EphB2 mediates social isolation-induced memory forgetting

The study investigates how adolescent social isolation (SI) impacts specific phases of memory and the molecular mechanisms involved. Prior work shows SI is detrimental to mental health and cognition, but it is unclear whether deficits stem from encoding, recall, or accelerated forgetting. The hippocampus, particularly CA1, is central to memory formation and fear memory circuitry, with dendritic spine structure and postsynaptic glutamate receptor complexes being critical for synaptic remodeling and function. EphB receptors, components of the glutamate receptor complex, regulate synapse formation, spine morphogenesis, and receptor recruitment, and have roles in cognition. The authors hypothesize that post-weaning SI accelerates forgetting of long-term fear memory via alterations in hippocampal synaptic plasticity mediated by EphB2 in CA1.

Background literature highlights the importance of social bonds for survival and cognition; SI is linked to poorer cognitive performance, faster decline, and depressive cognition. Post-weaning SI impairs spatial and social memory, but its effect on memory stages (encoding vs forgetting) has been unresolved. The hippocampal CA1 integrates inputs for memory formation and is essential for fear memory through its connections to cortex and amygdala. Dendritic spine abnormalities correlate with synaptic dysfunction and memory impairment. Postsynaptic glutamate receptor multiprotein complexes (AMPA, NMDA receptors) and EphB receptors determine synaptic transmission through receptor clustering. EphB receptors are key in synaptogenesis, spine morphogenesis, and have established roles in cognition and memory. This prior evidence motivates testing whether EphB2 changes underlie SI-induced memory deficits.

Animals: Thy1-GFP-M transgenic mice and EphB2 knockout mice (both sexes) backcrossed to the 129 strain. All procedures followed IACUC-approved protocols at Shanghai Jiao Tong University School of Medicine. Social isolation: Mice were randomly assigned to group housing (GH; 3–4 per cage) or social isolation (SI; single housing). Isolation started at postnatal week 4 (PW4) for durations of 2, 4, or 12 weeks. A 4-week SI model was used for most behavioral experiments. Fear conditioning and extinction: Pavlovian fear conditioning used context A (checkered walls; cleaned with 75% ethanol). Training: 3 min exploration, then two CS (28 s tone)–US (0.75 mA, 2 s footshock) pairings separated by 1 min; animals returned to home cage after 30 s. Tests: contextual memory (context A) at 1 h, day 1, and day 7 post-conditioning; auditory fear memory in context B (altered chamber, plexiglass floor, cleaned with 4% acetic acid) with 3 min exploration then 1 min tone. Extinction: unreinforced context exposures for 3 consecutive days. Freezing behavior recorded and quantified with Freeze Frame/FreezeView. Electrophysiology: Acute coronal brain slices (300 µm) prepared in ice-cold ACSF; recovery in oxygenated ACSF (95% O2/5% CO2) at 31 °C, then room temperature. mEPSCs recorded at −70 mV in presence of picrotoxin (100 µM) and TTX (1 µM). Paired-pulse ratio (PPR) recorded with bipolar concentric electrode in CA1; interstimulus intervals 25–400 ms. LTP: fEPSPs recorded in CA1 Schaffer collateral pathway; LTP induced with high-frequency stimulation (1 s 100 Hz train) at 70–80% of spike-evoking intensity; analysis in pCLAMP. Viral manipulations: Stereotaxic injections into CA1 (AP −1.7 mm, ML ±1.25 mm, DV −1.55 mm from skull surface). Viruses: AAV-EphB2 for overexpression; lentiviral shRNA (Lenti-sh-EphB2; sequence ACGGACAAGCTACAACACT) for knockdown. Post-injection, needle held 5 min to prevent backflow; recovery on 37 °C plate; experiments started 25–30 days later. Injection sites verified post hoc; off-target cases excluded. Immunohistochemistry: Slices blocked in 0.3% Triton X-100 PBS with 10% donkey serum; primary antibodies overnight at 4 °C (goat anti-EphB2 1:200); Alexa Fluor secondaries (1:200) for 2 h; confocal imaging (Leica TCS SP8). Spine classification with NeuronStudio using thresholds: AR_thin(crit)=2.5, HNR(crit)=1.3, HD(crit)=0.3 µm; protrusions length 0.2–3.0 µm; density = spine count/dendrite length. EphB2 fluorescence quantified in ImageJ with consistent thresholds. Western blotting: Hippocampi homogenized in CHAPS-based lysis buffer with inhibitors; SDS-PAGE and nitrocellulose transfer; antibodies included EphB1, EphB2, synaptophysin, PSD95, β-actin, GluN1, GluA1, GluA2, GluA6, GluN2A, GluN2B. Statistics: Mean ± SEM; Student’s t-tests for two-group comparisons; ANOVA with Tukey’s multiple comparisons for >2 groups; significance at P < 0.05.

• Behavioral memory: 2-week SI mice showed normal short- and long-term memory; 4-week SI mice exhibited impaired long-term fear memory evident 7 days after conditioning; 12-week SI mice showed earlier deficits including auditory fear at day 1. In 4-week SI vs GH: Context A (contextual) two-way ANOVA: time F(2,57)=3.522, P<0.05; group F(1,57)=4.767, P<0.05; interaction F(2,57)=2.743, P>0.05; Tukey: hour 1 and day 1 ns; day 7 P<0.05. Context B (auditory) time F(2,57)=3.432, P<0.05; group F(1,49)=8.377, P<0.01; interaction F(2,57)=3.284, P<0.05; Tukey: hour 1 and day 1 ns; day 7 P<0.01. Extinction did not differ between GH and SI; short-term memory (Y-maze) and sex differences were not significant.
• Morphology: Increased proportion of thin spines in CA1 of SI mice without change in total spine density; t-tests: total density t(36)=0.5472, P>0.05; mushroom t(36)=1.075, P>0.05; stubby t(36)=1.128, P>0.05; thin spines t=3.136, P<0.01.
• Synaptic physiology: mEPSC amplitude decreased in CA1 neurons from SI mice (t(29)=2.293, P<0.05) with unchanged frequency (t(29)=1.165, P>0.05); PPR unchanged (no presynaptic change). LTP in Schaffer collateral-CA1 pathway reduced in SI: two-way ANOVA for last 10 min fEPSP slope: group F(1,165)=51.72, P<0.0001; time F(10,165)=0.02342, P>0.05; interaction F(10,165)=0.01884, P>0.05.
• Molecular changes: Western blots showed decreased hippocampal GluN1, GluA1, GluA2, and notably EphB2 in SI mice. Immunofluorescence: EphB2 reduced in CA1 of SI mice, with no apparent change in cortex and CA3.
• Causality via EphB2: CA1-targeted lentiviral EphB2 knockdown in GH mice reduced freezing at day 7, increased thin spines, decreased mEPSC amplitude, and suppressed LTP—mimicking SI-induced deficits.
• Rescue experiments: CA1 AAV-EphB2 overexpression in SI mice increased EphB2 expression, restored long-term fear memory, normalized mEPSC and LTP, and corrected spine subtype distribution. Resocialization (2 weeks after 4-week SI) restored hippocampal EphB2 expression and increased freezing time compared to isolated mice, indicating environmental reversal of deficits via EphB2.

The findings demonstrate that adolescent social isolation accelerates forgetting of long-term fear memory without impairing initial acquisition or short-term memory. The memory loss correlates with reduced CA1 synaptic strength and plasticity, including decreased mEPSC amplitude, diminished LTP, and a shift toward thin spines, indicative of immature synapses. Molecularly, SI lowers levels of glutamate receptor subunits and specifically downregulates EphB2 in CA1. Manipulating EphB2 establishes causality: knockdown recapitulates SI-induced behavioral and synaptic phenotypes, while overexpression restores memory and synaptic function. Thus, EphB2 in CA1 is both necessary and sufficient to maintain long-term fear memory under social conditions, likely through its role in organizing postsynaptic receptor complexes and supporting spine maturation and synaptic plasticity. Environmental intervention (resocialization) rescues EphB2 levels and memory, underscoring the plastic nature of the deficit and highlighting EphB2 as a mechanistic link between social experience and memory persistence.

This study identifies EphB2-mediated synaptic mechanisms in hippocampal CA1 as critical for preventing social isolation-induced forgetting of long-term fear memory. Social isolation reduces EphB2 and glutamate receptor components, weakens synaptic transmission and plasticity, and biases spine morphology toward thin spines. Targeted EphB2 knockdown mimics, and EphB2 overexpression or resocialization reverses, the deficits, establishing EphB2 as a necessary and sufficient mediator. These findings suggest EphB2 as a potential therapeutic target to counteract memory loss associated with social isolation or loneliness. Future research could dissect downstream signaling pathways of EphB2, determine cell-type specificity within CA1 circuits, examine generalization to other memory domains, and explore pharmacological or environmental strategies to modulate EphB2 function.