This study shows that reduction of GIVA-PLA2
can prevent diverse Aβ-dependent functional impairments in a transgenic mouse model of AD. We zeroed in on this enzyme because our unbiased lipidomics analysis revealed a surprisingly selective increase in AA and its metabolites in brain tissues of hAPP mice, suggesting increased activity of this particular PLA2
isoform. We also showed that GIVA-PLA2
is expressed in brains of mice and that hAPP mice and AD patients have increased levels of phosphorylated (putatively activated) GIVA-PLA2
in the hippocampus, which is particularly vulnerable to AD in humans and to Aβ-induced neuronal deficits in transgenic mice2, 3, 12-15
releases fatty acids from phospholipids8
, the increased GIVA-PLA2
activity we detected in hAPP mice is consistent with the decreased levels of phospholipid-bound AA in AD brains40
, further underlining the potential relevance of our findings to the human condition. Increased GIVA-PLA2
activity in hAPP mice may also be related to the increases in isoprostanes (nonenzymatic AA metabolites that form during oxidative stress) in the brain and CSF of AD patients and hAPP mice7
and in 4-hydroxynonenals (oxidative aldehyde products of AA) in brains of AD patients41, 42
Interestingly, the translocation of GIVA-PLA2
to its phospholipid substrate is primarily regulated by intracellular calcium8
, and several lines of evidence suggest that Ca2+
-dependent signaling pathways are dysregulated in neurons of hAPP mice, particularly in the hippocampus1-3, 12-15
. Because transgene expression and Aβ levels were higher in the hippocampus than the cortex, the regional differences in aberrant GIVA-PLA2
activation may primarily reflect differences in Aβ levels and a threshold effect. Consistent with this interpretation, our in vitro
studies demonstrated dose-dependent neuronal GIVA-PLA2
activation by extracellular Aβ1−42
The concentration of any fatty acid product is dependent on the concentration of its precursor substrate and the enzymatic activity of the protein that metabolizes that precursor. The elevated levels of LTB4 and 14,15 EET in the cortex in the setting of normal AA levels there suggest that the activities of 5-LO, p450, LTA4 hydrolase, and soluble epoxide hydrolase may be increased in the cortex of hAPP-J20 mice compared to NTG mice.
Several lines of evidence suggest that GIVA-PLA2
activation contributes to Aβ1−42
-dependent neurotoxicity, which can range from synaptic dysfunction to neuronal cell death. Aβ-induced neurotoxicity requires AMPAR activity, which is regulated by surface AMPAR levels31, 32, 43
acutely elevated surface AMPAR levels in cultured neurons, and inhibiting GIVA-PLA2
blocked this process. Treatment with AA also increased surface AMPAR levels. Thus, GIVA-PLA2
and AA may contribute to Aβ neurotoxicity by mediating Aβ-induced increases in surface AMPAR expression (Supplementary Fig. 14
). Indeed, increased surface levels of AMPARs would be expected to increase neuronal excitability44
. Early neuronal excitability is thought to mediate acute Aβ1−42
and to be essential for delayed Aβ1−42
-dependent synaptic deficits30
and in vitro
evidence suggests that Aβ can elicit not only neuronal overexcitation, but also synaptic depression, and that both of these mechanisms may contribute to AD-related cognitive impairments3, 12, 13, 30, 32, 45
. Our results raise the possibility that the decreases in surface AMPAR observed after longer exposure to Aβ32, 39
may result from feedback inhibition (e.g., synaptic scaling46
) triggered by the acute increase in surface AMPAR expression and associated increases in neuronal excitability (Supplementary Fig. 14
). Additional potential causes include long-term depression–like mechanisms32, 39, 43
, and impairments of neuronal functions.
It is difficult to extrapolate from in vitro
to in vivo
conditions, and activation of GIVA-PLA2
leads to the production of many fatty acids with diverse biological activities8, 18
. Thus, it was not clear a priori
activation in hAPP mice or AD is beneficial or detrimental. For instance, AA and its metabolites participate in synaptic plasticity4
, cerebrovascular regulation5, 6
, oxidative stress7, 41, 42
, and inflammation5
, all of which have been implicated in the pathogenesis of AD1, 3, 5
To determine if reducing GIVA-PLA2 activity is beneficial or detrimental in the presence of abnormally elevated Aβ levels, we crossed hAPP mice onto a GIVA-PLA2-deficient background. Elimination or partial reduction of GIVA-PLA2 was well tolerated and effectively reduced learning and memory deficits, behavioral alterations, and premature mortality in hAPP mice. These findings suggest that GIVA-PLA2 contributes to the pathogenesis of these abnormalities and might be a useful target for therapeutic interventions in AD.
Mice and tissue
We analyzed sex-balanced groups of 2−6-month-old heterozygous transgenic and NTG mice from line J202, 12-15
mice were generated by Joseph Bonventre (Harvard Medical School, Boston)21
. All mice were on a C57/BL6 background. All experiments were approved by the Institutional Animal Care and Use Committee of the University of California, San Francisco. Anesthetized mice were perfused transcardially with saline. Mouse brains were processed for immunohistochemistry12
and lipid analysis11, 48
Frozen brain regions were pulverized in liquid nitrogen and spiked with deuteriated and odd chain metabolite surrogates48
prior to homogenization in ice cold methanol. Samples were diluted to 10% methanol with phosphate-buffered saline (pH7.4) and lipids were isolated by solid-phase extraction on Oasis HLB cartridges (Waters). Residues were dried under vacuum and reconstituted in the presence of internal standards allowing surrogate recoveries to be determined48
. Analytes were separated by reverse-phase liquid chromatography on a Waters UPLC equipped with a 2.1 × 150 mm, 1.7-μm C18-BEH Acquity column. A gradient of mobile phases A (water/0.1% acetic acid, wt/wt) and B (80:15 (v/v) acetonitrile/methanol/0.1% acetic acid, wt/wt) was used. Analytes were detected by negative-mode electrospray ionization with a tandem quadrupole mass spectrometer (QuattroMicro, Waters) operated in multireaction monitoring mode, and quantified against five-point calibration curves bracketing the observed concentrations. Quantitative analyte measurements were adjusted according to the recovery of structurally similar surrogate metabolites that were spiked into the samples before lipid extraction. Less than 2% of the spiked epoxide and hydroxyprostane rings were lost during sample preparation. The dehydration products of PGE2
) were quantified and combined with their parent PGs before statistical analysis.
Primary neuronal culture
Rat cortices (embryonic day 18) were digested with papain. Cells were plated in polylysine-coated wells and maintained in serum-free neurobasal medium supplemented with B27 and antibiotics. Half the medium was changed after 7 days in culture and 1 day before use. Cells were used after 14 days in culture. More than 95% of cells were neurons, as determined by staining with antibody against the neuron-specific marker NeuN (data not shown). Cell viability was assessed with trypan blue49
. Neuronal cell treatments included Aβ1−42
, AA, AACOCF3
, and BEL (Cayman Chemical), cell-secreted hAPP695
obtained from medium of Lipofectamine–tranfected HEK cells, NBQX, PD98059, and SB203580 (Sigma).
To prepare Aβ peptides, 1 mg of Aβ1−42 (Biopeptide) was solubilized in a glass tube with ice-cooled 1,1,1,3,3,3-hexafluoro-2-propanol (Sigma), incubated for 1 h at room temperature, allowed to evaporate over 24 h, and spun for 10 min on a speed vac. The remaining precipitate was solubilized with 100 μM DMSO in neurobasal medium, incubated at 4°C for 48 h, and centrifuged at 14,000 × g for 10 min at 4°C. Soluble oligomers were collected and resuspended in neurobasal medium. Concentrations of monomeric Aβ were determined by absorbance at 595 nm after addition of Coomassie reagent (Sigma).
RT reactions contained 300 ng of total RNA (DNase-treated) and random hexamer plus oligo d(T) primers. Diluted reactions were analyzed with SYBR green PCR reagents and an ABI Prism 7700 (Applied Biosystems). cDNA levels of hAPP and mGAPDH were determined relative to standard curves from pooled samples. The slope of standard curves, control reactions without RT, and dissociation curves of products indicated adequate PCR quality. The following hAPP primers were used: hAPP 5RF primer sequence (5'−3'): GAGGAGGATGACTCGGATGTCT; hAPP 6RR primer sequence (5'−3'): AGCCACTTCTTCCTCCTCTGCTA.
Snap-frozen brain tissues were homogenized with guanidine buffer followed by ELISA measurements of human Aβ peptides15
. The ELISA for Aβ1−42
detects this specific peptide. The ELISA for Aβ1-x
recognizes N-terminal fragments of Aβ containing the first 28 amino acids.
Frozen samples were homogenized in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, protease inhibitor cocktail (Roche), 10 μM pepstatin, 5 mM EDTA, 400 μM PMSF, and 0.2% Triton X-100), sonicated on ice, and centrifuged (5000 × g, 15 min). Protein concentration was determined by Bradford protein assay. Protein (40 μg) was loaded into each well of a 4−12% gradient SDS-PAGE gel. Gels were transferred to nitrocellulose membranes. Membranes were incubated with rabbit anti-GIVA-PLA2 (1:200, Santa Cruz Biotechnology), rabbit anti-phospho-GIVA-PLA2 (1:200, Abd Serotec), rabbit anti-calbindin (1:1000, Swant) in blocking buffer for 1 h. Secondary goat anti-rabbit antibodies (Chemicon) were used at 1:10,000 dilution. Protein bands were visualized with an ECL system (Pierce) and quantified densitometrically with Image J (NIH) software.
Sections were stained for GIVA-PLA2, Aβ, and GFAP with an avidinbiotin/peroxidase system (Vector Laboratories). For antigen retrieval, sections were boiled in 50 mM citric acid for 15 min. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in methanol for 15 min. Sections were blocked at room temperature for 1 h in 10% serum and 1% nonfat milk, and incubated overnight with a rabbit anti-GIVA-PLA2 antibody clone N-216 (1:100; Santa Cruz Biotechnology), mouse anti-Aβ1−5 (1:500, 3D6 antibody, Elan Pharmaceuticals), or rabbit anti-GFAP (1:500, Dako) at 4°C in blocking solution. Secondary biotinylated antibody was used at a dilution of 1:200 (Vector Laboratories). The chromagen was diaminobenzidine.
Biotinylation assay for AMPAR
Primary neuronal cultures were treated with Aβ or AA in artificial CSF (aCSF) or with vehicle for 10, 30, or 60 min and rinsed with ice-cold aCSF, and surface receptors were bound with EZ-link sulfo-NHS-SS-Biotin (Pierce). After rinsing with ice-cold glycine and then with ice-cold aCSF, cells were treated with immunoblotting lysis buffer (above) and scraped from the 24-well plate. Surface receptors were precipitated with avidinsepharose beads (Pierce). Samples from two wells were combined for western blot analysis with antibodies against GluR1 (1:1000, Chemicon) or GluR2 (1:1000, Chemicon)50
To minimize effects of social and environmental stressors on behavior, mice were housed singly under conditions of controlled temperature with a standard 12 h light/12h dark cycle. Before each behavioral assessment of mice in the elevated plus maze, open field, and Y-maze tests, residues were removed and the apparatus was cleaned with ethanol-wipes to standardize odors.
As described previously2, 12
, mice were trained to locate first a visible platform (days 1−3) and then a hidden platform (days 4−8) in a pool (122-cm diameter) filled with opaque water (18°C). Training consisted of two daily sessions 2−3 h apart, each consisting of two (visible) or three (hidden) 60-sec trials. During hidden platform training, the platform location was kept constant (in the center of the target quadrant) and the starting point was changed between trials. For all trials, the time (latency) and path length to reach the platform and the swim speed were recorded with a Noldus Instruments EthoVision video tracking system (San Diego Instruments). On days 6, 9, and 11 (before the first trial of the day), the platform was removed for a 60-sec probe trial, during which the proportion of time spent in the different quadrants and the number of target crossings were recorded. Mice that demonstrated a thigmotactic swim pattern during training were excluded from analysis (0−2 mice per genotype).
Open field activity
To assess activity levels, mice were placed individually into well-lit automated activity cages with rows of infrared photocells on each side interfaced with a computer (San Diego Instruments). Open field activity was recorded for two consecutive 5-min periods. Beam breaks were recorded automatically and used to calculate total fine and ambulatory movements, path lengths, and number of rearings at the center and periphery of the field.
Exploratory behavior was tested in a Y-shaped maze consisting of three identical arms, made of dark opaque polyvinyl plastic, with equal angles between each arm. For each trial, mice were placed individually into one arm and allowed to explore the maze for 6 min, during which the sequence and number of arm entries were recorded manually.
Elevated plus maze
Exploratory behavior and anxiety were assessed with an elevated, plus-shaped maze consisting of two open arms and two closed arms equipped with rows of infrared photocells interfaced with a computer (Hamilton-Kinder). In brief, mice were placed individually in the center of the maze and allowed to explore freely for 10 min. Beam breaks were recorded automatically and used to calculate the number of entries, distance traveled, and total time spent in each arm and the number of extensions over the edges of the open arms (edge pokes).
Slice preparation and electrophsyiology
Four-month-old C57/BL6J mice (n=4) were used for in vitro recordings. Mice were deeply anesthetized and decapitated. The brain was rapidly removed and placed in cold (~4°C) oxygenated slicing solution containing (in mM): 2.5 KCl, 1.25 NaH2PO4, 10 MgSO4, 0.5 CaCl2, 26 NaHCO3, 11 glucose, and 234 sucrose. A block of brain was fastened to the stage of a Vibratome-3000 (Vibratome, St. Louis, MO) with cyanoacrylate and 350 μm coronal slices were cut in slicing solution. Slices were incubated for ≥30 min at 30°C in standard aCSF containing (in mM): 2.5 KCl, 126 NaCl, 10 glucose, 1.25 NaH2PO4, 1 MgSO4, 2 CaCl2, and 26 NaHCO3 (pH ~ 7.4 when gassed with a mixture of 95% O2-5% CO2). Single slices were transferred to a submerged recording chamber where they were maintained at 30°C and perfused at a rate of ~2 mL/min. Recordings were made from neurons identified under infrared differential interference contrast video microscopy with a Zeiss Axioskop 2 FS plus microscope. Patch pipettes (4−6 MΩ) were pulled from borosilicate glass World Precision Instrument and filled with (in mM) 5 KCl, 135 K-gluconate, 2 NaCl, 10 HEPES, 4 EGTA, 4 MgATP, and 0.3 Na3GTP (pH ~7.3, 287 mosmol). Access resistance (RA) was monitored during recordings and neurons were discarded if RA > 15 MΩ. Drugs were applied locally via a local perfusion system (Automate Scientific).
For western blots, protein levels were normalized to mean levels in NTG controls. Unpaired, two-tailed t tests were used to assess differences between means for LC-MS/MS and western blot analyses and the effect of AACOCF3 on cell viability. Two-way ANOVA was used for western blot analysis of GIVA-PLA2 after Aβ treatments. One-way ANOVA with repeated measures was used for experiments assessing the dose- and time-dependent effects of Aβ and AA on cell viability. Behavioral data were assessed by two-way ANOVA with repeated measures when indicated. Tukey or Dunnett posthoc tests were used for multiple comparisons. Null hypotheses were rejected at the 0.05 level.