J Biol Chem, Vol. 274, Issue 27, 18916-18924, July 2, 1999

Oxidative Cross-linking of ApoB100 and Hemoglobin Results in Low Density Lipoprotein Modification in Blood
RELEVANCE TO ATHEROGENESIS CAUSED BY HEMODIALYSIS^*

Ouliana Ziouzenkova, Liana Asatryan, Mohammed Akmal^§, Ciro Tetta^¶, Mary Lou Wratten^¶, Gena Loseto-Wich, Günther Jürgens, Jay Heinecke^**, and Alex Sevanian^§§

From the  Department of Molecular Pharmacology and Toxicology, School of Pharmacy, ^§ Division of Nephrology, University of Southern California, Los Angeles, California 90033, the ^¶ Clinical and Laboratory Research Department, Bellco S. P. A., Mirandola 41037, Italy, the  Washington University School of Medicine, St. Louis, Missouri 63110, and  Karl-Franzens Universitat Graz, Graz A8010, Austria

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Human blood contains a form of minimally modified low density lipoprotein (LDL), termed LDL, whose origin remains unknown. Exploring the mechanism of formation, we found that LDL can be produced in plasma in the absence of oxygen following LDL incubation with oxidized hemoglobin species. A high degree of apolipoprotein B100 modification results from covalent association of hemoglobin with LDL involving dityrosine formation but not due to the malonaldehyde epitope formation. This was evidenced by the cross-reactivity of oxidized LDL with antibodies against hemoglobin that was accompanied by a 60-fold increase in dityrosine levels. In this study we found significantly higher LDL levels in the blood of hemodialysis patients, perhaps contributing to their greatly increased risk of atherosclerosis. The mechanism of LDL formation was studied during ex vivo blood circulation using a model system resembling clinical hemodialysis in terms of the induction of inflammatory responses. This circulation increased free hemoglobin and LDL levels compared with non-circulated blood without appreciable lipid peroxidation. Pronounced increases in LDL were found also during circulation of plasma supplemented with nanomolar hemoglobin levels. The increase in dityrosine content and presence of heme in LDL after blood circulation suggest that LDL is modified, in part, by hemoglobin-LDL conjugates containing heme. Thus, hemoglobin-mediated reactions leading to LDL oxidation in plasma can account for high LDL levels in hemodialysis patients.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The unlimited accumulation of oxidized LDL¹ in macrophages (1) and the ability of mildly oxidized LDL to promote inflammatory responses, proliferation of smooth muscle cells, and differentiation of monocytes in the arterial wall (2, 3) suggest a crucial role for oxidized LDL in atherogenesis. Oxidized LDL is present in atherosclerotic lesions (4), where various types of vascular cells can catalyze oxidative processes (5). Up to 10% of the plasma LDL consists of particles with increased net electronegative charge on apolipoprotein B100 (apoB100) and elevated levels of lipid peroxidation products (6-8). This LDL fraction, possessing some properties of a mildly oxidized LDL, is referred to as LDL (8). The proportion of LDL varies in different subjects (6) and is associated with dense LDL subfractions (9), identified as a risk factor for atherosclerosis progression (10). High LDL levels are considered potentially proatherogenic due to the high cytotoxicity and oxidizability of this LDL fraction (7, 9). Moreover, LDL may have a prolonged lifetime in the circulation determined by less effective binding of LDL to the normal LDL receptor (11).

Modification of LDL into LDL could be linked to oxidative stress induced during inflammatory events. Reactive oxygen species generated by inflammatory cells or cell-derived peroxidases have been proposed as agents responsible for LDL oxidation in the artery wall (5). Many of these species oxidize lipids to hydroperoxides (LOOH) and reactive aldehydes (12) or can specifically generate aldehydes from amino acids (13). These products can modify amino groups on apoB100, thereby increasing LDL electronegativity (12, 13). However, this modification requires relatively high concentrations of malonaldehyde (MDA) (14) or lipid hydroperoxides (15) in in vitro experiments. These reactions are thought to occur in extracellular matrices of the arterial wall, an environment that is depleted of inhibitory plasma proteins and plasma- or cell-derived antioxidants (5, 16).

In blood, radicals can be formed on heme proteins, such as hemoglobin (Hb) or myoglobin, following the oxidation with hydrogen peroxide released from activated white blood cells (17, 18). Hb-based protein radical was identified in vivo in animals under oxidative stress (19). It has been reported that Hb can effectively catalyze lipid peroxidation and LDL cross-linking in plasma-free medium in in vitro experiments (20-23). Whether oxidative reactions mediated by Hb lead to LDL formation in undiluted plasma or blood under inflammatory conditions is presently unclear.

Hb-mediated reactions can be implicated in oxidative stress during hemodialysis (HD), characterized by inflammatory reactions induced after contact of blood with hemodialysis (HD) membranes (24) and by occasional hemolytic complications (25). Oxidative stress induced during HD is believed to modify LDL and produce cardiovascular complications in HD patients (26, 27). However, enhanced lipid peroxidation in HD patients remains a controversial issue since many recent reports have failed to confirm the increase in plasma MDA or LOOH in these patients (28, 29). The level of apoB100 modification in HD subjects, including that arising from lipid peroxidation and LDL formation, has yet to be described.

We report here a link between LDL levels and oxidative stress during the HD-induced inflammation based on two novel findings as follows: 1) significantly increased LDL levels in HD patients, and 2) LDL formation in blood circulated ex vivo in a model HD system. Furthermore, we describe a novel mechanism for LDL formation in plasma which occurs without appreciable lipid peroxidation but involves conjugate formation between Hb and apoB100.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Reagents-- Sigma provided all reagents unless otherwise indicated. The reagents used were of analytical grade.

Patients-- Patients (19 male and 10 female, ages 26-84 years) who had various forms of end-stage renal failure and were on chronic hemodialysis therapy for 3.5-10.7 years (mean ± S.D. was 5.9 ± 2.3) and 14 healthy subjects (9 male and 5 female, ages 20-55 years) were recruited for this study from the vicinity of Los Angeles. None of the participants in this study were taking vitamin E supplement. HD patients were dialyzed three times a week for 3-4 h using polysulfone filters, and one patient was dialyzed using a cellulose acetate filter. Blood was obtained before the heparin injection.

Circulation-- Blood for experiments using the model hemodialysis system, and for the preparation of LDL, was obtained from overnight-fasted normolipidemic healthy donors. Blood was collected into tubes (VWR Scientific) containing EDTA (final concentration 1 mg/ml) or into the Vacutainer glass tubes containing sodium citrate (final concentration 2.6%), stored on ice, and used for the HD model circulation system within 1 h after withdrawal. The circulation line was comprised of 2 blood chambers and cuprophane and polysulfone mini-filters (Bellco S.p.A., Mirandola, Italy) (Fig. 1). Mini-filters (surface area 0.027 m²) resemble the HD filters that are commonly used in clinical practice. Blood was continuously circulated at a flow rate of 5 ml/min, using pump Masterflex C/L^TM (Cole-Palmer), in a temperature-controlled box at 37 °C protected from light. The circulation line was pre-washed with a 0.9% NaCl containing 5 units of heparin/ml (Elkins-Sinn, Cherry Hill, NJ) and with an initial 3 ml of blood. Approximately 30 ml of blood incubated in a glass beaker in the same box was used as a non-circulated control. Aliquots were obtained from circulated and non-circulated blood at 0, 2, and 4 h of incubation, where "0 h" was collected after approximately 4 min of circulation through the HD filters, representing the initial effects on blood during circulation. Plasma was prepared by centrifugation at 5,000 rpm for 10 min at 4 °C and stored on ice prior to isolation of LDL. The remaining plasma was stored at 70 °C. Concentrations of Hb species in plasma were determined according to Winterbourn (30).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Schematic of the model hemodialysis circulation system. The circulation line consisted of two chambers (open rectangles), and cuprophane (i) and one polysulfone (ii) mini-filters (closed rectangles), resembling the dialysis filters used in hemodialysis clinics. * represents the location where the line was disconnected to fill the circuit with blood (dashed rectangle) or to remove blood samples. Arrows indicate the direction of blood flow. A closed circle indicates a pump. Circulation was performed for 4 h at 37 °C in the dark.

LDL Determination-- LDL (d = 1.019-1.063) was separated by sequential ultracentrifugation as described by Hodis et al. (7). Within 24 h of isolation, LDL was separated by anion exchange chromatography into three fractions: native LDL (nLDL), LDL, and LDL² detected simultaneously at 280 nm (Perkin-Elmer LC-95 UV-detector) and at _ex 327 nm/_em 400 nm (RF-535 Shimadzu) (Fig. 2). Chromatography was performed using either an UNO^TM Q1 column (Bio-Rad) or Amersham Pharmacia Biotech Mono Q^R HR 5/5 column (Amersham Pharmacia Biotech, Uppsala, Sweden) following the method of Chang et al. (31).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   HPLC fractionation of LDL from control subject (A) of LDL exposed to the model HD system (B) and of LDL after oxidation with oxidized Hb species (C). LDL was isolated by ultracentrifugation and separated into three subfractions native LDL, LDL, and LDL2 that bear increasing electronegative charge using an anion exchange Amersham Pharmacia Biotech Mono QR HR 5/5 column as described under "Experimental Procedures." The lipoproteins were detected by absorbance at 280 nm (upper chromatogram) as well as by dityrosine-like fluorescence at ex 327 nm/em 400 nm (lower chromatogram). Note the marked increase in LDL fraction as well as the increase in dityrosine-like fluorescence for LDL circulated in the model HD system and in LDL reacted with oxidized Hb.

Lipid Peroxidation-- MDA content in LDL, containing 1 mg/ml EDTA and 0.01% butylated hydroxytoluene, was determined by the method of Wong et al. (32). HPLC analyses were performed on RP-18 column (250 × 4 mm, 5 µm) using phosphate buffer (50 mM; pH 6.8)/ethanol (60:40, v/v). MDA was detected using a Shimadzu RF-535 fluorometer set at _ex 525 nm/_em 550 nm, respectively. MDA-protein adducts were measured using polyclonal antibody against MDA-modified LDL from rabbit in a solid-phase sandwich fluorescence assay as described previously (33). Simultaneously apoB100 was determined using anti-apoB antibody (Behring AG, Germany). Results are given as ratios of the counts. Lipid hydroperoxides were measured by the method of Auerbach et al. (34). Concentrations were based on cholesterol or protein levels in LDL measured using cholesterol assay and the Protein MicroAssay (Bio-Rad) using bovine serum albumin as a standard, respectively.

Electrophoresis and Western Blotting-- SDS-PAGE of LDL (5-10 µg of protein) was performed according to Laemmli (35) for 45 min at 200 V using a Mini-PROTEAN II cell (Bio-Rad). Proteins were separated on 5 and 12% polyacrylamide layers (2:1 of the separating gel length) and stained using Coomassie Blue G-250 (Bio-Rad). For Western blotting, proteins were transferred to Immobilon-polyvinylidene difluoride membranes (Millipore) for 1 h at 100 V using a Mini Trans-Blot^R electrophoretic transfer cell (Bio-Rad). The membranes were immunoblotted with monoclonal antibodies to the -chain of Hb (Cortex Biochem, San Leandro, CA) at 1:20,000 dilution. Membranes were incubated with an horseradish peroxidase-conjugated secondary antibody (Bio-Rad) at 1:3,000 dilution and visualized by chemiluminescence using ECL (Amersham Pharmacia Biotech, Buckinghamshire, UK). Dot blots were carried out under the same conditions using antibodies to the - and -chains of Hb. Relative electrophoretic mobility of LDL samples was determined by means of agarose gel electrophoresis (Beckman Paragon system) as described previously (7).

Dityrosine Measurement-- Dityrosine content in LDL samples was measured by isotope dilution GC-MS using the method of Leeuwenburgh et al. (36). Samples were analyzed on a Hewlett-Packard 5890 gas chromatograph equipped with a 12-m DB-1 capillary column (0.2 mm diameter, 0.33-m film thickness, J & W Scientific) and a Hewlett-Packard 5988A mass spectrometer with extended mass range. Selected ion monitoring was obtained with the n-propyl heptafluorobutyryl derivative of both authentic and isotopically labeled amino acids in the negative ion chemical ionization mode with methane as the reagent gas.

MDA Enrichment of Plasma-- MDA stock solution (4 M) was prepared immediately before use by rapid acid hydrolysis of 0.2 ml of malonaldehyde bis(diethyl acetal) with 20 µl of 1 M HCl at room temperature according to Haberland et al. (14) and then diluted with 0.6 ml of phosphate-buffered saline. Plasma (3.96 ml), containing 1 mg/ml EDTA was incubated with 40 µl of the MDA stock solution (final concentrations in plasma are indicated in figure) for 2 h at 37 °C under argon, final pH 7.2.

Oxidation with Ferryl-Hb-- Human met-Hb was oxidized to ferryl-Hb for 1 min using a 6-fold excess of H₂O₂ (37). The reaction mixture was added dropwise to LDL (~6 mg/ml), containing 1000 units of catalase/ml under a stream of argon incubated at 0 °C. Hb was added at a 2:1, LDL/Hb ratio based on the tyrosine content of LDL (see below). To increase Hb concentrations, another portion of Hb was oxidized and added to reaction mixture, i.e. the procedure described above was repeated to reach the desired Hb concentrations. Control samples consisted of LDL only and LDL mixed with catalase and H₂O₂ at concentrations corresponding to that used for the highest amounts of Hb added to LDL. After addition of the desired oxidized Hb concentrations, the reaction mixture was incubated for 1 h under argon at 37 °C in the dark. LDL was then isolated by centrifugation as described. Prior to analysis, LDL was subjected to size exclusion chromatography using an Econo-pac 10 DG column (Bio-Rad). Concentrations of Hb species in reaction mixture were determined according to Miller et al. (20) and were calculated assuming LDL composition reviewed in Ref. 12 and that apoB100 consists of 152 mol of Tyr/mol of apoB100 (38).

Statistics-- Differences between treated groups were determined using Mann-Whitney Rank Sum Test, paired or unpaired Student's t tests, and the level of significance was set at p < 0.05. Data are shown as mean ± S.D. for three measurements, as one representative experiment, or as means of the results from experiments performed in duplicate or triplicate.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

High LDL Levels in HD Patients Are Not Associated with the Formation of MDA-Protein Adducts in LDL-- The extent of apoB100 modification in LDL samples was determined as the proportion of LDL in each subject's total LDL. LDL levels in HD patients were significantly higher than in healthy subjects (mean ± S.D., 5.7 ± 4.1% of the total LDL in HD versus 2.5 ± 1.0% in healthy controls) (Fig. 3A). In 30% of HD patients the LDL levels were extremely high (7.0-18.9%) and comparable to those reported after LDL oxidation in vitro with Cu²⁺ (31). Cholesterol levels in HD patients were slightly lower than those in healthy subjects (mean ± S.D., 105 ± 27 and 148 ± 27 mg/dl, respectively) and, therefore, cannot account for the elevated LDL levels.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Protein modification in healthy subjects and HD patients. Left panel, proportion of LDL- in LDL obtained from plasma of healthy subjects and HD patients. LDL was isolated by ultracentrifugation and subjected to anion exchange chromatography (UNOTM Q1 column) as described under "Experimental Procedures." Data represent the LDL content as a percent of total LDL. The proportion of LDL in HD patients (n = 29) was significantly higher than in healthy subjects (n = 14) as revealed by the Mann-Whitney Rank Sum Test (p < 0.029). Lines represent mean values. Right panel, comparison of MDA-protein adducts in total LDL and LDL subfraction from healthy subjects and HD patients. LDL from 11 HD patients was collected after chromatography, pooled, and concentrated using ultra fine filter (Ultrafree-15, Biomax 30K, Millipore). Combined LDL subfractions as well as total LDL (n = 15) were analyzed for MDA-protein adducts. Similar measurements were carried out using LDL (n = 10) and isolated LDL (n = 3) from healthy subjects. Anti-MDA-protein adducts/anti-apoB100 ratios in LDL of HD patients were not significantly different from those of healthy subjects (mean ± S.D. of anti-MDA-protein adducts/anti-apoB100 ratios were 0.011 ± 0.006 and 0.015 ± 0.005, respectively). Increases in MDA epitopes in LDL are shown as a percent of the anti-MDA-protein adducts/anti-apoB100 ratio, determined in total LDL from healthy subjects (right bars) and HD patients (left bars), respectively. Each measurement was performed in triplicate.

Total LDL and isolated LDL subfractions from HD patients or healthy subjects were analyzed for MDA-protein adducts using polyclonal antibodies to MDA-modified LDL (33). Surprisingly, the proportion of MDA-protein adducts in LDL of HD patients (n = 15) was similar to that in healthy subjects (n = 10). Mean ± S.D. of anti-MDA-protein adducts/anti-apoB100 ratios were 0.011 ± 0.006 and 0.015 ± 0.005, respectively. The proportion of MDA-protein adducts appears to be higher in LDL subfraction compared with total LDL; however, this trend does not reach statistical significance (Fig. 3B).

LDL Is Generated When Whole Blood Is Exposed to a Model HD System-- The effect of HD on LDL formation was studied using an ex vivo blood circulation system (see "Experimental Procedures"). Blood circulation in this model HD system leads to pronounced LDL formation. An approximately 2-fold increase in LDL over initial levels was found in blood after 4 h circulation (p < 0.0013) (Fig. 4). In contrast, LDL levels remained unchanged over this period in non-circulated blood (p = 0.24). LDL was formed using different types of anticoagulants (heparin, sodium citrate, and EDTA), as well as in the presence of LPS. LDL formation during 4 h circulation varied from 135 to 270% of the initial levels in blood obtained from different donors (n = 9) (Fig. 4, inset).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Time course of LDL formation in whole non-circulated blood (NC) and in whole (blood circulated in the model HD system C). The proportion of LDL increased during circulation (solid lines) in blood containing 1 mg/ml EDTA (C) as well as in citrated blood containing 5 ng/ml LPS (C, LPS). The latter condition mimics circulation under conditions with additional inflammatory cell activation. In non-circulated blood (dashed lines) with (NC, LPS) and without LPS (NC) the proportion of LDL did not change. Data represent the mean of two and three independent experiments without and with LPS, respectively. The LDL increase after 4 h of circulation was significant (**, p < 0.0013) as compared with circulated blood at 0 h incubation. The variability in LDL formation of circulated blood obtained from different donors is shown in the insert. LDL levels after 4 h circulation (C) were significantly higher (**, p < 0.0003, paired t test) compared with those in non-circulated blood (NC).

Lack of a Correlation between MDA Content in LDL and LDL Levels-- To determine if LDL formation involved reactions with reactive aldehydes produced during lipid peroxidation or following cyclooxygenase activation in cells, we measured the MDA content as well as MDA-protein adduct in LDL after circulation. MDA content in LDL was measured by HPLC after the acidic hydrolysis of LDL (32). The initial MDA content was 0.08 ± 0.05 mol/mol LDL (n = 5). After 4 h of blood circulation in the presence of LPS, the MDA content in LDL reached a level of 3.0 ± 3.2 mol/mol LDL (Fig. 5). The amount of MDA formed after blood circulation varied considerably among different donors (Fig. 5, upper inset). In contrast, circulation of blood without LPS leads to moderate increases in MDA, i.e. up to 1.0 ± 0.79 mol/mol LDL. In some LDL samples where the initial MDA content was higher than 1.5 mol/mol LDL, MDA levels in LDL were not affected by circulation. Increasing the MDA content in LDL did not increase significantly the formation of MDA-protein adducts measured by immune assay (Fig. 5, lower inset).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Time course of MDA accumulation in whole non-circulated blood (NC) and in whole blood circulated in the model HD system (solid lines, C). MDA content is shown in LDL isolated from circulated blood containing 1 mg/ml EDTA (C) or 5 ng/ml LPS (C, LPS). In non-circulated blood (dashed line) with (ND, LPS) and without LPS (NC) MDA content in LDL did not change significantly. Data are shown as the mean of 2 and 3 different experiments without and with LPS, respectively. The variability in the increase in LDL MDA content during 4 h circulation of blood obtained from different donors is shown in the insert. The change in MDA content in LDL from circulated blood was not significantly different from that of non-circulated blood (p = 0.28, paired t test).

To study the effect of higher MDA concentrations on LDL formation, plasma was incubated with different concentrations of MDA reagent. This led to the MDA incorporation of up to 12.4 mol/mol LDL but did not change the proportion of LDL (Fig. 6, upper panel). Neither the increase in MDA levels nor the initial MDA content of LDL correlated with the proportions of LDL that were formed (Fig. 6, lower panel). These findings show that formation of LDL during circulation cannot be explained on the basis of MDA formation or production of LDL-MDA conjugates, and other mechanisms likely account for the production of LDL.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Relationship between MDA content in LDL and the proportion of LDL. Upper panel, effect of MDA addition to plasma on LDL formation. MDA was added to plasma, and samples were incubated for 2 h at 37 °C under argon. The final plasma MDA concentrations are shown using a logarithmic scale. LDL was isolated from plasma and analyzed for MDA content and LDL proportions. MDA content in LDL was calculated based on the LDL cholesterol content. Pronounced increases in LDL-MDA levels (solid circles, solid line) were not accompanied by increased LDL (open circles, solid line). Lower panel, lack of a correlation between MDA content in LDL and the proportion of LDL. Values were determined using LDL obtained from circulated (closed circles) and non-circulated blood (open circles) as well as from plasma supplemented with MDA (open triangles, described in the upper panel). Relationship between the MDA content and LDL proportion in each of the investigated subgroups as well as in the combined groups (R2 = 0.003) was not significant (ns).

Free Hb Promotes LDL Formation-- A continuous increase in oxy-Hb to micromolar levels was found in blood during the extracorporeal circulation of blood (Fig. 7, left panel, solid lines), whereas the met-Hb levels were transiently increased during the circulation (Fig. 7, right panel). Hemolysis increased approximately 30-fold in the presence of LPS (triangles) and 2-fold without LPS (circles). A moderate increase in Hb was also observed in non-circulated blood. Clinical HD can also occasionally lead to hemolysis and can increase plasma-free Hb levels to 110-2400 mg/dl (correspondent to 0.17-3.8 µM) free Hb (25).

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Plasma concentration of oxy-Hb left panel) and met-Hb (right panel) in non-circulated blood and blood circulated in the model HD system. Hemoglobin release was investigated in blood without (circles) or containing LPS (triangles). Plasma was obtained from circulated (solid line, solid symbols) and non-circulated blood (dashed line, open symbols) after 0, 2, and 4 h of incubation. Concentrations of oxy- and met-Hb in plasma diluted 1:50 with 0.9% NaC1 were calculated spectrophotometrically (33), whereas ferryl-Hb was under detection limits.

To determine whether LDL formation in circulated whole blood was dependent on Hb, we measured LDL levels in plasma circulated with or without Hb. Hb added to plasma at final concentrations of 50, 100, and 150 ng/ml (0.8, 1.6, and 2.4 nM Hb, respectively), increased the proportions of LDL from 150, 550, and 450%, respectively, over initial levels during 4 h circulation in the model HD system (Fig. 8). This increase was similar to or higher than LDL formation during blood circulation (Fig. 4.). LDL formation appears to be influenced by plasma components because plasma from different donors responded differently when Hb was added. LDL formation in circulated plasma containing Hb was not accompanied by MDA formation (data not shown). LDL did not form in non-circulated plasma (Fig. 8, dashed line). These results indicate that nanomolar concentrations of Hb markedly increase LDL levels in plasma exposed to the model HD system, and the release of Hb at micromolar concentrations during the extracorporeal circulation of whole blood can account for LDL formation.

View larger version (17K): [in this window] [in a new window]   Fig. 8.   LDL formation during the circulation of plasma containing Hb. Plasma containing the indicated final concentrations of human Hb (95% met-Hb) was circulated in the model HD system (C) or incubated in glass for 4 h at 37 °C with (NC, Hb) or without Hb (NC). Data represent three experiments with plasma obtained from different donors. Data from non-circulated plasma shown as mean ± S.D. summarize the increase in LDL from non-circulated plasma containing 0.8, 1.6, and 2.4 nM Hb, respectively (NC, Hb, diamond, dashed line). The tendency (~10%) for LDL to increase was similar in plasma containing different Hb levels. LDL proportions in non-circulated plasma without Hb remained unchanged (closed circles). An increase in LDL proportions was found in circulated plasma (C, solid lines).

LDL oxidized by Hb in vitro has a number of specific properties compared with LDL oxidation mediated by Cu²⁺, Fe²⁺, and Fe³⁺. The radical species of Hb that forms by oxidation of metHb with H₂O₂ to ferryl-Hb (37) induces cross-linking of LDL (21) accompanied by an increase in dityrosine-like fluorescence at _ex 327 nm/_em 400 nm (21, 39). We observed the increase in dityrosine-like fluorescence in each LDL subfraction isolated from blood exposed to the circulation. This was performed by comparing the peak areas in the chromatograms from the fluorescence signal to that of the UV-visible signal (280 nm). One representative example is shown in Fig. 2B. During circulation of blood, the proportion of dityrosine-like fluorescence increased in the nLDL fraction (p = 0.02) compared with the nLDL recovered from non-circulated blood (Fig. 9). This effect is due to dityrosine formation rather than to a decrease in 280 nm absorption (due to oxidation of amino acids), since a positive linear correlation was found between peak areas obtained by fluorescence and UV detectors (Fig. 9, inset). Dityrosine-like fluorescence appears to be higher in the LDL fraction as compared with the nLDL fraction (p = 0.0005). However, further studies are needed to confirm this observation because this phenomenon can, in part, be due to protein oxidation. Nevertheless, the significant increase in dityrosine-like fluorescence within nLDL and LDL fractions was also observed during oxidation of LDL with Hb species in vitro (discussed below). The formation of LDL in blood and plasma supplemented with Hb as well as the increase in dityrosine-like fluorescence suggest that Hb-induced oxidation may modify LDL when whole blood is circulated in the model HD system.

View larger version (29K): [in this window] [in a new window]   Fig. 9.   Dityrosine-like fluorescence in nLDL and LDL. Dityrosine-like fluorescence ex 327 nm/em 400 nm (FU) and absorbance at 280 nm (AU) in LDL subfractions were determined as described in Fig. 2. The level of dityrosine-like fluorescence in each fraction was measured as the ratio of peak area obtained from fluorescence and UV detectors. Data represent mean ± S.D. of 12 independent experiments for circulated (C) and 7 for non-circulated stored blood (NC), respectively, using blood from the blood bank. Significantly higher levels of dityrosine-like fluorescence were found in LDL compared with nLDL fractions (**, p < 0.0005). After circulation in a model HD system, fluorescence increased in nLDL fraction (*, p < 0.02). However, fluorescence remained higher in LDL than in nLDL (, p < 0.02, paired t test). The fluorescence values (FU) correlated with the absorbance values (AU) in nLDL isolated from circulated (closed circles, r = 0.82, p < 0.0001, n = 21) as well as non-circulated blood (open circles, r = 0.76, p < 0.0004, n = 17). The regression line including all data for nLDL is shown in the insert.

Oxidized Hb Species Promote LDL Formation-- To investigate further the mechanism of LDL formation reactions between oxidized Hb species and isolated LDL was studied. Oxidized Hb species were produced by the rapid reaction of met-Hb with H₂O₂ added to LDL (37). Concentrations of Hb species after the reaction with LDL are shown in Table I. The reaction between LDL and the oxidized Hb species was performed under an argon stream to decrease the contribution of lipid peroxidation products to LDL modification. LDL then was isolated from the reaction mixture by ultracentrifugation and purified by size exclusion chromatography. Electrophoretic analysis of freshly isolated and purified LDL, using SDS-PAGE under reducing conditions, revealed that it was not contaminated by Hb or other proteins (Fig. 10). The oxidized Hb catalyzed LDL formation in a concentration-dependent manner (Table I). As expected, the levels of lipid peroxidation products (LOOH and MDA) as well as MDA-protein adducts on LDL were low and insufficient to account for the marked increases in LDL levels (up to 38% of total LDL). Modification of LDL charge was confirmed by comparing the electrophoretic mobility of oxidized LDL with that of control LDL. The presence of dityrosine in Hb-modified LDL was detected as increase in dityrosine-like fluorescence and by isotope dilution GC-MS analysis (36) (Table II). Dityrosine levels in LDL were 60- and 80-fold higher in LDL samples containing 9 and 17% of LDL, respectively, as compared with control LDL. An increase in specific fluorescence was found in all LDL fractions (Fig. 2C and Table II). These data strongly suggest that tyrosyl radicals were involved in the formation of covalent bonds in LDL exposed to oxidized Hb, and this correlated with LDL formation.

View this table: [in this window] [in a new window]   Table I Concentration of Hb species and level of LDL modification after in vitro oxidation with Hb Oxidation was performed as described under "Experimental Procedures." Hb concentrations are shown as ratio of LDL tyrosines/Hb, mol/mol. Concentration of Hb species was determined spectrophotometrically and calculated according to Miller et al. (20). After reaction, LDL was re-isolated by centrifugation and purified by gel chromatography. Purity of LDL was confirmed by electrophoresis (Fig. 10). Proportion of mildly modified LDL (LDL and LDL2) as well as MDA, MDA-protein adducts, and LOOH content was measured in purified LDL. Data represent mean ± S.D. of three measurements obtained from one representative experiment. The experiments were performed at least in triplicate.

View larger version (56K): [in this window] [in a new window]   Fig. 10.   Hb-mediated oxidation and the apo-B100 modification. LDL was oxidized with oxidized Hb species as described under "Experimental Procedures." The characteristics of control and oxidized LDL are described in Tables I-III. Hb concentrations are shown as the molar ratio of tyrosine (Tyr). Changes in electrophoretic mobility in LDL were determined using 0.5% agarose gels (A). Cross-linking with the same LDL samples was assessed using SDS-PAGE (B). SDS reducing buffer consisted of 62.5 mM Tris-HCl, 20% glycerol, 2% SDS, 5% -mercaptoethanol, and 0.5% bromphenol blue, Western blot analysis (C) revealed a dose-dependent cross-reactivity of apo-B100 with monoclonal antibodies to the Hb -chain in samples oxidized with Hb. Western blotting was performed using high concentrations of protein in blocking solution (1% of human albumin (Biocell Laboratories, Rancho Dominguez, CA), 0.5% Tween 20). This cross-reactivity was confirmed by means of dot-blots using monoclonal antibodies to the Hb -chain (D) as well as to the -chain (data not shown).

View this table: [in this window] [in a new window]   Table II Formation of dityrosines (DIT) after in vitro oxidation with oxidized Hb species Experimental conditions are described in Table I. Dityrosines (DIT) and tyrosines (Tyr) were measured by GS-MS according to the method of Leeuwenburgh et al. (36). Formation of DIT in LDL subfractions was estimated by the increase in fluorescence (Ex/Em 327/400 nm, and by the increase in this fluorescence normalized to protein concentration (AU 280) (see Fig. 2). Data represent mean ± S.D. of two or three measurements in one representative experiment. The experiments were performed at least in triplicate. (u.d.l., under detection limit; ND, not determined).

Electrophoretic analysis of LDL oxidized by the Hb species confirmed previous reports (20, 21) about the cross-linking of LDL protein (Fig. 10A). Although feasible, the formation of LDL-LDL conjugates is not likely to account for LDL formation, since there were few aggregates in the LDL subfraction as was observed using electron microscopy, and the LDL proportion remained the same after the centrifugation for 1 h at 110,000 rpm (Beckman Airfuge, rotor A 110) (data not shown). On the other hand, it is conceivable that the formation of inter-molecular bonds can affect LDL charge. When Hb-modified LDL was subjected to Western blotting, it was found to cross-react with monoclonal antibodies to Hb (Fig. 10C). The increase in immunoreactivity was proportional to the Hb concentrations in the reaction mixture (Fig. 10, C and D). This suggests the presence of Hb- or Hb-derived fragments on the protein component of LDL.

Hb-oxidized LDL exhibited spectral characteristics distinct from those of nLDL. The characteristic Fe(III)-heme absorption maximum at 406 nm increased in dose-dependent manner in Hb-oxidized LDL but not in control LDL incubated without Hb (Fig. 11, left panel). The right panel in Fig. 11 shows spectra for LDL isolated from circulated (C) and non-circulated (NC) blood. A slight increase for the heme absorption maximum at 406 nm was observed in LDL isolated from circulated blood at the beginning of circulation, and a further increase was seen at the end of the circulation. After reduction of this LDL with dithionite (Fig. 11, inset), the absorption maximum shifted to 415 nm. When the sample was oxidized, an absorbance peak at 406 nm maximum was produced. In contrast, the 406-nm peak was very low in LDL from non-circulated blood, indicating the absence of heme species.

View larger version (20K): [in this window] [in a new window]   Fig. 11.   Soret region of the LDL spectra. To exclude low molecular weight heme components, all LDL preparations were re-isolated and purified by size exclusion chromatography. Left panel, spectral scans of control freshly isolated LDL (bold line) and LDL recovered after reaction with increasing concentrations of oxidized Hb species (1:2, 1.5:1, 4:1 molar ratio of Hb/LDL-Tyr (shown in Tables IIII). A characteristic Fe(III)-heme absorption maximum at 406 nm is shown to increase in a dose-dependent manner in oxidized LDL but not in control LDL incubated without Hb (bold line). Right panel, spectra for LDL isolated from circulated (C) and non-circulated (NC) blood at the beginning (bold lines, 0) and after 4 h incubation (fine lines, 4). A slight increase in heme absorption maximum at 406 nm is seen in LDL isolated from circulated blood at the beginning of circulation (C, 0; dashed arrow), and a further increase at the end of the circulation interval (solid arrow). Low absorption at 406 nm was found in non-circulated blood (NC, 0 or 4). Inset, LDL isolated from circulated blood (C, 4) was reduced by dithionite leading to the shift in the absorption maximum from 406 (dashed arrow) to 415 nm (solid arrow) within the 3 min after reduction. Oxygenation of the sample for 3 min shifted the absorption maximum from 415 to 406 (solid arrow). A partial aggregation of LDL contributed to the increased absorption of the sample.

The presence of Hb- or Hb-derived fragments on apoB100 resulted in an increase in protein content of LDL as shown by an increase in the protein/cholesterol ratio that correlated with the amount of added Hb (Table III). This was proportional to increase in LDL. It is possible that the increase in protein/cholesterol ratio may be due, in part, to oxidation of cholesterol and other lipids, thus depleting the amount of cholesterol relative to protein. However, this effect should be minimal since low levels of lipid peroxidation took place under the reaction conditions. It appears that oxidative processes do not lead to the binding of Hb tetramers because the increase in protein molecular mass (43 kDa) did not correspond to the molecular mass of the Hb tetramer at 64 kDa.

Incubation of LDL for 4 h at 37 °C with 100 ng/ml met-Hb also lead to LDL formation. Initial LDL levels increased from 0.87 ± 0.23% to 8.78 ± 4.24% of total LDL in the absence of plasma and to 1.87 ± 0.02 in the presence of 5% plasma, respectively. Altogether, these results indicate that LDL is generated mainly via the binding of Hb to LDL and that this modification can take place in blood or plasma during circulation in a model hemodialysis system.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The main goals of this study were to determine the mechanism by which a potentially proatherogenic LDL subfraction can be formed in blood and thereby to establish the link between LDL levels in blood and oxidative stress. This is a controversial issue, since it has been widely held that LDL oxidation is unlikely to occur in plasma due to the abundance of free radical scavengers and antioxidants.

View larger version (8K): [in this window] [in a new window]   Reaction 1.

Free Hb can potentially mediate the LDL formation, since Hb protein radicals were found in blood in vivo (19) and are well known in terms of initiating lipid peroxidation, particularly in LDL (20-23). Hb can be oxidized to its met and/or ferryl state by reacting with lipid hydroperoxides (40, 41) or, readily, with H₂O₂ (18, 37). The latter can be generated under inflammatory conditions by inflammatory cells (42), erythrocytes (37), and platelets (43) or during the autoxidation of Hb (44). Oxidation to the ferryl state has been reported to be accompanied by tyrosyl radical formation (18, 19) (Reaction 1).

Hb-derived radicals can initiate lipid peroxidation and promote oxidative cross-linking of LDL (20, 21). In agreement, we found that Hb-mediated oxidation caused cross-linking and an increase in electrophoretic mobility of LDL (Fig. 10, A and B, and Table I). Hb-mediated reactions yielded also a marked and dose-dependent increase in LDL and LDL² levels (Table 1) with a preferential conversion to LDL, whereas the proportion of LDL² was ~10 times lower. Moreover, this process proceeded readily under argon atmosphere and occurred in the presence of very low levels of LOOH, MDA, and MDA-protein adduct levels in LDL (Table I). These were surprising findings because it is well documented that increase in electronegative charge and probably modification of LDL to LDL, for example during Cu²⁺-mediated oxidation, is due to the formation of Schiff's bases between aldehydes and positively charged amino groups in LDL (12). One of the major aldehydes involved in LDL modification is MDA (12). Thus, oxidative conversion of LDL by Hb is distinct from that catalyzed by Cu²⁺, which leads to the simultaneous formation of LDL, LDL² subfractions (31), strong apoB100 fragmentation (21), and proceeds via a lipid peroxidation-dependent mechanism (12). Our findings suggest that LDL modification by Hb occurs through a mechanism independent of lipid peroxidation.

Intermolecular cross-linking between apoB100 and Hb during Hb-catalyzed oxidation might be a novel mechanism for direct apoB100 modification leading to LDL formation. Three findings point to intermolecular cross-linking between Hb and apoB100. First, after reaction with oxidized Hb species, the purified LDL cross-reacted with monoclonal antibodies to Hb-- and --chains (Fig. 10). Second, spectrophotometric analysis of Hb-oxidized LDL revealed the absorbance characteristics of Fe(III)-heme (Fig. 11, inset). Third, there was a significant dose-dependent increase in protein/cholesterol ratio of LDL (Table III). In a previous report the cross-linking of LDL, or cross-linking within the Hb molecule, was explained on the basis of dityrosine formation, measured indirectly as increase in dityrosine-like fluorescence (21). In this study, we found that Hb-oxidized LDL is enriched in dityrosine in a concentrationdependent manner. This was shown by increased dityrosine-like fluorescence in nLDL and LDL fractions (Fig. 2) and confirmed using GC-MS analysis (Table II). Dityrosine formation represents an oxidation product of possible radical reaction (18) that could mediate cross-linking of LDL with Hb fragments or conjugates.² Reaction 2 describes a postulated Hb radical addition reaction to apoB100(Tyr) resulting in covalent binding to apoB upon electron transfer to the oxoferryl moiety. Reactions of Hb radical species with a host of plasma or tissue proteins via cross-linking of Hb fragments to proteins is plausible (45).

An important observation is that Hb-mediated LDL generation can occur in blood or plasma under specific conditions that mimic inflammation. Circulation of blood in the model HD system can facilitate oxidative reactions by activating inflammatory cells that come in contact with the large surface area of HD membranes (24) (Fig. 1). In our study, circulation of blood or plasma enriched with Hb leads to a significant increase in LDL (Fig. 4). This LDL was essentially identical to LDL formed by reactions with oxidized Hb with respect to the increase in dityrosine-like fluorescence and absorption characteristic for heme (Figs. 2 and 10). These effects were less pronounced in LDL oxidized during HD circulation than in those oxidized with Hb due to the negligible concentration of oxidized Hb species, particularly ferryl-Hb, during the conditions of HD circulation. Similarly, the contribution of MDA to LDL formation during circulation was negligible (Fig. 5 and Table I), and there was no relationship between MDA, MDA-protein adducts, and LDL levels. It has often been held that the MDA content of total LDL does not reflect the local increase in plasma MDA concentrations. However, the addition of MDA to plasma at concentrations up to 10 mM increased MDA levels in LDL but did not lead to increased LDL formation (Fig. 6). In contrast, nanomolar levels of Hb induced significant LDL formation.

We examined the LDL formation in vivo in patients on clinical HD, a treatment procedure associated with occasional hemolysis arising from physical breakage of erythrocytes (25) or metabolic blockage of the pentose phosphate shunt (46), and with inflammation (24). We found significantly higher LDL levels in HD patients compared with those in healthy subjects (Fig. 3, left panel). Low levels of MDA-protein adducts in LDL found in HD patients (Fig. 3, right panel) on the one hand and the abundance of free Hb during clinical HD on the other hand may explain the formation of LDL by postulated Hb-dependent mechanism in HD patients. High LDL proportions represent a potentially pro-atherogenic condition that may account for accelerated atherogenesis and cardiovascular mortality among over 850,000 HD patients (47).

Hb-oxidized LDL enriched in LDL shares many chemical properties with LDL isolated from human blood. They have similarly low levels of lipid hydroperoxides (8), MDA (11), MDA-protein adducts, along with a marked increase in LDL electronegativity (Table I). An earlier study found that naturally occurring LDL was associated mostly with the denser LDL particle fractions, which are more oxidized and more atherogenic (9). Cross-linking of LDL with Hb fragments or conjugates may lead to LDL formation with a more dense character. This may explain also a paradoxical finding (8) that some amino acid (glycine (+75%), serine (+46%), and alanine (+34%)) levels were significantly elevated in LDL isolated from human blood compared with the nLDL. It has been reported that microhemolysis occurs after erythrocyte attachment to the sub-endothelium, activated platelets, fibrin strands, or due to sudden tortuosities in the blood stream, each representing common hemostatic processes (48). Further studies will be required to determine the extent to which Hb-mediated reactions can occur in plasma in vivo. HD is an example of a condition producing oxidative stress that is related to activation of inflammatory cells and hemolysis and provides the opportunity to identify vascular events that initiate LDL oxidation and predispose individuals to an increased risk of atherosclerosis.

	Acknowlegments

We thank Drs. E Cadenas, C. Giulivi, and F. Ursini for helpful discussions. We also thank L. Zurbrugg and ASCP K. Chan for their helpful assistance in obtaining blood samples and G. Ledinski for excellent technical assistance.

	FOOTNOTES

^* This work was supported by National Institutes of Health Grant HL50350, American Heart Association Award 1176-FI1, IMI/Bellco Project Grant P0527 (to O. Z.), Fogarty International Fellowship 1 F05 TW05340-01 (to L. A.), Austria Science Fund-Special Research Center Biomembranes Project F00710, and Jubiläumsfonds Österreichichen National Bank Project 6941 (to G. J.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^** Established Investigator of the American Heart Association.

^§§ To whom correspondence should be addressed: 1985 Zonal Ave., PSC 612, Los Angeles, CA 90033. Tel.: 323-442-2770; Fax: 323-224-7473; E-mail: asevan{at}thevine.net.

² Oxidation of other residues may also contribute to modified charge as well as to oxidative cross-linking reactions, and these require further investigation in order to determine the extent to which dityrosine adducts contribute to the formation of LDL particles.

	ABBREVIATIONS

The abbreviations used are: LDL, low density lipoprotein; nLDL, native LDL; LDL, mildly oxidized LDL subfraction isolated from human plasma; LDL², LDL subfraction with higher electronegative charge than LDL; apoB100, apolipoprotein B100; LOOH, lipid hydroperoxides; MDA, malonaldehyde; HD, hemodialysis; Hb, hemoglobin; LPS, lipopolysacharide; HPLC, high performance liquid chromatography; GC-MS, gas chromatography-mass spectrometry; PAGE, polyacrylamide gel electrophoresis.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

1. Steinberg, D., Parthasarathy, S., Carew, T. E., Khoo, J. C., and Witztum, J. L. (1989) N. Engl. J. Med. 320, 915-924[Medline] [Order article via Infotrieve]
2. Navab, M., Berliner, J., Watson, A. D., Hama, S. Y., Territo, M. C., Lusis, A. J., Shih, D. M., Van Lenten, B. J., Frank, J. S., Demer, L. L., Edwards, P. A., and Fogelman, A. M. (1996) Atrerioscler. Thromb. Vasc. Biol. 16, 831-842[Abstract/Free Full Text]
3. Quinn, M. T., Parthasarathy, S., Fong, L. G., and Steinberg, D. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 2995-2998[Abstract/Free Full Text]
4. Yla-Herttuala, S., Palinski, W., Rosenfeld, M., Parthasarathy, S., Carew, T., Butler, S., Witztum, J., and Steinberg, D. (1989) J. Clin. Invest. 84, 1086-1095
5. Berliner, J. A., and Heinecke, J. W. (1996) Free Radical Biol. & Med. 20, 707-727[CrossRef][Medline] [Order article via Infotrieve]
6. Cazzolato, G., Avogaro, P., and Bittolo-Bon, G. (1991) Free Radical Biol. & Med. 11, 247-253[CrossRef][Medline] [Order article via Infotrieve]
7. Hodis, H., Kramsch, D., Avogaro, P., Bittolo-Bon, G., Cazzolato, G., Hwang, J., Peterson, H., and Sevanian, A. (1994) J. Lipid Res. 35, 669-677[Abstract]
8. Sevanian, A., Bittolo-Bon, G., Gazzolato, G., Hodis, H., Hwang, J., Zamburlini, A., Maiorino, M., and Ursini, F. (1997) J. Lipid Res. 38, 419-428[Abstract]
9. Sevanian, A., Hwang, J., Hodis, H., Cazzolato, G., Avogaro, P., and Bittolo-Bon, G. (1996) Arterioscler. Thromb. Vasc. Biol. 16, 784-793[Abstract/Free Full Text]
10. Krauss, R. M. (1998) Am. J. Cardiol. 81, 13B-17B[CrossRef][Medline] [Order article via Infotrieve]
11. Demuth, K., Myara, I., Chappey, B., Vedie, B., Pech-Amsellem, M. A., Haberland, M. E., and Moatti, N. (1996) Arterioscler. Thromb. Vasc. Biol. 16, 773-783[Abstract/Free Full Text]
12. Esterbauer, H., Gebicki, J., Puhl, H., and Jürgens, G. (1992) Free Radical Biol. & Med. 13, 341-390[CrossRef][Medline] [Order article via Infotrieve]
13. Hazen, S. L., Gaut, J. P., Hsu, F. F., Crowley, J. R., d'Avignon, A., and Heinecke, J. W. (1997) J. Biol. Chem. 272, 16990-16998[Abstract/Free Full Text]
14. Haberland, M. E., Fogelman, A. M., and Edwards, P. A. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 1712-1716[Abstract/Free Full Text]
15. Kim, J. G., Sabbagh, F., Santanam, N., Wilcox, J. N., Medford, R. M., and Parthasarathy, S. (1997) Free Radical Biol. & Med. 23, 251-259[CrossRef][Medline] [Order article via Infotrieve]
16. Rice-Evans, C., Leake, D., Bruckdorfer, K. R., and Diplock, A. T. (1996) Free Radical Res. 25, 285-311[Medline] [Order article via Infotrieve]
17. Davies, M. J. (1991) Biochim. Biophys. Acta 1077, 86-90[CrossRef][Medline] [Order article via Infotrieve]
18. Giulivi, C., and Cadenas, E. (1998) Free Radical Biol. & Med. 24, 269-279[CrossRef][Medline] [Order article via Infotrieve]
19. Maples, K. R., Kennedy, C. H., Jordan, S. J., and Mason, R. P. (1990) Arch. Biochem. Biophys. 277, 402-409[CrossRef]