Genetic Ablation of Calcium-independent Phospholipase A2γ Leads to Alterations in Hippocampal Cardiolipin Content and Molecular Species Distribution, Mitochondrial Degeneration, Autophagy, and Cognitive Dysfunction*

Genetic ablation of calcium-independent phospholipase A2γ (iPLA2γ) results in profound alterations in hippocampal phospholipid metabolism and mitochondrial phospholipid homeostasis resulting in enlarged and degenerating mitochondria leading to autophagy and cognitive dysfunction. Shotgun lipidomics demonstrated multiple alterations in hippocampal lipid metabolism in iPLA2γ−/− mice including: 1) a markedly elevated hippocampal cardiolipin content with an altered molecular species composition characterized by a shift to shorter chain length molecular species; 2) alterations in both choline and ethanolamine glycerophospholipids, including a decreased plasmenylethanolamine content; 3) increased oxidized phosphatidylethanolamine molecular species; and 4) an increased content of ceramides. Electron microscopic examination demonstrated the presence of enlarged heteromorphic lamellar structures undergoing degeneration accompanied by the presence of ubiquitin positive spheroid inclusion bodies. Purification of these enlarged heteromorphic lamellar structures by buoyant density centrifugation and subsequent SDS-PAGE and proteomics identified them as degenerating mitochondria. Collectively, these results identify the obligatory role of iPLA2γ in neuronal mitochondrial lipid metabolism and membrane structure demonstrating that iPLA2γ loss of function results in a mitochondrial neurodegenerative disorder characterized by degenerating mitochondria, autophagy, and cognitive dysfunction.

Genetic ablation of calcium-independent phospholipase A 2 ␥ (iPLA 2 ␥) results in profound alterations in hippocampal phospholipid metabolism and mitochondrial phospholipid homeostasis resulting in enlarged and degenerating mitochondria leading to autophagy and cognitive dysfunction. Shotgun lipidomics demonstrated multiple alterations in hippocampal lipid metabolism in iPLA 2 ␥ ؊/؊ mice including: 1) a markedly elevated hippocampal cardiolipin content with an altered molecular species composition characterized by a shift to shorter chain length molecular species; 2) alterations in both choline and ethanolamine glycerophospholipids, including a decreased plasmenylethanolamine content; 3) increased oxidized phosphatidylethanolamine molecular species; and 4) an increased content of ceramides. Electron microscopic examination demonstrated the presence of enlarged heteromorphic lamellar structures undergoing degeneration accompanied by the presence of ubiquitin positive spheroid inclusion bodies. Purification of these enlarged heteromorphic lamellar structures by buoyant density centrifugation and subsequent SDS-PAGE and proteomics identified them as degenerating mitochondria. Collectively, these results identify the obligatory role of iPLA 2 ␥ in neuronal mitochondrial lipid metabolism and membrane structure demonstrating that iPLA 2 ␥ loss of function results in a mitochondrial neurodegenerative disorder characterized by degenerating mitochondria, autophagy, and cognitive dysfunction.
Mitochondria are complex subcellular organelles that orchestrate the integration of multiple energy-producing and signaling pathways, which in turn modulate neuronal excitability, transmission, plasticity, and apoptosis (reviewed in Ref. 1). Increasing evidence has implicated mitochondrial dysfunction as a critical mechanism underlying the pathologic development of many progressive neurodegenerative disorders, including Alzheimer disease, Parkinson disease, amyotrophic lateral sclerosis, and Huntington diseases (2)(3)(4)(5)(6).
Mitochondrial membranes undergo rapid cycles of membrane fusion and fission thereby integrating mitochondrial membrane molecular dynamics with a complex repertoire of interwoven mitochondrial bioenergetic and signaling functions (7). Mature cardiolipins are dimeric doubly negatively charged phospholipids that contain a markedly increased volume in their aliphatic chains in comparison to their polar head group. This is largely accomplished by the initial synthesis of nascent short chain length cardiolipin molecular species and their subsequent remodeling to longer chain length highly unsaturated acyl chain moieties catalyzed by one or more phospholipases and subsequent transacylase and/or acyl transferase activities. The precise regulation of mitochondrial cardiolipin molecular species is necessary to facilitate diverse mitochondrial functions, including integration of cellular bioenergetics, cellular signaling, and mitochondrial membrane fusion and fission.
In the central nervous system, phospholipase A 2 s (PLA 2 s) 3 play critical roles in cellular growth, lipid homeostasis, and second messenger generation (8,9). PLA 2 s catalyze the cleavage of acyl groups from glycerophospholipids, generating free fatty acids and lysophospholipids, thereby initiating dual pathways of signal transduction (10). The released polyunsaturated fatty acids can be further metabolized to numerous biologically active lipid second messengers with discrete biologic functions * This work was supported, in whole or in part, by National Institutes of Health (11,12). Moreover, the production of lysolipids initiates a parallel arm of this signaling pathway through regulating the electrophysiologic properties of neuronal membranes, modulating capacitative calcium influx and serving as precursors of signaling lipids such as platelet-activating factor and lysophosphatidic acid (13,14).
Under physiologic conditions, PLA 2 s generate lipid signal second messengers necessary for critical neuronal functions, including neurotransmitter release, long-term potentiation, and cognitive function (15). Conversely, phospholipases also participate in the pathologic sequelae of neuronal ischemia, axonal dystrophy (e.g. infantile neuroaxonal dystrophy (INAD)), and Alzheimer disease (16,17). However, the types of phospholipases and the biochemical mechanisms that mediate these responses in neuronal tissues are at their earliest stages of understanding.
In previous studies, calcium-independent phospholipase A 2 (iPLA 2 ) was identified as the predominant phospholipase activity present in rat hippocampus accounting for over 70% of the measurable phospholipase A 2 activity (15). Moreover, inhibition of iPLA 2 activity by BEL prevented long-term potentiation, and inhibition of long-term potentiation could be rescued with eicosa-5,8,11-trienoic acid but not eicosa-8,11,14-trienoic acid (15). However, with the discovery of multiple new members of the iPLA 2 family (PNPLA1-9 (HUGO nomenclature)), which are each inhibited by BEL, identification of the one or more specific iPLA 2 s responsible for the observed effects in these early studies became more complex.
The prominent roles of mitochondrial phospholipases in regulating neuronal homeostasis are exemplified by their pleiotropic effects on mitochondrial bioenergetics and signaling (18,19). Through modulating the structure, composition, and organization of mitochondrial membrane constituents, phospholipases participate in the generation and maintenance of highly specialized membrane scaffolds that are necessary for efficient mitochondrial bioenergetic and signaling functions (7, 20 -23). Moreover, alterations in phospholipase activity can potentially modulate mitochondrial bioenergetic efficiency through the production of fatty acids that regulate uncoupling protein activity (24 -26). Thus, alterations in cardiolipin (CL) content and molecular species composition regulate electron transport chain efficiency, apoptosis, and mitochondrial signaling (reviewed in Refs. 27,28).
In prior studies, we identified, purified and cloned a novel calcium-independent phospholipase A 2 activity, termed iPLA 2 ␥ (also known as PNPLA8 by HUGO nomenclature) that is remarkable for the presence of dual mitochondrial and peroxisomal localization sequences (29,30). This phospholipase has the unique property of acting predominantly as a phospholipase A 1 in the presence of phospholipid substrates containing polyunsaturated fatty acids (e.g. arachidonic acid) at the sn-2 position thereby generating 2-arachidonyl lysophosphatidylcholine, which represents a central node in metabolic signaling (31). Cardiac myocyte-specific overexpression of iPLA 2 ␥ in mice results in alterations in mitochondrial ultrastructure that is accompanied by organelle dysfunction (32). Genetic ablation of iPLA 2 ␥ resulted in the generation of viable progeny that demonstrated decreased growth and cold intolerance and pos-sessed defects in ascorbate-stimulated Complex IV function (33).
In the current study, a novel mitochondrial neurodegenerative phenotype was identified in iPLA 2 ␥ Ϫ/Ϫ mice characterized by markedly enlarged heteromorphic structures, which were particularly prominent in the hippocampus, major changes in hippocampal lipids, including cardiolipin content and molecular species composition, and the presence of increased levels of oxidized lipid molecular species. Isolation of the enlarged structures identified them as degenerating mitochondria through proteomic, lipidomic, and electron microscopic analyses thereby demonstrating the essential role of iPLA 2 ␥ in neuronal mitochondrial structure and function.
Generation of the iPLA 2 ␥ Ϫ/Ϫ Mouse-The iPLA 2 ␥ Ϫ/Ϫ mice used in this study were generated as previously described (33). Interbreeding of heterozygous offspring was used to generate homozygous knockouts and wild-type (WT) littermates used in all studies. All experiments were performed through comparisons of littermate wild-type and knockout mice matched by age and sex. Animals were housed in standard cages with ad libitum access to food and water on a 12-h light/dark cycle.
Northern Blot Analysis-Total RNA was isolated from mouse brain sections utilizing an RNeasy tissue kit purchased from Qiagen (Valencia, CA), and Northern blotting was performed as previously described (29).
Electrospray Ionization Mass Spectrometric Analyses of Lipids-Lipid extractions and multidimensional ESI/MS analyses were performed essentially as described previously utilizing a TSQ Quantum Ultra Plus triple-quadrupole mass spectrometer (Thermo Fisher Scientific) with an automated nanospray apparatus (Nanomate HD, Advion Bioscience Ltd.) and Xcalibur system software (36). Enhanced shotgun lipidomics analyses of CL were performed using the Mϩ1/2 isotopologue approach with a mass resolution setting of 0.3 Th as previously described (37). Typically, a 2-to 5-min period of signal averaging was employed for each mass spectrum. A mass resolution of 0.7 Thomson was employed for acquisition of all mass spectra (except as noted above for CL). Analyses of oxidized lipid species were performed by using the multiple reaction monitoring (MRM) methodology previously described by Domingues et al. (38).

Purification of Hippocampal Mitochondria and Resolution of the Enlarged Heteromorphic Lamellar Structures by Buoyant
Density Centrifugation-Hippocampal mitochondria were fractionated using a discontinuous sucrose gradient similar to methods previously described (39). In brief, mouse hippocampus was dissected, homogenized with 10 strokes of a glass Dounce homogenizer in ice-cold 210 mM mannitol, 70 mM sucrose, 2 mM HEPES, pH 7.4, 0.1 mM EDTA buffer, centrifuged at 500 ϫ g for 10 min, followed by 7000 ϫ g centrifugation for 10 min with three washes to obtain a mitochondrial pellet. The pellet was loaded onto a discontinuous sucrose gradient of 15, 23, 32, and 60% sucrose in 10 mM MOPS, pH 7.2, 1 mM EDTA and spun at 134,000 ϫ g for 1 h. Bands were then collected, diluted in excess buffer, and centrifuged at 10,000 ϫ g for MALDI and proteomic analyses.
Proteomic Analyses of the Resolved Mitochondrial Membrane Fractions-Approximately 10 g of the mitochondrial fraction collected from the 32% or the 60% sucrose cushion obtained from iPLA 2 ␥ Ϫ/Ϫ mice hippocampi were subjected to electrophoresis on a 12% Tris-SDS-polyacrylamide gel after mixing with the same volume of loading buffer and boiling for 10 min. Protein bands were visualized with SimplyBlue TM SafeStain (Invitrogen), excised, and destained with 50% acetonitrile and 100 mM NH 4 HCO 3 . After destaining, gel slices were dried using a SpeedVac concentrator, reduced by 10 mM dithiothreitol, and alkylated with 50 mM iodoacetamide at room temperature. After removing excess iodoacetamide, the gel pieces were dehydrated in acetonitrile (CH 3 CN), rehydrated and washed in 0.1 M NH 4 HCO 3 , and dehydrated in acetonitrile. Digestion with trypsin (Promega, Madison, WI) was performed (20 ng/l trypsin) for 30 min. Any excess protease solution was removed, and 40 l of 50 mM NH 4 HCO 3 was added. The resultant peptides were desalted on a C18 minicolumn prior to analysis using a 4800 MALDI-TOF/TOF mass spectrometer operated in the positive ion mode (Applied Biosystems, Foster City, CA) using ␣-cyano-4-hydroxycinnamic acid as matrix. Mass spectra were recorded in the positive ion reflector mode over an m/z range of 800 -4000. After a TOF MS scan, the nine most abundant ions were subjected to product ion analyses using collision-induced dissociation. The voltages of source 1, the collision cell, and the collision cell offset were 8.0, 7.0, and Ϫ0.045 kV, respectively. 1500 consecutive laser shots were averaged, and the resultant MS and tandem MS data were analyzed using GPS Explorer TM Software v2.0 with a Mascot Search Engine against SwissProt databases. All computer-generated matched sequences were verified by manual interpretation of the tandem mass spectra.
Immunohistochemistry and Immunoblotting-Aliquots of tissue homogenate and mitochondrial extracts (10 g) were resolved on 7.5% SDS-polyacrylamide gels, transferred onto a polyvinylidene fluoride membrane, incubated with antibody directed against the protein of interest, and detected using a peroxidase-conjugated secondary antibody. Primary antibodies used for immunoblotting were anti-porin (Calbiochem, 1:200) and anti-COX IV (Invitrogen, 1:500). Tissue processing and immunohistochemistry were performed as previously described (16). In brief, mice were anesthetized by intraperitoneal injection of a ketamine/xylazine mixture, and tissues were fixed by intracardiac perfusion with phosphate-buffered saline followed by 4% paraformaldehyde. Harvested brain was post-fixed overnight in 4% paraformaldehyde and then cut into 2-to 3-mm coronal sections for routine paraffin embedding. For immunohistochemistry analyses, 6-m sections were stained with mouse monoclonal antibody anti-ubiquitin 1510 (Chemicon, 1:10,000). Species-specific biotinylated secondary antibody was utilized followed by avidin-horseradish peroxidase, and the chromogenic substrate diaminobenzidine tetrahydrochloride (Vector Laboratories, Burlingame, CA) for detection of horseradish peroxidase. Sections were then counter-stained with hematoxylin.
Transmission Electron Microscopy-Electron microscopy was performed as previously described (16). In brief, dissected tissue or mitochondrial fractions were fixed overnight in chilled 3% glutaraldehyde in cacodylate. Dissected tissues (1 mm) were postfixed in buffered OsO 4 , dehydrated in graded alcohol solutions and propylene, embedded in Epon, and examined by light microscopy after toluidine blue staining. Thin sections cut onto Formvar-coated slot grids and stained with uranyl acetate and lead citrate were examined with a JEOL 1200 electron microscope.
Locomotor Activity/Behavioral Exploration-Locomotor activity was evaluated in transparent (47.6 ϫ 25.4 ϫ 20.6 cm high) polystyrene enclosures over a 1-h period as previously described (40) using computerized, photo-beam instrumentation (Hamilton-Kinder, LLC, Poway, CA). General activity variables (total ambulations and rearings) along with measures of emotionality, including time spent, distance traveled, and entries made in a central zone, as well as distance traveled in the periphery were analyzed.
Evaluation of Neurologic Impairment in iPLA 2 ␥ Ϫ/Ϫ Mice-The Morris water navigation test was used to assess spatial learning and memory by training mice to locate a platform in a pool of opaque (room temperature) water as previously described (41). The procedure included cued (visible platform, varied location), place (platform submerged, fixed location), and probe (platform removed) trials. Trials were videotaped and swim paths were tracked and recorded by a computerized system (Polytrack, San Diego Instruments, San Diego, CA), which calculated escape path length (distance traveled to find the platform) and latency (time to reach the platform). Swimming speeds were also calculated. Mice received four trials per day for two consecutive days of cued training with the platform being moved to a different location for each trial within a day in the presence of very few distal cues. An intertrial interval of 60 s was used with a mouse being allowed to remain on the platform for 30 s before being removed. Place trials commenced 3 days after completing the cued condition, which involved administering 4 trials per day for 5 consecutive days (60-s maximum for a trial) in the presence of several salient distal cues. A 60-s intertrial interval was used, and the platform remained in the same location for all trials. Approximately 1 h after completion of the place trials on the 5th test day, the platform was removed and a 60-s probe trial was administered to evaluate retention of the platform location. Search behavior accuracy was assessed by analyzing the time spent in the target quadrant by itself as well as the target quadrant time versus the time spent in each of the other pool quadrants, and by the number of crossings made over the platform location (platform crossings).
Statistical Analyses-Analysis of variance models were used to analyze the behavioral data. Typically, the statistical models included two between-subjects variables (genotype and gender) and sometimes one within-subjects (repeated measures) variable, such as blocks of trials or test sessions. When analysis of variance tests with repeated measures were conducted, the Huynh-Feldt adjustment of ␣ levels was used for all withinsubjects effects containing more than two levels to protect against violations of the sphericity/compound symmetry assumptions underlying this analysis of variance model. The Bonferroni correction was used to maintain ␣ levels at 0.05 when multiple comparisons were made (i.e. p ϭ 0.05/number of comparisons).

iPLA 2 ␥ Message Is Abundantly Expressed in All Examined
Regions of WT Mouse Brain and Is Absent in iPLA 2 ␥ Ϫ/Ϫ Mice-First, we used Northern analysis to demonstrate the presence of iPLA 2 ␥ message in multiple regions of WT mouse brain, including cortex, cerebellum, hippocampus, and brainstem. As anticipated, iPLA 2 ␥ Ϫ/Ϫ mice did not contain any detectable message in these regions demonstrating the complete knockout of iPLA 2 ␥ in neuronal tissues (Fig. 1). Collectively, these results demonstrate the presence of robust amounts of iPLA 2 ␥ in neuronal tissues, which is completely ablated in iPLA 2 ␥ Ϫ/Ϫ mice.
Electrospray Ionization Mass Spectrometric Analyses of the Hippocampus in iPLA 2 ␥ Ϫ/Ϫ Mice-Because genetic ablation of phospholipase activity would be anticipated to result in alterations of the lipidome, we exploited the power of shotgun lipidomics with multidimensional mass spectrometry to assess lipid alterations in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice. This region was selected because of its vital importance in learning and memory and from prior work that had demonstrated both the abundance of iPLA 2 activity in the hippocampus as well as the impact of pharmacologic inhibition of iPLA 2 on hippocampal function (15,42,43). Furthermore, in the current study, morphologic alterations appeared to be maximally present in the hippocampus (see below). Shotgun lipidomic analysis of hippocampal lipid extracts from control and iPLA 2 ␥ Ϫ/Ϫ mice demonstrated that cardiolipin molecular species in iPLA 2 ␥ Ϫ/Ϫ mice (the signature [Mϩ1] 2Ϫ isotopologues indicated by asterisks are indicative of doubly negatively charged cardiolipins) were substantially increased and the distribution of molecular species shifted toward lower m/z molecular species in iPLA 2 ␥ Ϫ/Ϫ mice in comparison to the WT littermates ( Fig. 2A). The marked differences in cardiolipin molecular species composition in iPLA 2 ␥ Ϫ/Ϫ mice suggest an important role of iPLA 2 ␥ in CL remodeling in the hippocampus. Multiple cardiolipin molecular species in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice were increased (Fig. 2B). Prominent alterations included a 2-fold increase in CL molecular species containing arachidonate (Fig. 2C) and a 3-fold increase in CL molecular species containing shorter chain length aliphatic groups (m/z Ͻ 720 in iPLA 2 ␥ Ϫ/Ϫ mice (Fig. 2D)). In addition, shotgun lipidomics revealed a 1.5-fold increase in arachidonyl-containing PC molecular species in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice as well as a 2-fold increase in arachidonyl-containing phosphatidylethanolamine molecular species (Fig. 3, A and B). Conversely, a 30% decrease in the content of plasmenylethanolamine molecular species was present (Fig. 3B). Finally, a 25% increase in total ceramide was present in iPLA 2 ␥ Ϫ/Ϫ mice (Fig. 4).

Electron Microscopic Evidence of Enlarged Mitochondria, Heteromorphic Membrane Structures, and Identification of Inclusion
Bodies-Electron microscopic evaluation of iPLA 2 ␥ Ϫ/Ϫ mice was remarkable for the appearance of large bizarrely shaped structures, which appeared to be of mitochondrial origin based upon the presence of graded alterations in the folding architecture of internal membranes. These structures were observed throughout the brain, but were most prevalent in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice (Fig. 5A). No evidence of similar structures was present in WT littermates. These membranous structures in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice varied greatly in size, morphology, and ultrastructure (Fig. 5A). Occasionally nearly normal mitochondria (dark arrows) were found together with hybrid forms containing peculiar membranous aggregates and mitochondria (light arrows). Compared with normal mitochondria (Fig. 5B, far left image), iPLA 2 ␥ Ϫ/Ϫ hippocampus contained a diverse range of structures varying in size and morphology, including structures that closely resembled greatly enlarged and swollen mitochondria that were up to 10 m in diameter (Fig. 5B). These structures typically contained sheets of membrane and were found in stacked or whorled configurations.
Spheroids are typical histologic features of axonal neuropathies such as INAD resulting from missense mutations of iPLA 2 ␤ in humans that were also observed in iPLA 2 ␤ Ϫ/Ϫ mice (16). In the current study, similar spheroid structures, characterized by the presence of a prominent cleft that interrupted the tubulovesicular material, were observed in brain sections from iPLA 2 ␥ Ϫ/Ϫ mice (Fig. 5C). These results suggest that these two iPLA 2 family members share some overlapping functional roles, although it should be noted that structures resembling enlarged or distorted degenerating mitochondria were not observed in neuronal tissue sections from iPLA 2 ␤ Ϫ/Ϫ mice.  2 ␥ mRNA in wild type and its absence in the iPLA 2 ␥ ؊/؊ mouse brain. Northern blot analysis using RNA isolated from cortex, cerebellum, hippocampus, and brain stems of wild-type (W) and iPLA 2 ␥ Ϫ/Ϫ (K) are shown using a [ 32 P]dCTP-labeled 2-kb iPLA 2 ␥ cDNA (iPLA 2 ␥) probe that flanks the region encoding the lipase active site. Hybridization with a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe in each corresponding lane is shown for comparison. Results are representative of separate Northern analyses of RNA extracted from multiple tissues of three WT and three iPLA 2 ␥ Ϫ/Ϫ male animals 6 -8 months of age. iPLA 2 ␥ ؊/؊ -induced Mitochondrial Neurodegeneration DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 35635
This may be partially explained by the predominantly cytosolic localization of iPLA 2 ␤ and/or its association with plasma membrane constituents through its N-terminal ankyrin repeat domains. Finally, transmission electron micrographs of hippocampal tissue revealed ultrastructural evidence of autophagy. Fig. 5D shows a typical enlarged autophagosomal vacuole (arrow) surrounded by a dense delimiting membrane that contains several organelles.
Immunohistologic Identification of Ubiquitin-inclusion Bodies in Discrete Brain Regions of iPLA 2 ␥ Ϫ/Ϫ Mice-Although the gross anatomy of the brain in iPLA 2 ␥ Ϫ/Ϫ mice appeared normal and hematoxylin and eosin staining did not reveal apparent histologic differences between WT and iPLA 2 ␥ Ϫ/Ϫ mice, immunohistologic staining of the internal capsule revealed numerous small ubiquitin-positive inclusions (Fig. 6A). These ubiquitin-positive inclusions were observed in neuropil regions of the deep cerebellar nuclei suggesting localization to distal processes. Ubiquitin-positive inclusions were also found in white matter tracts, including the dentate gyrus and spinal cord gray matter (Fig. 6). These findings indicate impaired clearance of ubiquitinated proteins in neuronal processes. Throughout these regions, enlarged membranous structures similar to those previously shown were also present, although not to the same extent as those present in the hippocampus.
Identification of the Heteromorphic Structures as Degenerating Mitochondria by Sucrose Density Gradient Purification-To unambiguously identify the nature of these unusual structures present in the hippocampus of the iPLA 2 ␥ Ϫ/Ϫ mouse, we reasoned that the buoyant density of these structures was less than that of typical mitochondria and thus they should be resolved by sucrose density gradient fractionation. Analysis of hippocampal homogenates separated by discontinuous buoyant density gradient centrifugation revealed that, although nearly all of the WT membranous material was enriched at the 60% sucrose interface (normal mitochondrial sedimentation), membranous material was markedly enriched in the 32% sucrose fraction in homogenates from iPLA 2 ␥ Ϫ/Ϫ mice. The 32 and 60% interfacial material from the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice were further characterized by electron microscopy, shotgun lipidomics, and proteomics. First, electron microscopic examination demonstrated that the 32 and 60% interfacial materials from iPLA 2 ␥ Ϫ/Ϫ mice contained enlarged and distorted membrane bodies thereby establishing the identity of the interfacial material as the targeted enlarged membranous structures identified in whole tissue sections (Fig. 7, A and B). Importantly, in hippocampus from wild-type littermates, only trace amounts of protein were present at the 32% sucrose interface, and electron microscopy demonstrated only normal-sized mitochondria with typical cristae lacking any internal membranous structures (Fig. 7B). In contrast, the corresponding 32% sucrose interface of iPLA 2 ␥ Ϫ/Ϫ samples contained distorted mitochondrial structures similar to those observed by electron microscopy of hippocampal tissue sections. SDS-PAGE demonstrated that the protein composition of the material at the 32% interface from the iPLA 2 ␥ Ϫ/Ϫ mouse hippocampus closely resembled that of the proteins present at the 60% interface from wild-FIGURE 2. Altered levels of cardiolipin molecular species in the hippocampus of iPLA 2 ␥ ؊/؊ mice. A, hippocampal lipid extracts of wild-type and iPLA 2 ␥ Ϫ/Ϫ mice were prepared using a modified Bligh and Dyer procedure. Negative ion electrospray ionization mass spectra were acquired using a QqQ mass spectrometer as described under "Experimental Procedures." Equal amounts of a spiked CL internal standard (tetra 14:0 CL ([Mϩ1] 2Ϫ isotopologue at m/z 619.9) were added to each sample, and spectra were normalized to the intensity of the CL internal standard (not shown)). Peak heights represent the relative intensity (%) of the individual molecular species to the intensity of the CL internal standard. The asterisks indicate doubly charged CL plusone isotopologues whose ion peak intensities were utilized to quantify individual CL molecular species as described previously (37,78). The results indicate an increase in multiple CL species in the iPLA 2 ␥ Ϫ/Ϫ mice and especially those containing shorter chain length aliphatic chains. B, the levels of total CL molecular species were calculated in comparison with internal standard after 13 C deisotoping as described previously (37,78). A 1.5-fold increase in total cardiolipin content was observed in the iPLA 2 ␥ Ϫ/Ϫ hippocampus in comparison to WT littermates (Total CL, p Ͻ 0.05). C, a nearly 2-fold increase in a major arachidonyl-containing CL species (20:4 -20:4 -18:1-18:1) was observed (20:4 CL, p Ͻ 0.02). D, the majority of the increase in total CL was due to a 3-fold increase in the proportion of low molecular weight cardiolipin species (Ͻ720 CL, p Ͻ 0.002). All experiments were performed with four male WT and iPLA 2 ␥ Ϫ/Ϫ mice between 7 and 10 months of age. iPLA 2 ␥ ؊/؊ -induced Mitochondrial Neurodegeneration type mouse hippocampus that contained structurally normal mitochondria (Fig. 8A). Proteomic analysis of the 32% fraction from iPLA 2 ␥ Ϫ/Ϫ mice identified multiple mitochondrial proteins (Fig. 8B) thereby confirming that the membranous structures in the iPLA 2 ␥ Ϫ/Ϫ mouse were mitochondrial in origin. Similarly, lipidomic analyses demonstrated a significant enrichment of cardiolipin in the 32% fraction from the iPLA 2 ␥ Ϫ/Ϫ mouse (Fig. 8C). Furthermore, alterations in CL content and molecular species composition were evident in both the 32 and 62% sucrose fractions from iPLA 2 ␥ Ϫ/Ϫ mice in comparison to their WT littermates (Fig. 9A). The 32% fraction from iPLA 2 ␥ Ϫ/Ϫ mice also contained the traditional mitochondrial markers porin and COX IV by Western blot analysis (Fig. 9B). Collectively, the ultrastructural, proteomic, and lipidomic data strongly support the conclusion that the unusual membrane structures in the iPLA 2 ␥ Ϫ/Ϫ mouse hippocampus consist of grossly enlarged lipid-laden mitochondria with accumulated membranous structures, indicating a defect in normal mitochondrial membrane homeostasis that was accompanied by marked mitochondrial degeneration.
Identification of Increased Free Radical Formation in Hippocampal Tissue from iPLA 2 ␥ Ϫ/Ϫ Mice-To gain insight into the biochemical mechanisms leading to the appearance of enlarged mitochondria and associated membranous structures, we used MRM to identify oxidized ethanolamine glycerophospholipids (monitoring the Mϩ16 m/z fragments of the anticipated fatty acid anions). The results demonstrated a 4-fold increase in oxidized phosphatidylethanolamine glycerophospholipids (Fig. 10). Inefficient operation of the electron transport chain, due to the presence of alterations in phospholipid composition and dynamics of the mitochondrial inner membrane leads to the generation of reactive oxygen species. These Mouse hippocampal tissues were obtained from wild-type or iPLA 2 ␥ Ϫ/Ϫ mice fed ad libitum. The same hippocampal lipid extracts as those used in Fig. 2 were subjected to ESI/MS analyses in both the positive-ion and negative-ion modes in the presence of a small amount of LiOH as described under "Experimental Procedures." Spectra were normalized to the internal standards at m/z 680.6 for PC (A) and at m/z 686.5 for PE (B) and are represented as relative intensity (%) of the highest peak at m/z 766.7 in A and the highest peak at m/z 790.6 in B for direct comparison between WT and KO spectra. A, a 1.5-fold increase in arachidonyl-containing PC species indicated by asterisks was observed in the iPLA 2 ␥ Ϫ/Ϫ mice in comparison to WT littermates (p Ͻ 0.02). B, major PE plasmalogen species are indicated by brackets in the spectra. A 30% decrease in PE plasmalogen content was observed in the iPLA 2 ␥ Ϫ/Ϫ mice (p Ͻ 0.001) along with a nearly 2-fold increase in D18:0 -20:4 PE species (p Ͻ 0.001). All experiments were performed with male WT and iPLA 2 ␥ Ϫ/Ϫ mice between 7 and 10 months of age. iPLA 2 ␥ ؊/؊ -induced Mitochondrial Neurodegeneration DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51 results do not exclude other mechanisms but, rather, provide an integrated mechanism for the excessive production of free radicals and the resultant degeneration of mitochondria in iPLA 2 ␥ Ϫ/Ϫ mice (see "Discussion").
Spatial Learning and Memory Deficits in iPLA 2 ␥ Ϫ/Ϫ Mice-Results from the Morris water navigation test show that iPLA 2 ␥ Ϫ/Ϫ mice exhibited spatial learning and memory performance deficits. However, slower swimming speeds were also observed in iPLA 2 ␥ Ϫ/Ϫ mice suggesting the presence of possible sensorimotor and/or motivational disturbances thus making it difficult to determine whether the acquisition deficits during the place (spatial learning) trials were due to learning and memory impairments, compromised nonassociative functions, or both. For example, iPLA 2 ␥ Ϫ/Ϫ mice were significantly impaired with regard to escape path length (genotype main effect: p ϭ 0.002) during the place trials compared with the WT mice (Fig. 11A), whereas differences between the groups were even greater with regard to escape latency (genotype main effect: p Ͻ 0.0005; data not shown). However, the iPLA 2 ␥ Ϫ/Ϫ mice also had significantly slower swimming speeds (genotype main effect: p ϭ 0.001) compared with the WT controls (Fig.  11B), suggesting inferior swimming capabilities in the iPLA 2 ␥ Ϫ/Ϫ mice, which provided an explanation for the large between-group differences with regard to the escape latency data. Moreover, although path length is unlikely to be affected by differences in swimming speeds per se, the latter data do raise the possibility that the significantly greater path lengths of the iPLA 2 ␥ Ϫ/Ϫ mice might not have been due solely to spatial learning impairments but that alterations in other nonassociative functions may have contributed as well. Results from the cued trials (Fig. 11C), which were conducted before the place trials, provided evidence that the impaired swimming capabil- The same lipid extracts described in Figs. 2 and 3 were diluted in chloroform/ methanol and directly infused into an ESI ion source for ESI/tandem MS analyses as described under "Experimental Procedures." Spectra were normalized to internal standards and represented as the relative intensity (%) of the internal standard peak for direct comparisons between WT (upper panel) and KO (lower panel) spectra. A 25% increase in total ceramide content was observed in the iPLA 2 ␥ Ϫ/Ϫ (KO, p Ͻ 0.05, bar graph) compared with wild-type controls (WT) with most of the increase due to an increase in 18:0 molecular species at m/z 564.7 (p ϭ 0.02); n ϭ 3 per group. All experiments were performed with male WT and iPLA 2 ␥ Ϫ/Ϫ mice between 7 and 10 months of age. Open arrows identify structures with features of both mitochondria and membranous bodies. B, a series of images illustrating the range of mitochondrial and mitochondrial-like membranous body sizes observed in iPLA 2 ␥ Ϫ/Ϫ hippocampus. Numerous enlarged structures had an onion-like appearance with concentric whorls of membranous material. C, electron micrograph of a dystrophic axon in iPLA 2 ␥ Ϫ/Ϫ mice showing the presence of tubulovesicular elements with a "typical cleft" similar to lesions previously seen in the iPLA 2 ␤ Ϫ/Ϫ mouse (16). These clefts typically are admixed with sheets of membrane that are sometimes found in stacked or whorled arrangements. D, an example of a morphologic structure that has characteristics of an autophagosome (arrow). Microscopists were blinded to the identity of the analyzed groups; n ϭ 3 per group.
ities of the iPLA 2 ␥ Ϫ/Ϫ mice had relatively little impact on the path length data. For example, a significant genotype by blocks of trials interaction (p ϭ 0.028), and subsequent pairwise comparisons indicated that the groups only differed on the last block of trials when the iPLA 2 ␥ Ϫ/Ϫ mice had significantly longer path lengths (p ϭ 0.001), even though the groups also differed significantly in swimming speeds across the first three blocks of trials but not the last (data not shown). Analysis of the probe trial data provided more evidence of cognitive impairment in iPLA 2 ␥ Ϫ/Ϫ mice, particularly with regard to retention capabilities. For example, the probe trial results documented the significantly impaired performance of the iPLA 2 ␥ Ϫ/Ϫ mice on variables pertaining to both a highly resolved retention of the platform location (i.e. platform crossings; genotype main effect: p ϭ 0.003) as well as on variables reflecting a more generalized retention of the platform location, including time spent in the target quadrant (genotype main effect: p ϭ 0.04) that had contained the submerged platform, and spatial bias (Fig. 11, D and  E). Importantly, the WT mice showed spatial bias for the target quadrant by spending significantly more time in that quadrant compared with the time spent in each of the other quadrants (p Ͻ 0.007), while the iPLA 2 ␥ Ϫ/Ϫ mice showed no such bias. These results suggest that the WT mice learned and retained the platform location in contrast to the iPLA 2 ␥ Ϫ/Ϫ mice that showed little evidence of accurate retention. In contrast to the results from the water navigation task, the iPLA 2 ␥ Ϫ/Ϫ mice did not differ from WT controls on any variables from the locomotor activity/exploratory behavior test suggesting that their poor spatial learning and memory performance was not due to general malaise or torpor (data not shown).

DISCUSSION
The results of the present study identify multiple chemical, ultrastructural, and functional changes in the hippocampus of iPLA 2 ␥ Ϫ/Ϫ mice including: 1) increased cardiolipin content and altered molecular species distribution; 2) decreased plasmalogen content; 3) increased arachidonate content; 4) accumulation of oxidized lipid molecular species; 5) increased ceramide content; 6) activation of the ubiquitin-proteasome pathway and ultrastructural evidence for autophagy; 7) cognitive dysfunction; and 8) altered mitochondrial lipid homeostasis resulting in the presence of multiple heteromorphic membranous structures of mitochondrial origin. The concurrent nature of each of these alterations in iPLA 2 ␥ Ϫ/Ϫ mice identifies the obligatory role of iPLA 2 ␥ in physiologic mitochondrial phospholipid metabolism, phospholipid composition, and membrane molecular dynamics. The observed mitochondrial neuropathic phenotype of the iPLA 2 ␥ Ϫ/Ϫ mouse is of particular significance considering the known genetic origins of alterations in membrane lipid composition and FIGURE 6. Spheroid formation in discrete brain regions present in iPLA 2 ␥ ؊/؊ mice. Although the iPLA 2 ␥ Ϫ/Ϫ brain appeared normal by gross anatomical examination, and hematoxylin and eosin staining did not reveal apparent differences relative to WT, immunohistologic staining for ubiquitin in brain regions from 10-to 12-month-old male iPLA 2 ␥ Ϫ/Ϫ animals revealed numerous small ubiquitin-positive inclusions (arrows) in the internal capsule (A), molecular layer of the dentate gyrus in the hippocampus (B), polymorphic layer of the dentate gyrus in the hippocampus (C), and the spinal cord gray matter (D). In parallel studies, spheroids were rarely observed in WT brains (not shown). Immunostaining was labeled with a brown chromogen. Nuclei were stained with a hematoxylin counterstain. The bar represents 10 m. FIGURE 7. Transmission electron micrographs of hippocampal mitochondrial fractions. Hippocampal tissue from WT and iPLA 2 ␥ Ϫ/Ϫ mice were dissected, homogenized, and centrifuged to obtain a 7000 ϫ g mitochondrial fraction. This fraction was then subjected to discontinuous sucrose gradient centrifugation and the bands at the 32 and 60% sucrose interfaces were collected as described under "Experimental Procedures." After dilution in homogenization buffer and centrifugation at 10,000 ϫ g, 3% glutaraldehyde in cacodylate buffer was then added to the resultant pellets for subsequent analyses by transmission electron microscopy. A, electron microscopy of the 60% sucrose fractions from WT and iPLA 2 ␥ Ϫ/Ϫ (KO) samples illustrating the presence of multilaminar mitochondrial structures in the hippocampus of the iPLA 2 ␥ Ϫ/Ϫ mouse. B, representative material from the 32% sucrose interface from WT and iPLA 2 ␥ Ϫ/Ϫ mitochondrial pellets illustrating dramatic differences in organellar size and morphology; n ϭ 3 per group. All experiments were performed with male WT and iPLA 2 ␥ Ϫ/Ϫ mice between 10 and 12 months of age. iPLA 2 ␥ ؊/؊ -induced Mitochondrial Neurodegeneration DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51 metabolism manifest in the development of Alzheimer disease and other neurodegenerative processes (44). Numerous studies have demonstrated that PLA 2 s, and iPLA 2 ␥ in particular, play major roles in lipid metabolic pathways vital for normal neuronal function (8,12,14). Previously, we demonstrated that iPLA 2 ␥ displays regioselective PLA 1 activity with phospholipid substrates containing polyunsaturated fatty acids (e.g. arachidonic acid at the sn-2 position) resulting in the generation of 2-arachidonoyl lysolipids, which serve as central nodes in eicosanoid generation and downstream signaling (31). In addition, previous studies demonstrated that iPLA 2 ␥ promotes agonist-stimulated release of arachidonic acid in HEK293 cells (45) and the selective release of arachidonic acid from plasmalogens in ventricular myocytes (46). Considering that both eicosanoids and lysolipids mediate multiple cellular signaling pathways, the observed neurologic pathologies due to iPLA 2 ␥ ablation are not unanticipated and likely result from the accumulation of lipid substrates for iPLA 2 ␥ within the mitochondria. Alterations in mitochondrial membrane composition are likely associated with compromised mitochondrial function analogous to mitochondrial function observed in adenoviral-mediated short hairpin RNA iPLA 2 ␥ knockdown previously demonstrated in renal proximal tubular cells (47).
Cardiolipin is a critical mitochondrial phospholipid that facilitates protein supercomplex formation in the mitochondrial inner membrane allowing optimal electron transport chain function. Decreases in cardiolipin content and alterations in molecular species composition have previously been demonstrated in aging as well as multiple pathologic disorders (48 -50). Newly synthesized cardiolipin typically undergoes remodeling to incorporate longer and more highly unsaturated acyl chains (51). Under normal conditions, CL remodeling typically is initiated by the action of a phospholipase to generate lyso-CL that is subsequently re-acylated by either transacylation or acyl-CoA-mediated acyltransferase activity (52). This remodeling serves to promote optimal electron transport efficiency and bioenergetic capacity in muscle where the predominant molecular species of cardiolipin is tetralinoleoyl (18:2)-CL (53). However, a diverse array of hundreds of CL molecular species (including those containing arachidonic acid) exist in brain that likely participate in neuronal signaling functions The accumulation of shorter chain acyl species of CL in hippocampus from iPLA 2 ␥ Ϫ/Ϫ mice are consistent with a defect in cardiolipin remodeling with an expected reduction in mitochondrial bioenergetic efficiency, inefficient electron coupling, and alterations in cellular signaling. The observed increase in hippocampal cardiolipin content and increases in shorter aliphatic chain molecular species are likely direct consequences of decreased hydrolysis and remodeling of nascent cardiolipin, which appears to be a direct result of iPLA 2 ␥ loss of function. We point out that myocardium from iPLA 2 ␥ Ϫ/Ϫ mice contains reduced cardiolipin levels containing an increased relative proportion of molecular species with arachidonic and docosahexenoic acids (33). This is most likely due to the profound differences in brain and heart cardiolipin-remodeling enzymes, including alterations in transacylation and acyl transferase activity. The lack of accumulation of lipids in myocardial mitochondria suggests that compensatory mechanisms for hydrolyzing cardiolipin are present in myocardial mitochondria that are not effective in hippocampus. Collectively, these results are consistent with previous evidence in Drosophila demonstrating an important role of an iPLA 2 homolog in cardiolipin remodeling (54).
Many lines of research have linked oxidative stress with cerebrovascular disease, dementia, and Alzheimer disease (4,(55)(56)(57). Certain cellular phospholipids, such as plasmalogens, ethanolamine glycerophospholipids, and those containing highly unsaturated acyl groups, are particularly susceptible to oxidative damage. Brain contains an abundance of plasmalogens, which serve important roles in defining the electrophysiologic properties of neuronal membranes (58). A primary mechanism for the radical-mediated oxidation of plasmalogens occurs through the stabilization of three electron resonance structures present in the vinyl ether linkage resulting in its sus- FIGURE 8. Proteomic and lipidomic analyses of the iPLA 2 ␥ ؊/؊ mitochondrial fraction from hippocampus. Hippocampal tissue was homogenized and centrifuged to obtain a 7000 ϫ g mitochondrial fraction that was then loaded onto a discontinuous sucrose gradient. A, the mitochondrial fractions collected from the homogenate (Hom), 32 and 60% sucrose cushions of wildtype (W), and iPLA 2 ␥ Ϫ/Ϫ (K) hippocampus were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. Note the similarities in the proteinbanding pattern of the 32% iPLA 2 ␥ Ϫ/Ϫ fraction with those of the 60% WT and iPLA 2 ␥ Ϫ/Ϫ fractions. B, for proteomic analyses, individual bands from the iPLA 2 ␥ Ϫ/Ϫ 32% fraction were excised, destained, and trypsinized to obtain peptides that were then subjected to MALDI MS analysis and data-dependent tandem MS as described under "Experimental Procedures." A summary of identified proteins and their corresponding bands by SDS-PAGE is illustrated with established mitochondrial markers indicated by asterisks. C, total cardiolipin content of the 32 and 60% sucrose fractions from WT (closed bars) and iPLA 2 ␥ Ϫ/Ϫ (open bars) hippocampal mitochondria. Negative ion ESI/MS were acquired using a QqQ mass spectrometer for quantitation of cardiolipin species as described under "Experimental Procedures." Note the significant decrease in CL present in the iPLA 2 ␥ Ϫ/Ϫ 60% sucrose interface (normal mitochondria) and an increase in CL content of the iPLA 2 ␥ Ϫ/Ϫ 32% sucrose interface (light mitochondria) relative to WT. Lanes represent total cardiolipin in the 60 and 32% sucrose fractions of wild-type (W) or iPLA 2 ␥ Ϫ/Ϫ (K) hippocampal mitochondria. All experiments were performed with male wild-type and iPLA 2 ␥ Ϫ/Ϫ mice between 10 and 12 months of age. ceptibility to oxidative cleavage generating 2-acyl lysolipids as demonstrated in early lipidomics experiments (59). Importantly, the 2-acyl lysolipids formed from plasmalogen free radical oxidation likely have distinct pathologic sequelae in comparison to those generated by iPLA 2 ␥-catalyzed sn-1 chain PLA 1 hydrolysis of diacyl phospholipids containing polyunsaturated fatty acids at the sn-2 position. It is important to note that decreased plasmenylethanolamine and an increased content of oxidized phosphatidylethanolamines (PEs) has been reported during aging as well as during neuropathologic states (60,61). Clearly, disruption of iPLA 2 ␥-mediated signaling and membrane lipid remodeling leads to mitochondrial degeneration, increased oxidative stress, and precipitation of neurologic dysfunction.
Notably, genetic ablation of iPLA 2 ␥ resulted in a robust increase in cellular ceramide content. Increases in ceramides have been previously associated with apoptosis and oxidative stress (62). Specifically, ceramides have been proposed to result in organized structures in the mitochondrial outer membrane facilitating the release of cytochrome c promoting apoptosis through activation of caspase 3.
In recent years, attention has increasingly focused on mitochondrial dysfunction as a causative factor in neurodegeneration and the pathogenesis of Alzheimer, Parkinson, Hunting- FIGURE 9. Cardiolipin and protein mitochondrial markers in hippocampal mitochondrial fractions. A, hippocampi from wild-type (WT) and iPLA 2 ␥ Ϫ/Ϫ (KO) were dissected and homogenized, and mitochondrial fractions were isolated using a discontinuous sucrose gradient. Lipidomic analyses were performed using ESI/MS as described in detail under "Experimental Procedures." The total ion intensity of individual CL molecular species was calculated as previously described (53) based upon peak intensities of doubly charged plus-one isotopologues and the theoretical 13 C isotopic distribution. Multiple alterations in cardiolipin species mass and distribution were observed in the 60% fraction and 32% fraction isolated from WT and KO mice. The identities of the major individual species m/z (identified as described in Ref. 8-month-old male WT and iPLA 2 ␥ Ϫ/Ϫ mice. B, Western analysis of WT and iPLA 2 ␥ Ϫ/Ϫ hippocampal homogenates and sucrose density gradient fractions for the mitochondrial markers porin and COX IV. Western analysis was performed on hippocampal tissue homogenates and isolated mitochondrial fractions (10 g/lane) from 8-month-old WT and iPLA 2 ␥ Ϫ/Ϫ male mice. Antiporin and anti-COX IV antibodies were used to visualize porin and COX IV following incubation with horseradish peroxidase-conjugated secondary reagents as described under "Experimental Procedures." Homog ϭ homogenate (WT) and iPLA 2 ␥ Ϫ/Ϫ (KO); 32% -7000g ϭ mitochondria present at the 32% sucrose interface after 7000 ϫ g centrifugation of the WT and KO homogenates; 60% -7000g ϭ mitochondria present at the 60% sucrose interface after 7000 ϫ g centrifugation of the WT and KO homogenates; arrows indicate the locations of bands corresponding to porin or COX IV. Results are representative of three separate Western analyses. iPLA 2 ␥ ؊/؊ -induced Mitochondrial Neurodegeneration DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51 ton, amyotrophic lateral sclerosis, hereditary spastic paraplegia, and cerebellar diseases (63,64). Mitochondria are centrally involved in neurotransmission as well as in the metabolic plasticity necessary for neuronal survival (65). Many studies have described abnormal, compromised, or enlarged mitochondria in aging, multiple neurologic diseases, and apoptosis (49,51,53,54). Aging is associated with a decreased capacity to generate ATP by oxidative phosphorylation due primarily to diminished complex I and IV activity in mitochondria (6,64,66). Increases in mitochondrial size and decreased membrane potential are present during aging and likely lead to mitochondrial dysfunction through oxidative damage (67). The enlarged mitochondria observed during aging often contain concentric multilamellar structures that were observed in a neuronal cell line highly susceptible to oxidative stress (63) as well as in patients with Barth syndrome. These structures likely originate from an imbalance in lipid synthesis, remodeling, and catabolism lead-ing to the presence of lipid-laden structures that were isolated and characterized by proteomics and lipidomics in this study.
The presence of ubiquitin-immunoreactive inclusion bodies is a characteristic pathologic marker of motor neuron disease (68,69). Alterations in the ubiquitin-proteasome system and autophagy through lysosomes are increasingly recognized as playing important roles in neurodegeneration (70 -72). The importance of lipids in protein ubiquitination, endocytosis, sorting, and degradation has been well documented (73). Elimination of aged and dysfunctional mitochondria protects cells from their potentially harmful effects while allowing cellular survival in the presence of modest mitochondrial damage. However, in the present case it is clear that the prevalence of the enlarged mitochondrial autophagocytic structures likely reflects contributions from their increased generation or decreased clearance.
The present studies demonstrate that ablation of iPLA 2 ␥ produces neurologic impairment, including performance deficits in spatial learning and memory and swimming as assessed with the Morris water maze test. Collectively, these studies identify the importance of iPLA 2 ␥ in maintenance of normal brain function with particular reference to aspects of behavior and cognition mediated in large part by the hippocampus. Of note is the presence of defects in cognitive performance and spatial recognition in patients with Barth syndrome (74).
In addition to iPLA 2 ␥, two other members of the patatin family have previously been implicated in the pathogenesis of neurodegenerative disorders. Ablation of the patatin homolog neuropathy target esterase in Drosophila and mice was previously shown to result in neurodegeneration (75,76). Secondly, a mutation in human iPLA 2 ␤ results in INAD (77) characterized by diffuse cerebellar atrophy and abnormal iron deposition in the medial and lateral globus pallidum. Recently, ablation of iPLA 2 ␤ in a mouse model was demonstrated to recapitulate the major features of INAD (16). However, neither INAD in humans nor iPLA 2 ␤ Ϫ/Ϫ mice demonstrated the pleiotropic alterations of mitochondrial structure or displayed autophagy as indicated by electron microscopy in this study.
In summary, the data presented identify the obligatory role of iPLA 2 ␥ in neuronal mitochondrial structure and function. FIGURE 11. Spatial learning and memory deficits in iPLA 2 ␥ ؊/؊ mice in the Morris water navigation test. A, acquisition performance during place trials iPLA 2 ␥ Ϫ/Ϫ mice (open squares) demonstrated impaired acquisition (spatial learning) performance by exhibiting significantly longer escape path lengths versus WT controls (closed circles) across the blocks of trials ( † , p ϭ 0.002). Groups differed significantly beyond Bonferroni correction in blocks 2, 3, and 5 (*, p Ͻ 0.009) and in block 4 ( # , p Ͻ 0.02). B, swimming performance during place trials. iPLA 2 ␥ Ϫ/Ϫ mice had significantly reduced swim speeds versus WT controls ( † , p Ͻ 0.0001), where differences exceeded Bonferroni correction for blocks 1, 2, 4, and 5 (*, p Ͻ 0.004) and for block 3 ( # , p Ͻ 0.02). C, cued trials performance. Mice were evaluated in the cued trials before conducting the place trials to assess the presence of compromised nonassociative functions in the mice that might affect subsequent acquisition performance during the place condition. An analysis of variance on the path length data yielded a significant genotype by blocks of trials interaction (p ϭ 0.028), but subsequent pairwise comparisons showed that the groups did not differ during the cued trials except for Block 4 when the iPLA 2 ␥ Ϫ/Ϫ mice (open squares) had significantly longer path lengths compared with the WT (closed circles) controls (*, p ϭ 0.001). D, assessing retention of the precise platform location during the probe trial. iPLA 2 ␥ Ϫ/Ϫ mice (open bar) had significantly fewer platform crossings than WT (closed bar) controls (p ϭ 0.003) suggesting that the iPLA 2 ␥ Ϫ/Ϫ mice were impaired in terms of retaining an exact location of the platform. E, determination of time in the pool quadrants to assess retention performance. iPLA 2 ␥ Ϫ/Ϫ mice also exhibited impaired probe trial performance in terms of a more generalized retention of the platform location by spending significantly less time in the target (TGT) quadrant compared with WT mice (p ϭ 0.04). The retention deficits in the iPLA 2 ␥ Ϫ/Ϫ mice were further documented by comparisons with WT mice that showed a spatial bias for the TGT quadrant and spent significantly (*, p Ͻ 0.007) more time in it compared with quadrants that were to the right (RGT), left (LFT), or opposite (OPP) the target quadrant, whereas the iPLA 2 ␥ Ϫ/Ϫ mice showed no such spatial bias. All experiments were performed with 10 WT and 9 iPLA 2 ␥ Ϫ/Ϫ mice at 4 months of age.
These alterations result from profound changes in mitochondrial lipid metabolism resulting from the absence of iPLA 2 ␥ catalytic activity. These results thus present a compelling case for an obligatory role of iPLA 2 ␥ in normal neurologic function and underscore the diverse array of biochemical and neuropathologic abnormalities resulting from iPLA 2 ␥ loss of function. Understanding the contributions of alterations in iPLA 2 ␥ processing and catalytic activity as well as those resulting from non-synonymous mutations in neurodegenerative disorders represents a new frontier in the investigation of these disease states.