Modification of human low-density lipoprotein by the lipid peroxidation product 4-hydroxynonenal

Biochimica et Biophysics Acta 875 (1986) 103-114 Elsevier

103

BBA 52078

Modification of human low-density lipoprotein by the lipid peroxidation product 4-hydroxynonenal Gimther

Jiirgens a, Johanna

Lang b and Hermann

Esterbauer

b.*

u Institute of Medical Biochemistrv, Universrty of Gmr. Harrachgasse 21 /III and h Institute of Biochemistv. Uniuersity of Grm, Schuhertstrasse 1, A-8010 Grar (Austria) (Received

Key words:

Lipid peroxidation;

May 29th, 1985)

4-Hydroxynonenal;

Atherosclerosis;

LDL

The effects of the lipid peroxidation product 4-hydroxynonenal on freshly prepared human low-density lipoprotein (LDL) were studied. At a fixed LDL concentration (5.7 mg/ml) the amount of 4-hydroxynonenal incorporated into the LDL increased with increasing aldehyde concentration from 28-30 (0.2 mM) to 140 (1 mM) mol per mol LDL, whereas at a fixed aldehyde concentration (0.2 mM) its incorporation into LDL decreased with increasing LDL concentration from 48 (1 mg LDL/ml) to 26 (12 mg LDL/ml) mol 4-hydroxynonenal bound per mol LDL. Of the total hydroxynonenal taken up 78% was bound to the protein and 21% to the lipid moiety; the remaining 1% was dissolved as free aldehyde in the lipid fraction. Amino acid analysis of the apolipoprotein B revealed that 4-hydroxynonenal attacks mainly the lysine and tyrosine residues and to a lesser extent also serine, histidine and cysteine. Treatment of LDL with 4-hydroxynonenal results in a concentration-dependent increase of the negative charge of the LDL particle as evidenced by its increased electrophoretic mobility. Moreover, 4-hydroxynonenal treatment leads to a partial conversion of the apolipoprotein B-lOOTintohigher molecular weight forms most probably apolipoproteins B-126 and B-151. Compared to malonaldehyde, 4-hydroxynonenal exhibits a much higher capacity to modify LDL and it is therefore believed that this aldehyde is a more likely candidate for being responsible for LDL modification under in vivo lipid peroxidation conditions.

Introduction Human low-density lipoprotein (LDL), a distinct population of the lipid protein complexes of the plasma, plays the major role in supplying cells of tissues and organs with cholesterol. Since the basic studies on the binding and uptake of human LDL by cultivated fibroblasts via a receptor-mediated process [l], the influence of the charge of LDL, and the nature of the groups on the lipoprotein responsible for its binding to the LDL receptor have been the issue of extensive examinations. Subsequently, it could be shown that the lysine * To whom correspondence

should be addressed.

0005-2760/86/$03.50 0 1986 Elsevier Science Publishers

and arginine residues of the LDL apolipoprotein B are accountable for the binding mentioned above [2,3]. Modification of 30% or more of the a-amino groups of lysine residues of LDL by acetoacetylation completely abolishes its binding to the LDL receptor ]2]. A series of other studies proved that modified forms of LDL, being more negatively charged than native LDL, cause accumulation of large amounts of cholesterol esters, when offered to macrophages. This was described for LDL, modified chemically by acetylation [4,5], and by treatment with the lipid peroxidation product, malonaldehyde [6,7]. All these modifications have in common that the positive charges of the E-amino group of lysine

B.V. (Biomedical

Division)

104

residues of the apolipoprotein B are partly altered, which results in an increase in the net negative charge of LDL. Recently, it was suggested that malonaldehyde reacts specifically with some lysine residues, which are obviously involved in the binding of apolipoprotein B to the macrophage receptor [S]. A biological modification has been described for LDL, which was reisolated after incubation with endothelial cells [9] and it was shown that the increase of electrophoretic mobility of LDL mediated by endothelial cells is completely prevented if incubation is performed in the presence of antioxidants and if the transition metal ions of the medium are chelated with EDTA [lo]. The possibility that lipid peroxidation is involved in the modification of LDL mediated by endothelial cells was further strengthened by experiments dealing with the alteration of LDL by human umbilical vein endothelial cells or bovine aortic smooth muscle cells through free radical processes and lipid peroxidation [ll]. LDL became oxidized as evidenced by the generation of thiobarbituric acid-reactive material and developed an increased relative electrophoretic mobility as well as other changes in its features. It is known that lipid peroxidation generates, besides malonaldehyde, a great diversity of other reactive carbonyls [12,13]. 4-Hydroxynonenal is one of the major aldehydic lipid degradation products produced by peroxidizing liver microsomes and it is highly reactive towards proteins and other biomolecules [14-161. In a preliminary report [17] we have shown that, in vitro, 4-hydroxynonenal is incorporated into LDL much more rapidly than malonaldehyde. Incubation of LDL with 0.2 mM 4-hydroxynonenal resulted in the consumption of 28-29 mol 4-hydroxynonenal/mol LDL, whereas under the same experimental conditions virtually no malonaldehyde was bound to LDL. We report in this paper on the mechanism and kinetics of the reaction of 4-hydroxynonenal with LDL and on the electrophoretic and structural properties of the 4-hydroxynonenal-modified LDL. Materials and Methods Materials 4-Hydroxy-2,3-trans-nonenal

was

prepared

according to a chemical synthesis previously described [18]. 14C-labelled 4-hydroxynonenal was produced from [,U-‘4C]arachidonic acid by lipoxygenase-catalysed formation of its 15-monohydroperoxide. The 15-hydroperoxy-5,8,11,13-eicosatetraenoic acid (15-HPETE) was subsequently decomposed aerobically in the presence of ascor4-hydroxynonenal bate/Fe’+. The 14C-labelled produced in this way was isolated from other decomposition products by several chromatographic purification steps as described elsewhere

[191. 4-Hydroxynonenal and 14C-labelled 4-hydroxynonenal were stored as a solution in chloroform (10 mg/ml) at -20°C. To prepare aqueous solutions, samples of the chloroform solution were evaporated on a rotary evaporator (30°C) the residue was taken up in 0.1 M Tris-HCl buffer (pH 7.5), 0.16 M NaCl, 1 mg/ml EDTA and 0.1 mg/ml chloramphenicol. To remove traces of chloroform the buffer solution was degassed for a few minutes under vacuum. The exact 4-hydroxynonenal concentration was determined by the ultraviolet absorbance at 222 nm using 13600 as molar absorptivity coefficient [18]. Malonaldehyde was obtained by acid hydrolysis of 1,1,3,3-tetraethoxypropane [20]. The acidic malonaldehyde solution was adjusted with NaOH to about pH 7.5 and then diluted with an equal volume, yet doubly concentrated, the buffer described above. Chemicals, buffers and biochemicals were purchased from Sigma Chemie, Munchen (F.R.G.) or Merck, Darmstadt. [ U-‘4C]Arachidonic acid was from New England Nuclear, Dreieich, F.R.G. Human LDL was prepared from freshly drawn EDTA plasma by stepwise ultracentrifugation within a density cut of 1.020-1.050 g/cm3 and finally purified by gelfiltration on Bio-Gel A 15m (60 x 1.5 cm) using a 0.1 M Tris-HCl buffer (pH 7.5) containing 0.16 M NaCl, 1 mg/ml EDTA and 0.1 mg/ml chloramphenicol. Isolated LDL was stored under nitrogen at 4°C and used within 2 weeks. The purity and composition of the LDL fractions received was determined as described earlier [21]. Cholesterol estimation was performed using the CHOD-iodide method from Merck. Total lipoprotein concentration was estimated by dry weight determination [22]. In one experiment the

105

plasma was divided into two parts. During the isolation of one half, 1 mg/ml sodium azide (present during all preparation steps of LDL) was added. Incubation procedure and measurement of 4-hydroxynonenal incorporation The incubation medium contained varying concentrations (l-22 mg/ml) of LDL in 0.1 M Tris-HCl buffer (pH 7.5), 0.16 M NaCl, 1 mg/ml EDTA and 0.1 mg/ml chloramphenicol. 4-Hydroxynonenal was added in the same buffer to give a final concentration ranging from 0.1 to 10 mM. The mixture was incubated in small Erlenmeyer flasks in a water bath at 37’C or ambient temperature for up to 5 h. To follow the kinetics of the incorporation of 4-hydroxynonenal into the LDL, samples (0.5 ml) were withdrawn at different time periods, mixed with an equal volume of acetonitrile/acetic acid 96: 4 (v/v) and then centrifuged. The unreacted aldehyde in the clear supernatant was measured by HPLC on an ODS column as described [17,23]. Control incubations, where LDL was omitted, were also performed for each set of experiment to correct for possible self decomposition of the aldehyde. The amount of 4-hydroxynonenal incorporated per mg LDL was calculated from the 4-hydroxynonenal concentration of the control sample (Chydroxynonenal, LDL omitted) minus 4-hydroxynonenal concentration in the experimental sample (Chydroxynonenal + LDL). To convert this figure into mol 4-hydroxynonenal per mol LDL, an LDL molecular weight of 2.5 . 10h was assumed [21]. Incubations of LDL with malonaldehyde (0.2 mM final concentration) were performed under conditions as described for 4-hydroxynonenal. To measure the amount of malonaldehyde taken up by the LDL, samples (1 ml) of the incubation mixture were mixed with an equal volume of acetonitrile and centrifuged. The free malonaldehyde present in the clear supernatant was determined by HPLC on an aminophase column as described [20]. Amino acid analysis After dialysis against distilled water containing 0.1 mg/ml EDTA, pH adjusted to 7.5 with NaOH, native and 4-hydroxynonenal-modified LDL were

lyophilized and delipidated with chloroform/ methanol 2 : 1 (v/v) as described by others [2]. The apolipoprotein received was then treated with dinitrofluorobenzene to convert reactive amino and hydroxy groups into the acid-stable dinitrophenyl derivatives [2,24]. The dinitrophenylated apolipoprotein B was subjected to 4 M hydrochloric acid hydrolysis in sealed, evacuated tubes at 110°C for 24 h. Amino acids were analyzed on a Biotronik Model LC 7000 amino acid analyser (Biotronik, F.R.G.) using the lithium citrate buffer programme and norvaline as internal standard. The number of amino acids of apolipoprotein B (lysine, histidine, serine, tyrosine) modified by 4-hydroxynonenal was calculated from the amount of each amino acid residue found in the 4-hydroxynonenal-treated sample minus the amount of each amino acid residue of dinitrophenylated apolipoprotein B of native LDL. The calculation of the number of amino acids/m01 LDL was based on a total protein component of 500000 g/mol LDL [8]. For control purposes, amino acid composition of apolipoprotein B, not treated with dinitrofluorobenzene, was also estimated. Gel filtration and electrophoresis To study the effect of 4-hydroxynonenal on the molecular weight of LDL or the formation of aggregates, native and 4-hydroxynonenal(1 mM, 7 mM) -treated LDL samples were separated on a Bio-Gel A 15 m column (60 x 1.5 cm) with 0.1 M Tris-HCl buffer, (pH 7.5) 0.16 M NaCl, 1 mg/ml EDTA and 0.1 mg/ml chloramphenicol. The column effluent was monitored at 280 nm and fractions of 1.3 ml were collected. In cases where LDL was incubated with “C-labelled 4-hydroxynonenal, separation of free and LDL bound 4-hydroxynonenal was performed on a Sephacryl S 200 (Pharmacia) column (30 X 0.6 cm) with 0.1 M Tris-HCl buffer (pH 7.5) containing 0.16 M NaCl, 1 mg/ml EDTA and 0.1 mg/ml chloramphenicol. Agarose gel electrophoresis of native and 4-hydroxynonenal-treated LDL was performed with the lipidophor system kindly supplied by Immuno AG (Vienna, Austria). SDS electrophoresis in 3.75% polyacrylamide gels was conducted according to the method of Weber and Osborn [25] slightly modified as described elsewhere [26]. The native and 4-hy-

106

droxynonenal-treated LDL were dialyzed as described above for the amino acid analysis. The lipoprotein was then lyophilized and delipidated with chloroform/methanol 2 : 1 (v/v) as described [2]. Prior to electrophoresis some samples (50 ~1) were treated with 5 ~1 of a 10% mercaptoethanol solution and heated. Staining was performed with Coomassie blue R 250 in 25% isopropanol, 10% acetic acid at 55’C, destaining in 7% acetic acid at 55’C. Catalase, thyroglobulin and ferritin (Pharmacia) were used as standards for molecular weight determination. Other procedures Acetylation of LDL was performed by adding acetic acid anhydride in multiple small portions to the LDL solutions [28]. Free sulfhydryl groups of native and 4-hydroxynonenal-treated LDL (0.8 mM, 5 mM, 4 h incubation time) were estimated with 5,5’-dithiobis(Znitrobenzoic acid) according to Cardin et al. [29]. The radioactivity measurements were made on a Beckman Instrument (Model1 LS 3800) with 5 ml Aquasol (NEN) as scintillation cocktail. All values were quench corrected. Free 4-hydroxynonenal dissolved in the lipid phase of LDL was estimated by HPLC of the lipid extract essentially as described [17,23].

decomposition reaction was 1.4, 2.8, 4.2 and 7%. The quantity of 4-hydroxynonenal incorporated into the LDL was calculated from the difference between control incubations (6hydroxynonena1, LDL omitted), which were performed for each set of experiments, and the experimental sample i.e., incubations of 4-hydroxynonenal + LDL. Assuming a molecular weight of 2.5 . IQ6 for LDL [21] it can be calculated from the values of Fig. 1 that 613 -I_45 mol 4-hydroxynonenal (mean + S.D. obtained from four different LDL preparations) were incorporated into 1 mol LDL after 8 h of incubation with 7 mM 4-hydroxynonenal. In an attempt to estimate the maximum 4-hydroxynonenal binding capacity of LDL, incubation conditions were varied with respect to the aldehyde and the LDL concentration. At a fixed LDL concentration (5.7 mg/ml) the increase of the 4-hydroxynonenal concentration from 0.2 to 1 mM and 10 mM led to a 4- and 43-fold increase of the initial rates (calculated from the 30 min values)

Results and Discussion Time course of the incorporation of I-hydroxynonenal into LDL Incubation of human LDL with 4-hydroxynonenal resulted in a time-dependent decrease of the aldehyde concentration, as measured by HPLC analysis of the free unreacted aldehyde, present in the reaction mixture at different time intervals (Fig. 1). The rate and extent of the consumption of 4-hydroxynonenal by LDL showed good reproducibility and was more or less the same in LDL samples prepared from blood of different donors. Control incubations in which LDL was omitted showed no measurable decrease of the 4-hydroxynonenal concentration when the initial aldehyde concentration was 1 mM or higher. A slight decrease of the aldehyde concentration however occurred in 0.2 mM 4-hydroxynonenal solutions; after 1, 2, 4 and 6 h the loss due to this unspecific

INCUBATION

TIME. HOURS

Fig. 1. Time course of the consumption of 4-hydroxynonenal (HNE) by LDL. 7 mM 4-hydroxynonenal were incubated with LDL (22 mg/ml) in closed 25 ml Erlenmeyer flasks (final volume 10 ml) in a shaking water bath at 37°C for 8 h. At the indicated time points, samples (0.5 ml) were withdrawn and added to an equal volume of acetonitrile/acetic acid (96:4) and centrifuged. The clear supernatant was diluted lo-fold with distilled water, and 4-hydroxynonenal was then estimated by HPLC. Each value represents the mean+S.D. from four separate LDL samples prepared from different donor pools.

107

A 1200 -

1000 -

1

2 INCUBATION

3

L

5

6

TIME, HOURS

Fig. 2. Kinetics of the incorporation of 4-hydroxynonenal into LDL. 0.2 mM (o), 1 mM (m) and 10 mM (A) 4-hydroxynonenal (HNE) were incubated in the absence (control) and presence of LDL 5.7 mg/ml (experimental). At the indicated time points the free aldehyde present in the control and experimental sample was estimated by HPLC as in Fig. 1. The amount of 4-hydroxynonenal incorporated into the LDL was calculated from the difference of the control and experimental sample. The insert is a magnification of the 0.2 mM experiment.

of aldehyde consumption and the total amount of incorporated aldehyde increased from 28-30 (0.2 mM) to 140 (1 mM) and 1200 (10 mM) mol, 4-hydroxynonenal per mol LDL (Fig. 2). With the highest concentration of 4-hydroxynonenal the reaction slowed down after the initial rapid phase but was still not complete after 5 h of incubation. Such a high degree of modification also results in a strong effect on the molecular structure and solubility of the LDL. LDL solutions exposed to 10 mM 4-hydroxynonenal became turbid and, finally, the lipoprotein precipitated after prolonged stand-

ing. We have not attempted to estimate the absolute maximum number of 4-hydroxynonenal molecules which could be bound to the LDL molecule, since this would have required aldehyde concentrations which rapidly lead to a precipitation of LDL; moreover, such high concentrations are far above any physiological relevance. At a fixed aldehyde concentration (0.2 mM) the rate and extent of aldehyde consumption increased as expected when the LDL concentration increased from 1 to 12 mg/ml (Table I). The number of 4-hydroxynonenal molecules bound per mol LDL was inversely related to the LDL concentration. Thus, a solution of 1 mg LDL/ml gave a binding of 48 mol 4-hydroxynonenal/mol LDL, whereas a solution of 12 mg/ml gave only 26 mol 4-hydroxynonenal/ LDL. The reason is most likely the establishment of an equilibrium between association and dissociation of the 4-hydroxynonenal-LDL complex. Such an equilibrium would be influenced by the molar ratio of the reactants 4-hydroxynonenal and LDL and shifted to the side of association with increasing free aldehyde concentrations. A plot of mol 4-hydroxynonenal bound per mol LDL against the free aldehyde concentration present in the reaction medium (4 h reaction time) also suggests such a relationship, since the curve (not shown) has a hyperbolic shape and asymptotically approaches a constant binding of approximately 50 mol4-hydroxynonenal/ LDL. The average from all ten binding experiments (Fig. 2, Table I; Ref. 17) performed with 0.2 mM 4hydroxynonenal and l-12 mg LDL/ml, which is equal to a 42- and 500-fold molar excess of the aldehyde, was 36 _t 8.7 mol 4-hydroxynonenal incorporated into 1 mol LDL. This value agrees well with the reported binding maximum of 30-35 mol malonaldehyde/LDL found after exposure of LDL to 100 mM malonaldehyde for 3 h [7]. The incorporation of 4-hydroxynonenal into LDL was also determined with “C-labelled 4hydroxynonenal. For this purpose LDL (22 mg/ ml) was incubated 4 h with 5.0 and 5.87 mM 4-hydroxy[‘4C]nonenal; thereafter, LDL was separated from the unreacted aldehyde by Sephacryl S-200 gel chromatography and its radioactivity was counted. Based on the specific radioactivity of 4-hydroxy[‘4C]nonenal (8560 dpm/pmol), the incorporation was 244 and 285 mol 4-hydroxynon-

108

TABLE EFFECT

I OF LDL CONCENTRATION

ON THE INCORPORATION

OF 4-HYDROXYNONENAL

LDL (l-12 mg/ml) was incubated ‘with 0.2 mM 4-hydroxynonenal (HNE) at 37°C. 4-hydroxynonenal was measured by HPLC and the amount bound to LDL was calculated where LDL was omitted. LDL

Incubation

(mg/mI)

2

2 free

bound

(PM)

(PM)

0 1 2 3 5 10 12

After 1. 2 and 4 h the remaining free from the difference to the control sample

time(h)

free

bound

(ILM) 194

(PM) _

179 172 163 150 116 105

15 22 31 44 78 89

HNE/LDL (mol/mol)

4 HNE/LDL (mol/moI)

free

bound

(PM)

(PM)

HNE/LDL (moI/mol)

_

183

_

_

167

_

_

37 27 26 22 20 19

172 157 141 119 79 71

11 26 42 64 104 112

27 32 35 32 26 23

148 130 115 86 49 40

19 37 52 81 118 127

48 46 43 40 29 26

enal/LDL in the 5.0 and 5.87 mM incubation systems. The corresponding values estimated by the HPLC method were 225 and 250 mol 4-hydroxynonenal/ LDL. Identification of the LDL binding sites for 4-hydroxynonenal 4-Hydroxynonenal is a strong electrophilic reagent which could react with a number of nucleophiles such as the sulfhydryl, amino or hydroxy groups present in the LDL polypeptide side chains. Additionally, incorporation into LDL could occur by a reaction with the amino group of phosphatidylethanolamine or be simply due to the dissolving of the lipophilic aldehyde in the lipid domain of the LDL molecule. In order to distinguish between these possibilities LDL (22 mg/ ml) was incubated with 14C-labelled 4-hydroxy. nonenal (5.86 mM, 29377 dpm) for 4 h. Tht 4-hydroxynonenal-modified LDL was then sep. arated from the unreacted aldehyde by gel chro. matography on Sephacryl S 200. The LDL was clearly separated from the unreacted aldehyde as judged by cholesterol estimation and radioactivity measurements of the fractions eluting from the column. 57.2% of the radioactivity eluted with the low-molecular-weight fraction and 42.8% was associated with the LDL fraction, which corresponds to an incorporation of 285 mol 4-hydroxynon-

enal/mol LDL. The total recovered radioactivity was 98 f 3%. The LDL was then subjected to delipidation and the radioactivity of the protein and lipid fraction was estimated again, revealing that 87% of the radioactivity incorporated into the LDL complex was bound to the protein, whereas the remaining fractions of 13% was in the lipid material. Repetition of this experiment with 1 and 5 mM 4-hydroxynonenal (22 mg LDL/ml, 4 h) with different LDL samples gave, for the protein fraction, 78.3 f 5.3% (mean k S.D. (n = 6), range 73.4-89%) and, for the lipid fraction, 21.6 f 5.2%. Analysis of the lipid material by HPLC showed that 96.5% of the HNE was covalently bound to the lipids, whereas 3.5% was present as free aldehyde. This clearly shows that a small amount of the aldehyde is in fact trapped in its free form in the lipid core of the LDL molecule. Therefore, from the total 4-hydroxynonenal incorporated into the LDL, about 78% is bound to the protein, 21% to the lipid and 1% is dissolved in the lipid phase. Acetylation of the LDL by acetic anhydride abolished the incorporation of the aldehyde into the lipoprotein to a large extent [17]. Since this reagent reacts with amino groups of lysine and with thiol, imidazole and aliphatic and aromatic hydroxyl groups [49], it appeared that these groups play a major role for the binding of the aldehyde to the protein moiety. This could then be con-

109

firmed by amino acid analysis of the apolipoprotein B prepared from native and 4-hydroxynonenal-treated LDL The amino acid composition of native LDL was in good agreement with the values reported in the literature [2]. The amino acid residues per mol LDL obtained by our analysis were (mean + S.D. from three analyses, based on a molecular weight of 500000 [8] for the total LDL protein component): Asp, 494 f 11; Thr, 288 + 7; Ser, 348 + 10; Glu, 579 _t 14; Gly, 221 f 12; Ala, 280 f 3; Val, 256 &-21; Met, 16 k 8; Ile, 280 & 4; Leu, 551 * 17; Phe, 235 + 2.7; Lys, 368 + 5; His, 114 k 5; Arg, 151 + 4; Tyr, 160 + 4.

TABLE EFFECT

The amino acid composition obtained for 4-hydroxynonenal-treated LDL was more or less in the range of the standard deviation of the native LDL, indicating that the linkages of 4-hydroxynonenal to the amino acid side chains were cleaved under the acidic conditions of protein hydrolysis. The same observation was also made in the case of malonaldehyde modified LDL [7]. Therefore, apolipoprotein B of 4-hydroxynonenal-treated LDL was reacted with dinitrofluorobenzene prior to acid hydrolysis to convert functional groups not blocked by 4-hydroxynonenal into acid stable dinitrophenyl derivatives. Similarly, apolipoprotein B from native LDL not exposed to 4-hydroxynon-

II OF 4-HYDROXYNONENAL

ON THE AMINO

ACID

COMPOSITION

OF LDL

LDL (22 mg/ml) was incubated for 4 h at 37°C as described, in the presence (5 mM or 0.8 mM) and absence of 4-hydroxynonenal (HNE). After gel filtration and dialysis the samples were delipidated, treated with dinitrofluorobenzene (DNFB), hydrolyzed with HCl at 1lO’C for 24 h and analyzed. The values given for native LDL were obtained from LDL samples (mean f S.D., three experiments) not treated with 4-hydroxynonenal or DNFB. The calculation is based on a molecular weight of 2.5.106 for LDL and a total protein component of 500000 g/mol LDL. ‘mol 4-hydroxynonenal/mol LDL’ was calculated from the amount of 4-hydroxynonenal consumed in the reactidn and determined by the HPLC method. ‘Percent 4-hydroxynonenal in LDL protein’ was estimated in a separate experiment with 4-hydroxy[‘4C]nonenal. LDL was isolated by Sephacryl S200, delipidated and the radioactivity of the protein fraction was measured. For the 0.8 mM incubation system the value was assumed to be the same. On average the LDL protein contains 78.3 k5.3% (n = 6) of the total radioactivity incorporated into LDL. ‘mol 4-hydroxynonenal bound to LDL protein’ was calculated from mol 4-hydroxynonenal/mol LDL, assuming that 80.58 is protein bound. ‘Amino acid residues modified’ were calculated from the number of amino acids modified and the mol 4-hydroxynonenal bound to LDL protein. Amino acid residues per mol LDL native

Lysine Histidine Serine Tyrosine Cysteine

DNFB-treated

HNE (5.0 mM)and DNFB-treated

HNE (0.8 mM)and DNFB-treated

21.9 4.6 319.3 28.8 _

61.2 12.1 342.2 79.4

41 .o 9.4 331.8 14.7

_ _ _ _

0 0 0 0 0 0

45.3 7.5 22.9 50.6 1.9 128.2

19.1 4.8 12.5 25.9 1.7 64.0

_ _

_ _

225.4 80.5

61.0 (80.5)

_

_

181.4

49.1

_

_

368+ 114* 348 + 160+ 2.5

5 5 10 4

HNE-modified Lysine Histidine Serine Tyrosine Cysteine Total mol HNE/mol LDL % HNE in LDL protein mol HNE bound to LDL protein Amino acid residues modified (mol/mol HNE)

amino acids per mol LDL

0.71

1.3

110

enal was subjected to dinitrophenylation. The number of amino acids attacked by 4-hydroxynonenal was then calculated from the difference between the experimental sample, i.e., LDL pretreated by 4-hydroxynonenal, and subsequently dinitrophenylated and the control sample, which was only treated by dinitrofluorobenzene (DNFB) (Table II). The analysis showed that 4-hydroxynonenal had reacted mainly with lysine, serine and tyrosine and to a minor extent with histidine. From the well-known reactivity of 4-hydroxynonenal towards thiol groups [27,28] it was expected that the free cysteine SH groups present in the LDL molecule [29] had also reacted with 4-hydroxynonenal. Our native LDL contained 2.5 free sulfhydryl groups and 1.9 (5 mM) and 1.7 (0.8 mM) of them were blocked by 4-hydroxynonenal. Including the cysteine residues, the total number of amino acids which had reacted with 0.80 and 5 mM 4-hydroxynonenal was 64 and 128 mol per mol LDL. The number of 4-hydroxynonenal molecules bound to the LDL protein, estimated by HPLC analysis and measurement of the radioactivity contained in the LDL protein after incubation with 5 mM 4-hydroxy[i4C]nonenal, was 181. Formation of higher molecular weight apolipoprotein B and LDL aggregates Investigation of apolipoprotein B from native LDL by SDS-polyacrylamide gel electrophoresis gave a single band, as reported by others [26,30]. The band corresponds to an apparent molecular weight of 550 000, designated as B-100. Apolipoprotein B prepared from LDL pretreated with 1 mM 4-hydroxynonenal showed, in the SDS electrophoresis, in addition to B-100, two bands corresponding to higher molecular weight proteins (Fig. 3). The position of the two new bands relative to the B-100 band resembled very closely the two higher molecular weight bands (690000 and 830000) found in LDL pretreated with 200 mM malonaldehyde [26]. The incubation conditions (1 mM 4-hydroxynonenal) which led to the formation of higher molecular weight apolipoprotein B, did not change the molecular weight of the lipoprotein, as shown by Bio-Gel A-15m gel-permeation chromatography. The LDL sample treated with 1 mM 4-hydroxynonenal showed an elution profile, identical to the native LDL, when chro-

matographed in Bio-Gel A-15 m. Intermolecular cross-linking, therefore, cannot be the cause for the formation of higher molecular weight apolipoprotein B. We assume that intramolecular selfassociation of apolipoprotein B, as has been proposed for malonaldehyde-treated LDL [26], is involved in this process. The treatment of the samples with mercaptoethanol did not change the electrophoretic patterns in the SDS electrophoresis. LDL exposed to a rather high concentration (7 mM) of 4-hydroxynonenal clearly showed a high molecular weight LDL fraction in the Bio-Gel elution profile, which was about 8% of the total LDL. Obviously, such high concentrations lead to intermolecular cross-links or changes of the structural parameters in a way which causes aggregate

1

2

Fig. 3. SDS-polyacrylamide gel electrophoresis of the protein moiety of native LDL (1) and 4-hydroxynonenal-modified LDL (2). Native and 4-hydroxynonenal-treated (1 mM, 4 h, 37OC) LDL was dialyzed and delipidated. 50 pg of protein were applied on each gel. Staining was performed with Coomassie blue R 250.

111

formation. It seems worth noting that LDL which had been isolated from the same plasma int he presence of sodium azide (1 mg/ml) and subsequently exposed to 7 mM 4-hydroxynonenal contained about 30% of higher molecular weight LDL aggregates. Sodium azide is known to induce free radical reactions and lipid peroxidation within the lipid domain of LDL [31] and this process also leads to high molecular weight aggregates, as evident from the Bio-Gel elution profile. Electrophoretic behaviour of 4-hydroxynonenal-modified LDL To test whether incorporation of 4-hydroxynonenal leads to changes of the electrophoretic mobility, agarose gel electrophoresis was performed. LDL (22 mg/ml) incubated for 4 h in the presence of 5 mM 4-hydroxynonenal showed a markedly increased electrophoretic mobility, indicating an increased negative charge of the molecule (Fig. 4). A significantly increased electrophoretic mobility was also observed with LDL exposed to 1 mM 4-hydroxynonenal. The lowest

4-hydroxynonenal concentration leading to a measureable increase of the electrophoretic mobility was about 0.1-0.2 mM. The relative electrophoretic mobility, i.e., ratio of migration rate of modified LDL to the reference LDL [ll] was 1.5, 1.1 and 1.05 for LDL treated with 5, 1 and 0.2 mM 4-hydroxynonenal. The increase of the electrophoretic mobility most likely results from the binding of 4-hydroxynonenal to the E-amino group of lysine (see Table II) which abolishes positive charges of the protein and thereby increases the net negative charge. A small contribution to the increase of net negative charge could also result from the modified histidine residues, whereas binding of 4-hydroxynonenal to serine and tyrosine residues should have no effect on the charge of the molecule. As compared to native LDL, the shape of the lipoprotein band on agarose gel electrophoresis was unchanged with LDL modified by low 4-hydroxynonenal concentrations (0.2-l mM), whereas heavy modification (a 5 mM) showed that some material remained at the start and some was smeared over the gel as a wide band. This again indicates the formation of high molecular weight aggregates upon incubation with high concentrations of 4-hydroxynonenal. Conclusion

Fig. 4. Agarose gel electrophoresis of native and 4-hydroxynonenal-modified LDL. LDL (22 mg/ml) was incubated with 0 (control), or 1 or 6 mM 4-hydroxynonenal for 4 h at 37°C. Native LDL (control), lanes 1 and 3. 4-Hydroxynonenal-treated LDL, lane 2 (6 mM) and lane 4 (1 mM). The arrows indicate the start position.

Involvement of free radical reactions and lipid peroxidation in the multistep process of the development of atherosclerotic lesions has been suggested [6-11,17,26]. Support for this hypothesis comes from several lines of investigations. (a) The polyunsaturated fatty acids of all lipoprotein classes including low-density lipoprotein (LDL) are highly sensitive towards oxidative degradation [31-331. (b) Cultured endothelial cells (rabbit aorta, human umbilical vein) are capable of inducing peroxidation in LDL in the absence of antioxidants (butylated hydroxytoluene, vitamin E) and transition metal chelators [10,13,34]. Endothelial cell modified-LDL exhibits increased electrophoretic mobility as compared to native LDL [lo] and is cytotoxic towards human skin fibroblasts [35]. (c) LDL, being more negatively charged than its native counterpart, is not recognized and taken up

112

by the classical LDL receptor but rapidly by the ‘scavenger’ receptor [7,10] of macrophages, which has been claimed to be identical to the acetyl-LDL receptor [9,36]. This stimulated uptake of altered LDL particles could explain the transformation of monocyte macrophages into lipid-laden foam cells [7,10,37]. (d) The recognition of LDL by the classical receptor is dependent on a limited number of lysine residues of the apolipoprotein B [2,3]. Chemical modification of the a-amino group of the lysine residues with the lipid peroxidation product, malonaldehyde, yields an LDL with an increased electrophoretic mobility [26] and biological properties similar to the endothelial cell-modified LDL [7,8], i.e., diminished uptake by the classical B,Ereceptor and stimulated recognition by the ‘scavenger’ receptor of monocyte macrophages. (e) Malonaldehyde and malonaldehyde-like substances are produced during peroxidative degradation of polyunsaturated fatty acids in many biological systems, including tissues, isolated cells, subcellular fractions and isolated lipids [38,39]. Malonaldehyde is also generated by aggregating blood platelets [6,7,40,41]. Malonaldehyde and lipid peroxide levels are increased in the serum of patients suffering from intravascular thrombosis [42,43] and in the LDL fraction of older individuals [44]. All the above suggests, but of course does not prove, that lipid peroxides and/or some of their break-down products such as malonaldehyde could lead to pathological alterations of the LDL and thus be involved in artherogenesis. Lipid peroxidation is a very complex phenomenon [45] and pathological effects could in principle be due to reactive free radicals continuously produced during chain propagation, to diffusible and reactive non-radical lipid degradation products and to structural alterations of the lipid protein complexes. Which of these mechanisms is predominant may depend on the conditions and systems studied. However, the principle of a unicausality for damage caused by lipid peroxidation is unlikely. In most studies concerned with lipid peroxidation in biological systems its extent and rate is measured by the amount of malonaldehyde or malonaldehyde-like substances as estimated by the thiobarbituric acid assay [39]. This assay is sensi-

tive and its major advantage is, at first glance, its simplicity. In retrospect, however, this could have hampered the insight in the complexity of the lipid peroxidation process and how it affects biolgoical structures, since it unintentionally made the easily measurable malonaldehyde the centre of attention. For example, it was believed that malonaldehyde is one of the major mutagenic constituents in peroxidized lipids, and only very recently [46] was it shown that purified malonaldehyde is much less mutagenic as compared to the other lipid peroxidation products. We have shown [12-14,451 that peroxidizing liver microsomes produce, in addition to free malonaldehyde, a great diversity of other aldehydes, i.e., alkenals, 2-alkenals, 2,4-alkadienals, ketoaldehydes and 4-hydroxyalkenals. The total amount of these aldehydes is about the 2-fold of the amount of malonaldehyde [45]. In the biological sense, 4-hydroxynonenal is of particular interest, since it is a major product and has the capacity to modify biomolecules [12-16,27,28] including LDL much more effectively than malonaldehyde. Compared to 4-hydroxynonenal, malonaldehyde is rather unreactive towards proteins. For example, human serum exposed 1 h to a 0.1 mM aldehyde solution, incorporated only about 5% of malonaldehydes but 80% of the 4-hydroxynonenal [12]. Similar large differences were observed with bovine serum albumin, glutathione, cysteine [12] and LDL [17]. 4-Hydroxynonenal is highly reactive towards thiols [12,27,28,47,48] and it was therefore expected that the aldehyde binds, by the Michael type reaction, to the cysteine residues of the LDL. Unexpectedly, the aldehyde also interacted with lysine and tyrosine and to a lower extent with histidine and serine residues of LDL. The molecular mechanism of these reactions remains to be elucidated and will be the subject of further studies. From the ratio of amino acids modified/m01 4-hydroxynonenal it seems possible that the stoichiometry of the reaction changes as a function of the aldehyde concentration (Table II). The 4-hydroxynonenal-modified LDL exhibits properties very similar, if not identical, to those observed with LDL treated with high concentration (100-200 mM) of malonaldehyde [7,26]. Thus, 4-hydroxynonenal (0.1 to 6 mM) treatment increases the negative charge and electrophoretic

113

mobility of LDL in a concentration-dependent manner and also leads to a partial conversion of apolipoprotein B into two higher molecular weight forms, which are most likely identical with the B-126 and B-151 bands in malonaldehyde-modified LDL [26]. preliminary results revealed that 4-hydroxynonenal-modified LDL shows a reduced uptake by the classical LDL receptor of fibroblasts (Jessup, W., Jurgens, G., Lang, J., Esterbauer, H. and Dean, R.T., unpublished data). With the present knowledge we do not know whether in vivo 4-hydroxynonenal could be produced in the vascular tissue at concentrations (0.1-0.2 mM) sufficient to interact with the passing LDL and thereby increase its negative charge and electrophoretic mobility. It should be stressed, however, that 4-hydroxynonenal is a highly lipophilic compound and, if formed from the polyunsaturated fatty acids within the lipid core of LDL itself, would remain there and reach concentrations high enough to attack the lipid-embedded parts of the protein. Malonaldehyde on the other hand, is a hydrophilic compound which would immediately diffuse from the place of its origin in lipid phase into the surrounding aqueous phase as shown for peroxidizing liver microsomes [13]. The statements made above together with the results described clearly show that lipid peroxidation products other than malonaldehyde must be considered in discussing the significance of lipid peroxidation for modification of LDL. Acknowledgements These studies were performed persuant to a contract with the National Foundation for Cancer Research, Bethesda, MD, U.S.A. Part of this work was also supported by the ijsterreichischer Fonds zur Forderung der Wissenschaftlichen Forschung (Project No. 5158) and by the Jubilaumsfonds der Gsterreichischen Nationalbank. We appreciate the technical assistance of Gunther Radspieler, Gerhard Ledinski and Gerd Kager. References 1 Goldstein, J.L. and Brown, M.S. (1977) Annu. Rev. Biothem. 46, 897-930 2 Weisgraber, K.H., Innerarity, T.L. and Mahley, R.W. (1978) J. Biol. Chem. 253. 9053-9062

3 Mahley, R.W., Innerarity, T.L. and Weisgraber, K.H. (1980) Ann. N.Y. Acad. Sci. 348, 265-277 4 Goldstein, J.L., Ho, Y.K., Basu, S.K. and Brown, M.S. (1979) Proc. Natl. Acad. Sci. USA 76, 333-337 J.L. (1981) Annu. Rev. Bio5 Brown, M.S. and Goldstein, them. 52, 223-261 6 Fogelman, M.A., Shechter, I., Seager, J., Hokom, M., Child, J.S. and Edwards, P.A. (1980) Proc. Natl. Acad. Sci. USA 77, 2214-2218 7 Haberland, M.E., Fogelman, M.A. and Edwards. P.A. (1982) Proc. Natl. Acad. Sci. USA 79, 1712-1716 8 Haberland, M.E., Olch, C.L. and Fogelman, A.M. (1984) J. Biol. Chem. 259, 11305-11311 9 Henriksen, T., Mahoney, E.M. and Steinberg, D. (1981) Proc. Natl. Acad. Sci. USA 78, 6499-6503 10 Steinbrecher, U.P., Parthasarathy, S., Leake, D.S., Witzum, J.L. and Steinberg, D. (1984) Proc. Natl. Acad. Sci. USA 81, 3883-3887 11 Morel, D.W., Di Corleto, P.E. and Chisolm, G.M. (1984) Arteriosclerosis 4, 357-364 12 Esterbauer, H. (1982) in Free Radicals, Lipid Peroxidation and Cancer (McBrien, D.C.H. and Slater, T.F., eds.), pp. 101-122. Academic Press, London 13 Esterbauer, H.. Cheeseman, K.H., Dianzani, M.U.. Poli, G. and Slater, T.F. (1982) B&hem. J. 208, 129-140 14 Benedetti, A., Comporti, M. and Esterbauer, H. (1980) B&him. Biophys. Acta 620, 281-296 15 Benedetti, A., Barbieri, L., Ferrali, M., Casini, A.F., Fullceri, R. and Comporti, M. (1981) Chem. Biol. interact. 35, 331-340 16 Schauenstein, E., Esterbauer, H. and Zollner, H. (1977) in Aldehydes in Biological Systems: Their Natural Occurrence and Biological Activities, Pion Ltd., London 11 Jtirgens, G., Lang, J., Esterbauer, H. and Holasek, A. (1984) IRCS Med. Sci. 12, 252-253 18 Esterbauer, H. and Weger, W. (1967) Monatsh. Chem. 98, 1994-2000 19 Fauler, G. (1984) Thesis, University of Graz, Austria 20 Esterbauer, H., Lang, J., Zadravec, S. and Slater, T.F. (1984) Methods Enzymol. 105, 319-328 21 Jiirgens, G., Knipping, M.J., Zipper, P., Kayushina, R., Degovics, G. and Laggner, P. (1981) Biochemistry 20, 3231-3237 22 Jtirgens, G. (1983) Clin. Chem. 29, 1856 23 Lang, J. and Esterbauer, H. (1985) Anal. Biochem., in the press 24 Wolfsy, L. and Singer, S.J. (1963) Biochemistry 2, 104-116 25 Weber, K. and Osborn, M. (1969) J. Biol. Chem. 244, 440664412 26 Bittolo Bon, Cozzolato, G. and Avogadro. P. (1983) Artery 12, 74-80 27 Esterbauer, H., Zollner, H. and Scholz, N. (1975) Z. Naturforsch. 30~. 466-473 28 Esterbauer, H., Ertl, A. and Scholz, N. (1976) Tetrahedron 32, 285-289 29 Cardin, A.D., Witt, K.R., Barnhart, C.L. and Jackson, R.L. (1982) Biochemistry 21, 4503-4511 30 Hong, J.L., Pflug, J. and Reichl, D. (1984) Biochem. J. 222, 49-55

114 31 &huh, J., Fairclough, G.F., Jr. and Haschemeyer, R.H.(1978) Proc. Nat]. Acad. Sci. USA 75, 3173-3177 32 Nichols, A.V., Rehnborg, C.S. and Lindgren, D. (1961) J. Lipid Res. 2, 203-207 33 Lee, D.M. (1980) B&hem. Biophys. Res. Commun. 95, 1663-1672 34 Morel, D.W., Cathcart, M.K. and Chisholm, G.M. (1983) J. Cell. Biol. 97, 427a (abstract) 35 Hessler, J.R., Morel, D.W., Lewis, J.L. and Chisholm, G.M. (1983) Arteriosclerosis 3, 215-222 36 Henriksen, T., Mahoney, E.M. and Steinberg, D. (1983) Arteriosclerosis 3, 149-159 37 Ross, R. (1981) Arteriosclerosis 1, 293-311 38 Slater, T.F. (1972) Free Radical Mechanism in Tissue Injury, Pion Ltd., London 39 Lang, J., Heckenast, P., Esterbauer, H. and Slater, T.F. (1984) in Oxygen Radicals in Chemistry and Biology (Bors, W., Saran, M. and Tait, D., eds.), pp. 351-354, Walter de Gruyter and Co., Berlin 40 Hammerstrom, S. and Farlardeau, P. (1977) Proc. Natl. Acad. Sci. USA 74, 3691-3695

41 Lang, J., Jtirgens, G. and Esterbauer, H. (1984) HoppeSeyler’s Z. Physiol. Chem. 365, 931-932 42 Santos, M.T., Valles, J., Aznar, J. and Vilches, J. (1980) J. Clin. Pathol. 33, 973-976 43 Yagi, K. (1984) Bio Essays 1, 58-60 44 Hagihara, M., Nishigaki, P.I., Maseki, M. and Yagi, K. (1984) J. Gerontol. 39, 269-272 45 Poli, G., Dianzani, M.U., Cheeseman, K.H., Slater, T.F., Lang, J. and Esterbauer, H. (1985) Biochem. J. 227, in the press 46 Marnett, L.J., Hurd, H.K., Hollsten, M.C., Levin, D.C., Esterbauer, H. and Ames, B.N. (1985) Mutation Res. 148. 25-35 47 Schauenstein, E. and Esterbauer, H. (1979) in Submolecular Biology and Cancer, Ciba Foundation Series 67, pp. 225-244, Excerpta Medica, Amsterdam 48 Schauenstein, E., Taufer, M., Esterbauer, H., Kylianek, A. and Seelich, T. (1971) Monatsh. Chem. 102, 517-529 49 Glazer, A.N. (1976) in The Proteins, Vol. II, 3rd Edn. (Neurath, H., Hill, R.L. and Boeder, C.L., eds.), pp. 2-103, Academic Press, New York