Cholesteryl ester acyl oxidation and remodeling in murine macrophages: formation of oxidized phosphatidylcholine.

Cholesterol is an essential component of eukaryotic cell membranes, regulating fluidity and permeability of the bilayer. Outside the membrane, cholesterol is esterified to fatty acids forming cholesterol esters (CEs). Metabolism of CEs is characterized by recurrent hydrolysis and esterification as part of the CE cycle; however, since recombinant 15-lipoxygenase (15-LO) was shown to oxidize cholesteryl linoleate of LDL, there has been interest in CE oxidation, particularly in the context atherogenesis. Studies of oxidized CE (oxCE) metabolism have focused on hydrolysis and subsequent reverse cholesterol transport with little emphasis on the fate the newly released oxidized fatty acyl component. Here, using mass spectrometry to analyze lipid oxidation products, CE metabolism in murine peritoneal macrophages was investigated. Ex vivo macrophage incubations revealed that cellular 15-LO directly oxidized multiple CE substrates from intracellular stores and from extracellular sources. Freshly harvested murine macrophages also contained 15-LO-specific oxCEs, suggesting the enzyme may act as a CE-oxidase in vivo. The metabolic fate of oxCEs, particularly the hydrolysis and remodeling of oxidized fatty acyl chains, was also examined in the macrophage. Metabolism of deuterated CE resulted in the genesis of deuterated, oxidized phosphatidylcholine (oxPC). Further experiments revealed these oxPC species were formed chiefly from the hydrolysis of oxidized CE and subsequent reacylation of the oxidized acyl components into PC.


Preparation of lipid vesicles
Synthetic lipid vesicles (LVs) were formed in aqueous buffer essentially by the method of Chapman and Trelease ( 20 ). Briefl y, a mixture of CE(d 4 -18:2), palmitoyl-oleoyl-phosphocholine, and cholesterol (0.60, 0.15, and 0.10 mol, respectively) was dried under a stream of N 2 and diluted in ether (0.5 ml). This lipid composition corresponded to a neutral lipid to amphipathic lipid ratio similar to that of VLDL. Calcium-and magnesium-free (Ca/Mg-free) HBSS (1.6 ml) was added, and the biphasic system was sonicated under a stream N 2 . When the ether evaporated completely, the aqueous phase became uniformly turbid, indicating a stable emulsion of lipid vesicles.

Human lipoprotein
Whole blood was obtained from healthy volunteers with approval from the Colorado Multi-institutional Review Board. Platelet-poor plasma was prepared from whole blood as previously described ( 21 ) and centrifuged at 45,000 rpm for 10 min at 18°C. The top layer of chylomicrons was discarded, and the remaining chylomicron-free, platelet-poor plasma was used immediately. This material is hereafter referred to as human lipoprotein (LP).

Resident peritoneal macrophage incubations
All research involving animals was done in accordance with Public Health Service policy and was approved by the Colorado Multi-institutional Review Board. Murine peritoneal macrophages were isolated essentially as previously described ( 22 ). Briefl y, 5 ml of Ca/Mg-free HBSS was injected into the peritoneal cavity of euthanized mice; the cavity was massaged and gently agitated before the lavage fl uid was recovered. Macrophages in peritoneal lavage fl uid were counted and diluted with a convenient amount (no less than 0.5 vols) of calcium-and magnesium-containing HBSS. Cells were then aliquoted at 5 × 10 5 cells/well into 24-well plates, which were incubated at 37°C for 1 h. Nonadherent cells were removed by washing once with HBSS. To the adherent cells, 0.5 ml of HEPES buffered HBSS (10 mM; pH 7.2) was added along with 25 l of synthetic lipid vesicles (4.7 mol/l CE (d 4 -18:2) fi nal conc.), 1 l of human LP (approximately 3 mol/l CE fi nal conc [ 1 ]), or d 4 -18:2 (4.7 M fi nal conc.). Incubations with d 4 -18:2 included fatty acid-free BSA (0.5% w/w) in the medium (MP Biomedicals, Solon, OH). After incubation at 37°C for various times, the medium was removed, and the cells were washed with 500 l of medium. Adherent cells were then scraped twice with 500 l of 90% MeOH in water containing SnCl 2 (2 mM). Samples remained at room temperature for 1.5 h to allow complete reduction of hydroperoxide species before lipids were extracted. This step halted enzymatic activity by denaturing proteins and prevented further propagation of peroxy-radical oxygenation. Time courses were performed twice, each time using pooled macrophages from three mice and duplicate time points. When human LP was used, the two time courses were performed with material derived from the blood of separate donors.

Lipid extractions
Neutral lipids were extracted as previously described using 75:25 (v/v) isooctane:ethyl acetate ( 17 ). Phospholipids were extracted with a modifi ed Bligh and Dyer procedure replacing chloroform with dichloromethane ( 19 ). Free fatty acids and oxylipins peroxidation was observed as the dominant lipid modification suggesting direct CE oxidation by intracellular . However, chiral CE oxidation products were not shown, and therefore oxidation of CEs by radical mechanisms could not be ruled out. In fact, nonenzymatic LDL oxidation results in a similar profi le of lipid oxidation products, with oxCEs being the dominant species. Furthermore, it is not fully understood how the formation of extracellular oxidized LDL could proceed by direct oxidation of CE by intercellular 15-LO. Thus, it remains unclear whether 15-LO, in the intact cell, plays an important role in direct CE oxidation Previous studies demonstrated that oxCEs were substrates for hydrolysis, even preferred substrates over their nonoxidized counterparts ( 11 ). These investigations went on to show that CE oxidation enhances free cholesterol effl ux ( 12 ); however, the fate of the newly freed oxylipins was not examined. In separate studies, where particularly high levels of oxidized CEs were formed during coincubations of LDL with 15-LO-expressing murine fi broblasts, oxidized PL (oxPL) species were observed as minor products ( 8 ). This fi nding suggested that radical propagation had occurred or that 15-LO was directly oxidizing PLs ( 13 ). An alternate mechanism of oxPL formation, which was not examined, was further metabolism of the abundant oxCE species leading to the esterifi cation of oxidized acyl chains in the PL pool. These separate observations of oxCE hydrolysis and the genesis oxPL in conditions of high oxCE suggest that some proportion of oxPL might originate from the hydrolysis and remodeling of abundant oxCEs. Oxidized PLs, particularly oxidized phosphatidylcholine (oxPC), are ligands for macrophage scavenger receptors ( 14 ) implicated in vascular plaque progression ( 15 ) where oxCEs are often abundant ( 16,17 ).
Using isotopically labeled substrate coincubated with murine peritoneal macrophages, we determined whether cellular 15-LO was capable of directly oxidizing CE within endogenous lipid bodies as well as deuterated CE originating from the culture medium. By LC-MS/MS we examined the resulting oxCE profi le for evidence of 15-LO-specifi c oxidation products of isotopically labeled and unlabeled CEs. To determine if 15-LO acts as a CE oxidase in vivo, CE oxidation products were examined in freshly harvested murine macrophages. To investigate biosynthetic coupling of CE oxidation to oxPL synthesis, we examined PLs isolated from murine macrophages after incubation with isotopically labeled CE substrate. To determine if this pathway was specifi c to oxCE remodeling, we compared oxPL products derived from CE oxidation and remodeling to oxPL generated by macrophage incubation with free d 4 -18:2.  ( 5 ). In contrast, auto-oxidation of CE(18:2) produced an array of nonspecifi c oxidation products lacking regio-and stereospecifi city. Therefore, detailed characterization of CE oxidation products (oxCE) provides insight into mechanisms of CE oxygenation. To determine if macrophage 15-LO oxygenated CE delivered in extracellular LVs, CE(d 4 -18:2)-laden LVs were incubated with murine peritoneal macrophages for 45, 90, 180, or 1200 min. After incubation, the medium was removed, and the cells were washed before intracellular lipids were extracted and analyzed by NP LC-MS/MS.

Materials
After shorter incubation times, the array of d 4 -oxCEs was indicative of 15-LO-driven oxygenation and auto-oxidation. The presence of numerous, deuterated oxidation products (all previously characterized [ 17 ]) indicated some nonspecifi c auto-oxidation occurred; however, the specifi c regioisomer, CE(d 4 -13-ZE-HODE), was the most abundant product, suggesting involvement of 15-LO in CE(d 4 -18:2) oxidation ( Fig. 1A ). When this compound was isolated and rechromatographed on a chiral column, a clear excess of the S enantiomer was observed, confi rming oxygenation of the CE(d 4 -18:2) substrate by 15-LO to CE(d 4 -13( S )-ZE-HODE) ( Fig. 1A, inset). When the incubation medium supernatant was analyzed separately (data not shown), only auto-oxidation products were detected, suggesting that 15-LO activity was confi ned to the macrophage cytoplasm. As incubation times increased, the preponderance of CE(d 4 -13( S )-ZE-HODE) was diminished relative to other oxidation products. After 20 h of substrate incubation, total oxidation products had clearly increased, but there was no longer an excess of the chiral 15-LO-specifi c product ( Supplementary Fig. IA ), suggesting nonspecifi c oxidation had overwhelmed the specifi c product profi le. The inclusion of certain antioxidants was found to attenuate this auto-oxidation, but 15-LO-specifi c products were still readily formed ( Supplementary Fig. II ).

Oxidation of human lipoprotein CEs by macrophage 15-LO
To determine if CE bound in LP particles could be oxidatively modifi ed by murine macrophage 15-LO, human LP was coincubated with adherent macrophages for 45, 90, 180, or 1200 min. Neutral lipids were then extracted and analyzed by NP LC-MS/MS. The near absence of nonspecifi c CE oxidation products at 45 min ( Fig. 2A ), and even after 20 h (Supplemental Fig. 1B ), indicated that very were isolated by solid phase extraction (C 18 ) as previously described ( 23 ).

Chromatography
Separation of CEs and oxCEs by normal phase (NP) LC-MS/MS as well as chiral separations of hydroxylated-CEs were performed as previously reported ( 17 ). Free fatty acids and oxylipins (e.g., HETEs and HODEs) were separated using reversed-phase LC-MS/MS (RP LC-MS/MS) as previously reported ( 23 ).
Phospholipids (PLs) were separated by two methods. To isolate PLs according to headgroup, NP LC was performed as previously described ( 24 ). To separate PLs species by acyl composition, a RP LC-MS/MS was used. Lipid extracts were loaded onto a 150 × 3 mm, 2.6 m, Kinetex C 18 column (Phenomenex, Torrance, CA), and PLs were eluted with an isocratic fl ow of 5:3:1:

Mass spectrometry
Online MS/MS of CEs was performed essentially as previously described using an AB Sciex API 3200 triple quadrupole mass spectrometer (AB Sciex, Toronto, Canada) ( 17 ). Briefl y, ammonium acetate ( Online MS/MS of phospholipids was performed on an AB Sciex 5500 triple quadrupole linear ion trap hybrid mass spectrometer operating in negative ion mode essentially as previously described ( 24 ). All species were detected using MRM transitions for deprotonated molecular ions decomposing to specifi c product carboxylate anions with the exception of phosphatidylcholine species, where the parent ions were molecular acetate adducts [M+OAc] Ϫ . A comprehensive table of parent and product ions monitored is presented in Supplementary Table I. In exploratory experiments (chiefl y NP LC), all transitions were included in the MRM method; in subsequent experiments (chiefl y RP LC), the number of transitions included was reduced (only monitoring palmitate-and oleate-containing PC species) to allow for greater sensitivity. To accommodate over 400 MRM transitions while maintaining a functional sampling rate (i.e., duty cycle <6 s), a "scheduled MRM" approach was used during certain experiments. This involved monitoring selected transitions during a discrete retention time window corresponding to the known elution behavior of specifi c PL analytes, thus limiting the total MRM transitions monitored during a given portion of the chromatogram.
Collisionally induced dissociation spectra were acquired online during HPLC by including an enhanced product ion scan in the duty of the mass spectrometer. Ion source and collisionally induced dissociation parameters were the same as those used for MRM analyses. Product ion spectra were acquired from m / z 250 to 870 scanning at 1,000 Da/s. MS 3 spectra were acquired using an activation energy of 0.1 V, scanning products ions from m / z 150 to 300 at 1,000 Da/s.

Freshly harvested peritoneal macrophages contain 15-LO-specifi c oxCEs
To determine if murine resident peritoneal macrophages contained 15-LO-specifi c oxCEs in vivo, experiments were carried out to minimize macrophage activation and quench enzymatic activity upon harvesting. Lavage fl uid (Ca/Mg-free HBSS) collected from wild-type mice ( n = 4) was mixed with one volume of ice-cold MeOH containing SnCl 2 (2 mM) immediately after collection from the peritoneal cavity. Lipids were then extracted and analyzed by NP LC-MS/MS. The CE profi le was dominated by the polyunsaturated species CE(18:2) and CE(20:4). 15-LO-specifi c oxidation products of CE(18:2) and CE (20:4) were consistently present at low levels ( Fig. 4 ). When these products were isolated and rechromatographed on a chiral column, they showed high enantiomeric excess of the S confi guration, confi rming enzymatic oxygenation ( Fig. 4, insets). These results indicated that little auto-oxidation of CEs had occurred. This lack autooxidation was consistent with previous observations that plasma is an effective antioxidant ( 25 ). On the other hand, enzymatic oxidation products were readily apparent. Both CE(18:2) and CE(20:4) were specifi cally oxygenated by 15-LO forming CE(13( S )-HODE) and CE(12( S )-HETE), respectively ( Fig. 2A ). Oxygenation of these two polyunsaturated CEs by 15-LO was rather robust. Using previously generated calibration curves ( 17 ) to convert product/substrate peak area ratios to product/substrate relative abundance, maximal conversion of CE(18:2) to CE(13( S )-HODE) was determined to be 3.1 ± 1.3%, and conversion of CE(20:4) to CE(12( S )-HETE) was 9.6 ± 3.9% (mean ± SEM; n = 4; 20 h incubation).
To assess CE uptake, we compared the peak areas of CEs and oxCEs from macrophages exposed to human LP with control macrophages exposed to synthetic LVs containing CE(d 4 -18:2). Macrophages exposed to human LP contained orders of magnitude more CE(18:2) and CE(20:4) ( Fig. 2B ). Although de novo CE synthesis cannot be ruled out, the remarkable increase in CE content was most likely due to uptake or perhaps to adhesion of human LP. The 15-LO-specifi c products CE(13( S )-HODE) and CE(12 ( S )-HETE) were also markedly increased (approximately 100-fold over control cells; Fig. 2C ), strongly suggesting that a portion of the CE derived from the human LP had gained access to macrophage cytoplasm where it was oxygenated by 15-LO.
To determine unambiguously if 15-LO was responsible for oxygenation of CE(18:2) and CE(20:4), human LP was  ( Fig. 5A ). Incorporation of d 4 -HODE into other PL classes (e.g., PE, PS, PI, or PG) was not detected. To further characterize these phospholipid metabolites of CEs, MS 2 and MS 3 spectra were acquired as d 4 -HODE-containing PCs eluted from the HPLC ( Fig. 5C,D ). The resulting product ion spectra were consistent with the structures proposed in Fig. 5B . Specifi cally, MS 2 product ion spectra of PC acetate adducts, [M+OAc] Ϫ , contained the expected carboxylate anions, including the ions at m / z 299 corresponding to d 4

Remodeling of oxCE is a robust pathway of oxidized phosphocholine formation
The formation of d 4 -oxPC from CE(d 4 -18:2) substrate suggested that oxidation and remodeling of CEs might form oxPC in macrophages. However, the complex network of cellular acyl remodeling pathways could provide small amounts of 15-LO-specifi c CE oxidation products were present in freshly harvested murine peritoneal lavages.

Acyl-remodeling of oxidized CEs
Studies of further metabolism, specifi cally hydrolysis and reacylation of isotopically labeled CE-derived fatty acyl groups, were carried out by incubating adherent macrophages with CE(d 4 -18:2) LVs for 20 h. To assess hydrolysis of d 4 -oxCEs, the medium and the cells were collected together and free fatty acids were isolated by solid-phase extraction before analysis by RP LC-MS/MS. These analyses revealed the presence of free d 4 -9-HODE and d 4 -13-HODE ( Supplementary Fig. IV ), suggesting that oxidized CE(d 4 -18:2) was hydrolytically cleaved during incubation with macrophages, a fi nding consistent with previous studies ( 11 ). Alternatively, CE(d 4 -18:2) may have been hydrolyzed before oxygenation. In separate experiments, to determine if oxidized acyl chains could be esterifi ed into macrophage PLs, the incorporation of [1-14 C]13-HpODE was determined by TLC. These experiments, detailed in the supplementary materials, indicated this substrate was indeed incorporated into macrophage PLs, mainly PC ( Supplementary Fig. V ). These fi ndings were consistent with two previous studies demonstrating preferential incorporation of 13-HODE into PC of epithelial cells ( 26,27 ).

Direct oxidation of multiple CE substrates by macrophage 15-LO
Previous studies demonstrating direct oxygenation of CE(18:2) were performed using recombinant 15-LO and therefore did not address enzymatic oxidation of CE substrate in the intact cell ( 5,28,29 ). Separate studies, examining CE oxidation in intact cells (e.g., murine peritoneal macrophages, fi broblasts, and human monocytes), demonstrated that 15-LO expression increased oxidized CE content; however, direct evidence of enzymatic CE oxidation (i.e., chiral oxCEs) was not shown ( 9,10,30 ). Because 15-LO is capable of oxidizing substrates other than CE, and considering that hydroperoxy species can initiate radical propagation, the oxidation of CE observed may have been an indirect effect of the lipoxygenase. Furthermore, 15-LO is an intercellular enzyme, and therefore its interaction with substrate originating outside the cytoplasm is not straightforward. To determine whether CEs, both alternate routes of d 4 -oxPC formation, which did not involve remodeling of oxidized CE(d 4 -18:2), such as oxidation of preformed PC(16:0/d 4 -18:2) or oxidation of free d 4 -18:2 and subsequent esterifi cation into PC. To determine if d 4 -oxCE remodeling contributed signifi cantly to d 4 -oxPC formation or if the above-mentioned alternate routes were responsible for the observed d 4 -oxPC, murine peritoneal macrophages were incubated in parallel with LVs containing CE(d 4 -18:2) or an equivalent molar amount  lipid bodies were substrates for enzymatic oxidation. As previously reported using recombinant 15-LO, CE(13-S )-HODE) was the product of CE(18:2) oxygenation. Additionally, it was observed in the macrophage that CE(20:4) was oxidized, specifically forming CE(12-S )-HETE) in an even more efficient reaction than the oxidation of CE(18:2). Because the 15-LO null macrophages produced neither product, this single enzyme was likely responsible for the oxidation of both CE(18:2) and CE(20:4) ( Fig. 3 ). Although these results demonstrated that 15-LO has dual CE substrate specifi city, there was no evidence for the dual regiospecifi city normally observed during peroxidation of free fatty acids. For example, 15-LO exclusively formed endogenous and from the extracellular environment, could be directly modifi ed by 15-LO within intact cells, isotopically labeled CE(d 4 -18:2) LVs were incubated with murine peritoneal macrophages. This allowed metabolism of CE(d 4 -18:2) to be analyzed independent of endogenous CEs, which are present in lipid bodies within murine peritoneal macrophages. These experiments revealed that CE(d 4 -18:2) taken up from the medium was directly oxygenated by intracellular 15-LO as evidenced by the deuterated, chiral oxidation product, CE(d 4 -13( S )-HODE) ( Fig. 1A ).
Nondeuterated CEs were also oxidized by 15-LO during macrophage incubations, indicating that CEs in endogenous  ( Fig. 4 ) strongly suggests that 15-LO-catalyzed CE oxidation occurs in the animal and therefore may play a role in CE metabolism in vivo, but it remains to be determined whether this pathway is relevant to LDL oxidation and atherogenesis within the arterial wall.

Interactions of Acyl remodeling pathways generate oxPC from oxCE
Considering that cellular cholesterol is in dynamic equilibrium between free and esterifi ed forms and that there is evidence that oxCEs are included in this cycle ( 33,34 ), we hypothesized that remodeling of abundant oxCEs in intact macrophages could lead to the generation of other oxidized lipid species, particularly oxPLs. The detection of free d 4 -HODE isomers after macrophage incubation with CE(d 4 -18:2) supported the hypothesis that oxCEs are substrates for CE hydrolases. This reaction itself may have biological implications; indeed, the hydrolysis products of CE(12-HETE) from CE(20:4) (Supplementary Fig. IIIB ) despite the fact that this mouse enzyme produces 12-and 15-HETE from free arachidonate ( 31 ). This curious fi nding suggests that binding interactions of the CE acyl chain within the 15-LO catalytic pocket are remotely infl uenced by the presence of the lipophilic cholesterol moiety. Cholesteryl esters are highly insoluble in aqueous solutions and are almost exclusively found inside lipid bodies (or synthetic variants) or the membranes engaged in their synthesis. In order to oxidize CEs, 15-LO must associate with the amphipathic portion of these macromolecular structures, which may induce conformational changes in the enzyme enhancing regiospecificity. Alternatively, CE within lipid bodies may be presented to 15-LO in a different manner than the association of free fatty acids.
To examine whether oxidation of extracellular CE would extend to substrate bound in actual LP particles, adherent macrophages were incubated in medium containing human LP. Marked increases in cellular CE content after LP incubation indicated that the macrophages had likely taken up signifi cant amounts of LP, that LP had adhered to the outer surfaces of the macrophages, or both ( Fig. 2B ). Comparable increases in 15-LO-specifi c oxCEs, nearly 100-fold over control macrophages ( Fig. 2C ), suggested that some of the CE derived from human LP had accessed the intracellular compartment, where 15-LO directly oxidized the newly taken up CE substrate. These ox-CEs were readily formed without the need for potent experimental stimuli (e.g., macrophage stimulation), consistent with the activity of 15-LO, which requires minimal activation ( 32 ). Therefore, the concomitant increases in the level of available substrate and 15-LO-specifi c CE oxidation products suggested that 15-LO CE oxidase activity is partly regulated by simple substrate availability. These observations further implicate direct 15-LO catalyzed CE oxidation as an initiating event during in vitro LDL oxidation. However, because evidence of chiral products eroded over time, this phenomenon may be diffi cult to track in vivo using oxidized product stereochemistry. The occurrence of chiral oxCEs in freshly harvested murine peritoneal  both 15-LO-specifi c oxCEs, 13( S )-HODE and 12( S )-HETE, are ligands for peroxisome proliferator-activated receptor-␥ and the newly defi ned G protein-coupled 12( S )-HETE receptor, respectively ( 35,36 ). Although the hydrolysis of oxCEs has previously been reported, the fate of the newly freed oxidized acyl components has not been examined.
We speculate that oxidation and remodeling of cellular CE might effi ciently form oxPC for two reasons. First, CEs are particularly good targets for oxidation, as evidenced by high levels of oxCE in human atheromata and the ease of experimentally forming oxCEs reported here and elsewhere ( 17 ). Indeed, the primary CE species [i.e., CE(18:2) and CE(20:4)] are both substrates for 15-LO ( 38 ) and susceptible to free-radical mediated oxidation. Second, the required enzymatic machinery for CE conversion to oxPC appears to be localized to cytoplasmic lipid bodies. Both 15-LO and neutral cholesterol hydrolase associate with lipid bodies ( 38,39 ), and oxCEs have been shown to be preferred substrates for hydrolysis ( 11 ). Additionally, the newly described lysophosphatidylcholine acyltransferases 1 and 2 are also localized to lipid bodies where they catalyze acylation of lysophosphatidylcholine at the monolayer ( 40 ). The results described here demonstrate that remodeling of oxCE indeed contributes to PC oxidation and suggests that the presence of abundant oxCE may infl uence overall PC oxidation levels in the macrophage. Because oxCEs can accumulate signifi cantly under certain circumstances (e.g., in the atheromatous plaque), their degradation and remodeling into biologically active compounds such as oxPC, even to a partial extent, may have biological implications. Indeed, this work suggests that a portion of the atherogenic oxPC, which has been the focus of considerable attention ( 15 ), might originate from 15-LO-initiated oxidation of susceptible CEs and subsequent metabolism of these unique species.