1004
Chem. Res. Toxicol. 2005, 18, 1004-1011
Desferrioxamine Inhibits Production of Cytotoxic Heme to Protein Cross-Linked Myoglobin: A Mechanism to Protect against Oxidative Stress without Iron Chelation Brandon J. Reeder* and Michael T. Wilson Department of Biological Sciences, Central Campus, University of Essex, Wivenhoe Park, Colchester, Essex, CO4 3SQ, United Kingdom Received December 10, 2004
The heme group of myoglobin can form a covalent bond to the protein when met (ferric) myoglobin is reacted with peroxides under acidic conditions. This heme to protein cross-linked species is highly pro-oxidant and found in the urine of patients with rhabdomyolytic-associated acute renal failure. Desferrioxamine, an iron-chelating agent used in the treatment of iron overload, is reported to be partially effective at preventing kidney failure following rhabdomyolysis. In this article, we show that in addition to its capacity as an iron chelator, desferroxamine can inhibit the peroxide-induced formation of heme to protein cross-linked myoglobin and decreases the pro-oxidant activity of both native and heme to protein cross-linked myoglobin. The mechanism of peroxidation and of heme to protein cross-linking involves the formation of ferryl intermediate (Fe4+dO2-), and it is by the reduction of this intermediate to the ferric form that desferrioxamine can exert inhibitory effects. The concentrations at which desferrioxamine inhibits the formation of heme to protein cross-linked myoglobin and prevents the pro-oxidant activity of native and oxidatively modified myoglobins are comparable to the concentrations used for in vivo studies of iron-related oxidative stress. Thus, the ameliorative effects of treatment of posthemolytic events by desferrioxamine cannot be exclusively assigned to its ability to chelate free iron.
Introduction The respiratory heme proteins myoglobin (Mb)1 and hemoglobin (Hb) can react with peroxides to form higher oxidation states such as the ferric and ferryl species. The ferryl oxidation state of Mb and Hb promotes the oxidation of lipids by the abstraction of a hydrogen atom from the lipid, forming a lipid radical (1-4). These heme proteins are believed to play a role in some oxidative stress reactions such as reperfusion injuries after ischemia and acute renal failure associated with muscle damage (rhabdomyolysis) (5-7). Recent evidence points to the importance of peroxide-induced heme to protein cross-linked Mb (Mb-X) under conditions of oxidative stress. Mb-X is a stable marker for peroxide-induced ferric-ferryl redox cycling and has been found to be a useful marker for oxidative stress in vivo (8, 9). Highly relevant is the discovery that Mb-X itself is significantly more cytotoxic than native Mb, with an enhanced ability to oxidize free and membrane-bound lipids. Mb-X increases in vitro low density lipoprotein oxidation and enhances human fibroblast cell death 5-fold with respect to unmodified Mb (10, 11). In conditions such as rhabdomyolysis, free and membrane-bound lipids are oxidized by a noncyclooxygenase mechanism to generate potent vasoconstrictor compounds termed F2-isoprostanes within the rhabdomyolytic kidney * To whom correspondence should be addressed. Tel: +44-1206 873333 ext. 2119. Fax: +44-1206 872592. E-mail:
[email protected]. 1 Abbreviations: DFO, desferrioxamine; DTPA, diethylenetriaminepentaacetic acid; Mb, myoglobin; Mb-X, heme to protein cross-linked Mb; Hb, hemoglobin.
(12-14). Lipid oxidation and isoprostane formation have been proposed to proceed by an as yet undefined radicalbased mechanism (15). Both free iron and Mb can initiate lipid oxidation reactions in vitro. Mb-X has been identified in the urine of hospitalized human patients suffering from rhabdomyolytic-associated acute renal failure (7) and in the urine and kidneys of animal models of rhabdomyolysis (Reeder, Moore, and Wilson, unpublished data). These data indicate that Mb undergoes ferricferryl redox cycling in rhabdomyolytic kidneys, implying that Mb has been involved in oxidative events. Furthermore, we have found that a substantial fraction of the Hb in cerebrospinal fluid following an aneurysmal subarachnoid hemorrhage contains the heme cross-linked form of this protein (16). The importance of intact hemes on oxidative stress is illustrated by heme oxygenase knockout mice that have a greatly decreased ability to degrade heme groups (17). These knockout mice are more susceptible to the complications following rhabdomyolysis and experience higher levels of plasma creatinine and lactate dehydrogenase as compared to normal mice, indicating a lower renal function and higher cellular damage, respectively. This normally nonfatal model of rhabdomyolysis was associated with a high mortality rate with the heme oxygenase knockout mice (17). Conversely, overexpression of heme oxygenase-1 inhibits arterial contractions induced by Hb and can reduce vasospasm after subarachnoid hemorrhage in rats (18). These studies support the view that it is the intact heme and not free iron that is the toxic agent during posthemolytic oxidative stress.
10.1021/tx049660y CCC: $30.25 © 2005 American Chemical Society Published on Web 05/07/2005
Desferrioxamine and Heme to Protein Myoglobin
Desferrioxamine (DFO, DFO mesylate, deferoxamine, deferiprone, desferal) is a transition metal chelator with particular affinity for ferric iron and is currently used primarily in the clinical treatment of patients with elevated levels of iron, such as that caused by regular blood transfusions required to treat thalassaemia major (19-21). A number of studies have used DFO to examine the role of “free” iron (i.e., not heme-bound forms) in the mechanism of oxidative stress (22-24). Because DFO is able to bind transition metals such that their catalytic activity is inhibited (25-27), amelioration of the symptoms of oxidative stress by DFO is frequently attributed to the capacity of DFO to act as an iron chelator and thus the importance of free iron in oxidative stress is inferred. While it is clear that oxidative stress can be initiated by free iron and that under such circumstances DFO is an effective inhibitor of iron-related oxidative stress, it is also the case that DFO can impact on oxidative reactions independently of its ability to chelate iron. For example, DFO has been reported to act as a substrate for peroxidases (27) and a scavenger of radicals (27-29). It can also act as an electron donor, decreasing peroxide-induced oxidation of membranes by Mb and Hb (30-33). The effects of DFO on the formation and decay of the ferryl intermediate of Mb and its effect on the pro-oxidant activity of Mb in vitro have not been studied in detail. To date, there has been no investigation of the effect of DFO on Mb-X formation and pro-oxidant activity. It is, therefore, important to determine whether established therapeutic agents such as DFO have any effect on the concentrations and pro-oxidant activities of cytotoxic compounds such as Mb-X. Such data could clarify the relative contributions of heme proteins and free ironinduced oxidative stress. In this study, we have examined the effect of DFO on the formation of Mb-X. We show that DFO inhibits peroxide-induced heme to protein cross-linking in Mb. We show that the mechanism of inhibition involves reduction of the ferryl heme that we have previously demonstrated is an essential prerequisite for cross-linking (34). We have also examined the effect of DFO on the pro-oxidant activity of native Mb and Mb-X. DFO effectively inhibits both native Mb and Mb-X-catalyzed oxidation of phosopholipid liposomes at concentrations comparable to those used in previous in vitro and in vivo studies and in therapies of iron overload. The results show that DFO can diminish Mb-induced oxidative stress in two distinct way: (i) by inhibition of Mb conversion to Mb-X and (ii) by blocking the pro-oxidant activity of both Mb and MbX.
Experimental Procedures Chemical Reagents and Supplies. Equine Mb, 30% H2O2, soybean lipoxygenase (EC 1.13.11.12), linoleic acid, and DFO mesylate (C25H48N6O8‚CH4O3S) were purchased from Sigma (Poole, United Kingdom). Trifluoroacetic acid (TFA) was obtained from Fischer Chemicals (Loughborough, United Kingdom). Preparation of Mb. Met (ferric) Mb was prepared by filtration through a Sephadex G-25 column and stored at 4 °C. No further purification steps were performed, and Mb was used within 3 days of preparation. Preparation of 13-L-Hydroperoxy 9-11, cis-trans Octadecadienoic Acid (HPODE). The lipid hydroperoxide HPODE was generated from linoleic acid by incubating with soybean 15-lipoxygenase as previously described (35). The
Chem. Res. Toxicol., Vol. 18, No. 6, 2005 1005 HPODE was kept in ethanol at -80 °C until required and introduced to the experimental reaction as a 1% ethanol solution (final concentration). Measurement of Heme to Protein Cross-Linking by Reverse Phase HPLC Analysis. Concentrations of Mb-X were determined by analyzing samples on an Agilent HP1100 HPLC fitted with a diode array spectrophotometer. A Zorbax StableBond 300 C3 reverse phase column was used (250 mm × 4.6 mm, 5 µm pore size) fitted with a C3 guard column (12 mm × 4.6 mm). The solvents used were as follows: A, 0.1% TFA; B, acetonitrile containing 0.1% TFA. The gradient profile was 0-10 min; 35% B, 10-15 min; 35% B increasing to 37% B, 15-16 min; 37% B increasing to 40% B, 16-20 min; 40% B increasing to 43% B, steady for 5 min then increasing to 95% B, 25-30 min. The column temperature was 25 °C, the pump flow rate was 1 mL min-1, and the sample injections were 25 µL. Mb-X amounts were determined from the integral of the total peak area at 400 nm between 19 and 25 min, assigned to Mb-X. The peak areas were converted into concentration by multiplying the values by a constant determined from the addition of apoMb/ Mb-X mixtures. The apoMb/Mb-X mixture was prepared using the acid-butanone extraction method (8) to remove noncovalently bound heme and dialyzed against water overnight to remove trace solvent and for 2 h against 5 mM sodium acetate buffer, pH 5. The Mb-X concentration of this standard was determined optically using a Varian Cary 5E spectrophotometer [408nm ) 76 mM-1 cm-1, pH 6 (9)]. Rate Constants Measurement. The peroxide-induced formation of ferryl Mb and the subsequent ferryl reduction were followed optically at 408 and 425 nm using an Agilent HP8453 diode array spectrophotometer fitted with a multicell carriage. The time course of the optical changes was fitted to a model (36) defining the rate constants for ferryl formation (k1) and ferryl decay (k2):
A)
( )( χ-1 χ
)
Amax k1Amax (e-k2t - e-k1t) + (1 - e-k1t) (1) k1 - k2 χ
where Amax ) maximum absorbance of ferryl species and 1/χ ) the absorbance offset to account for spectral changes due to oxidative modifications to the heme moiety. The fits were calculated by the least-squares method using the Microsoft Excel solver program. This model represents a modification of the Bateman equation where compound A (met) f compound B (ferryl) f compound C (modified met). Preparation of Lecithin Liposomes. Soybean phospholipids (50 mg, lecithin “asolectin”, 19% phosphatidylcholine, Sigma, type II-S P-5638) were dissolved in chloroform in a glass round-bottomed flask. The chloroform was evaporated under a stream of argon gas. This was repeated, and the lecithin was dried thoroughly, leaving a fine lecithin film on the glass surface. The lecithin was suspended in 10 mL of sodium phosphate buffer (5 mM) so that the concentration of phospholipids was 5 mg mL-1. Oxygen was removed from the solution using a nitrogen aspirator fitted with a high vacuum pump. The lecithin suspension was freeze-thawed four times to fracture the phospholipid layers into leaflets. These phospholipid leaflets were stored at -80 °C until required. Liposomes were formed from the leaflets using a Northern Lipids stainless steel extruder fitted with a Whatman nucleopore drain disk (25 mm diameter) and two layers of Whatman nucleopore polycarbonate membranes (25 mm diameter with 0.1 µm pore size). The lecithin was forced through the membranes using pressurized nitrogen gas (∼20 bar). The membranes and drain disk were replaced, and the collected lecithin was forced through the membranes a further 10 times. The liposomes were stored at 4 °C and used within 2 days of preparation. Measurement of Liposomes Oxidation. An Agilent 8453 diode array spectrophotometer was used to measure the oxidation of liposomes. Mb or Mb-X was incubated with the liposomes (200 µg/mL) at 37 °C at pH 7.4 [25mM sodium phosphate containing 25 µM diethylenetriaminepentaacetic acid (DTPA)].
1006
Chem. Res. Toxicol., Vol. 18, No. 6, 2005
Figure 1. Effect of DFO on the time course of ferryl formation and decay. Ferric Mb (10 µM) was reacted with H2O2 (10 µM) in sodium phosphate buffer, pH 7. The time course of ferryl formation and subsequent autoreduction was monitored at 425 and 408 nm at 1 min time intervals. Different concentrations of DFO were added to the Mb directly before the addition of peroxide (A, 0 µM; B, 1 µM; C, 2 µM; D, 3 µM; E, 5 µM; F, 7.5 µM; G, 10 µM; H-K, 20, 30, 50, and 75 µM; and L-Q, 100, 200, 300, 500, 750, and 1000 µM).
Reeder and Wilson
Figure 2. Changes in Mb spectrum following the reaction with hydrogen peroxide in the presence of DFO. Ferric Mb (10 µM) has a Soret peak maximum at 408 nm and peaks in the visible region at 502 and 630 nm. After the addition of H2O2 in the presence of DFO (100 µM), a proportion of the Mb is oxidized to ferryl Mb following a time course identical to line L in Figure 1. Shown is the ferric/ferryl spectrum at the maximum ferryl concentration (∼5 µM) and the final Mb spectrum after the reaction has come to completion. The spectrum of this final Mb species is almost identical to the initial ferric Mb spectrum with evidence of slight heme degradation (∼5%) as shown by the difference spectrum (inset).
Oxidation of liposomes was measured by the appearance of lipidbased conjugated dienes measured optically at 234 nm. To correct for changes in light scattering, a three-point baseline drop was calculated from optical measurements taken at 220, 234, and 255 nm. The lipid hydroperoxide HPODE [234nm ) 2.5 × 104 M-1 cm-1 (37)] was used to calculate the extinction coefficient using the above method ( ) 1.45 × 104 M-1 cm-1).
Results Effect of DFO on the Ferrylmyoglobin Reduction Rate. The effect of DFO on the time course of H2O2induced formation and decay of ferryl Mb is shown in Figure 1. In the absence of DFO (Figure 1, line A), the spectrum of Mb changes from the met (ferric) oxidation state with a peak maximum at 408 nm in the Soret region to that of the ferryl state exhibiting a prominent shoulder at 425 nm (Figure 2). The presence of the ferryl oxidation state can be verified by its reaction with sulfide that gives a distinctive peak at 617 nm (38, 39) (data not shown). This confirmed that the ferryl state was >95% populated 25 min after the reaction was initiated under the described conditions. Once H2O2 is consumed, a second phase is observed in which the spectrum of Mb approaches the original absorbance intensities of ferricMb. The addition of DFO changes the time course of ferryl formation and decay, as reported by others (33, 40), decreasing the extent of ferryl formation and enhancing ferryl decay. The effect of DFO on the rate constants of ferryl formation and decay was determined by fitting the time course of the optical changes to a mathematical AfBfC model (36) (eq 1), the results of which are shown in Figure 3. DFO does not significantly affect the rate constant of ferryl formation (Figure 3, 9). Ferryl decay, however, is strongly affected by DFO, increasing the rate constant for ferryl reduction to ferric (met) as DFO concentrations increase (Figure 3, b). The rate constant for ferryl decay increases over 700-fold, from 1.6 × 10-5 s-1 in the absence of DFO to 1.2 × 10-2 s-1 in the presence of 1 mM DFO. Identical results were obtained from experiments in which the ferryl species was formed at pH 10, where it
Figure 3. The effect of DFO on the rate constant for ferryl formation and ferryl decay. Time courses from Figure 1 were fitted to a model to determine the rate constant for ferryl formation (9) and ferryl reduction (b) as a function of DFO addition (see text).
is stable for hours, and any excess peroxide removed by catalase prior to rapid change of the pH to 7 in the presence of DFO. The kinetics of reduction was as represented in Figure 3. Reactions between DFO-iron complexes and peroxide at pH 7 must therefore be insignificant for the time courses that we report in Figure 1. Detection and Quantification by Reverse Phase HPLC of Mb-X. Cross-linking of heme to protein follows peroxide-induced formation of a radical and a protonated ferryl iron that leads to the generation of a covalent bond between the protein and the heme moiety (9, 34). The formation of ferric Mb-X is pH-dependent, being greater at lower pH values (8, 34). The effect of H2O2 on the composition of oxidatively modified Mb can be determined quantitatively using reverse phase HPLC (Figure 4). Without H2O2, the heme and globin components of Mb elute separately, heme at 14.6 min and protein at 22.1 min (Figure 4A, i and ii). This separation of heme from the globin occurs because the acidic HPLC solvents used
Desferrioxamine and Heme to Protein Myoglobin
Figure 4. Reverse phase HPLC of ferric Mb before and after reaction with hydrogen peroxide. (A) Chromatograph of Mb (25 µL injection of 100 µM ferric Mb) monitored optically at 400 (top row) and 280 (bottom row) nm. Unmodified heme, absorbing at both 280 and 400 nm, elutes at 14.6 min (i). The apoprotein component, absorbing only at 280 nm, elutes at 22.1 min (ii). (B) Chromatograph of ferric Mb after incubation with 2× excess H2O2. Numerous oxidatively modified hemes are seen to elute between 7 and 13.5 min (iii). Mb-X elutes between 19 and 25 min with a large peak at 23.6 min (iv). (C) Chromatograph of ferric Mb after incubation with 10× excess H2O2. Incubations were at pH 5 and 25 °C, and HPLC conditions and methods are as those described in the Experimental Procedures.
(∼pH 2) disrupts the tertiary structure of the heme pocket and the histidine-iron ligand that holds the heme within the protein. Under acidic conditions, the heme may be considered to be “free”, i.e., not covalently bound to the protein. After reaction with 2× excess H2O2, the unmodified heme content is decreased by approximately half and numerous other oxidatively modified free hemes are observed eluting before the unmodified heme peak (Figure 4B, iii). These hemes have distinctive spectral properties, all showing a peak at approximately 720 nm under the acidic conditions (34). This spectral property is indicative of a heme that has had the conjugated structure of the porphyrin ring disrupted and are similar to the green “chlorin” type hemes (41). Heme that is
Chem. Res. Toxicol., Vol. 18, No. 6, 2005 1007
Figure 5. Inhibition of heme to protein cross-link formation by DFO. Ferric Mb (100 µM) was reacted with H2O2 (0-1000 µM) at pH 5 (25 mM sodium acetate containing 100 µM DTPA) in the presence of various concentrations of DFO (b, 0 µM; 9, 50 µM; 2, 100 µM; O, 200 µM; 0, 300 µM; 4, 500 µM; and *, 1000 µM). Heme protein cross-linking (Mb-X) was measured by reverse phase HPLC (see the Experimental Procedures). The extent of cross-linking as a function of H2O2 concentration is shown in A and as a function of DFO concentration in B (b, 50 µM; 9, 100 µM; 2, 150 µM; [, 200 µM; O, 250 µM; 0, 300 µM; 4, 500 µM; ], 750 µM; and *, 1000 µM). In B, the Mb-X concentration has been normalized with respect to the extent of Mb-X formation in the absence of DFO.
covalently bound to the globin (Mb-X) elutes between 19 and 25 min with a large peak at 23.6 min (Figures 4B and 3C, iv). It is probable that protein damage, generated by radicals during the peroxide reaction, is largely responsible for the heterogeneity of the Mb-X peak, although at this stage we cannot rule out heterogeneity of the heme to the protein bond. Mb-X has similar spectral properties to the oxidatively modified free hemes plus a protein component absorbing at 280 and 210 nm (34). The addition of 10× excess H2O2 further decreases the unmodified heme content and also the oxidatively modified free heme and Mb-X content due to general degradation of the porphyrin ring (Figure 4C). DFO Inhibits Peroxide-Induced Heme to Protein Cross-Linking and Peroxide-Induced Mb-X Degradation. Figure 5A shows the extent on the formation of Mb-X as a function of H2O2 concentration. Concentrations of Mb-X were determined by reverse phase HPLC as described in the Experimental Procedures section. In the absence of DFO, increasing H2O2 concentration increased the yield of Mb-X. The formation of Mb-X reaches a maximum yield around 500 µM H2O2 (41.8%). Higher peroxide concentrations result in a decrease in the yield
1008
Chem. Res. Toxicol., Vol. 18, No. 6, 2005
Figure 6. DFO inhibits heme to protein cross-linking and the degradation of heme protein cross-linked Mb. Ferric Mb (100 µM) was reacted with H2O2 without (b) or with (O) 2 mM DFO. Mb-X concentrations were determined by HPLC. In the absence of DFO, Mb-X is degraded by peroxide concentrations higher than 500 µM. In the presence of DFO, Mb-X formation is inhibited at concentrations of H2O2 below 1 mM. Above this concentration, Mb-X degradation is inhibited by DFO such that the Mb-X concentration is higher than that seen in the absence of DFO.
of Mb-X. This decrease is not due to a lower production of Mb-X but rather an increase in the degradation of the heme. The addition of DFO to the Mb/H2O2 reaction mixture generally decreases the extent of Mb-X formation. For example, at 100 µM H2O2, inclusion of 1 mM DFO decreased the extent of Mb-X formation 7-fold from 18.4 to 2.7 µM. Figure 5B reports the data from Figure 5A, normalized such that the extent of Mb-X formation in the absence of DFO at each H2O2 concentration represents 100% Mb-X formation. This figure confirms that DFO is an effective inhibitor of Mb-X formation, so long as the concentration of peroxide is low (
