (Fig. 4C). After 5 min the reaction is completed and only the Fe3+ doublet remains. In order to chemically confirm the reduction of Fe3+ by the drug, we tested the ability of the specific Fez+ chelator 4,7diphenyl-1,lO-phenanthroline to compete with and displace iron from adriamycin after an anaerobic incubation of the ferric chelate of the drug. Because the drug and Fez+-phenanthroline absorb in the same region, we extracted the Fez+ complex of phenanthroline with CHC13-CH30H (4:1, v/v). Fig. 6 shows the results of the HPLC analysis of the organic extract obtained after a 1.5-h anaerobic incubation of adriamycin with Fe3+. Panel A of the figure shows that a Fez+phenanthroline standard eluted as a single symmetric peak absorbing at 535 nm. The sample in which adriamycin and Fe3+ had been incubated anaerobically yielded a peak with the same retention time (panel B ) . Identical results were obtained on aerobic incubation. Simultaneous analysis by fluorimetric detection showed that less than 5% of the drug had been extracted in the organic phase and eluted from the column with the solvent front. Quantitation through the use of an external standard curve indicated that all the iron initially added as Fe3+ had been reduced to Fez+. Panel C shows that a solution of ferric acetohydroxamate, when assayed for the presence of Fez+, yielded minimal amounts of phenanthroline-iron. Similar amounts of Fez+ (1 pM) were found in buffer alone (Fig. 6D) or adriamycin alone (data not shown).
Because of the reaction with 0 2 elucidated by the spectroscopic studies, one would expect the reaction of Fe3+ with adriamycin to be associated with O2 consumption. Measurement of Oz concentration by polarography did in fact show that addition of Fe3+ to adriamycin triggered the consumption of Oz. At pH 7.4 and 0.2 mM adriamycin, the addition of 0.5 eq of Fe3+ resulted in a rate of 2.4 pmol/liter/min. In addition, the rate of Oz consumption was a function of adriamycin concentration yielding a linear double reciprocal plot (Fig. 7). As in the case of ferricytohcrome c reduction by adriamycin-Fe3+-ADP (8), the rate also progressively increased at more basic pH. The effect of pH on the rate of oxygen consumption by both adriamycin and its iron complex is shown in Fig. 8. The presence of iron shifted the apparent pK from 9.65 to 7.95. Thus, in the physiologic pH range the Fe3+-adriamycin complex reduces O2 while adriamycin alone does not. The central role of Fe3+ reduction in the reaction leading to O2 consumption is clearly indicated by the experiment shown in The concentrations used in the experiment were 200 +M adriamycin, 100 p~ Fe(AHA)3, and 300 p~ BPS.  Fig. 9. Addition of the water-soluble 4,7-diphenyl-l,lO-phenanthroline sulfonate (bathophenanthroline, BPS) blocked 0 2 consumption either before or after the reaction with the complex was initiated. The known respective affinities of adriamycin and BPS for Fe3+ rule out the possibility that, at the concentrations used (see legend to Fig. 9), the effect of BPS could be due to successful competition with adriamycin for ferric iron.

Iron-mediated Electron Transfer from
Fez+ chelates can either react with O2 to yield superoxide and H202 or bind O2 in a superoxo-or peroxo-like form (15). In some systems, O2 coordinated to iron in a superoxo-or peroxo-like form is susceptible to nucleophilic displacement (16). In the case of adriamycin (200 p~) , however, the addition of excess of NaN3 (10 mM), NaCN (10 mM), or NaCl (100 mM) with respect to iron did not inhibit 0 2 consumption by Fe3+-adriamycin. To rule out the possibility that O2 might still be bound to iron but displaceable as either superoxide or H202 rather than 02, nucleophile addition was repeated at the same concentrations used above, but now in the presence of superoxide dismutase and catalase. The introduction of the two enzymes (superoxide dismutase, 120 units/ml; catalase, 400 units/ml) into the oxygraph cell resulted in the immediate evolution of 02, that was, however, unrelated to the presence or absence of the nucleophiles. The latter observation prompted an analysis of the enzyme effects shown in Table I.
Catalase caused a 35% drop in Oz consumption. Since catalase converts each 2 eq of H202 to 1 eq of 02, this indicates that HzOz can account for 70% of the O2 consumed. The initial rates of O2 uptake were affected to the same extent by addition of catalase alone or catalase and superoxide dismutase. Superoxide dismutase alone did not appear to have any measurable specific effect, but boiled superoxide dismutase did decrease the rate of O2 consumption. This surprising decrease could be due to the release of copper and zinc ions which interfere with iron-mediated O2 reduction, or the denatured protein itself may bind the drug-iron complex altering the rate of Oz consumption. From the above experiments, it appears that the observed O2 consumption is not due to the formation of a stable adriamycin-iron-O2 complex between O2 and the drug-iron chelate, but to a reaction that yields Hz02 as the predominant product. The present investigation demonstrates that iron binds to adriamycin and that these complexes cycle to reduce molecular 0,. Optical absorption, EPR, and Mossbauer spectroscopic studies as well as Fez+ chemical extraction experiments all indicate that the Fe3+ bound to adriamycin is reduced to Fez+. The isomer shift values observed in the Mossbauer spectrum indicate that the Fez+ formed is high spin S = 2.
The EPR experiments demonstrate that an oxidized adriamycin free radical is formed as Fe3+ is reduced to Fez+. An unusual aspect of this reaction sequence is the slow rate of Fe3+ reduction, as indicated by the apparent rate constant of the spectral shift seen in Fig. 2. This behavior is confirmed by the EPR and Mossbauer experiments (Figs. 3 and 4) that show a similar slow rate for the disappearance of the Fe3+ signals. This observation is unusual because electron transfer reactions are fast processes, as in the case of Fe3+-thiol complexes (17). The rapidity of the subsequent reaction (Fig. 2) after O2 addition suggests that the Fe3+ reduction is the ratelimiting step in the reaction sequence. This is further corroborated by the ability of BPS to immediately stop the O2 uptake by the drug-metal chelate. Perhaps the closest precedent in the chemical literature for these observations is the reaction of o-quinones and catechols (0-hydroquinones) with iron. The recent report of phenanthrenequinone-phenanthrenecatecholate complexes is of particular interest. These complexes can exhibit either strong apparent charge transfer bands without Fe3+ reduction or, in contrast, full reduction of Fe3+ with formation of a radical semiquinone ligand depending upon the dielectric strength of the solvent (18). Parallel detailed studies do not appear to have been done on hydroxyquinones analogous to adriamycin. However, this study indicates that the Fe3+-adriamycin complex at neutral pH also exhibits both a strong charge transfer band and slow electron transfer from the ligand to the iron.
The second step in this reaction cycle is the reduction of oxygen accompanied by oxidation of the ferrous iron. Our polarographic experiments, Table I, indicate that 70% of the Oz consumed is transformed to H202. Recently we have shown that H2OZ reacts with the complex leading to the oxidative cleavage of DNA. The latter suggests the complication that H202 is also a transient product able to react with the ironchelate of adriamycin. Thus, while it is clear that HzOz is a product, the formation of other reactive O2 species, such as hydroxyl radical, may also occur. As a result of electron transfer to molecular oxygen Fe3+-adriamycin is regenerated, and in the presence of excess adriamycin this cycle of Oz reduction will continue.
In recent years, oxidative damage to macromolecular targets has been advocated as a mechanism responsible for the cytotoxicity and/or the cardiac toxicity of adriamycin (2-4). The cycle of iron-mediated electron transfer from adriamycin to molecular oxygen results in the formation of reduced oxygen and oxidized adriamycin radicals which may mediate the therapeutic and toxic effects of the drug.