Ablation of endophilin A1 in CA1 neurons disrupts the E/I balance and increases epilepsy susceptibility.

(A) Experimental protocol of PTZ kindling. Saline serves as negative control for PTZ.

(B) Representative images of convulsive behavior in each group.

(C) Seizures were monitored and scored after each injection. Nine 10-week-old mice were used in each condition. Seizure scores are shown as means ± S.E.M. Statistical test: repeated measures ANOVA followed by Tukey post-hoc test at each time point. * p < 0.05, ** p < 0.01, *** p < 0.001, EndoA1-/-_PTZ vs EndoA1+/+_PTZ at each time point.

(D) The representative mCherry-positive hippocampal CA1 neuron used for electrophysiological recording.

(E-F) Representative traces of evoked excitatory postsynaptic currents (eEPSC) and quantification of mean eEPSC amplitude (E), and paired-pulse recording (F) from 10 control (Ctrl, uninfected) and 10 EndoA1 KO (mCherry and Cre) pyramidal neurons of 4 independent animals.

(G-H) Representative traces of evoked inhibitory postsynaptic currents (eIPSCs) and quantification of mean eIPSC amplitude (G), and paired-pulse recording (H) from 9 control and 9 EndoA1 KO pyramidal neurons of 4 independent animals.

(I) Effect of EndoA1 KO on the E/I ratio from 9 control and 9 EndoA1 KO pyramidal neurons of 4 independent animals.

(J) Representative recordings of miniature excitatory postsynaptic currents (mEPSCs) traces (+10 mV) from control or EndoA1 KO CA1 pyramidal cells.

(K) Cumulative probability and quantification of mEPSC amplitude from 7 control and 7 EndoA1 KO pyramidal neurons of 3 independent animals.

(L) Cumulative probability and quantification of mEPSC frequency from 7 control and 7 EndoA1 KO pyramidal neurons of 3 independent animals.

(M) Representative recordings of miniature inhibitory postsynaptic currents (mIPSCs) traces (–70 mV) from control or EndoA1 KO CA1 pyramidal cells.

(N) Cumulative probability and quantification of mIPSC amplitude from 7 control and 7 EndoA1 KO pyramidal neurons of 3 independent animals.

(O) Cumulative probability and quantification of mIPSC frequency from 7 control and 7 EndoA1 KO pyramidal neurons of 3 independent animals. Statistical test: two-tailed Student’s t-test in E-I & K, L, N, O. ns, non-significant. All data shown are scatterplots with means ± S.E.M.

EndoA1 KO causes defects in the formation and stabilization of inhibitory synapses.

(A-B) Immunofluorescence staining and confocal microscopy of gephyrin (GPN) in the hippocampal region of 3-week-old (P21) and 10-week-old (P60) EndoA1+/+ or EndoA1-/- mouse brains. Scale bar, 50 μm.

(C) Quantification of the signal intensity of GPN in (A) and (B). n = 6 mice for each genotype. (D-E) Same as (A-B) except that slices were immunostained for VGAT, (D) for P21, (E) for P60. Scale bar, 50 μm.

(F) Quantification of VGAT signal intensity in (D) and (E). n = 6 mice for each genotype.

(G-H) Same as (A-B) except that slices were immunostained for GABAAR γ2 (γ2). (G) for P21, (H) for P60. Scale bar, 50 μm.

(H) Quantification of γ2 signal intensity in (G) and (H). n = 6 mice for each genotype. Statistical test: unpaired two-tailed Student’s t-test. ns, non-significant. All data are scatterplots with means ± S.E.M. Abbreviations: CA1-3, cornus ammonis region 1-3; DG, dentate gyrus; DAPI staining of cell nuclei is shown in blue.

Endophilin A1 is distributed in the iPSD and regulates the assembly of the inhibitory postsynaptic machinery.

(A) Cultured EndoA1fl/fl hippocampal neurons were transfected with construct expressing GFP or GFP-2A-Cre on DIV12, fixed and immunostained for surface GluA1 (sGluA1) or GABAAR γ2 (sγ2) on DIV16. Shown are representative confocal microscopy images. Lower panels are magnification of boxed regions in the upper panels. Scale bar, 10 μm in the upper panels and 5 μm in the lower panels.

(B) Quantification of the signal intensity (left) and cluster number (right) of GPN puncta in dendrites in (A). GFP: 15 neurons; Cre: 18 neurons, from 3 independent cultures. Statistical test: unpaired two-tailed Student’s t-test. ns, non-significant.

(C-D) Cultured EndoA1fl/fl hippocampal neurons were co-transfected with constructs for GFP or GFP-2A-Cre and Flag vector or Flag-endophilin A1 (Flag-EndoA1) on DIV12, and immunostained for surface γ2 on DIV16. Shown are representative confocal images (C) and quantification of surface γ2 in dendrites (D). Scale bar, 2 μm. GFP: 18 neurons; Cre: 19 neurons; Cre + EndoA1: 18 neurons, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test.

(E-F) EndoA1fl/fl hippocampal neurons were infected with adeno-associated virus (AAV) carrying mCherry or mCherry-2A-Cre on DIV3 and surface biotinylated on DIV16. Total and surface proteins were detected by SDS-PAGE and immunoblotting. N = 3. Statistical test: unpaired two-tailed t-test.

(G) Hippocampal neurons were transfected with construct expressing GFP to label neuronal morphology on DIV12, then immunostained with antibodies against EndoA1 and GPN on DIV16. Shown are representative z stack maximum projection images captured by structured illumination microscopy (SIM). Lower panels are insets of boxed regions in the upper panel. Dendrites and spines are outlined in white. Scale bars, 2 μm in upper panels and 0.5 μm in magnified images.

(H) Same as (G) except that neurons were immunostained for EndoA1 and surface γ2.

(I) Same as (G) except that neurons were immunostained for EndoA1 and VGAT.

(J) Immunoisolation from the membrane fraction of mouse brain tissues with antibodies against GABAAR α1 (α1) or EndoA1.

(K) Immunoisolation from EndoA1+/+ and EndoA1-/- brain lysates with antibodies to GABAAR α1. Shown are representative immunoblot (left) and quantification of neuroligin 2 (NL2), GPN and GABAAR γ2 co-immunoisolated with α1 (right). N = 6 independent experiments. Statistical test: unpaired two-tailed Student’s t-test.

(L) PSD fractions were isolated by differential centrifugation from mouse brain lysates and analyzed by immunoblotting with antibodies to EndoA1 and synaptic proteins. S1, homogenates; PSD, postsynaptic density. SYP, synaptophysin. N = 7 independent experiments. Statistical test: unpaired two-tailed Student’s t-test. All data represent scatter plots with means ± S.E.M.

Endophilin A1 interacts with gephyrin and promotes its clustering in postsynaptic neurons.

(A) Silver staining of proteins co-immunoprecipitated from EndoA1+/+ mouse brain lysates with antibodies to endophilin A1.

(B) Gene Ontology (GO) enrichment analysis of EndoA1-interacting proteins identified by mass spectrometry.

(C-D) HEK293T cells co-transfected with constructs expressing Myc-tagged GPN or HA- tagged NL2 and Flag-tagged EndoA1 for 48 h were lysed for Flag IP. Input and bound proteins were analyzed by immunoblotting with antibodies against Myc, HA and Flag.

(E) Co-IP from brain lysates with antibodies to EndoA1.

(F) HEK293T cells overexpressing Flag-tagged endophilin As and Myc-tagged GPN were lysed for co-immunoprecipitation.

(G) EndoA1fl/fl hippocampal neurons were co-transfected with constructs expressing GFP and Flag vector, or GFP-2A-Cre and Flag vector, Flag-EndoA1, Flag-EndoA2 or Flag-EndoA3 on DIV12, fixed on DIV16 and immunostained for GPN, GFP and Flag. Shown are representative confocal microscopy images. Scale bar, 2 μm.

(H) Quantification of GPN signal intensity (left) and cluster number (right) in dendrites in (G). 36 neurons for GFP + vector; 34 neurons for Cre + vector; 38 neurons for Cre + EndoA1; 29 neurons for Cre + EndoA2; 27 neurons for Cre + EndoA3, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test.

(I) Hippocampal neurons were co-transfected with constructs expressing GFP and vector or Flag-EndoA1 on DIV12, fixed and immunostained for GPN on DIV16. Shown are representative confocal microscopy images of GPN in the soma (left) and dendrites (right). Scale bars, 5 μm in the upper panels and 2 μm in the lower panels.

(J) Quantification of GPN signal intensity (left) and cluster number (right) in the soma and dendrites. 34 neurons for GFP + vector; 37 neurons for GFP + EndoA1, from 3 independent cultures. Statistical test: unpaired Student’s two-tailed t-test. All data represent scatterplots with means ± S.E.M.

Both BAR and SH3 domains are required for endophilin A1-mediated organization of iPSD.

(A) EndoA1fl/fl hippocampal neurons expressing GFP or Cre were immunostained for GPN, surface γ2 and GFP on DIV16. Shown are representative confocal microscopy images. Scale bars, 5 μm in left panels and 1 μm in magnified images.

(B-C) Quantification of the mean intensity of GPN (green) and surface γ2 (red) and their colocalization in (A). 21 neurons for GFP, 22 neurons for Cre, from 3 independent cultures. Statistical test: unpaired two-tailed Student’s t-test. All data represent scatterplots with means ± S.E.M.

(D) DIV12 hippocampal neurons were co-transfected with constructs expressing GFP and vector or Flag-EndoA1, fixed on DIV16 and immunostained for surface γ2. Shown are representative confocal microscopy images of surface γ2 in the soma (left) and dendrites (right). Scale bars, 5 μm in the upper panels and 2 μm in the lower panels.

(E) Quantification of surface γ2 mean intensity (left) and cluster number (right) in the soma and dendrites. Soma: 20 neurons for GFP + vector, 20 neurons for GFP + EndoA1; Dendrites: 35 neurons for GFP + vector, 30 neurons for GFP + EndoA1, from 3 independent cultures. Statistical test: unpaired two-tailed Student’s t-test. All data represent scatterplots with means ± S.E.M.

(F) Schematic representation of EndoA1 and GPN domain structure and diagrams of fragments used for interaction mapping.

(G) GST-tagged EndoA1 fragments conjugated to Glutathione-Sepharose beads were incubated with lysates of HEK293T cells overexpressing Myc-tagged GPN. Input and bound proteins were analyzed by immunoblotting with antibodies against Myc. Bottom panel: Coomassie blue-stained SDS-PAGE gel showing GST fusion proteins. FL, full length.

(H) HEK293T cells co-transfected with constructs encoding Myc-tagged GPN and Flag-tagged EndoA1 fragments were lysed for Flag IP. Input and bound proteins were analyzed by immunoblotting with antibodies against Myc and Flag.

(I) Binding of His-tagged EndoA1 to GST-tagged GPN fragments in pull-down assay. Bound proteins were analyzed by SDS-PAGE and immunoblotting with anti-His antibodies. Input and bound proteins were analyzed by immunoblotting with antibodies against His. Bottom panel: Coomassie blue-stained SDS-PAGE gel showing GST fusion proteins. FL, full length. Right panel: Quantification of EndoA1 binding to GPN domain. N = 4 independent experiments. Statistical test: one-way ANOVA by Tukey post hoc test. All data represent scatterplots with means ± S.E.M.

(J) DIV12 EndoA1fl/fl hippocampal neurons were co-transfected with constructs expressing GFP and Flag vector or GFP-2A-Cre, and Flag vector or Flag-EndoA1 FL, Flag-EndoA1 ΔSH3, Flag-EndoA1 aa172-352, fixed and immunostained for GPN on DIV16. Shown are representative confocal microscopy images of GPN. Scale bar, 2 μm.

(K) Quantification of mean intensity (left) and cluster number (right) in dendrites are shown for GPN in (J). 41 neurons for GFP + vector; 44 neurons for Cre + vector; 35 neurons for Cre + FL; 24 neurons for Cre + ΔSH3; 27 neurons for Cre + aa172-352, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test. All data represent scatterplots with means ± S.E.M.

(L) Same as (I) except that neurons were immunostained for surface γ2. Scale bar, 2 μm.

(M) Quantification of mean intensity (left) and cluster number (right) in dendrites are shown for surface γ2 in (K). 38 neurons for GFP + vector; 31 neurons for Cre + vector; 26 neurons for Cre + FL; 33 neurons for Cre + ΔSH3; 36 neurons for Cre + aa172-352, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test. All data represent scatterplots with means ± S.E.M.

Actin polymerization is required for stabilization of inhibitory postsynapses.

(A-D) DIV12 cultured hippocampal neurons transiently expressing GFP were treated with DMSO (vehicle control), 1 μM Latrunculin A (LatA) or 1 μM Nocodazole (Noco) for 60 min (A) or 120 min (C) on DIV16, respectively. Changes in postsynaptic proteins GPN (red) and PSD95 (green) were detected by confocal microscopy. Quantitative data shown were the mean intensity (left) and cluster number (right) of GPN and PSD95 at 60 min (B) and 120 min (D). Ctrl: 20 neurons for 60 min; 29 neurons for 120 min; LatA: 20 neurons for 60 min; 33 neurons for 120 min; Noco: 23 neurons for 60 min; 22 neurons for 120 min, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test. Scale bar, 5 μm.

(E-H) Same as (A-D) except that neurons were immunostained for GPN (red) and surface γ2 (green). Changes in GPN and surface γ2 at 60 min (E) or 120 min (G) were detected by confocal microscopy. Quantitative data shown were the mean intensity (left) and cluster number (right) of GPN and surface γ2 at 60 min (F) and 120 min (H). Ctrl: 35 neurons for 60 min; 24 neurons for 120 min; LatA: 34 neurons for 60 min; 20 neurons for 120 min; Noco: 34 neurons for 60 min; 24 neurons for 120 min, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test. Scale bar, 5 μm.

(I) Levels of synaptic proteins normalized to control neurons following LatA or Noco treatment in (A,C,E,G).

(J-K) The colocalization between surface γ2 and GPN signals in dendrites in (E) and (G). Statistical test: one-way ANOVA by Tukey post hoc test. All data are scatterplots with means ± S.E.M.

Both the membrane-binding and p140Cap-binding capacities of endophilin A1 are required for iPSD assembly and E/I balance.

(A) Cultured hippocampal neurons from EndoA1fl/fl mice were co-transfected with construct expressing GFP and Flag vector or GFP-2A-Cre and Flag vector, Flag-EndoA1 wild-type (WT), mutants (KKK-EEE or Y343A) or p140Cap, and analyzed on DIV16 by immunofluorescence staining for GPN (green) and GFP (red). Scale bar, 2 μm.

(B) Quantification of GPN mean intensity (left) and cluster number (right) in (A). 35 neurons for GFP vector; 43 neurons for Cre + vector; 32 neurons for Cre + WT; 33 neurons for Cre + KKK-EEE; 36 neurons for Cre + Y343A; 33 neurons for Cre + p140Cap, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test.

(C) Same as (A) except that neurons were immunostained for surface γ2 on DIV16. Scale bar, 2 μm.

(D) Quantification of surface γ2 mean intensity (left) and number (right) in (C). 44 neurons for GFP + vector; 42 neurons for Cre + vector; 32 neurons for Cre + WT; 35 neurons for Cre + KKK-EEE; 35 neurons for Cre + Y343A; 32 neurons for Cre + p140Cap, from 3 independent cultures. Statistical test: one-way ANOVA by Tukey post hoc test.

(E) Experimental scheme for seizure scoring of EndoA1fl/fl mice stereotactically co-injected with AAVs carrying mCherry and Cre, and EGFP and EndoA1 or its mutants (KKK-EEE and Y343A) in the hippocampal CA1 region. Mice were intraperitoneally administered PTZ for 12 days after AAV injections, then seizures were scored.

(F) PTZ-induced seizures in mice injected with the indicated AAVs were scored for 30 min. Ctrl, 7 mice; Cre, 7 mice; Cre + WT, 7 mice; Cre + KKK-EEE, 6 mice; Cre + Y343A, 7 mice.; Statistical test: repeated measures ANOVA followed by Tukey post-hoc test at each time point.

* p < 0.05, ** p < 0.01, *** p < 0.001, red color is for Cre vs Ctrl, green color is for Cre + WT vs Cre. All data are means ± S.E.M.

(G-K) AAVs expressing Cre (AAV-mCherry-2A-Cre) and GFP (AAV-EGFP) or Cre and EndoA1 WT or mutant (AAV-EGFP-2A-EndoA1, AAV-EGFP-2A-EndoA1 KKK-EEE, AAV-EGFP-2A-EndoA1 Y343A) were stereotaxically injected into the CA1 regions of EndoA1fl/flmouse brain at P0. Acute hippocampal slices were prepared on P14-P21 for dual recording of evoked excitatory postsynaptic currents (eEPSC) and inhibitory postsynaptic currents (eIPSC). Shown are representative traces and pairwise comparisons of noninfected (Ctrl) and infected neurons in the same slice. Scale bar = 50/50, 50/50, 50/50, 50/50, 40/20 (pA/ms) in (G); Scale bar = 100/100, 100/400, 100/100, 100/200, 300/100 (pA/ms) in (H); Scale bar = 50/50, 50/50, 50/50, 50/50, 50/50 (pA/ms) in (I); Scale bar = 100/100, 200/100, 100/100, 100/200, 100/200 (pA/ms) in (J). Statistical test: unpaired two-tailed Student’s t-test. ns, non-significant; All data are scatterplots with means ± S.E.M. All number of recording were from more than 7 pyramidal neurons of 3 independent animals.

Model for the role of endophilin A1 at the inhibitory postsynapses and E/I balance.

(A) Endophilin A1 directly interacts with the inhibitory postsynaptic scaffold protein gephyrin to promote organization of iPSD and synaptic recruitment/stabilization of the GABAARs via its plasma membrane association and actin polymerization-promoting activities. (B) Loss of endophilin A1 impairs inhibitory synaptic neurotransmission and disrupts the balance of excitatory/inhibitory activities in the neural circuits, leading to enhanced epileptic seizure susceptibility.