Cytochrome c is an oxidative stress–activated plasmalogenase that cleaves plasmenylcholine and plasmenylethanolamine at the sn-1 vinyl ether linkage

Plasmalogens are phospholipids critical for cell function and signaling that contain a vinyl ether linkage at the sn-1 position and are highly enriched in arachidonic acid (AA) at the sn-2 position. However, the enzyme(s) responsible for the cleavage of the vinyl ether linkage in plasmalogens has remained elusive. Herein, we report that cytochrome c, in the presence of either cardiolipin (CL), O2 and H2O2, or oxidized CL and O2, catalyzes the oxidation of the plasmalogen vinyl ether linkage, promoting its hydrolytic cleavage and resultant production of 2-AA-lysolipids and highly reactive α-hydroxy fatty aldehydes. Using stable isotope labeling in synergy with strategic chemical derivatizations and high-mass-accuracy MS, we deduced the chemical mechanism underlying this long sought-after reaction. Specifically, labeling with either 18O2 or H218O, but not with H218O2, resulted in M + 2 isotopologues of the α-hydroxyaldehyde, whereas reactions with both 18O2 and H218O identified the M + 4 isotopologue. Furthermore, incorporation of 18O from 18O2 was predominantly located at the α-carbon. In contrast, reactions with H218O yielded 18O linked to the aldehyde carbon. Importantly, no significant labeling of 2-AA-lysolipids with 18O2, H218O, or H218O2 was present. Intriguingly, phosphatidylinositol phosphates (PIP2 and PIP3) effectively substituted for cardiolipin. Moreover, cytochrome c released from myocardial mitochondria subjected to oxidative stress cleaved plasmenylcholine in membrane bilayers, and this was blocked with a specific mAb against cytochrome c. Collectively, these results identify the first plasmalogenase in biology, reveal the production of previously unanticipated signaling lipids by cytochrome c, and present new perspectives on cellular signaling during oxidative stress.

Plasmalogens are phospholipids that contain a cis-vinyl ether linkage at the sn-1 position of the glycerol backbone and are essential for physiologic cellular function (1,2). Typically, plasmalogens are highly enriched in polyunsaturated fatty acids, especially arachidonic acid, at the sn-2 position (3,4). The unique properties of plasmalogens influence membrane structure and molecular dynamics, modulate the function of integral membrane proteins, and contribute to the phospholipase-mediated release of signaling fatty acids during cellular stimulation (5)(6)(7)(8)(9). Deficiencies in plasmalogen content have identified important roles for these specialized phospholipids in modulating signal transduction (10), cell proliferation (11), cholesterol trafficking (12), and sensitivity to oxidative stress (13,14). Moreover, human genetic mutations leading to deficiencies in plasmalogen biosynthesis (e.g. PEX7 and alkylglycerone phosphate synthase (AGAPS)) unambiguously demonstrate that the absence of plasmalogens results in severe human disease (15)(16)(17). However, the cellular mechanisms mediating these phenomena, the enzymes that catalyze the cleavage of the sn-1 vinyl ether in plasmalogens (plasmalogenases), and mechanistic identification of the biologic functions of plasmalogens have remained largely enigmatic.
Mammalian myocardium contains substantial amounts of plasmalogens (Ͼ25% of total myocardial phospholipids and Ͼ60% of sarcolemmal phospholipids) that are highly enriched in arachidonic acid (AA) 3 at the sn-2 position. Furthermore, plasmalogen molecular species contain the majority of AA in many human cell types. Accordingly, delineation of the molec-ular pathways mediating the cellular metabolism of plasmalogens is of paramount importance to understand their roles in cellular functions. However, despite intense efforts, an enzyme catalyzing the cleavage of the vinyl ether linkage in plasmenylcholine or plasmenylethanolamine, plasmalogenase, has eluded identification for nearly a century.
Oxidative stress is a major participant in mediating many of the pathologic processes present in diabetic cardiomyopathy, myocardial infarction/reperfusion, and the progression of heart failure (18 -20). Mitochondria occupy over 30% of cardiac volume (21) and are the primary source of reactive oxygen species generated during myocardial ischemia/reperfusion injury (22)(23)(24)(25). Furthermore, during ischemia/reperfusion, rapid changes in mitochondrial membrane potential and permeabilization (or rupture) of the outer membrane occur, leading to the release of cytochrome c (cyt c) (26 -28). The release of cyt c from the intermembrane space allows its access to multiple signaling proteins, scaffolding complexes, and distal intracellular membrane compartments (29 -31).
Under physiologic conditions, cyt c is bound to the inner mitochondrial membrane where it transfers electrons from complex III to complex IV. As an electron transfer protein, cyt c possesses a high redox potential (ϩ260 mV) and a compact tertiary structure containing a covalently bound heme iron. Upon interaction with negatively charged phospholipids (32), especially cardiolipin (33), cyt c undergoes a conformational alteration with subsequent displacement of the axial Met-80 that is accompanied by a dramatic decrease in its redox potential to Ϫ400 mV. This prevents cyt c from serving as an electron transporter and allows access of hydrogen peroxide (produced by mitochondrial superoxide dismutase) to the heme moiety, conferring peroxidase activity to the conformationally altered cyt c. Previous work has shown that cardiolipin (CL)-bound cytochrome c can oxidize the polyunsaturated linoleoyl acyl chains in cardiolipin (34,35), which can subsequently be released by calcium-independent phospholipase A 2 ␥ (36).
Based upon the known chemical sensitivity of the plasmalogen vinyl ether linkage to oxidation (37), we hypothesized that the peroxidase activity of cardiolipin-activated cytochrome c could oxidatively metabolize the vinyl ether bond in plasmalogens, leading to an intermediate that was susceptible to hydrolytic cleavage. Herein, we report that CL-activated cyt c catalyzes the oxidatively enabled hydrolytic cleavage of the vinyl ether linkage of plasmenylcholine and plasmenylethanolamine in the presence of H 2 O 2 or alternatively by oxidized CL. The products of the reaction, 2-acyl-lysophospholipids and ␣-hydroxy fatty aldehydes, were identified either directly by highresolution accurate-mass (HRAM) MS and fragmentation analyses or through synergistic chemical derivatization strategies and HRAM MS of derivatized precursor ions and their fragmentation-induced product ions. Importantly, we demonstrate that oxidized CL can activate the plasmalogenase activity of cyt c even in the absence of H 2 O 2 . To the best of our knowledge, this is the first identification of a protein that can catalyze the cleavage of the vinyl ether linkage of plasmalogens, thereby identifying multiple previously unknown cellular lipid metabolic pathways that participate in cellular metabolism, signaling, and cell fate decisions.

Conformationally altered cytochrome c is a potent plasmalogenase
Because CL-activated cytochrome c is a potent lipid peroxidase in the presence of H 2 O 2 (38), based upon energetic and molecular orbital considerations, we hypothesized that cyt c could catalyze the cleavage of the vinyl ether linkage present in plasmalogens. To test this hypothesis, we purified equine heart cytochrome c from Sigma by Mono S FPLC using an NaCl gradient. The purified cyt c migrated as a single band by SDS-PAGE and was Ͼ99% pure by silver staining (Fig. S1). The purified cyt c was incubated with small unilamellar vesicles (SUVs) comprising 1-O-1Ј-(Z)-octadecenyl-2-arachidonoyl-sn-glycero-3-phosphocholine (plasmenyl-SAPC) as host containing 10 mol % bovine heart CL (predominantly tetra-18:2 CL) as guest in the absence or presence of H 2 O 2 for 10 min at 37°C. We focused on CL because it is markedly enriched in mitochondria and has been previously demonstrated to bind to cyt c inducing a conformational change that activates its cryptic peroxidase activity (34, 39 -42). In initial experiments, reactions were monitored for the production of LPC by shotgun lipidomics. Minimal plasmalogenase activity was observed in the absence or presence of H 2 O 2 alone (Fig. 1A). Similarly, incubations with cyt c alone also resulted in barely detectable cleavage of the plasmenylcholine vinyl ether linkage (minor peak at m/z 566.4 by ESI-MS/MS) (Fig. 1A). Addition of H 2 O 2 dramatically activated cyt c plasmalogenase activity, resulting in the robust generation of 2-AA-LPC (Fig. 1A). In contrast, similar experiments with diacyl SAPC-CL SUVs did not result in appreciable production of either 1-stearoyl-LPC or 2-AA-LPC in the presence of cyt c and H 2 O 2 (data not shown). Incubation of SUVs comprising 1-O-1Ј-(Z)-octadecenyl-2-arachidonoyl-sn-glycero-3phosphoethanolamine (plasmenyl-SAPE) (45 mol %) containing 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) (45 mol %) and 10 mol % bovine heart CL in the presence of cyt c resulted in the production of 2-AA-LPE, which was markedly increased in the presence of H 2 O 2 (Fig. 1B).

Substantiation of cytochrome c as a plasmalogenase
Further characterization of the plasmalogenase activity of cytochrome c revealed that production of 2-AA-LPC from plasmenyl-SAPC depended upon the amount of purified cytochrome c ( Fig. 2A) with a maximum apparent turnover number of Ϸ6 at low concentrations of cyt c (1 M). The reaction was also dependent on the concentration of H 2 O 2 (Fig. S2A) as well as on the mol % cardiolipin present in the host plasmenyl-SAPC vesicles (Fig. S2B). Kinetic analysis of reaction products at a fixed concentration of cytochrome c (5 M) in the presence of H 2 O 2 (200 M) demonstrated a linear increase in 2-AA-LPC and 2-AA-LPE production utilizing plasmenyl-SAPC/1palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine or plasmenyl-SAPE/POPC SUVs, respectively, each containing 10 mol % CL as guest to mimic physiologic conditions (Fig.  S2C). The initial catalytic rates of vinyl ether cleavage were ϳ50 nmol of 2-AA-LPC/min⅐mg of cyt c and ϳ60 nmol of 2-AA-LPE/min⅐mg of cyt c.

Plasmalogenase activity of cytochrome c
We next examined the susceptibility of the plasmalogenase reaction to various inhibitors. Importantly, a cyt c-specific mAb resulted in the nearly complete inhibition of H 2 O 2 -dependent cyt c plasmalogenase activity (Fig. 2B). Moreover, cyanide and azide inhibited the plasmalogenase activity as has been shown for inhibition of the peroxidase activity of cyt c (43, 44) (Fig. 2B). Although the singlet oxygen scavengers lycopene and ␤-carotene did not appreciably affect cyt c-mediated plasmalogenase activity, the lipid hydroperoxyl scavenger ␣-tocopherol markedly inhibited the measured plasmalogenase activity (Fig.  2C). Interestingly, bifonazole, a cytochrome P450 inhibitor that inhibits the peroxidase activity of cyt c (45), potently inhibited cyt c-mediated plasmalogenase activity (Fig. 2C). To rule out the possible contribution of free heme or free iron to the observed plasmalogenase activity, hemin or ferric chloride was substituted for cyt c in the presence of hydrogen peroxide. Incubations with these reagents did not result in appreciable plasmalogenase activity (Fig. S3A). Additional experiments with catalase and nitric-oxide synthase revealed that these heme proteins exhibited Ͻ10% and no plasmalogenase activity compared with cyt c, respectively (Fig. S3B), suggesting that cytochrome c is relatively specific for cleavage of the plasmalogen vinyl ether linkage. We specifically point out that these results do not preclude the possibility that other heme proteins may also be capable of cleaving the plasmalogen vinyl ether bond. Collectively, these results firmly identify cytochrome c as an enzymic mediator of the cleavage of the plasmalogen vinyl ether linkage.

Identification of the regiospecificity of the lysolipid reaction products from plasmenylcholine and plasmenylethanolamine
To establish the regiospecificity of the arachidonoyl moiety in the lysolipids produced by cyt c-catalyzed plasmalogenase activity, full-mass MS 2 analyses of the lysolipid products and authentic synthetic 2-AA-LPC and 2-AA-LPE standards were compared. Excellent matches of the full-mass MS 2 spectra The mAb directed against cyt c was preincubated with cyt c prior to the assay for plasmalogenase activity. C, effect of singlet oxygen scavengers ␤-carotene (␤-Car) and lycopene (Lyc), the lipid peroxyl radical scavenger ␣-tocopherol (␣-Toc), or the cytochrome P450 inhibitor bifonazole (BFZ) on the plasmalogenase activity of cytochrome c. Following addition of internal standards (17:0 LPC and 14:1 PC), reaction products were extracted and analyzed as described under "Experimental procedures."

Plasmalogenase activity of cytochrome c
(shown as mirror plots) were obtained between the lysolipid reaction products and corresponding authentic 2-AA-LPC and 2-AA-LPE standards (Fig. 3, A and D). For 2-AA-LPC, the presence of a high abundance product ion at m/z 147.1 (i.e. sodiated ethylene phosphane) demonstrated the regiospecificity of the arachidonate group at the sn-2 position as described previously (46). We also performed high-mass-accuracy MS 1  A, full-mass MS 2 analysis of both plasmalogenase assay product and synthetic 2-AA-LPC at m/z 566.2 was performed on a triple-quadrupole mass spectrometer. An excellent match of the two MS 2 spectra are displayed as a mirror plot, including the presence of high-abundance product ion at m/z 147.1 (i.e. sodiated ethylene phosphate), which indicates the regiospecific acylation at the sn-2 position of LPC species (71). B and C, HRAM MS 1 and MS 2 mass spectrometric analysis of the assay product at m/z 566.3223 from A was also performed on an LTQ-Orbitrap mass spectrometer. The measured m/z values of the product ions (at m/z 507.2491 and 383.2559) resulting from loss of trimethylamine and phosphocholine, respectively, confirm the identity of the phosphocholine headgroup in LPC with each within Ϯ2 ppm of theoretical values. In panel B, the peak corresponding to m/z 532.3380 is the 17:0 LPC internal standard. D, MS 2 analysis of both the plasmalogenase assay product and synthetic 2-AA-LPE at m/z 524 (sodium adduct) was performed on a triple-quadrupole mass spectrometer. The MS 2 spectra demonstrate the presence of diagnostic fragment ions (prominent m/z 164 (sodiated phosphoethanolamine) and m/z 383 (sodiated arachidonoyl glycerol) peaks) indicative of 2-AA-LPE (46). E and F, HRAM MS 1 and MS 2 mass spectrometric analysis of the assay product at m/z 524.2752 from D was also performed on an LTQ-Orbitrap mass spectrometer showing fragment ions corresponding to the accurate mass Ϯ2 ppm of the aforementioned product ions. In panel E, the peak corresponding to m/z 448.2436 is the 14:0 LPE internal standard.

Plasmalogenase activity of cytochrome c
ppm of their calculated theoretical masses (Fig. 3, B, C, E, and F).

Identification of the oxidized sn-1 chain aliphatic constituents
To determine the reaction products resulting from oxidative cleavage of the vinyl ether bond, we utilized synergistic derivatization strategies to avoid the rapid postsource decay of the anticipated aldehyde reaction products. After incubation of cyt c with plasmenyl-SAPC/CL (90:10) SUVs in the presence of H 2 O 2 , the resultant aldehydes were derivatized with O-benzylhydroxylamine (BOA). High-mass-accuracy MS spectra of aldehyde BOA derivatives demonstrated an intense molecular ion at m/z 390.3365 (Fig. 4A), indicative of the anticipated product of the C18:0 ␣-hydroxyaldehyde BOA adduct. Fragmentation of the molecular ion at m/z 390.3365 resulted in the appearance of a predominant peak at 372.3242 resulting from loss of H 2 O (Fig. 4B). Similar BOA derivative ions were obtained when plasmenyl-SAPE/POPC/CL vesicles were used as substrate (Fig. 4, C and D). To further confirm the presence of the hydroxyl group in the resultant aldehyde product, dim-ethylglycine (DMG) was used to derivatize the hydroxyl groups. An intense molecular ion at m/z 370.3328 corresponding to the mass of the C18:0 ␣-hydroxyaldehyde DMG derivative was routinely observed (Fig. 4E). Similar results were obtained when plasmenyl-SAPE was used as substrate (Fig. 4F). Collectively, these results demonstrate that cytochrome cmediated oxidative cleavage of the vinyl ether bond of plasmalogens generates lysophospholipids and ␣-hydroxy fatty aldehydes as products.

Activation of cyt c-mediated plasmalogenase activity by oxidized CL alone or by phosphatidylinositol phosphates in the presence of H 2 O 2
Because cardiolipin is likely oxidized by cyt c during the plasmalogenase reaction, we examined whether CL preoxidized with cyt c in the presence of H 2 O 2 could activate the plasmalogenase activity of cyt c in the absence of exogenous H 2 O 2 in the reaction mixture. First, cardiolipin was oxidized by cyt c in the presence of hydrogen peroxide (35). Second, oxidized CL molecular species were purified by HPLC and analyzed by Plasmalogenase activity of cytochrome c HRAM MS. Spectra revealed the presence of multiple oxidized molecular species with an approximate 1:1 mixture of hydroperoxides and hydroxides, similar to results described previously (35). Next, we incorporated the HPLC-purified cyt c-oxidized CL (10 mol %) as guests into plasmenyl-SAPC SUV hosts by sonication and then measured the plasmalogenase activity of cyt c using these SUVs as substrate containing oxidized CL in the absence of additional H 2 O 2 . Remarkably, inclusion of cyt c-oxidized CL fully reconstituted the plasmaloge-nase activity even in the absence of exogenous H 2 O 2 (

Plasmalogenase activity of cytochrome c
does not appreciably bind to vesicles in the absence of cardiolipin it readily binds to plasmenyl-SAPC LUVs containing either nonoxidized or oxidized CL (Fig. 5A, lower panel). These results are consistent with the CL-dependent activation of cyt c plasmalogenase activity (Fig. S2B).
Next, we hypothesized that other negatively charged phospholipids such as PI(4,5)P 2 and PI(3,4,5)P 3 (which are enriched in the plasma membrane) could also act to transform cyt c into a potent plasmalogenase after its release from the mitochondrial intermembrane space. Accordingly, we examined the ability of 5 mol % PI(4,5)P 2 or PI(3,4,5)P 3 as guests in host plasmenyl-SAPC SUVs in the presence of H 2 O 2 to activate cyt c plasmalogenase activity. The results demonstrated robust activation of plasmalogenase activity with phosphatidylinositol phosphates similar to that observed with CL, suggesting that released cyt c activated by association with negatively charged lipids has multiple membrane targets after its release from the mitochondria (Fig. 5B).

Mitochondrial cytochrome c mediated plasmalogenase activity
Next, we examined whether cyt c released from rabbit heart mitochondria by outer membrane permeabilization could cleave the vinyl ether linkage of plasmenyl-SAPC in exogenous SUVs containing CL. Incubation of rabbit heart mitochondria in the presence of KCl and calcium phosphate resulted in swelling of the mitochondria and the robust release of cyt c as demonstrated by Western blot analysis of the mitochondrial supernatant with a specific anti-cyt c mAb (Fig. 5C). After removal of the mitochondria by centrifugation, the resultant supernatant was incubated with plasmenylcholine vesicles containing 10 mol % oxidized CL in the absence or presence of a specific mAb against cyt c. Similar to incubations with purified cyt c, addition of the anti-cyt c mAb to stressed rabbit heart mitochondrial supernatants resulted in marked inhibition of plasmalogenase activity, indicating that cyt c was responsible for the observed activity (Fig. 5C).

Human heart mitochondria contain abundant plasmalogen molecular species
Mammalian myocardium is known to be enriched in plasmalogen molecular species (7). Previously, subcellular fraction-ation and mass spectrometric analysis of phospholipids from canine myocardium revealed substantial enrichment of plasmalogens in sarcolemmal and sarcoplasmic reticular membranes (3,4). To determine whether mitochondrial membranes contained plasmalogens representing a proximal target for cytochrome c plasmalogenase activity, we measured choline and ethanolamine plasmalogens in human heart mitochondria by shotgun lipidomics. The results demonstrated that human heart mitochondria contain an abundance of both ethanolamine and choline plasmalogens enriched in AA (Fig. 5D). These results strongly suggest that cytochrome c localized to human heart mitochondria generates plasmalogen-derived lipid second messengers during oxidative stress. 18 (47) or DMG (for derivatization of hydroxyl groups). These reactions were each selected for the high ionization efficiency of their derivatives and their ability to substantiate the identities of the proposed reaction products (48). Control incubations of cyt c with 10 mol % oxidized CL/plasmenyl-SAPC vesicles in the presence of supplemental 16 (47), one of the oxygen atoms in molecular oxygen must be incorporated into the aliphatic chain of the aldehyde product, likely as an ␣-hydroxyl group that readily loses water upon fragmentation. In contrast, reactions performed in H 2 18 O did not result in   (Fig. 8, B and C). Fragmentation of this ion (m/z ϭ 372.3367) yielded ions at both m/z 104.0709 and m/z 106.0749 with the latter ion indicating the presence of 18 O in protonated DMG that was originally present in the hydroxyl group of the ␣-hydroxyaldehyde (Fig. 9, B and C). Importantly, incubations with both 18 (Fig. 8D). Fragmentation of this molecular ion generated a predominant protonated DMG product ion at m/z 106.0750 (Fig. 9D). Remarkably, incubations with hydrogen [ 18 O]peroxide did not result in significant (Ͻ5%) incorporation of 18 O into the DMG derivative of the C18:0 ␣-hydroxyaldehyde (Fig. S4). We specifically point out that exchange of the aldo, keto, and gem-diol forms of the ␣-hydroxyaldehyde in water likely contributes to the observed distribution of the 18 O label present in the protonated DMG product ion. Mechanistically, these studies suggest that cyt c likely abstracts an electron from the stabilized resonance form of the vinyl ether bond to form a carbonbased radical that reacts with molecular oxygen to form a peroxy or epoxide intermediate (Fig. 10). Next, the positive charge on the vinyl ether oxygen is relieved by addition of H 2 O to form a hemiacetal, which then spontaneously rearranges at neutral pH to yield the resultant ␣-hydroxyaldehyde and 2-acyl-lysolipid products.

Discussion
Despite nearly a century of investigation, the identity of the enzyme(s) responsible for the cleavage of the vinyl ether linkage in plasmenylcholine and plasmenylethanolamine has remained

Plasmalogenase activity of cytochrome c
an enigma. Herein, we report the first molecular identification of a plasmalogenase in biology. The reaction mechanism is unique in that it utilizes the initial oxidation of the vinyl ether linkage by conformationally altered cyt c activating its cryptic peroxidase activity. Stable isotope experiments demonstrated that this leads to the addition of molecular oxygen to the ␣-carbon followed by the addition of water, yielding a hemiacetal intermediate that spontaneously rearranges at neutral pH to generate a 2-acyl-lysolipid and an ␣-hydroxyaldehyde (Fig. 10). Furthermore, the present study demonstrates that the vinyl ether linkage in plasmenylcholine and plasmenylethanolamine can be readily cleaved by cyt c in the presence of negatively charged phospholipids (e.g. tetra-18:2 CL, PI(4,5)P 2 , or PI(3,4,5)P 3 ) and H 2 O 2 or by oxidized CL alone.

Plasmalogenase activity of cytochrome c
Previous studies have demonstrated the susceptibility of negatively charged lipids (e.g. CL and phosphatidylserine) to bind and induce conformational alterations that transform cyt c from an electron carrier to a potent peroxidase that can extract the bisallylic protons in CL (34,38,49). The resultant allylic radicals readily add molecular oxygen, resulting in the generation of CL hydroperoxides that can then be reduced to hydroxides (34). In contrast, incubations with zwitterionic diacyl PC and PE containing polyunsaturated fatty acyl chains were not effective in inducing cyt c-mediated oxidation in our hands and in agreement with previous studies (50). Our results demonstrate that cyt c is a highly active plasmalogenase after conformational alterations by negatively charged lipids that displace the axial Met-80, allowing access of H 2 O 2 or organic peroxides (e.g. CL hydroperoxides) to activate its peroxidase activity, which is typical for many members of the peroxidase family of enzymes.
The present results demonstrate that 18 O from 18 O 2 was predominantly incorporated into the hydroxyl group of the ␣-hydroxyaldehyde, whereas 18 O from H 2 18 O was delivered to the aldehyde carbon. This isotopic distribution indicates that the reaction sequence is likely initiated from a stable resonance form of vinyl ethers where an electron is transferred from the ␤-vinyl ether carbon (oxidation) to cyt c to form the allylic radical intermediate, which can then readily add molecular oxygen (Fig. 10). Next, the positive charge on the ether oxygen is relieved by addition of water to form the hemiacetal, which at neutral pH rearranges to form the lysolipid and the ␣-hydroxyaldehyde. It is also likely that after addition of molecular oxygen an epoxide can be formed, which then stereospecifically ringopens to form the hemiacetal with subsequent rearrangement to the identified 2-acyl-lysolipid and ␣-hydroxyaldehyde reaction products (Fig. 10).
Typically, peroxidase activity results in incorporation of oxygen from H 2 O 2 into the oxidized substrate. However, we could not observe any detectable incorporation of oxygen from H 2 18 O 2 into the ␣-hydroxyaldehyde product. Thus, these stable isotope studies identify an unusual reaction coordinate where molecular oxygen is incorporated into the ␤-vinyl ether carbon whose hydrogen cannot be directly abstracted by a radicalmediated mechanism. Instead, a stable resonance form is utilized where an electron is extracted, effecting the oxidation of the vinyl ether linkage by interactions with activated cytochrome c. Moreover, it is unusual for peroxidases to directly incorporate molecular oxygen without the use of a radical intermediate (OH ⅐ or RO ⅐ ), which would lead to stable isotope scrambling.
Previously, a mammalian enzyme capable of hydrolytic cleavage of lysoplasmalogens through a nonoxidative mechanism has been identified (51). However, despite intense efforts to identify plasmalogen hydrolysis by this protein (Tmem86b, HUGO 28448), the results demonstrated that this enzyme does not possess plasmalogenase activity. Accordingly, Tmem86b plays a role in lysoplasmalogen catabolism in liver (51) likely with the obligatory participation of the lysoplasmalogen sn-2 hydroxy group through internal assistance that appears necessary for catalysis as originally proposed 50 years ago (52). In addition, leukocyte myeloperoxidase has been demonstrated to indirectly hydrolyze the vinyl ether linkage of plasmalogens through production of highly acidic HOCl or HOBr to generate ␣-haloaldehydes (53)(54)(55)(56)(57). Thus, the myeloperoxidase cleavage of plasmalogens is acid-catalyzed and is not mechanistically related to the oxidatively enabled hydrolytic cyt c-catalyzed plasmalogenase reaction described herein.
The demonstration of plasmenylcholine and plasmenylethanolamine in mitochondrial membranes suggests that cytochrome c can regulate/modulate multiple mitochondrial bioenergetic and signaling functions through its plasmalogenase activity. In addition, a substantial portion of myocardial mitochondria are in direct contact with the sarcoplasmic reticulum (SR) by tethering to mitofusin1 (58). Because the SR is highly enriched in plasmalogen molecular species (4) and tethered to mitochondria, the activation of cyt c plasmalogenase activity released from the mitochondrial intermembrane compartment in the immediate vicinity of the SR is anticipated to result in plasmalogen cleavage and alterations in SR calcium homeostasis. Furthermore, the existence of subsarcolemmal mitochondria in myocardium is well known (59). Accordingly, we propose that subsarcolemmal mitochondria can release cytochrome c in close spatial juxtaposition to the sarcolemmal membrane, which is highly enriched in plasmalogen molecular species containing AA (3), identifying numerous potential signaling functions. For example, lysolipids are known to precipitate alterations in electrophysiologic function, including ventricular tachyarrhythmias and sudden death (60), whereas arachidonic acid and its metabolites modulate ion channel kinetics (61,62). Moreover, the highly reactive fatty acyl ␣-hydroxyaldehyde product is likely to covalently react with multiple nucleophiles to form adducts, which could be detrimental to the function of critical enzymes (63,64). Finally, it is known that healthy cells can release small amounts of cyt c from mitochondria (65,66), which we speculate could serve signaling functions between mitochondria and other intracellular organelles/membranes.
Recently, we have demonstrated that cyclooxygenase (COX)-2, but not COX-1, mediates the oxidation of 2-AA-LPC and 2-AA-LPE to generate specific eicosanoid lysolipids, which are potentially important signaling molecules (67). The previously unknown pathway identified in the current studies provides an alternative route to the generation of 2-AA-lysolipids for oxidation to biologically active signaling metabolites (Fig.  11). Alternatively, because plasmalogens are often highly enriched in AA, the hydrolysis of 2-AA-containing lysolipids by lysophospholipases can release arachidonic acid for multiple downstream metabolic and signaling pathways mediating proinflammatory and vasoactive effects among others (68 -71).
In regard to ischemia/reperfusion, it is well established that the majority of cell death occurs in the center of the infarct zone with varying degrees of cell survival in the border zones. Recent studies have demonstrated that beneath a certain threshold level of damage to mitochondria, cardiac myocytes are able to remove these damaged mitochondria through mitophagy and survive (72). However, when the level of damage exceeds that which culling can adequately repair, cardiac myocyte death ensues. Thus, the salvage of cardiac myocytes in the border Plasmalogenase activity of cytochrome c zone of an infarct is a primary goal of modern cardiology research. The present study identifies both a likely mechanism modulating mitochondrial communication and regulation of cardiac metabolism and signaling in normal myocardium as well as a potential therapeutic target for the salvage of damaged cardiac myocytes in border zones or in failing hearts. The potential implications of inhibiting this reaction to salvage myocardium to decrease morbidity and mortality have profound medical significance.
Previous studies have demonstrated a dramatic reduction (ϳ60% decrease) of plasmalogen molecular species in brains of patients with Alzheimer's disease (AD) (e.g. see Ref. 73). Furthermore, there have been numerous studies demonstrating that mitochondrial dysfunction precedes the deposition of amyloid and that mitochondrial oxidative stress or defective axonal transport is the proximal mediator of cognitive dysfunction in AD (e.g. see Refs. 74 and 75). Moreover, previous studies have identified the presence of ␣-hydroxyaldehydes in patients with AD, and the amount of ␣-hydroxyaldehydes was highly correlated with the state of cognitive dysfunction (76). Accordingly, we propose that cytochrome c-mediated degradation of plasmalogens due to increased oxidative stress is a potential mechanism responsible for the decrease in plasmalogens, membrane dysfunction, altered intramembrane proteolysis, and resultant cognitive abnormalities in patients with AD.
Collectively, the present study identifies conformationally activated cytochrome c as the first plasmalogenase in biology.
Furthermore, cyt c initiates the cleavage of the vinyl ether bond in plasmalogens by a previously undescribed oxidatively enabled hydrolytic mechanism, generating a hemiacetal that spontaneously rearranges to produce 2-acyl-lysolipids and ␣-hydroxyaldehydes. This process has the catalytic potential to facilitate regulation of mitochondrial bioenergetics and signaling as well as communication with distal membrane compartments following release of cyt c from mitochondria during oxidative stress.

Chromatographic purification of cytochrome c
Equine cytochrome c was purchased from Sigma and further purified by FPLC utilizing a Mono S column equilibrated with 20 mM HEPES, pH 7.4, containing 0.1 mM DTPA. Briefly, 0.5 mg of solid equine cytochrome c was dissolved in 1 ml of 20 mM HEPES, pH 7.4, containing 0.1 mM DTPA and applied to a Mono S column equilibrated with the same buffer. Bound cytochrome c was eluted from the Mono S FPLC column using a linear gradient of NaCl (0 -500 mM) over 25 column volumes. Fractions (0.5 ml) were collected and assayed for plasmalogenase activity using plasmenylcholine/CL SUVs in the presence of hydrogen peroxide as described below. Larger scale purifications of cytochrome c (50 mg) were performed utilizing an SP Sepharose column (1.5 ϫ 15 cm) equilibrated with 20 mM HEPES, pH 7.4, containing 0.1 mM DTPA. Bound cytochrome c was eluted with a linear NaCl gradient (500 mM final concentration) over 12 column volumes (360 ml). Eluted cytochrome c was concentrated using Amicon centrifugal filter cartridges (3000 molecular weight cutoff). The concentration of cytochrome c was determined by the difference in absorbance between fully oxidized cytochrome c and reduced cytochrome c utilizing sodium dithionite (⑀ 550 ϭ 21.1 mM Ϫ1 cm Ϫ1 ) (80). Absorbance determinations were made using a PerkinElmer Life Sciences LAMBDA 25 UV/visible double beam spectrophotometer with UV WinLab software.

Assay of plasmalogenase activity
For most reactions, 200 l of purified cytochrome c (10 M) in 10 mM potassium phosphate, pH 7.0, containing 0.25 mM DTPA was incubated with small unilamellar vesicles of plasmenyl-SAPC (250 M) or plasmenyl-SAPE (125 M)/POPC (125 M) containing 10 mol % bovine heart cardiolipin (predominantly tetra-18:2 CL) with the indicated concentrations of hydrogen peroxide for 10 min at 37°C. The concentration of hydrogen peroxide was measured using an extinction coefficient, ⑀ 240 , of 43.6 M Ϫ1 cm Ϫ1 . For some reactions, cardiolipin preoxidized by cytochrome c and H 2 O 2 or phosphatidylinositol phosphates (PI(4,5)P 2 and PI(3,4,5)P 3 ) were substituted for nonoxidized cardiolipin. Reactions were terminated by addition of 1 ml of methanol, the samples were placed on ice, and 1 ml of chloroform was added. Next, appropriate internal standards (5 nmol of 17:0 LPC and 50 nmol of 14:1 PC for reactions with plasmenylcholine and 5 nmol of 14:0 LPE and 50 nmol of 16:1 PE for reactions with plasmenylethanolamine) were added. Phase separation was induced by addition of 900 l of 25 mM NaCl followed by vigorous vortexing and centrifugation to effect phase partitioning. The water-methanol layer was re-extracted with chloroform and combined with the original chloroform layer from the first extraction, and samples were dried under a nitrogen stream.

Mass spectrometric identification and quantitation of 2-acyl-lysophospholipids
Mass spectrometric analysis was performed using a TSQ Quantum Ultra mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with an automated nanospray apparatus (Nanomate HD, Advion BioSciences, Ithaca, NY) for direct infusion MS as described previously (81). The capillary temperature of the mass spectrometer was set at 150°C. The spray voltage and gas pressure of the Nanomate were set at 1.3 kV and 1.1 p.s.i., respectively. Mass spectrometric analyses including full-mass MS 1 spectra and tandem mass spectrometric analyses were all performed in the positive-ion mode. Identification of 2-AA-LPC and 2-AA-LPE was performed by ESI-MS and/or ESI-MS/MS mass spectrometric analyses of the sodiated molecular ion at m/z 566 for 2-AA-LPC, the protonated molecular ion at m/z 502 for 2-AA-LPE, or the sodiated molecular ion at m/z 524 for 2-AA-LPE. The regiospecificity of AA-LPC and AA-LPE was determined by the ratio of product ions as described previously (46,71). The sodiated 2-AA-LPC and 2-AA-LPE were also analyzed using a hybrid mass spectrometer (LTQ-Orbitrap, Thermo Fisher Scientific) by direct infusion in the positive-ion mode with sheath, auxiliary, and sweep gas flows of 4, 1, and 0 arbitrary units, respectively. The capillary temperature was set at 275°C, the electrospray voltage was 4.0 kV, and tube lens were set to 100 V. The instrument was calibrated as instructed by the manufacturer prior to use. The isolation width was set to Ϯ1 thomson, normalized collision energy was set at 30 eV for 2-AA-LPC and 22 eV for 2-AA-LPE, and resolution was 30,000 at m/z 400. Instrument control and data acquisition were performed using Thermo Xcalibur version 2.1 software.

Derivatization and mass spectrometric identification of ␣-hydroxy fatty aldehydes
Reaction products from incubations of plasmenyl-SAPC SUVs containing 10 mol % bovine heart CL in the presence or absence of cyt c and H 2 O 2 were extracted into an equal volume of cyclohexane/diethyl ether (4:1). ␣-Hydroxy fatty aldehydes Plasmalogenase activity of cytochrome c were derivatized with DMG in a new tube by a modified procedure (48) in the presence of hexadecanol (1 nmol) and analyzed in the positive-ion mode. Briefly, to the dried cyclohexane/diethyl ether extract, 5 l of a mixture of N,N-dimethylglycine HCl (1 M) and DMAP (1 M) were added in anhydrous chloroform followed by addition of 10 l of N-(3-dimethylaminopropyl)-NЈ-ethylcarbodiimide hydrochloride (0.25 M) in anhydrous chloroform, 200 l of anhydrous chloroform, and 100 mg of 4-Å molecular sieve followed by vortexing and incubation at 45°C for 60 min. Reactions were stopped by drying the solvent under an N 2 stream followed by resuspension in 200 l of isopropyl alcohol and analysis by LC-MS/MS as described below. ␣-Hydroxyaldehydes were also derivatized with BOA in the presence of 1 ml of 0.1 M borax buffer, pH 9.4, containing 1 nmol of fatty aldehyde internal standard (hexadecanal-d 9 ) as described previously (47). Briefly, 2 mol of BOA was added to each reaction in borax buffer and incubated for 3 h at room temperature. An equal volume of cyclohexane/diethyl ether (4:1, v/v) was then added with vortexing and centrifuged to separate organic and aqueous layers. The upper organic layer was transferred to a new tube, 1 ml of cyclohexane/diethyl ether was added to the remaining aqueous phase, and the mixture was vortexed and subsequently centrifuged to separate the layers. The organic phases were combined and dried under a nitrogen stream prior to mass spectral analysis.
The BOA-and DMG-derivatized ␣-hydroxyaldehydes were analyzed by LC-MS/MS using a Surveyor HPLC system and an LTQ-Orbitrap mass spectrometer. The BOA-or DMG-derivatized ␣-hydroxyaldehydes were separated with a C18 reversephase column (Kinetics EVO C18, 5 m, 150 ϫ 2.1 mm) at 22°C with a flow rate of 200 l/min. A linear gradient of solvent A (water with 10 mM ammonium acetate, pH 4.3) and solvent B (isopropyl alcohol) was used as follows: 0 min, 25% B; 5 min, 25% B; 20 min, 100% B; 25 min, 100% B; 25.1 min, 25% B; 30 min, 25% B. The sample injection volume was 10 l, and the autosampler tray temperature was set at 4°C. The spray voltage in the ESI source was 4.1 kV. The sheath gas flow rate was 40. The capillary temperature was 270°C. The BOA-and DMGderivatized ␣-hydroxyaldehydes were also analyzed by tandem MS. The collision energy used was 30 eV (collision-induced dissociation) for the BOA-derivatized sample and 35 eV (highenergy collision dissociation) for the DMG-derivatized sample with an isolation width of Ϯ1.5 thomsons.

Measurement of incorporation of 18 O from 18 O 2 and H 2 18 O into plasmalogenase reaction products
Plasmenyl-SAPC vesicles containing 10 mol % cardiolipin in 10 mM potassium phosphate buffer, pH 7.0, containing 0.25 mM DTPA was 1) purged with 18 18 O atom enrichment). Concentrated cytochrome c (1 mM) was then injected into the vial to 10 M final concentration and vortexed. After incubation for 30 min at 37°C, the reaction was terminated by addition of 0.8 ml of cyclohexane/diethyl ether (4:1) and vortexed. The upper organic layer was removed, and the remaining aqueous layer was re-extracted with 0.8 ml of cyclohexane/diethyl ether (4:1). The com-bined organic layers were then dried under a nitrogen stream and subjected to derivatization and LC-MS/MS analysis as described above.

Isolation of rabbit heart mitochondria, induction of mPTP opening, and release of cytochrome c
Male New Zealand rabbits were sedated with ketamine/xylazine (35/5 mg/kg) prior to euthanasia with pentobarbital (150 mg/kg) and removal of the heart under a protocol approved by the Washington University Animal Studies committee. After thorough washing with PBS to remove excess blood, the heart was finely diced with a razor blade prior to homogenization using a Teflon homogenizer (150 rpm ϫ 10 passes) in mitochondrial isolation buffer (MIB) (10 mM HEPES, pH 7.4, containing 0.21 M mannitol and 0.07 M sucrose) containing 5 mg/ml fatty acid-free BSA. Following pelleting of myofibrils/nuclei/ cell debris by low speed centrifugation (1000 ϫ g for 7 min), mitochondria were isolated from the resultant supernatant by differential centrifugation at 10,000 ϫ g for 10 min. Mitochondria were washed three times with MIB without BSA before use. Release of cyt c was induced by incubation of mitochondria in MIB containing 75 mM KCl and 75 M CaHPO 4 for 35°C for 30 min. Swollen mitochondria were then pelleted by centrifugation at 10,000 ϫ g for 10 min to obtain the supernatant fraction. Proteins in the resultant supernatant and remaining mitochondria (pelleted by centrifugation) were resolved by SDS-PAGE (14% gel), transferred to PVDF membranes, and probed for cytochrome c content by ECL Western blot analysis utilizing a monoclonal cyt c antibody and anti-mouse IgG-HRP conjugate as described previously (83). The supernatant was analyzed for plasmalogenase activity using CL oxidized by cyt c in the absence or presence of a mAb (1 g) directed against cyt c as described above.

Collection of human heart tissue and isolation of mitochondria
Human heart tissue was obtained from Mid-America Transplant (St. Louis, MO) and the Washington University Translational Cardiovascular Tissue Core. Tissue from the left ventricle and left ventricular apex of nonfailing human hearts was obtained from organ donors having a heart deemed unsuitable for transplantation due to donor age, epicardial (nonocclusive) coronary vascular disease, or high risk behavioral profile. All tissue was trimmed of excess fat, washed in cold PBS, and then placed in ice-cold MIB prior to isolation of mitochondria by differential centrifugation as described previously (84).

Determination of plasmenyl-PC and plasmenyl-PE content of human heart mitochondrial phospholipids by shotgun lipidomics
Isolated human heart mitochondria were extracted into chloroform using a Bligh-Dyer extraction method in the presence of internal standards (14:1 PC and 16:1 PE) as described previously (85). Phosphatidylethanolamine phospholipids were derivatized with fluorenylmethoxycarbonyl (Fmoc) chloride as described previously (86). Acid hydrolysis of plasmalogens was performed as described above, cleaving only vinyl ether lipids Plasmalogenase activity of cytochrome c without affecting either diacyl or alkyl ether phospholipids. The resultant lipid extracts were analyzed utilizing a TSQ Quantum Ultra Plus triple-quadrupole mass spectrometer as described above.

Binding of cyt c to LUVs
LUVs were prepared by extrusion of 1-O-1Ј-(Z)-octadecenyl-2-arachidonoyl-sn-glycero-3-phosphocholine (1.25 mM) with or without nonoxidized CL (10 mol %) or previously oxidized CL (10 mol %) in 10 mM potassium phosphate buffer, pH 7.0, containing 0.25 mM DTPA and 48 mM sucrose through a 0.8-m filter. LUVs (0.2 ml) were diluted 1:5 in 10 mM potassium phosphate, pH 7.0, containing 0.25 mM DTPA and 25 mM KCl prior to addition of purified cyt c (2.5 M final concentration). Following incubation for 5 min at room temperature, the samples were then centrifuged at 100,000 ϫ g at 15°C for 1 h. The supernatants were removed, and the pelleted LUVs were resuspended in 0.5 ml of buffer prior to analysis of cyt c content by SDS-PAGE and immunoblot analysis as described above.

Oxidation of cardiolipin by cytochrome c and hydrogen peroxide
Bovine heart cardiolipin was oxidized by cyt c as described previously (87) with minor modifications. Briefly, CL SUVs (250 M) in 10 mM potassium phosphate buffer, pH 7.0, containing 0.25 mM DTPA were incubated with cyt c and hydrogen peroxide at 37°C for 40 min. Cytochrome c (10 nmol) and H 2 O 2 (50 nmol) were added at 10-min intervals to achieve 50 and 250 M final concentrations, respectively, at the end of the incubation. Following addition of KCl to 100 mM, oxidized CL was extracted into chloroform following the Folch extraction method (82).