Saturated free fatty acids and association with memory formation

The study addresses how lipid metabolites—particularly free fatty acids (FFAs) derived from phospholipids—contribute to learning and memory formation, processes classically attributed to protein and gene regulation. Neuronal membranes are rich in phospholipids, which are essential for synaptic function, membrane dynamics, and neurotransmission. Enzymatic remodeling by phospholipases generates lipid metabolites (diacylglycerols, inositol trisphosphate, lysophospholipids, and FFAs) that modulate membrane properties and act as signaling molecules. While polyunsaturated FFAs like arachidonic acid (AA) have established roles in neurotransmission and long-term potentiation, recent in vitro data suggest stimulation predominantly generates saturated FFAs, raising the hypothesis that saturated FFAs may also be critical in learning and memory. The purpose of this study is to map brain-wide distributions of FFAs and phospholipids and determine their activity-dependent changes during auditory fear conditioning (AFC), and whether these changes depend on NMDA receptor activation required for memory consolidation.

Prior work has shown that lipids in the brain serve structural, signaling, developmental, and metabolic functions and are crucial for synaptic vesicle cycling and neurotransmission. Canonical phospholipids typically carry saturated acyl chains (e.g., palmitic, stearic) at sn-1 and unsaturated acyls (e.g., arachidonic acid, DHA) at sn-2. Activation of phospholipase A2 (PLA2) releases unsaturated FFAs such as AA, which modulate neurotransmitter release, membrane fluidity, synaptic transmission, and LTP, and contribute to inflammation. Protein acylation—myristoylation (myristic acid) and palmitoylation (palmitic acid)—is prevalent in the nervous system and essential for synaptic plasticity, affecting trafficking and function of key synaptic proteins (e.g., PSD95, AMPA/NMDA receptors). Despite AA’s recognized role, its neuronal abundance is relatively low compared to saturated FFAs, and recent isotope-tagging studies (FFAST) in vitro found predominant generation of saturated FFAs upon stimulation, motivating investigation of saturated FFAs in learning. NMDA receptor-dependent mechanisms underlie memory consolidation, and NMDA antagonism (CPP) blocks associative learning, providing a framework to link lipid changes with behavioral outcomes.

Design: Male Sprague–Dawley rats (6–8 weeks, 380±15 g) underwent auditory fear conditioning (AFC) under four conditions: saline-unpaired, saline-paired, CPP-unpaired, CPP-paired (n=8 per condition; total 32 animals). CPP (10 mg/kg i.p.) or saline was administered before AFC. Paired AFC: five 5 kHz, 75 dB, 20 s tones co-terminated with footshock (0.6 mA, 0.5 s) after 120 s habituation; unpaired groups received the same number of tones and shocks without temporal association (≥20 s separation). After AFC, animals recovered for 2 h before sacrifice. Tissue collection: Animals were deeply anesthetized and transcardially perfused with ice-cold oxygenated ACSF to remove blood and rapidly chill the brain. Brains were snap-frozen in liquid nitrogen, stored at −80 °C, sectioned (80–100 µm) in a −20 °C cryostat, and six regions were microdissected on ice: central amygdala (CeA), basolateral amygdala (BLA), prefrontal cortex (PFC), ventral hippocampus (VH), dorsal hippocampus (DH), and cerebellum (CB). Samples remained frozen during processing. Lipidomics overview: Two targeted approaches were used: (1) FFAST (Free Fatty Acid Stable isotope Tagging) to quantify 18 FFAs with nanomolar sensitivity; (2) targeted LC–MS/MS of 135 phospholipid species across five classes (PA, PC, PE, PG, PS), using negative-mode diagnostic fragmentation to assign acyl composition (sn-1/sn-2 species-level without positional isomer resolution). FFA extraction and FFAST labeling: Tissue homogenized in cold acid; FFAs extracted with chloroform/methanol and dried. Carboxyl groups were derivatized using CDI and isotopic FFAST tags: experimental samples labeled with FFAST-124 or FFAST-127; internal standards (18 FFA standards) labeled with FFAST-138. Labeled paired/unpaired samples from matched brain regions were combined with internal standards for multiplexed analysis. FFA LC–MS/MS: AB Sciex 5500 QTRAP with Shimadzu Nexera UHPLC; positive ESI, MRM transitions for each labeled FFA; C18 column at 65 °C; gradient of water/acetonitrile with 0.1% formic acid; 6 min runs. Quantification used calibration curves (0.05–25 ng/mL; R²>0.99) and internal standard normalization; concentrations normalized to tissue weight (pmol/mg). Phospholipid extraction and LC–MS/MS: Bligh–Dyer extraction with internal standards (PC/PE/PS/PA/PG 17:0/17:0). Negative-mode MRM on same instrument with HILIC column at 65 °C; gradient of acetonitrile/50 mM ammonium formate (pH 3.55). Class-based calibration (0.15–5 µg/mL; LLOD ~0.15 µg/mL, LLOQ ~0.60 µg/mL on-column) with internal standards; acyl fragment ions used to determine species-level composition. Behavior: Freezing to tone measured at 0 h and 24 h post-AFC. Statistics by one-way ANOVA with Sidak multiple comparisons. Data processing and statistics: Peak integration with Multiquant MQ4. Custom Python scripts (Pandas, Numpy, Seaborn, SciPy) for quantification, outlier removal by median filtering, means±SEM, stacked barplots, heatmaps. Fold-change (paired vs unpaired) significance by two-tailed Student’s t-test (p<0.05 threshold). Multivariate analysis: Isomap non-linear dimensionality reduction on normalized profiles (18-FFA vectors for 24 samples: 6 regions × 4 conditions) to 3D, grouping by condition. Hierarchical clustering heatmaps for region/class responses.

• Baseline distribution: FFAs and phospholipids were heterogeneously distributed across brain regions, with highest concentrations (per mg tissue) in amygdala (CeA/BLA) and prefrontal cortex; lowest in dorsal hippocampus and cerebellum. Saturated FFAs (C6:0–C24:0) predominated across regions; among unsaturated FFAs, DHA (22:6), arachidonic acid (20:4), and oleic acid (18:1) were most abundant.
• AFC-induced changes (saline-treated): Paired AFC caused significant brain-wide increases in several FFAs, most notably saturated myristic acid (C14:0) and to a lesser extent palmitic acid (C16:0), along with increases in AA (20:4). Phospholipid classes generally decreased (PA, PC, PE, PS) with a slight increase in PG, indicating activity-dependent phospholipid remodeling.
• NMDA receptor dependence: In CPP-treated animals, paired AFC resulted in decreases across FFAs and phospholipids, and blocked long-term memory consolidation (reduced freezing at 24 h). Behaviorally, both saline and CPP groups showed similar freezing immediately after AFC (0 h), but CPP significantly reduced freezing at 24 h, indicating NMDA receptor-dependent consolidation. Lipids were assayed 2 h post-AFC, when FFA profiles already diverged between saline and CPP despite similar short-term behavior, linking FFA changes to consolidation.
• Regional specificity: The strongest FFA responses to AFC occurred in the amygdala, particularly the CeA, with myristic acid showing the largest absolute and fold increases. PFC also showed broad increases; hippocampal regions showed smaller absolute FFA changes but relatively higher AA responses compared to other FFAs. CPP treatment markedly diminished these changes across regions, especially in CeA and DH.
• Phospholipid–FFA relationships: Decreases in certain phospholipid species containing myristoyl chains (e.g., PA/PC/PS with 20:4_14:0; 22:6_14:0; 16:0_14:0) were observed in CeA, potentially reflecting substrates for phospholipase-mediated release of myristic acid. However, overall acyl distributions in FFAs did not mirror those in phospholipids, suggesting enzymatic specificity and region-dependent substrate utilization. Free palmitic acid did not simply track its high abundance in PC.
• Multivariate profiling: Isomap NLDR of FFA profiles grouped samples by the four experimental conditions (saline/CPP × paired/unpaired), correlating with long-term memory outcomes. Similar clustering was not observed for phospholipid classes, indicating more complex, region-specific phospholipid responses.
• Sample sizes and statistics: n=8 animals per condition per analysis; significance for lipid fold-changes by two-tailed t-test (p<0.05). Behavioral ANOVA with Sidak multiple comparisons showed significant CPP-induced impairment of freezing at 24 h.

The study demonstrates that learning-associated neural activity during AFC rapidly remodels the brain’s phospholipid metabolite landscape, with robust increases in saturated FFAs—especially myristic acid—in regions critical for fear learning (amygdala and PFC). These changes are NMDA receptor-dependent and precede the consolidation of long-term memory, as evidenced by their presence 2 h post-AFC when short-term behavioral responses are still indistinguishable between saline- and CPP-treated animals. The data suggest that saturated FFAs, alongside arachidonic acid, likely contribute to synaptic plasticity mechanisms required for consolidation. Potential mechanisms include: (1) provision of substrates for protein acylation (myristoylation and palmitoylation) that regulate synaptic protein localization, receptor organization (e.g., PSD95, AMPA/NMDA receptors), and signaling; (2) modulation of membrane biophysical properties influencing vesicle trafficking and neurotransmitter release; and (3) activation of lipid-signaling pathways. The mismatch between FFA and phospholipid acyl distributions implies selective phospholipase activity (PLA1 and/or PLA2 on canonical or non-canonical phospholipids) and region-specific enzymology during learning. The strong amygdala response aligns with its established role in fear conditioning circuitry, while smaller hippocampal responses and distinct AA profiles reflect regional functional specializations. Overall, the findings extend the understanding of lipid involvement in memory from a focus on polyunsaturated FFAs to a prominent role for saturated FFAs in NMDA-dependent consolidation.

This work provides the first comprehensive, brain-wide mapping of activity-dependent changes in FFAs and phospholipids during learning, showing that auditory fear conditioning elicits robust, NMDA receptor-dependent increases in saturated FFAs (notably myristic and palmitic acids) and specific decreases in phospholipid species, with strongest effects in the amygdala and prefrontal cortex. The results implicate saturated FFAs as critical contributors to memory acquisition and consolidation, likely via effects on synaptic protein acylation and membrane dynamics. Future directions include: directly linking FFA generation to acyl-CoA pools and protein acylation dynamics during learning; resolving positional isomers and lipid localization with advanced MS imaging; dissecting the roles of specific phospholipases (PLA1 vs PLA2) and lipid-metabolizing enzymes; expanding to different learning paradigms, sexes, and developmental stages to assess generalizability.

• Positional isomers of phospholipids (sn-1/sn-2) could not be resolved; species assignments reflect acyl composition but not definitive positional configurations, limiting substrate–product inference.
• Phospholipids are far more abundant than FFAs, complicating direct substrate–product mass balance assessments; multiple lipid pools (lysophospholipids, acylglycerols) may contribute to FFA changes.
• Measurements were performed 2 h post-AFC; temporal dynamics before and after this time point were not resolved.
• Only male mid–late adolescent rats were studied; sex- and age-related differences in lipid responses and behavior may limit generalizability.
• The study correlates FFA changes with consolidation but does not directly demonstrate causal mechanisms (e.g., protein acylation states were not measured in parallel).
• CPP affected some basal FFA levels regionally; while activity-dependent changes were analyzed, pharmacological off-target or systemic effects cannot be entirely excluded.