INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

HEMOGLOBIN (Hb), when used as an oxygen-carrying blood substitute outside the intracellular milieu of the red cell, undergoes oxidation-reduction reactions that lead to both the generation and the consumption of highly reactive oxygen and nitrogen species (1). Autoxidation of oxyhemoglobin (Fe²⁺) is known to lead to the formation of the non-oxygen-carrying ferrihemoglobin (met; Fe³⁺) and superoxide anion (O₂·), which in turn dismutates into hydrogen peroxide (37). Hb consumes nitric oxide (NO) in a series of redox reactions to form nitrate (51) and thus effectively removes NO as a biological transducer of vascular relaxation (1, 3, 44). Methemoglobin and oxyhemoglobin react with hydrogen peroxide to catalyze the removal of hydrogen peroxide, without being consumed in the process (a property that qualifies Hb as a pseudoperoxidase enzyme), and to generate ferrylhemoglobins (Fe⁴⁺) as a transient intermediate (23). Ferrylhemoglobin is a strong oxidizing agent that is believed to mediate the peroxidation of lipids, proteins, carbohydrates, and nucleic acids (20, 25).

Early clinical use of cell-free Hb as a blood substitute was limited by the significant in vivo toxicities observed in clinical trials (5, 50) and in animal studies (21, 35). These toxicities present as dysregulation of hemostasis and include hypertension, bradycardia (5, 50), decreased glomerular filtration rate and renal plasma flow (10), mild prolongation of partial thromboplastin time (50), disseminated intravascular coagulation with resultant thrombosis (21, 35), and ischemic parenchymal damage (21, 35). Clinical experience with the current generations of highly purified Hb-based products has identified new toxicity and efficacy problems. These include vasoconstriction, a short intravascular half-life, rapid autoxidation (34), and potential free radical-mediated toxicity (1).

The endothelial cells lining the vasculature are particularly sensitive targets for oxidation products generated by cell-free Hb, because of direct and continuous contact of the endothelium with circulating blood. Endothelial cell damage will disrupt normal vascular control of coagulation, vessel tone and tension, and vessel permeability (17, 41). Several in vitro studies on Hb interactions with cells in culture show that Hb either can be directly cytotoxic or can alter the endothelial cell state in a number of ways. Concentrations of Hb that are not directly cytotoxic (<100 µM) damage endothelial cells by disrupting cytoskeletal structures (16), by sensitizing endothelial cells in culture to hydrogen peroxide (6) and other cytotoxic agents (43), and by inducing genes that protect against oxidative damage, such as heme oxygenase (8, 38, 39) and transferrin (7, 8). Concentrations of Hb >100 µM are directly cytotoxic to endothelium (39, 52), with an apparent correlation between cytotoxicity and the extent of Hb oxidation. Native human Hb oxidized to its ferric form is more cytotoxic toward vascular endothelium than oxyhemoglobin (39). The cytotoxicity of oxyhemoglobin is dependent on the rate of Hb autoxidation and correlates with the time in culture and the presence of the iron chelator deferoxamine but not catalase, a powerful scavenger of hydrogen peroxide (39). Oxidation of Hb generates a number of products that could mediate cellular cytotoxicity including oxygen free radicals, hydrogen peroxide (43), higher oxidation states of Hb such as the ferryl heme (Fe⁴⁺) (16, 39), and free heme released after protein damage (6).

In this study, the effects of native Hb on hydrogen peroxide cytotoxicity toward bovine aortic endothelial cells (BAEC) in culture were compared with the effects of two chemically modified Hb, developed as blood substitutes, that differ from native Hb in their susceptibility to oxidation by hydrogen peroxide (12, 39). We document that ferrylhemoglobin, formed by the reaction of hydrogen peroxide with the chemically modified Hb, is a mediator of cytotoxicity. This may provide a plausible mechanism for some of the in vivo toxicities seen with infusion of these products (30). These findings may also be relevant to current efforts directed toward the design of a new generation of Hb-based blood substitutes with a reduced propensity to enter into oxidative reactions and thus protect against heme-mediated toxicity (49).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. Acridine orange, ethidium bromide, and the lactate dehydrogenase (LDH) assay kit were from Sigma Chemical (St. Louis, MO). [methyl-³H]Thymidine (4 Ci/mmol) was from ICN Biomedicals (Costa Mesa, CA). Dulbecco's modified Eagle's medium (low glucose), fetal bovine serum, heparin, and L-glutamine were from Life Technologies (Grand Island, NY).

Isolation of BAEC. Endothelial cells were isolated from the vessel wall of bovine aorta (Mt. Airy Meat Locker, Mt. Airy, MD) as previously described (15). The purified endothelial cell population exhibited the characteristic cobblestone, nonoverlapping morphology of confluent monolayers and the presence of uniformly distributed acetylated low-density lipoprotein uptake, identified with the fluorescent probe 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (Biomedical Technologies, Stoughton, MA) (55). BAEC were maintained in culture with Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum, 0.05 mg/ml heparin, 0.025 mg/ml ascorbate and 2 mM L-glutamine in a 5% CO₂ incubator at 37°C (BAEC-DMEM). Experimental data were obtained from BAEC in the 2nd to 10th passages.

Measurement of DNA fragmentation and LDH release. DNA fragmentation was quantitated using a filtration assay to separate intact chromatin from DNA fragments (13, 36). Briefly, 2 × 10⁴ BAEC were plated per well of a 96-well tissue culture plate with BAEC-DMEM containing 0.2 µCi [³H]thymidine. After BAEC were cultured for 2 days, the medium containing [³H]thymidine was removed and replaced with fresh medium. The BAEC were placed back in culture overnight to wash out residual free [³H]thymidine before being cultured with the indicated treatments for 6 or 24 h. Each treatment condition was performed in replicates of five. In experiments to determine the effect of Hb on hydrogen peroxide cytotoxicity, Hb was added to BAEC immediately before the addition of hydrogen peroxide. At the end of the incubation the plates were stored at 70°C for at least 24 h. After the plates were thawed, BAEC and the medium were aspirated onto glass fiber filters with a microcell harvester (Skatron Instruments, Sterling, VA). The radioactivity on the filters was quantitated with the LKB Betaplate Beta counter (Wallac, Gaithersburg, MD). A control sample of BAEC incubated in BAEC-DMEM was used to determine 100% intact chromatin. Percent DNA fragmentation was calculated as the average difference in intact DNA between the test and control samples divided by the control (13, 36).

Preparation of ferrylhemoglobin. A stock solution of 30% (wt/vol) hydrogen peroxide was added to 4 mM -DBBF prewarmed to 37°C to yield a final concentration of 40 mM hydrogen peroxide. The solution was incubated at 37°C for 1 min and then loaded onto a 5-ml Sephadex G-25 column preequilibrated with 50 mM sodium phosphate, pH 7.2 at 4°C. The heme-containing fraction eluting from the column in the void volume was collected, analyzed spectrally, and stored on ice until use. Samples of the reaction mixture were monitored spectrally over the range from 380 to 700 nm using a Hewlett-Packard 8452A diode-array spectrophotometer. The proportion of oxyhemoglobin and its oxidation products, including the ferrylhemoglobin, were determined according to earlier published work (2). Spectral analysis of the column-purified material revealed that 45-50% of the heme was in the ferryl state and free of residual hydrogen peroxide.

Assessment of changes in cellular morphology. BAEC were cultured in six-well tissue culture dishes without and with the indicated additives. An equal volume of PBS containing 0.01 mg/ml ethidium bromide and 0.003 mg/ml acridine orange was added to the BAEC culture at the end of 6 h and incubated in the dark at room temperature for 5 min (4, 18). Fluorescence staining of BAEC was visualized with an inverted microscope (Diaphot 300, Nikon, Melville, NY) using a fluorescein filter set.

Statistical analysis. Data are expressed as means ± SE for replicate experiments. The differences between treatment groups were assessed by one-way ANOVA followed by Student's unpaired t-test. Statistical significance was defined as P < 0.05 to reject a null hypothesis. All statistical calculations were performed with JMP version 3.2 for the Macintosh (SAS Institute, Cary, NC).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Peroxide-induced killing of BAEC. Hydrogen peroxide exhibited a steep dose-response curve for inducing cell death in BAEC (Fig. 1), in keeping with the well-established deleterious effects of peroxide on tissues and organs (29). Analysis of variance of the DNA fragmentation and LDH release data showed that the increasing cytotoxicity observed with increasing hydrogen peroxide was statistically significant (P < 0.001 and P = 0.0017, respectively). Both DNA fragmentation and LDH release were observed within 6 h after the addition of 1 mM hydrogen peroxide to the cell culture system, indicating that both apoptotic and necrotic cell death had occurred. This mixture of cell death pathways was confirmed by observed changes in cellular morphology (Fig. 2B). The condensed and fragmented nuclei stained with acridine orange alone (green color, indicated in Fig. 2B by arrows) are characteristic of apoptotic cell death. The enlarged nuclei stained with ethidium bromide (red or orange color in Fig. 2) are characteristic of necrotic cell death (4).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Dose dependence of hydrogen peroxide cytotoxicity to bovine aortic endothelial cells (BAEC). BAEC were cultured with indicated concentrations of hydrogen peroxide for 6 h. BAEC and culture supernatants were then collected, and DNA fragmentation and lactate dehydrogenase (LDH) release were assessed. Each data point is mean ± SE for 6 experiments. * Significantly different from control.

View larger version (169K): [in this window] [in a new window]   Fig. 2.   Morphology changes in BAEC after culture with hydrogen peroxide and modified hemoglobins. BAEC were incubated for 6 h and then viewed under a fluorescent microscope after staining with acridine orange (green) and ethidium bromide (red/orange) as described in MATERIALS AND METHODS. A: control BAEC cultured with fresh medium. B: BAEC cultured with 1 mM hydrogen peroxide. C: BAEC cultured with 200 µM -cross-linked hemoglobin (-DBBF). D: BAEC cultured with 1 mM hydrogen peroxide and 200 µM -DBBF. Arrows in B-D indicate endothelial cells with apoptotic morphology (condensed nuclei with or without ethidium staining). Arrow in C, top right points to apoptotic cell that has detached from cell culture plate.

Direct effects of Hb on BAEC. Incubation of BAEC with any of the three Hb preparations (over concentration range 20-200 µM) for 6 h consistently produced a <20% increase in DNA fragmentation and LDH, values that are below the sensitivity of these assays to detect significant cytotoxicity under our conditions. Consequently, we observed no significant increase in cytotoxicity compared with control BAEC incubated in the absence of added Hb. The morphology changes induced by -DBBF in BAEC were characteristic of all three cell-free Hb preparations and indicated that <15% of the BAEC had undergone apoptotic cell death (Fig. 2C). Those BAEC that had undergone apoptotic cell death had rounded up or detached from the culture dish and exhibited condensed or fragmented nuclei stained with acridine orange. These BAEC also excluded ethidium bromide, which represents an earlier stage of apoptosis in which the plasma membrane is still intact. The morphology changes induced by HbA₀ and poly--DBBF were identical to that seen with -DBBF (data not shown).

Cytotoxicity of Hb and peroxide mixtures. The cytotoxic effects of Hb and hydrogen peroxide mixtures at several different concentrations of hydrogen peroxide, ranging from 1:1 to 1:20 molar ratios of heme to peroxide, were examined (Fig. 3). Mixtures of hydrogen peroxide and HbA₀ containing subcytotoxic concentrations of hydrogen peroxide (0.2 mM) were not cytotoxic. At higher concentrations of hydrogen peroxide, which gave nearly complete cytotoxicity at 24 h, the addition of HbA₀ completely inhibited cytotoxicity. -DBBF and poly--DBBF appeared to have a small inhibitory effect on the DNA fragmentation and LDH release induced by 1 mM hydrogen peroxide, but this apparent decrease was not statistically significant. Like HbA₀, -DBBF and poly--DBBF did not enhance cytotoxicity at subcytotoxic levels of hydrogen peroxide.

View larger version (36K): [in this window] [in a new window]   Fig. 3.   Hemoglobin effects on hydrogen peroxide-induced cytotoxicity toward BAEC. BAEC were cultured for 6 (A and C) and 24 (B and D) h without (open bars) and with 0.2 (hatched bars), 1 (crosshatched bars), and 2 (filled bars) mM hydrogen peroxide in absence and presence of 200 µM of nonmodified hemoglobin (HbA0), -DBBF, and polymerized -DBBF (poly--DBBF). DNA fragmentation (A and B) and LDH release (C and D) were assessed and expressed as a percentage of total cellular DNA or LDH, respectively. Each bar represents mean ± SE of 3 experiments. * Significantly different (unpaired Student's t-test) from those without added hemoglobin (No additions) at same hydrogen peroxide concentration.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Dose dependence of -DBBF on hydrogen peroxide cytotoxicity. BAEC were incubated for 24 h without or with 1 mM hydrogen peroxide and increasing concentrations of -DBBF. DNA fragmentation is expressed as a percentage of total DNA. Each data point is mean ± SE of 4 experiments. * Concentrations of -DBBF at which combination of -DBBF and hydrogen peroxide induced a significantly greater amount of DNA fragmentation than -DBBF alone (unpaired Student's t-test).

View larger version (14K): [in this window] [in a new window]   Fig. 5.   Spectral analysis of ferryl--DBBF. A: spectra were recorded for a single sample of -DBBF before addition of hydrogen peroxide (Oxy Hb), 1 min after addition of a 10-fold molar excess of hydrogen peroxide (Ferryl Hb), and 60 min after addition of hydrogen peroxide (Met Hb). B: spectrum recorded for a sample of ferryl--DBBF after chromatographic removal of residual hydrogen peroxide as described in MATERIALS AND METHODS.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Cytotoxicity correlates with ferryl--DBBF levels. In a representative experiment (1 of 3), -DBBF was incubated with hydrogen peroxide for indicated times at 37°C. -DBBF-hydrogen peroxide mixture was then added to BAEC and cultured for 24 h. Total heme concentration in culture medium was 100 µM. Percentage of heme in ferryl form () was determined spectrally as described in MATERIALS AND METHODS and represents a single determination. At end of 24-h culture, LDH release () and DNA fragmentation () were determined and expressed as a percent of total (mean ± SD, n = 5 replicates as described in MATERIALS AND METHODS). Inset, LDH release plotted as a function of ferryl--DBBF concentration.

Direct cytotoxicity of ferryl--DBBF. To rule out the possibility that residual hydrogen peroxide in the reaction mixture with -DBBF was responsible for the cytotoxicity, we prepared ferrylhemoglobin in which the hydrogen peroxide was removed by column chromatography. Spectral analysis of the Hb after removal of hydrogen peroxide showed that the preparation contained between 45 and 50% ferryl--DBBF (Fig. 5B). The removal of hydrogen peroxide enhanced the stability of the ferryl form. At the end of a 60-min incubation with hydrogen peroxide continuously present, ferryl--DBBF completely converted back to methemoglobin (Fig. 5A). In contrast, reaction of -DBBF with hydrogen peroxide followed by chromatographic removal of the hydrogen peroxide yielded a preparation of ferryl--DBBF in which <5% of ferryl--DBBF converted to methemoglobin after 60 min. This ferryl--DBBF preparation showed a high degree of cytotoxicity toward BAEC as assessed by DNA fragmentation (Fig. 7). This preparation was consistently two- to threefold more cytotoxic than oxy--DBBF alone. Ferryl--DBBF-induced cytotoxicity was caused entirely by apoptosis, because LDH release was not significantly increased above control levels (18.2 ± 1.6 vs. 14.5 ± 1.3%, without and with 100 µM ferryl--DBBF, respectively; n = 3). Morphological examination of the BAEC revealed that the ferryl--DBBF had induced apoptotic cell death in the BAEC (Fig. 8), consistent with apoptotic cell death induced by the -DBBF and hydrogen peroxide mixture (Fig. 2D). This further supports a role for ferrylhemoglobin as a mediator of cytotoxicity under these conditions.

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Ferryl--DBBF-induced cytotoxicity in BAEC. DNA fragmentation was determined after incubation of BAEC with indicated concentrations of ferryl--DBBF (residual hydrogen peroxide removed; ) for 24 h. In separate experiments, DNA fragmentation after 24-h incubation of BAEC with untreated -DBBF () was determined for comparison of cytotoxicity. Each experiment is mean ± SE of 3 and 4 experiments for ferryl--DBBF and nontreated -DBBF, respectively.

View larger version (202K): [in this window] [in a new window]   Fig. 8.   Morphology changes in BAEC after culture with ferryl--DBBF. BAEC were incubated with 50 µM ferryl--DBBF for 6 h and then viewed under a fluorescent microscope after being stained with acridine orange (green) and ethidium bromide (red/orange) as described in MATERIALS AND METHODS. Arrows indicate endothelial cells with apoptotic morphology.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The endothelial cells lining the vasculature are particularly sensitive to oxidative stress (41). When endothelial cells are damaged, normal vasculature control of coagulation, vessel tone, and vessel permeability are disrupted (17). Several in vitro studies have demonstrated that coculture of Hb with endothelial cells generates an oxidative stress. Incubation of endothelial cells with Hb concentrations <100 µM induces the biosynthesis of protective enzymes (7, 42), whereas Hb concentrations >100 µM induce cell death (39, 52). These studies have identified several potential candidates for the agent(s) that is responsible for the Hb-induced endothelial cell cytotoxicity but have not directly established the critical factors.

Cellular toxicity induced by cell-free Hb appears to be a consequence of its oxidation state. In the absence of added oxidants, HbA₀ induced <20% increase in endothelial cell cytotoxicity, in agreement with results obtained in other laboratories using stroma-free Hb (6, 52). However, Motterlini et al. (39), using HbA₀, reported much higher levels of cytotoxicity in porcine aortic endothelial cells in the absence of serum. Methemoglobin (Fe³⁺), formed by the oxidation of HbA₀, induces greater cytotoxicity than oxyhemoglobin (Fe²⁺) (52). Inhibitors of iron oxidation such as deferoxamine (16, 40) and scavengers of reactive oxygen species such as superoxide dismutase and catalase (43) protect against Hb-induced cytotoxicity and thus support a role for oxidation-reduction reactions. The release of free heme or iron after oxidation of Hb and the subsequent uptake of the iron by endothelial cells may also be a necessary part of the cytotoxicity process. Uptake of free heme is cytotoxic to endothelial cells in that preincubation with Hb or free heme sensitizes endothelial cells to hydrogen peroxide-induced cytotoxicity (7, 9). Intracellular heme iron could then catalyze the formation of several cytotoxic and highly reactive oxygen and nitrogen species from superoxide, hydrogen peroxide, and nitric oxide endogenously produced by endothelial cells (17, 33).

At the lower heme concentrations used in our experiments (20 µM), all three cell-free Hb preparations attenuated hydrogen peroxide-induced cytotoxicity, even concentrations of hydrogen peroxide higher than those likely to be produced during inflammatory reactions (26). This effect is likely caused by the pseudoperoxidase activity of these proteins (31). Hb consumes hydrogen peroxide in a series of reactions that involve cycling of Hb between the methemoglobin form (Fe³⁺) and the ferrylhemoglobin form (Fe⁴⁺) (23). The pseudoperoxidase activity of hemoproteins that involves the intermediacy of ferryl heme is well recognized and has been shown to be a real phenomenon in biological systems (19, 45). Modified Hb exhibit a slower reaction rate for conversion of ferrylhemoglobin back to methemoglobin, but they still have a sufficiently active peroxidase activity (20) to protect the endothelial cells against cytotoxicity from hydrogen peroxide.

We showed previously (2, 48) that oxidation of both -DBBF and poly--DBBF by hydrogen peroxide gives rise to a more stable ferryl intermediate that persists for long periods of time and that could potentially contribute to the cytotoxicity of the Hb. Comparison of HbA₀ with chemically modified Hb revealed that at higher concentrations of these modified Hb the production and persistence of ferrylhemoglobin in solutions generated a distinct mode of cytotoxicity that offset its protective effects. The importance of ferrylhemoglobin as a mediator of cytotoxicity was supported by the correlation between cytotoxicity and the concentration of ferrylhemoglobin present at the start of the incubation and the cytotoxicity induced by ferrylhemoglobin after chromatographic removal of hydrogen peroxide.

The transient nature of ferrylhemoglobin has made it difficult to definitively show that ferrylhemoglobin mediates cytotoxicity. Ferrylheme iron and ferrylhemoglobin have been shown to catalyze the peroxidation of lipids in cell-free systems (22, 27, 48). In one study, -DBBF was more effective than HbA₀ in the peroxidation of lipid vesicles (48). Studies on hydrogen peroxide cytotoxicity toward endothelial cells have proposed that the formation of ferryl iron within the cell, even in the absence of exogenously added Hb, was responsible for mediating hydrogen peroxide cytotoxicity (33). The stability of the ferryl iron in -DBBF has provided an excellent tool for demonstrating the direct cytotoxic effects of ferrylhemoglobin and has demonstrated a good correlation with the levels of ferrylhemoglobin and the extent of cell death induced by these preparations.

The reaction of Hb/myoglobin with hydrogen peroxide involves a two-electron oxidation process. One oxidation equivalent has been shown to reside on the globin as a transient radical located on an aromatic amino acid residue, whereas the second, longer lived oxidizing radical is believed to be the oxyferryl complex (Fe⁴⁺==O) (22, 23, 25). The globin-based radical and ferrylhemoglobin were recently detected by electron paramagnetic resonance in normal human blood (54). In this study the authors suggest that the source of peroxide in blood required for the reaction was the dismutation of superoxide produced via the autoxidation of intraerythrocytic Hb. It is interesting to note that this reaction occurred despite the presence of normal blood-reducing mechanisms (54). Under conditions of ischemia and reperfusion, patients have a diminished ability to control the oxidative reactions of Hb, thus raising the concern that the use of cell-free Hb as a blood substitute may in fact lead to an in vivo production and persistence of ferrylhemoglobin in tissue. Ferrylheme is able to oxidize critical targets such as unsaturated fatty acids or membrane lipoproteins and to contribute to postischemic reperfusion tissue injury (47). This hypervalent heme species has, therefore, all the attributes necessary to be considered a potential mediator of free radical damage to endothelial cells. Our data provide the first evidence that a peroxide-mediated oxidation of chemically modified Hb produces a persistent ferrylheme that is capable of inducing toxicity in endothelial cells in culture and that, unlike the case of native Hb, intermolecular and intramolecular cross-linking may have locked the proteins into a conformation that is more susceptible to the formation of a stable ferryl radical (12, 47). This work may also have some relevance to current research efforts by many in the field to suppress or control Hb oxidation reactions. Site-directed mutagenesis has recently been used to suppress the autoxidation reactions of myoglobin, a blood substitute prototype, and its reactions with nitric oxide, an important biological transduction molecule (18a). Other, more direct chemical strategies are aimed at cycling ferrylheme back to ferric heme, either by simulating a catalase-like activity in hemoproteins using an active redox compound such as nitroxides (32) or by addition of Trolox, an analog of vitamin E (24, 45).