Increased PLA₂-dependent fatty acid levels in brain tissues of hAPP mice. Lipids were extracted from hippocampal or cortical homogenates from hAPP and NTG mice (n=12 per genotype) and analyzed by quantitative LC-MS/MS. Fatty acid levels (pmol) were normalized to total protein (mg). a–f, Hippocampal and cortical levels of (a) AA, (b) total AA-derived metabolites, (c) PGE₂ and its nonenzymatic PGB₂ degradation product, (d) LTB₄, (e) 14,15-EET and its 14,15-DHET metabolite, and (f) total fatty acids. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 vs. NTG or as indicated by bracket (t test; means ± s.e.m.).

GIVA-PLA₂ levels in hAPP mice and humans with AD. (a–f) Coronal sections of cortex and hippocampus (a–c) or hippocampus (d–f) in 6-month-old NTG (a,d), hAPP (b,e), and GIVA-PLA₂-deficient (c,f) mice. Scale bar (a–c) 1 mm, (d–f) 250 μm. g–i, Hippocampal and cortical levels of GIVA-PLA₂ in mice were determined by western blot analysis with a rabbit polyclonal antibody. g, Representative western blot showing levels of phosphorylated (p) and unphosphorylated (u) GIVA-PLA₂. Tubulin served as a loading control, NIH3T3 cells as a positive control (+), and cortex from GIVA-PLA₂-deficient mice as a negative control (−). h,i, Hippocampal (h) and cortical (i) GIVA-PLA₂ levels determined by densitometric analysis of western blot signals (n=12 mice per genotype and brain region; age: 6 months). j, hAPP mRNA levels in hippocampus and cortex of hAPP mice determined by quantitative RT-PCR (n=7 mice; age: 2−4 months). k, l, Levels of Aβ_1-x (k) and Aβ₁₋₄₂ (l) (ng per g of tissue) in hippocampus and cortex of hAPP mice determined by ELISA (n=8 mice; age, 2−4 months). m–n, Levels of phosphorylated GIVA-PLA₂ protein in the CA1 hippocampal region in patients with mild, moderate, or severe AD and in nondemented, age-matched controls (C) were determined by western blot analysis. m, Representative western blot. n, Western blot signals were quantitated densitometrically and normalized to tubulin (n=4−8 cases per group). *P<0.05, ****P<0.0001 (t test; mean ± s.e.m.), **P<0.01 vs. control (Tukey test; mean ± s.e.m.).

Inhibition of GIVA-PLA₂ prevents Aβ₁₋₄₂ toxicity in primary neuronal cultures. Primary rat neurons were treated with Aβ₁₋₄₂ oligomers as indicated after 14 days in vitro. Quantitative results were obtained from three wells per condition in five independent experiments and normalized to untreated controls. a, Levels of phosphorylated and unphosphorylated GIVA-PLA₂ in cell lysates were determined by western blot analysis. Aβ increased levels of phosphorylated GIVA-PLA₂ in a dose- and time-dependent manner (P<0.0001 by two-way ANOVA, mean ± s.e.m.). b–d, Percentage of viable cells determined by trypan blue exclusion and counting of unlabeled cells. b, Aβ caused neuronal cell death in a dose- and time-dependent manner (P<0.001 by repeated-measures ANOVA, mean and s.e.m.). c, AA also led to neuronal death (P<0.01 by repeated-measures ANOVA). d, Pretreatment of cells with AACOCF₃, a GIVA-PLA₂-specific inhibitor, for 10 min ameliorated Aβ-induced neuronal death (P<0.01 by repeated-measures ANOVA, P<0.001 at 6 and 12 h by paired t test). e, Surface levels of GluR1 were assessed by biotinylation assay 10 min after the indicated treatments. Aβ₁₋₄₂ (10 μM) increased surface levels of GluR1 compared with Aβ₄₂₋₁, an effect that could be blocked with AACOCF₃ pretreatment and replicated with AA (mean ± s.e.m). f, Surface levels of GluR1 decreased to baseline levels after 30 and 60 min of exposure to Aβ₁₋₄₂ or AA (mean ± s.e.m.). ****P<0.0001 versus Aβ₄₂₋₁ (Tukey test).

GIVA-PLA₂ reduction improves learning and memory in hAPP mice. a,b, Mice (n=8−15 per genotype; age: 4−6 months) were tested in the cued (a) and hidden (b) platform components of the Morris water maze. In both components, hAPP/PLA₂^+/+ mice differed from all groups without hAPP (P<0.0001, by repeated-measures ANOVA and Tukey test). In the hidden platform component, hAPP/PLA₂^+/+ mice learned more poorly than hAPP/PLA₂^+/− mice (P<0.05) and hAPP/PLA₂^−/− mice (P<0.01, by repeated-measures ANOVA and Tukey test). c, Representative swim paths of individual mice with indicated genotypes during the last probe trial (platform removed). The blue square indicates the target location during hidden platform training. d, Number of times mice crossed this target location during the last probe trial compared with how many times they crossed corresponding locations in non-target quadrants. Only hAPP/PLA₂^+/+ mice failed to show a clear target location preference. A two-way ANOVA of target/nontarget ratios revealed significant effects of hAPP (P<0.0001) and PLA₂ (P<0.001) and a significant interaction between hAPP and PLA₂ genotype (P<0.001). *P<0.05, **P<0.01, ***P<0.001 vs. nontarget locations; ^###P<0.001 (Tukey test, mean ± s.e.m.).

GIVA-PLA₂ reduction prevents hyperactivity, abnormal anxiety/exploration-related behavior, and premature mortality in hAPP mice. a–g, Mice (n=8−15 mice per genotype, age: 4−6 months) were tested in the open field (a,b), the Y maze (c,d), and the elevated plus maze (e–g). Genotype effects and interactions were assessed by two-way ANOVA. a, Total movements (P<0.001 for hAPP effect, P<0.0001 for PLA₂ effect, P<0.01 for interaction). b, Total rearings (P<0.001 for hAPP effect, P<0.05 for PLA₂ effect, and P<0.05 for interaction). c, Total entries (P<0.05 for hAPP effect, P<0.0001 for PLA₂ effect, and P<0.001 for interaction). d, Total alternations (P<0.0001 for hAPP effect, P<0.001 for PLA₂ effect, and P<0.0001 for interaction). (e) Percent time spent in open arms (P<0.0001 for hAPP effect, P<0.05 for PLA₂ effect, and P=0.259 for interaction). (f) Percent distance traveled in open arms (P<0.0001 for hAPP effect, P<0.05 for PLA₂ effect, P=0.237 for interaction). g, Edge pokes (P<0.0001 for hAPP effect, P<0.01 for PLA₂ effect, P=0.133 for interaction). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 vs. PLA₂^+/+ mice or as indicated by brackets (Dunnett test, mean ± s.e.m.). h, Kaplan-Meier survival analysis of 189 mice revealed premature mortality in hAPP/PLA₂^+/+ mice (P<0.001 by log-rank chi-square test), but not in hAPP/PLA₂^+/− or hAPP/PLA₂^−/− mice.

Reduction or removal of GIVA-PLA₂ did not affect hAPP or Aβ levels in hAPP mice. a, Hippocampal hAPP protein levels (relative to GADPH) in hAPP mice with different levels of GIVA-PLA₂ expression (n=6 per group; age, 4−6 months) were determined by western blotting and densitometric analysis of signals. b–d, Hippocampal and cortical levels of Aβ_1-x (b) and Aβ₁₋₄₂ (c) and Aβ₁₋₄₂/Aβ_1-x ratios (d) in 4−6-month-old hAPP mice with different levels of GIVA-PLA₂ expression (n=6−10 per group; age, 4−6 months) were determined by ELISA and expressed as ng per g of tissue.