Formation of N ε-(Hexanonyl)lysine in Protein Exposed to Lipid Hydroperoxide

The objectives of this study were to estimate the structure of the lipid hydroperoxide-modified lysine residue and to prove the presence of the adducts in vivo. The reaction of lipid hydroperoxide toward the lysine moiety was investigated employingN-benzoyl-glycyl-l-lysine (Bz-Gly-Lys) as a model compound of Lys residues in protein and 13-hydroperoxyoctadecadienoic acid (13-HPODE) as a model of the lipid hydroperoxides. One of the products, compound X, was isolated from the reaction mixture of 13-HPODE and Bz-Gly-Lys and was then identified as N-benzoyl-glycyl-N ε-(hexanonyl)lysine. To prove the formation of N ε-(hexanonyl)lysine, named HEL, in protein exposed to the lipid hydroperoxide, the antibody to the synthetic hexanonyl protein was prepared and then characterized in detail. Using the anti-HEL antibody, the presence of HEL in the lipid hydroperoxide-modified proteins and oxidized LDL was confirmed. Furthermore, the positive staining by anti-HEL antibody was observed in human atherosclerotic lesions using an immunohistochemical technique. The amide-type adduct may be a useful marker for the lipid hydroperoxide-derived modification of biomolecules.

During lipid peroxidation, biomolecules such as proteins or aminolipids can be covalently modified by lipid decomposition products. For the case of aliphatic aldehydes (alkanals) such as 1-hexanal or 1-nonanal, the N ⑀ -amino groups of the lysine residues in protein can be modified through the formation of a Schiff base. ␣,␤-Unsaturated aldehydes (alkenals) such as acrolein or 4-hydroxy-2-nonenal react with lysine, cysteine, and histidine through a Michael-type addition (1,2). On the other hand, lipid hydroperoxide might covalently react with protein without serious decomposition of its structure (3). Keto fatty acid (4), which is one of the products by lipoxygenase reaction, can also react with protein and amino acids as previously suggested (5)(6)(7). In addition, the pyrrole compounds from long chain epoxides and lysine were identified (8). However, the mechanism of lipid hydroperoxide-derived protein modification is not so clear.
To estimate the structure after lipid hydroperoxide-derived lysine modification, the reaction of 13-hydroperoxyoctadecadienoic acid   1 with N-benzoyl-glycyl-L-lysine (Bz-Gly-Lys) was investigated. In this study, a novel compound, N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine (named HEL), was identified as one of the lipid hydroperoxide-modified lysine residues. The formation of HEL in lipid hydroperoxide-modified proteins including oxidatively modified LDL was confirmed using the specific antibody to the HEL residue. In addition, the HEL moiety was detected in human atherosclerotic plaques by immunohistochemical approach. As far as we know, the formation of an amide-type adduct has not been previously reported. This novel adduct derived from lipid hydroperoxide may become an initial marker for the oxidative damage of biological molecules in vivo.
Preparation of Lipid Hydroperoxides-13-HPODE was prepared by the enzymatic reaction of lipoxygenase with linoleic acid (9,10). 15-Hydroperoxyeicosatetraenoic acid (15-HPETE) was prepared as described previously (9,10). Methyl linoleate hydroperoxide (MLOOH) was prepared by the reaction of soybean lipoxygenase with methyl linoleate (ML). A 200-mg sample of ML and sodium deoxycholate (1.62 g) was dissolved in 240 ml of 200 mM borate buffer (pH 9.0). Lipoxygenase (100 mg, Sigma type I-B) was added to the solution and reacted for 3 h at room temperature. The formed peroxide was extracted twice with an * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed. equal amount of chloroform/methanol (1:1). The collected chloroform layer was evaporated. The obtained peroxide was purified by thin layer chromatography (TLC) and developed with n-hexane/ether (6:4). The peroxide was extracted with CHCl 3 and then evaporated. The amount of MLOOH was calculated from the molar coefficient, ⑀ 234 nm ϭ25000 M Ϫ1 cm Ϫ1 using the value of linoleic acid hydroperoxide (11). Reaction Conditions-Bz-Gly-Lys (5 mM) and lipid-free BSA (5 mg/ ml) were typically incubated with 13-HPODE, 15-HPETE, or MLOOH (5 mM) at 37°C in 0.1 M phosphate buffer (pH 7.4) for 3 days. The lipid hydroperoxide-modified proteins were isolated by ethanol precipitation as already described (9,10).
Isolation and Structural Determinations of Compound X-To obtain 13-HPODE-modified Bz-Gly-Lys, 5 mM Bz-Gly-Lys was incubated with 5 mM 13-HPODE for 3 days at 37°C in phosphate buffer and then freeze-dried. The sample was extracted with methanol to remove the large amounts of inorganic salts. The extract was evaporated, dissolved in H 2 O, and then applied to gel filtration chromatography (TOYOPEAL HW-40F, 1.5 ϫ 50 cm) with H 2 O as an eluent at a flow rate of 0.8 ml/min. The fractions (5 ml each) were monitored by absorbance at 234 nm and lipofuscin-like fluorescence (excitation, 350 nm; emission, 420 nm) using a JASCO Ubest-50 UV-visible spectrophotometer and Hitachi F2000 fluorescence spectrophotometer, respectively. The fluorescent fractions 29 -34 were used for further identification because the fluorescence might be considered as a marker of lipid amine adducts. The fluorescent fractions were concentrated and then applied to a Sep-Pak cartridge (Waters) with 0 -100% methanol (20% stepwise) elution. The 20% methanol fraction was used for the isolation of the modified lysine derivative, because it had the strongest fluorescence. The fraction was next applied to reversed-phase HPLC (Develosil ODS-HG-5 (8 ϫ 250 mm), Nomura Chemical Co.) and then fractionated using gradient elution (solvent A, 0.1% acetic acid/CH 3 CN (7/3); solvent B, 0.1% acetic acid/CH 3 CN (1/1)) at a flow rate of 2.0 ml/min. The gradient program was as follows: 0 min (B 0%), 10 min (B 0%), 50 min (B 100%), 60 min (B 100%), and 61 min (B 0%). The elution was monitored by UV absorbance at 234 nm. The peak (retention time 30 min) was further purified by repeated reversed-phase HPLC. The obtained compound X weighed 1.3 mg. Spectral data of the isolated compound X are as follows: 1 H NMR (CD 3 OD) (ppm) 0.80 (t, J ϭ 6.9Hz, 3  Synthesis of N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine-For the first step, benzoyl-glycine (1 eq.) and the N ⑀ -(carboxybenzoyl)lysine methyl ester (1 eq.) were conjugated in dimethylformamide (DMF) with EDC (1.1 eq.) in the presence of an enhancer, NHS (1.1 eq.), as described previously (12) with some modifications. After an overnight reaction at room temperature, the reaction mixture was dissolved with ethyl acetate and then washed with equal amounts of 1 N HCl, water, 5% NaHCO 3 , and then water. The residual ethyl acetate layer was passed through Na 2 SO 4 for dehydration. The eluent was concentrated, and the crude N-benzoyl-glycyl-N ⑀ -(carboxybenzoyl)lysine methyl ester was crystallized with water/ethanol at 4°C for 3 h. The removal of carboxybenzoyl from the purified peptide was performed using Pd-C under H 2 for 3 h at room temperature in water/methanol. The obtained N-benzoyl-glycyl-L-lysine methyl ester was purified by preparative reversedphase HPLC (Develosil ODS-5 (20 ϫ 250 mm), Nomura Chemical Co.) using 0.1% trifluoroacetic acid, CH 3 CN (5/3) as the eluent. Hexanoic acid and the N-benzoyl-glycyl-L-lysine methyl ester were conjugated with EDC and NHS as described previously. The reaction mixture was washed as already described, and the residual product was purified by reversed-phase HPLC on a Develosil ODS-5 (20 ϫ 250 mm) using 0.1% trifluoroacetic acid, CH 3 CN (5/3) as the eluent. The N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine methyl ester was treated with 0.25 N NaOH at 37°C for 1 h to remove the methyl ester. The obtained compound was purified by reversed-phase HPLC on the column using 0.1% trifluoroacetic acid, CH 3 CN (5/3) as the eluent. The identification was performed by 1 H NMR and mass spectroscopy. The spectral data of the synthetic N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine are as follows: 1  N-Benzoyl-glycyl-N ⑀ -(hexanonyl(D-11))lysine derivative was prepared using D-11-hexanoic acid as follows. Briefly, benzoyl-glycyl-Llysine was conjugated with D-11-hexanoic acid using EDC and NHS as coupling reagents (12). The obtained benzoyl-glycyl-N ⑀ -(hexanonyl)-lysine was isolated and purified by reversed-phase HPLC on a Develosil ODS-HG-5 (8 ϫ 250 mm) equilibrated with 0.1% trifluoroacetic acid, CH 3 CN (5/3) at a flow rate of 2.0 ml/min. The elution was estimated by UV absorbance at 234 nm, and identification of the synthetic deuterified hexanonyl compound was performed by liquid chromatographymass spectrometry (LC-MS).

Synthesis of N-acetyl-glycyl-N ⑀ -(hexanonyl)-L-lysine Methyl Ester-
The N-acetyl-glycyl-N ⑀ -(hexanonyl)-L-lysine methyl ester (N ⑀ -hexanonyl AGLME) was prepared by conjugation between AGLME and hexanoic acid using EDC as the coupling reagent and NHS as the enhancer, as described previously. The synthetic compound could not be separated with ethyl acetate/water fractionation because of its high water solubility. Therefore, the reaction mixture was diluted with 0.1% trifluoroacetic acid and passed through a Sep-Pak cartridge. The cartridge was washed with 0.1% trifluoroacetic acid, and the products were then eluted with 0.1% trifluoroacetic acid, CH 3 CN (1/1). The eluent was concentrated and applied to preparative reversed-phase HPLC (Develosil ODS-5 (20 ϫ 250 mm)) using 0.1% trifluoroacetic acid, CH 3 CN (7:3) as the eluent. The elution was monitored by absorbance at 215 nm. The peak was collected and concentrated. The obtained N-acetyl-glycyl-N ⑀ -(hexanonyl)-L-lysine methyl ester (N ⑀ -hexanonyl AGLME) was identified by 1 H NMR and mass spectroscopy (LC-MS). The spectral data of N ⑀ -hexanonyl AGLME are as follows: 1  Amino Acid Analysis-Samples were hydrolyzed with 6 N HCl in vacuo at 105°C. The hydrolysates were dried, dissolved in citrate buffer (pH 2.2), and then applied to an amino acid analyzer, JLC-500 (JEOL).
Preparation of Antibody against Hexanonyl Keyhole Limpet Hemocyanin-The conjugation of hexanoic acid with proteins was performed as follows. Hexanoic acid (2.3 mg), EDC (4.5 mg), and sulfo-NHS (5 mg) were dissolved in 400 l of dimethylformamide, and the reaction mixture was incubated for 24 h at room temperature. To the solution, 0.95 ml of KLH or BSA (10 mg in 0.1 M phosphate buffer (pH 7.4)) was added and further incubated for 4 h at room temperature. The obtained hexanonyl proteins were dialyzed against phosphate-buffered saline (PBS) for 3 days at 4°C. The hexanonyl KLH was emulsified with an equal volume of complete Freund's adjuvant to a final concentration of 0.5 mg/ml, and 1 ml of the solution was then intramuscularly injected into a New Zealand White rabbit. After 4 weeks, 1 ml of the hexanonyl KLH emulsified with an equal volume of incomplete adjuvant (0.5 mg/ml) was injected as a booster every 2 weeks until an adequate antibody generation occurred. Hexanonyl BSA was used for the evaluation of the antibody generation specific to hexanonyl protein.
Oxidized lipid-modified proteins were prepared as follows. Lipid (linoleic acid and arachidonic acid, 5 mM; cardiolipin, 1 mg/ml) was oxidized by 5 mM ascorbic acid and 0.05 mM FeCl 3 for 24 h at 37°C in PBS containing 20% methanol. To the reaction mixture, lipid-free BSA (final concentration, 5 mg/ml) was added and further incubated at 37°C for 3 days. To isolate the modified proteins, an equal amount of CHCl 3 : CH 3 OH (2:1) was added, vigorously mixed, and then centrifuged for 10 min at 4°C. The lower layer was discarded, and an equal amount of CHCl 3 was added and mixed. After centrifugation, the lower layer was discarded again. To the residual upper layer, 9 volumes of ice-cool ethanol was added and kept for 45 min at 4°C. After centrifugation, the pellet was dissolved in water with sonication. The protein solution was dialyzed against water for 2 days at 4°C. Aldehyde-modified proteins were prepared as already described (9,10). The preparation of lipid hydroperoxide-modified proteins was as follows. The hydroperoxides (13-HPODE or 15-HPETE) were incubated with lipid-free BSA at 37°C in 0.1 M phosphate buffer (pH 7.4) for 3 days. Oxidized BSA was prepared by the incubation of lipid free-BSA with hydrogen peroxide/ metal ion (9,10). Modified proteins were dialyzed against PBS at 4°C for 2 days. The concentration of all modified proteins was measured by a BCA assay kit (Pierce).
Modification of LDL-Human LDL was isolated from healthy volunteers using density centrifugation (15). The modification of LDL was performed by copper ion and 2,2Ј-azobis(2-amidinopropane)dihydrochloride (AAPH) as described previously (9). The modification by copper ion was performed by incubation of LDL (0.2 mg/ml) with 50 M CuSO 4 in PBS at 37°C. AAPH-induced oxidation of LDL was carried out by incubation of AAPH (0 -5 mM) with LDL (0.2 mg/ml) in PBS at 37°C for 24 h. The reaction was terminated by the addition of 10 M butylated hydroxytoluene and 1 mM EDTA. The measurements of lipid peroxidation were performed by the following two methods. The generation of thiobarbituric acid reactive substance was measured as described previously (9, 10). The formation of lipid peroxide was measured by a Determiner LPO kit (Kyowa medix), a colorimetric method based on the reaction of lipid peroxides with a methylene blue derivative in the presence of hemoglobin (16).
LC-MS Measurement-The sample was applied to a liquid chromatograph on a Develosil ODS-HG-5 (4.6 ϫ 250 mm), which was connected with a mass spectrometer (PLATFORM II, VG Biotech). The separation was performed by a two-pump gradient. The solvent A for AGLME was 0.1% acetic acid; solvent B for AGLME was CH 3 CN. For the Bz-Gly-Lys system, solvent A was 0.1% acetic acid, CH 3 CN(7/3), and solvent B was 0.1% acetic acid, CH 3 CN (3:7). The gradient programs were as follows: AGLME, 0 min, A 100%; 70 min, A 30%; 75 min, A 30%; 80 min, A 100%. Bz-Gly-Lys, 0 min, A 100%; 30 min, A 0%; 35 min, A 0%; 40 min, A 100%. The electrospray ionization (positive) mode was used for the detection. For the measurements of the N ⑀ -hexanonyl derivative of Bz-Gly-Lys, deuterified hexanonyl Bz-Gly-Lys was added to samples at a concentration of 19 M before the analysis as an internal standard.
Enzyme-linked Immunosorbent Assay (ELISA)-Indirect noncompetitive ELISA was performed as already described (9,10). Briefly, 50 l of antigen in PBS was dispensed into a well and kept at 4°C overnight. After blocking with Block Ace (Dainihon Seiyaku, Osaka, Japan), 100 l of antiserum (1/5000 in PBS containing 0.5% BSA) was added to the well. The binding of the antibody on the coated antigen was evaluated as already described (9,10).
The cross-reactivity of the low molecular weight compound with antibody was investigated by indirect competitive ELISA (9, 10). As a coating agent, 50 l of hexanonyl BSA (0.5 g/ml) was pipetted onto wells and kept at 4°C overnight. At the same time, 50 l of antiserum (1/2500 in PBS containing 1% BSA) and 50 l of sample were mixed in an Eppendolf tube and reacted at 4°C overnight. The plate was washed, and 90 l of the reacted solution was pipetted onto a well. The binding of the residual antibody on coated hexanonyl BSA was estimated as described previously (9,10).
Reaction between Preincubated 13-HPODE and Lysine Residue- The   FIG. 1. Separation of lipid-Lys adducts derived from the reaction of 13-HPODE and Bz-Gly-Lys. A, the reaction mixture was concentrated and extracted with methanol. The extract was applied to gel filtration chromatography (HW-40). The elution was monitored by absorbance at 234 nm and fluorescence (excitation, 350 nm; emission, 420 nm). Fractions 29 -34 were used for further isolation. B, the concentrated fractions were further fractionated using Sep-Pak cartridge. An aliquot of the 20% methanol fraction was applied to reversed-phase HPLC on a Develosil ODS-HG-5 (4.6 ϫ 250 mm, Nomura Chemical Co.) using a linear gradient of the two-solvent system at a flow rate of 0.8 ml/min. The elution was monitored by absorbance at 234 nm. Solvent A (0.01% trifluoroacetic acid) and solvent B (CH 3 CN) were used for the gradient. The gradient employed was as follows: B 0% to B 60% in 60 min, B 60% to B 60% in 5 min, B 60% to B 0% in 10 min. Immunohistochemical Analysis-Tissue sections were prepared from frozen arteries (8 m thick). Before immunostaining, frozen sections were fixed by incubation in ice-cold acetone for 20 min. Sections were incubated with 10% normal goat serum in PBS (20 min) to block nonspecific binding before staining and then with primary antibody (1:350 dilution) for 1 h at room temperature. Sections were incubated with 5% normal rabbit serum or the anti-HEL antibody preabsorbed with hexanonyl BSA instead of the primary antibody as negative controls. Immunostaining was performed with anti-rabbit antibody peroxidase-label (1:50 dilution, DAKO) with hydrogen peroxide and 3,3diaminobenzidine tetrahydrochloride as chromogen. Sections were counterstained with aqueous hematoxylin.

Isolation of Lipid Hydroperoxide-modified Lysine
Derivative-To search for the specific hydroperoxide-derived lysine modification, we used Bz-Gly-Lys as the substrate, and the isolation of the 13-HPODE-modified lysine derivative was performed. To remove large amounts of unreacted Bz-Gly-Lys, the reaction mixture was concentrated and then applied to gel filtration chromatography using TOYOPEAL HW-40 (TOSOH) as a gel. The fractions were monitored by absorbance at 234 nm and lipofuscin-like fluorescence (Fig. 1A). As a result of the HPLC analysis of each fraction, fractions 35-40 contained the unreacted substrate, Bz-Gly-Lys. Fractions 29 -34 had a lipofuscin-like fluorescence, which could be considered as a marker of lipid decomposition product-modified molecules. Therefore, the fluorescent fractions were used for further isolation. The mixed fractions 29 -34 were applied to a Sep-Pak cartridge with stepwise methanol elutions. The HPLC profile of the 20% methanol fraction is shown in Fig. 1B. A compound, labeled X, was then isolated by repeated reversed-phase HPLC.
Identification of N ⑀ -(Hexanonyl)lysine Derivatives-The mo-lecular weight of X, 405, was confirmed by FAB-MS. Compound X was hydrolyzed with 6 N HCl at 105°C in vacuo and submitted for amino acid analysis. Interestingly, both Gly and Lys were completely recovered from the acid hydrolysates of compound X (Gly/Lys ϭ 0.96). This suggested that the bond between the lipid-derived structure and Bz-Gly-Lys was acid liable such as an amide bond or a Schiff base. The structure of X was elucidated using 1 H NMR. The proposed structure of X with the parent molecules is shown in Fig. 2A. To confirm the structure of compound X, the synthesis of the N ⑀ -(hexanonyl)lysine adduct was performed by carbodiimide conjugation of the lysine derivative with hexanoic acid. The instrumental analysis of the synthetic N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine almost agreed with that of isolated compound X. Neither the isolated compound X nor the synthetic N ⑀ -hexanonyl adduct had any fluorescence. The time-dependent changes in Bz-Gly-Lys during incubation with 13-HPODE were examined. As shown in Fig. 2B, a loss of Bz-Gly-Lys was observed in a time-dependent fashion, and the formation of N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine, compound X, was confirmed. The conversion yield of compound X from the loss of Bz-Gly-Lys after a 3-day incubation with 13-HPODE was 5.6%.

FIG. 3. Comparison of HPLC-MS between synthetic N ⑀ -hexanonyl AGLME and 13-HPODE-modified AGLME.
The structure of N ⑀ -hexanonyl AGLME is shown in the chart (A). The aliquots of the reaction mixture and synthetic compound were analyzed by reversed-phase HPLC on a Develosil ODS-HG-5 column as described under "Experimental Procedures" (B).
To further confirm the formation of the N ⑀ -(hexanonyl)lysine named "HEL" in the 13-HPODE-modified Lys, the AGLME was incubated with 13-HPODE, and the formation of the HEL derivative was investigated. After a 3-day incubation, an aliquot of the reaction mixture was applied to LC-MS. The product, which shows m/z 358 as an (M ϩ H) ϩ ion, corresponding to N ⑀ -hexanonyl AGLME (M r 357), was eluted at a retention time of 40.79 min, and this completely agreed with the elution time of the synthetic N ⑀ -hexanonyl AGLME (Fig. 3). The mass charts of both the product and the synthetic N ⑀ -hexanonyl AGLME showed the same fragmentation pattern.
The Mechanism for Formation of HEL-HEL was formed by the reaction of AGLME with 15-HPETE as well as 13-HPODE (Fig. 4, A and B). The reaction of methyl linoleate hydroperoxide with AGLME could also generate the HEL derivative (Fig.   4C). The relative ratio of the formation of HEL from 13-HPODE, 15-HPETE, and MLOOH was 1:0.49:0.47. These results suggested that the adduct should be formed from not only free fatty acid hydroperoxides but also esterified fatty acid hydroperoxides, such as cholesteryl ester hydroperoxide and phosphatidylcholine hydroperoxide.
Hexanal, one of the lipid decomposition products, can react with lysine and form a Schiff base. However, HEL does not have a Schiff base in its structure. To deny the participation of 1-hexanal in the amide bond formation, the formation of N ⑀hexanonyl AGLME during incubation of AGLME with aldehyde was examined by LC-MS. HEL was not formed from the reaction of 1-hexanal with AGLME.
Furthermore, the catalytic activity of peroxide on the formation of the amide bond was also investigated. The N ⑀ -hexanonyl adduct was not generated by the reaction of 5 mM AGLME with 5 mM hexanoic acid (1-hexanal) in the absence or presence of 5 mM tert-butyl hydroperoxide. This result revealed that the formation of HEL was not derived from the decomposition prod-  8. Formation of antigenic materials in proteins modified by lipid peroxidation products. Linoleic acid, arachidonic acid, and cardiolipin were oxidized by ascorbate/iron, and the obtained oxidized lipids were then reacted with BSA for 3 days at 37°C. Oxidized BSA was prepared by the treatment of lipid-free BSA with H 2 O 2 /iron/EDTA as already described (9). 13-HPODE-and 15-HPETE-modified BSAs were made as described (9,10). These modified proteins including native BSA were coated at a concentration of 0.01 mg/ml. After blocking, the antiserum (1/5000) was added to the wells. The evaluation of binding was performed as described in the Fig. 5 legend. ucts, hexanal or hexanoic acid. The existence of other unknown precursors is suggested.
Preparation of Antibody to HEL-It is difficult to detect HEL moiety in protein molecules or tissue samples by chemical methods because HEL is unstable for acid hydrolysis from its amide linkage and does not have any specific absorbance. Therefore, the preparation of an antibody specific to HEL was planned. The antibody was prepared by injection of hexanonyl KLH as an immunogen, and the production of the antibody, which reacts with hexanonyl BSA, was observed. A detailed characterization of the obtained antiserum was then performed. At first, we examined the cross-reactivity of the antibody with amide-type synthetic adducts (CH 3 -(CH 2 ) n -CO-NH-Lys; n ϭ 0 -9) by ELISA. As shown in Fig. 5, hexanonyl (n ϭ 4) protein has been strongly recognized. Heptanonyl (n ϭ 5) and pentanonyl (n ϭ 3) proteins could be reacted with the antibody to a lesser extent. It is important that propanonyl (n ϭ 1) BSA could not be bound by the antibody, suggesting that the antibody can be used for the oxidative modification of -6 fatty acids but not -3 ones. Similar results were obtained by indirect competitive ELISA (data not shown). In addition, a carboxyalkylated protein (HOOC-(CH 2 ) n -CO-NH-Lys) such as glutaric acid-BSA (n ϭ 3) or azelaic acid-BSA (n ϭ 7) could not be recognized by the anti-HEL antibody.
Lipid oxidation leads to the formation of reactive aldehydes such as 1-hexanal, malondialdehyde, and 4-hydroxy-2-nonenal. These reactive aldehydes can react with biological molecules such as proteins (1, 2). The cross-reactivity of aldehyde-modified proteins with the anti-HEL antibody was examined by ELISA. As shown in Fig. 6, aldehyde-modified proteins used were not recognized by the antibody, whereas hexanonyl BSA was.
To prove that the epitope of the antibody is N ⑀ -(hexanonyl)lysine (HEL), the cross-reactivity of the synthetic peptide-containing HEL moiety with the antibody was investigated by competitive ELISA. Fig. 7 shows that N-acetyl-glycyl-N ⑀ -(hexanonyl)lysine methyl ester and N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine could be recognized by the antibody. This suggested that the antibody is specific to the N ⑀ -(hexanonyl)lysine moiety.
Formation of HEL by Peroxidized -6 Fatty Acids-As shown in Fig. 8, the formation of HEL was observed by incubation of BSA with ascorbate/Fe 2ϩ -oxidized linoleic acid, arachidonic acid, and cardiolipin. The treatments of BSA with 15-HPETE as well as 13-HPODE also generated antigenic compounds. The result may suggest that the N ⑀ -(hexanonyl)lysine, HEL, becomes a marker for the oxidative modification of lysine by oxidized -6 fatty acids including phospholipids containing esterified -6 fatty acids. In addition, oxidized BSA, which was prepared by the oxidation of protein by hydrogen peroxide/iron, could not generate the antigenic materials.
The Effect of the Preincubation of 13-HPODE on the Formation of HEL-To clarify whether the formation of HEL becomes a marker for the oxidative damage of protein by lipid hydroperoxide, the effect of the preincubation of 13-HPODE on the formation of N ⑀ -(hexanonyl)lysine was examined by chemical and immunochemical methods (Scheme I). The loss of 13-HPODE during preincubation was evaluated (as 13-HODE) by FIG. 9. Effect of preincubation of 13-HPODE on the formation of HEL moiety. The 13-HPODE was incubated in 0.1 M phosphate-buffered saline (pH 7.4) at 37°C at various intervals. An aliquot of the solution was harvested and analyzed. A, the residual amounts of 13-HPODE were estimated after conversion to the 13-HODE, followed by HPLC analysis. B, squares, the preincubated 13-HPODE was incubated with Bz-Gly-Lys for 3 days at 37°C, and the formation of N ⑀ -hexanonyl compounds was estimated by LC-MS. B, triangle, the preincubated 13-HPODE was incubated with BSA for 3 days at 37°C, and the formation of HEL residues in the protein was evaluated by ELISA. SCHEME I. Flow diagram of the preincubation experiment. The 13-HPODE was prepared from linoleic acid by lipoxygenase. The obtained 13-HPODE was incubated at neutral pH. An aliquot of the preincubated solution was used for the estimation of the residual 13-HPODE. The preincubated solution was divided, one portion was further incubated with Bz-Gly-Lys, and the other was further incubated with BSA. The detailed protocol is shown under "Experimental Procedures," and the result is summarized in Fig. 9. reduction with NaBH 4 followed by HPLC analysis. The preincubated 13-HPODE was reacted with Bz-Gly-Lys and further incubated for 3 days at 37°C. The formation of HEL moiety was evaluated by LC-MS. During a 3-day preincubation, about 90% of 13-HPODE was decomposed (Fig. 9A). The formation of HEL from Bz-Gly-Lys and 13-HPODE was also decreased with increasing preincubation time (Fig. 9B, squares). The preincubated 13-HPODE was also mixed with lipid-free BSA, and the formed HEL moiety in BSA was then estimated by the anti-HEL antibody. A similar result was observed using BSA as a FIG. 10. Formation of immunoreactive compounds in copper-oxidized LDL. LDL was oxidized by 50 M copper ion at 37°C at various intervals. The oxidation was terminated by adding EDTA/ butylated hydroxytoluene. Control means the omission of copper ion. A, the oxidation of LDL was evaluated by the formation of thiobarbituric acid reactive substrates (TBARS). The value was expressed as an equivalent of malondialdehyde (MDA). B, the formation of lipid hydroperoxide during the oxidation of LDL with copper ions was evaluated by a Determiner LPO kit (16). The value was expressed as an equivalent of cumene hydroperoxide. C, the formation of immunoreactive materials was estimated by ELISA using anti-HEL antibody as described in the Fig. 5 legend.
FIG. 11. Formation of immunoreactive materials in azo compound-oxidized LDL. LDL was incubated with various concentration of AAPH at 37°C for 24 h. The oxidation was terminated by adding EDTA/butylated hydroxytoluene. A, the oxidation of LDL was evaluated by the formation of thiobarbituric acid reactive substrates (TBARS). B, the generation of lipid hydroperoxide was estimated as described in the Fig. 10 legend. C, the formation of immunoreactive materials was estimated by ELISA using anti-HEL antibody. substrate (Fig. 9B, triangles). The result was in agreement with the data of the negative reactivity against aldehyde-modified proteins by the antibody (Fig. 6). These results suggested that the N ⑀ -(hexanonyl)lysine was derived from 13-HPODE itself or its slightly modified compounds but not from considerably decomposed compounds such as aldehydes.
The Formation of HEL in Oxidatively Modified LDL-The oxidative modification of LDL has been considered as a plausible inducer of atherosclerosis. We investigated the formation of the HEL moiety in copper-oxidized LDL using the antibody. During the incubation of LDL with copper ion (CuSO 4 ), lipid peroxidation proceeded as evaluated by the formation of thiobarbituric acid reactive substance (Fig. 10A) and lipid peroxide (Fig. 10B). As shown in Fig. 10C, antigenic compounds in LDL were increased with increasing incubation time. Using another initiator, AAPH, similar results were observed (Figs. 11, A-C).
The Presence of HEL Moiety in Human Atherosclerotic Lesion-HEL moiety was formed during the oxidation of LDL (Figs. 10 and 11). To investigate the presence of HEL moiety in atherosclerotic lesion, immunohistochemical detection was performed. As shown in Fig. 12A, HEL positive staining in human atherosclerotic lesion was observed, whereas the use of a nonspecific antibody (normal rabbit serum) in the procedure caused the disappearance of the positive staining in the lesion (Fig. 12B). In addition, no positive staining was observed with the anti-HEL antibody preabsorbed with hexanonyl BSA (data not shown). The result suggests that the immunopositive materials to anti-HEL antibody are present in the human atherosclerotic lesion. DISCUSSION Lipid decomposition products can modify biological materials. Among the products, the reactivities of malondialdehyde and 4-hydroxy-2-nonenal have been investigated in detail. The chemical structures of modified amino acid residues were identified in vitro (1). The antibodies raised to malondialdehyde or 4-hydroxy-2-nonenal-modified protein were prepared, and the immunopositive materials were detected in various tissues such as atherosclerotic plaques (17,18). These findings suggested that the modification of lipid decomposition products (advanced lipid peroxidation end products) occurred in vivo. On the other hand, considerable amounts of lipid hydroperoxides exist in vivo (19).
The fluorescence formation from lysine modification by linoleic acid hydroperoxide was previously observed (20). It has also been reported that the lipid hydroperoxide can react with protein, followed by the formation of the lipid-protein covalent adduct (3). However, the precise structures of the lipid hydroperoxide-lysine adduct are unknown. We report the identification of a novel adduct, N ⑀ -(hexanonyl)lysine, which has an amide bond between the N ⑀ -amino group and lipid-derived part, from the reaction mixture of 13-HPODE and Bz-Gly-Lys. As far as we know, the formation of an amide bond during lipid peroxidation has not been reported. The chemical approach for the detection of N ⑀ -(hexanonyl)lysine, HEL, in a protein is probably difficult because the HEL moiety can be hydrolyzed with 6 N HCl, which is often used for the detection of modified amino acid residues. Therefore, to detect HEL immunochemically, we prepared an antibody against the hexanonyl protein.
The antibody reacted with N-benzoyl-glycyl-N ⑀ -(hexanonyl)lysine (compound X) as well as the N ⑀ -(hexanonyl)lysine residue in the protein. Using the anti-HEL antibody, the formation of N ⑀ -(hexanonyl)lysine in both the 13-HPODE-modified protein and 15-HPETE-modified protein was proven (Fig. 8). This result suggests that the formation of the HEL may be a good marker for the oxidative modification by oxidized -6 fatty acids such as linoleic acid or arachidonic acid. The lipid hydroperoxide can become a precursor of the N ⑀ -hexanonyl adduct, whereas it is unknown whether the reaction of the lipid hydroperoxide and lysine proceeds directly or indirectly.
We also observed the appearance of immunoreactivity against the anti-HEL antibody during the oxidation of LDL. The uptake of oxidized LDL by macrophage can be considered as one of the plausible contributors for foam cell formation, which may initiate atherosclesosis. The presence of HEL moiety in human atherosclerotic lesion was immunohistochemically proven by the anti-HEL antibody. However, the chemical identification of the HEL moiety in atherosclerotic lesion was not performed. More detailed studies are needed for the elucidation of the formation of the HEL moieties in vivo.
Kim et al. (21) prepared antibodies against lipid hydroperoxide-modified protein and reported the positive staining in early atherosclerotic lesion, whereas the precise epitopes were not so clear. Recently, we prepared antibodies to 13-HPODE-or 15-HPETE-modified proteins, which cannot recognize aldehyde-modified proteins (9,10). The result assumed that the lipid hydroperoxide-specific modification can occur. However, N-benzoyl-glycyl-N ⑀ -(hexanonyl)-L-lysine (compound X) and the hexanoyl protein cannot cross-react with the antibodies raised against the 13-HPODE-or 15-HPETE-modified KLH (data not shown). This result agreed with the fact of the requirement of the carboxyl terminus of the lipid moiety in an adduct for the appearance of the antigenicity (9, 10). On the FIG. 12. Immunohistochemical detection of HEL moieties in human atherosclerotic lesions. Tissue sections (aorta) were prepared from frozen arteries (8 m thick). Sections were incubated with 10% normal goat serum in PBS (20 min) to block nonspecific binding before staining and then with anti-HEL antibody (1:350 dilution) (A). Sections were incubated with 5% normal rabbit serum instead of the primary antibody as negative controls (B). Immunostaining was performed with anti-rabbit antibody peroxidase-label (1:50 dilution, DAKO) and with hydrogen peroxide and 3,3-diaminobenzidine tetrahydrochloride as chromogen. Sections were counterstained with aqueous hematoxylin. other hand, the synthetic carboxyalkylamides (HOOC-(CH 2 ) n -CO-NH-Lys), glutaric acid-BSA and azelaic acid-BSA, were not reacted with the anti-HEL antibody (Fig. 6). The anti-HEL antibody may become a better tool for lipid hydroperoxidederived oxidative modification than the anti-13-HPODE-KLH and 15-HPETE-KLH antibody because the anti-HEL antibody recognized the CH 3 terminus of the lipid-Lys adduct, which is formed from the reaction of lysine residues and peroxidation products, derived from not only free fatty acids but also esterified fatty acids (Fig. 4).
In summary, we isolated and identified a novel lipid-Lys adduct, N ⑀ -(hexanonyl)lysine, from the reaction between 13-HPODE and Lys, although the detailed mechanism of formation of the N ⑀ -hexanonyl linkage remains unknown. We also showed the preparation of the antibody to HEL and the formation of HEL in oxidatively modified LDL using the immunochemical method. The immunopositive materials were also observed in human atherosclerotic lesions. The adduct may become a marker for the initial stage of oxidative damage of biomolecules.