SORLA interacts with EphA4 and suppresses EphA4 activation in response to ligand stimulation. (A) SORLA coimmunoprecipitates with EphA4. EphA4 or APP was immunoprecipitated with the indicated antibodies and immunoblotted for SORLA, EphA4, or APP. (B) SORLA/EphA4 complexes can be reconstituted by exogenous expression. EphA4 was coexpressed with GST or SORLA-GST in HEK293T or HEKswAPP cell lines, and glutathione Sepharose precipitates were immunoblotted for EphA4, APP (left), or GST (right). A single representative experiment is shown in A and B. (C–E) SORLA overexpression attenuates EphA4 activation. (C) HEK293 cells stably expressing EphA4 were transfected with control or SORLA constructs and incubated with ephrinA1-Fc (A1-Fc) for the time indicated. EphA4 immunoprecipitates were immunoblotted for pY602 phosphorylation or total EphA4. Data in D represents mean ± SE from five independent experiments (*, P < 0.05; Student’s t test). (E) WT or SORLA-overexpressing transgenic (SORLA TG) cortical neurons were stimulated with ephrinA1-Fc, and control or EphA4 immunoprecipitates were immunoblotted for EphA4 pY602 or total EphA4 as indicated. Lysate inputs are shown (right). (F) Cumulative data from E; graphs represent mean ± SE from seven independent experiments (*, P < 0.05; Student’s t test).

SORLA overexpression attenuates ephrin-mediated growth cone collapse. (A and B) Characterization of ephrinA1-mediated growth cone collapse in hippocampal neurons. WT (A) or SORLA TG (B) hippocampal neurons were treated with Fc control or ephrinA1-Fc for 30 min at DIV3 and stained for tubulin (green) and F-actin (red). Boxed regions are magnified as indicated. Bars, 10 µm. (C and D) Evaluation of growth cone collapse in WT, transgenic SORLA (TG) or knockout (KO) hippocampal neurons. (C) WT, SORLA TG, and KO hippocampal neurons were treated with Fc or ephrinA1-Fc for 30 min and stained for F-actin and tubulin. Growth cones at the tips of axonal processes were imaged and scored into collapsed and noncollapsed categories. The percentage of growth cone collapse was calculated in a minimum of three independent dissections, with each data point in the graph representing a separate dissection. (D) Fold growth cone (GC) collapse observed with ephrinA1-Fc treatment over Fc control for WT, TG, or KO neurons calculated for each experiment in C. In C and D, graphs represent mean ± SE; *, P < 0.05; ****, P < 0.00001; Student’s t test.

SORLA overexpression attenuates ephrinA1-induced EphA4 clustering. (A) WT hippocampal neurons at DIV3, untreated or exposed to control Fc for 30 min or ephrinA1-Fc for the indicated time, were fixed and stained for EphA4 (green), F-actin (red), or tubulin (blue). Note that ephrinA1-Fc treatment for 5 min results in EphA4 clustering without growth cone collapse. Bar, 5 µm. (B–D) Evaluation of EphA4 clustering in WT, SORLA TG, and KO hippocampal neurons. (B) WT, SORLA TG, and KO hippocampal neurons were treated with ephrinA1-Fc for 5 min and stained for EphA4 (green) and tubulin (blue). Boxed regions are magnified in the adjacent panels. Bar, 10 µm. (C) EphA4 clustering within F-actin–rich areas was examined by microscopy in WT, SORLA TG, and SORLA KO hippocampal neurons at DIV3 by staining for EphA4 (green), F-actin (red), and tubulin (blue). Bar, 5 µm. EphA4 clusters ≥200 nm within F-actin–enriched growth cone regions were scored in a minimum of four independent dissections/experiments (D); graph represents mean ± SE, from individual experiments; *, P < 0.05; **, P < 0.003; Student’s t test. (E) Total EphA4 fluorescence intensities from individual growth cones scored in D were plotted; graph represents mean ± SE.

EphA4 interacts with the SORLA YWTD/EGF-like domain, and EphA4/SORLA interaction is attenuated by a T947M AD-associated SORLA mutation. (A) Schematic diagram of SORLA domains; an AD-associated T947M mutation resides within the YWTD/EGF-like region. (B) The SORLA YWTD/EGF-like region interacts with EphA4. Purified EphA4-FLAGhis6 was incubated with various SORLA domain-Fc fusion proteins immobilized on protein G Sepharose, and coprecipitation with EphA4 was evaluated by immunoblotting. Adjacent graph depicts relative EphA4 coprecipitation (YWTD/EGF set to 1.0). Results are from three independent experiments; graph represents mean ± SE. (C) Purified EphA4 was evaluated for coprecipitation with WT or a T947M variant of the SORLA YWTD/EGF domain fused to Fc as in B. The adjacent graph depicts relative EphA4 binding with the WT construct set to 1.0 in four replicates from two independent experiments, mean ± SE; **, P < 0.005; Student’s t test. (D and E) The SORLA T947M mutation attenuates SORLA/EphA4 interaction and effects on EphA4 activation. HEK293 cells stably expressing EphA4 were transfected with control, WT, or T947M SORLA expression constructs, and EphA4 immunoprecipitates from lysates were evaluated for SORLA coprecipitation (D) or ephrinA1-Fc induced EphA4 activation (E) by immunoblotting. Graphs in D depict T947M coprecipitation relative to WT (set to 1.0; mean ± SE; **, P < 0.005; Student’s t test) from six replicates in four independent experiments. Graphs in E show relative pY602/EphA4 ratios with control vector treatments set to 1.0 (mean ± SE from five replicates in three independent experiments; *, P < 0.015; paired Student’s t test).

SORLA suppresses Aβ-mediated EphA4 activation and attenuates cognitive and synaptotoxic deficits associated with oligomeric Aβ injection in SORLA TG mice. (A and B) SORLA TG mice show a slight reduction in EphA4 activation at steady state. (A) EphA4 immunoprecipitates from crude synaptosome fractions in 3-mo-old WT and SORLA TG mice were subjected to immunoblotting, and pY602/EphA4 ratios were plotted in B. Graph represents mean ± SE from WT, n = 12, and SORLA TG, n = 12 animals (*, P < 0.04; Student’s t test). (C–E) SORLA suppresses Aβ-mediated EphA4 activation. Aβ_1–42 oligomers (C) were applied to WT or SORLA TG cortical neurons for 2 h, and Eph4 immunoprecipitates (D) were evaluated for EphA4 activation. (E) Relative fold pY602/EphA4 ratios were evaluated for WT and SORLA TG (TG) neurons (graphs depict mean ± SE from three replicates from two independent experiments; *, P < 0.04; Student’s t test). (F and G) SORLA suppresses spatial memory defects associated with hippocampal Aβ injection. WT and SORLA TG mice (n = 12, each group; n = 48 total, two independent assay runs) were stereotactically injected with PBS or Aβ into the hippocampus and subjected to 5 d of navigational training to find a hidden platform. (F) Escape latency decreased for all groups except WT Aβ-injected animals (red). Graph represents mean ± SE (*, P < 0.03; **, P < 0.008; two-way ANOVA). (G) After training, a probe test was performed to determine whether animals could accurately recall and return to the platform area; the number of entries was automatically recorded for the experimental groups. Significance was determined using one-way ANOVA (mean ± SE; *, P < 0.05). (H and I) SORLA attenuates Aβ-mediated EphA4 activation. Hippocampal tissue from WT (H) and SORLA TG (I) mice injected with PBS or Aβ were subjected to biochemical fractionation after behavioral analysis, and EphA4 immunoprecipitates from membranes/total synaptosomes were immunoblotted for activated EphA4 (pY602) and total EphA4. pY602/EphA4 ratios were normalized to PBS injected samples (set to 1.0). Graphs represent mean ± SE; *, P < 0.05; Student’s t test.

EphA4 redistribution to PSD clusters and LTP impairment is attenuated in SORLA TG mice. (A and B) Hippocampal Aβ injection induces EphA4 redistribution to PSD95 puncta. WT and SORLA TG mice were injected with PBS or Aβ and subjected to immunohistological staining for Aβ (brown) using MOAB-2 antibody (A) or costaining for EphA4 (red) and PSD95 (green; B). DAPI (blue) insets in B depict the dentate gyrus region imaged. Bars: (A) 300 µm; (B) 100 µm. Boxes in A and B are magnified in adjacent panels. EphA4/PSD95 co-staining puncta (yellow) are marked by yellow circles in magified panels in B. (C) Graph indicates percentage of red/green overlapping puncta scored for five 100-µm-square regions from hippocampal sections (mean ± SE; **, P < 0.001; Student’s t test). (D and E) SORLA overexpression restores Aβ-induced LTP impairment. Mean fEPSP slopes normalized over baseline values are shown for WT and SORLA TG acute slices under steady-state conditions with or without 500 nM Aβ as indicated. Baseline recordings were taken 20 min before LTP induction (0 time point), and 60 min after LTP induction. Graph represents mean ± SE; *, P < 0.05. (E) Representative fEPSP recordings obtained before (black) and after (gray) high-frequency stimulation. (F) Cumulative fEPSP slopes between 50 and 60 min after LTP induction (mean ± SE, ≥5 slices per mouse/three mice per genotype/treatment category). Statistical values were calculated using two-way ANOVA; *, P < 0.05.

EphA4 activation is observed in human AD and correlates with decreased EphA4/SORLA interaction. (A) Total lysates were generated from a representative control and AD cohort and immunoblotted with the indicated antibodies. (B and C) Elevated EphA4 activity is observed in AD. (B) EphA4 was immunoprecipitated from total synaptotosome fractions generated from human frontal cortex and immunoblotted for activated EphA4, pY602, or SORLA as indicated. (C) Activated EphA4 (pY602/EphA4 ratios) as determined in B were plotted for control (n = 23) and AD (n = 30) patient cases. Graph represents mean ± SE (**, P < 0.02; Student’s t test). (D) Enhanced SORLA/EphA4 interaction correlates with attenuated EphA4 activation. SORLA/EphA4 and pY602/EphA4 ratios were individually plotted for control (black) and AD (red) patients. R² (goodness of fit) values are shown for control (black) and AD (red) data groups; the p-value (significant deviance of the slope from 0 by linear regression analysis) for the AD group was found to be significant.