Hemin causes mitochondrial dysfunction in endothelial cells through promoting lipid peroxidation: the protective role of autophagy

Abstract

The hemolysis of red blood cells and muscle damage results in the release of the heme proteins myoglobin, hemoglobin, and free heme into the vasculature. The mechanisms of heme toxicity are not clear but may involve lipid peroxidation, which we hypothesized would result in mitochondrial damage in endothelial cells. To test this, we used bovine aortic endothelial cells (BAEC) in culture and exposed them to hemin. Hemin led to mitochondrial dysfunction, activation of autophagy, mitophagy, and, at high concentrations, apoptosis. To detect whether hemin induced lipid peroxidation and damaged proteins, we used derivatives of arachidonic acid tagged with biotin or Bodipy (Bt-AA, BD-AA). We found that in cells treated with hemin, Bt-AA was oxidized and formed adducts with proteins, which were inhibited by α-tocopherol. Hemin-dependent mitochondrial dysfunction was also attenuated by α-tocopherol. Protein thiol modification and carbonyl formation occurred on exposure and was not inhibited by α-tocopherol. Supporting a protective role of autophagy, the inhibitor 3-methyladenine potentiated cell death. These data demonstrate that hemin mediates cytotoxicity through a mechanism which involves protein modification by oxidized lipids and other oxidants, decreased respiratory capacity, and a protective role for the autophagic process. Attenuation of lipid peroxidation may be able to preserve mitochondrial function in the endothelium and protect cells from heme-dependent toxicity.

the toxicity of free heme has been documented in several disease states. Hemolytic anemias such as sickle cell disease (SCD) and thalassemia result in red blood cell lysis and subsequent release of heme from hemoglobin (37, 42, 53, 70, 87). Hemin has also been suggested to contribute to atherosclerosis, since regions of high turbulence could contribute to red blood cell destruction (64). Physiologically, free heme concentrations in the blood are maintained at low levels (0.1–1 μM) (6) by the high binding affinity of proteins such as serum albumin, hemopexin, and haptoglobin (24, 35, 49, 54, 65, 77). Damage from heme release is also limited by heme oxygenases, which catalyze heme degradation (43). Iron released from heme is then sequestered by iron regulatory proteins such as ferritin and transferrin (82). Tight control and detoxification of free heme and iron is then critical since both either free iron or heme react with hydrogen peroxide to generate highly oxidizing species capable of initiating lipid peroxidation (29).

Pathologically, high levels of heme and iron release occur in acute conditions such as severe hemolytic crisis in sickle cell disease (up to 20 μM heme) (8, 37), or thalassemia (50–280 μM hemin) (70, 76). Vascular-related pathology is seen in sickle cell anemia and can include vascular occlusion due to platelet activation leading to hypercoagulation, which can lead to thromboembolism (3, 9, 62) and decreased endothelial function (1, 2, 51). Iron-related cardiac disease is a leading cause of death in the case of thalassemia (10, 20). In addition to hemolytic diseases, heme release can occur in the vasculature in areas vulnerable to shear stress (41, 80). This is interesting since these are the same areas prone to lesion formation in atherosclerosis. Indeed, some have postulated that hemin could be an important mediator of lipid peroxidation and damage during atherogenesis (4).

Hemin is also used pharmacologically in the treatment of porphyria, and in this case the heme is prepared in a sterile saline solution and introduced intravenously into the patients (79). This results in hemin concentrations up to 100 μM heme (79), which has the potential to overwhelm the body's capabilities to bind and metabolize free heme. The side effects of this treatment include phlebitis (46, 79) and coagulopathy characterized by thrombocytopenia (59).

The toxic effects of free heme and heme proteins is well appreciated with evidence that it scavenges nitric oxide (NO) (23, 67, 72) decreasing blood flow (74) and further exacerbating the severity of ischemic episodes and potential damage to the endothelium. Although it is clear that hemin can be cytotoxic to the endothelium (59), the mechanisms involved remain unclear. Several possibilities include direct activation of proapoptotic cell death pathways or the secondary production of reactive species and damage to cellular proteins leading to cellular dysfunction.

The pro-oxidant effects of heme have been demonstrated in a number of clinical and animal models. For example, patients with increased free heme and iron overload have increased levels of lipid hydroperoxides and lipid oxidation products such as malondialdehyde (MDA) (4, 70, 87) and low levels of serum antioxidants such as ascorbate and α-tocopherol (52, 58). One mechanism through which heme has been suggested to be toxic to the endothelium is by initiating the oxidation of LDL (low density lipoprotein), as supported by studies in cell and animal models (4, 55). However, the mechanism by which oxidized lipids mediate this cytotoxicity in the endothelium is less well understood. We hypothesized that hemin would induce lipid peroxidation and generate highly reactive lipid peroxidation products, which are capable of modifying proteins critical in cell survival.

We and others have shown that reactive lipid species are capable of inhibiting respiration in intact cells and isolated mitochondria (27, 83). We reasoned that because the mitochondria is a major intracellular source of both hydrogen peroxide and high levels of unsaturated phospholipids, which are susceptible to lipid peroxidation, then mitochondrial dysfunction would be a primary mechanism mediating hemin-dependent cell death. Furthermore, we hypothesized that the autophagy-lysosomal system is an important mechanism by which proteins damaged by hemin-induced reactive lipid species are processed within the cell. Autophagic degradation of lipid-protein adducts has been shown to be critical in the removal of proteins modified by 4-hydroxynonenal (4-HNE) (28), but it is not known if autophagy is activated by hemin and protects against mitochondrial damage in endothelial cells. In this article, we investigated whether autophagy may be an important means of protecting the mitochondria from hemin-induced damage.

In the present study, we determined whether hemin causes the formation of reactive lipid peroxidation products, which can modify proteins. We tested the hypothesis that hemin induces endothelial cell death by causing mitochondrial dysfunction and that post-translational modification by reactive lipid species contributes to this toxicity. Furthermore, we demonstrate activation of mitophagy and a potential role of autophagy in protecting endothelial cells from hemin-induced cell death.

MATERIALS AND METHODS

Reagents.

Hemin chloride, oligomycin, antimycin A, FCCP, and MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) were all obtained from Sigma Aldrich (St. Louis, MO) and were of the highest quality offered. JC-1 dye (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolyl-carbocyanine iodide), Lipofectamine 2000, and Mitotracker Deep Red were purchased from Invitrogen (Carlsbad, CA). Mito-RFP (fluorescent protein-based reagent containing the leader sequence of E1alpha pyruvate dehydrogenase fused to TagRFP) was purchased from Invitrogen. Protease inhibitor cocktail complete mini was from Roche (Indianapolis, IN). HEPES, sucrose, potassium chloride, and EGTA were obtained from Fisher Scientific (Fairlawn, NJ). BODIPY-arachidonic acid (BD-AA) was synthesized as described (25) and suspended in an ethanol vehicle for cell treatments. Antibodies against caspase 9, cleaved caspase 3, and β-actin were purchased from Cell Signaling Technologies (Boston, MA). LAMP-1 antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). MnSOD antibody was purchased from Enzo Life Sciences (Farmingdale, NY). Complex IV antibody was purchased from Invitrogen. VDAC antibody was obtained from MitoSciences (Eugene, OR). 3-Methyladenine (3-MA) and LC3 antibody were purchased from Sigma Aldrich. Full range rainbow molecular weight markers and secondary antibodies were obtained from GE Life Sciences (Buckinghamshire, UK). Full range rainbow molecular weight markers and secondary antibodies were obtained from GE Life Sciences. GFP-LC3 construct was a generous gift from Noburo Mizushima, and the RFP (mCherry)-LC3 construct from Xuejun Jiang.

Preparation of hemin stocks.

Hemin can be introduced into the circulation through a number of mechanisms where it is then capable of initiating oxidation reactions which damage cellular components including the endothelium (4, 8). Once released from a heme protein, hemin participates in a series of equilibrium reactions forming dimers and other aggregates (11, 12, 84). This can be effectively modeled in cell culture experiments through the preparation of a monomeric form, which can then interact with the biological milieu.

A 10–25 mM stock of hemin chloride was prepared on the day of use for each experiment. DMSO was used as a vehicle control, such that the same volume of vehicle was added to all treatment groups. Hemin chloride was dissolved in DMSO and verified to be in the monomeric form by measuring the absorbance at 406 nm (ϵ = 170 mM/cm) (73). Hemin was added directly from the DMSO stock to the medium before addition to cells.

Cell culture and treatments.

Bovine aortic endothelial cells (BAEC) were isolated as reported (22) and grown in DMEM containing 10% FBS, 4 mM L-glutamine, 100 units of penicillin, and 100 μg streptomycin. All treatments were performed in reduced serum DMEM (0.5% FBS) unless otherwise noted. To eliminate variability in mitochondrial function seen with cell aging, only BAEC from passages 5–8 were used for the purposes of this study.

Cell viability assays.

Cell viability was determined using lactate dehydrogenase (LDH) or 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. LDH activity in cell lysates and medium was determined by monitoring the rate of oxidation of NADH at 340 nm. The percent cytotoxicity was determined by dividing the rate of change in the absorbance in the medium by the combined LDH activity in the medium and lysate. Cell viability was determined by the MTT proliferation assay after treatment with hemin for 4–24 h as noted. Briefly, the reduction of MTT dye was monitored at 550 nm as an indicator of viable cells. The values were expressed relative to control, which was represented as 100% viability.

Measurement of cellular ATP by HPLC.

Nucleotide extraction and HPLC analysis were performed by a method described previously (69). Briefly, cell lysates were obtained from BAEC following treatment by scraping the cells after addition of 5% perchloric acid. This lysate was centrifuged (20,800 g for 10 min at 4°C), and the precipitated protein pellet was resuspended in 1 ml of 0.5 M NaOH. Protein concentrations were determined by the Bradford assay. Supernatant was transferred into an Eppendorf tube and neutralized by precipitating ClO₄ with K₂HPO₄. The suspension was vortexed, kept on ice for 10 min, and then centrifuged (as above) to remove salt. The supernatant was used immediately or stored at −80°C until analysis.

Measurement of Δψ (mitochondrial inner membrane potential) using JC-1.

The mitochondrial membrane potential was assessed using JC-1 dye. Briefly, BAEC grown in 48-well plates were serum deprived overnight in DMEM containing 0.5% FBS. Cells were treated for 30 min or 4 h with hemin in reduced serum culture medium. Following treatment, cells were washed with medium and incubated with 7.7 μM JC-1 dye for 30 min. Cells were then washed with PBS, and red/green fluorescence was measured using a fluorescence plate reader with excitation/emission filters suitable for rhodamine (560/595 nm) and fluorescein (484/535 nm). Results are expressed as the ratio of red to green fluorescence.

Western blot analysis.

Cell lysate proteins were separated by SDS-PAGE (10% or 12.5% gels) and transferred to nitrocellulose membranes. Protein amounts were determined using the Bradford method, and equal protein amounts were loaded. Uniform protein loading was verified by Ponceau S staining of Western blots. Membranes were blocked in Tris-buffered saline with 0.05% Tween 20 (TBST) containing 5% nonfat dry milk powder for 1 h. Western blots were probed with primary antibodies overnight, washed three times with TBST, and then incubated with the appropriate secondary antibodies for 2 h. Membranes were then washed with TBST three times, before developing with SuperSignal West Dura chemiluminescent substrate.

Detection of lipid-protein adducts.

To detect lipid-protein adducts in cells treated with Bt-AA or BD-AA, proteins from lysates were separated using 10% SDS-PAGE with nonreducing conditions. To detect lipid-protein adducts with Bt-AA, proteins were transferred to nitrocellulose and blots were probed with Streptavidin-HRP. The band intensity was calculated using AlphaEase imaging software. To detect lipid-protein adducts with BD-AA, following electrophoresis, proteins bound to reactive products of BD-AA were visualized by in-gel fluorescence imaging of the BODIPY signal (blue laser, 520BP40 filter setting) using a Typhoon fluorescence scanner. The fluorescent signal intensity for each lane was determined using Image Quant analysis software.

A pulldown procedure using neutravidin was used to enrich for proteins modified by oxidized Bt-AA in whole cell lysates from BAEC treated with 10 μM Bt-AA and 25 μM hemin. Cells were lysed in a 10 mM Tris (pH 7.4) buffer containing 1% Triton, 10 μM DTPA, and protease inhibitor cocktail. After cell lysis, protein concentrations were determined using the Bradford method and 500 μg of whole cell lysate were used per affinity precipitation reaction. Neutravidin beads (100 μl of slurry per reaction) were prewashed with 20 mM Tris-HCl, pH 7.4, containing 10 μM DTPA (Tris-DTPA buffer) six times. Cell lysates were mixed with neutravidin beads and incubated overnight at 4°C on a rotator mixer. Beads were then washed with 500 μl of 0.1 M glycine pH 2.8 six times, followed by three washes with 500 μl of Tris-DTPA buffer to remove unbound proteins. Proteins bound to the avidin resin were eluted by heating the beads to 70°C in heat block for 10 min in 100 μl of 2X laemmli buffer containing β-mercaptoethanol. Samples were centrifuged at 16,595 g for 10 min at 25°C, and supernatants were used for further analysis.

Transfection with GFP and RFP constructs.

For experiments using GFP or mCherry-LC3 (2 μg DNA per 1 × 10⁶ cells), BAEC were transfected in serum free complete DMEM for 6 h. Following transfection, media was changed to 10% FBS complete DMEM for 24 h. Cells were then trypsinized and plated at 5 × 10⁴ per well in a 4-well chambered coverslip. The following day, BAEC were treated with compounds as indicated and imaged using live cell fluorescence confocal microscopy. All treatments were performed in triplicate, and representative images were chosen for figures.

For experiments using both RFP-mito and GFP-LC3, BAEC were transfected with RFP-mito (Cell lights; Invitrogen) at 40 particles per cell (PPC) overnight, followed by transfection with GFP-LC3 (2 μg DNA per 1 × 10⁶ cells) for 6 h. The media was then changed to 10% FBS complete DMEM for 24 h. Cells were trypsinized and plated at 2.5 × 10⁴ per well in a 8-well chambered coverslip. BAEC were trypsinized and plated onto glass coverslips for imaging 24 h before experiment. BAEC were treated with DMSO vehicle or 25 μM hemin for 3 h. Cells were imaged using a confocal fluorescence microscope.

Mitochondrial isolation.

Following treatment, mitochondria were isolated from BAEC using digitonin and nagarse. Briefly, one confluent 6-well plate was used for each preparation. Cells were scraped in 1 ml of mitochondrial isolation buffer containing 50 mM HEPES, 0.1 M sucrose, 0.1 M potassium chloride, and 1 mM EGTA. Digitonin was added to cell suspension to a final concentration of 346 μM. Tubes were mixed by inversion, and suspension was centrifuged at 500 g for 1 min. Supernatant was discarded and pellet was reconstituted in 500 μL mitochondrial isolation buffer. Nagarse (20 μg) was added to each sample, and samples were mixed by inversion for 30 s. BSA (4 mg) was added to each sample to quench the proteolytic activity of nagarse. Each sample was centrifuged at 10,000 g for 5 min at 4°C. The supernatant was discarded, and the pellet was washed in 500 μL mitochondrial isolation buffer containing 4 mg BSA. A final wash was performed in mitochondrial isolation buffer containing 1 mM PMSF and protease inhibitor cocktail. Suspension was filtered through a Bio-Rad microspin column to remove large debris, and flow through was centrifuged at 10,000 g for 10 min and supernatant removed. Final pellet was reconstituted in 25 μl of 10 mM Tris, 0.1% Triton containing 1 mM PMSF, and protease inhibitor cocktail. Pellet was extracted on ice for 10 min in this solution. Protein concentrations were determined using the Bradford Assay, and samples were loaded at equal protein concentrations onto a 10% SDS-PAGE gel. Adducts with BD-AA were detected by in-gel fluorescence using a Typhoon fluorescent scanner.

Extracellular flux assay.

Oxygen consumption rates were determined using an XF Extracellular Flux Analyzer (XF24) from Seahorse Biosciences (North Billerica, MA). BAEC were plated at an optimum seeding density of 40,000 cells/well in specialized XF24 tissue culture plates 24 h before the assay for all experiments. Cells were washed with unbuffered DMEM before assay and allowed to equilibrate in this medium for 1 h before performing assay. All extracellular flux assays were performed in unbuffered medium. The conditions for determining mitochondrial function in BAEC with the XF24 analyzer were described previously by our laboratory (15).

Statistical analysis.

Results are reported as means +/- SEM. The Student's t-test was used to compare values between two treatment groups. A P value of < 0.05 was considered statistically significant.

RESULTS

Hemin causes cell death and decreases mitochondrial membrane potential in BAEC.

Following treatment with hemin (10–50 μM) for 24 h, LDH release was detected at 25 and 50 μM hemin indicating significant cell death at these concentrations (Fig. 1A). As an additional high throughput measure of both cell viability and mitochondrial function, reduction of the dye MTT by BAEC was used at the 4- and 24-h time points (Fig. 1B). At the 4-h time point, 50 μM hemin showed a significant decrease in MTT reduction, whereas at the 24-h time point, both 25 and 50 μM showed a profound decrease in signal. LDH release measures late apoptosis and necrosis, so to determine whether the mechanism of cell death was due to apoptosis, we measured caspase 3 and -9 after 4 h. Procaspase 9 was found to be significantly depleted at 25 and 50 μM hemin, and the appearance of cleaved caspase 9 and -3 was also seen at these concentrations, consistent with cell death being initiated though the mitochondrial intrinsic cell death pathway (Fig. 1C). We therefore hypothesized that hemin causes mitochondrial damage before activation of the cell death pathway. At lower concentrations hemin can induce cytoprotective pathways through induction of the synthesis of heme oxygenase-1 (HO-1) (5). To determine whether this is occurring our conditions cells were treated with hemin for 4 h and the levels of HO-1 determined by Western blotting (Fig. 1D). As expected HO-1 was induced by hemin, but this was largely attenuated by exposure to the highest concentrations at which cell death occurred. Apoptosis requires ATP availability; otherwise necrotic cell death is initiated (60). To test for the effect of hemin on ATP, cells were incubated with the oxidant for 3 h before lysis in percholoric acid and measurement of ATP by HPLC. As shown in Fig. 1E only at the highest concentration of hemin (50 μM) were ATP levels decreased, whereas cell death was observed at both 25 and 50 μM (Fig. 1, A and B), suggesting sufficient ATP is available for apoptotic cell death. Changes in mitochondrial membrane potential are indicative of mitochondrial dysfunction and also early cell death. We measured mitochondrial membrane potential using JC-1 dye. As seen in Fig. 1F, hemin decreased the ratio of red to green fluorescence after 4 h in a dose-dependent manner. This indicates that hemin decreases mitochondrial membrane potential in BAEC.

Fig. 1.

Hemin causes apoptotic cell death and decreases mitochondrial membrane potential in bovine aortic endothelial cells (BAEC). A: cell death determined using lactate dehydrogenase (LDH) release assay in BAEC treated with hemin for 24 h. B: cell viability determined using the (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT) assay 4 or 24 h after hemin treatment. C: caspase 9 (pro and cleaved) and cleaved caspase 3 detected using immunoblot of lysates from BAEC treated with hemin for 4 h. D: heme oxygenase-1 (HO-1) protein levels determined using immunoblot of BAEC lysates from cells treated for 4 h with hemin. E, right: effects of hemin on mitochondrial membrane potential was determined by 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolyl-carbocyanine iodide (JC-1) fluorescence after 30 min or 4 h treatment with hemin in BAEC. E, left: ATP content in BAEC was measured following 2 h treatment with hemin. BAEC were treated with DMSO (vehicle control) or 10–50 μM hemin for 2 h before harvesting lysates. Lysates were analyzed by HPLC against an internal standard to quantify the amount of ATP. *P < 0.05 vs. control; n ≥ 3. AU, arbitrary units. #P ≤ 0.05 vs. 10 µM hemin.

Hemin causes cellular bioenergetic dysfunction.

To determine whether decreases in bioenergetic capacity occur in response to hemin exposure (10–50 μM), we assessed oxygen consumption in BAEC treated for 3 h (before caspase activation is detectable). As shown in Fig. 2A, the metabolic profile was determined using inhibitors of mitochondrial function as we have described previously (15). As shown in Fig. 2, A and B, only the highest concentration of hemin caused a decrease in the basal oxygen consumption rate (OCR). Oligomycin was next added to the cells, which inhibits the ATP synthase and decreases OCR to the extent the cells utilize ATP. Hemin decreased the inhibitory effect of oligomycin, which is consistent with an increase in proton leak and a decrease in ATP linked respiration (Fig. 2, C and D). The maximal respiratory rate, determined after the addition of the uncoupler FCCP (1 μM), was decreased at 25 and 50 μM hemin (Fig. 2E). The reserve capacity, the difference between OCR in the basal immediately before oligomycin addition and maximal state, was decreased by hemin in a concentration-dependent manner (Fig. 2F). To elucidate whether mitochondrial dysfunction precedes apoptotic signaling, we next followed the time course for the loss of caspase and found that this only occurred after 4 h exposure to hemin (Fig. 3A), and this was confirmed also for caspase 3 (result not shown). To follow the time course of hemin-dependent effects on mitochondrial function hemin (25 μM) was injected onto the cells in the instrument and a mitochondrial function assay performed at 0.5, 1, 2, and 3 h exposure. Shown in Fig. 3B is the complete time course for control and hemin treated cells over 3 h. The initial effect of adding hemin is to transiently suppress basal respiration, which then slowly recovers by the 3-h time point consistent with Fig. 2. The earliest detectable change is an increase in ATP linked and proton leak, which is evident at 0.5 h (Fig. 3, C and D). Decreases in maximal and reserve capacity occur at later time points (Fig. 3, E and F).

Fig. 2.

Hemin impairs cellular bioenergetics. The reagents oligomycin (O), FCCP (F), and antimycin A (A) were added sequentially to determine the effect of hemin on specific mitochondrial function parameters. Oligomyocin was used as an inhibitor of ATP-synthase. FCCP, a mitochondrial uncoupler, was used to elicit maximal oxygen consumption through the mitochondria. Antimycin A, an inhibitor of Complex III, was added to inhibit the electron transport chain, allowing for the determination of oxygen consumption linked to the respiratory chain. A: BAEC were treated for 3 h with hemin, followed by a 1-h equilibration in assay medium (thus assay was performed 4-h post-initial hemin exposure). After this, oxygen consumption was measured using an extracellular flux analyzer as detailed in the methods. Representative BOFA assay is shown. B: basal oxygen consumption in BAEC following hemin treatment. C: oxygen consumption attributed to proton leak. D: ATP-linked oxygen consumption. E: maximal oxygen consumption determined using FCCP to stimulate oxygen consumption. F: oxygen consumption attributed to reserve capacity. G: nonmitochondrial oxygen consumption. *P < 0.05 vs. control; n ≥ 3.

Fig. 3.

Hemin induces mitochondrial dysfunction prior to apoptotic signaling. A: BAEC were treated with 25 μM hemin for 0–4 h before determination of caspase 9 by immunoblot. B: BOFA assay performed after injection of 25 μM hemin. The trace shown is from an experiment where hemin was injected (at arrow indicated by “H”), and basal oxygen consumption was measured over time for 3 h post-hemin injection. At the end of the assay, the compounds oligomycin (O), FCCP (F), and antimycin A (A) were injected to determine mitochondrial function parameters. This same protocol was also adapted to inject O, F, and A at 0.5, 1, and 2 h post-hemin injection (as marked with dashed lines, data not shown). With the use of this data, the mitochondrial parameters were determined at each time point (0.5, 1, 2, and 3 h) following 25 μM hemin treatment. C: ATP-linked oxygen consumption measured in DMSO vehicle treated control cells (○) and 25 μM hemin (■) at 0.5, 1, 2, and 3 h postinjection. D: proton leak. E: maximal oxygen consumption (FCCP-stimulated rate). F: reserve capacity. *P < 0.05 vs. control; n ≥ 3.

Hemin-induced lipid peroxidation results in protein adduct formation and cell death, which is inhibited by antioxidants.

Lipid peroxidation products such as HNE and isoprostanes have been shown to affect mitochondrial function (26, 27, 30, 34, 38, 40, 44). We have previously used biotinylated lipids to monitor protein adduct formation with reactive lipid species in cells (14, 25, 61). To determine whether hemin induces the formation of reactive lipid species we used a biotin-tagged derivative of arachidonic acid (Bt-AA). BAEC were treated with 10–50 μM hemin and 10 μM Bt-AA for 4 h. Proteins bound to biotinylated lipid species were detected using a Streptavidin-HRP Western blot (Fig. 4A). Protein adducts with Bt-AA were observed at 10–50 μM hemin as an increase in biotinylated proteins compared with the untreated control, but not in BAEC treated with Bt-AA alone (Fig. 4, A and B). The faint bands observed in the samples from BAEC treated with Bt-AA alone are similar to that seen in the control, which correspond to the endogenous biotin containing carboxylases. To determine whether these adducts could be reversed by a chain-breaking antioxidant, we pretreated BAEC with α-tocopherol (100 μM) for 1 h before treatment with Bt-AA and hemin. As shown in Fig. 4, A and B, α-tocopherol was effective at decreasing the protein modification by reactive lipid species seen at all concentrations of hemin.

Fig. 4.

Hemin induces protein modification by oxidized lipids, and cytotoxicity of hemin is inhibited by antioxidant pretreatment. A: BAEC were pretreated with α-tocopherol for 1 h before stress with hemin for 4 h. Reactive lipid species were visualized by probing for proteins attached to a biotin moiety using immunoblots probed with streptavidin HP. Representative Western blot image is shown. B: quantitation of signal from 3 immunoblots. C: viability of BAEC following pretreatment with antioxidants for 1 h and treatment with hemin for 16 h, determined using the MTT assay. *P < 0.005 vs. control; #P < 0.005 vs. hemin only; n ≥ 3.

Next we tested whether lipid peroxidation plays a role in the cytotoxicity of hemin using a structurally diverse range of antioxidants. The phenolic lipid peroxyl radical scavengers, α-tocopherol (0–100 μM), trolox (0–100 μM), and probucol (0–50 μM), were added to the cells 1 h before the addition of hemin (0–50 μM) and then viability measured by the MTT assay 16 h later. We found that all of the antioxidants inhibited hemin-induced cell death, and in Fig. 4C we show the concentrations that were most effective for each compound at 25 μM hemin. The differences in relative potency of the antioxidants are most likely due to the different accessibility to the cell. However, the ability of the antioxidants to inhibit cell death was essentially lost at 50 μM hemin. These data suggest that other hemin-dependent protein modifications may be occurring, which are resistant to lipid radical scavengers. To test this we measured both protein carbonyl formation and thiol oxidation in response to increasing concentrations of hemin after 4-h exposure (Fig. 5). Hemin exposure caused both protein carbonyl formation and thiol oxidation to increase in a concentration-dependent manner. Carbonyl formation, which also measures protein-lipid, adducts, was inhibited by a-tocopherol (Fig. 5, A and B), whereas the antioxidant had a minimal effect on protein thiol oxidation (Fig. 5, C and D).

Fig. 5.

Hemin induces oxidative protein modification. A: BAEC were pretreated with DMSO vehicle (•) or α-tocopherol (□) for 1 h before stress with hemin for 4 h. Cells were then treated with BODIPY-hydrazide as detailed in methods. Carbonyl adducts were visualized using in-gel fluorescence scanned with a Typhoon fluorescence imager. B: quantitation of adducts determined using ImageQuant software. C: BAEC were pretreated with α-tocopherol for 1 h before stress with hemin for 4 h. Lysates were incubated with BODIPY-IAM and separated using SDS-PAGE. Representative in-gel fluorescence image is shown. D: quantitation of total reactive protein thiol. *P < 0.05 vs. control; #P < 0.05 vs. hemin only; n = 3.

Reactive lipid species modify mitochondrial proteins.

To monitor the induction and location of intracellular lipid peroxidation by hemin we incorporated a fluorescent tag, BODIPY (BD), onto arachidonic acid through conjugation with the carboxyl group on the fatty acid (25). We have shown previously that this does not change the ability of the unsaturated fatty acid to be a substrate for lipid peroxidation (25, 39). BAEC were incubated with BD-AA and hemin for 2 h followed by treatment with mitotracker red (500 nM) after which confocal images were taken. Treatment of cells with BD-AA alone resulted in a diffuse signal with no clear association with mitochondria (Fig. 6A). On addition of hemin (25 μM) for 2 h, the intensity of the distribution of the BD-AA staining changed with the formation of punctuate areas of signal, some of which showed a strong colocalization with mitotracker, and others which were distinct (Fig. 6A, bottom).

Fig. 6.

Oxidized products of BODIPY arachidonic acid (BD-AA) localize to the mitochondria after hemin treatment. A: BAEC were treated with 10 μM BD-AA and 25 μM hemin for 2 h before imaging. Cells were washed with low serum media and costained with mitotracker red. Confocal images of BAEC after 2 h, imaged for BODIPY and mitotracker red fluorescence, are shown. Arrow denotes the area of field enlarged at bottom. B: BAEC were pretreated with 10 nM mitochondrially targeted antioxidant Mito-Q for 1 h, followed by 4 h treatment with hemin. Protein adducts were visualized using in-gel fluorescence of BODIPY moiety. C: quantitation of total BODIPY fluorescence intensity from all bands. *P < 0.05 vs. control; #P < 0.05 vs. hemin only; n = 3.

These data suggest lipid peroxidation could occur in the mitochondria, and to test this the mitochondrially targeted antioxidant, Mito-Q (10 nM), was added to BAEC 1 h before hemin (25 and 50 μM) treatment with BD-AA for 4 h, and protein adducts were measured. As shown in Fig. 6, B and C, some basal signal is seen with BD-AA alone; however, addition of 25 or 50 μM hemin increased protein-lipid adducts, and these were significantly attenuated by MitoQ. To test for the formation of protein adducts in mitochondria, a mitochondrial preparation was prepared from BAEC using treatment of cells with digitonin to remove the plasma membrane followed by exposure to the protease nagarse to remove nonmitochondrial proteins. The efficiency of this preparation was demonstrated using marker proteins and enzyme assays (Fig. 7, A–D). Beta-actin was used as a marker of cytosolic protein and was essentially absent in the mitochondrial preparation (Fig. 7A). Consistent with these data citrate synthase specific activity, a mitochondrial matrix enzyme, was increased, and LDH activity, a cytosolic enzyme, decreased (Fig. 7, B and C). With the use of Western blotting of Lamp-1 a marker of the lysosome and VDAC and Complex IV subunit I, MnSOD as markers for mitochondria it is clear that the preparation is enriched for mitochondria with minimal contamination from either lysosomal or cytoplasmic proteins. The mitochondrial fractions were next analyzed for protein adduct formation by BD-AA after cells were treated with hemin (4 h) with or without 1 h 100 μM α-tocopherol pretreatment. As found with the total cell lysate protein adduct formation using Bt-AA (Fig. 4, A and B), mitochondrial oxidized lipid-protein adducts were significantly increased by the combination of BD-AA and hemin and attenuated by α-tocopherol (Fig. 7, E and F).

Fig. 7.

Mitochondrial proteins are modified by oxidized lipids following hemin treatment. A: validation of mitochondrial isolation purity. Cells were permeabolized with digitonin followed by nagarse treatment. Intact mitochondrial proteins were localized to the high speed pellets. Western blots of lysates and mitochondria fractions are shown from no detergent (a), digitonin only (b), and digitonin/nagarse (c) probed with an antibody specific for β-actin. B: citrate synthase as measured by the generation of thionitrobenzoate. C: LDH activity measuring the rate of oxidation of NADH at 340 nm. D: Western blots of whole cell lysates (C) and mitochondria (M) fractions with antibodies specific for LAMP-1, voltage-dependent anion channel (VDAC), subunit 1 of cytochrome c oxidase, and MnSOD. E: BAEC were pretreated with 100 µM α-tocopherol for 1 h before hemin addition. After treatment with 10 µM BD-AA and 25 µM hemin for 4 h, mitochondrial fractions were isolated from BAEC. A representative in-gel fluorescence visualization of BODIPY signal from mitochondrial fractions separated by SDS-PAGE is shown. F: quantitation of BODIPY fluorescence normalized to amount of VDAC (mitochondrial marker protein). Data represent means ± SE; n = 3. *P ≤ 0.002 vs. whole cell lysates. **P ≤ 0.006 vs. BD-AA alone. #P < 0.05 vs. BD-AA + hemin.

To verify the modification of a specific mitochondrial protein, we performed a pulldown experiment for lipid modified proteins after treatment of cells with Bt-AA, with and without hemin (25 μM), and tested for the enrichment of the mitochondrial protein voltage-dependent anion channel, (VDAC). As shown in Fig. 8, VDAC was present in the cell lysate and the total levels in the cell were unchanged in any of the treatment groups. After pulldown with neutravidin resin, treatment with Bt-AA alone resulted in increased levels of VDAC compared with control, indicating a low level of mitochondrial lipid peroxidation occurring in these cells. This was greatly enhanced in the combined exposure of Bt-AA and hemin, indicating significant modification of VDAC through a lipid peroxidation-dependent mechanism (Fig. 8B).

Fig. 8.

VDAC is modified by reactive lipid species. Cells were treated with 10 μM biotinylatedarachidonic acid (Bt-AA) and 25 μM hemin for 4 h. Proteins bound to biotin were pulled down using neutravidin resin. A: Western blot of lysates, flowthrough, and pulldown fractions probed with an antibody specific for VDAC. B: quantitation of band intensity corresponding to VDAC. Data represents mean ± SE; n = 3. *P < 0.05 vs. Bt-AA treated.

Protection from hemin-dependent bioenergetic dysfunction using α-tocopherol.

To determine whether the lipid radical scavenging antioxidants can protect against mitochondrial damage the protocol for determining cellular bioenergetic function in Fig. 2 was used. BAEC were pretreated with α-tocopherol (100 μM, 1 h) before treatment with hemin for 3 h. After incubation with hemin, with and without α-tocopherol, the antioxidant and hemin were removed, cells washed with unbuffered DMEM and equilibrated in unbuffered DMEM for 1 h, and the mitochondrial function assay was performed (Fig. 9A). As shown before, hemin increased proton leak (Fig. 9C) and decreased maximal and reserve capacity (Fig. 2, E and F), and these effects were significantly attenuated by α-tocopherol. In this experiment a small but significant effect on basal respiration was noted, which was also inhibited by α-tocopherol.

Fig. 9.

Vitamin E (Vit E; α-tocopherol) attenuates mitochondrial dysfunction due to hemin in BAEC. BAEC were treated with 100 μM α-tocopherol (1 h) before 25 μM hemin (4 h). A: BAEC were washed with unbuffered DMEM to remove treatment reagents, and oxygen consumption over time was measured using the XF24 Extracellular Flux Analyzer. B: basal oxygen consumption. C: oxygen consumption rate attributed to proton leak. D: maximum OCR determined using the FCCP stimulated rate. E: reserve capacity calculated from the difference between the basal and maximal OCR rates for each group. *P < 0.005 control vs. hemin treated; #P < 0.05 vitamin E + hemin vs. hemin only; n = 3 to 4 per group.

Hemin causes BD-AA accumulation in cells and activates autophagy.

To determine whether hemin is inducing autophagy, lysates from BAEC treated with 25 μM hemin for 0–4 h were analyzed by Western blot for levels of LC3-I and LC3-II. As shown in Fig. 10A, LC3-I levels did not significantly change with 25 μM hemin treatment at any of the timepoints tested. However, LC3-II was found to be increased at 2–4 h after hemin treatment (Fig. 10A), suggesting autophagy is induced in endothelial cells treated with hemin after mitochondrial dysfunction is initiated (Fig. 3). As an additional assessment of autophagy, cells were transfected with GFP-LC-3 and treated with hemin (25 μM) for 3 h before analysis by confocal microscopy. As shown in Fig. 10B with representative cells the GFP-LC-3 is diffuse in the control cells but becomes punctate in the presence of hemin, similar to the response to the established inhibitor of autophagy, chloroquine. To test whether lipid protein adducts are targeted to the autophagosome, cells were transfected with mCherry-LC-3 and treated with hemin and BD-AA (green) for 3 h. Consistent with the GFP-LC3 result, numerous puncta were formed on addition of hemin (Fig. 10C). As shown before BD-AA-dependent signal shows a diffuse pattern, localization with a disrupted mitochondrial network, and puncta not associated with mitotracker. In this experiment it is clear that many of the hemin-induced autophagosomes contain lipid-protein adducts (Fig. 10C).

Fig. 10.

Hemin increases autophagosome accumulation in BAEC. BAEC were treated with 25 μM hemin for 1, 2, or 4 h. A: Western blot probed with anti-microtubule-associated protein 1 light chain (LC)3 antibody. Quantification of band intensity expressed as the ratio of LC3-II to LC3-I. *P < 0.05 vs. control; n = 3. B: confocal microscopy of green fluorescence protein (GFP)-LC3 transfected BAEC following treatment with DMSO control for 3 h (a), 25 μM hemin for 3 h (b), or 10 μM chloroquine (positive control) for 5 h (c). Arrows indicate cells undergoing autophagy. C: BAEC were transfected with mCherry-LC3 and then treated with 25 μM hemin for 3 h in the presence of 10 μM BD-AA. Cells were imaged by confocal microscopy. Representative images are shown.

Hemin induces mitophagy.

To determine whether mitophagy was occurring on exposure to hemin, mitochondria were tracked by transfection of the BAEC with RFP-mito (red fluorescent protein conjugated to pyruvate dehydrogenase) and autophagy by GFP-LC3 . A typical cellular response is shown in Fig. 11 in which the reticular mitochondrial network is apparent with RFP-mito. The GFP-LC3 signal was diffuse and did not colocalize with RFP-mito in control cells (Fig. 11A, top). Addition of hemin induced puncta formation, and some cells were found to undergo mitophagy. In these cells, the reticular pattern of the mitochondria was lost and the autophagosomes (seen with GFP-LC3) colocalized with the RFP-mito signal (Fig. 11A, bottom).

Fig. 11.

Hemin increases mitophagy in BAEC. A: BAEC were transfected with red fluorescence protein (RFP)-mito followed by transfection with GFP-LC3. BAEC were transferred to glass coverslips for imaging 24 h before experiment. BAEC were treated with DMSO vehicle or 25 μM hemin for 3.5 h. Live cells were imaged using a confocal fluorescence microscope. Arrow designates a cell undergoing mitophagy. B: LDH assay following 24 h cotreatment with 0–50 μM hemin and 1 mM. C: LDH assay from BAEC cotreated with 1 mM 3-methyladenine (3-MA) and 0–50 μM hemin for 3 h. Following treatment, the media containing compounds were removed and cells were allowed to incubate in low serum media for an additional 24 h before assay. *P < 0.05 vs. hemin only.

To determine whether the autophagic process is a protective mechanism in hemin toxicity, BAEC were cotreated with 1 mM 3-MA (3-methyladenine), an inhibitor of autophagy, and 25 μM hemin for 24 h. Following treatment, as observed before, 25 and 50 μM hemin caused significant cell death. Cotreatment with 3-MA did not potentiate this cell death. Short durations of hemin were found to not elicit cell death in these cells, so we hypothesized that autophagy may delay toxicity upon exposure to hemin. To test this, we treated BAEC with hemin for 3 h, removed the media, and then determined toxicity after 24 h. With the use of this protocol, hemin alone did not elicit cell death at any of the concentrations tested (Fig. 11B). However, in those cells treated concurrently with the autophagy inhibitor, 3-MA, significant cell death was observed (Fig. 11C). These data suggest that autophagy is initiated upon hemin exposure as a protective mechanism. However, if cells undergo extended exposure to hemin, this mechanism is not sufficient to prevent cell death.

DISCUSSION

The location of the endothelium within the vascular lumen makes it a prime target for toxins, xenobiotics, oxidants, and reactive metals in the circulation (4, 7, 71). Not surprisingly, the vasculature contains a number of cytoprotective pathways designed to prevent heme- and iron-dependent cytotoxicity such as hemopexin and the ability to induce HO-1 (6, 49). However, under numerous pathological conditions these pathways become overwhelmed, resulting in exposure to high local concentrations of hemin. Hemin may also contribute to cardiovascular disease since it has been shown to form covalent adducts with lipoproteins in human subjects (47, 48). This is interesting because in areas vulnerable to atherosclerotic lesions, the turbulent blood flow results in hemolysis and a regional increase in free heme (56, 64). In addition, patients with sickle cell anemia and β-thalassemia are at higher risk of cardiovascular complications (1, 20). Under conditions of severe skeletal muscle damage, the release of the heme protein myoglobin into the circulation results in increased lipid peroxidation, and the resulting isoprostanes derived from oxidation of arachidonic acid cause renal failure (32, 50). Furthermore, in cardiac ischemia-reperfusion myoglobin from the cardiomyocytes is released into the circulation and contributes to oxidative stress (17). Once released from the cellular environment heme proteins such as myoglobin and hemoglobin are less stable, and on reaction with hydrogen peroxide pro-oxidant derivatives of the proteins are made, which initiate lipid peroxidation (29, 68). As part of this process the heme prosthetic group, which is not covalently bound to the protein, dissociates and, because it is hydrophobic, partitions into cellular membranes (36). Although it has been shown that hemin is toxic to endothelial cells the molecular mechanisms and targets in the cell remain largely undefined (4). In this article, we evaluate the effects of hemin, an iron protoporphyrin, on mitochondrial function and one of the key cytoprotective pathways responsible for removing damaged proteins and organelles, the lysosomal-autophagy pathway.

In the first series of experiments we confirmed that hemin causes cytotoxicity in endothelial cells through an apoptotic process, and this is associated with a progressive decrease in mitochondrial membrane potential with some loss of ATP at the highest concentration tested (Fig. 1). These data, and those shown in Fig. 3, clearly indicate modification of mitochondrial function occurs early and precedes cell death. We found that hemin increased the cleavage of caspases 9 and -3 (Fig. 1C), indicating hemin causes apoptosis through the intrinsic pathway and that caspase 9 is active. Interestingly, only 30 min exposure to hemin caused a decrease in mitochondrial ATP linked respiration and increased proton leak (Fig. 3) which is consistent with a decrease in mitochondrial membrane potential. Interestingly, proton leak has been shown to be increased in isolated mitochondria by lipid peroxidation products such 4-hydroxynonenal, and we have observed the same response on exposure of endothelial cells to DMNQ, which produces hydrogen peroxide (15, 27, 66). In addition, the maximal bioenergetic capacity, assessed by the addition of the uncoupler FCCP, was also decreased at later time points. This parameter can be used to determine the reserve capacity, which is the available bioenergetic capacity that cells have available to combat oxidative stress. At higher concentrations of hemin, mitochondrial function is highly depressed, and at later time points (16–24 h) these cells are no longer viable (Figs. 1 and 2). These data indicate that the induction of apoptosis through the intrinsic pathway occurs after mitochondrial dysfunction become evident and at approximately the same time that increased autophagy can be detected (Fig. 10).

To test the hypothesis that lipid peroxidation is playing an essential role in mediating hemin-dependent cytotoxicity we used cells supplemented with tagged arachidonic acid and demonstrated the formation of protein lipid adducts, which was inhibited by lipid radical scavengers (Fig. 4). The formation of protein lipid adducts appears to be playing a key role in mediating cell death since three structurally distinct lipid radical scavengers attenuated the cytotoxicity of 25 μM hemin but at had little effect at 50 μM (Fig. 4). We hypothesized that at the higher concentration of hemin protein modification could be occurring through a lipid peroxidation independent pathway. To test this we measured both protein carbonyl formation, which measures both reactive oxygen species and lipid peroxidation-dependent protein modification, and protein thiol oxidation (Fig. 5). Protein carbonyl formation was significantly, but not completely, inhibited by α-tocopherol at all concentrations of hemin, whereas the antioxidant had minimal effects on protein thiol oxidation. These data suggest that several mechanisms of protein modification occur on exposure to hemin, ultimately leading to cell death. As a further confirmation of the link between lipid peroxidation, mitochondrial dysfunction, and cell death we found that α-tocopherol significantly inhibited the respiratory dysfunction induced by 25 μM hemin (Fig. 9). These data support that hypothesis that lipid oxidation plays an important role in the initial mitochondrial dysfunction and cell death caused by hemin in endothelial cells.

To determine whether lipid peroxidation products formed during 25 μM hemin treatment were targeted to a specific subcellular compartment, we used the fluorescently labeled fatty acid, BD-AA and followed the localization using the tracker dye Mitotracker red (Fig. 6A). After 2 h, BD-AA was found to colocalize with the Mitotracker dye. More BODIPY fluorescence was observed in the mitochondria with hemin treatment, suggesting the accumulation of oxidized lipids in this compartment. The ability of hemin to increase the formation of protein-reactive lipid species and decrease mitochondrial function suggested mitochondrial proteins may themselves be targets of reactive lipid species following hemin treatment. To address this question, we pretreated cells with the mitochondrially targeted antioxidant (13, 31), Mito-Q, before stress with hemin. This mitochondrially targeted antioxidant has been shown to inhibit lipid peroxidation in mitochondria both in vitro and in vivo (31). Hemin was found to increase protein adducts with BD-AA in whole cell lysates, and this protein modification was inhibited by 10 nM Mito-Q at both 25 and 50 μM hemin (Fig. 6, B and C). Further supporting modification of mitochondrial proteins by reactive lipids, in mitochondria isolated from BAEC, we found that hemin increased protein adducts with BD-AA. With the use of the same concentration of α-tocopherol found to protect mitochondrial function and cell viability, adducts with BD-AA were decreased to the same level as cells untreated with 25 μM hemin (Fig. 7, E and F).

A key event during intrinsic apoptosis is the formation of the mitochondrial permeability transition pore (MPT) and release of cytochrome c. Although the exact structure of the MPT is not known, VDAC has been suggested to be one component of the multi-protein complex. In addition, VDAC has been shown to be important for substrate availability to the mitochondria from the cytosol (86). The open and closed states of VDAC have also been shown to be tightly regulated with VDAC closure being favored early in apoptosis before cytochrome c release (16). To determine whether one of the mitochondrial protein targets of hemin-induced reactive lipid species was VDAC, we used Bt-AA and enriched for modified proteins using neutravidin resin. We found that the combination treatment of Bt-AA and hemin robustly increased the levels of modification of VDAC in the pulldown samples compared with the control or treatment of Bt-AA alone (Fig. 8). No changes in VDAC protein levels were detected in whole cell lysates or the flow through fractions, suggesting VDAC expression is not changed with hemin treatment, but rather protein modification by reactive lipid species is induced upon exposure to hemin. It is important to note that the toxic effects of hemin on mitochondrial function and cell death cannot be ascribed to VDAC modification alone.

Several of the more reactive lipid peroxidation products have been shown to inhibit proteasomal function, a predominant means of clearing adducted proteins in the cell (18, 63). An alternative mechanism of clearing modified proteins and organelles is through the autophagic-lysosomal pathway (19). We hypothesized that hemin treatment results in increased autophagic flux, and this may allow for the uptake of lipid-protein adducts. In Fig. 10A, we demonstrated that hemin increased the formation of LC3-II, a marker of autophagy over time, with a significant increase observed at 2–4 h. This was further validated by transfecting endothelial cells with a GFP-LC3 construct. Upon treatment with the positive control, chloroquine, or treatment with hemin, GFP-LC3 fluorescence changed from a diffuse pattern to distinct puncta (Fig. 10B). Autophagic vacuolization requires the presence of LC3 in the autophagosomal membrane (21, 33, 81), so these puncta correspond to the autophagic vacuole containing GFP-LC3. Once it was observed that hemin potently induced autophagy, further experiments were focused on determining whether lipid peroxidation products colocalize within autophagosomes. As was observed with GFP-LC3, mCherry-LC3 fluorescence formed clear puncta in hemin-treated BAEC but not in the control cells (Fig. 10C). Interestingly, BD-AA signal also appeared punctate in BAEC treated with hemin. Furthermore, when the green fluorescence from BODIPY and the red fluorescence from mCherry-LC3 were overlaid, a clear colocalization was observed in hemin-treated cells (Fig. 10C). As shown in Fig. 11 in which many of the autophagosomes colocalized with RFP-mito, mitophagy is initiated following hemin treatment. Taken together, these data suggest that reactive lipid species formed during exposure to high concentrations of hemin colocalize with autophagic vacuoles, promote mitophagy, and point toward a means of lipid-protein degradation independently of the proteasome. To test how autophagy inhibition affects hemin-induced toxicity, 3-MA, an inhibitor of autophagy, was used. In contrast with the protection afforded by α-tocopherol, inhibiting autophagy with 3-MA exacerbated toxicity consistent with the hypothesis that damaged proteins can contribute to cell death (Fig. 11).

In summary, we have demonstrated that hemin induces protein modification by lipid peroxidation products and causes decreases in mitochondrial function. The mitochondrial dysfunction and protein damage elicited by hemin occurs before apoptotic cell death and inhibition of lipid peroxidation partially corrects the bioenergetic defect as well as prevents cell death. Furthermore, hemin induces autophagic vesicle accumulation, whereas autophagy itself was found to be protective against hemin induced toxicity. Hemin induced modification of mitochondrial proteins by lipid peroxidation products, as observed using tagged derivatives of arachidonic acid. Additionally, data in this article point toward a possible role of autophagy in clearing proteins modified by reactive lipid species. This method of degrading modified proteins could be important in disease states where highly electrophilic compounds are produced, which result in the inhibition of proteasomal function (63, 78, 85). Further research into the role of autophagy in removing the cell of lipid-protein adducts (28, 45) as well as damaged mitochondria (57, 75, 88) could enable the design of better therapies in hemolytic diseases as well as cardiovascular diseases such as heart failure and ischemia-reperfusion injury.

Taken together, these data suggest that the endothelial cell death elicited by hemin is due to mitochondrial dysfunction caused by posttranslational modification of proteins by reactive lipid and oxygen species. Endothelial damage and vascular dysfunction associated with hemolytic disease may be improved by antioxidant therapy aimed at preventing lipid peroxidation and the resulting mitochondrial dysfunction.

GRANTS

This work was supported by American Heart Association (AHA) Predoctoral Fellowship 0815177E (to A. N. Higdon); National Heart, Lung, and Blood Institute Grant R01-HL-096638 and AHA Scientist Development Grant 0635361N (to A. Landar); National Institutes of Health (NIH) Grant R01-NS064090 and a Veterans Affairs merit award (to J. Zhang); NIH Grants ES/HL10167 and DK75865 (to V. M. Darley-Usmar); and University of Alabama at Birmingham -University of California San Diego O′Brien Core Center for Acute Kidney Injury (Grant No. NIH/NIDDK 1P30 DK 079337) for mitochondria preparations and analysis.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: A.N.H., A.L., J.Z., and V.M.D.-U. conception and design of research; A.N.H., G.A.B., B.K.C., X.O., and M.S.J. performed experiments; A.N.H., G.A.B., B.K.C., X.O., M.S.J., and J.Z. analyzed data; A.N.H., M.S.J., J.Z., and V.M.D.-U. interpreted results of experiments; A.N.H., G.A.B., B.K.C., X.O., and M.S.J. prepared figures; A.N.H. drafted manuscript; A.N.H., M.S.J., J.Z., and V.M.D.-U. edited and revised manuscript; A.N.H., G.A.B., B.K.C., X.O., M.S.J., A.L., J.Z., and V.M.D.-U. approved final version of manuscript.