Localization of nitration and chlorination sites on apolipoprotein A-I catalyzed by myeloperoxidase in human atheroma and associated oxidative impairment in ABCA1-dependent cholesterol efflux from macrophages.

We recently reported that apolipoprotein A-I (apoA-I), the major protein component of high density lipoprotein, is a selective target for myeloperoxidase (MPO)-catalyzed nitration and chlorination in both and serum of subjects with cardiovascular disease. We further showed that the extent of both apoA-I nitration and chlorination correlated with functional impairment in reverse cholesterol transport activity of the isolated lipoprotein. Herein we used tandem mass spectrometry to map the sites of MPO-mediated apoA-I nitration and chlorination in vitro and in vivo and to relate the degree of site-specific modifications to loss of apoA-I lipid binding and cholesterol efflux functions. Of the seven tyrosine residues in apoA-I, Tyr-192, Tyr-166, Tyr-236, and Tyr-29 were nitrated and chlorinated in MPO-mediated reactions. Site-specific liquid chromatography-mass spectrometry quantitative analyses demonstrated that the favored modification site following exposure to MPO-generated oxidants is Tyr-192. MPO-dependent nitration and chlorination both proceed with Tyr-166 as a secondary site and with Tyr-236 and Tyr-29 modified only minimally. Parallel functional studies demonstrated dose-dependent losses of ABCA1-dependent cholesterol acceptor and lipid binding activities with apoA-I modification by MPO. Finally tandem mass spectrometry analyses showed that apoA-I in human atherosclerotic tissue is nitrated at the MPO-preferred sites, Tyr-192 and Tyr-166. The present studies suggest that site-specific modifications of apoA-I by MPO are associated with impaired lipid binding and ABCA1-dependent cholesterol acceptor functions, providing a molecular mechanism that likely contributes to the clinical link between MPO levels and cardiovascular disease risk.

Oxidative modification has been linked to changes in the function of a number of proteins with examples of both loss and gain of function (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). The oxidative modification of low density lipoprotein (LDL), 1 for example, is believed to play a crucial role in the initiation and progression of atherosclerosis (13)(14)(15)(16). Oxidatively modified LDL, but not native LDL, is taken up by macrophages via scavenger receptors that are not regulated by cellular cholesterol content to produce cholesterolladen foam cells. Although the exact mechanisms of LDL oxidation that are proatherogenic in vivo are not known, myeloperoxidase (MPO)-mediated modification appears to be a likely contributor to the process (17). MPO is a heme peroxidase that is produced by various phagocytes and utilizes hydrogen peroxide and chloride to generate chlorinating oxidants like HOCl to kill pathogens (18). Human monocytes utilize this enzyme and the co-substrates hydrogen peroxide and nitrite to generate nitrating oxidants capable of initiating lipid peroxidation and protein nitration (19). Importantly recent reports confirm that MPO promotes both protein nitration and lipid peroxidation in vivo (20 -22) and that both MPO and nitrotyrosine levels in blood are strong predictors of an increased risk of cardiovascular disease (23)(24)(25).
High density lipoprotein (HDL) also has an important role in atherosclerosis. Unlike LDL, however, HDL has antiatherosclerotic effects related to the maintenance of vascular homeostasis (26 -28). One major antiatherosclerotic function of HDL is its participation in the reverse cholesterol transport process by acting as an acceptor of free cholesterol taken from peripheral tissues for ultimate delivery to the liver and other steroidogenic tissues.
Several laboratories have reported that the oxidation of HDL in vitro alters its cholesterol acceptor function with both enhancement and inhibition in efflux noted, depending upon the oxidation system used (29 -34). Until recently, whether or not the major protein component of HDL, apolipoprotein A-I (apoA-I), was oxidized in vivo and if so by which pathway(s) were unknown. We have recently showed that apoA-I is a selective target for nitration and chlorination in human serum and within human atherosclerotic lesions (35). Those experiments also found that serum apoA-I nitrotyrosine and chlorotyrosine content is increased in subjects with cardiovascular disease and that, in apoA-I isolated from cardiovascular disease patients, higher degrees of modification were inversely correlated with lower ABCA1-dependent reverse cholesterol transport activity of the lipoprotein particle (35). Finally those studies showed that MPO selectively associates with apoA-I under physiologic conditions, such as in serum and within human atheroma, presumably through a specific interaction site mapped to the helix 8 region of the apoA-I sequence (35). Those data and subsequent confirmatory data from another laboratory (36) clearly demonstrate that apoA-I is nitrated and chlorinated in vivo and link this modification with proatherogenic changes in apoA-I function.
The purpose of the present studies was to establish the molecular events associated with MPO-mediated modification of apoA-I and specifically relate those events to alterations of apoA-I function. We mapped the sites of MPO-mediated nitration and chlorination with tandem mass spectrometry and now demonstrate the specific modification of two regions of the protein. We also established the hierarchy of modifications and report a co-localization of the preferred residue modified by both MPO-catalyzed chlorination and nitration with the previously identified MPO interaction site in the helix 8 region. A strong correlation was noted between dose-dependent progression of MPO-mediated site-specific modifications of apoA-I and loss of both ABCA1-dependent reverse cholesterol transport and inhibition of apoA1 lipid binding. We also showed that peroxynitrite (ONOO Ϫ ) preferentially nitrates apoA-I at a site distinct from that modified by MPO and with no significant functional impairment in cholesterol efflux activity of the lipoprotein. Finally we used our in vitro observations to design targeted LC-tandem MS experiments that directly detected the nitration and chlorination of specific tyrosine residues in apoA-I isolated from human atheroma tissue. As a result, our data suggest a link between the degree of MPO-catalyzed sitespecific apoA-I modifications and the loss of important antiatherosclerotic functions of HDL. MPO-catalyzed oxidative modification of apoA-I in the artery wall may thus contribute to the clinical association between MPO and cardiovascular disease.

HDL and ApoA-I Modification Reactions
Whole blood was drawn from healthy donors who gave written informed consent for a study protocol approved by the institutional review board of the Cleveland Clinic Foundation. HDL was isolated from human plasma (density range, 1.063-1.210 g/ml) using differential ultracentrifugation and extensively dialyzed in 50 mM phosphate buffer (pH 7.0) with 100 M diethylenetriamine pentaacetic acid (DTPA) in the dark at 4°C. Delipidated and purified apoA-I was purchased from Biodesign International (Saco, ME) and used without further purification.
The MPO-mediated modification reactions were carried out in a 50 mM phosphate buffer, pH 7.0, containing 100 M DTPA, 1 mg/ml protein (apoA-I), 57 nM purified human MPO (donor: hydrogen peroxide, oxidoreductase, EC 1.11.1.7; A 430 /A 280 ratio of 0.79), and either 1 mM nitrite (for the nitration reactions) or 100 mM chloride (for the chlorination reactions). The myeloperoxidase reactions were initiated by adding hydrogen peroxide at varying concentrations (0 -200 M) and carried out at 37°C for 1 h. These reaction conditions included physiologically relevant amounts of MPO, chloride, and nitrite and hydrogen peroxide concentrations that ranged from physiologic to pathologic. The peroxynitrite and HOCl reactions were similarly carried out at 37°C for 1 h in 50 mM phosphate buffer, pH 7.0, containing 100 M DTPA with the peroxynitrite and HOCl added to give final concentrations between 0 and 200 M. In all reactions, the concentrations of the key reactants were verified spectrophotometrically using molar extinction coefficients of 170 cm Ϫ1 mM Ϫ1 at 430 nm for MPO, 39.4 cm Ϫ1 M Ϫ1 at 240 nM for hydrogen peroxide, 350 cm Ϫ1 M Ϫ1 at 292 nM for HOCl (NaOCl), and 1670 cm Ϫ1 M Ϫ1 at 302 nM for peroxynitrite.
The modified proteins were taken immediately for the apoA-I functional studies described below. An aliquot of each reaction was removed, precipitated with acetone, and separated by SDS-PAGE for the mass spectrometry experiments.

Mass Spectrometry Experiments
Protein Digestion-The protein bands were digested according to an in-gel digestion procedure (37). Briefly the protein bands were cut from the gel and washed in 50% ethanol, 5% acetic acid prior to tryptic digestion with a modified, sequencing grade trypsin (Promega, Madison, WI) overnight at room temperature. The peptides were extracted from the gel, evaporated to dryness, and reconstituted in either 1% acetic acid or 0.1% formic acid for analysis by capillary column HPLCelectrospray ionization-mass spectrometry.
Identification of the Modification Sites-The detailed mapping and detection of the nitration and chlorination sites was carried out using an LCQ Deca ion trap mass spectrometer system (ThermoFinnigan, San Jose, CA) equipped with a nanospray ionization source (Protana, Odense, Denmark). The source was operated under microspray conditions at a flow rate of 200 nl/min. The digests were analyzed by reversed-phase capillary HPLC using a 50-m-inner diameter column with a 15-m-inner diameter tip purchased from New Objective Corp. (Woburn, MA). The column was packed with ϳ6 cm of C 18 packing material (Phenomenex, Torrence, CA) and eluted using a 45-min gradient of increasing acetonitrile (2-70%) in 50 mm acetic acid. The data were acquired in the data-dependent mode, recording a mass spectrum and three collision-induced dissociation (CID) spectra in repetitive cycles (37). The program Sequest was used to compare all CID spectra recorded with the sequence of human apoA-I (National Center for Biotechnology Information (NCBI) accession number 229513) and considering the appropriate changes in the tyrosine residue masses of ϩ34 Da for chlorination and ϩ45 Da for nitration.

Site-specific Quantitation of Protein Modification
Site-specific quantitation experiments were performed on a QToFmicro-mass spectrometry system (Waters, Milford, MA) equipped with a CapLC HPLC system (Waters) with autoinjector. The electrospray ionization source was operated under microspray conditions at a flow rate of 600 nl/min. The digests were analyzed by reversed-phase capillary HPLC using a 75-m-inner diameter column with a 19-m-inner diameter nanospray source capillary. The column was packed with ϳ15 cm of 10-m C 18 packing material (Phenomenex) and eluted using a 45-min gradient of increasing acetonitrile (2-70%) in 0.1% formic acid. Quantitation was achieved by using the Native Reference Peptide method developed for the site-specific quantitation of post-translational modifications (38,39) in the selected ion monitoring mode. For the selected ion monitoring experiments, full scans from m/z 300 to m/z 1600 were acquired with the time-of-flight mass analyzer of the quadrupole timeof-flight instrument, and mass chromatograms were constructed based on the m/z value of the different peptide ions of interest. The apoA-I peptides ATEHLSTLSEK and QGLLPVLESFK were used as native reference peptides. The relative quantity of each analyte peptide was determined by dividing the chromatographic peak area of that analyte peptide by the chromatographic peak area of the reference peptide.

ApoA-I Functional Studies
Cholesterol Efflux-The cholesterol efflux experiments were performed according to established procedures (40,41). Subconfluent RAW264.7 cells in 24-well dishes were cholesterol-loaded and labeled overnight in 0.5 ml of DGGB (Dulbecco's modified Eagle's medium supplemented with 50 mM glucose, 2 mM glutamine, and 0.2% bovine serum albumin) containing [ 3 H]cholesterol-labeled acetylated low density lipoprotein (AcLDL). The [ 3 H]cholesterol-labeled AcLDL was prepared by incubating [ 3 H]cholesterol for 30 min at 37°C with the AcLDL and diluted in DGGB to give a final concentration of 50 g/ml AcLDL with 0.33 Ci/ml [ 3 H]cholesterol. The day after labeling, the cells were washed three times in phosphate-buffered saline, 0.2% bovine serum albumin and incubated with 0.5 ml of DGGB with or without 0.1 mM 8-Br-cAMP for 16 h. After the 16-h incubation, 50 g/ml HDL protein in 0.5 ml of DGGB with or without 8-Br-cAMP was added to each well. After a 4-h incubation at 37°C, 100 l of medium was removed and centrifuged, and the radioactivity was counted as a measure of the effluxed cholesterol in the media. The respective cells from each well were extracted with hexane/isopropanol (3:2, v/v), and the radioactivity was determined as a measure of the cholesterol retained in the cell. The percentage of cholesterol effluxed was calculated as the radioactivity in the medium divided by the total radioactivity (medium radioactivity plus cell radioactivity).
Lipid Binding-An LDL aggregation assay, modified from a previ-ously described assay (42), was used to test apoA-I lipid binding. In a 96-well assay plate, 75 g of LDL was mixed with or without 3 g of control or modified apoA-I in a final volume of 200 l of reaction buffer (50 mM Tris-HCl, pH 7.4, with 150 mM NaCl and 2 mM CaCl 2 ). Each reaction was performed in triplicate. The plate was then incubated in a SpectraMax plate reader (Amersham Biosciences) at 37°C for 10 min. A 20-l aliquot of diluted phospholipase C, derived from Bacillus cereus (Sigma catalog number P7147) and sufficient to induce aggregation in 1 h, or a buffer control was added to each well to hydrolyze the phospholipid polar head groups to make the LDL surface hydrophobic and initiate aggregation. Aggregation at 37°C was monitored by absorbance at 478 nm, read every 2 min for a period of 1 h. ApoA-I lipid binding activity results in the inhibition of the LDL aggregation and was calculated from the aggregation rate (⌬A/min) during the rapid phase that occurred after a short time lag. The aggregation rate for the modified apoA-I was normalized to the rate of native apoA-I.

Detection of Modified apoA-I in Human Atheroma
ApoA-I Isolation-Human atheroma tissue was isolated from aortas obtained at autopsy within 10 h of death. The tissue was immediately rinsed in ice-cold phosphate-buffered saline supplemented with 100 M DTPA and immediately frozen in 65 mM sodium phosphate, pH 7.4, with 100 M DTPA and 100 M butylated hydroxytoluene under N 2 at Ϫ80°C until analysis. The DTPA was included as a metal chelator, and butylated hydroxytoluene was included as an antioxidant to prevent artifactual oxidation of the sample during the subsequent handling steps.
Fatty streaks and intermediate lesions of human thoracic aorta were powdered using a stainless steel mortar and pestle at liquid nitrogen temperatures and mixed with phosphate-buffered saline containing 100 M DTPA and a protease inhibitor mixture (Sigma catalog number P8340) for 10 h at 4°C. The suspended lesion sample was centrifuged, and the supernatant was used for apoA-I purification. For the purification, apoA-I was bound to an anti-HDL IgY resin (GenWay Biotech, San Diego, CA) and eluted in 0.1 M glycine (pH 2.5), and the eluate was neutralized by addition of 1 M Tris (pH 8.0). The neutralized sample was dissolved in sample loading buffer without heating, run in a 12.5% SDS-polyacrylamide gel (Criterion, Bio-Rad), and detected by Coomassie Blue staining. The SDS-PAGE analysis revealed that Ͼ90% of the protein recovered from the column was apoA-I.
LC-Tandem MS Analysis-The immunoaffinity-isolated apoA-I band was cut from the gel and digested with trypsin as described above. The LC-tandem MS experiments used a ThermoFinnigan LTQ linear ion trap mass spectrometer with a Surveyor HPLC pump and autosampler system. Samples were injected onto a 10-cm ϫ 75-m-inner diameter capillary column that was eluted with a linear gradient of acetonitrile in 50 mM acetic acid at ϳ1 l/min. Selected reaction monitoring experiments were used to record the product ion spectra of molecular ions of peptides containing each modified tyrosine residue characterized in the mapping experiments. Chromatograms for those peptides were reconstructed by plotting the fragmentation of the molecular ions to the most abundant product ions in the respective CID spectra. Detection of the appropriate peptide was verified by the CID spectrum that was recorded at that retention time.

Statistical Analyses
Statistically significant differences were determined by either a oneway analysis of variance using a Tukey-Kramer multiple comparisons test or a Student's t test. Statistical significant differences are reported when p Ͻ 0.05.

Mapping the Nitration and Chlorination Sites in ApoA-I-
Initial experiments focused on determining the tyrosine nitration sites in apoA-I produced by treating HDL with both an enzymatic modification system (MPO/H 2 O 2 /NO 2 Ϫ ) with varying concentrations of hydrogen peroxide and a non-enzymatic system using varying concentrations of peroxynitrite. After each treatment, the proteins in HDL were precipitated with cold acetone and separated by SDS-PAGE, and the apoA-I band was cut from the gel for in-gel digestion with trypsin. The digest was analyzed by capillary column HPLCtandem mass spectrometry using the data-dependent mode of the ion trap mass spectrometry system. Approximately 2000 CID spectra were recorded, and these spectra were searched for the spectra of modified peptides. The search routine was focused on the amino acid sequence of apoA-I and used a tyrosine residue mass difference of ϩ45 Da to find the spectra of the nitrated peptides. These analyses detected peptides that covered 95% of the protein sequence, including peptides that contained all seven of the tyrosine residues in the mature apoA-I sequence. Two peptides containing nitrotyrosine residues (Tyr-192 and Tyr-166) were found in HDL treated with MPO/H 2 O 2 /NO 2 Ϫ at concentrations of hydrogen peroxide Ͻ50 M. At higher hydrogen peroxide concentrations (Ͼ100 M), additional nitrotyrosine-containing peptides (Tyr-29 and Tyr-236) could also be detected. The three remaining tyrosine residues are contained in peptides that were detected and sequenced in these analyses, but no corresponding nitrated form was found under any reaction condition that was tested. As a comparison, a similar reaction of HDL with 100 M peroxynitrite gave nitrotyrosine modifications of the apoA-I at three tyrosine residues: Tyr-166, Tyr-18, and Tyr-236. Again the remaining tyrosine residues were detected in these analyses exclusively in the un-nitrated form.
The CID spectra of the five nitrated peptides are shown in Fig. 1. For each peptide, the identity was established by the series of product ions recorded in the respective CID spectrum. A component of these CID spectra is the characteristic residue mass of the nitrotyrosine moiety (208 Da). Overall the combination of the peptide molecular weight measurements, the peptide sequence information in the CID spectra, and the known apoA-I amino acid sequence allowed unambiguous assignment of the nitrotyrosine positions.
Subsequent experiments used complementary protein chlorination systems, with both enzymatic (MPO/H 2 O 2 /Cl Ϫ ) and non-enzymatic (HOCl) reactions, to modify the HDL. Four sites of chlorination were detected in both of these reactions: Tyr-192, Tyr-166, Tyr-29, and Tyr-236. These are the same tyrosine residues that were nitrated by the MPO/H 2 O 2 /NO 2 Ϫ reaction. As noted for the nitration sites, chlorination of Tyr-192 and Tyr-166 was detected at hydrogen peroxide concentrations Ͻ50 M, whereas chlorination of Tyr-29 and Tyr-236 required Ͼ100 M hydrogen peroxide. The Tyr-18 nitration site seen with peroxynitrite treatment was not chlorinated by either the MPO-or HOCl-mediated reaction. Fig. 2 contains the CID spectra of the four chlorinated peptides. These CID spectra are characterized by a fragmentation pattern that includes the residue mass of the chlorotyrosine (197 Da for the more abundant 35 Cl isotope). As note above for the nitrated peptides, the combination of the peptide molecular weight, the information in the CID spectrum, and the apoA-I amino acid sequence gave a clear-cut assignment of the chlorotyrosine positions.
Quantitative Analyses Identify the Preferred Modification Sites-A comprehensive map of the nitration and chlorination sites in apoA-I is shown in Fig. 3. One observation made during the initial mapping experiments was a potential hierarchy in the various modifications with two of the MPO-mediated nitration/chlorination sites being modified prior to the other two, one MPO-mediated modification site that was not modified by peroxynitrite, and one peroxynitrite modification site that was not modified by MPO. As a result, quantitative experiments were designed to specifically determine the order of the nitration and chlorination sites.
These quantitative experiments used our previously described Native Peptide Reference method to follow the disappearance of the different regions of apoA-I, represented as the respective tyrosine-containing peptides formed by the trypsin digestion (38,39). HDL was treated under nitration or chlorination conditions with increasing amounts of H 2 O 2 in the MPO-mediated reactions or increasing amounts of peroxynitrite and HOCl in the non-enzymatic reactions. The reactions were stopped by protein precipitation with Ϫ20°C acetone and separated by SDS-PAGE. The apoA-I bands were cut for tryptic digestion, and the digests were analyzed by capillary column LC-electrospray ionization-MS. The pro-gression of MPO-mediated modification of the individual sites in apoA-I was monitored by plotting mass chromatograms for each tyrosine-containing peptide and calculating the respective peak area ratios relative to the unmodified native reference peptide. As shown in Figs. 4B and 5B, the progression of the MPO-mediated modification, with the increasing concentrations of H 2 O 2 , at each site in the apoA-I produced a decreased recovery of the respective unmodified peptides in the digest. These data revealed that Tyr-192 serves as the preferred MPO-catalyzed nitration and chlorination site followed by Tyr-166 and Tyr-29 and with only a minor degree of modification of Tyr-236 at higher levels of oxidant. A comparable pattern of oxidative modification was also seen in the HOCl reaction. Remarkably the relative effectiveness of the HOCl reaction was significantly less than that of the MPO-catalyzed chlorination reaction. Specifically the MPO-catalyzed oxidation reaction produced greater modification at every concentration of H 2 O 2 examined relative to molar equivalent amounts of HOCl. In contrast, the doseresponse characteristics of the peroxynitrite reaction differed both in the efficiency of the modification reaction and in the clear preference for nitration of Tyr-18 and the absence of nitration at Tyr-192.
Oxidatively Modified ApoA-I Is Functionally Impaired-Figs. 4A and 5A also show the dose-response effects of modification of HDL on ABCA1-mediated cholesterol efflux properties. HDL was treated with the MPO/H 2 O 2 /NO 2 Ϫ system, peroxynitrite, the MPO/H 2 O 2 /Cl Ϫ system, or HOCl. Cholesterol efflux was then measured by incubation of the treated samples with cholesterol-loaded murine macrophage RAW264.7 cells in the presence and absence of pretreatment with 8-Br-cAMP. In the absence of 8-Br-cAMP treatment, RAW264.7 cells do not express an appreciable level of ABCA1 and support ABCA1independent cholesterol efflux to HDL but no cholesterol efflux to apoA-I (40, 43-45). 8-Br-cAMP treatment of RAW264.7 cells induces ABCA1 mRNA and protein allowing ϳ2-fold higher cholesterol efflux to HDL and significant levels of cholesterol efflux to lipid-free apoA-I (40,41,43,44). Therefore, the presence or absence of 8-Br-cAMP pretreatment allows one to measure both ABCA1-dependent and -independent cholesterol efflux. As seen in Figs. 4A and 5A, the MPO-mediated nitration and chlorination reactions and the HOCl treatment produced dose-dependent losses of the ABCA1-dependent cholesterol efflux to HDL without affecting ABCA1-independent efflux to HDL. Overall the rank order of efficiency for functional impairment by various modification reactions was MPO-mediated chlorination Ͼ HOCl chlorination Ͼ MPO-mediated nitration Ͼ Ͼ peroxynitrite nitration.
Control reactions of HDL treated with H 2 O 2 alone showed no decrease in ABCA1-dependent cholesterol efflux, demonstrating the critical nature of the MPO-catalyzed peroxidase reaction. The same tyrosine residues were also modified, and a similar pattern of decreased ABCA1-dependent cholesterol efflux was seen in control experiments using lipid-free apoA-I treated with the complete MPO-mediated modification systems (data not shown). These results are consistent with protein modification, as opposed to lipid modification, being responsible for the loss of efflux activity.
The effect of MPO-mediated modification on the lipid binding characteristics of apoA-I was also tested. These experiments measured the lipid binding activity of apoA-I by monitoring the ability of the apoA-I to inhibit the aggregation of LDL that is treated with phospholipase C. The apoA-I inhibition is due to its ability to coat the modified hydrophobic LDL through a lipid binding process that is an initial step in apoA-I-mediated cholesterol efflux. As shown in Fig. 6, LDL treated with the phospholipase C produces a time-dependent aggregation that is significantly reduced by the lipid binding activity of unmodified apoA-I or apoA-I that was pretreated with hydrogen peroxide alone (Fig. 6A). Pretreatment of the apoA-I with the MPOmediated nitration and chlorination systems (MPO/H 2 O 2 /NO 2 Ϫ and MPO/H 2 O 2 /Cl Ϫ , respectively) significantly inhibited this reduction with the nitration reaction giving a 10% inhibition and the chlorination reaction giving a 35% inhibition. The identical modified or control apoA-I preparations were tested for ABCA1-dependent lipid efflux acceptor activity, and the decreases in the lipid binding activity of apoA-I correlated directly with observed losses of ABCA1-dependent efflux acceptor activity (Fig. 6B).
Specific ApoA-I Modification Sites Are Found in Vivo-The LC-tandem MS studies of apoA-I modified in vitro by MPO described above produced a roster of modification sites for evaluation in vivo. Fig. 7 shows a series of selected reaction monitoring chromatograms from the LC-tandem MS analysis of apoA-I that was isolated from human atheroma tissues. The elution of peptides containing nitration at the two primary nitration sites, Tyr-192 and Tyr-166, are seen in the chromatograms shown in Fig. 7, A and C, respectively. The CID spectra (Fig. 7, B and D, respectively) recorded at these retention times provided unambiguous proof of the correct identify of these nitrated peptides. The amounts of the nitrated peptides, relative to the respective unmodified pep- . All values were normalized to the ABCA1-dependent cholesterol efflux obtained with unmodified HDL. Statistically significant differences (p Ͻ 0.05) relative the same treatment dose are designated with an asterisk. B, site-specific quantitation of apoA-I nitration determined using LC-MS. The proteins in the modified HDL reaction were precipitated with cold acetone, separated by SDS-PAGE, and detected by Coomassie Blue staining. The apoA-I was cut from the gel and digested by trypsin. The progression of the nitration reaction was followed quantitatively using the native reference peptide method. The peak area ratio of each tryptic peptide containing the respective tyrosines of interest to the native reference peptide was measured. The percent modification of each peptide was determined based on the decrease of the amount of each peptide relative to an untreated control. tides, could be estimated as 9% for the Tyr-192-containing peptide and 0.2% for the Tyr-166-containing peptide by integrating the area of each chromatographic peak. These values must be considered estimates because the relative LC-MS responses of the nitrated versus un-nitrated peptides have not been determined. Similar experiments targeting the secondary nitration sites identified through the in vitro experiments could not detect the Tyr-29-and Tyr-236-containing peptides in the nitrated forms (data not shown). This inability to detect these nitrated peptides is consistent with the relatively poor efficiency of the MPO-mediated nitration of these sites.
We also attempted to verify the presence of site-specific chlorinated peptides. A peptide containing chlorination at the Tyr-192 position could not be detected (data not shown) despite the identification of this site as the preferred site of chlorination in in vitro models. The peptide containing chlorination at the Tyr-166 position could be detected (Fig. 7B, bottom panel), although the CID spectrum (Fig. 7D, bottom panel) showed a superimposed spectrum of peptide ions containing chlorination of Tyr-166 and oxidation of Tyr-166 to give trihydroxyphenylalanine. This oxidized peptide has a molecular mass that is 2 Da lower than the corresponding chlorinated species. As a doubly charged ion, however, the chlorinated and oxidized peptides differ by 1 Da in the m/z scale of the mass spectrometer. This m/z difference cannot be distinguished in the 2-Da acceptance window of the first stage of mass analysis in the ion trap detector. The resulting singly charged fragment ions differ by 2 Da and can be distinguished in the CID spectrum. DISCUSSION A growing body of evidence suggests important links between protein nitration, MPO, and atherosclerosis. Nitrated proteins are enriched within atherosclerotic lesions (46,47), and there is a strong clinical association between elevated levels of nitrotyrosine in serum proteins and cardiovascular disease (24). An important source of nitrating species is the enzyme MPO, a protein found in human atherosclerotic lesions (48). MPO is elevated in patients with cardiovascular disease (23,25,49) and is able to produce nitrating oxidants in vivo (20 -22). Recent studies also demonstrate that serum MPO levels serve as a strong and independent predictor of endothelial dysfunction in human subjects (50), suggesting a mechanistic link between oxidation, inflammation, and cardiovascular disease. Although we have recently demonstrated that apoA-I is a selective target of nitration and chlorination in human serum and atheroma tissue FIG. 5. Dose-dependent apoA-I chlorination and the impairment of reverse cholesterol transport activity. HDL was chlorinated in a series of either MPO/H 2 O 2 /Cl Ϫ reactions with increasing amounts of H 2 O 2 or HOCl as indicated for 1.5 h. A, the effect of the respective chlorination reactions on the ABCA1-dependent reverse cholesterol efflux of the HDL. The efflux was determined using the same methods described in Fig. 4 for the nitration product. Statistically significant differences (p Ͻ 0.05) relative the same treatment dose are designated with an asterisk. B, site-specific quantitation of apoA-I chlorination determined using LC-MS. The methods used to detect and characterize these sites are given in Fig. 4.   FIG. 6. ApoA-I modification inhibits lipid binding coordinately with ABCA1-dependent cholesterol acceptor activity. A, ApoA-I lipid binding activity was assayed by its ability to inhibit phospholipase C (PCL)-induced LDL aggregation over a 1-h time course at 37°C. 3 g/ml apoA-I or H 2 O 2 -modified apoA-I decreased LDL aggregation to approximately the same extent. Compared with unmodified apoA-I, the complete MPO/H 2 O 2 /Cl Ϫ modification system and to a lesser extent the MPO/H 2 O 2 /NO 2 Ϫ system led to apoA-I with decreased ability to bind to the phospholipase C-treated LDL and inhibit its aggregation. The values represent the means of triplicate wells. The ability of apoA-I to inhibit LDL aggregation was dose-dependent (data not shown). B, lipid binding was calculated from the initial slopes from the experiment shown in A and normalized to the value for unmodified apoA-I (x axis). The identical apoA-I preparations were used to assay ABCA1-dependent lipid efflux from 8-Br-cAMP-treated RAW264.7 cells in triplicate (as shown in Fig. 5B) and normalized to the value for unmodified apoA-I (y axis). MPO modifications led to coordinate reductions in both lipid binding and ABCA1-dependent cholesterol acceptor activity (linear regression, r 2 ϭ 0.96; p Ͻ 0.0001).

FIG. 7.
Detection of nitrated peptides in apoA-I isolated from human atheroma tissue by LC-tandem mass spectrometry. Selected reaction monitoring chromatograms for the detection in vivo of unmodified, nitrated, and chlorinated peptides after in-gel tryptic digestion of apoA-I immunoaffinity-purified from human atheroma tissue. A, the detection of the unmodified and nitrated forms of peptide containing the favored Tyr-192 site determined by the in vitro experiments. B, the CID spectrum recorded at this retention time to confirm the identity of the peptide and position of the nitration. C, the detection of the unmodified, nitrated, and chlorinated forms (top to bottom, respectively) of the peptide containing the secondary Tyr-166 site. Again the CID spectra recorded at this retention time (D) confirm the identity of the peptides and the position of the nitration and chlorination. As described in the "Discussion," the CID spectrum of the nitrated peptide is unambiguous, while the CID spectrum of the molecular ion of the putative chlorinated peptide also shows significant overlap with the trihydroxyphenylalanine-containing form of this peptide that co-elutes and has a similar molecular weight. (35), very little is known about the molecular events and consequences of apoA-I oxidation.
Other investigators have reported functional alterations to HDL in vitro following tyrosylation, methionine oxidation, and exposure to chlorinating oxidants (29 -34, 51, 52). These reports, however, have been contradictory with both increased and decreased efflux reported. For example, modification of HDL by tyrosyl radical-generating systems has been shown to increase the ability of HDL to remove cholesterol from cultured cells with the formation apoA-I-apoA-II heterodimers being the primary contributors to this effect (29,30). Subsequent studies used the administration of tyrosyl radical-oxidized HDL, produced in vitro, to reduce the extent of atherosclerotic lesion formation in apolipoprotein E-deficient mice (51). Similarly oxidation of HDL with lipid hydroperoxides has been shown to oxidize methionine residues in apoA-I and increase the ability of the apoA-I to bind lipid (52). In contrast, treatment of HDL with either HOCl or enzymatic HOCl-generating systems, including MPO, has been reported to inhibit cholesterol efflux (31)(32)(33).
Our recent studies have shown that MPO is able to bind to apoA-I in human serum and atherosclerotic tissue and that apoA-I is significantly enriched in both nitrotyrosine and chlorotyrosine compared with total protein (35), observations that are consistent with MPO-catalyzed oxidation of apoA-I in vivo. Thus examination of the underlying structural and functional alterations to apoA-I that accompany HDL modification by MPOgenerated oxidants is of considerable interest. While three chlorination sites and a chlorination motif were recently described for HOCl-modified apoA-I (34), no such data exist for apoA-I nitration or the corresponding functional alterations that result. Determining the site of modification is important because a key aspect of many protein oxidation reactions is the apparent selectivity of the modification reactions, including both the proteins that are modified and the site of the modification. In fact, it has been proposed that proteins contain inherent structural elements that facilitate and direct oxidative modification as a component of the recognition and removal of aged proteins (1,2). As a result, understanding the site of modification is critical to understanding the biological effects.
The tandem mass spectrometry experiments characterized a set of nitrated and chlorinated peptides in apoA-I that are produced by MPO-mediated modification of HDL in vitro. These modifications occur in two general regions of the protein sequence, a 71-amino acid region spanning helices 8 through 10 in the C-terminal half of the protein and a 12-amino acid region in the non-helical region near the N terminus. In fact, the quantitative experiments demonstrate that the peptides containing Tyr-192 and Tyr-166 (helices 8 and 7, respectively) are far more extensively modified than the other tyrosine-containing peptides. Those experiments also showed that, while the HOCl modified the same residues that were modified in the MPO-mediated reactions, peroxynitrite exposure to HDL did not effectively modify the helix 8 and 7 regions of apoA-I.
The oxidative modification of apoA-I by MPO produced a corresponding loss of both ABCA1-mediated reverse cholesterol transport activity and lipid binding efficiency. Two aspects of the cholesterol efflux data are particularly noteworthy. First, the degree of modification seen in the site-specific peptide quantitation experiments for both the MPO/H 2 O 2 /NO 2 Ϫ and MPO/H 2 O 2 /Cl Ϫ systems paralleled the loss of ABCA1-dependent cholesterol efflux. Second, distinctions were seen in the extent to which the ABCA1-dependent cholesterol efflux was affected. The peroxynitrite-mediated modification reaction did not inhibit ABCA1-dependent cholesterol efflux acceptor activity to any significant degree, while the MPO-mediated protein nitration reaction with HDL produced a dose-dependent reduction of ABCA1-dependent cholesterol efflux acceptor activity. Moreover the MPO-mediated protein chlorination reaction of HDL was a more effective modification system and ultimately produced complete inhibition of the ABCA1-dependent cholesterol efflux acceptor activity. A similar treatment of HDL with increasing amounts of HOCl was modestly less effective than the MPO-mediated protein chlorination system. Treatment of isolated apoA-I also resulted in impaired efflux, thus eliminating lipid oxidation as an explanation for the functional changes.
At this time, the exact nature of the link between the specific nitration and chlorination sites seen in these experiments and the altered function of HDL and apoA-I is not clear. In fact, it is possible that these particular modifications primarily represent markers for the general progression of the HDL modification by MPO with the functionally critical modification being an as yet unidentified amino acid. However, previous work from our laboratory found that in the MPO-mediated modification of serum albumin, ϳ75% of a consumed tyrosine site was converted to the corresponding nitrotyrosine-containing peptide, (38) indicating that the predominant product of the MPOmediated nitration is the nitrotyrosine product. In addition, the most prevalent nitration and chlorination sites are in helix 8 and helix 7, parts of apoA-I that are important to the lipid acceptor function. One might speculate that the addition of either the nitrate group or chlorine to the Tyr-192 and Tyr-166 positions alters the helix bundle structure of lipid-free apoA-I in some way. For example, the electronegativity of the nitro and chloro groups could stabilize the interaction with lysine and arginine residues in adjacent helices and not permit the helix-opening conformational change needed for lipid binding and ABCA1-dependent lipid acceptor activity.
It is also not clear what factors direct the modification sites. The Tyr-192 site that is identified in our experiments as the preferential site of MPO-mediated nitration and chlorination fits the published HOCl-mediated chlorination motif, YXXK/ KXXY (34). The ability of this motif to direct HOCl-mediated chlorination is reportedly based on formation of an N⑀ chloramine intermediate on the lysine residue (34). Our mapping data are not consistent with this motif. Not only does the secondary site of nitration and chlorination (Tyr-166) not fit this motif, but other sites that fit the motif (Tyr-115 and Tyr-236) are either not modified or are poorly modified. That such a sequence motif might direct MPO-mediated modification chemistry seems unlikely. The chemical rationale provided for the YXXK motif as a facilitator of aromatic halogenation is based upon the idea that N⑀ amino lysine residues favorably position a presumptive N⑀ monochloramine intermediate for electrophilic addition to the aromatic ring of tyrosine. However, aromatic halogenation catalyzed by chloramines is only favored at low pH, proceeding via a Cl 2 intermediate (53). In addition, nitrogen dioxide, the proximate oxidant species formed by the MPO/H 2 O 2 /NO 2 Ϫ system (20), is relatively inefficient at protein nitration, proceeding via generation of a tyrosyl radical intermediate. Thus, enzymatic production of 2 mol of oxidant are required per mol of nitration making the position of the lysine residue of limited significance.
An alternative explanation for the MPO-dependent preferential site of modification is that the structure of the apoA-I places Tyr-192 and Tyr-166 residues in a favorable position relative to an MPO-apoA-I interaction site in helix 8 that we have identified (35), and it is this spatial relationship that serves as the primary determinant of the modification sites. Indeed the favored Tyr-192 modification site is directly within this mapped MPO-apoA-I interaction region. As far as we are aware, the present studies are the first to note distinct patterns of preferential site-specific nitration of a target protein by MPO versus peroxynitrite. Further, while an MPO-apoA-I interaction site does not explain the similar site selectivity of the modification by reagent HOCl versus the MPO/H 2 O 2 /Cl Ϫ system, our data demonstrate a significant increased efficiency of the MPO-versus HOCl-mediated oxidative modification and functional impairment of the lipoprotein, consistent with a direct MPO-apoA-I interaction, and reduced diffusion distance between oxidant source and target.
Finally these data are the first report of the direct detection of the specific nitrated and chlorinated sites in vivo. The modified peptides were detected in apoA-I isolated from human atheroma tissues. The favored MPO-mediated modification site identified in the in vitro experiments, Tyr-192, was found in the nitrated form only, whereas the secondary site was found in the nitrated, chlorinated, and a trihydroxyphenylalanine form. Indeed this trihydroxyphenylalanine form nearly masks the chlorinated peptide due to the similar molecular weight and retention time. It is intriguing to consider the extent of these modification reactions. The in vivo detection shown in Fig. 7 is consistent with 9% of this one site (Tyr-192) being nitrated in this sample. As described elsewhere, the nitrotyrosine and chlorotyrosine content of apoA-I isolated from human atheroma also indicates a high degree of modification (35). Indeed the nitro-and chlorotyrosine values for apoA-I in human atheroma are consistent with up to one-half of HDL particles containing at least one modified tyrosine residue in some patients (35). The data presented in this report, in turn, clearly show that nitrated and chlorinated HDL becomes functionally impaired with reduced ABCA1-dependent reverse cholesterol transport activity.
In summary, these data establish a hierarchy for the MPOmediated modification of apoA-I in which specific tyrosine residues are modified. In addition, these experiments also demonstrated a link between the progression of the MPO-mediated modification reaction and a loss of the lipid binding and reverse cholesterol transport functions of apoA-I. Thus, our results further support a new paradigm in the oxidation hypothesis of atherosclerosis whereby impaired HDL function is an important participant (14). Previous studies of the oxidation hypothesis have primarily focused on the oxidative modification of LDL and the acquisition of scavenger receptor recognition as a primary mechanism responsible for enhanced lipid deposition in the vessel wall. These results, however, are consistent with an equally plausible additional mechanism through which lipid deposition is enhanced by a disruption of efflux processes from the vessel wall.