The unique reactivity of EKODE lipid peroxidation products allows in vivo detection of inflammation

Significance

The identification of epoxyketooctadecenoic acids (EKODEs) reveals significant insights into lipid peroxidation and inflammatory diseases. EKODEs, derived from linoleic acid, display unique electrophilic properties, interacting with biological nucleophiles. Our study shows EKODEs form stable adducts with cysteine residues via a rapid Michael addition followed by epoxide opening at an unexpected site. This mechanism generates advanced lipoxidation end products, useful as biomarkers for oxidative stress and inflammation. By developing polyclonal antibodies targeting EKODE-Cys adducts, we offer a powerful tool for detecting and monitoring inflammatory diseases, enhancing diagnostic and therapeutic approaches for oxidative stress-related conditions.

Abstract

Lipid peroxidation is a complex biochemical process associated with oxidative stress, and its products play crucial roles in cellular signaling and the pathophysiology of many diseases. Among the diverse array of lipid peroxidation (LPO) products, epoxyketooctadecenoic acids (EKODEs) have emerged as intriguing molecules with potential impacts on inflammatory diseases. EKODEs arise from linoleic acid reacting with reactive oxygen and nitrogen species present during inflammation. A hallmark of many LPO products is an electrophilic chemical functionality that can react with different biological nucleophiles to form adducts that impact a broad swath of physiologic processes. Here, we present the identification of reactivity patterns exhibited by the EKODE class of LPO products that arise due to the unique chemistry of the EKODE electrophiles, namely α, β-unsaturated epoxyketones of variable regiochemistry. Our initial investigations with models of the EKODE reactive core showed that surrogates of lysine did not react, and histidine nucleophiles formed reversible Michael adducts. However, when models of cysteine nucleophiles were tested, a unique reactivity profile emerged where rapid Michael addition was followed by slow rearrangement and epoxide opening at an unpredicted electrophilic site, affording what we postulated to be an advanced lipoxidation end product (ALE). After confirming the EKODE reactivity in model systems, we produced polyclonal antibodies of a stable epitope of the EKODE-based ALE and used these antibodies to investigate an approach for in vivo monitoring of inflammatory disease progression.

Oxidative stress can lead to structural changes in biomolecules. In this regard, polyunsaturated fatty acids (PUFAs) are particularly susceptible to peroxidation (1). The resulting lipid hydroperoxides (LOOHs) can further evolve into two types of lipid peroxidation (LPO) products: oxidative cleavage products (OCPs) and oxidative addition products (OAPs). OCPs, such as 4-hydroxy-2-nonenal (4-HNE) and 4-oxo-2-nonenal (4-ONE), are the diffusible end products of peroxidation (2). In contrast, OAPs, such as EKODEs, result from the oxygenation and rearrangement of bis-allylic functionalities in PUFAs (3) (Fig. 1A). EKODEs have been detected in vivo (4), and concentrations in plasma range from 1 to 500 nM with a mean value of 50 nM in 24 human subjects (5, 6). No studies have investigated the concentrations of EKODEs as components of phospholipids in native membranes, but EKODEs are the most abundant OAPs from linoleic acid as generated under in vitro oxidation conditions (7). As such, it is reasonable to postulate that observed in vivo concentrations dramatically underrepresent the prevalence of these LPO products. EKODEs display prominent biological activities, including stimulating corticosterone production, aldosterone secretion, and adrenal steroidogenesis in rat adrenal cells (8, 9). EKODEs have been further found to effectively activate the pathways controlled by the antioxidant response element promoter (10).

Fig. 1.

EKODE structures and their surrogates and the plausible mechanism of reactivity with thiols. (A) Lipid peroxidation products of linoleic acid. (B) Reaction sites for compounds A and B as model compounds and compounds C-D as surrogate molecules were investigated. (C) Time course of the reactions of N-acetyl-cysteine-methyl ester with trans-EKODE-(E)-Ib and compound A. (D) Time course of the reactions of N-acetyl-cysteine-methyl ester with trans-EKODE-(E)-IIb and compound B. (E) Proposed chemical mechanism of formation for EKODE ALE via a ring-opening reaction of Michael addition products with thiol-containing nucleophiles.

Advanced lipoxidation end products (ALEs) are complex, stable molecules formed through the reaction of carbonyl species (RCS) of OCPs and OAPs with proteins, amino acids, or nucleic acids resulting in condensation reactions generating new molecules that have been implicated in the pathogenesis of various chronic diseases including diabetes, atherosclerosis, and neurodegenerative disorders (11–13). Recent studies have highlighted the role of ALEs in modulating cellular signaling pathways, thereby influencing inflammation, apoptosis, and autophagy mechanisms (14). Therefore, the identification and quantification of ALEs hold significant diagnostic and therapeutic potential in managing oxidative stress-related diseases (15, 16). For example, OCPs can extensively modify human serum albumin (HSA), the most abundant protein in human blood, on multiple nucleophilic residues such as cysteine, histidine, and lysine (17), which has implicated HSA in the detoxification of LPO products. The origin of the work presented here started with an examination of the reactions of EKODEs with HSA. As will be discussed below, mass spectrometry (MS)-based peptide mapping showed peptides with adducted EKODE that primarily corresponded to cysteine as the key reactive nucleophile (SI Appendix, Table S1). In addition, during chemical investigations of the inherent chemical reactivity of EKODE molecules, we identified a unique, unpredicted reactivity profile, which we hypothesized would lead to the accumulation of EKODE-derived ALEs in tissues undergoing unregulated inflammation and concomitant LPO. These promising data motivated us to explore the possibility of the electrophilic reaction of EKODEs with a broader range of protein targets.

Immunochemical methods offer a robust and facile approach for monitoring changes in covalent adduct and protein modifications. In the studies described here, we show a unique reactivity of the EKODE family of LPO products wherein rapid Michael addition is followed by the formation of stable epoxide-opening ALEs with cysteine nucleophiles. To further investigate, EKODE-cysteine adduct haptens were linked to the large carrier protein, keyhole limpet hemocyanin (KLH), for antibody production (Fig. 3). We used the above system to raise an antibody against EKODE-Cys adducts. We subsequently show this antibody exhibits high specificity and sensitivity for covalent adducts of EKODE with proteins. Using this approach, we show that EKODE-Cys adducts are accumulated under oxidative stress conditions, and in a heart perfusion experiment mimicking an ischemic heart, a clear link between the levels of EKODE-Cys adducts is observed as a function of time after induced ischemia (Fig. 4). Thus, the generated polyclonal antibody can find extensive applications as a valuable tool for evaluating EKODE-Cys adducts as a biomarker of oxidative stress both in vitro and in vivo.

Fig. 3.

Chemical synthesis of EKODE-Cys-spacer-KLH conjugate for antibody production. (A) (a) N-acetyl cysteine methyl ester in 20% EtOH/PBS, (b) EDC in DMSO, (c) NHS in DMSO, (d) aminocaproic acid in DMSO/PBS, and (e) KHL/BSA, EDC, NHS. (B) characterization of antigens and coating agents using MALDI-TOF and TNBSA assay. (C) Serum antibody titer in rabbits immunized with EKODE-Cys-spacer-KLH haptens. (D) Cross-reactivity of LPO products-modified proteins with anti-EKODE Ib-Cys and anti-EKODE IIb-Cys antibody.

Fig. 4.

Immunochemical detection of EKODE II-Cys adducts in cells and tissues. (A) Detection and quantification of EKODE II-Cys adducts in M17 cells treated with H₂O₂, n = 3. (B) Detection and quantitation of EKODE II-Cys adducts in human brain tissue at different ages. (C) Distribution of EKODE IIb-Cys adducts in mouse tissues. (D) Detection and quantitation of EKODE-Cys adducts in ischemic mouse heart using anti-EKODE IIb-Cys antibody (β-actin was used as the loading control), n = 3. Data represent mean ± SD.

Results

Mechanistic Studies of the Reactions of EKODE Model Compounds with Nucleophilic Amino Acids.

EKODE I and II and the corresponding model compounds (A and B) and surrogate electrophiles (C, D, and E) were synthesized (Fig. 1 A and B and SI Appendix, Figs. S1–S7). Compounds A and B were incubated with amino acid surrogates in 50 mM PBS buffer (pH 7.4) containing 20% acetonitrile to facilitate solubility and mimic a membrane-interface environment (Fig. 1B). The resulting products were isolated, characterized, and summarized (SI Appendix, Fig. S8). The reaction of compounds A and B with 10 equivalents of n-butylamine at 37 °C for 24 h resulted in no detectable amounts of Schiff base or Michael adducts. This result is in agreement with our previous proteomics study, which found no lysine adducts in a model protein, apomyoglobin, treated with EKODEs (18). These results can be explained by the highly reversible nature of lysine adducts, which can often only be isolated and detected after reduction with NaBH₄ (19, 20). Regardless, in the reaction of compound B with imidazole, two diastereomers of Michael adducts were identified, which is consistent with a previous report for compound A (7). Michael adducts generated from histidine by nucleophilic attack at the β-carbon of the α, β-unsaturated carbonyl has been extensively studied (21). The sulfhydryl group of cysteine residue is the most reactive nucleophile in proteins. It has been broadly reported that thiol groups can attack the epoxides, with ring opening regiochemistry typically occurring at the most electron-deficient position, α to the carbonyl (22, 23). Surrogate electrophiles were designed to specifically test the reactivity profiles of the EKODEs and provide mechanistic insights into adduct formation. Surrogate C with n-butanethiol only displayed Michael addition, while D and E resulted in the expected epoxide opening product, alpha to the carbonyl (Fig. 1B and SI Appendix, Fig. S8).

The reaction of compound A with n-butanethiol produced a ring-opening product at the α position (to the olefin) within 72 h, though the Michael adduct was too unstable to be purified. In contrast, the reaction of compound B resulted in a stable Michael adduct in 1 h and a ring-opening product at the β position with respect to the carbonyl group after 48 h, which was confirmed by ¹H–¹H correlation spectroscopy (HHCOSY). The peaks at 3.52 ppm correspond to the proton of the OH group at C5 (Fig. 1E). This was confirmed by treating the product with CH₃OD before NMR analysis, which completely removed the peaks. The correlation between the OH peaks and the peaks at 4.48 ppm suggested that the OH group is on C5, which is supported by the fact that there is no correlation between peaks at 4.48 ppm and the peaks in the 1 to 2 ppm range (SI Appendix, Fig. S9A). A previous report has shown that OCPs can undergo advanced reactions that involve Michael addition to an α, β-unsaturated carbonyl, followed by aldol condensation (24). Therefore, it is reasonable to postulate that EKODE II initially forms a Michael adduct, which subsequently undergoes an intramolecular rearrangement through a six-membered ring intermediate to form the ring-opening product relaying the sulfur substitute at the β position of the epoxy ring.

Real-time kinetics of the reaction between 10 μM of EKODEs and EKODE model compounds and 100 μM N-acetyl-cysteine-methyl ester were monitored by changes in α, β-unsaturated absorbance using UV spectroscopy at 234 nm, which corresponds to α, β unsaturated ketone. The results show a fast loss of the chromophore, followed by slow recovery of the α, β-unsaturated carbonyl (Fig. 1 C and D). Our interpretation, based on the model proposed above, is that Michael addition of the cysteine is rapid (Fig. 1E), followed by a slow epoxide ring-opening reaction that results in partial recovery of the α, β unsaturated ketone. That there is not full recovery of the absorbance at 234 nm can either be attributed to incomplete epoxide opening or a second Michael addition with cysteine in equilibrium. Additionally, the reaction was monitored for 3 d by LC–MS with 10 μM solution of trans-EKODE-(E)-IIb and 100 μM N-acetyl-cysteine-methyl ester solution in acetonitrile/water. The selected ion currents of trans-EKODE-(E)-IIb and EKODE-Cys adduct were extracted from the total ion current (TIC) using an extraction window of 1.0 Da (m/z 310 ± 0.5 for trans-EKODE-(E)-IIb, and m/z 488 ± 0.5 for EKODE-Cys) (SI Appendix, Fig. S9B).

Mass Spectrometric Characterization of the Modification of HSA by trans-EKODE-(E)-IIb.

10 μM HSA was incubated with 1 mM EKODE at 37 °C overnight in PBS containing 20% acetonitrile. The modified protein was then precipitated and resolubilized in ammonium bicarbonate (AMBIC) buffer, followed by proteolytic digestion using trypsin (SI Appendix, Fig. S10A). The resulting tryptic peptides were analyzed using information-dependent acquisition (IDA) by high-resolution mass spectrometry (HRMS) to reveal modification sites. HSA peptides were subjected to peptide mapping by BioPharmaView software for the presence of +310.2144 mass shift, which corresponds to the addition of EKODE via Michael addition or +292.4131 for the formation of Schiff base with lysine residue. The protein sequence coverage obtained was 89.7%. When HSA was exposed to 1 mM EKODE, 7 modified peptides were identified, and the modification sites were Cys-34, His-67, His-146, His-338, His-440, His-464, and His-535 (Fig. 2A and SI Appendix, Table S1). The observed monoisotopic m/z of the peptide precursors agreed with the theoretical value with an error of less than 10 ppm. In addition, multiple charge states for the peptide containing Cys-34 (+3 and +4 charge) and His-338 (+3 and +4 charge) were identified. No EKODE-modified lysine residue was identified, consistent with the results in our model study that EKODEs were not reactive with lysine or n-butylamine (SI Appendix, Fig. S8)

Fig. 2.

Modification of HSA by trans-EKODE-(E)-IIb. (A) HSA modification sites (PDB ID: 1AO6). (B) LC–MS analysis of EKODE-modified His-338 peptide. Extracted HRMS scan spectrum at the retention time of 5.64 min; (B–D) Zoomed-in HRMS scan of highlighted area. (C) MS/MS product ion scan of peptide containing EKODE-modified His-338 (m/z 593.3595, +3).

The mass spectrum was also manually examined to confirm the site of modification. When an extraction window of 12.5 mDa was used to extract the peptide containing EKODE-modified His-338 (m/z 593.3595, +3), a distinctive peak was observed at the retention time of 5.64 min (SI Appendix, Fig. S10 B and C). The high resolution of the mass spectrometer allowed the use of a narrow extraction window, increasing the specificity. The HRMS scan at 5.64 min was then extracted using an extraction window of 0.12 min to display the mass spectrum. The highlighted peaks correspond to the multiple charge states of His-338-modified peptide, and the zoomed-in spectra are shown in Fig. 2B. The charge states can be confirmed by the isotopic distribution and match charge states +4, +3, and +2. To validate the site of modification, the product ions of the matched peptide were examined. The MS/MS product ion scan of the peptide with EKODE-modified His-338 matched the y- and b- product ion series. The presence of matched product ions y₁₀⁺ and b₂⁺ indicates that the modification site is the histidine residue (Fig. 2C). To identify the most reactive residues, lower concentrations of EKODE (100 and 500 μM) were incubated with 10 μM HSA and subjected to LC–MS analysis. The obtained sequence coverages were 90.4% and 87.7%. When HSA was exposed to 500 μM of EKODE, the modification was found on Cys-34, His-67, His-146, His-338, and His-440 residues. Next, the EKODE concentration was reduced to 100 μM, the lowest ratio where EKODE modification could be detected. At 100 μM, only Cys-34, His-67, His-146, and His-440 were modified, which thus represent the most reactive residues of HSA toward EKODE modification (SI Appendix, Table S2).

Chemical Strategy for EKODE-Cys-Spacer-KLH Conjugate Preparation and Characterization of Antibodies Against EKODE-Cysteine Adduct.

EKODEs were incubated with N-acetyl-cysteine-methyl ester overnight in PBS and 20% ethanol. Purification by column chromatography over silica removed excess reagents, resulting in a combination of EKODE-cysteine Michael adduct and ring-opening product. The terminal carboxylic acid of EKODE-cysteine adducts was activated by 1-ethyl-3-(-3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) to form an active O-acylisourea intermediate, which then reacted with N-hydroxysuccinimide (NHS) to produce a stable amine-reactive NHS ester (Fig. 3A). The amine-reactive NHS ester was mixed with aminocaproic acid to generate an EKODE adduct-spacer conjugate, which mimics the lysine residues on carrier proteins. Therefore, the same reaction conditions were used for the conjugation of EKODE-adduct-spacer conjugates with KLH. The conjugate with bovine serum albumin (BSA) was also produced as the coating agent for immunoassays (SI Appendix, Fig. S12). The molecular weight of intact BSA was around 66,343 Da, but after conjugation, it shifted to ~76,848 Da for EKODE Ib-Cys-spacer-BSA conjugate (indicating a mean of 17.5 conjugations per BSA) and ~78,525 Da for EKODE IIb-Cys-spacer (indicating a mean of 20.3 conjugations per BSA) (SI Appendix, Fig. S11). The protein conjugates were further evaluated by 2,4,6-trinitrobenzene sulfonic acid (TNBSA) assay to determine the number of free ε-amino groups from lysine residues. The conjugation efficiency was calculated by comparing TNBSA with matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry (Fig. 3B).

The KLH conjugates were used to immunize rabbits (Cocalico Biologicals). The antibody titer was monitored by ELISA using the BSA-derived conjugates as a coating agent and goat anti-rabbit IgG (HRP conjugated) as the secondary antibody. The antibody titer increased steadily over 91 d and reached a plateau around day 70 (Fig. 3C). The specific IgG fractions were enriched, and the BSA conjugates of EKODE-Cys adducts were immobilized on aldehyde-activated beaded agarose resin through the Schiff base bond formed with primary amines. The resulting affinity resin was then used for immunocapturing specific antibodies from the final bleed. The unwanted endogenous antibodies were eliminated by washing the resin-bound complex, and the desired IgG fraction was dissociated at low pH. Epitope characterization of the purified antibodies was performed using competitive ELISA. Negative competitors were synthesized by incubating EKODEs with N-acetyl-histidine-methyl ester or 4-HNE with N-acetyl-cysteine-methyl ester (SI Appendix, Fig. S12C). The anti-EKODE-Cys antibodies were preadsorbed with competitors at different concentrations. The mixture was applied in ELISA with the same coating agents and secondary antibody system used to measure antibody titer. Results show that EKODE-His adducts and HNE-Cys adducts did not inhibit antibody binding (SI Appendix, Fig. S12).

Next, BSA was incubated with EKODEs and other common LPO products, including 4-HNE, 4-ONE, acrolein, and 2-octenal. The resulting BSA adducts were resolved using SDS-PAGE and transferred to the membrane for western blot analysis. As shown in Fig. 3D, the antibody did not recognize aldehyde-modified BSA, whereas EKODE adducts displayed a concentration-dependent reactivity. In addition, the anti-EKODE-IIb-Cys antibody was able to differentiate the two EKODE isomers, which was consistent with competitive ELISA results. It is worth mentioning that while anti-EKODE IIb-Cys antibody exhibited excellent specificity toward corresponding EKODE IIb isomers, both EKODE-Ib-Cys and EKODE-IIb-Cys adducts showed a similar level of inhibition toward anti-EKODE-Ib-Cys antibody, indicating that anti-EKODE-Ib-Cys antibody cannot differentiate the two EKODE isomers. BSA, a model protein containing a free cysteine residue Cys-34, was incubated with linoleic acid and an iron/ascorbate-based free-radical-generating system. The formation of EKODE-Cys adducts was monitored immunochemically for 24 h, with a constant increase in the amount of EKODE-derived modification on cysteine residue over time (SI Appendix, Fig. S13).

Immunochemical Detection of EKODE-Cysteine Adducts in Different Cells, Tissues, and Ischemic Heart.

Human neuroblastoma cells (M17) were treated with increasing concentrations of H₂O₂ (0.1 to 0.5 mM) for 24 h to mimic the oxidative damage derived from oxidative stress. Treatment with a high concentration of H₂O₂ (0.5 mM) resulted in loss of cell viability. In contrast, lower concentrations of H₂O₂ did not induce notable cell death, and the cell lysate was analyzed by western blot. Endogenous cysteine adducts of both EKODE-Ib and EKODE-IIb were identified in the control groups that were not treated with H₂O₂ (Fig. 4A). A careful examination of these results shows a similar reactivity profile, which was predictable based on the results shown in Fig. 3D. Based on these results in conjunction and the observed specificity of the EKODE-IIb antibodies together with its unique reactivity, we decided to primarily investigate EKODE-II as an ALE through western analysis moving forward. To assess whether the cell culture western blot were reflected in vivo, the anti-EKODE-IIb-Cys antibodies were tested with human brain tissues from different age groups (Fig. 4B). Interestingly, we observed an age-dependent accumulation of immunoreactivity. To investigate whether the observed immunoreactivity was protein-specific, we investigated nine separate tissue types, and observed tissue-specific immunoreactivity. While the identity of the corresponding proteins is not known, we found it compelling that we observe specificity products within each tissue. Next, we tested whether we could observe an increase in immunoreactivity in tissues experiencing oxidative stress. Specifically, we induced ischemia in perfused rat hearts. Following baseline equilibration, we transitioned to global no-flow ischemia. Hearts were then quick-frozen and homogenized for western blot analysis. Most EKODE-modified proteins in control groups (0 min) were identified at 37 to 50 kDa, consistent with the result in the tissue distribution experiment (Fig. 4C). There was no change in the abundance of EKODE-Cys adduct after 5 min of heart ischemia. However, at 15 and 30 min, the level of adduct formation increased dramatically: 274% and 517%, respectively (Fig. 4D).

Discussion

Understanding the complexity of protein modification with LPO products remains a major challenge. Many LPO products contain an α, β-unsaturated carbonyl as the primary functional group, which is generally known to undergo Michael addition and Schiff base formation (25). Some bifunctional aldehydes, such as 4-HNE, can crosslink proteins via simultaneous conjugation with nucleophilic protein residues (Cys, His, Lys) on the bond and Schiff base condensation with Lys on the carbonyl (18). Here, we described a mechanistic model of the reactivity between EKODEs and biological nucleophiles (Fig. 1B and SI Appendix, Fig. S8). The reactions of EKODEs and EKODE model compounds with N-acetyl-cysteine-methyl ester were monitored by UV spectroscopy (Fig. 2 C and D). For both EKODE-I & -II rapid Michael addition dominates the early stage of the adduct-forming reaction. In the case of EKODE-I molecules, the ring-opening reaction occurs at differential rates when comparing compound A with trans-EKODE-(E)-Ib. In contrast, the absorption recovery and consequently ring-opening reactions of trans-EKODE-IIb and compound B were similar. Our interpretation of these results is that in the case of the EKODE-II series the ring-opening reaction is occurring intramolecularly, wherein for EKODE-I molecules ring-opening can be better explained through an intermolecular model based on our current data. In addition, the ring-opening products of EKODE II were found to be more stable long term. As such, our studies prioritized these stable EKODE II adducts.

The LC–MS analysis of the reaction of EKODE II with N-acetyl-Cysteine-methyl ester showed that the formation of adducts was regioselective and kinetically controlled, with the early stage of the reaction dominated by the appearance of two out of a possible four stereoisomers of the Michael reaction. Given that the reaction may be hindered by the increased steric effect in the ring-opening product, this result agrees with the previous findings that the Michael addition of the EKODE I model compound with imidazole is stereoselective (7). After 3 d, the Michael adducts disappeared, and a new broad peak was detected, tentatively assigned as a mixture of diastereomers of the ring-opening products (SI Appendix, Fig. S9B). Also, we detected a trace amount of what we have tentatively determined to be an EKODE adduct with two additional cysteines, though we have not performed confirmation experiments on this observation. Regardless, this indicates that the ring-opening products of EKODEs might undergo an additional Michael addition with cysteine or other nucleophiles, which could cause protein cross-linking, which has been observed with other OCPs (26).

Beyond the mechanism of the reaction, it has been reported that HSA can be modified by LPO products on multiple nucleophilic residues, including cysteine, histidine, and lysine (17). In this report, a solution of HSA (10 μM) was exposed to various concentrations of EKODE, and we found that at high concentrations of EKODE (1 mM), 1 cysteine and 6 histidine residues were modified, while at medium concentration of EKODE (500 μM), only five modifications, including one on a cysteine residue and 4 on histidine residues could be detected. At the lowest investigated concentration of EKODE (100 μM), the modifications were limited to 1 cysteine and 3 histidine residues. Previous research has shown that HNE modifies HSA, with the most reactive site being Cys-34 (Michael adduct), followed by Lys-199 (Schiff base), His-67, and His-146 (Michael adduct) (27, 28). Cys-34 provides the most substantial fraction of free thiols in serum, and it works as a scavenger for ROS and an endogenous detoxifying agent for reactive carbonyl species (29). Our study showed that EKODEs and HNE exhibit comparable selectivity by readily reacting with Cys-34. If HSA were to undergo adduct formation with an EKODE-II molecule and for a stable ALE, such a covalent modification would dampen the ability of HSA to play its antioxidant role (SI Appendix, Fig. S13). Our data also suggest that the reactions of EKODEs with a variety of cysteines could lead to the formation of irreversible and stable products that may accumulate in the body over time. Therefore, EKODE-derived adducts of HSA at Cys-34 may serve as a valuable biomarker for human diseases associated with oxidative stress.

Encouraged by the previous results, we aimed to develop a tool that could give us a snapshot of the relevance of EKODE adducts in vitro and in vivo. Polyclonal antibodies were generated to EKODE-Ib and -IIb via the generation of a KLH-conjugate of a purified and characterized EKODE adduct hapten. Epitope characterization with competitive ELISA was performed against negative competitors for each purified antibody. To ensure the specificity of the antibodies, anti-EKODE-Cys antibodies were further evaluated by western blot analysis with BSA modified by various LPO products and checked for cross-reactivity. The high specificity of the antibodies might be attributed to the antibody purification step where specific BSA conjugates of EKODE-Cys adducts rather than generic protein G. This allowed the antibodies to recognize the entire structure of the EKODE-Cys adduct, including the EKODE backbone and cysteine moiety, and differentiate it from other LPO-induced protein modifications and modifications at other protein residues. Our experiments showed that when BSA was modified with linoleic acid under nonenzymatic oxidation conditions, there was a noticeable increase in a new protein band with a size between 100 and 150 kDa, observed after 2 h of incubation, which could be a BSA dimer, possibly formed via a cross-linking reaction with an EKODEs (30) (SI Appendix, Fig. S13).

The central nervous system is vulnerable to damage caused by oxidative stress due to its high level of PUFA, high rate of oxygen consumption, and comparatively limited endogenous defense mechanisms against oxidation. The reactive LPO products derived from oxidative damage of polyunsaturated lipids readily modify proteins and DNA. A major focus of the field in this area is the identification and determination of the relevance of these adducts in neurodegenerative disease and aging (31–33). We applied the anti-EKODE-Cys antibodies to identify EKODE adducts in the nervous systems. The levels of EKODE-Cys adduct considerably increased when the cells were exposed to H₂O₂. Exposure to 0.1 mM H₂O₂ for 24 h resulted in a 100% increase in the levels of EKODE-IIb-Cys adduct.

Next, using these antibodies, we investigated the correlation of EKODE-Cys adducts with the brain tissues at different ages. As shown in Fig. 4B, the identities of EKODE-modified proteins appear to be consistent across different age groups, but the amount of modification showed a steady increase with age. For example, the concentrations of EKODE adducts in middle-aged individuals (30 to 40 y) are over 300% higher than in young individuals (Fig. 4B). The presence of these higher levels of EKODE-Cys adducts supports the oxidative stress theory of aging, which attributes the phenomenon of aging to endogenously generated oxidants and the subsequent oxidative damage of biomolecules (34). Since the ring-opening reaction of EKODE with cysteine is irreversible and the resulting adduct is stable, EKODE-Cys adducts may continue accumulating through lifetime.

These results indicated that EKODE-Cys adducts might serve as biomarkers for aging and related diseases such as Alzheimer’s and Parkinson’s diseases. It is worth mentioning that studies using antibodies against HNE-derived protein adducts showed no correlation with aging (35). This might be explained by the fact that EKODEs are generated under less severe oxidative conditions and are, therefore, more sensitive to the normal aging process. However, further research is needed to fully understand the relationship between EKODE-Cys adducts and aging, as other factors, such as diet and relative rates of protein turnover, may also affect their levels.

When we investigated other tissues, we found that adducts accumulated heavily in the liver and heart, whereas no traceable levels were detected in the intestine or spleen. Since liver is the primary organ for detoxification, many studies have been published using immunochemical approaches to detect hepatic protein adducts and use them as biomarkers for oxidative stress in liver diseases such as alcoholic liver diseases (ALD) (36, 37) (Fig. 4C). A particularly striking result we observed is shown in Fig. 4D and illustrates the time-dependent accumulation of EKODE-Cys adducts in a model of cardiovascular disease, specifically a series of heart perfusion experiments to model cardiac ischemia (38). These results suggest a strong association between EKODE-modified proteins and heart ischemia. It is likely that the accumulation of EKODE-Cys adducts during ischemia may result from elevated ROS production, leading to EKODE-modified cardiac proteins. Additionally, the reduced disposal rate of LPO products may cause the accumulation of EKODE-Cys adduct (39). One possibility could be the depletion of glutathione (GSH) caused by ischemia (40, 41), and accumulation of oxidized glutathione (GSSG), which in turn reduces the activity of glutathione S-transferase (GST), the enzyme responsible for the detoxification of LPO products (42, 43). We have previously shown that GSH plays a key role in scavenging ROS and forming covalent conjugates with OCPs such as HNE (44). Although the GSH adduct of EKODE has not yet been reported, it is highly plausible that GSH could react with EKODE via a similar mechanism. Regardless, based on the results presented here, we provide a compelling argument to further study the origin and fates of the EKODE LPO products.

Materials and Methods

Kinetic Studies of the Reaction of EKODEs and EKODE Model Compounds with N-Acetyl-Cysteine-Methyl Ester.

10 μM EKODEs/EKODE model compounds and 100 μM N-acetyl-cysteine-methyl ester were dissolved in PBS containing 20% acetonitrile. UV quartz cuvettes with 10 mm path length were filled with the reaction mixture and capped. UV analysis was conducted on a Lambda 25 UV/VIS Spectrophotometer (PerkinElmer, Waltham, MA) equipped with a cell changer and Peltier temperature controller. The spectrometer was programmed to maintain the reactions at 37 °C and scan the UV absorbance from 250 nm to 210 nm every 30 min at a scan speed of 480 nm/min. The absorbance at 234 nm was used to monitor the reaction’s disruption and recovery of the α, β-unsaturated carbonyl.

Mass Spectrometric Characterization of the Modification of HSA by trans-EKODE-(E)-IIb.

10 μM HSA was incubated with various concentrations of trans-EKODE-(E)-IIb (100/500/1,000 μM) in PBS containing 20% acetonitrile overnight. 400 μL of prechilled ethanol/ethyl acetate (1:1) was added to the centrifuge tube containing 100 μL of the reaction mixture, and the tube was placed on ice for 30 min. After centrifugation for 15 min at 14,800×g, the supernatant was discarded, and the precipitate was washed with cold acetonitrile (3 × 400 μL). The protein pellet was reconstituted in 100 μL of 50 mM AMBIC, followed by the addition of 9 μL of trypsin solution (0.5 mg/mL in AMBIC). The samples were then incubated overnight at 37 °C and the digestion was quenched by adding 10 μL of 10% TFA (v/v).

LC–HRMS Analysis Using the IDA Method.

LC–HRMS analysis employed a HPLC system consisting of Shimadzu LC20AD pumps and a SIL-HTC autosampler (Columbia, MD). A Kinetex XB-C18 column (2.1 mm × 50 mm, 2.6 μm, 100 Å, Phenomenex, Torrance, CA) was used for LC separation at a flow rate of 0.4 mL/min. Mobile phase A was 0.2% formic acid (FA) in water, and mobile phase B was 0.2% FA in acetonitrile. The needle was rinsed with a solvent consisting of 0.1% TFA in 50% acetonitrile in water (v/v/v). The multistep gradient was as follows: 0 to 1 min (5% B), 1 to 10 min (5% to 90% B), 10 to 12 min (90% B), 12 to 12.5 min (90% to 5% B), and 15 min (5% B). 5 μL of the reaction mixture from each time point was resolved by LC prior to MS analysis on an API 5600 TripleTOF mass spectrometer (AB Sciex, Foster City, CA) by the IDA method. The ion source parameters in positive turbo ion spray mode were as follows: curtain gas 30 psi, GAS1 45 psi, GAS2 45 psi, ion spray voltage 5,500 V, and source temperature 500 °C. The TOF scan of peptide precursors was conducted at an MS range of m/z 100 and 1,600, and the product ion spectrum acquisition was triggered for precursors with an intensity above 150 cps. The TripleTOF instrument was autotuned daily using a tuning solution (AB Sciex) to ensure MS accuracy. APCI positive calibration standard (AB Sciex) was delivered at a speed of 500 μL/min for 2 min by the calibration delivery system (CDS) every five sample injections for mass calibration. The acquired data were processed using BioPharmaView software (AB Sciex) for peptide mapping. Variable modifications of C₁₈H₃₀O₄ (+310.2144 Da) on Cys, His were included for peptide identification. The matching tolerance for peptide precursor and product ion was set at 10 ppm and 0.10 Da.

Synthesis of Antigens Against EKODE-Cysteine Adduct.

Preparation of EKODE-Cys adducts.

140 mg (0.452 mmol) of trans-EKODE-(E)-Ib or trans-EKODE-(E)-IIb was dissolved with 1 mL of 20% acetonitrile in PBS buffer (pH 7.4). N-acetyl-L-cysteine methyl ester (85.6 mg, 0.484 mmol) was then added, and the reaction was stirred at RT overnight, at which time the reaction mixture was extracted with 5 mL of ethyl acetate three times. The organic layer was combined, dried over sodium sulfate, and filtered. The filtrate was concentrated and briefly purified by silica gel chromatography using hexanes/ethyl acetate (1:1) to give a mixture of ring-opening product and Michael adduct as a transparent oil (163 mg, 74.1% yield). The mixture of EKODE-Cys adducts was characterized by LC–MS, which showed a peak at m/z 488.20 ([M + H]⁺) and 510.15 ([M + Na]⁺).

Preparation of EKODE-Cys-NHS ester.

163 mg (0.335 mmol) of EKODE-Cys adducts were dissolved with 4 mL of dimethyl sulfoxide (DMSO) in a round-bottom flask, followed by the addition of EDC (642 mg, 3.34 mmol) and NHS (963 mg, 8.37 mmol). After stirring at RT overnight, 25 mL of water was added to the reaction and extracted with 25 mL of ethyl acetate three times. The combined organic layer was dried on sodium sulfate and filtered. After removal of the solvent by rotavap, column chromatography with hexanes/ethyl acetate/methanol (5:5:1) yielded EKODE-Cys-NHS ester as a transparent oil (143 mg, 74.2% yield). The product was analyzed by LC–MS, and m/z was found at 607.15 ([M + Na]⁺).

Preparation of EKODE-Cys-spacer.

EKODE-Cys-NHS ester (143 mg, 0.245 mmol) was dissolved in 1.5 mL DMSO. A solution of aminocaproic acid (321 mg, 2.45 mmol) in 3.5 mL PBS was added. The reaction mixture was stirred vigorously at RT for 4 h, then 5 mL of water was added and extracted with ethyl acetate (5 × 25 mL). The combined organic layer was dried over Na₂SO₄, and the solvent was removed in vacuo. The residual oil was purified by column chromatography over silica (hexanes/ethyl acetate/methanol, 5:5:1) to give EKODE-Cys-spacer as a transparent oil (62.6 mg, 42.6% yield). The [M + H]⁺ and [M + Na]⁺ of the product were found at m/z 601.30, and m/z 623.30.

Preparation of EKODE-Cys-spacer-NHS ester.

The same procedure as for the synthesis of EKODE-Cys-NHS ester was used. 31 mg (51.6 μmol) of EKODE-Cys-spacer was dissolved in 1 mL of DMSO, followed by addition of EDC (90 mg, 516 μmol) and NHS (148 mg, 1.29 mmol). The reaction mixture was stirred at RT overnight, and then 5 mL of water was added and extracted with ethyl acetate (5 × 25 mL). The dried organic layer was concentrated and purified by column chromatography to yield EKODE-Cys-spacer-NHS ester as a transparent oil (22 mg, 61.0% yield). The product was subjected to LC–MS analysis, and [M + H]⁺ and the [M + Na]⁺ of the product were found at m/z 698.35, and m/z 720.35.

Preparation of EKODE-Cys-Spacer-Protein Conjugates.

200 μL of EKODE-Cys-spacer-NHS ester solution in DMSO (10 mg/mL, 14.3 mM) was added to 466 μL of protein solution (BSA or KLH 4.35 mg/mL in PBS). After stirring at RT for 4 h, the reaction mixture was dialyzed in dialysis cassettes (10 K molecular weight cut off) against 2 L of 30% DMSO in PBS with gentle stirring at 4 °C. The buffer was changed to 2 L of 20% DMSO in PBS after 2 h of dialysis and subjected to 4 h of additional dialysis. The reaction mixture was then dialyzed in PBS overnight.

Characterization of EKODE-Cys-spacer-BSA conjugates by MALDI-TOF.

The MALDI matrix solution was prepared by dissolving 1 tube of sinapinic acid (1 mg) in 100 μL of 30% acetonitrile in water with 0.1% trifluoroacetic acid (TFA). 1 μL of EKODE-Cys-spacer-BSA conjugate was applied onto a stainless steel MALDI sample block, followed by the addition of 1 μL of MALDI matrix solution and mixed by pipette tips. The samples were dried at RT and subjected to MALDI-MS analysis, which was performed using a Bruker BiFlex III MALDI-TOF equipped with a pulsed nitrogen laser source (λ = 337 nm) and used at an acceleration voltage of 28 kV operated in the reflection mode (Voyager Biospectrometry Workstation, PerSeptive Biosystems, Framingham, MA). The number of modified lysine residues was determined based on the addition of molecular weight due to the conjugation of EKODE-Cys-spacer, and the extent of lysine modification was calculated based on the fact that BSA contains 59 lysine residues.

Production and Characterization of Anti-EKODE-Cys Antibodies.

Immunization and ELISA analysis of antibody titers.

The EKODE-Cys-spacer-KLH conjugates were sent to Cocalico Biologicals (Stevens, PA) for antiserum production. The rabbits were immunized on days 14, 21, 49, and 77. Test bleeds were collected on days 14, 35, 56, 70, and 84 from the initial inoculation, and exsanguination was performed on day 91 for production bleed. To determine anti-EKODE-Cys antibody levels in rabbit blood serum, the corresponding EKODE-Cys-spacer-BSA conjugates were used as coating agents. 50 μL of BSA conjugate solution (20 μg/mL in PBS) was added to each well of sterilized Baxter ELISA plate and incubated at 4 °C overnight. The coating solution was discarded, and the plate was washed three times using 300 μL of wash buffer (50 mM pH 7.4 PBS with 0.05% v/v Tween-20). The remaining active sites on the plate were blocked by adding 300 μL of 3% (w/v) BSA in PBS to each well and followed by incubation at 37 °C for 2 h. The plate was then washed with washing buffer (3 × 300 μL), followed by the addition of 50 μL of rabbit serum from each test bleed or prebleed (before injection of antigen as negative response control), which were prediluted 1:10,000 with 1% BSA in PBS (w/v). 1% BSA in PBS without serum was used as a blank, and each sample was analyzed in duplicate. The plate was then incubated at 37 °C for 2 h and washed with washing buffer (3 × 300 μL). The stock solution of HRP conjugated goat anti-rabbit IgG secondary antibody (0.4 mg/mL in water/glycerol, 1:1) was diluted 1:5,000 with PBS, and 100 μL of the diluted solution was added to each well. The plate was incubated at RT for 1 h and washed three times. 1-step ABTS substrate solution (150 μL) was added to each well to reveal the immunocomplex bound to the plate. After incubating at RT for 30 min, the reaction was quenched by 100 μL of 1% SDS solution in water, and the absorbance of the samples was measured at 405 nm by Bio-Rad 450 Microplate reader (Hercules, CA).

Purification of antibody via immunoaffinity capture.

To obtain the specific anti-EKODE-Cys antibodies from the crude antiserum, EKODE-BSA conjugates were prepared by mixing 100 μL of 0.5 mM trans-EKODE-(E)-Ib or trans-EKODE-(E)-IIb solution in ethanol with 400 μL of 1 mg/mL BSA solution in PBS, followed by incubation at 37 °C overnight. The resulting EKODE-BSA conjugates were immobilized on the MicroLink protein coupling column using the following procedures: the columns were centrifuged at 1,000×g for 2 min to remove the storing buffer. Then, 300 μL of coupling buffer (0.1 M sodium phosphate, 0.15 M NaCl, pH 7.2) was added to suspend the resin, followed by centrifugation at 1,000×g for 1 min to remove the buffer. This step was repeated two more times to equilibrate the column. Subsequently, 150 μL of the EKODE-BSA conjugates and 150 μL of the coupling buffer were added to the column and mixed by gentle swirling. 2 μL of sodium cyanoborohydride solution (5 M in 0.01 M NaOH) was then added, and the column was incubated overnight at 4 °C with gentle end-over-end mixing. After washing the column with coupling buffer (3 × 300 μL), the reaction was stopped by adding 2 × 300 μL of the quenching buffer (1 M Tris, 0.05% NaN₃, pH 7.4). 200 μL of the quenching buffer and 4 μL of sodium cyanoborohydride solution were then applied to the column and incubated at RT for 30 min to block the active binding sites on the resin. After washing the resin containing immobilized EKODE-BSA conjugates three times with 300 μL of washing buffer (1 M NaCl, 0.05% NaN₃), 300 μL of the antiserum was added to the column and incubated overnight at 4 °C with gentle end-over-end mixing. The resin-bound immunocomplex was washed three times with 300 μL of washing buffer containing 0.05% Tween-20, followed by washing with a coupling buffer (3 × 300 μL). 100 μL of elution buffer (pH 2.8, contains primary amine) was then slowly added and incubated at RT for 10 min. The dissociated antibodies were collected by centrifugation, and the resulting solution was neutralized immediately by adding 5 μL of 1 M Tris, pH 9.0.

Characterization of anti-EKODE-Cys antibodies by competitive ELISA.

The competitor EKODE-His adducts were synthesized by mixing trans-EKODE-(E)-Ib or trans-EKODE-(E)-IIb (10 mg, 32.3 μmmol) with N-acetyl-L-histidine methyl ester (34 mg, 161 mmol) overnight at RT in 1 mL of 20% acetonitrile in PBS. HNE-Cys adducts were obtained via the reaction of HNE (20 mg, 128 mmol) with N-acetyl-L-cysteine methyl ester (45 mg, 254 mmol) in 1 mL of chloroform with overnight stirring at RT. Both negative competitors and EKODE-Cys adducts were subjected to eight serial dilutions with PBS. 50 μL of the diluted competitors and positive control (PBS with no competitor) were incubated with 50 μL of purified antiserum (1:1,000 diluted with 1% BSA in PBS) overnight at 4 °C. The resulting antibody–competitor mixture was analyzed by ELISA under the same conditions used to measure antibody titers. The measured absorbance values for duplicate samples were averaged, and the relative values to a positive control (B/B₀) were plotted against the log concentration.

Characterization of anti-EKODE-Cys antibodies by western blot.

To evaluate the cross-reactivity of anti-EKODE-Cys antibodies with the protein adducts of other LPO products, the BSA adducts were prepared by incubating 90 μL of 1 mg/mL BSA solution in PBS with 10 μL of 1 mM or 100 μM LPO products (EKODEs, HNE, ONE, acrolein, octenal) in ethanol at 37 °C for 24 h. 1 μL of modified BSA was mixed with 9 μL of water and 10 μL of 2× loading buffer containing 2-mercaptoethanol, and 8 μL of the mixture was separated by 15% SDS-PAGE at 100 V (gel was prepared according to standard protocols). The resolved proteins were then electrotransferred onto Immobilon PVDF membrane at 30 V overnight at 4 °C and incubated with a blocking agent (10% nonfat milk in 0.1% Tween-20 in Tris-buffered saline (TBST)) with gentle shaking for 1 h at RT. After washing with TBST for 3 × 5 min, the anti-EKODE-Cys antibodies (diluted 1:1,000 in 1% milk in TBST) were then applied to the membrane and incubated overnight at 4 °C. The blots were then rewashed with TBST for 3 × 10 min, followed by the addition of a secondary antibody (HRP-linked goat anti-rabbit IgG, 1:10,000 diluted in TBST). After incubating at RT for 1.5 h and washing for 3 × 10 min, the blots were developed by ECL detection reagent for 5 min and visualized by exposure to X-ray film.

Immunochemical Detection of Endogenous EKODE-Cys Adducts in Biological Samples.

Immunochemical detection in human neuroblastoma cells.

Human neuroblastoma cell line M17 was maintained in serum-free Opti-MEM media (Invitrogen, Gaithersburg, MD) with 5% donor calf serum and 1% penicillin/streptomycin. A H₂O₂ solution of 30% (w/w) in water was added to a final concentration of 0.25 mM at 80% - 90% cell confluence. The cells were collected after 24 h of H₂O₂ treatment, and cell lysis was performed in lysis buffer with protease inhibitor phenylmethanesulfonyl fluoride (PMSF) and phosphatase inhibitor NaF. The protein concentration was determined using a BCA protein assay, and 20 μg of protein was resolved in duplicate by SDS-PAGE and analyzed by western blot using anti-EKODE-Cys antibodies. The membrane was then incubated in stripping buffer at RT for 30 min to remove the primary and secondary antibodies. After washing with TNBST for 3 × 10 min, the blots were probed by β-actin as an internal control. Quantitation of the blots was conducted by a computer-assisted scanning system (Quantity One 4.3, Bio-Rad).

Immunochemical detection in human brain tissues.

Frozen frontal cortex samples were homogenized in lysis buffer containing protease inhibitor cocktail. After determining the protein concentration with BCA assay, 30 μg of protein was resolved using SDS-PAGE. Proteins were then transferred to the Immobilon PVDF membrane for western blot analysis using anti-EKODE-Cys as the primary antibody and β-actin as an internal control. The quantitative result generated from Quantity One was plotted against the age of the subjects. To assess whether the western blot results can reflect the level of EKODE adducts, 20 μL of 1 mg/mL homogenate of human brain tissue was incubated with 10 μM or 100 μM of trans-EKODE-(E)-Ib overnight at 37 °C and analyzed by western blot using an anti-EKODE-Ib-Cys antibody.

Immunochemical detection in various mouse tissues.

C57BL/6 mouse (2-mo-old) was anesthetized with avertin, and organs were quickly excised. All experiments were approved by the Institutional Animal Care and Use Committee at Case Western Reserve University. The tissues were homogenized in lysis buffer containing a protease inhibitors cocktail. After determining protein concentration with BCA assay, 30 μg of protein was resolved using SDS-PAGE and subjected to immunoblot using anti-EKODE-Cys antibodies.

Immunochemical detection in rat heart ischemia.

Adult male Sprague-Dawley rats (300 to 350 g) were fed ad libitum for 8 to 12 d with standard laboratory chow before experiments. The heart perfusion experiments were performed as previously described (38). The control group was a 15-min perfusion with no ischemia. The different ischemic perfusion groups were based on 15-min perfusion to allow for baseline equilibration, followed by various times (5, 15, and 30 min) of global no-flow ischemia. Each group consisted of three rats (n = 3 × 4), and hearts were quick-frozen at the end of the protocol for each group. Powdered frozen organs were homogenized in the lysis buffer with protease inhibitors. Protein concentration was measured by BCA assay, and 30 μg of protein was separated using SDS-PAGE. Proteins were then transferred to the Immobilon PVDF membrane for western blot analysis using anti-EKODE-Cys as the primary antibody and β-actin as an internal control. Quantitation of the blots was performed by Quantity One.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.