The phospholipase A2 activity of peroxiredoxin 6[S]

Peroxiredoxin 6 (Prdx6) is a Ca2+-independent intracellular phospholipase A2 (called aiPLA2) that is localized to cytosol, lysosomes, and lysosomal-related organelles. Activity is minimal at cytosolic pH but is increased significantly with enzyme phosphorylation, at acidic pH, and in the presence of oxidized phospholipid substrate; maximal activity with phosphorylated aiPLA2 is ∼2 µmol/min/mg protein. Prdx6 is a “moonlighting” protein that also expresses glutathione peroxidase and lysophosphatidylcholine acyl transferase activities. The catalytic site for aiPLA2 activity is an S32-H26-D140 triad; S32-H26 is also the phospholipid binding site. Activity is inhibited by a serine “protease” inhibitor (diethyl p-nitrophenyl phosphate), an analog of the PLA2 transition state [1-hexadecyl-3-(trifluoroethyl)-sn-glycero-2-phosphomethanol (MJ33)], and by two naturally occurring proteins (surfactant protein A and p67phox), but not by bromoenol lactone. aiPLA2 activity has important physiological roles in the turnover (synthesis and degradation) of lung surfactant phospholipids, in the repair of peroxidized cell membranes, and in the activation of NADPH oxidase type 2 (NOX2). The enzyme has been implicated in acute lung injury, carcinogenesis, neurodegenerative diseases, diabetes, male infertility, and sundry other conditions, although its specific roles have not been well defined. Protein mutations and animal models are now available to further investigate the roles of Prdx6-aiPLA2 activity in normal and pathological physiology.

article will review the unique PLA 2 activity of Prdx6 and its roles in normal and pathologic physiology.

PRDX6 AS A PLA 2 A moonlighting protein
The presence of a novel PLA 2 activity with an important role for lung phospholipid metabolism and with characteristics subsequently associated with Prdx6 was discovered as a result of experiments that used isolated lungs and epithelial cells in culture (23,24). The protein was identified and subsequently cloned through use of a novel inhibitor of PLA 2 activity and assay at acidic pH (25). Although the originally published molecular sequence for Prdx6 was based on its PLA 2 activity (25)(26)(27), it shortly thereafter became apparent that the protein had homology to the peroxiredoxin family, and, like other family members, it also expressed peroxidase activity (4,28). Based on that finding, several subsequent articles reviewing PLA 2 enzymes regarded Prdx6 as a peroxidase and, therefore, not a PLA 2 (12,13). However, it is now well recognized that many proteins can perform more than one function, and, although their trivial names may reflect only one activity (for example, peroxiredoxin implies a peroxidase), the "unnamed" second function may be of equal importance (29)(30)(31). This multifunctional activity has led to the appellation "moonlighting" protein. Dual functionality for a protein may include one enzymatic activity and a structural function, as shown by the crystallins, for example (32), but also might include two distinct enzymatic activities such as cytochrome c that functions in the TCA cycle as well as in the pathway to cellular apoptosis (33). While Prdx6 is called a peroxiredoxin, it has both PLA 2 as well as GSH peroxidase (GPx) activities. Interestingly, Prdx6 also expresses a third enzymatic activity, LPCAT, that is coupled to its PLA 2 activity (10), and perhaps a fourth (nonenzymatic) function, namely, a molecular chaperone activity as recently described for a Prdx6 homolog in cyanobacterium (34).
Evidence that Prdx6 is a PLA 2 There are several lines of evidence confirming the PLA 2 activity of Prdx6. First, activity of recombinant Prdx6 is demonstrated by the liberation of a FFA from the sn-2 position of phospholipids (2,4,5,27,(35)(36)(37)(38). Although a prior review of PLA 2 enzymes characterized the Prdx6-PLA 2 activity as "trace" (13), measured rates of activity range from 100 to 130 nmol/min/mg for unmodified Prdx6 to almost 2 µmol/min/mg protein for Prdx6 that has been phosphorylated (35,36,39). While this activity is considerably lower than the activities expressed by snake venoms and other secreted PLA 2 enzymes, it is similar in magnitude to that shown by other intracellular PLA 2 enzymesfor example, human platelet PLA 2 (40), Ca 2+ -independent PLA 2 from a macrophage cell line (41), lysosomal PLA 2 (LPLA 2 ) (42), and Photobacterium damsclae PLA 2 (43).
As a fourth line of evidence, PLA 2 activity, measured under conditions that are relatively specific for Prdx6 (acidic pH, absence of Ca 2+ , and inhibition by MJ33), has been demonstrated in lung tissue and cellular homogenates. The PLA 2 activity in freshly isolated lung homogenate was 0.15 nmol/min/mg protein and, in alveolar type 2 epithelial cells that were stimulated with a phorbol ester, was 0.7 nmol/min/mg protein (24,52), a value 7-fold greater than the activity of Ca 2+ -independent PLA 2 from U937 cells stimulated by concanavalin A (53). This activity is markedly decreased in homogenates of lungs and cells from Prdx6 null mice, confirming its association with Prdx6 (9).
The fifth line of evidence for Prdx6-PLA 2 activity is the physiologic effects that are associated with genetic manipulation of Prdx6-PLA 2 expression in mice (9,38). These effects, discussed in detail below, include alterations of lung phospholipid turnover, altered cell membrane repair in oxidative stress, and failure of NADPH oxidase type 2 (NOX2) activation. Specificity for the role of the PLA 2 activity of Prdx6 in the altered physiology has been confirmed by specific mutation of the PLA 2 active site in Prdx6 (2,10,(54)(55)(56).

Tissue and cellular/subcellular distribution of Prdx6
Prdx6 is a widely distributed protein that is expressed throughout the plant and animal kingdoms. In mammals, Prdx6 is expressed in virtually all organs, with the highest expression levels in lungs, brain, kidneys, and testes (26,57). At the cellular level, the highest expression within the lung is in epithelium and alveolar macrophages with lower, but significant, expression in microvascular endothelial cells (26,58). Within cells, the protein is largely cytosolic, but it is also expressed in acidic organelles [lysosomes and lysosomal-related organelles (LROs)] (27,38,(59)(60)(61), as shown in Fig. 1. Prdx-PLA 2 specific activity is nearly 3-fold greater in lung LROs [lamellar bodies (LBs)] as compared with lung homogenate (38). The cellular and subcellular localization of Prdx6 is important for understanding the physiologic role of its PLA 2 activity. Prdx6 targeting to acidic organelles in lung epithelial cells is independent of lipid binding but requires a peptide sequence (amino acids [31][32][33][34][35][36][37][38][39][40] for binding to the 14-3-3 protein as a chaperone; the chaperone function of 14-3-3 requires its phosphorylation by MAPK (59,62,63). Mutation of Ser32 in the targeting sequence abolishes Prdx6 binding to 14-3-3 and its targeting to LB (62).

Structural basis for multiple enzymatic activities of Prdx6
What special properties of Prdx6 allow it to express two essentially unrelated (peroxidase and PLA 2 ) activities? We have proposed, based on the published X-ray crystal structures of the oxidized and reduced protein [Protein Data Bank (PDB) ID codes 1PRX and 2V2G] (64,65), that the mode for binding of substrate to Prdx6 results in its positioning for either reduction at the peroxidatic site (amino acids surrounding C47 in Prdx6) or for hydrolysis by the PLA 2 catalytic triad (S32-H26-D140) as shown in Fig. 2 (6, 7). The peroxidatic site is at the base of a shallow pocket, while the catalytic triad for hydrolysis is at the protein surface. The distance between Ser32 in the catalytic triad and the peroxidatic Cys47 in the pocket is 28 Å, and there are a number of positively charged residues forming a groove between the two active sites (65). The key feature of the protein from an evolutionary standpoint is the development of a structure that allows binding of the lipid with access to these two distinct enzymatically active sites in the protein. Substrate binding by Prdx6 is discussed further below.
A major difference between the two LPLA 2 enzymes (aiPLA 2 and LPLA 2 ) is that Prdx6, although located in lysosomes and LROs, also is present in the cytosol; actually, the percentage of cellular aiPLA 2 content is greater in cytosol than in acidic organelles, although its concentration is greater in the latter (24,27). Furthermore, its activity, while maximal at acidic pH under basal conditions, is also present at neutral pH with oxidized phospholipid as the substrate or with reduced substrate following phosphorylation of the enzyme (see below). Thus, cytosolic Prdx6 has PLA 2 activity toward cell membranes and can function as a PLA 2 in both cytosolic (neutral pH) as well as lysosomal (acidic pH) compartments. [Although LPLA 2 can hydrolyze oxidized truncated phosphatidylcholine (PC), but not PC, at neutral pH, this activity is much less than activity at pH 4 (42).] The appellation aiPLA 2 , then, does not fully describe the enzymatic activity of the protein, although it does represent a convenient means to refer to the unique PLA 2 activity associated with Prdx6. Thus, aiPLA 2 represents a "hybrid" with characteristics of both intracellular Ca 2+ -independent and LPLA 2 types, as well as some characteristics not shared by either the type 4 or 5 enzymes. For now, ai-PLA 2 might reasonably be classified with the type 4 intracellular Ca 2+ -independent PLA 2 enzymes pending future insights leading to a more rigorous reclassification of this large and diverse group of proteins. An important caveat related to this classification is that aiPLA 2 activity is not sensitive to bromoenol lactone (BEL) (24,27,54), an inhibitor of other type 4 PLA 2 enzymes. We have routinely assayed the PLA 2 activity of Prdx6 based on the release of radiolabeled palmitate (either 9,10-3 H-palmitate or 1-14 C-palmitate) from the sn-2 position of dipalmitoyl phosphatidylcholine (DPPC) (23,24,27,50,54). In our assay, the substrate is presented as unilamellar liposomes containing labeled DPPC, egg PC, cholesterol, and phosphatidylglycerol (PG) in molar ratio 50:25:15:10, but use of a simpler liposome formulation (e.g., omission of egg PC) certainly would be reasonable. Incubation is generally for 1 h followed by TLC to separate the phospholipid species; the DPPC spot is scraped from the plate and radioactivity is analyzed by scintillation counting. The native protein shows limited PLA 2 activity when assayed at pH 7-8 (activity 0-50 nmol/min/mg protein), but considerably greater activity when assayed at pH 4 (100-130 nmol/min/ mg protein) (54,60). Although several "kits" are commercially available for assay of various PLA 2 activities, there is no kit available, to our knowledge, with demonstrated specificity for aiPLA 2 . We have used a fluorescence assay with bis-BODIPY-C 11 -PC as the fluorophore substituting for radiolabeled PC (25,27), but key issues have been stability of the substrate under the acidic conditions required for optimal measurement of activity for the nonphosphorylated protein and the high background fluorescence for the manufactured substrate.

Substrate specificity
aiPLA 2 activity is greatest with PC as substrate; activity is decreased 40% with the substitution of phosphatidylethanolamine, decreased another 60% with phosphatidylglycerol, and further decreased with other head groups (inositol and serine) as the PL substrate (27). The K m for the aiPLA 2 reaction with PC as substrate is 350 µM (27). Activity of aiPLA 2 toward PC shows no preference for the sn-2 acyl group, but is decreased substantially (60%) with substitution of an alkyl linkage; however, there is no activity with 1-O-hexadecyl-2-acetyl-glycerophosphocholine (platelet activating factor), i.e., no PAFAH activity (27). The protein also does not express PLA 1 or lysophospholipase activities, but does express LPCAT activity (see below). Activities of recombinant protein based on the human or rat amino acid sequence or of protein isolated from either rat or bovine lungs are essentially similar (4,27).

Effect of the oxidation of substrate or enzyme
Oxidized phospholipid [oxidized sn-2 linoleic or arachidonic acid (AA)] is a substrate for aiPLA 2 (8,27). The activity at acidic pH is similar for reduced and oxidized substrate. However, in contrast to the reaction with reduced substrate, where activity is significantly decreased at pH 7 as compared with pH 4, PLA 2 activity with oxidized substrate is similar at acidic and neutral pH (7,27,56). This gain of aiPLA 2 activity at cytosolic pH with oxidized lipids as substrate is important for understanding the role of this enzyme in the repair of peroxidized cell membranes (Fig. 3).
Similar to this effect of oxidized substrate, PLA 2 activity of Prdx6 at neutral pH is increased by oxidation of the protein (e.g., by treatment with H 2 O 2 ), while activity at pH 4 is unaffected (unpublished observations from our laboratory and ref. 78). A possible (but untested) mechanism for this effect is that formation of the sulfinic acid intermediate associated with Prdx6 oxidation facilitates its binding to substrate. That is, the interactions of reduced Prdx6 with oxidized substrate or oxidized Prdx6 with reduced substrate both enhance protein-substrate binding and thereby enhance enzymatic activity.
The addition of GSH to the assay mixture in vitro also has a positive influence on the measured aiPLA 2 activity at neutral pH (60). This effect is specific for GSH because addition of GSSG or several other (nonreducing) sulfhydryls had no effect (unpublished observations). Although not known precisely, we propose that the mechanism for this effect of GSH might be (paradoxically) oxidation of Prdx6 protein. Recombinant Prdx6 can be stabilized during storage by spontaneous formation of a sulfenylamide (5), presumably by interaction of the sulfenic intermediate at C47 with an adjacent amino acid in the 3D structure; analysis of the crystal structure of the sulfinic acid form of Prdx6 suggests that the interacting amino acid may be His39 (65). Sulfenylamide formation may protect the protein against irreversible oxidation. By mass spectroscopy, this amide linkage is "broken" by addition of GSH, allowing autooxidization of the protein to the sulfinic form (unpublished observations). Thus, oxidation of the protein by treatment either with an oxidant (H 2 O 2 ) or with GSH can result in increased aiPLA 2 activity in vitro.

Prdx6 phosphorylation
While increased PLA 2 activity is associated with oxidation of either the substrate or the protein, phosphorylation of Prdx6 (at T177) has a much greater effect on activity. PLA 2 activity of phosphorylated Prdx6 was 1.3-1.7 µmol/min/ mg protein at both acidic and neutral pH (36,39,79); these results were generated with protein that was less than fully phosphorylated so that the PLA 2 activity of phosphorylated Prdx6 could be 2 µmol/min/mg protein. Thus, phosphorylation of Prdx6 increases its PLA 2 activity by 15to 20-fold at pH 4 and (because of the lower baseline) to a much greater extent at pH 7 (35,36,39,79). Physiologically, phosphorylation of cytosolic Prdx6 can occur through MAPK (ERK or p38) activity (54). The increase in PLA 2 activity with phosphorylation reflects a conformational change in the protein as determined by tryptophan fluorescence and NaI fluorescence quenching (39). We have postulated that phosphorylation converts the protein to a less rigid molten globular state, allowing the conformational change (39).

Substrate binding and aiPLA 2 activity
Physiologic activation of aiPLA 2 activity requires the binding of Prdx6 to its substrate as an initial event, and analysis of the binding of this enzyme to substrate provides the basis for understanding PLA 2 activity under various intracellular conditions (7,80). Binding of the enzyme to the surface of the phospholipid vesicle (the substrate) is the first step for enzymatic activity, and conditions that promote binding are reflected in increased enzymatic activity (81). The isoelectric point for Prdx6, calculated as well as measured, is 5.7-6.0 (82,83), indicating that Prdx6 has a positive bulk charge under acidic conditions. Thus, Prdx6 binding to liposomal substrate is significantly increased by the negative charge imparted to liposomes with the inclusion of PG or phosphatidylserine in the lipid mixture (6). By ultrafiltration analysis, significant binding of protein to the negatively charged liposomes was demonstrated following incubation at pH 4, but binding was reduced markedly with incubation at pH 7 (7). Studies with protein mutagenesis have shown that the surface amino acid sequence H26-W31-S32 of Prdx6 is the site for phospholipid binding to Prdx6 (6). The increased binding at acidic pH presumably reflects protonation of a surface amino acid, most likely His26.
A recent crystallographic study (PDB ID code 2V2G) indicated that the distance between the H26 and D140 residues in Prdx6 is relatively great (9.7 Å) and suggested that a phospholipid substrate actually may bind to the flat surface across the PLA 2 active site and peroxidatic active site of the other monomer in the normal dimeric configuration of the protein (65). However, the experimental demonstration of PLA 2 activity in the Prdx6 monomer indicates that dimerization of the protein is not necessary for activity (5). Furthermore, zero-length chemical cross-linking and homology modeling studies have shown that several regions of reduced human Prdx6 are in a substantially different conformation from that shown for the crystal structure of the peroxidase catalytic intermediate (84). Study of the catalytic triad in some other proteins also has shown considerable mobility of the active-site Ser residue (85). In support of plasticity of the Prdx6 protein, considerable conformational change has been demonstrated with its phosphorylation (39). Additional crystallographic studies of native Prdx6 that is bound to substrate under acidic conditions or to phosphorylated protein would help to clarify the binding paradigm.
While the binding of Prdx6 to phospholipid under acidic conditions is clear, a major fraction of intracellular Prdx6 is localized to the cytosol (25,27,59), where the neutral pH supports neither aiPLA 2 binding to membranes nor its enzymatic activity under "resting" conditions. However, the cytosolic protein does bind to oxidized substrate at pH 7 so that oxidation of cell membrane phospholipids leads to their binding to Prdx6 and subsequent PLA 2 activity (6, 7).
Phosphorylation of Prdx6 also promotes its binding to substrate at pH 7 and allows physiological activity of cytosolic protein (39). The constant (K D ) for Prdx6 binding to substrate, estimated from fluorescence measurements, is 5.6 µM for phosphorylated protein compared with 24.9 µM for the nonphosphorylated WT (39). Phosphorylation changes the conformation of Prdx6 from its native state, resulting in the exposure of hydrophobic residues; the change in conformation results in a significant increase in PLA 2 activity at both neutral and acidic pH (39).

Inhibitors of aiPLA 2 activity
aiPLA 2 activity is insensitive to most agents that are commonly used to inhibit other PLA 2 enzymes, such as p-bromophenacyl bromide, arachidonoyl trifluoromethyl ketone, and BEL (25,27,38,54,56,79). However, several effective inhibitors of aiPLA 2 activity have been identified. Because aiPLA 2 activity is serine-dependent, the enzyme is inhibited nonspecifically by the serine "protease" inhibitor, diethyl p-nitrophenyl phosphate (DENP) (4,27). Another potent and more specific inhibitor of aiPLA 2 activity is MJ33 (usually as the Li + salt) (9, 23-25, 52, 86). This competitive inhibitor is an analog of the phospholipid tetrahedryl transition state that specifically interferes with the catalytic turnover by PLA 2 enzymes that are bound to the enzyme-substrate interface (51). Inhibition of aiPLA 2 activity is >90% at an inhibitor to substrate concentration of 1 mol% (24,52). MJ33 also inhibits pancreatic PLA 2 and several other PLA 2 enzymes, although the affinity of the inhibitor for these other enzymes appears to be less than for Prdx6 (51).
We also have identified two endogenous proteins that inhibit aiPLA 2 activity and might play a role in its physiologic regulation. The first such inhibitor is lung surfactant protein A (SP-A), a multimeric protein (26 kDa monomer) that is synthesized and secreted by lung epithelial cells as a component of lung surfactant (87). SP-A was shown to inhibit a PLA 2 that is secreted in the venom of the Habu snake (Trimeresurus viridis) (88) and subsequently was shown to bind to Prdx6 and inhibit its PLA 2 activity (60,61,79). The K i for inhibition of aiPLA 2 activity is 10 µg/ml (0.4 µM) (60). SP-A and Prdx6 are both present in lung LBs, and the inactivation of LB SP-A resulted in increased LB aiPLA 2 activity (61). This "unlocking" of Prdx6-PLA 2 activity provides evidence that SP-A can regulate aiPLA 2 activity in the intact cell. We have identified the amino acid sequences of Prdx6 and SP-A (10 and 16 amino acid sequences, respectively) that are responsible for interaction of the two proteins and have shown that a 16 amino acid SP-A peptide mimics the effect of the whole SP-A protein on aiPLA 2 activity (79). Our more recent studies have demonstrated effective inhibition by a 9 amino acid sequence derived from the 16 amino acid peptide (unpublished observations).
A second endogenous protein that binds to Prdx6 and inhibits its aiPLA 2 activity is p67 phox , a component of the NOX2 activation cascade (35,37). p67 phox binds strongly to phosphorylated Prdx6, but weakly to the nonphosphorylated protein (35). Phosphorylated p67 phox , the form of the protein required for activation of NOX2, does not bind to Prdx6 (35). The interaction of phosphorylated Prdx6 with p67 phox may play a role in the regulation of NOX2 activation (see below).
Of note, none of these four inhibitors of aiPLA 2 activity (DENP, MJ33, SP-A, and p67 phox ) have an effect on the peroxidase activity of Prdx6, clearly showing that these two activities of Prdx6 are independent.

Mutations of Prdx6 for the study of its PLA 2 activity
Several mutations of Prdx6 have been shown to be useful for the evaluation of its enzymatic and binding activities: 1) S32A or H26A mutations prevent binding of the protein to phospholipids, thereby abolishing PLA 2 (and phospholipid hydroperoxide peroxidase) activities; peroxidase activity with small peroxides such as H 2 O 2 is unaffected (6,10,55). 2) S32T mutation has no effect on Prdx6 binding to phospholipids or its enzymatic activities (Thr can substitute for Ser), but binding to 14-3-3 is abolished, resulting in the failure of Prdx6 trafficking to acidic organelles (62).
3) Mutation of the D140A component of the catalytic triad abolishes PLA 2 activity; there is no effect of this mutation on phospholipid binding or on other Prdx6-related activities (6,56). 4) C47S mutation results in the total loss of peroxidase activity (phospholipid hydroperoxides and H 2 O 2 ) but no effect on PLA 2 activity (4, 10, 56, 58). 5) T177 mutation to Ala or Glu prevents Prdx6 phosphorylation and the subsequent increase in aiPLA 2 activity; mutation to Glu results in increased basal aiPLA 2 activity (36). 6) D31A mutation abolishes LPCAT activity, but there is no effect on aiPLA 2 or peroxidase activities (10). 7) The double mutation L145/L148 prevents dimerization and results in loss of peroxidase activity, but PLA 2 activity is intact; a single mutation of either Leu has an intermediate effect (5).
Summary of the reported mutations for study of specific physiological effects of Prdx6: C47S eliminates all peroxidase activities; D140A eliminates PLA 2 activity; D31A eliminates LPCAT activity; T177A prevents protein phosphorylation; L145A/L148A prevents protein dimerization; S32A prevents lipid binding; and S32T allows lipid binding but prevents the transport of Prdx6 to acidic organelles. It is important to emphasize that the frequently used S32A mutation of Prdx6 does abolish its PLA 2 activity, but, through its effects on lipid binding, also abolishes PHGPx activity; this mutated protein expresses peroxidase activity for H 2 O 2 and other small hydroperoxides, but has no activity toward the perhaps more important phospholipid hydroperoxide substrate. Thus, the loss of a physiological effect in the presence of S32A-Prdx6 does not unequivocally indicate a role for aiPLA 2 activity without additional evidence-e.g., inhibition of the effect with MJ33.

Regulation of lung phospholipid turnover
Lung surfactant, a complex lipid-protein product that is essential for mammalian life, promotes a low surface tension at the air-liquid interface of the lung alveoli, thereby minimizing the "work" of breathing and facilitating uniform lung ventilation (90). PC is the major biochemical component of the lung surfactant, comprising approximately 75% of the secreted product; approximately 2/3 of PC is the dipalmitoyl form (DPPC) that functions as the primary surface-active material (52,91). Lung surfactant has a complex life cycle, as it is synthesized by alveolar granular pneumocytes (epithelial type 2 cells), stored in the LBs of the pneumocytes, secreted into the alveolar space by exocytosis, taken up by the same cells through receptormediated endocytosis, and reprocessed in LB for degradation or resecretion (52,(92)(93)(94). The measured half-time for DPPC in the alveolar space has been estimated, based on the label in palmitate, at approximately 10 h for small animal models and at 28 h for primates (91,92). Extracellular (alveolar) DPPC that is recycled to LBs is degraded and/or remodeled prior to resecretion (92). LB are LROs that maintain an acidic pH (approximately pH 5) (95) and thus can support the PLA 2 activity of Prdx6. The physiological "purpose" of this recycling pathway is not understood but is related most likely to quality and/or quantity control of the lung surfactant material.
Prdx6-PLA 2 activity is involved in both the degradation and the remodeling of surfactant phospholipids (9,23,24,38,50,52). Degradation of DPPC involves PLA 2 activity as an initial step with subsequent metabolism of the liberated palmitate and lysoPC fractions. Remodeling is accomplished through PLA 2 activity to generate lysoPC, followed by reacylation with an acylCoA, catalyzed by LPCAT, to regenerate PC; the remodeling reactions represent the classical Lands cycle for PC synthesis (94). Because Prdx6 expresses both PLA 2 and LPCAT activities, it is, therefore, a complete remodeling enzyme.
DPPC in lung surfactant also can be catabolized by alveolar macrophages (96). This pathway represents a scavenging pathway for removal of "damaged" or excessive phospholipids and is not part of the elegant surfactant recycling mechanism. Uptake of phospholipids by macrophages for degradation is through phagocytosis rather than the receptor-mediated pathway for recycling. As Prdx6 is expressed at a significant level in alveolar macrophages (54), aiPLA 2 could play a role in this degradative (scavenging) pathway, although that has not yet been studied. LPLA 2 also is expressed in alveolar macrophages, and this, rather than aiPLA 2 , may be the major macrophage-related phospholipid degradative enzyme (97,98).
The phenotype of Prdx6 null lungs demonstrates the role of aiPLA 2 activity in phospholipid homeostasis (9). In mice aged 9-11 weeks, the PC content of Prdx6 null compared with WT control was increased by 73% in the lung, 29% in the lung LBs, and 58% in lung broncho-alveolar lavage fluid (BALf). Unlike lungs from WT mice where the PC content normalized to lung protein or body weight remains relatively constant with time, lungs from Prdx6 null mice show a linear increase in lung PC, lung DPPC, and lung total phospholipid; at 1 year of age, PC in null mice compared with control is greater by 200% in the lung and 180% in the BALf. In experiments to evaluate the metabolism of lipids, lungs from WT mice degraded 54% of internalized DPPC over a 2 h observation period, while lungs from Prdx6 null mice degraded only 5%. To study PL synthesis, mice were injected with radiolabeled palmitate into the tail vein; when studied 24 h later, there was 73% less radioactivity in LB PL from null lungs as compared with WT (9).
WT lung alveolar type II epithelial cells in primary culture showed a marked decrease in PC synthesis following their treatment with MJ33, similar to that seen in Prdx6 null lungs (24). Inhibition of aiPLA 2 activity prevented phospholipid remodeling with mechanical stress with the lung cell line A549 in culture (99). For the opposite effect, transgenic Prdx6 overexpressing mice showed a 17-25% decrease in PC content of lungs, LBs, and BALf and a 36% increase in the degradation of radiolabeled DPPC that had been administered to lungs and internalized by lung cells (38). Thus, Prdx6 plays an important role in the turnover of lung surfactant phospholipids.

Antioxidant defense and cell membrane repair
Exposure of cells to oxidants [termed reactive oxygen species (ROS)] results from normal metabolism as well as in association with toxic exposures. Organisms express a variety of antioxidant enzymes, including several peroxidases, to counter oxidant "stress." Thus, the GPx activity of Prdx6 as a scavenger of H 2 O 2 and hydroperoxides has received attention as an antioxidant enzyme (100,101). However, the PLA 2 activity of Prdx6 also plays a major role in antioxidant defense related to its role in the "repair" of peroxidized cell membranes (2,55,56,58). Because the repair of peroxidized cell membrane phospholipids represents the reversal of the effects of oxidative stress, it can be considered as an antioxidant activity.
An important role for PLA 2 activity in the repair of peroxidized cell membranes was suggested some time ago (102,103), but the specific PLA 2 involved was not known and identification of aiPLA 2 as specific for these reactions is relatively recent. WT lungs and endothelial cells show repair of peroxidized cell membrane phospholipids over several hours following removal of the oxidant (Fig. 4); however, Prdx6 null lungs that express neither the peroxidase nor the PLA 2 activities of Prdx6 fail to show significant cell membrane repair following a lipid peroxidation event (7,56,58). Based on the high rate constant (K 2 > 10 6 /M/s) for the phospholipid hydroperoxide reductase (19), we had assumed based on theoretical calculations (104) that this reaction would predominate in peroxidized cell membrane repair. However, experimentally, that has proved not to be the case, and specific mutations of Prdx6 that abolish either the peroxidase or the PLA 2 activity of the protein have indicated that the aiPLA 2 activity of Prdx6 can contribute approximately equally to the PHGPX activity for the maximal rate of cell membrane repair (55,56,58).
While the peroxidase activity of Prdx6 directly reduces a peroxidized membrane phospholipid (to its corresponding alcohol), the Prdx6-PLA 2 activity liberates the oxidized fatty acid by the hydrolysis of the sn-2 acyl bond; the lysoPC that is generated by this reaction can be reacylated through the LPCAT activity of Prdx6 (10). Based on kinetic studies, the sequential deacylation and reacylation reactions for Prdx6 occur without release of lysoPC from attachment to the enzyme (10). Thus, these linked activities of Prdx6 (PLA 2 , LPCAT) can replace the oxidized fatty acid of a phospholipid with a nonoxidized one, thereby regenerating the reduced phospholipid.
This repair function linked to Prdx6-PLA 2 activity has been demonstrated for acute oxidative stress, but it is possible that the membrane repair function of Prdx6 also plays an important physiological role under chronic conditions. For example, decreased reproductive efficacy has been demonstrated in aging mice and is accelerated by the absence of Prdx6-PLA 2 activity (105, 106) (see below). This finding may reflect a role for Prdx6 in protection against the chronic effects of cell membrane lipid peroxidation.

Cell signaling: activation of NOX
NOX2 is a multicomponent enzyme that is expressed by professional phagocytes (polymorphonuclear leukocytes, or PMN, macrophages), microvascular endothelium, and other cell types. This enzyme complex has the primary function to generate the superoxide anion that plays an important role in bacterial killing as well as in normal cell signaling related to cell proliferation, cell migration, immune function, aging, stem cell self-renewal, and other homeostatic processes (107). NOX2 itself is activated by a signaling pathway that has long been proposed to require PLA 2 activity, although neither the specific PLA 2 nor the relevant product were known (108). Initially, the enzyme called group IV (cytosolic) PLA 2 was considered for this role (109), but this is unlikely based on studies with cPLA 2 null mice (110). More recent studies have demonstrated that Prdx6 PLA 2 activity is required for NOX2 activation, at least in PMN, macrophages, and lung microvascular endothelial cells (54,111). Following an activating signal, Prdx6 is phosphorylated and "translocates" to the plasma membrane (54); this translocation reflects the marked increase in the ability of Prdx6 following its phosphorylation to bind to phospholipids in membranes (7). Inhibition or "knock out" of Prdx6-PLA 2 activity essentially abolishes NOX2 activation in response to activating stimuli (54). The mechanism for the role of Prdx6 in NOX2 activation is through aiPLA 2 -mediated generation of lysoPC that, in turn, is converted to lysophosphatidic acid (LPA); LPA reacts with a cell membrane receptor (LPA receptor 1), leading to liberation of Rac (112), a crucial component of the NOX2 activation pathway (113).
As described above, p67 phox binds to phosphorylated Prdx6 and inhibits its PLA 2 activity (35), or, conversely and not yet tested, binding of phosphorylated Prdx6 to p67 phox may inhibit the phosphorylation of the latter protein. Further study is required to determine whether this interaction constitutes a physiological mechanism for regulating NOX2 activity.
An example of the physiological role of activated NOX2 is the response of pulmonary microvascular endothelium to altered mechanical stress; these cells in culture on going from a flow to no-flow condition (simulated ischemia) show a brisk aiPLA 2 -dependent generation of ROS that has been associated with endothelial cell growth, cell proliferation, and vascular remodeling, among other effects (114). Activation of NOX2 by aiPLA 2 activity also has a major role in the inactivation of a G-protein-coupled opioid receptor (115).
In addition to NOX2, Prdx6 has been shown to facilitate the activation of NADPH oxidase type 1 (NOX1) (116). Prdx6 can bind to and stabilize the expression level of NOXa1, a homolog of p67 phox and a cytosolic activating protein for NOX1. Activation of NOX1 was not seen in the presence of S32A mutant Prdx6, reflecting either the loss of aiPLA 2 activity or the failure of lipid binding by the Fig. 4. Role of the PLA 2 activity of Prdx6 in the recovery from lipid peroxidation by the isolated perfused mouse lung. Isolated mouse lungs were perfused with tert-BOOH for 1 h and then with oxidantfree perfusate for the subsequent 2 h. C47S-Prdx6 and D140A-Prdx6 were "knock-in" mice that express only the PLA 2 or only the peroxidase activity of Prdx6, respectively. Membrane lipid peroxidation was measured by the fluorescence of 1,3-bis(diphenylphosphino) propane (DPPP). Full recovery from lipid peroxidation was seen between 2 and 3 h after removal of the oxidant. No recovery from lipid peroxidation was seen in the Prdx6 null mouse, while intermediate recovery was seen in the absence of either the PLA 2 or the peroxidase activity. Thus, both activities of Prdx6 are required for a maximal rate of recovery from lipid peroxidation. Reproduced from ref. 56 with permission. mutant protein. A possible role for Prdx6 phosphorylation in the binding and stabilization of NOXa1 was not determined. The role of aiPLA 2 activity to stabilize the expression levels of the cofactor was suggested by MJ33-mediated inhibition of the effect (116); however, the mechanism for this effect of aiPLA 2 activity on NOXa1 is unclear. Because Rac appears to be an important cofactor for NOX1 activation (117), a possible mechanism for the involvement of Prdx6 in NOX1 activation is the generation of active Rac through aiPLA 2 -generated lysoPC, as for NOX2. That possibility has not been tested.
The role of aiPLA 2 in the activation of NOXs indicates that Prdx6 can function either as a prooxidant or an antioxidant enzyme. Prdx6 does indeed appear to serve both functions. On the one hand, Prdx6 is crucial for the activation of NOX2, a major physiological source of the oxidant superoxide anion that plays an essential role in cell signaling; on the other hand, the peroxidase activity of Prdx6 can scavenge ROS (H 2 O 2 ), while its ability to reverse membrane oxidation through its peroxidase as well as PLA 2 activities can prevent the consequences of oxygen radical overproduction. Thus, Prdx6 can be regarded as an enzyme to "normalize" the intracellular content of ROS; that is, it is involved in the generation of ROS for cellular signaling, but also in decreasing excessive cellular H 2 O 2 levels and in protection against the side effects of ROS overproduction.

aiPLA 2 IN PATHOPHYSIOLOGY
A role for Prdx6 has been described for a wide range of both human pathologies as well as for animal models of disease. For the most part, evidence for the role of Prdx6 is based on the finding of an increased or a decreased expression level for the protein, occurring either as a result of the disease or through genetic manipulation. In some instances, an altered Prdx6 expression level may merely reflect the role of inflammation that leads to induction of Prdx6 along with other antioxidant enzymes. Thus, there currently are relatively few well-documented examples where expression of Prdx6 plays a specific role in the pathophysiology of disease, and fewer still where there is an apparent role specifically for Prdx6-PLA 2 activity.

Acute lung injury and sepsis
Acute lung injury (ALI; or acute respiratory distress syndrome) represents either a final common pathway for many etiologies that primarily affect the lung or a secondary response to systemic diseases (118). Pneumonia, either bacterial or viral, is the most common primary lung disease resulting in this syndrome, while sepsis is a common cause of secondary ALI. Hallmarks of ALI include lung inflammation associated with an influx of PMN and increased generation of ROS, in large part through activation of the enzyme NOX2 in PMN, alveolar and tissue macrophages, lung endothelium, and possibly other cell types (118). Excessive ROS production can lead to tissue injury through oxidation of cellular proteins and lipid peroxidation of cell membranes.
Prdx6-PLA 2 activity can play two major roles in ALI. First, through its role in the activation of NOX2, aiPLA 2 can contribute to the pathophysiology of lung damage in ALI (54). Second, through its effects on cell membrane phospholipids, aiPLA 2 participates in the repair of oxidative injury (56). Although the inhibition of aiPLA 2 activity may reduce the ability of Prdx6 to repair peroxidized cell membranes, repair through reduction of phospholipid hydroperoxides, i.e., the PHGPx activity of Prdx6, would be unaffected. Thus, inhibition of aiPLA 2 activity by administration of MJ33 led to a marked reduction of lung injury including the extent of lipid peroxidation in mouse models of ALI associated with the intratracheal administration of bacterial lipopolysaccharide, ischemia-reperfusion injury, or exposure to hyperoxia (19,86,119,120).

Cancer and carcinogenesis
Elevated expression levels for Prdx6 have been demonstrated in association with a broad variety of human cancers, including lymphoma and cancer of the breast, esophagus, lung, liver, ovaries, pancreas, bladder, thyroid, gingiva, tongue, skin, and mesothelium, among others (121)(122)(123)(124)(125)(126)(127)(128)(129)(130)(131)(132)(133). Increased expression of Prdx6 correlated with a poorer prognosis for breast cancer and large B-cell lymphoma, progression of bladder cancer, and increased invasiveness of retinoblastomas (124,(133)(134)(135), but correlated with improved survival of patients with hepatocellular carcinoma (123) and some types of pancreatic adenocarcinoma (136). The correlation of Prdx6 expression with poor prognosis for breast cancer was associated with several SNPs of Prdx6, although these SNPs were unrelated to the active site for aiPLA 2 activity (124). This effect of Prdx6 expression was supported by studies with human breast cancer cells in vitro showing increased proliferation and invasiveness associated with Prdx6 overexpression (125). Similarly, overexpression of Prdx6 in lung and gastric cancer cells in vitro led to an increased rate of tumor cell growth, cell migration, and cellular invasion (126)(127)(128)132). These adverse effects of Prdx6 in the gastric cancer cell line were reversed by treatment with a microRNA (miR-24-3p) (128). To briefly summarize these studies of human cancers and cancer cells, the increased expression of Prdx6 in cancers has frequently been associated with increased invasiveness and a poorer prognosis, although the opposite effect also has been reported.
As for human tumors, Prdx6 overexpression increased the growth and invasiveness of established experimental tumors in virally and chemically induced mouse skin carcinoma; however, the presence of Prdx6 inhibited the formation of new tumors (132). Based on the correlation of tumor number with lipid peroxidation in the skin carcinoma model, increased susceptibility to tumor formation could reflect a low-level chronic inflammation (132), although that remains open to further investigation. Assuming that chronic low-level lipid peroxidation does support carcinogenesis, both aiPLA 2 and peroxidase activities of Prdx6 would be involved because both contribute to the repair of peroxidized cell membranes (56). Thus, Prdx6 might have a dual role in carcinogenesis by suppressing the formation of new tumors, but stimulating the growth of existing tumors.
On the other hand, a reasonable mechanism for increased tumor growth and invasiveness associated with Prdx6 overexpression has not been proposed. Prdx6 prevented hyperglycemia-mediated apoptosis in lens epithelial cells (137), and decreased apoptosis in tumors could support an increased rate of tumor growth. However, to the contrary, Prdx6 supported, rather than inhibited, TNFmediated apoptosis in hepatocellular carcinoma cells in vitro and also in primary bronchial epithelial cells and a gastric cell line (123,128,138). This latter effect would tend to dampen rather than increase tumor growth and invasiveness as observed for Prdx6 overexpression. Additional studies are clearly required to resolve these issues.
Several studies have specifically evaluated other roles of Prdx6 enzymatic activities in carcinogenesis. Knockdown of presenilin 2 resulted in increased tumor growth in a mouse model of lung carcinoma; this effect was attributed to a secondary increase in aiPLA 2 activity, although the assay used to determine enzymatic activity was nonspecific (139). The studies described above relating Prdx6 to apoptosis in hepatocellular carcinoma cells indicated that its PLA 2 activity was responsible for the effects (123); however, that conclusion is suspect because it was based on the use of an inhibitor, BEL, that does not inhibit aiPLA 2 activity (26,27,54). In a study of lung cancer cells, mutations S32A and C47S in Prdx6 both abolished the effect of WT Prdx6 on tumor metastasis, while MJ33 treatment had no effect (130); although the authors interpreted these data to show involvement of both PLA 2 and peroxidase activities, a more likely interpretation is that the phospholipid hydroperoxide peroxidase (PHGPx) activity of Prdx6, but not the PLA 2 activity, is responsible for the results. In a study of melanoma cells, the suppression of cell growth by Prdx6 "knockdown" was rescued by transfection with a Prdx6 construct that expresses aiPLA 2 , but not peroxidase activity (129). The proposed mechanism for the activation of melanoma tumor growth was through AA release resulting from aiPLA 2 activity (129); while that mechanism is feasible, recombinant Prdx6-PLA 2 does not show specificity for AA-containing phospholipids (27). Additionally, as described above, mutation of S32, purportedly to confirm the role of PLA 2 activity, also abolishes PHGPx activity, further confounding the results.
An increase in the number and size of urethane-induced adenocarcinomas and the growth of A549 and NCI-H460 human lung cancer cells in mice was attributed to activation of the JAK2/STAT3 pathway, but the deletion of either the GPx activity (C47S-Prdx6) or the aiPLA2 activity (S32A-Prdx6) of Prdx6 had no effect on phosphorylated (activated) JAK2/STAT3 expression in these cells (127,140). Although the authors concluded that either enzymatic activity by itself could support activation of the pathway, this conclusion is suspect because either mutation (C47S or S32A) would have abolished PHGPx activity leaving as remaining activities only the hydrolysis of PC (through aiPLA 2 activity of the C47 mutant) or the reduction of H 2 O 2 or similar small oxidant (through GPx activity of the S32 mutant); it is unclear how either one of these activities alone could activate the JAK2/STAT3 pathway. Other activators proposed for the effect of aiPLA 2 activity on tumor growth include (in different studies) MAP or Src family kinases (127,129,130), urokinase plasminogen activator or its receptor (125,130), or various other signaling proteins (125), although definitive evidence for their possible role in aiPLA 2 activation was not presented.
Based on these various studies, a preliminary conclusion is that aiPLA 2 activity can support tumor growth, invasiveness, and metastasis. These effects of Prdx6 may vary with cell type. However, some studies present contradictory information. Further, the mechanism for these proposed effects of Prdx6 is unclear. Thus, additional and better controlled studies will be required to fully understand the importance and the mechanistic basis for the role of ai-PLA 2 in carcinogenesis.

Diseases of the CNS
Prdx6 overexpression has been demonstrated in a variety of CNS neurodegenerative diseases and related animal models including Parkinson's disease (141), amyotrophic lateral sclerosis (142), Alzheimer's disease (143), and several other types of dementia (141,144,145). Brains from patients with Alzheimer's disease showed an increased number and increased staining intensity of Prdx6-positive astrocytes that were closely involved with diffuse plaques (143). Overexpression of Prdx6 protected against neuropathology in a mouse model of prion disease (146) and served as a biomarker of traumatic brain injury (147,148). Prdx6 expression was markedly increased in the spinal cord of mice with experimental autoimmune encephalomyelitis, as a model of multiple sclerosis and transgenic overexpression of the protein in this experimental model resulted in a significant decrease in clinical severity (149). However, none of these studies specifically evaluated which activity of Prdx6 was responsible for the findings, and increased Prdx6 expression may have merely reflected an ongoing oxidant stress.
Several studies have attempted to evaluate the role of aiPLA 2 activity in mouse models of CNS pathology. Overexpression of Prdx6 was shown to potentiate the neurodegeneration associated with administration of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, a mouse model of Parkinson's disease (150), and the CNS pathology in a mouse model of Alzheimer's disease (151). Although the assays for Prdx6-PLA 2 activity in these studies were not definitive, the administration of MJ33 provided evidence that the adverse effects of Prdx6 in these models might have reflected an increase in the aiPLA 2 activity in specific brain cells.

Diabetes
Prdx6 null mice develop a phenotype similar to early stage human type 2 diabetes manifested by reduced glucosedependent insulin secretion and increased insulin resistance (152). A role for the loss of aiPLA 2 activity in the pathogenesis of these metabolic manifestations might be expected, but has not yet been tested.

Male infertility
The aiPLA 2 activity of Prdx6 may play a role in male fertility. Spermatozoa from Prdx6 null mice have impaired fertilizing ability that is accentuated during aging (105,106), and treatment of sperm from wild-type mice with MJ33 decreased their motility, viability, and fertilization rates (153). These results suggest that the decrease in fertility with aging in mice is related to decreased expression of aiPLA 2 activity. The mechanism for these effects of aiPLA 2 may reflect its role in ROS-mediated signaling that is required for the subsequent phosphorylation events that lead to sperm capacitation (154), although the effects of an increase in lipid peroxidation products also would be possible. As for the relevance of these results to humans, infertility in human males has been associated with decreased expression of Prdx6 (155). Furthermore, treatment of human sperm with MJ33 decreased their capacitation (156). Thus, based on the results obtained with mouse models, aiPLA 2 activity has an important role in the male reproductive system. The role of changing aiPLA 2 expression in the decreased fertility with human aging has not yet been studied.

Other pathophysiology
Altered expression of Prdx6 has been reported for several miscellaneous diseases. Lower Prdx6 expression correlated with a younger age for human cataract surgery while treatment with Prdx6 delayed cataract formation in animal models (157)(158)(159) and protected against cytotoxicity in retinal pericytes (137). Prdx6 content in blood was increased significantly with experimental human hypobaric hypoxia as a model for acute mountain sickness (160). Prdx6 content in blood serum was increased in 50% of patients with Crohn's disease (161). Tissue and plasma levels of Prdx6 have been proposed as a biomarker for human abdominal aortic aneurysm (162). Lung Prdx6 expression was increased in an animal model of silicosis (163). Transfection with Prdx6 in rats ameliorated the experimental skin injury associated with ionizing radiation (164). Despite these investigations (mostly involving proteomic techniques) describing Prdx6 association with disease, a possible etiologic relationship to Prdx6, and specifically to its aiPLA 2 activity, has not been confirmed for any of these conditions.

SUMMARY AND CONCLUSIONS
In summary, Prdx6 is an intracellular Ca 2+ -independent PLA 2 enzyme (called aiPLA 2 ), and this protein also expresses GPx and LPCAT activities. aiPLA 2 activity has an important role in several physiologic functions. These functions include the phospholipid metabolism of lung alveolar type 2 epithelial cells, although the role of this enzyme in the lipid metabolism of other organ and cell types has not yet been investigated. aiPLA 2 activity also has a major physiological role in utilizing the remodeling pathway to repair peroxidized cell membrane phospholipids associated with oxidative stress. Additionally, the aiPLA 2 -mediated generation of lysoPC for conversion to lysoPA is important in the complex pathway for the activation of NOX2. Although aiPLA 2 activity has been associated with a variety of diseases, including ALI and sepsis, cancer and carcinogenesis, degenerative CNS diseases, diabetes, and male infertility, among others, additional evidence is necessary for a full appreciation of its pathophysiologic significance.
The author thanks the many other collaborators who have participated in the study of Prdx6 during the past 25 years, especially Drs. Sheldon Feinstein, Yefim Manevich, David Speicher, Shampa Chatterjee, Yongchen Wu, and Hamidur Rahaman, as well as Mr. Chandra Dodia for his tireless devotion to this project. The author thanks Dr. Cristian O'Flaherty for comments regarding male infertility and Ms. Dawn Williams for help in preparation of the manuscript.